Inflorescence architecture: A developmental genetics approach

THE VOL. 65

BOTANICAL

REVIEW

OCTOBER--DECEMBER 1999

NO. 4

Inflorescence Architecture: A Developmental Genetics Approach* SUSAN SINGER, l JOHN SOLLINGER, 1 SONJA M A K I , 2 JASON FISHBACH, l B R A D SHORT, 1 CATHERINE REINKE, 1 JENNIFER FICK, 1 L A U R A COX, l A N D R E W M C C A L L , ~ AND HEIDI M U L L E N ~

lDepartment of Biology Carleton College Northfield, MN55057, U.S.A. 2Department of Horticulture Clemson University Clemson, SC 29634, U.S.A.

I. II. III. IV.

Abstract . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Overview of Pea Inflorescence Development . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Methods . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . A. Plant Material . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . B. Growth Conditions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . C. Plant Age . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . D. Scanning Electron Microscopy . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . E. Classification of Stems (Branches) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . V. Results and Discussion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . A. The Regulation of First-Order Axis Development by DETand VEG1 . . . . . . . . B. The Regulation of First-Order Axis Development by UNI . . . . . . . . . . . . . . . . . C. The Role of NEP in Second-Order Axis Development . . . . . . . . . . . . . . . . . . . . D. The Role of PIMin Third-Order Axis Development . . . . . . . . . . . . . . . . . . . . . . 1. The Effect ofpim Mutation on Inflorescence Architecture . . . . . . . . . . . . . . . 2. broccoli Enhances the pim Phenotype . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

386 386 388 388 388 390 390 390 391 392 392 392 395 398 398 403

* Gene symbols used in this article: For clarity a common symbolization is used for genes of all species discussed in this article. Genes are symbolized with italicized capital letters. Mutant alleles are represented by lowercase, italicized letters. In both cases, the number immediately following the gene symbol differentiates among genes with the same symbol. If there are multiple alleles, a hyphen followed by a number is used to distinguish alleles. Protein products are represented by capital letters without italics.

Copies of this issue [65(4)] m a y be purchased from the NYBG Press, The New York Botanical Garden, Bronx, NY 10458-5125, U.S.A. Please inquire as to prices.

TheBotanical Review65(4): 385-410, October-December 1999 9 1999 The New York Botanical Garden 385

386

THE BOTANICAL REVIEW

E. The Role of UNlin Third-Order Axis Development . . . . . . . . . . . . . . . . . . . . . . VI. Summary . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . VII. Acknowledgments . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . VIII. Literature Cited . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

405 407 407 408

I. Abstract We are characterizing a suite ofPisum sativum mutants that alter inflorescence architecture to construct a model for the genetic regulation of inflorescence development in a plant with a compound raceme. Such a model, when compared with those created for Antirrhinum majus and Arabidopsis thaliana, both of which have simple racemes, should provide insight into the evolution of the development of inflorescence architecture. The highly conserved nature of cloned genes that regulate reproductive development in plants and the morphological similarities among our mutants and those identified in A. majus and A. thaliana enhance the probability that a developmental genetics approach will be fruitful. Here we describe six P. sativum mutants that affect morphologically and architecturally distinct aspects of the inflorescence, and we analyze interactions among these genes. Both vegetative and inflorescence growth of the primary axis is affected by UNIFOLIATA, which is necessary for the function of DETERMINATE (DET). DET maintains indeterminacy in the first-order axis. In its absence, the meristem differentiates as a stub covered with epidermal hairs. DETinteracts with VEGETA TIVE1 (VEG1). VEG1 appears essential for second-order inflorescence (I2) development. vegl mutants fail to flower or differentiate the I2 meristem into a rudimentary stub. det vegl double mutants produce true terminal flowers with no stubs, indicating that two genes must be eliminated for terminal flower formation in P. sativum, whereas elimination of a single gene accomplishes this in A. thaliana and A. majus. NEPTUNE also affects 12development by limiting to two the number of flowers produced prior to stub formation. Its role is independent of DET, as indicated by the additive nature of the double mutant det nep. UNI, BROC, and PIM all play roles in assigning floral meristem identity to the third-order branch, pim mutants continue to produce inflorescence branches, resulting in a highly complex architecture and aberrant flowers, uni mutants initiate a whorl of sepals, but floral organogenesis is aberrant beyond that developmental point, and the double mutant uni pim lacks identifiable floral organs. A wild-type phenotype is observed in broc plants, but broc enhances thepim phenotype in the double mutant, producing inflorescences that resemble broccoli. Collectively these genes ensure that only the third-order meristem, not higher- or lower-order meristems, generates floral organs, thus precisely regulating the overall architecture of the plant.

II. Introduction Investigations of the diversity in inflorescence architecture among angiosperms have led to extensive typologies focused on final structure, with more limited emphasis on the development of the inflorescence (Weberling, 1989; Coen & Nugent, 1994; Grimes, 1996; Souer et al., 1998). Developmental studies can contribute substantially to our understanding of evolution (Diggle, 1992; Wray, 1994). For example, Tucker (1997) has devoted her career to comparative floral developmental studies in the Fabaceae (Leguminosae), leading to clear distinctions between parallel and convergent evolution of floral traits. Cladistic analysis in the Fabaceae has been enhanced by the complementation of ontogeny and molecular sequence analysis (Chappill, 1995; Doyle, 1997). Our approach to understanding inflorescence architecture is to analyze developmental mutants. The goal is to understand the genetic regu-

