by differences between individuals within cultivars, and 48.8% was explained by
differences among cultivars. The limited contribution to the intra-accession variance
of sex-linked markers in dioeciousCannabisis somehow surprising, given the
frequency with which such markers have been identified and exploited (see next
Section). Data obtained by SSR markers basically confirmed this partition of
variation (Gilmore and Peakall 2003 ). A similar extent of polymorphism of among-
and within-population variability and of heterozygosity was detected in a study
using AFLP to identify a group of threeCannabisvarieties asfiber and drug plants,
and the degree of reduction of such variation upon inbreeding of the plant material
(Datwyler and Weiblen 2006 ).
The composition of genetic variation obtained by the different studies is rela-
tively variable, but in general, it can be concluded that it reflects the great variability
present in theCannabisgermplasm, which is fully compatible with the reproductive
habits of a dioecious, outbred species such asC. sativa, and suggests the existence
of a widely-shared gene pool with limited genetic separation within different
groups.
Microsatellite studies were thefirst to reveal specific features of theC.sativa
genome. Hsieh et al. ( 2003 ) isolated thefirstC. sativa-specific microsatellite loci
containing a simple sequence repeat motif of 6 bp (CACCAT), with variations in
repeat unit length from 3 to 40; in the genotypes analyzed, the range found for these
loci was from 1 to 4 alleles per locus, suggesting the multi-locus nature of the
markers. Alghanim and Almirall ( 2003 ) identified GA/CT as the most common
motif in Cannabis genome, representing 50% of the total eight different
microsatellite repeats identified, followed by three nucleotide repeat motifs
(GTT/CAA, AAG/TTC and GAT/CTA) and by other minor motifs (GT/CA,
CAT/GTA, ACG/TGC and GGA/CCT). In the eleven loci found to be polymorphic
and reliable for scoring the different alleles in a population of 41Cannabissamples,
the number of alleles per locus ranged from 3 to 9 and the expected heterozygosity
ranged between 0.368 and 0.710. These SSRs proved to be effective in uniquely
identifying 27 profiles in theCannabissamples tested, discriminating the identities
of duplicates and unique samples. All these features identified these groups of
microsatellite markers as an excellent forensic tool, of potential use also for genetic
mapping.
Again in 2003, Gilmore and Peakall profiled 93Cannabisplants representing 9
drug and 6fiber accessions of different origins, usingfive microsatellite loci: a total
of 79 alleles were detected, allowing the attribution of a unique genotypic profile to
89 individuals, leaving therefore only 4 aside which derived from a single drug type
accession. PCA results suggested a much lower genetic diversity among drug-type
accessions when compared tofiber-type accessions, as expected by the particularly
strict selection procedures applied to drug strains, and confirming the differences
observed by the RAPDs analyses described above (Forapani et al. 2001 ), and also
matching the observations based on Single Nucleotide Polymorphisms (SNPs) (see
below). AMOVA analysis carried out on the basis of SSR data, showed that the
contribution of within-accessions variation to the total genetic variance observed
was 73%, while only 21% was due to intra–accession differences and 6% to
322 C. Onofri and G. Mandolino