A Morphological and Quantitative Characterization of Early Floral Development in Apple (Malus × domestica Borkh.)

Abstract

Apple is an important crop and a focus of research worldwide. However, some aspects of floral commitment and morphogenesis remain unclear. A detailed characterization of bourse shoot apex development was undertaken to provide a framework for future genetic, molecular and physiological studies. Eight morphologically distinct stages of shoot apex development, prior to winter dormancy, were defined. Based on measurements of meristem diameter, two stages of vegetative development were recognized. Vegetative meristems were flat, and either narrow (stage 0) or broad (stage 1). Pronounced doming of the apex marked stage 2. During stage 3, the domed meristem initiated four to six lateral floral meristems and subtending bracts before converting to a terminal floral meristem (stage 4). The terminal floral meristem proceeded directly with bractlet and sepal initiation, while lateral floral meristems initiated bractlets (stage 5). Sepal initiation began on the basal lateral flower (stage 6) and continued in an acropetal direction until all floral meristems had completed sepal initiation (stage 7). In this study, only stage 0 and stage 7 apices were observed in dormant buds, indicating that stages 1–6 are transient. The results suggest that broadening of the apex (stage 1) is the first morphological sign of commitment to flowering.

INTRODUCTION

In the post‐genomics era, it is increasingly simple to clone genes from non‐model species. However, issues of what constitutes an orthologous gene, and what the exact functions of orthologous genes are, remain difficult. Comparing loss/gain of function phenotypes or gene expression patterns with those characterized in model species can provide useful information about the mechanisms of gene action, but only if there is a firm understanding of the developmental processes in the species being studied. In a landmark paper, Smyth and co‐workers undertook a detailed characterization of floral development in wild‐type Arabidopsis thaliana, providing reference observations that define 12 distinct developmental stages (Smyth et al., 1990). This work has provided a developmental context in which gene expression patterns and mutant phenotypes can be compared and interpreted in Arabidopsis. Despite the utility of this approach, many non‐model systems lack comparable developmental analyses. One such example is apple (Malus × domestica Borkh.), a crop of worldwide horticultural importance, and a useful system to study flowering in woody angiosperms.

Elucidating the genetic and environmental factors that control the transition to flowering and floral development in apple is an active area of research. A number of genes that control the transition to reproductive development have been identified and characterized in model species such as Arabidopsis and Antirrhinum (Mandel et al., 1992; Weigel et al., 1992; Blazquez et al., 1997; Bradley et al., 1997; Hempel et al., 1997). Apple orthologues of several of these genes have been isolated (Sung and An, 1997; Sung et al., 1999, 2000; Wada et al., 2002). However, the developmental context in which these genes act is still poorly understood because the published descriptions of floral morphogenesis in apple are inadequate. For instance, a number of authors use the terms inflorescence and flower interchangeably, and most fail to describe the sequence of lateral flower initiation (Pratt et al., 1959; Fulford, 1966b; Buban and Faust, 1982; Hirst and Ferree, 1995).

In apple, flowers are produced on terminal inflorescences (Fig. 1). New growth is initiated from one or more axillary meristem(s) proximal to the terminal inflorescence (Crabbé and Escobedo Alvarez, 1991). During the growing season, these axillary meristems initiate a series of vegetative leaves, bud scales and leaf primordia that comprise the bourse shoot. A bourse shoot may either terminate in an inflorescence (Fig. 1) or remain vegetative (Fulford, 1966a).

Fig. 1. Apple floral bud before and after bud break. A, Schematic illustrating the organization of a winter dormant floral bud. In the previous season, the terminal meristem initiated a series of vegetative leaves (not represented), then bud scales (curved lines), transition leaves (cross‐hatched) and leaf primordia (filled), before terminating in an inflorescence. Subsequent vegetative growth is from a ‘bourse’ shoot (circled) that develops from the axil of one of the vegetative primordia. B, Macroscopic image of floral bud after bud break. Flowers and expanded vegetative leaves (VL) were initiated in the previous season; new leaves (L1–L3) are initiated by the bourse shoot (arrowhead) terminal meristem.

