|Home | About | Journals | Submit | Contact Us | Français|
Fritillaria-type female gametophyte development is a complex, yet homoplasious developmental pattern that is interesting from both evolutionary and developmental perspectives. Piper (Piperaceae) was chosen for this study of Fritillaria-type female gametophyte development because Piperales represent a ‘hotspot’ of female gametophyte developmental evolution and have been the subject of several recent molecular phylogenetic analyses. This wealth of phylogenetic and descriptive data make Piper an excellent candidate for inferring the evolutionary developmental basis for the origin of Fritillaria-type female gametophytes.
Developing ovules of Piper peltatum were taken from greenhouse collections, embedded in glycol methacrylate, and serially sectioned. Light microscopy and laser scanning confocal microscopy were combined to produce three-dimensional computer reconstructions of developing female gametophytes. The ploidies of the developing embryos and endosperms were calculated using microspectrofluorometry.
The data describe female gametophyte development in Piper with highly detailed three-dimensional models, and document two previously unknown arrangements of megaspore nuclei during early development. Also collected were microspectrofluorometric data that indicate that Fritillaria-type female gametophyte development in Piper results in pentaploid endosperm.
The three-dimensional models resolve previous ambiguities in developmental interpretations of Fritillaria-type female gametophytes in Piper. The newly discovered arrangements of megaspore nuclei that are described allow for the construction of explicit hypotheses of female gametophyte developmental evolution within Piperaceae, and more broadly throughout Piperales. These detailed hypotheses indicate that the common ancestor of Piperaceae minus Verhuellia had a Drusa-type female gametophyte, and that evolutionary transitions to derived tetrasporic female gametophyte ontogenies in Piperaceae, including Fritillaria-type female gametophyte development, are the consequence of key nuclear migration and patterning events at the end of megasporogenesis.
Perhaps one of the most complex developmental sequences expressed among angiosperm female gametophytes is the Fritillaria-type ontogeny. This developmental pattern, which begins with a tetrasporic coenomegaspore, involves a free-nuclear division in which three haploid megaspore nuclei initiate mitosis and produce two triploid restitution nuclei (instead of six haploid nuclei; Bambacioni, 1928; Maheshwari, 1950; Johri et al., 1992; Batygina, 2006). As far as is known there is no example of mitosis in any group of eukaryotes in which fewer nuclei are present after a mitotic division is initiated than before. In spite of the peculiarity of this division, Fritillaria-type female gametophyte development appears to have independently evolved in angiosperms at least seven times: in Liliaceae (Bambacioni, 1928; Romanov, 1957), Piperaceae (Kanta, 1962; Nikiticheva, 1981; Prakash and Kin, 1982; Prakash et al., 1994), Plumbaginaceae (Dahlgren, 1916; Haupt, 1934), Euphorbiaceae (Carano, 1925, 1926; Tateishi, 1927; Kapil, 1961), Cornaceae (D'Amato, 1946; Chopra and Kaur, 1965), Tamaricaceae (Mauritzon, 1936; Sharma, 1939, 1940; Battaglia, 1941, 1943; Johri and Kak, 1954) and Asteraceae (Fagerlind, 1939; Maheshwari and Srinivasan, 1944; Howe, 1964). A modified form of this ontogeny, Plumbagella-type female gametophyte development, is reported in Plumbagella (Dahlgren, 1915; Haupt, 1934; Fagerlind, 1938; Boyes, 1939).
The repeated origin of such a complex ontogenetic sequence makes Fritillaria-type female gametophyte development an ideal subject for an evolutionary developmental investigation. Yet, basic knowledge of this ontogeny in angiosperms is limited, and the developmental basis for the origin of this female gametophyte type has never been explored. Although several lineages of angiosperms contain members with Fritillaria-type female gametophyte development (see above), it was decided to study this ontogeny within Piperales. There are reports in Piperales of transitions from monospory to bispory (Johnson, 1900c; Quibell, 1941; Raju, 1961) and tetraspory (Johnson, 1900a, b, 1914; Swamy, 1944; Lei et al., 2002) and at least six of the ten standard female gametophyte ontogenies recognized by Maheshwari (1950) are known to exist in Piperales (Fig. 1) (Maheshwari, 1950; Davis, 1966; Johri et al., 1992; Batygina, 2006; Madrid and Friedman, 2008). Piperales has also been the subject of several recent phylogenetic analyses (Nickrent et al., 2002; Zanis et al., 2003; Neinhuis et al., 2005; Wanke et al., 2007a, b). This wealth of developmental diversity and robust phylogenetic sampling make Piperales an excellent candidate in which to study female gametophyte developmental evolution, and Piper an informative clade in which to study the evolution of Fritillaria-type female gametophyte development.
