An altered body plan is conferred on Arabidopsis plants carrying dominant alleles of two genes

One of the fundamental differences between plant and animal development is that animal body plans are established during embryogenesis, while plant embryos contain only a small fraction of the final body plan. In plants, the adult body plan is established during postembryonic development by the activity of two groups of proliferative cells: the root and the shoot apical meristems. The root apical meristem gives rise to the root system and the shoot apical meristems (SAMs) give rise to the shoot. The developmental pattern generated from the SAM is segmental, and therefore the architecture (body plan) can be described in terms of reiterative units referred to as metamers (White, 1984). In the case of the shoot, each metamer consists of an internode (stem segment) subtended by a node. The node is generally composed of a leaf and a branch emerging from the axil of the leaf. Metamers are produced sequentially by the SAM. Many variations of metameric structure exist, including variation of the internodal length, suppression of leaf or branch development, or transformation of leaves and branches into specialized structures. The morphology of sequential metamers may change over the course of the plant’s development through the influences of both endogenous developmental signals and environmental cues. Thus, plant development exhibits a higher degree of plasticity than animal development.

Arabidopsis thaliana is a model genetic organism for the study of meristem development and body plan determination. In Arabidopsis, three types of metamers have been described (Schultz and Haughn, 1991): Type 1 metamers consist of a very short internode subtended by the leaf and an axillary branch at the leaf axil. These metamers form first and are organized into a basal rosette of leaves. Type 2 metamers are similar to type 1, except for the elongated internodes between leaves. They develop after type 1 metamers and give rise to the basal, vegetative portion of the flowering stems. Type 3 metamers consist of long internodes subtended by individual flowers. They are the last type of metamer to form and thus constitute the distal portions of the flowering stems. The development of higher order branches in the axils of leaves is usually suppressed in Arabidopsis until the transition to flowering. Once initiated, these axillary branches generally produce only type 2 and type 3 metamers. A diagrammatic representation of the basic body pattern of the flowering Arabidopsis plant is provided in Fig. 1A.

There are several mutations that modify the Arabidopsis body plan. For example, the revoluta (rev) locus alters development of type 1 and 2 metamers. In rev plants, axillary meristems frequently fail to develop in the axils of rosette and cauline (aerial) leaves, leading to branchless plants (Talbert et al., 1995). Plants homozygous for the embryonic flower (emf) mutation lack type 1 metamers and produce only type 2 and 3 metamers on the main axis. Therefore, emf plants lack the basal rosette of leaves and germinate into leafand flower-bearing stems directly (Sung et al., 1992; Bai and Sung, 1995; Yang et al., 1995). In contrast to the emf plants that have suppressed development of the basal rosette, an increase in the number of type 1 metamers is observed in mutants and ecotypes of Arabidopsis that are delayed in the transition to flowering (Martinez-Zapater et al., 1994; Haughn et al., 1995). Interestingly, most of these late-flowering lines are affected only in the primary apical meristem; at the transition to flowering, the axillary meristems produce only type 2 and type 3 metamers just as they do in early-flowering lines. A few examples of lateflowering lines, including florens (F), that produce type 1 metamers on lateral branches have been reported (Koornneef et al., 1994). These lines are characterized as having aerial rosettes in the axils of cauline leaves. This phenotype was attributed to the extreme lateness of flowering in these lines. However, the genetic basis of the aerial-rosette phenotype was not investigated.

In an effort to identify other specific genes that influence the basic body plan of Arabidopsis, we have examined a variety of land races for alterations in body pattern. In this report, we describe a late-flowering ecotype of Arabidopsis, Sy-0, in which axillary meristems produce type 1, type 2 and type 3 metamers, resulting in the formation of aerial rosettes at the nodes of the flowering stem (Fig. 1B). We provide evidence that this phenotype is mediated by dominant alleles of two unlinked genes. A consequence of this altered developmental pattern is that vegetative development is prolonged past the time of the transition to flowering, resulting in an increase in the life span of the parent plant.