Due to the horticultural importance of apple, much of the work on flowering has focused on the later stages of floral morphogenesis and fruit development (for a review, see Pratt et al., 1988). There have been a number of studies on the timing of the transition from vegetative to floral development, but much of the early literature focused on the numbers, type, and rate of lateral organ appearance rather than events occurring at the shoot apex itself (Fulford, 1965, 1966a, b). Doming of the apex has been identified as an indication of the transition from vegetative to reproductive development, but it is unclear if this is truly the first morphological sign of floral commitment (Buban and Faust, 1982; Fulford, 1966b; Luckwill and Silva, 1979). Pratt et al. (1959) noted that a broadening of the apex precedes doming, although this observation was not quantified. To date, there has been no comprehensive developmental analysis of the morphogenetic events between vegetative development and floral organ initiation.

In this paper, scanning electron micrographs and images of sectioned shoot apices are presented, illustrating the progression from vegetative to floral development. Based on these, eight morphologically distinct stages of shoot apex development prior to winter dormancy are defined. We discuss these stages with reference to meristem identity. This detailed description provides morphological markers that are more indicative of the developmental state of the shoot apex than temporal references, such as dates, degree days, or days relative to blooming. Quantitative analyses of meristem diameter over time and transitions between meristem identity states are presented. These data provide quantitative support for the observations of Pratt et al. (1959) and suggest that broadening of the apex is the first morphological sign of the transition to flowering. It is believed that this study will provide a framework for the interpretation of future molecular, physiological and developmental studies.

MATERIALS AND METHODS

Samples were collected from 15‐year‐old ‘Royal Gala’/MM.106 trees growing at the HortResearch Lawn Road orchard near Havelock North, New Zealand. For each collection, approx. 50 bourse shoots (Fig. 1B) were removed from spurs (short shoots) carrying flowers/fruit. One bourse shoot per branch was collected from random positions on well‐exposed branches. Two to three branches were sampled per tree, and 16–25 different trees were sampled at each time‐point. A previous study of Royal Gala grown in the same geographical location found no correlation between heat accumulation and the onset of floral bud development (McArtney et al., 2001), thus sampling dates are presented as days after full bloom (DAFB). Full bloom is defined as the date when 80 % of the terminal (king) flowers on spurs are open. At 21 DAFB, long (>50 mm) and short (≤50 mm) bourse shoots could be distinguished from one another. From this point onwards, only short bourse shoots were collected to minimize variation. Collections were made at approx. 2‐week intervals from –14 DAFB (21 Sep. 2001) until 141 DAFB (22 Feb. 2002). Additional samples were collected at 189 and 280 DAFB (11 Apr. and 11 Jul. 2002, respectively). For the last four collections (127, 141, 189 and 280 DAFB), the sample size was increased to approx. 70 buds per collection.

Scanning electron and light microscopy

Expanded leaves and/or bud scales were removed under a binocular microscope until six to ten leaf primorida remained at the shoot tip. Dissected apices were fixed in either 3 % glutaraldehyde in 0·05 m phosphate buffer for scanning electron microscopy (SEM), or FAA (3·7 % formaldehyde, 5 % acetic acid, 50 % ethanol) for embedding in paraffin. Within each of the first ten collections, approx. 20 samples were fixed for SEM and 30 for light microscopy. In the last four collections, approx. 20 samples were fixed for SEM and 50 for light microscopy. Samples for SEM were dehydrated in acetone, critical‐point dried in liquid CO₂, and the remaining leaf primorida were removed under a binocular microscope to reveal the shoot apex. Samples were mounted on stubs and sputter coated with 25 nm gold (SCD‐050; Bal‐Tec, Balzers, Liechtenstein). Specimens were examined on a Cambridge 250 Mark III scanning electron microscope (Cambridge Instruments, Cambridge, UK) operated at 20 kV, and photographed using 35 mm film. Tissue embedded in paraffin was sectioned at 10 µm (Leica microtome, Wetzlar, Germany), heat mounted, stained with saffranin/fast green (Ruzin, 1999), and photographed using a 35‐mm camera attached to an Axioplan microscope (Zeiss, Jena, Germany). Figures were assembled using Adobe Photoshop (Adobe Systems, Mountain View, CA, USA).