All members of the genus Piper reported until now show the Fritillaria-type (Joshi, 1944; Swamy, 1944; Maugini, 1953; Murty, 1959; Yoshida, 1960; Kanta, 1962; Nikiticheva, 1981; Prakash and Kin, 1982; Prakash et al., 1994) and female gametophyte development in Piper peltatum was investigated. Nuclear positioning, nuclear migration and nuclear division throughout the syncytial stages of female gametophyte development were documented. The ploidy levels of several cell types within the mature female gametophyte and the seed were also measured. The goals were to characterize better the specific developmental processes that are associated with female gametophyte development in Piper, and use these detailed ontogenetic insights to generate hypotheses about how diverse patterns of female gametophyte development may have evolved in Piperaceae.
Piper peltatum is a herbaceous shrub that is native to Mexico, Central America and South America. Piper peltatum may grow up to 3 m in height, and has stems with distinct nodes that may become woody at the base. The leaves are peltate and about 40 cm across. Developing infloresences of P. peltatum were collected from greenhouse collections at the University of Colorado in the springs and summers of 2006 and 2007. Each inflorescence was immersed in a modified PIPES buffer (60 mm PIPES, 25 mm HEPES, 5·4 mm CaCl2, 1·97 mm MgCl2) where it was dissected into smaller pieces that each contained about 12 flowers. These infloresence portions were then chemically fixed.
Developing flowers were placed in a fixative solution containing either 3:1 95 % ethanol:acetic acid or 4 % paraformaldehyde, 2·5 % gluteraldehyde, and 4 % acrolein in a modified PIPES buffer (above) for 12 h. Specimens were dehydrated in a graded ethanol series, infiltrated, embedded with glycol methacrylate (JB-4 embedding kit; Polysciences, Warrington, PA, USA), and serially sectioned into 5-μm-thick ribbons using a Microm HM330 rotary microtome and glass knives. Sections from flowers that were fixed in the 3:1 solution were stained with a solution of 0·2 µg mL−1 of 4′,6-diamidino-2-phenylindole (DAPI) and 0·1 mg mL−1 phylenediamine in 0·05 m Tris (pH 7·2). Sections from flowers that were fixed with paraformaldehyde, gluteraldehyde and ethanol were stained with 0·1 % toluidine blue. All slides were viewed with a Zeiss Axiophot microscope using brightfield, differential interference contrast, and epifluoresence optics. Digital micrographs were taken with a Zeiss Axiocam digital camera. Post-processing of all images was done with the Adobe Creative Suite 2 software package (San Jose, CA, USA). Digital image manipulations were restricted to processes that applied to the entire image unless otherwise noted in specific figure legends.
Selected 5-μm-thick serial sections of developing female gametophytes that had been fixed in paraformaldehyde, gluteraldehyde and acrolein were photographed using a Leica TCS SP2 AOBS laser scanning confocal microscope. Images were created by exciting the toluidine blue stain at 633 nm with a helium/neon laser, with detectors set to absorb within the 640–700 nm range. Each 5-μm-thick section was photographed along the z-axis to create a stack of eight to ten 0·5-μm-thick optical sections. Z-stacks from each section were aligned and modelled using the IMOD software package (Kremer et al., 1996).