GENETICS OF INFLORESCENCE DEVELOPMENT

387

lation of inflorescence development, beginning with the identification of key regulatory genes. Ultimately a comparative analysis of these genes among species will provide a complementary perspective on the evolution of inflorescence patterns. The relationship between single gene mutations and evolutionary events is a continuing source of debate. Hilu (1983) surveyed the literature on mutations affecting development in flowering plants and concluded that the predominance of single-gene mutations may affect macroevolution, leading to new taxa. The cloning of the CYCLOIDEA (CYC) gene in Antirrhinum majus (Asteridae) has focused the discussion on the evolution of zygomorphic flowers (Luo et al., 1996). CYC is required for normal development of zygomorphic flowers. Donoghue et al. (1998) conducted an expanded phylogenetic analysis of the Asteridae and concluded that zygomorphic flowers arose at least eight times and that there were at least nine reversals to actinomorphy. Not all reversals were loss-of-function mutations. They suggest that several developmental pathways with different genes (not necessarily CYC loss of function) will be found to account for derived actinomorphy. This illustrates the potential for a genetics approach to distinguish between parallel and convergent evolution, much as Tucker (1997) has done with developmental morphology. Doebley and Lukens (1998), however, emphasize that protein function is highly conserved and that transcriptional regulators play key roles in plant development. Plant form may evolve by alterations in cis-regulatory elements in genes or transcription regulators. They propose that the predominant mechanism in the evolution of form is change in the cis-regulatory elements of transcription regulator genes. Temporal and spatial changes in gene expression can result in substantial morphological change. Genes affecting reproductive development in Arabidopsis thaliana and A, majus have been shown to be highly conserved among the angiosperms (Frohlich & Meyerowitz, 1997). Floral meristem identity genes affect inflorescence architecture by initiating floral development. The MADS-box floral genes (transcription regulators), which include the ,4. thaliana floral meristem identity gene,4PET,4L,41 (,4P1) and its homolog, SQUAMOS`4 (SQU,4), in,4. majus, first began to diverge from one another before the origin of seed plants (Mandel et al., 1992; Munster et al., 1997; Purugganan, 1997). LE,4FY(LFY), an ,4. thaliana floral meristem identity gene, is another highly conserved regulatory gene with homologs in the basal angiosperms and Gnetales (Weigel et al., 1992; Frohlich & Meyerowitz, 1997). The LFF/FLO homologs are essential for normal development in plants with cymose inflorescences (the sympodial growth of the Solonaceae) and racemose inflorescences (simple and compound). LFYand its homolog in ,4. majus, FLORIC,4UL,4 (FLO), are expressed primarily, though not exclusively, in floral meristems, while homologs in Pisum (UN1), Nicotiana (NFL), Impatiens, and Petunia (,4LF) are expressed in vegetative meristems as well (Kelly et al., 1995; Hofer et al., 1997; Poteau et al., 1997; Souer et al., 1998). A vegetative phenotype, however, is found only in P. sativum uni mutants, in which compound leaves become simple (Hofer et al., 1997). Our work is aimed at understanding how highly conserved genes are used to create diverse inflorescence architectural pattems. We are elucidating and comparing the genetic regulation of inflorescence development in Pisum sativum (garden pea) with ,4. thaliana and ,4. majus, two model systems for molecular genetic analysis of flowering (reviewed by Weigel, 1995; Yanofsky, 1995; Amasino, 1996). P. sativum was selected for several reasons. As a result of the long history of the pea as both a model genetic system and an agriculturally important crop, numerous mutants alter reproductive development. The inflorescence of the pea is one step more complex than the simple racemes of,4. thaliana and,4, majus. Thus comparative analysis may provide insight into an underlying genetic mechanism that distinguishes simple and compound racemes. Our long-range plan to extend our work to other family members will be facilitated by compara-

388

THE BOTANICAL REVIEW

tive data on more than 200 taxa within the Fabaceae, which is broader than any other group (Endress, 1994; Tucker & Douglas, 1994). Although P. sativum is a highly derived and domesticated legume, it shares basic inflorescence architectural features with its wild relatives (P. abyssinicum, P. humile, and P. elatius) and with sweet peas (Lathyrus). The indeterminacy of the inflorescence is most likely the basal state in the Fabaceae (Tucker, 1998). Also, some of our mutations appear to "unmask" ancestral traits found in more basal Fabaceae. I I I . Overview of Pea Inflorescence Development The pea is a quantitative, long-day plant with flowering time regulated by both a grafttransmissible floral promoter and inhibitor with meristem sensitivity to the signal(s) controlled by the geneLF(LATEFLOWERING) (Reid et al., 1996; Weller et al., 1997). The transition from vegetative to inflorescence development parallels an increasing complexity in leaf structure. Commitment to inflorescence development precedes the initiation of nodes that will contain second-order inflorescence meristems (I2 meristems) (Ferguson et al., 1991). The firstorder inflorescence (Ii) is comparable to the preceding vegetative (V1) axis, except that I2 rather than V2 meristems are initiated in each leaf axil (Figs. 1-2). Multiple axillary meristems may be initiated at a node, but in wild-type peas only one develops into a shoot. An I~ meristem is indeterminate. Determination for inflorescence development found in cultured meristems provides evidence that the meristem that initiates the I1 is in a developmentally distinct state from the V1 even though flowers are initiated only on the third-order axis (Ferguson et al., 1991). An I2 meristem initiates one or more floral meristems before terminating in a stub covered with epidermal hairs (Hole & Hardwick, 1976; Singer et al., 1990). The result is a compound raceme. In this article we use a developmental genetics approach to construct a model for regulation of inflorescence architecture in pea (see Table I for a summary of the P. sativum mutants reported here). Descriptive and experimental analysis of the following mutants provides the basis for this model, vegetative1 (vegl) fails to make floral structures or I2 stubs (Reid & Muffet, 1984). determinate (det) is a homeotic mutation that converts an I1 meristem to an I2 meristem, so that the primary axis of the plant terminates as a stub (Singer et al., 1990; Ferguson et al., 1991). Unlike terminal flower (tfl) in A. thaliana and centroradialus (cen) in A. majus, suppression of indeterminacy in the main axis by det does not result in the production of a terminal flower (Shannon & Meeks-Wagner, 1991; Alvarez et al., 1992; Bradley et al., 1996). neptune (nep) increases the number of flowers produced by an I2 meristem. PROLIFERA TING INFLORESCENCE MERISTEM (PIM), a floral meristem identity gene, is necessary for the third-order meristem of the inflorescence to develop into a flower. Thepim mutant phenotype is more severe in the broccoli (broc) background, much like the interaction reported between cauliflower (cal) and apl in A. thaliana (Kempin et al., 1995). UNI, the homolog of LFY/FLO, is another floral meristem identity gene, but it also plays a role in leaf development (Hofer et al., 1997). Mutations in these genes alter the branching patterns and floral development of pea inflorescences.

IV. Methods A. PLANTMATERIAL Seeds heterozygous for det were provided by the late G. A. Marx (New York State Agricultural Experiment Station, Geneva, NY). Heterozygotes were continuously selected and

GENETICS OF INFLORESCENCE DEVELOPMENT

389

Fig. 1. Wild-type pea inflorescence architecture. Growth is indeterminate on the first-order I~ axis. The I2 branch produces floral meristems before terminating as a stub (s).

Fig. 2. Schematic of the wild-type plant in Figure 1. Each arrow represents a shoot apical meristem; the number in the arrow indicates branch order, with one being the primary axis. 12 meristems terminate as a stub covered with epidermal hairs, whereas I~ meristems are indeterminate. V2 meristems can reiterate the overall pattern o f a V~ meristem. V = vegetative meristem; I = inflorescence meristem; F = floral meristem.