Vegetative meristem measurements

Vegetative (flat) meristem diameter was measured from calibrated digital images of median longitudinal sections (Leica MZFLIII stereomicroscope equipped with a DC200 digital camera; Wetzlar, Germany). From –14 to 67 DAFB, all samples were vegetative and 25–28 meristems were measured per collection. As the proportion of flowering apices increased during the season, the proportion of vegetative meristems per collection decreased. From each of the 127, 141, 189 and 280 DAFB collections, 40–50 individuals were sectioned and three to six vegetative meristems were observed and measured in each collection.

Preliminary analysis of the data pooled from all sampling dates indicated a bimodal distribution of meristem diameters, with a boundary between the two modes at approx. 130 µm. Therefore, meristems with a diameter ≤130 µm were defined as stage 0 and those >130 µm as stage 1. For each sampling date from –3 to 109 DAFB, mean meristem diameter was calculated separately for stage 0 and stage 1 meristems. Due to the small number of vegetative meristems in the 127, 141, 189 and 280 DAFB collections, these time‐points were not included when calculating the change in mean meristem diameter over time. However, these samples were included when calculating the proportion of stage 0 and stage 1 vegetative meristems within the population. Data were analysed using Origin 6.0 (Microcal Software, Inc., Northampton, MA, USA).

RESULTS AND INTERPRETATION

Stages of floral morphogenesis in apple

Based on our analysis of SEMs and sections of bourse shoot apices, eight morphologically distinct stages from vegetative development to floral organ initiation were defined.

Initially, at –14 DAFB, all bourse shoot meristems were flat, and narrow (Fig. 2A and B). By the second collection, –3 DAFB, some of the meristems appeared significantly broader (Fig. 2C and D). Based on measurements of meristem diameter (see Materials and Methods), we classified flat meristems as either narrow (stage 0) or broad (stage 1). From –3 until 127 DAFB, both stage 0 and stage 1 meristems were observed.

Fig. 2. Scanning electron micrographs and images of sectioned vegetative and inflorescence meristems. SEM of a narrow, flat vegetative meristem (stage 0) (A), a broad, flat vegetative meristem (stage 1) (C) and a domed inflorescence meristem (stage 2) (E). Median longitudinal section through a stage 0 (B), a stage 1 (D) and a stage 2 meristem (F). Bars = 100 µm.

From 77 to 127 DAFB, we observed pronounced doming of the apex in the majority of individuals sampled (Fig. 2E and F). In domed apices, the youngest primordia are positioned higher on the apical dome than older primordia, in contrast to flat meristems in which all leaves are inserted at the same level (compare F with B and D in Fig. 2). Domed apices were designated stage 2.

Domed apical meristems initiate four to six bracts, each subtending a lateral floral meristem (Fig. 3A and B). The first two to three bracts have a wide lamina and basal stipules similar to vegetative leaves; later bracts are narrow and devoid of stipules. Bracts and floral meristems are initiated in the same phyllotaxy as the vegetative leaves, with a divergence angle of approx. 137° (Fig. 3B). Each lateral floral meristem is initiated from the flanks of the domed apical meristem and is initially concealed by a subtending bract. We define stage 3 as the point at which the first lateral floral meristem becomes visible. Each lateral floral meristem initially appears as a narrow fold of tissue between the bract and the domed meristem before becoming elliptical (compare ‘FM’ in A and B in Fig. 3).

Fig. 3. Scanning electron micrographs of stages 3–7. Side‐on (A) and top‐down (B) view of stage 3 domed inflorescence meristem (INFM) and lateral floral meristems (FM) subtended by bracts (B). C, A stage 4 apex in which the terminal floral meristem (tFM) has initiated a terminal bract (tB) and one bractlet (b*). Lateral floral meristems each initiate a pair of bractlets (marked ‘*’). D, A stage 5 apex in which the terminal floral meristem has initiated five sepals (S1–S5), and pairs of bractlets elongate on lateral floral meristems. E, During stage 6, the lateral flowers initiate sepals (S), beginning with the basal‐most and continuing in an acropetal direction. F, By stage 7, all lateral flowers have initiated sepals. Br, bract removed. Bars = 100 µm for A–F.