Ecotype Sy-0 was originally collected from the Isle of Skye, UK, and obtained from the Arabidopsis Information Service Stock Center, accession number 1204 (Frankfurt, Main, FRG). Seeds for Landsberg erecta (Ler) were obtained from A. R. Kranz, while seeds for the lateflowering lines co-2 and fca-1 were obtained from the Arabidopsis Biological Resource Center at Ohio State University, Ohio, USA. FRI-Sf2 and FLC-Sf2 alleles from Sf2 ecotype that were introgressed into Ler background (4 and 7 backrosses respectively, Lee et al., 1994) and were kindly provided by R. Amasino, were used for the genetic analysis. All plants were grown at 22°C, 65-85% relative humidity under fluorescent illumination supplemented with incandescent light (100-150 μE m⁻² sec⁻¹) on a 2:1 mixture of Jiffy Mix (Jiffy Products of America, Batavia, IL) and perlite, with continuous soil moisture by 10% Hoagland solution (Hoagland and Arnon, 1938). Long-day grown plants were provided with the continuous illumination, while short-day grown plants were kept 8 hours in light and 16 hours in dark.

Aerial rosettes were induced in Ler plants by growing seedlings under continuous light for 6 days, after which they were transferred to short-day conditions.

For vernalization experiments, seeds were surface-sterilized with 30% (v/v) bleach and 0. 5% (v/v) Triton X-100 and sown on 0. 8% (w/v) agar-solidified MS medium (Murashige and Skoog, 1962). Seedlings were grown at 4°C under h∼5 μE m⁻² sec⁻¹ of fluorescent light for 30 days, after which they were transferred to a soil mixture and grown as described.

Time to flowering was measured as days from germination until flower buds became visible in the middle of the vegetative rosette.

Specimens for light microscopy were fixed in 5% (w/v) glutaraldehyde in 0.05 M potassium phosphate buffer at room temperature. Fixed material was dehydrated in a graded ethanol series and embedded in LR White resin. 2 μm sections were cut on a Porter Blum MT-2 ultramicrotome, affixed to glass microscope slides and stained with toluidine blue. Sections were photographed under bright-field illumination using a Zeiss Axioskope.

DNA was prepared from pools of F₃ plants derived from individual F₂ plants, grown in a liquid culture. DNA extraction and gel blot analysis were performed as described (Lee et al., 1993). Microsatellite DNA was amplified according to the protocol from Bell and Ecker (1994).

When grown under continuous light, the Landsberg erecta (Ler) strain of Arabidopsis develops a vegetative rosette of seven to eight basal leaves. Approximately 2 weeks after germination, the SAM converts to reproductive development, giving rise to a highly branched stem system (Fig. 2A). The primary flowering stem bears two to three cauline leaves at the lower nodes (type 2 metamers) and numerous flowers in the distal region (type 3 metamers). In the axils of both cauline and rosette leaves, axillary branches develop (Fig. 2B). These branches reiterate the developmental pattern of the main flowering stem by producing a few cauline leaves and many flowers. In the axils of these cauline leaves, higher-order branches form that reiterate the same developmental pattern.

In the Sy-0 ecotype of Arabidopsis, this growth pattern is altered in two ways: (1) the number of leaves within the basal vegetative rosette is increased, and (2) rosettes of leaves develop at the nodes of the primary stem (Fig. 2C). The time to flowering in Sy-0 plants occurs approximately two months after seed germination, resulting in formation of a basal rosette of 70–80 leaves (Fig. 2D). After the transition to reproductive development, an elongating stem bearing 10–12 cauline leaves becomes visible. In the axils of cauline leaves, additional leaves appear (Fig. 2E). The internodes between these leaves do not elongate, resulting in the development of type 1 metamers. As a consequence, a rosette of up to 18 leaves develops in the axils of cauline leaves, similar in structure to the original basal rosette. These rosettes develop at the axils of all leaves, but they are most easily seen in the aerial portion of the plant. Consequently, we will refer to them as aerial rosettes. A close up of an aerial rosette borne at a flowering-stem node of an Sy-0 plant is shown in Fig. 2F. Later in development, metamers of type 2 and 3 form sequentially, giving rise to an elongated axis with cauline leaves and flowers (Fig. 2G). Thus, branches contain only type 2 and 3 metamers in Ler (Fig. 2B), while development of the type 2 and 3 metamers is preceded by the formation of the type 1 metamers in Sy-0.

Visual analysis of development of the aerial rosettes on Sy-0 plants indicates that shoot meristem development has been altered in some way relative to that of most ecotypes. To determine where alterations have occurred, we compared morphological changes at the shoot apex in Ler and Sy-0 plants (Fig. 3).

