Genetic analysis of the control of flowering time in Arabidopsis thaliana
Arabidopsis is a facultative long-day plant, flowering earlier under long than short photoperiods. Many genetic mutations have been identified that affect the timing of the floral transition in Arabidopsis (Redei, 1962; Koornneef et al., 1991; Koornneef et al., 1998b) . These genetic mutations affect genes that are involved in responses to environmental stimuli and the autonomous control of flowering.
Natural genetic variation in flowering
Two methods have been used to identify genes that are involved in the genetics of the transition to flowering. The first approach involves the analysis of natural genetic variation in flowering-time control that occurs between different accessions of Arabidopsis, and the second is based on the isolation of flowering-time genetic mutants after mutagenesis.
Both genetic approaches have contributed to the understanding of genes that control the genetical transition to flowering, and have been complimentary to each other. For example the FRIGIDA (FRI) gene was initially identified by analysis of natural variation and not in mutagenesis experiments (Clarke and Dean, 1994) . Homologues of some Arabidopsis flowering-time genes have been found in other species, including rice, barley, wheat, Pharbitis and Brassica napus, demonstrating that parts of the genetic network controlling flowering in Arabidopsis are highly conserved, even in distantly related angiosperms (Takeba, 1966; Vince-Prue, 1985; Lin, 2000c; Mouradov et al., 2002) . This genetic analysis of Arabidopsis mutants and accessions led to the identification of four major flowering-time genetic pathways: Photoperiod; Vernalisation; Gibberellin and Autonomous (Mouradov et al., 2002).
Natural variation in flowering time
There is large variation in the timing of the floral transition between different accessions of Arabidopsis. For example, accessions collected from mountainous regions or higher latitudes are often winter annuals, flowering only after vernalisation stimuli, whereas ecotypes collected from milder climates do not require a prolonged exposure to winter conditions in order to flower (Simpson et al., 1999; Henderson et al., 2003).
Flowering Locus C
Crossing winter and summer annuals demonstrated that these differ at two major effect loci, called FLOWERING LOCUS C (FLC) and FRIGIDA (FRI) (Clarke and Dean, 1994; Clarke et al., 1995; Alonso-Blanco et al., 1998; Michaels and Amasino, 1999; Sheldon et al., 1999) . Winter annuals contain dominant alleles of both loci, which delay flowering until vernalisation (Simpson et al., 1999 ; see section 184.108.40.206).
Surprisingly crossing summer annual accessions also demonstrated the existence of natural-genetic variation between the accessions. For example, the Landsberg erecta (Ler) and Columbia (Col ) accessions are both early flowering and show similar flowering responses, but in F 2 populations created by intergressing these accessions late flowering plants segregate. This suggests that these two accessions contain different combinations of alleles that confer flowering time variation (Lister and Dean, 1993) . Similar results were obtained in crosses between the Cape Verde islands and Ler accessions (Alonso-Blanco et al., 1998).
Mutations affecting flowering time
Much of the early work to identify mutants affected in flowering time was based on screening for late-flowering mutants (Redei, 1962; Koornneef et al., 1991; Koornneef et al., 1998a) . In these populations the co, gigantea (gi) and luminidependens (ld) mutants were the first to be identified (Redei, 1962) . One of the reasons that these early screens identified only mutations that delay flowering was that the progenitor accessions (Ler; Col ; Wassilewskija (Ws)) were all early flowering under the screening conditions used. More recently screening was also carried out using later flowering accessions leading to the identification of early-flowering mutants (Michaels and Amasino, 1999) , also early-flowering accessions screened under non-inductive conditions yielded different early flowering mutants (Noh and Amasino, 2003).
The construction of double mutants carrying two mutations that delay flowering enabled the positions of genes into genetic pathways (Figure 1.1). If double mutants flowered at a similar time to single mutants, then the affected genes were placed in the same genetic pathway. Conversely if mutations showed an additive effect upon flowering time, so that a double mutant flowered later than either single mutant, then the mutations were proposed to affect different pathways (Koornneef et al., 1991; Coupland, 1995a).
The genetic analysis was supported by physiological data. So for example, mutations positioned in one pathway (later called the autonomous pathway) based on genetic criteria all confirmed a response to vernalisation, and those placed in a second pathway (later called the photoperiod pathway) did not confer a vernalisation response but were day length insensitive. This combination of physiological and genetic analysis identified the four main pathways responsible for flowering-time control in Arabidopsis.
Early flowering mutants identified genes required to repress flowering (Sung et al., 2003) . These mutants include early flowering 3 and 4 (elf3; elf4); terminal flower 1 (tfl1); early flowering in short days (efs); early in short days 4 (esd4); flc and mads affecting flowering1/flowering locus m (maf1/flm) (Zagotta et al., 1996; Liljegren et al., 1999; Michaels and Amasino, 1999; Sheldon et al., 1999; Soppe et al., 1999; Hicks et al., 2001; Ratcliffe et al., 2001; Scortecci et al., 2001; Doyle et al., 2002; Reeves et al., 2002) . Some of these genes were highly pleiotrophic and later proved to have general effects in chromatin remodelling, including terminal flower 2 (tfl2); embryonic flower 1 and 2 (emf1; emf2) and fertilisation-independent endosperm (fie) (Sung et al., 1992; Larsson et al., 1998; Ohad et al., 1999; Aubert et al., 2001; Yoshida et al., 2001).