390

THE BOTANICAL REVIEW

Table I Pisum sativum inflorescence architecture genes

Gene

Mutant phenotype

Reference

No flowers, no 12stub Converts It meristem to I2 meristem terminating as a stub Multiple pods per inflorescence branch in both nep-1 and nep-2 alleles PIM Floral meristem develops as anflorescence meristem BROC Enhancespim phenotype; broc alone is wild type UNI Reduces compound leaves to simple leaves; flowers contain mainly leaflike sepals and carpels

VEG1 DET NEP

Reid & Murfet, 1984 Singer et al., 1990 Singer, unpubl. Reid et al., 1996 Singer, unpubl. Hofer et al., 1997; Marx, 1987

selfed for nine generations prior to the selection of the mutant and wild-type lines used in this study. Marx also provided st (line B88-181, a mutation causing reduced stipules), multipodded det, and nep seed. pim was a spontaneous mutant isolated in our greenhouse out of an early-flowering line (line C76-505) from Marx. The mutant was outcrossed into a number of different lines to assess genetic background effects on phenotype and was allowed to selfpollinate for at least seven generations to near isogeneity. To determine genetic interactions with other developmental genes, double mutants were constructed using pim as the pollen parent. Seeds heterozygous for vegl were provided by I. C. Murfet (University of Tasmania, Hobart, Australia). uni (line JI 2171 T) was procured from the John Innes Centre (Norwich, England) as a heterozygote. The uni allele, tac, which alters leaf development but not floral development (Marx, 1987), was not used in these studies. B. GROWTH CONDITIONS Unless otherwise noted, seeds were sown in a soilless potting mix (Prime-Gro 7, ThermO-Rock East, New Eagle, PA) in 10 cm pots in growth rooms. Long-day conditions (18 hours light/6 hours dark) were maintained with high-pressure sodium vapor and metal halide lights (1:1 ratio) for 8 hours (600-700 ~.mol photons/m2/second I of photosynthetically active radiation) and low-level incandescent bulb irradiation for 10 hours. Day/night temperatures were 21.5/18.5 C. Plants were fed weekly with 200 ppm of Plantex 20:10:20 fertilizer (Plantex Corp., Brampton, Ontario, Canada). C. PLANT AGE Plant age was determined by counting the number of nodes acropetally from the first node above the cotyledons to the last unfolded sepal pair. The first node to contain an I2 meristem is termed the NFI (node of floral initiation). Typically all nodes initiated after the NFI also have developed I2 branches. V2 buds are capable of initiating inflorescence axillaries (I3 branches in this case), but only after initiating vegetative nodes. D. SCANNING ELECTRON MICROSCOPY Shoot apices were dissected under a stereoscope and fixed in 2% gluteraldehyde in Millonig's buffer (1961), dehydrated in an acetone series, and critical-point dried with CO2 in a

GENETICS OF INFLORESCENCE DEVELOPMENT

391

Fig. 3. Schematic ofa det plant. Labels are as in Figure 2. In this mutant, the 11meristem becomes an I2 meristem that directly initiates floral meristems and terminates as a stub.

Balzers Union CPD020 apparatus. Samples were mounted on stubs and coated with gold in a Hummer Jr. Sputter coater before viewing in a JEOL JSM-350 SEM at 15 kV. Photomicrographs were taken on Polaroid Type 55 film, or images were digitized with slow-scan acquire system from JEOL. E. CLASSIFICATIONOF STEMS (BRANCHES) In wild-type plants, the I1 stem is indeterminate and has compound leaves with axillary meristems that produce 12 inflorescences. I2 meristems have suppressed leaf development, produce floral (F) meristems in their axils, and terminate as rudimentary stubs. Early in floral development, third-order floral meristems initiate zygomorphic flowers with an outer whorl of five sepals, a second common whorl of five petals and five stamens, a third whorl of five stamens, and a carpel in the fourth whorl (Tucker, 1989). Some of our mutations give rise to second-, third- and fourth-order branches that do not fit the description o f l b 12, or F branches found in the wild type. The classification of mutant stem complexes was based on the presence of floral, infloral, and vegetative traits inclusive to a given axis. Floral traits scored included the presence of floral organs (stamens, carpels, petals, and sepals), whorled phyllotaxy, intemode compression, lack of axillary structures, and determinate growth. Infloral traits included nodes with modified or suppressed leaves and termination as a stub. Vegetative structures included complex leaves (or parts thereof), distichous phyllotaxy, internode elongation, and apical meristem indeterminacy, as did first-order inflorescence stems. When the stipules at the seventh node above the NFI unfolded, the apex

392

THE BOTANICAL REVIEW

of the third- and fourth-order branches were categorized according to their developmental fate: termination as a stub; termination as a "flower" (determinate or indeterminate); or indeterminate vegetative stem.

V. Results and Discussion A. THE REGULATION OF FIRST-ORDER AXIS DEVELOPMENT BY DET AND VEG1 Analysis of det plants is consistent with the interpretation that DET is necessary to maintain indeterminacy in the I1. In the absence of DET, the terminal (first-order) meristem initiates two or three I2 meristems before forming a rudimentary stub (Fig. 3; Singer et al., 1990). The terminal meristem of wild-type plants ceases growth as flowers and fruits develop, but it maintains its meristematic state. This is in contrast to the det terminal meristem (Figs. 4-6). Removal of flowers and fruits results in resumed meristematic growth in wild-type plants but not in det plants (Maki & Singer, unpubl.). det does not affect the vegetative meristem, which is consistent with the interpretation that I1 and V1 meristems are developmentally distinct. Murfet (1989) has suggested that a minimum number of vegetative nodes (possibly 10) is required before det apices terminate, even in very early flowering lines. This is consistent with the interpretation that the transition from a V1 to an I1 meristem represents a phase change. Given that det has no effect on the vegetative phenotype, it should affect the vegl mutant phenotype only ifvegl, which never initiates floral organs (Figs. 7-8), has made the transition to an I~ meristem. To test this, we constructed det vegl double mutants and obtained a surprising result. The plants flowered, and no stubs formed on any of the axes (Figs. 9-11). Thus, elimination of the function of two genes is necessary for terminal flower formation in P. sativum. In A. thaliana and A. majus, a single gene (the homologs TFL and CEN, respectively) maintains indeterminacy in the terminal meristem (Shannon & Meeks-Wagner, 1991 ; Bradley et al., 1997). It is possible that either DET or VEG1 is homologous to TFL and CEN; however, since VEG1 appears to be responsible for I2 stub formation, DET is the more likely candidate. Alternatively, the DET VEG1 gene combination that suppresses determinacy may have evolved independently. Stebbins (1974) has argued that a cymose inflorescence was basal in the angiosperms. Suppression of terminal flower formation appears to have occurred before the Fabaceae arose, since racemes are basal in this family (Tucker, 1998). The compound raceme observed in Pisum may be accounted for by the suppression of second-order flower development by a second gene. TFL has been proposed to have a more global effect on shoot architecture by extending both the vegetative and reproductive phases, so that senescence occurs before a terminal flower forms (Ratcliffe et al., 1998). CEN, however, is only expressed in later inflorescence stages. DET function appears to be restricted to the I1 and does not alter the number of vegetative nodes before flowering. B. THE REGULATIONOF FIRST-ORDER AXIS DEVELOPMENT BY UNI Unlike DET, UNI function is necessary for normal vegetative development. Two mutant alleles of UNlhave altered leaf morphology--simple leaves or conversion of tendrils to leaflets (Marx, 1987). We have observed a second phenotypic effect ofuni on the first-order axis. After initiating several I2 axillary meristems, uni plants terminate with the formation of a stub, just like det plants (Figs. 12-14). The same phenotype is observed in uni det plants, which are indistinguishable from uni plants. Thus DETprobabty functions downstream of UNI. This is