During the vegetative phase, the shoot apical meristem in Ler initiates leaf primordia on its flanks. At this stage, there is no sign of axillary meristem development in the axils of leaf primordia (Fig. 3A). Upon transition from the vegetative to reproductive phase, the primary shoot apical meristem converts from a vegetative to an inflorescence meristem and begins to initiate flower meristems (Fig. 3B). Flower meristems were identified based on their more rounded appearance relative to the ellipsoid shape of leaf primordia. Concomitant with the transition of the primary shoot apical meristem to the reproductive phase, dense, meristematic cells that represent axillary meristems become visible in the axils of the youngest leaf primordia (Fig. 3B). The primary inflorescence then emerges from the basal vegetative rosette as a consequence of increased internode elongation between several of the youngest developing leaves and between developing flowers (Fig. 3C). The axillary meristems, developing in the axils of the cauline leaf primordia, initiate a few leaf primordia and then begin producing flowers (Fig. 3D). Later in development, the internodes between these leaves elongate resulting in a branch with cauline leaves.

There are at least two possible ways that the development of the axillary meristems could be altered from the pattern described above to give rise to the aerial rosettes in Sy-0 plants: (1) the axillary meristems could initiate prematurely and produce leaf primordia during the vegetative stage of the plant’s development; or (2) axillary meristems could initiate at the transition to flowering as in the Ler strain, but remain longer in a vegetative phase before (Fig. 3F). Developing meristems in the axils of cauline-leaf primordia are shown in Fig. 3G. They initiate leaf primordia and produce a spiral of rosette leaves (Fig. 3H) before undergoing the transition to reproductive development. Thus, the aerial-rosette phenotype arises due to the initiation of many more leaf primordia by the axillary meristems when compared to most earlyand late-flowering strains of Arabidopsis. The Sy-0 developmental pattern represents a heterochronic shift in the transition of axillary meristems from vegetative to reproductive development.

To determine the genetic basis of Sy-0 phenotype, we crossed Sy-0 to Ler plants. The 23 F₁ plants tested were all late flowering and had aerial rosettes, indicating that the aerialrosette phenotype is dominant (Table 1). The F₂ progeny consisted of plants that had a wide range of flowering times. In general, plants with the latest flowering times also developed aerial rosettes. The number of leaves within the aerial rosettes varied among the aerial-rosette-bearing progeny. In moderately late-flowering plants, aerial rosettes contained 3 leaves, while in the latest-flowering plants aerial rosettes had 10-12 leaves. The number of leaves per node tended to be highest in basal nodes and lower in more apical vegetative nodes. The segregation ratio between aerial-rosette-bearing plants (ar) and aerial-rosette-lacking plants (nar) in this population was approximately 9:7. These data fit best with the model that an interaction between dominant alleles of two unlinked genes is needed for the aerial-rosette phenotype. In addition, a ratio of approximately 1:3 (ar:nar) was observed among the progeny of a test-cross (BC₁) between F₁ plants and Ler. This is consistent with the expected ratio for two independently assorting genes. As in the case of the F₂ population, only plants with the later flowering times developed aerial rosettes (Fig. 4). In both segregating populations, plants lacking aerial rosettes had a wide variation of flowering times, but in general they flowered earlier then plants that developed aerial rosettes. Among the plants lacking aerial rosettes were some that flowered later than the Ler parent (Fig. 4), suggesting that one or both genes implicated in aerial-rosette formation also cause a late-flowering phenotype when present alone.

Segregation ratios between aerial-rosette-bearing and -lacking plants from F₂ and BC₁ populations were consistent with the possibility that dominant alleles of two unlinked genes are necessary for aerial-rosette formation. If the two dominant gene hypothesis is valid, then it should be possible to reconstitute a double-dominant phenotype by crossing plants that carry dominant alleles of single genes and lack aerial rosettes. A complementation experiment was performed by setting up diallel crosses between eight non-aerial-rosette-bearing plants from the BC₁ population (Fig. 5). These plants differed in their flowering time, ranging from early flowering (two weeks after germination) to late flowering (one month post germination). Diallel analysis revealed several crosses whose progeny segregated for the aerial-rosette phenotype, supporting the possibility that dominant alleles of two unlinked genes are responsible for the aerial-rosette phenotype.

The chromosomal locations of genes necessary for aerialrosette development was determined by using aerial-rosettebearing plants from the BC₁ population for the mapping analysis. RFLP analysis detected linkage between the aerialrosette phenotype and RFLPs on the tops of chromosomes 4 and 5, close to the RFLP g6844 on chromosome 4 (gene B) and to the m291 RFLP marker on chromosome 5 (gene A) (Fig. 6). This result provides additional evidence that dominant alleles of two unlinked loci are required for the aerial-rosette phenotype.