Stage 4 begins with the initiation of a bract that subtends the apical meristem itself (marked ‘tB’ in Fig. 3C). This contrasts with stage 3, in which newly initiated bracts subtend lateral meristems. The stage 4 terminal meristem initiates a single bractlet (second‐order bract), which is inserted higher on the apex than the primary bract (compare ‘tB’ and ‘b*’ in Fig. 3C). The lateral floral meristems initiate two bractlets each; these are visible as bulges on the periphery of the basal floral meristems (note asterisks in Fig. 3C). As the lateral meristems enlarge, the subtending primary bracts become displaced and take on a more upright appearance (compare A and C in Fig. 3). The shape of the apex changes from domed to slightly flattened (Fig. 3C).

During stage 5, the terminal meristem initiates five sepals in the same phyllotaxy as the vegetative organs (note the size difference between individual sepals in Fig. 3D). Once the terminal meristem has begun sepal initiation, pairs of bractlets on each of the lateral meristems begin to elongate, starting with those on the proximal lateral meristem and continuing in an acropetal direction.

Stage 6 is defined as an inflorescence in which the proximal meristem has begun sepal initiation, but the distal lateral meristem has initiated only bractlets (Fig. 3E). During stage 6, the terminal flower initiates petals and stamens (not shown). By stage 7, all lateral meristems have begun floral organ initiation (Fig. 3F).

Analysis of vegetative meristem diameter through one growing season

Analysis of meristem diameter data pooled from all sampling dates indicated a bimodal distribution with a boundary between the two modes at 130 µm (Fig. 4). For each sampling date from –3 to 109 DAFB, mean meristem diameter was calculated separately for stage 0 and stage 1 meristems (Fig. 5). In the initial collection (–14 DAFB), all sampled meristems were stage 0, with a mean diameter of 95 µm. Stage 1 meristems were observed between –3 and 96 DAFB, and the mean diameter increased during this period. In collections later than 96 DAFB, no stage 1 meristems were observed. From –3 DAFB, the mean diameter of stage 0 meristems decreased. It is suggested that this trend reflects a decrease in the number of large stage 0 meristems that are in transition to stage 1, rather than a decrease in the size of individual stage 0 meristems.

Fig. 4. Histogram showing the distribution of vegetative meristem diameters. The measurements of flat meristems from all collections were pooled. The data have a distribution with a boundary at approx. 130 µm.

Fig. 5. Mean diameter of stage 0 and stage 1 meristems from –14 to 109 days after full bloom (DAFB). Individual flat meristems were measured and classified as stage 0 (≤130 µm) or stage 1 (>130 µm). At –14 DAFB, only stage 0 meristems were observed. Stage 1 meristems were observed between –3 and 96 DAFB. Open triangles, stage 0; open circles, stage 1. Vertical bars represent s.e. for each mean.

Due to destructive sampling, it was not possible to follow the fate of individual meristems. However, it is hypothesized that the majority of individual meristems underwent transition from stage 0 to stage 1 during the season. The observation that very few meristems had a diameter between 120 µm and 140 µm suggests that individual meristems undergo a rapid transition from stage 0 to stage 1.

Transitions between meristem identity within a population

To examine transitions between meristem identities within a population, stages that represent a common terminal meristem identity were combined (see Discussion). The four meristem identity states are vegetative meristem (stage 0), vegetative meristem committed to become floral (stage 1), inflorescence meristem (stages 2 and 3) and floral meristem (stages 4–7). The observed frequency of individuals with a given meristem identity is represented for each collection (Fig. 6). As described in the previous section, all meristems were initially stage 0. Some individuals had undergone the transition to stage 1 before the second collection (–3 DAFB). The frequency of stage 1 meristems peaked at 53 DAFB. The decrease in the frequency of stage 1 meristems coincides with an increase in the frequency of inflorescence meristems (stages 2 and 3). The frequency of inflorescence meristems peaked between 96 and 109 DAFB, then decreased as the frequency of floral meristems increased. By 189 DAFB, the population dynamics appeared stable, with approx. 95 % of individuals floral and 5 % stage 0.

Fig. 6. Transitions between meristem identities. Stages representing a common terminal meristem identity were pooled into four groups: vegetative meristem (stage 0), vegetative meristem committed to floral development (stage 1), inflorescence meristem (stages 2 and 3) and floral meristem (stages 4–7). The frequency of individuals with a given meristem identity state was plotted over time. A, The frequency of vegetative meristems decreased exponentially until it became constant at approx. 5 %. B, The frequency of committed meristems peaked at 53 DAFB, and had decreased to zero by 109 days after full bloom (DAFB). C, The transition to inflorescence meristem occurred rapidly, with a sharp peak around 100 DAFB. D, The transition to terminal floral meristem identity occurred between 100 and 141 DAFB.