GENETICS OF INFLORESCENCE DEVELOPMENT

393

Figs. 4-6. Comparison of scanning electron micrographs of terminal meristems in wild-type and det plants. 4. A wild-type plant that had produced no new flowers in three weeks. The terminal meristem (M) is subtended by a leafprimordium (L) and an associated floral meristem (FM) formed by the I2. Magnification is 160x. 5. A terminal meristem dissected from a control wild-type plant that had ceased growth. Note the absence of epidermal hairs. Magnification is 280x. 6. Rudimentary stub (s) produced by the det terminal meristem. F labels the last flower initiated. It is formed in the axil of leaf L, which exhibits suppressed development. Magnification is 250x.

394

THE BOTANICAL REVIEW

Figs. 7-10. Comparison ofvegl and det vegl plants. 7. A vegl plant that has extensive vegetative axillary outgrowth at the nodes where I2 meristems would form in wild-type plants. Growth continues indefinitely. 8. The terminal (I0 meristem ofa vegl plant that has initiated 72 nodes. The morphology is equivalent to that of a vegetative plant. 9. The terminal meristem ofa det vegl double mutant plant terminates with the production of a flower. Since no stub or I~ meristem physically separates this flower and the last axillary flower, they often fuse as they develop, giving the appearance of a double flower~ 10. Scanning electron micrograph of the terminal flower ofa det vegl plant. The larger flower on the left is the last axillary flower, which is in close proximity to the smaller terminal flower on the right.

consistent with our hypothesis that DET and CEN/TFL are homologs. Expression of CEN is dependent on expression of FLO, the homolog of UNI (Bradley et al., 1996; Hofer et al., 1997). Terminal flower suppression results when FLO suppresses CEN. The interactions between FLO and CEN are indirect (Ratcliffe et al., 1998). In the pea, Ii identity is regulated by UNI and DET, which appear to function in the same pathway. VEGI may be necessary for the I i meristem to initiate an I2 meristem. It may be essential for stub formation.

GENETICS OF INFLORESCENCE DEVELOPMENT

395

Fig. 11. Schematic of a det vegl plant. Note that both the first- and second-order branches terminate directly in the formation of a flower and that no stubs are present. Labels are as in Figure 2.

Figs. 12-13. Morphology of a uni plant. 12. Termination of the uni Ii meristem. The aberrant "flower" is axillary in origin. 13. Scanning electron micrograph of a uni I~ meristem that has terminated as a stub (s). The It initiated an 12 meristem that gave rise to the "flower" (F*) prior to differentiating epidermal hairs and ceasing growth.

C. THE ROLE OF NEP IN SECOND43RDER AXIS DEVELOPMENT I2 meristems initiate one or two floral meristems before terminating as a stub. NEP appears to be responsible for limiting the number of flowers initiated by the I2 meristem before stub formation, nep-1 mutants can produce as many as five pods on the first inflorescence branch under long-day conditions before stub formation (Fig. 15). We have identified another allele

396

THE BOTANICAL REVIEW

Fig. 14. Schematic ofa uni plant. Compare the terminal phenotype with det (Fig. 3). The conversion of an Ii to an I2 is identical to that of the detplant. Flowers (F*) are also aberrant after a whorl of sepals is initiated. Labels are otherwise as indicated in Figure 2.

ofNEP (nep-2)that has a similar mutant phenotype in the multipodded det line obtained from Marx. The multipodded trait behaves as a recessive single gene. Complementation tests revealed that this multipodded gene is allelic to NEP. Thus we refer to nep as nep-1 and to the multipodded allele as nep-2. These alleles should be particulary useful in separating resourceallocation effects on architecture from basic parameters established genetically. Genetic and environmental effects were distinguished by growing wild-type, nep-1, and nep-2 plants in 10 cm and 15 cm pots (Table II). Wild-type plants always produced two flowers per node, whereas mutants produced more nodes than did the wild type in both 10 cm and 15 cm pots. Unlike the wild type, both nep-1 and nep-2 plants produced more flowers in 15 cm pots than in 10 cm pots. The larger pot size increases the root mass and, presumably, the resources available for growth. Thus, the nep-I and nep-2 mutations appear to release the growth of the I2 meristem from strict genetic control and allow for plasticity in development based on nutritional factors. Diggle (1992) suggests that developmental processes themselves are targets of natural selection and provides numerous examples illustrating complex relationships between development and evolution. It is intriguing that NEP plays a role in limiting F meristem production on an I2 in P. sativum and most likely its wild relatives; yet the perennial sweet pea, Lathyrus latifolius, produces large numbers of flowers and pods at each reproductive node for an extended period of time. One hypothesis is that the flower- and fruit-induced senescence in Pisurn is not present in L. latifolius. Wild-type P. sativum produces flowers over an extended period of time as each new reproductive node expands; however, in nep-! and nep-2 plants all flowers

GENETICS OF INFLORESCENCE DEVELOPMENT

397

Figs. 15--16. Phenotype of the nep-1 and nep-I det plants. 15. The It meristem of this nep-1plant continues to initiate nodes until developing flowers and pods induce senescence and meristem activity is halted (Figs. 4-5). Note that four flowers formed on the most basal I2.16. This nep-1 det plant shows an additive phenotype. The first two I2 branches produce five and four flowers, respectively. The I1 then morphologically becomes an 12, initiating two flowers before terminating as a stub (s).

