AFLP and MS-AFLP analysis of the variation within saffron crocus (Crocus sativus L.) germplasm

Abstract

The presence and extent of genetic variation in saffron crocus are still debated, as testified by several contradictory articles providing contrasting results about the monomorphism or less of the species. Remarkably, phenotypic variations have been frequently observed in the field, such variations are usually unstable and can change from one growing season to another. Considering that gene expression can be influenced both by genetic and epigenetic changes, epigenetics could be a plausible cause of the alternative phenotypes. In order to obtain new insights into this issue, we carried out a molecular marker analysis of 112 accessions from the World Saffron and Crocus Collection. The accessions were grown for at least three years in the same open field conditions. The same samples were analysed using Amplified Fragment Length Polymorphism (AFLP) and Methyl Sensitive AFLP in order to search for variation at the genetic (DNA sequence) and epigenetic (cytosine methylation) level. While the genetic variability was low (4.23% polymorphic peaks and twelve (12) effective different genotypes), the methyl sensitive analysis showed the presence of high epigenetic variability (33.57% polymorphic peaks and twenty eight (28) different effective epigenotypes). The pattern obtained by Factorial Correspondence Analysis of AFLP and, in particular, of MS-AFLP data was consistent with the geographical provenance of the accessions. Very interestingly, by focusing on Spanish accessions, it was observed that the distribution of the accessions in the Factorial Correspondence Analysis is not random but tends to reflect the geographical origin. Two clearly defined clusters grouping accessions from the West (Toledo and Ciudad Real) and accessions from the East (Cuenca and Teruel) were clearly recognised.

Fig 1. Factorial Correspondence Analysis evidencing the relationships among AFLP genotypes and MS-AFLP epigenotypes of different saffron crocus accessions.

1A) Factorial Correspondence Analysis showing multivariate relationships among AFLP genotypes of different accessions on the axes corresponding to first (x axis, 9.72% inertia) vs. second (y axis, 6.75% of inertia) main factors (1A I). The two populations Pop A and Pop B correspond to the main clusters identified by STRUCTURE analysis at K = 2 (1A II). Population A included just the S accessions and three different genotypes, while population B included both S and NS accessions and 9 different genotypes. Inside population B, samples included in the dashed line refer to accessions from Iran, India, Afghanistan and Turkey. 1B) Factorial Correspondence Analysis showing multivariate relationships among MS-AFLP epigenotypes of different accessions on the axes corresponding to first (x axis, 12.88% of inertia) vs. second (y axis, 10.15% of inertia) main factors (1B I). The two populations Pop C and Pop D correspond to the main clusters identified by STRUCTURE analysis at K = 2 (1B II). Population C included just Spanish accessions and 6 effective epigenotypes, while population D included both S and NS accessions and 22 effective epigenotypes.

Fig 2. FCA analysis based on the MS-AFLP epigenotypes of Spanish accessions.

Factorial Correspondence Analysis showing multivariate relationships among MS-AFLP epigenotypes of different saffron crocus accessions of Spanish provenance on the axes corresponding to first (x axis, 12.88% of inertia) vs. second (y axis, 10.15% of inertia) main factors. Only the points corresponding to accessions of Spanish origin have been plotted. Accessions from the WEST (Toledo and Ciudad Real) and from the EAST (Cuenca and Teruel) tended to cluster separately with just few exceptions. Climatic conditions between the two areas are very different and this may be reflected in the epigenetic composition.