DISCUSSION

To date, there has been no comprehensive analysis of early floral development in apple. Using scanning electron micrographs and sectioned material, a series of morphological landmarks that defines each stage of floral development is presented here. These reference observations will be useful in the interpretation of gene expression patterns and abnormal modes of development in apple.

Eight morphologically distinct stages of shoot meristem development prior to winter dormancy are defined. Early in the growing season, all meristems were vegetative, flat and narrow (stage 0). The majority of vegetative meristems broadened (stage 1), and eventually became domed (stage 2). The domed apical meristem initiated four to six lateral meristems, each subtended by a bract (stage 3), before converting to a terminal floral meristem (stage 4). The terminal flower proceeded directly with sepal initiation, while lateral floral meristems initiated bractlets (stage 5). Sepal initiation began on the basal lateral flower (stage 6) and continued in an acropetal direction until all floral meristems had completed sepal initiation (stage 7).

By observing the progression from vegetative to floral development, the identity of the terminal meristem at each stage can be deduced (summarized in Table 1). Stage 0 and stage 1 meristems both initiate leaves, and are therefore vegetative meristems. Stage 2 and stage 3 meristems are inflorescence meristems; they initiate bracts, and lateral meristems that will develop into flowers. During stages 4–7, the terminal meristem initiates a single bractlet, then floral organs, and is therefore a floral meristem.

Table 1.

Morphological features that mark the beginning of each stage and corresponding terminal meristem identity

Stage	Morphology of apex/lateral meristems	Terminal meristem identity
0	Apex is flat, ≤130 µm in diameter	Vegetative meristem
1	Apex is flat, >130 µm in diameter	Vegetative meristem, committed to floral development
2	Apex is domed	Inflorescence meristem
3	Apex initiates bracts and lateral floral meristems
4	Terminal meristem initiates own bract and bractlet, lateral floral meristems initiate bractlet pairs	Floral meristem
5	Terminal floral meristem initiates sepals	Floral meristem
6	Proximal lateral floral meristem initiates sepals	Floral meristem
7	All lateral floral meristems have initiated sepals	Floral meristem

Based on the bimodal distribution of flat meristem diameters, we suggest that there are two populations of vegetative meristems. While all vegetative meristems are initially stage 0, the majority become stage 1. In our study, only stage 0 and stage 7 apices were observed in dormant buds (280 DAFB), indicating that stages 1–6 are transient. Due to destructive sampling, the fate of a single meristem cannot be followed, but it appears that all meristems that become stage 1 eventually reach stage 7. This suggests that broadening of the apex is the first morphological indication of the transition to flowering and reflects commitment to floral development.

Previous descriptions of bourse shoot apex development have noted that meristem diameter increases prior to floral development (Pratt et al., 1959). Buban and Faust (1982) reported an increase in mitotic activity and changes in histological zonation prior to doming. An increase in mitotic frequency and changes in the morphology of the shoot apex are widely recognized as signs of the transition from vegetative to floral development in a wide range of species (Steeves and Sussex, 1989). The results presented here provide quantitative support for the concept that broadening of the apex is an indication of the transition to floral development in apple.

The observed increase in frequency of stage 1 between 39 and 53 DAFB agrees with the observation that histological changes at the shoot apex occur between 3 and 6 weeks after full bloom (Buban and Faust, 1982). Other estimates of when floral development begins are based on doming of the apex, or the appearance of floral organs, and are therefore much later than our estimates (Fulford, 1966b; Luckwill and Silva, 1979).

A number of well‐characterized genes are known to control the transition from vegetative to floral development in model species such as Arabidopsis and Antirrhinum (Mandel et al., 1992; Weigel et al., 1992). The Arabidopsis gene LEAFY (LFY) appears to regulate the earliest stages of the floral transition (Blazquez et al., 1997). Two LFY orthologues, AFL1 and AFL2, have been isolated from apple (Kotoda et al., 2000; Wada et al., 2002). AFL2 was constitutively expressed, whereas AFL1 was up‐regulated before doming was observed. In light of the results presented here, it would be interesting to compare AFL1 expression levels in narrow (stage 0) and broad (stage 1) vegetative meristems.