Table II Effect of genotype and pot size on number of flowers produced by second-order inflorescence meristems a Genotype

NEP nep-1 nep-2

Pot size (inches) 4 6 4 6 4 6

Mean number of flowers 4- SE (n) 2.0 + 2.0• 2.7 + 4.2 • 2.8 + 4.0+

0.0(15) 0.0(15) 0.2 (19) b 0.3 (20) b 0.1 (45) r 0.1 (28) c

a All data are for the first floral node of each plant. b Two-tailed t-test indicates that means are different (p < .0001). c Two-tailed t-test indicates that means are different (p < .0001 ).

form and set fruit over a reduced time period. Given the constraint of flower/fruit-induced senescence, an extended period of time for flowering may be more advantageous than obtaining a few more seeds from the nep plant. Such mutants make possible a more integrated approach to the study of the evolution of plant development. We investigated the interaction ofDETand NEP by creating the double mutant det nep-1 (Fig. 16). The phenotype is identical to the multipodded det line (containing nep-2) that we obtained from Marx. The phenotype is additive, with additional flowers forming directly in the axils of the first-order branch before the first-order meristem terminates as a stub. Visually, this illustrates the homeotic conversion of the I~ meristem to an I2 meristem. The pheno-

398

THE BOTANICAL REVIEW

Figs. 17-19. Morphology and architecture ofpim plants. 17. A whole-plant image of a reproductive pim plant. Outgrowth in each axillary is extensive because of extensive branching from the third-order branch and beyond. 18. Scanning electron micrograph of the stub (s) formed by the 12meristem ofapim plant. 19. The 12meristem ofapim plant initiated an I3 meristem (as opposed to a third-order floral meristem). It is indicated by a 1~ since it is the primary axillary meristem initiated by the I2. Unlike wild-type pea inflorescences, a secondary axillary bud (2~ also developed and was initiallyvegetative. In this case bract development at that axillary node is completely suppressed, but in some pim plants the bracts expand. The stem tissue on the far right is also part of the I2 and went on to terminate as a stub.

type of the second-order inflorescence branch is altered by nep-2 but not by det, which also has no effect on the I2 in the single det mutant. Based on these observations, it is clear that I1 and I2 meristems have distinct, genetically controlled programs. D. THE ROLE OF PIM IN THIRD-ORDER AXIS DEVELOPMENT

1. The Effect of pim Mutation on Inflorescence Architecture In the reproductive phase of P. sativum development, the third-order meristem initiates floral organs. PIMis critical in specifying floral meristem identity. It does not have a direct effect on I1 or I2 meristems; nor does it alter the node of floral initiation. The pim mutant replaces flowers with extensive inflorescence and vegetative branching (Fig. 17). The I2 terminates in a stub after producing one or two nodes (Fig. 18). Curiously, both primary and secondary buds form and develop in these I2 nodes. In wild-type P. sativum, secondary axillary buds are present in the vegetative axils, but they seldom develop. We have observed secondary axillary bud formation in wild-type I2 nodes, but these buds do not develop. In pim these secondary axillaries develop later as vegetative shoots, perhaps because functional floral signal(s) are no longer present (Fig. 19).

GENETICS OF INFLORESCENCE DEVELOPMENT

399

Fig. 20. Polar (A) and profile (B) schematics of the pim triad. In place of a determinate third-order wild-type flower, the pim mutant develops a shoot with opposing nodes before terminating as a stub or carpel or reverting to an indeterminate vegetative shoot. The fourth-order axes are nearly opposite each other and have mirrored developmental fates; that is, both will develop as triads with the same phenotypic possibilities as the third-order meristem. This nesting of triads can continue indefinitely, depending on growth conditions. CB = second-order common bract (development may be suppressed in some pim plants); B = bracteole; b = abaxial; d = adaxial.

Figs. 21-23. Examples ofpim triads, all subtended by third-order bracteoles (b). 21. Fourth-order vegetative axillaries failed to expand, and the third-order axis terminated as an aberrant flower (F). Here the bracteoles are slightly alternate. 22. A triad with a third-order stub (s) and fourth-order vegetative axillaries. 23. A triad with vegetative shoots (v) for the third- and fourth-order axillaries.

400

THE BOTANICAL REVIEW

Fig. 24. Inflorescence axillary ofapimplant with extensive vegetative branching and compound leaf formation. In this case, the I2 initiated two I3 meristems before terminating as a stub. Both I3 meristems developed vegetatively. This is the fundamental reproductive axis ofapimplant. Extensive branching of the axillary inflorescences leads to complex structures that incorporate morphological markers of vegetative, infloral, and floral phases of development.

In the third-order branch, single flowers are replaced with elaborate branch systems that exhibit amalgams of floral, infloral, and vegetative character states. The first lateral appendages of the third- and higher-order meristems are paired, opposite foliar structures consistent with Tucker's (1987) definition of bracteoles in other Fabaceae, except that bracteoles tend not to have shoots or buds. Morphologically the subtending, decussate lateral appendages in pim flowers (presepal) seem homologous with bracteoles, as seen in Sophorajaponicum (Tucker & Douglas, 1994). After producing one to several sets of paired bracteate nodes, each stem terminates either as a stub or as a flower, or else exhibits indeterminate vegetative growth, reiterating the entire V---*I--*"F" pattern. From out of the bracteate nodes grow axillary stems, which, in rum, produce more axillary stems before the growing point pursues one of the three terminal options: stub, carpel, or vegetative shoot (Figs. 20-23). These triad structures are distinct from those reported for Petunia, in which the inflorescence meristem bifurcates into an F and an I meristem (Souer et al., 1998). Tucker (1987) has found pseudoracemes that resemble the pim triads in 5 of about 32 papilionaceous tribes. While there is more variability in the fate of the three meristems in thepim triad, it is possible that PIMsuppressesthis phenotype found in other relatives and may have evolutionary significance for inflorescence architecture. The absence ofbracteoles and other leafy outgrowths in the third-order axis of wild-type P. sativum, and their presence in pim plants, is indicative of another aspect of PIM suppression of vegetative growth.

GENETICS OF INFLORESCENCE DEVELOPMENT

401

Fig. 25. Schematic ofapim plant. Development deviates from the wild type beginning with the thirdorder axis, which initiates two nearly opposite fourth-order meristems that generally have identical fates. Figure 20 illustrates the different developmental outcomes for these meristems. Two different fourthorder fates are used in this figure to illustrate the range of possibilities. While this figure illustrates a single triad, in many pim plants the nesting of the triads continues. F* = an aberrant flower; V4 = fourthorder vegetative shoot; I3 = third-order inflorescence meristem. Other labels are as in Figure 2.