A cross between lines 4 and 8 from the diallel matrix (see Fig. 5), produced F₁ progeny that segregated for the aerialrosette phenotype. Lines 4 and 8 were moderately late flowering themselves, but did not display the aerial-rosette phenotype. When lines 4 and 8 were allowed to self fertilize, their progeny segregated for late flowering in a semidominant fashion. For example, AA plants produced 68±4 rosette leaves, Aa plants produced 26±2 rosette leaves, while aa plants produced 7±1 rosette leaves before flowering. This raises the possibility that each complementary dominant allele implicated in aerial-rosette development delays flowering time when present alone. Additionally, aerialrosette-bearing plants were always the latest flowering in segregating populations. We have never observed an early flowering plant that developed aerial rosettes, supporting the hypothesis that genes conferring the late-flowering and the aerial-rosette phenotype in Sy-0 are associated. To test this hypothesis, we mapped the late-flowering genes in lines 4 and 8 relative to the RFLP markers that are linked to the genes involved in aerial-rosette formation. Both molecular markers, ld on chromosome 4 and m291 on chromosome 5, showed linkage to the genes causing delayed flowering in lines 4 and 8 (0/16 recombinants in both cases), suggesting that they cosegregate with genes involved in the aerial-rosette formation. Since the late-flowering and aerial-rosette phenotypes both involve delayed transition of meristems to reproductive development, we suggest that the same two genes are responsible for both phenotypes. We have tentatively assigned gene names to these two loci: gene A will be referred to as ART (aerial rosette) while gene B will be referred to as EAR (enhancer of aerial rosette).

This analysis also revealed that line 4 carried a dominant allele of EAR locus on chromosome 4 and that line 8 carried a dominant allele of ART on chromosome 5. Subsequently, lines homozygous for dominant alleles of ART and EAR were isolated. They are both late flowering (Table 2). Line ART/ART forms a basal rosette consisting of ∼65 leaves, while line EAR/EAR has ∼25 leaves (Table 2). Neither of these lines form aerial rosettes, but when crossed together they yield a uniform late-flowering, aerial-rosette-bearing F₁ progeny (Table 3).

The late-flowering phenotype of lines ART/ART and EAR/EAR raised the possibility that the ART and EAR genes could be alleles of previously identified late-flowering genes that map in their vicinity. The map location of EAR was indistinguishable from the location of the late-flowering FRI locus (Lee et al., 1993). To determine whether EAR can be replaced with the dominant FRI allele from Sf2 ecotype, we crossed line ART/ART with a line homozygous for a dominant allele of FRI gene. F₁ progeny of these two lines were late flowering and developed aerial rosettes (Table 3), suggesting that FRI can replace the dominant allele of gene EAR in causing the Sy-0 phenotype.

The interaction between ART and FRI is not the only one that yields the aerial-rosette phenotype. Another lateflowering gene, which maps ∼25 cM from the FRI locus, fca (Koornneef et al., 1991), interacts with the ART locus in a similar way. F₂ progeny of a cross between line ART/ART and a line homozygous for the fca allele segregated for aerialrosette-bearing plants in ∼ 3/16 ratio (Table 3), indicating that the recessive allele of the fca locus has the same effect as the dominant allele of the EAR gene. On the contrary, fca in combination with EAR did not segregate for the aerial rosettes. Since FCA does not map near either ART or EAR, it is apparent that ART is primarily responsible for the aerialrosette phenotype and that the effect is enhanced by other late-flowering loci.

We have demonstrated that the ability of Sy-0 plants to form aerial rosettes depends on the dominant alleles of two unlinked genes and that under our experimental conditions (continuous light) Ler plants were not able to form aerial rosettes. However, Ler plants can develop aerial rosettes under certain environmental conditions. If Ler plants are grown under long days until the initial transition to flowering occurs and then are shifted to short days, axillary meristems will develop aerial rosettes at the axils of cauline leaves (Laibach and Kribben, 1953). This demonstrates that by manipulating environmental conditions (day length) one can phenocopy Sy-0 in the Ler background.

Environmental conditions may also suppress the expression of the aerial-rosette phenotype in the Sy-0 plants. When Sy-0 seedlings are vernalized (kept at 4°C for 30 days) and then transferred to 23°C, they convert from vegetative to reproductive development after initiating only 25 leaves, and do not produce aerial rosettes at the axils of cauline leaves (Table 2).

