Association mapping of essential oil components in Dalmatian sage (Salvia officinalis L.)
2012
The mixed-model association mapping approach using kinship coefficients (K-matrix) and the population structure information (Q-matrix) was applied to identify molecular markers associated with eight major essential oil compounds in Dalmatian sage (Salvia officinalis L.). Twenty-five natural populations originating from Croatia (23) and Bosnia and Herzegovina (2), each consisting of 20 to 25 plants, were genotyped using AFLP and SSR markers. Four AFLP primer combinations yielded 559 polymorphic markers while by using 8 SSRs a total of 161 alleles were identified. Essential oils were analyzed using analytical gas chromatography (GC/FID) and gas chromatography-mass spectrometry (GC/MS) techniques. Out of 62 compounds detected eight were found in concentrations higher than 5% in at least one sample (cis-thujone, camphor, trans-thujone, 1, 8-cineole, β-pinene, camphene, borneol and bornyl acetate). Data on each of eight compounds were subjected to mixed-model analysis using a range of K, Q and K + Q models based on different methods for construction of K and Q matrices. Pairwise kinship coefficients were based either on AFLP data (Hardy, 2003), or SSR data (Ritland, 1996 ; Loiselle et al., 1995) and estimated using the SPAGEDI software (Hardy and Vekemans, 2007). Population structure matrix (Q) based on SSR data was obtained by (a) STRUCTURE (Pritchard et al., 2000), (b) principal components analysis (PCA ; Patterson et al., 2006), (c) principal co-ordinate analysis (PCoA) based on proportion-of-shared-alleles distance matrix, and (d) factorial correspondence analysis (FCA) as implemented in Genetix 4.05 (Belkhir et al. 2004). For each variable (essential oil compound) the models were fitted in ASReml 3 (Gilmour et al., 2009) and compared using BIC. In most cases, K+Q models showed better goodness of fit than the other models. The models including K-matrix based on AFLP data generally fitted better than models including K-matrix based on SSR data. The best fitting models for different compounds varied in the Q part. However, “Hardy + STRUCTURE” model was chosen best in four out of eight cases.
Keywords:
- Correction
- Source
- Cite
- Save
- Machine Reading By IdeaReader
0
References
0
Citations
NaN
KQI