The development and outgrowth of more than one axis from the high-order nodes provides additional complexity to the branching pattern. Generally, these secondary axillaries develop as vegetative shoots. One result is the presence of compound leaves in the inflorescence. Although their development lags significantly behind that of the primary axillaries, they can eventually outgrow the remainder of the inflorescence. Thus, there exists the potential of indefinite proliferation of the pim inflorescence (Figs. 24-25). Floral development is aberrant in comparison with the carefully documented development of wild-type flowers by Tucker (1989). The petals are most affected, and organ fusions occur regularly (Figs. 26-28). The first initiated petals in P. sativum are those that fuse as the keel. These two petals can be distinguished by color in certain genetic backgrounds and are almost always missing in pim. Petals are lacking to various degrees. If only one petal is present, it is invariably the vexillum. Even when, in the rare instance, all five petals are present, their arrangement is skewed, so that the wild-type imbricate pattern is not followed. There is also a great deal of organ fusion, especially between sepals and anthers, and anthers and carpels. In some cases accessory flowers or inflorescences form in the axil of the sepals, which morphologically take on more leaflike, specifically stipulelike, traits (Fig. 29). The phenotype of the pim mutant is consistent with Poethig's (1990) contention that juvenile, adult, and reproductive programs are discrete and can overlap. Inpim, insufficiencies in the reproductive program may permit the phenotypes of other programs to become more apparent. Alternatively, PIM may be required to suppress vegetative growth. Although reproductive development does not require the termination of a vegetative phase in maize, it may

402

THE BOTANICAL REVIEW

Figs. 26-29. Floral abnormalities in pim plants. 26. A wild-type flower with normal keel development. 27. A pim flower with aberrant keel formation. 28. Scanning electron micrograph illustrating fusion of sepal (se) and anther (a), as well as carpel (c) and anther. (Bar = 500 szm.) 29. Two inflorescence axillaries (I) based in the sepal (se) ofapim plant. Note the proximity of these two axillaries to the base of the carpel (c). be that the vegetative program prevents normal floral development (Bassiri et al., 1992; Evans & Poethig 1997). Many examples of vegetative development exist in pim inflorescences, from the lack of bract suppression in some 12 branches to the presence of compound leaves in later-order inflorescence branches. The relative activity or level of products of these different programs (vegetative, infloral, and reproductive, in the terminology used in this article) may result in the distinct developmental states reported by McDaniel et al. (1992) in the progression from vegetative to reproductive development. For example, a meristem can be determined for inflorescence development in P. sativum, and this determination is distinct from determination for flower development (Ferguson et al., 1991). In the case ofpim, the third- and higher-order branches appear to express components of the vegetative, inflorescence, and floral programs in a nonlinear fashion (Fig. 30).

GENETICS OF INFLORESCENCE DEVELOPMENT

403

F:

Fig. 30. Model for indirect phase transitions in wild-type and pim plants. Wild-type first-order vegetative axes become reproductively competent (i.e., I~ axes) and generate second-order inflorescences from which third-order flowers arise (outside heavy arrows). The lack of PIM product is first evident when the transition from I to F is incomplete (*). Rather than producing determinate flowers, pim mutants generate shoots with mixed identity and extensive triad branching. The supernumerary axes produce new amalgams of morphological structures through incomplete and unstable phase transitions. Eventually, axillary primordia completely "revert" to the vegetative phase. The dashed internal arrows indicate developmental transitions that have been observed inpim inflorescences. This illustrates that a linear progression through these phases is not obligatory. V = vegetative; I = infloral; F = floral.

Time of development of specific meristems and primordia appears to be as critical as position in this mutant, in which reversion can occur. Hempel and coworkers (1998) have shown in A. thaliana that developing primordia have a quantitative response to floral induction signals and that the fate ofprimordia can be modified in situ after initiation. PIMmay act coordinately with other floral meristem identity genes to provide floral identity signals, not only in sufficient quantity but also at the correct times. 2. broccoli Enhances the p i m Phenotype We have identified a single gene, the recessive mutation broccoli (broc), which has a normal phenotype in a wild-type background. The double mutant, pim broc, has a more extreme pim phenotype (Figs. 31-32). The branching that occurs after the initiation of the third-order meristem is more extensive, and flower development is minimal. The phenotype resembles a head of broccoli. Thus BROC and PIM may serve partially redundant functions in assigning floral meristem identity. Morphologically, pim resembles the apetala-1 mutant ofA. thaliana, which exhibits a more extreme phenotype when combined with the cauliflower (cal) mutation (Bowman et al., 1993). BROC and CAL are also potential homologs. The triple mutant apl cal tfl has the same phenotype as apl tfl plants. Thus tfl appears to inhibit the enhancement of the apl phenotype by cal. Based on phenotype, we believe that DET in P. sativum, CEN in A. majus, and TFL in A. thaliana have potential functional homologies. The triple mutantpim broc det is additive, unlike the apl cal tfl triple mutant, indicating that det does not interfere with the enhancement ofthepim phenotype by broc (Fig. 33). Thus, if homologies exist among these three genes in A. thaliana and P. sativum, the ways they coordinate to effect inflorescence architecture will differ.

404

THE BOTANICAL REVIEW

Figs. 31-33. Interactions ofPIM, BROC, and DET. 31. Apim broc double mutant in which extensive branching beyond the I2 and suppression of leafy structures yields a broccoli-like phenotype, broc alone has a wild-type phenotype. Contrast this figure with Figure 17.32. Scanning electron micrograph of the inflorescence of a pim broc plant. The phenotype is characterized by numerous floral or prefloral meristems and an occasional sepal (se). These meristems are tightly packed together and in the lower reproductive nodes fail to develop further. More expansion of leaflike structures and some floral parts are observed in progressively more apical reproductive nodes, as seen in Figure 31.33. An additive phenotype is observed in the brocpim det triple mutant. The I1 terminates as an I2 with the characteristic broccoli phenotype.

GENETICS OF INFLORESCENCE DEVELOPMENT

405

Figs. 34-36. Interactions between PIM and UNI. 34. Scanning electron micrograph of a uni plant reveals a nesting of carpels and some sepals within a large sepal whorl, c = carpel; se = sepal. 35. Shoots consisting of spirals of leaflike sepals formpim triads in thepim uni double mutant. No identifiable floral organs form. 36. Scanning electron micrograph of a pim uni reproductive axillary beginning with the fifth-order axis (e.g., 5~ = fifth-order branch). The labels are on bracts. At these higher-order branch points, the triad architecture is sometimes lost and the meristem produces a single axillary before terminating as a stub. That axillary produces one node before terminating as a stub, resulting in sympodial growth pattern with leafy bracts alternating 180 ~ apart. E. THE ROLE OF UNI IN THIRD-ORDER AXIS DEVELOPMENT

UNI is necessary for normal floral development, in addition to its role in vegetative and I1 development (Fig. 12). The I2 ofuni plants is normal and terminates with a stub. The I3 begins to initiate what resembles a whorl of sepals, and then a series of flowerlike structures begins to

406

THE BOTANICAL REVIEW

Fig. 37. Model illustrating spatially where genes presented in this article affect inflorescence architecture. Note that the role of DETis to block or delay the transition of the I1 meristem to an I2 fate. Here I2 is used in the first order to indicate a developmental fate rather than a branch position, as is done elsewhere in the article. UNlis necessary for leaf development throughout growth and also for DETactivity and normal floral organogenesis. VEG1 and NEP are important in 12 development, whereas PIM and BROC are needed for the development of the third-order meristem as a flower. Mutations in any of these genes alter inflorescence architecture in distinct ways and collectively should provide insight into the regulation of the development of this compound raceme.