Another Arabidopsis gene, APETELA1 (AP1), is expressed immediately after floral determination and is thought to act semi‐redundantly with LFY to promote floral development (Hempel et al., 1997). Sung et al. (1999) have cloned MdMADS2, a presumptive AP1 orthologue from apple, and show immunolocalization of MdMADS2 in tissue sections. These authors claim that MdMADS2 is expressed in the inflorescence meristem. However, the meristems they present appear broad and flat, and have not begun initiating floral meristems, similar to those defined as stage 1. The expression of an AP1 orthologue in a stage 1 meristem supports the hypothesis that broadening of the apex is the first step of floral commitment.

The observation that the apex domes (stage 2) between 96 and 109 DAFB agrees closely with previous reports (Luckwill and Silva, 1979; Hirst and Ferree, 1995; McArtney et al., 2001). The sharp peak in the observed frequency of inflorescence meristems during this period suggests that the transition from stage 1 to stage 2 occurs relatively synchronously within the population. These results are consistent with a model in which the transition to stage 1 is regulated largely by internal signals, and the transition from stage 1 to stage 2 is synchronized by environmental signals.

The transition from inflorescence to floral meristem also occurs synchronously within the population. This may reflect the relatively constant number of lateral flowers initiated. In Arabidopsis, the TERMINAL FLOWER (TFL) gene acts to maintain the inflorescence meristem in an indeterminate state (Bradley et al., 1997). The expression pattern of the apple TFL orthologues is likely to reflect the determinate nature of the apple inflorescence. Based on the TFL expression pattern in Arabidopsis, we would expect apple orthologue(s) to be expressed in the inflorescence meristem during stages 2 and 3, while lateral floral meristems are initiated, and down‐regulated in the inflorescence meristem as it is converted to a floral meristem during stage 4. Manipulation of apple orthologue(s) of TFL may be useful in developing horticultural varieties that do not require chemical or manual fruit thinning.

While it is clear that apple has a determinate inflorescence, the morphological terminology and sequence of floral development are controversial (Pratt et al., 1988). Bijhouwer (1924) described the apple inflorescence as having a tendency toward dichasial branching. In some cultivars, we have occasionally observed supernumery flowers in the axils of bractlets, but without the highly regular branching pattern that defines a dichasium (Weberling, 1989). Bracts are initiated by the inflorescence meristem and subtend lateral floral meristems, whereas bractlets are initated by the floral meristems. In most inflorescences no organs develop from the axils of bractlets. It is suggested that the apple inflorescence is a simple panicle, which is characterized by the main and lateral axes each terminating in a single flower. Paniculate inflorescences are relatively common within the Rosaceae (Evans and Dickenson, 1999a, b). The lateral flowers are initiated before the terminal flower, but the terminal flower is the most developmentally advanced.

It is concluded that there are two transient meristem identity states in the progression from vegetative to floral development (Fig. 7). The first transient state (stage 1) is characterized by broadening of the apex, which is proposed to be the first committed step towards floral development. In the study presented here, approx. 95 % of individuals reached stage 1. The results suggest that all stage 1 meristems become inflorescence meristems (stages 2 and 3), the second transient meristem state. Inflorescence meristems terminate in floral meristems (stages 4–7). It is believed that this study provides a contextual framework for understanding the genetic regulation of flowering in apple.

Fig. 7. Diagram representing meristem identity states in the progression from vegetative to floral development. In this study, approx. 95 % of vegetative meristems proceeded to stage 1, and 5 % remained at stage 0. All stage 1 meristems became inflorescence meristems (stages 2 and 3), and all inflorescence meristems terminated in floral meristems (stages 4–7). Based on these observations, it is concluded that stage 1 represents the first committed step towards floral development.

ACKNOWLEDGEMENTS

We thank Doug Hopcroft and Raymond Bennett for support with scanning electron microscopy, and Drs Eric Walton, Kim Snowden, Steve McArtney and Bruce Veit for comments and criticism. This research was funded by the New Zealand Foundation for Research Science and Technology.

Received: 23 December 2002; ; Returned for revision: 13 March 2003. Accepted: 17 April 2003    Published electronically: 12 June 2003