Thus, vernalized Sy-0 plants phenotypically resemble Ler plants by producing secondary inflorescences containing only type 2 and 3 metamers.

The body plan of the flowering Arabidopsis plant can be described in terms of the induction, identity and fate of shoot apical meristems during the course of the plant’s development. In both Sy-0 and Ler, the primary shoot apical meristem, formed during embryogenesis, gives rise initially to vegetative metamers. At this stage we found no anatomical evidence for the existence of axillary meristems in the axils of developing leaves of either the Ler or the Sy-0 strain. The transition to reproductive development involves a change in the primary shoot apical meristem from a leaf-producing meristem to a flower-producing meristem. Accompanying this identity change in the primary apical meristem is a basipetal wave of axillary meristem induction in the axils of existing leaf primordia and elongation of internodes between several of the youngest leaf primordia. As pointed out by Hempel and Feldman (1994), it is the imposition of these post-transition events on the youngest preexisting leaf primordia that gives rise to the type 2 metamer. Ultimately, the complex branching pattern characteristic of the mature flowering plant is a consequence of developmental processes carried out by the axillary meristems – each undergoing a vegetative phase of development before switching into flower-producing meristems. It is the timing of these identity switches at the axillary meristems that produces the observed difference in body plan between Sy0 and most other strains of Arabidopsis.

From our anatomical observations, we suggest that both Ler and Sy-0 conform to two basic rules that govern axillary meristem behavior in Arabidopsis.

1. Axillary meristems are initiated only after the primary meristem undergoes the transition to flowering.

2. Axillary meristems undergo an obligatory vegetative phase before converting to reproductive development.

While these rules apply to most strains of Arabidopsis, there are some notable exceptions. In Sy-0 and other late-flowering varieties, we have observed that axillary meristems may be initiated in the axils of the oldest mature leaves even before the transition to flowering (unpublished observation). Thus, rule number 1 may only apply to nodes that are in the proximity of the primary apex. Exceptions to rule number 2 involve cases where the axillary meristem converts directly to a floral meristem rather than to the normal inflorescence meristem. Terminal flower (tfl1) mutants produce solitary flowers in the axils of primary leaves (Shannon and Meeks-Wagner, 1991; Alvarez et al., 1992). In addition, transgenic plants that ectopically express either the leafy (LFY) gene or the apetala1 (AP1) gene also produce axillary flowers (Weigel and Nilsson, 1995; Mandel and Yanofsky, 1995). It should be noted that the primary shoot apical meristem may obey rule number 2 more stringently than the axillary meristems do. In the cases of the embryonic flowering mutant (emf) (Sung et al., 1992; Bai and Sung, 1995; Yang et al., 1995), as well as the tfl1 mutant and the LFY/AP1 overexpressors, the primary shoot apical meristem always produces a few vegetative nodes before undergoing the transition to reproductive development.

While both Sy-0 and Ler conform to the rules stated above, they differ in the developmental timing of the transition of meristems from vegetative to reproductive development. Ler is an example of an early flowering strain of Arabidopsis while Sy-0 can be classified as a late-flowering strain. These classifications are based on the number of rosette leaves produced by the primary shoot apical meristem before it undergoes the transition to reproductive development when the plant is grown under a long-day photoperiod. Mutational analysis of the Ler strain of Arabidopsis has revealed 15 different loci that are required for early flowering in long days (Koornneef et al., 1991). In addition, dominant alleles of the FRI locus confer the late-flowering phenotype in some natural ecotypes (Lee et al., 1993; Clarke and Dean, 1994). In general, these late-flowering lines are only affected in the transition of the primary shoot apical meristem to reproductive development. In these lines, axillary meristems convert from vegetative to reproductive development after the production of only two to three vegetative nodes and, as a consequence, produce only type 2 and type 3 metamers. In contrast, Sy-0 axillary meristems undergo a prolonged vegetative phase, resulting in the aerial-rosette phenotype. In order to understand the developmental basis for this heterochronic difference between Sy-0 and the other early and late-flowering strains of Arabidopsis, the nature of floral evocation in plants must be considered.