GENETICS OF INFLORESCENCE DEVELOPMENT

407

develop within these axils and the axils of later-initiated structures (Fig. 34; Hofer et al., 1997). Primarily sepal-carpel intermediates, carpels, and sepals form. Although we have observed an occasional anther, petals are absent. The phenotype is quite similar to the uni homolog,/fy, in A. thaliana (Weigel et al., 1992; Hofer et al., 1997). The double mutantpim uni branches likepim plants creating nested triads, although single axillaries initiated are usually suppressed in the wild type and often inpim. In place of flowers there are whorls of leaflike structures at later branch points (Figs. 35-36). uni appears to enhance the outgrowth of I2 bracts. Ifpim is the homolog of apl in A. thaliana, then the phenotype ofpim uni in P. sativum is consistent with the interactions reported between AP1 and LFY in A. thaliana (Bowman et al., 1993; Parcy et al., 1998). The double mutants apl lfy (there are multiple alleles of these A. thaliana genes) are morphologically similar to the pim uni plants. AP1 and LFY are believed to have overlapping and distinct functions. LFY has been shown to induce AP1 expression (Parcy et al., 1998). UNI is expressed in the vegetative phase ofP. sativum development and controls more than the floral meristem identity role proposed for PIM. Clearly the additive phenotypic effects seen in the aberrant inflorescences of pim uni plants indicate that UN! plays a role in flower development that is at least partially distinct from that of PIM.

VI. Summary The evolution and development of inflorescence architecture can be fruitfully investigated through analysis of mutants that appear to play a key regulatory role in inflorescence development. Our study of mutants in P. sativum has allowed us to identify genes that regulate key steps in compound raceme inflorescence development and to begin to explore their functions (Fig. 37). Not only are inflorescence and floral meristems developmentally distinct, but in a compound raceme like that ofP. sativum, inflorescences of different branch orders are also distinct. What is particularly valuable about this suite of mutants is that they affect several distinct stages of inflorescence development, which allows us to identify 11, I2, and F meristems in terms of key genes as well as of morphological traits. This has allowed us to explore the effects of changing certain inflorescence components on the overall architecture. Finding homologies between these genes and those that have been identified in A. majus and A. thaliana, which have simple rather than compound racemes, will further our understanding of the evolution of inflorescence architectures. A second avenue of pursuit is to take advantage of the extensive morphological data on Fabaceae and begin a comparative study to determine how gene function may have been modified, deleted, or added over time within this family.

VII. Acknowledgments We dedicate this article to the late Dr. Gerry Marx, who generously provided us with the mutants, advice, and gentle encouragement that launched our exploration of developmental regulation of inflorescence architecture in peas. Conversations about pea developmental mutants with Ian Muffet and Scott Taylor (University of Tasmania) and with Julie Hofer and Noel Ellis (John Innes Institute) have been most helpful. We thank the students on the Carleton Greenhouse Crew for excellent plant care. This work was supported by NSF RUI grant 9405799 and USDA grant 9103136 to Susan Singer.

408

THE BOTANICAL REVIEW

VIII. Literature Cited Alvarez, J., C. L. Gull, X-H. Yu & D. R. Smyth. 1992. TERMINAL FLOWER: A gene affecting inflorescence development in Arabidopsis thaliana. P1. J. 2:103-116. Amasino, R. 1996. Control of flowering time in plants. Curt. Opinion in Genet. & Developm. 6: 480-487. Bassiri, A., E. E. Irish & R. S. Poethig. 1992. Heterochronic effects of Teopod2 on the growth and photosensitivity of the maize shoot. P1. Cell 4: 497-504. Bowman, J. L., J. AIvarez, D. Weigel, E. M. Meyerowitz & D. R. Smyth. 1993. Control of flower development in Arabidopsis thaliana by APETALA1 and interacting genes. Development 119: 721-743. Bradley, D., R. Carpenter, L. Copsey, C. Vincent, S. Rothstein & E. Coen. 1996. Control of inflorescence architecture in Antirrhinum. Nature 379: 791-797. ---, O. Rateliffe, C. Vincent, R. Carpenter & E. Cnen. 1997. Inflorescence commitment and architecture in Arabidopsis. Science 275: 80-83. Chappill, J. A. 1995. Cladistic analysis of the Fabaceae: The development of an explicit phylogenetic hypothesis. Pp. 1-9 in M. D. Crisp & J. J. Doyle (eds.), Advances in legume systematics. Pt. 7. Phylogeny. Royal Botanic Gardens, Kew. Coen, E. S. & J. M. Nugent. 1994. Evolution of flowers and inflorescences. Development (suppl.): 107-116. Diggle, P. K. 1992. Development and the evolution of plant reproductive characters. Pp. 326-355 in R. Wyatt (ed.), Ecology and evolution of plant reproduction: New approaches. Chapman and Hall, New York. Dnebley, J. & L. Lukens. 1998. Transcriptional regulators and the evolution of plant form. P1. Cell 10: 1075-1082. Donoghue, M. J., R. H. Ree & D. A. Baum. 1998. Phylogeny and the evolution of flower symmetry in the Asteridae. Trends P1. Sci. 3:311-317. Doyle, J. J. 1997. A phylogeny of the chloroplast gene rbcL in the Fabaceae: Taxonomic correlations and insights into the evolution ofnodulation. Amer. J. Bot. 84: 541-554. Endress, P. K. 1994. Diversity and evolutionary biology of tropical flowers. Cambridge University Press, Cambridge. Evans, M. M. S. & R. S. Poethig. 1997. The viviparous8 mutation delays vegetative phase change and accelerates the rate of seedling growth in maize. PI. J. 12: 769-779. Ferguson, C. J., S. C. Huber, P. H. Hong & S. R. Singer. 1991. Determination for inflorescence development is a stable state, separable from determination for flower development in Pisum sativum L. Planta 185: 518-522. Frohlich, M. W. & E. M. Meyerowitz. 1997. The search for flower homeotic gene homologs in basal angiosperms and gnetales: A potential new source of data on the evolutionary origin of flowers. Int. J. PI. Sci. 158: S131-S142. Grimes, J. 1996. Branch apices, heterochrony, and inflorescence morphology in some mimosoid legumes (Leguminosae: Mimosoidea). Telopea 6: 729-748. Hempel, F. D., P. C. Zambryski & L. J. Feldman. 1998. Photoinduction of flower identity in vegetatively biased primordia. P1. Cell 10: 1663-1675. Hilu, K. W. 1983. The role of single-gene mutations in the evolution of flowering plants. Pp. 97-128 in M. K. Hecht, B. Wallace & G. T. Prance (eds.), Evolutionary biology. Plenum Press, New York. Hofer, J., L. Turner, R. Hellens, M. Ambrose, P. Matthews, A. Michael & N. Ellis. 1997. UNIFOL1ATA regulates leaf and flower morphogenesis in pea. Curr. Biol. 7: 581-587. Hole, C. C. & R. C. Hardwick. 1976. Development and control of the number of flowers per node in Pisum sativum, London Ann. Bot. 40: 707-722. Kelly, A. J., M. B. Bonnlander & D. R. Meeks-Wagner. 1995. NFL, the tobacco homolog of FLORICA ULA and LEAFY, is transcriptionally expressed in both vegetative and floral meristems. P1. Cell 7: 225-234. Kempin, S., B. Savidge & M. Yanofsky. 1995. Molecular basis of the cauliflower phenotype in Arabidopsis. Science 267: 522-525.

