Motivated by a previous study indicating that polymorphism at an indel, Ind2, within the Brassica nigra COL1 gene is significantly associated with flowering time, we searched for evidence of selection in a sample of 41 complete sequences of B. nigra COL1. The within-gene population recombination rate is moderate, and all neutrality tests used in the present study failed to detect departure from the standard neutral model or evidence of selection. The haplotype structure of the 5'-half of the gene is primarily associated with the demographic history of the species and more specifically with the split between European and Ethiopian populations, whereas the structure of the 3'-half reflects the polymorphism at Ind2. This could be the result of selection or a combination of recombination and migration during the history of the sample of sequences. Without additional information on polymorphism in flanking areas, these two alternatives are difficult to tell apart. If selection acted on the gene, we suggest that if the indel itself is not the target of selection, among the polymorphic sites cosegregating with the polymorphism at Ind2, replacement polymorphisms around sites 890 and 1260 are the most likely quantitative trait nucleotides within the gene.

The CONSTANS-like gene family has been shown to evolve exceptionally fast in Brassicaceae. In the present study we analyzed sequence polymorphism and divergence of three genes from this family: COL1 (CON-STANS-LIKE 1) and two copies of CO (CONSTANS), COa and COb, in B. nigra. There was a significant fourfold difference in overall nucleotide diversity among the three genes, with BniCOb having twice as much variation as BniCOL1, which in turn was twice as variable as BniCOa. The ratio of nonsynonymous-to-synonymous substitutions (dN/dS) was high for all three genes, confirming previous studies. While we did not detect evidence of selection at BniCOa and BniCOb, there was a significant excess of polymorphic synonymous mutations in a McDonald-Kreitman test comparing COL1 in B. nigra and A. thaliana. This is apparently the result of an increase in selective constraint on COL1 in B. nigra combined with a decrease in A. thaliana. In conclusion, a complex scenario involving both demography and selection seems to have shaped the pattern of polymorphism at the three genes.

Previous QTL mapping identified a Brassica nigra homolog to Arabidopsis thaliana CO as a candidate gene affecting flowering time in B. nigra. Transformation of an A. thaliana co mutant with two different alleles of the B. nigra CO (Bni COa) homolog, one from an early-flowering B. nigra plant and one from a late one, did not show any differential effect of the two alleles on flowering time. The DNA sequence of the coding region of the two alleles was also identical, showing that nucleotide variation influencing flowering time must reside outside the coding region of Bni COa. In contrast, the nucleotide sequence of the B. nigra COL1 (Bni COL1) gene located 3.5 kb upstream of Bni COa was highly diverged between the alleles from early and late plants. One indel polymorphism in the Bni COL1 coding region, present in several natural populations of B. nigra, displayed a significant association with flowering time within a majority of these populations. These data indicate that a quantitative trait nucleotide (QTN) affecting flowering time is located within or close to the Bni COL1 gene. The intergenic sequence between Bni COL1 and Bni COa displayed a prominent peak of divergence 1 kb downstream of the Bni COL1 coding region. This region could contain regulatory elements for the downstream Bni COa gene. Our data suggest that a naturally occurring QTN for flowering time affects the function or expression of either Bni COL1 or Bni COa.