UNIVERSITE DE BOURGOGNE THÈSE Yongbo LIU - Université de ...

aphanistrum from hybrids and R. sativus (Hedge et al. 2006). We could apply these criteria
to o ur r esults r ather th an tr ying t o dr aw c lear-cut a ssignation t o given s ubspecies. T he
frequency of yellow a nd i ntense yellow pe tals was ne arly 100% i n BG a nd D M, a nd a ll
silique di ameters m easured l ess t han 5.1 m m. The e xistence of regional hom ogeneity o f
flower color w as poi nted out it s ome in stance: ju st b right yellow in S cotland, ju st w hite
around Y ork, but mixed c olors ne ar R othamsted ( Cousens, pe rsonnal c ommunication);
intense yellow in Poitou and white in other Britany populations and North Western France,
but mixed color in the Rhone valley (authors’ personal observations). In contrast, BT and NM
had mixed color, and their siliques were most often wider than 5.1 m m in diameter. It could
be indirect evidence of gene flow from cultivated radish (for the white color) or oilseed rape
(for the intense yellow color) to wild radish as observed elsewhere (Kercher and Conner 1996;
Snow e t a l. 2010) . B esides s ilique di ameter a nd f lower c olor, ot her s ilique a nd f lower
characteristics were also scored in this study. In every case as well as globally (Fig. 4.5), these
traits showed more polymorphism or larger range of variation in BT and NM than in BG and
DM.
Vegetative growth and reproduction
Seed set per radish plant was very low in the oilseed rape field, 53 seeds per plant on average,
and this may result from the small population size (81 plants) and the low density (0.01 pl m -2 )
because wild radish is self-incompatible and need both a variety of self-incompatibility alleles
in the stand and active pollinators. Small population size and low density limit opportunities
for i ndividuals t o m ate effectively an d i nfluences m aternal r eproduction ( e.g. E lam et al .
2007). In addition, f oraging i nsects w ere p robably poor ly attracted b y rare r adish flowers
embedded i n huge a mount of oi lseed r ape f lowers. T hese conditions s hould ha ve be en
favorable for interspecific crosses due to the abundant oilseed rape pollen available in contrast
to the poor amount of wild radish pollen, except in the case there was a specific pollinator to
wild r adish. H owever, t here w as no he rbicide-resistant p rogeny, which i ndicated t hat
interspecific hybridization, if any, can be comprised between 0 and 0.2 % (95 % confidence
limits of the data), and not different between NM and BT+BG+DN. The frequency of such
interspecific h ybridization i n t he f ield w as f ound t o be e xtremely l ow ( Chèvre e t a l. 2000;
Thalmann et al. 2001; Warwick et al. 2003), although once observed at 0.5 % (Darmency et al.
1998). Genotype and population polymorphism for the interspecific barriers to hybridization
could explain such widely different estimates (Guéritaine and Darmency 2001; Guéritaine et
154