GENETICS OF INFLORESCENCE DEVELOPMENT

409

Luo, D., R. Carpenter, C. Vincent, L. Copsey & E. Coen. 1996. Origin of floral symmetry in Antirrhinum. Nature 383: 794-799. Mandel, M. A., C. Gustafson-Brown, B. Savidge & M. Yanofsky. 1992. Molecular characterization of the Arabidopsis floral homeotic gene APETALA1. Nature 360: 273-277. Marx, G. A9 1987. A suite of mutants that modify pattern formation in pea leaves. PI. Molec. Biol. Reporter 5:311-335. McDaniel, C. N., S. R. Singer & S. M. E. Smith. 1992. Developmental states associated with the floral transition. Developm. Biol. 153:59-69. Millonig, G. 1961. Advantage of a phosphate buffer for OsO4 solutions in fixation. J. Appl. Physics 32: 1637. Munster, T. J. Pahnke, A. Di Rosa, J. Kim, W. Martin, H. Saedlcr & G. Theissen. 1997. Floral homeotic genes were recruited from homologous MADS-box genes preexisting in the common ancestor of ferns and seed plants. Proc. Natl. Acad. USA 94: 2415-2420. Mnrfet, I. C. 1989. Flowering genes in Pisum. Pp. 10-18 in E. Lord & G. Bernier (eds.), Plant reproduction: From floral induction to pollination. American Society of Plant Physiologists, Rockville, MD. Parcy, F., O. Nilsson, M. A. Busch, I. Lee & D. Weigel. 1998. A genetic framework for floral patterning. Science 395: 561-566. Poethig, R. S. 1990. Phase change and the regulation of shoot morphogenesis in plants. Science 250: 923-930. Poteau, S., D. Nichols, F. Tooke, E. Coen & N. Batty. 1997. The induction and maintenance of flowering in Impatiens. Development 124: 3343-3351. Purugganan, M. D. 1997. The MADS-box floral homeotic gene lineages predate the origin of seed plants: Phylogenetic and molecular clock estimates. J. Molec. Evol. 45: 392-396. Rateliffe, O. J., I. Amaya, C. A. Vincent, S. Rothstein, R. Carpenter, E. S. Coen & D. J. Bradley. 1998. A common mechanism controls the life cycle and architecture of plants. Development 125: 1609-1615. Reid, J. B. & I. C. Murfet. 1984. Flowering in Pisum: A fifth locus, veg. Ann. Bot. 53: 369-382. , , S. R. Singer, J. L. Weller & S. A. Taylor. 1996. Physiological-genetics of flowering in Pisum. Seminars Cell & Developm. Biol. 7: 455-463. Shannon, S. & D. R. Meeks-Wagner. 1991. A mutation in the Arabidopsis TFLI gene affects inflorescence meristem development. PI. Cell 3: 877-892. - & .1993. Genetic interactions that regulate inflorescence development in Arabidopsis. P1. Cell 5: 639-655. Singer, S. R., L. P. Hsiu ng& S. C. Huber. 1990. Determinate (det) mutant ofPisum sativum (Fabaceae: Papilionoideae) exhibits an indeterminate growth pattern. Amer. J. Bot. 77:1330-1335. Souer, E., A. van der Krol, D. Kloos, C. Spelt, M. Bliek, J. Mol & R. Koes. 1998. Genetic control of branching pattern and floral identity during Petunia inflorescence development. Development 125: 733-742. Stebbins, G. L. 1974. Flowering plants: Evolution above the species level. Harvard University Press, Cambridge, MA. Tucker, S. C. 1987. Pseudoracemes in papilionoid legumes: Their nature, development and variation. J. Linn. Sot., Bot. 95: 181-206. 9 1989. Overlapping organ initiation and common primordia in flowers ofPisum sativum (Fabaceae: Papilionoideae). Amer. J. Bot. 76: 714-729. ---. 1997. Floral evolution, development, and convergence: The hierarchical-significance hypothesis. Int. J. Plant Sci. 158: S143-S161. ---. 1998. Floral ontogeny in legume genera Petalostylis, Labichea, and Dialium (Caesalpinoideae: Cassieae), a series in floral reduction. Amer. J. Bot. 85: 184-208. - & A. W. Douglas. 1994. Ontogenetic evidence and phylogenetic relationships among basal taxa of legumes. Pp. 11-32 in I. K. Ferguson & S. C. Tucker (eds.), Advances in legume systematics. Pt. 6. Structural botany. Royal Botanic Gardens, Kew. Weberling, F. 1989. Morphology of flowers and inflorescences. Cambridge University Press, Cambridge. Weigel, D. 1995. The genetics of flower development: From floral induction to ovule morphogenesis. Annual Rev. Genet. 29: 19-39.

410

THE BOTANICAL REVIEW

Weigel, D., J. Alvarez, D. R. Smyth, M. F. Yanofsky & E. M. Meyerowitz. 1992. LEAFYcontrols floral meristem identity in Arabidopsis. Cell 69: 843-860. Weller, J. L., J. B. Reid, S. A. Taylor & I. C. Murfet. 1997. The genetic control of flowering in pea. Trends P1. Sci. 2: 1360-1385. Wray, G. A. 1994. Developmental evolution: New paradigms and paradoxes. Developm. Genetics 15: 1-6. Yanofsky, M. F. 1995. Floral meristems to floral organs: Genes controlling early events in Arabidopsis flower development. Annual Rev. P1. Physiol. & P1. Molec. Biol. 46: 167-188.