Most models for the control of flowering time assume the existence of chemical signals that are produced in vegetative tissues but act at the meristems to either inhibit or induce the transition to reproductive development (Martinez-Zapater et al., 1994; Haughn et al., 1995). In theory, the timing of the transition to reproductive development for an individual meristem will depend on the levels of inductive and inhibitory signals impinging on that meristem and on the competence of the meristem to respond to these signals. Competence to respond (sensitivity) to flowering signals may be particularly important for the developing axillary meristems, which undergo an obligatory vegetative phase before switching to reproductive development – even in a context where the primary apical meristem has already undergone the transition to flowering.

If it is assumed that sufficient floral evocation signal is present at the apex once the primary meristem has made the switch to reproductive development, then it is reasonable to suggest that the developing axillary meristems are initially insensitive to this signal, and so produce vegetative structures. For most early and late strains of Arabidopsis, this period of insensitivity is short-lived; whereas in Sy-0, the phase of insensitivity is prolonged. This model for explaining the aerialrosette phenotype is illustrated in Fig. 7. We propose that the interactions of the two genes identified in this study produces the prolonged insensitivity to floral evocation signals by the developing axillary meristems in Sy-0. While we can not rule out the alternative possibility that the phenotype is due to a deficiency in the concentration of flowering signal at the axillary meristems (as proposed by Koornneef et al., 1994), it seems unlikely that a level of a global flowering signal sufficient to trigger reproductive development in the primary apical meristem would not also be present at the axillary meristems only a few cells away. On the contrary, the aerial-rosette phenotype can be produced in Ler by shifting plants from long day to short day just after the transition to flowering. In this case, the prolonged vegetative phase for axillary meristems could be due to a change in the floral signal caused by the shift to short-day photoperiod rather than a change in sensitivity to the signal by the meristem.

Lines homozygous for either the ART or the EAR dominant allele are late flowering. Both genes map to the regions where other known late-flowering genes map, raising the possibility that they are alleles of already identified loci. ART locus maps on chromosome 5 in the vicinity of two other semidominant late-flowering genes: FLC and CO. In addition to the semidominant late-flowering loci, there are two other candidate loci that map near ART: EMF and KNAT1. A recessive allele of the EMF gene suppresses vegetative development of the primary SAM to the initiation of only 1-3 leaf primordia before flowering (Sung et al., 1992; Bai and Sung, 1995; Yang et al., 1995). A dominant allele of ART locus from Sy-0 confers exactly the opposite phenotype: it prolongs the vegetative phase of SAMs. Since loss-of-function and gain-of-function alleles of regulatory genes often have the opposite phenotypes, it is possible that ART represents a dominant gain-of-function EMF allele. Alternatively, ART could be a dominant allele of the KNAT1 locus. Even though mutant alleles of KNAT1 are not available, the expression pattern of KNAT1 wild-type allele revealed its localization within the vegetative meristem (Lincoln et al., 1994). If the KNAT1 gene is a regulator of vegetative development, ART from Sy-0 could be a dominant allele of it. Finally, ART could be a new late-flowering gene defining a novel locus on chromosome 5.

EAR, on the contrary, maps to chromosome 4, in a region near the FRI locus (Lee et al., 1993). We have obtained two pieces of circumstantial evidence that EAR is an allele of the FRI locus: EAR and FRI are both semidominant late-flowering genes, and the FRI allele from Sf2 ecotype can reconstitute the Sy-0 phenotype when combined with ART from chromosome 5 (Table 3).

Current models of the genetic pathways that control flowering time are subdivided into pathways that are influenced by day length and vernalization (Martinez-Zapater et al., 1994; Haughn et al., 1995). Our results indicate that the aerial-rosette phenotype is reversed by vernalization. Since neither FRI nor fca disrupt vernalization and ‘day-length’ sensing pathways (Lee et al., 1993; Koornneef et al., 1991), it is likely that the ART locus is vernalization sensitive. Recombination of the ART allele with other late-flowering loci will help to resolve whether the ART locus can be fit into existing schemes for the control of flowering time.

V. G. thanks Dr Michael Freeling and Dr Jerry Kermicle for inspiring her interest in plant development and genetics. We thank Claudia Lipke and Kandis Elliot for the photographic and art work. We thank Dr Ray Evert and members of his laboratory for helping V. G. with the histology. We thank Patric Masson, Rick Amasino, Donna Fernandez and members of Kathy Barton’s laboratory for critical reviews of this manuscript. The work presented here was funded by grants from the Department of Energy/National Science Foundation/US Department of Agriculture Collaborative Research in Plant Biology Program (no. BIR92-20331).