Sections from flowers fixed in the 3:1 solution were stained with DAPI (see above) for 1 h at room temperature in a light-free environment. Microspectrofluorometric measurements of relative DNA levels of DAPI-stained nuclei were performed within 2 h. Measurements were made with a Zeiss MSP 20 microspectrophotometer with digital microprocessor coupled to a Zeiss Axioskop microscope equipped with epifluorescence (HBO100-W burner). An ultraviolet filter set (model number 48702) with excitation filter (365 nm, bandpass 12 nm), dichroic mirror (FT395) and barrier filter (LP397) was used with a Zeiss Plan NeoFluar ×40 objective. Before each recording session, the photometer was standardized by taking a reading of fluorescence emitted from a fluorescence standard (GG 17), and this reading was taken to represent 100 relative fluorescence units (RFU). At the completion of each session, an additional reading was made of the fluorescence standard to confirm that little or no deviation in the relative fluorescence value obtained from the fluorescence standard had occurred during the period of data recording. Relative DNA content for each nucleus was determined by summing the individual fluorescence values of each of the serial sections through that nucleus. A net photometric value for each section of a nucleus was obtained by recording an initial reading of the nucleus and subtracting a background value obtained from cytoplasm proximal to the nucleus. Thus, background fluorescence from the glycolmethacrylate was removed from the photometric analysis of relative DNA content.
The embryological literature listed in Table 1 was reviewed with respect to nuclear positioning, nuclear division, and wall formation during megasporogenesis and megagametogenesis. Data from the present study of P. peltatum were used as a template to identify equivalent female gametophyte nuclear positioning events in other previously studied species of Piperaceae. Character distribution and estimation of ancestral and derived states were inferred using the programs MacClade (Maddison and Maddison, 1992) and Mesquite (Maddison and Maddison, 2008). Characters were traced using the most parsimonious resolving option in MacClade and with the maximum likelihood analysis module in Mesquite. The recent angiosperm phylogenies of Nickrent et al. (2002), Zanis et al. (2003), Neinhuis et al. (2005) and Wanke et al. (2007a, b) were used for character mapping. Each clade was coded as either tetrasporic and the character being present, tetrasporic and the character being absent, or not tetrasporic. The absence of a character was described with two character states because the character could be absent for two distinct reasons. If a clade was tetrasporic and did not have the character, this was due to patterns of nuclear migration and nuclear division. However, if a clade was not tetrasporic, then it was not capable of carrying out the actions described that involve four megaspore nuclei, and this was due to preceding patterns of megasporogenesis. By using three character states, it was possible to account for each of these explanations.
The megasporocyte of P. peltatum is densely cytoplasmic and initially contains a single vacuole that is positioned on the chalazal-side of the megasporocyte nucleus (Fig. 2A, B). Smaller vacuoles accumulate throughout the cytoplasm, and this correlates with an increase in megasporocyte volume (Fig. 2C, D; see Video 1 in Supplementary data, available online). After attaining a length of approx. 25 µm, meiosis I occurs without cytokinesis. Out of the 272 ovules examined in this study, only one two-nucleate female gametophyte was observed (Fig. 2E, F; see Video 2 in Supplementary data), suggesting that meiosis II quickly follows meiosis I.
The axes of division in meiosis II are perpendicular. After this division, the female gametophyte is, on average, ≈30 µm long (n = 9) and has four tetrahedrally arranged megaspore nuclei (Fig. 3A–C; see Video 3 in Supplementary data, online). Dense areas of cytoplasm, that may be remnants of meiotic spindles, are found between megaspore pairs. However, full-fledged phragmoplasts were never observed and wall formation did not occur (Fig. 3A–C). As cytoplasmic connections between megaspore nuclei dissipate, separate vacuoles coalesce to form a single large vacuole, and megaspore nuclei migrate from their tetrapolar tetrahedral orientation (Fig. 3A–C) to a more evenly spaced tetrapolar cruciate (planar) arrangement (Fig. 3D–F; see Video 4 in Supplementary data). In this configuration, each of the four nuclei continues to occupy a separate pole of the cell, but all four nuclei are positioned within a single plane of symmetry. A small volume of cytoplasm, extending to the wall of the syncytium, always surrounds each nucleus at this stage (Fig. 3D–F) and the entire cell is, on average, ≈37 µm long (n = 10).
When the female gametophyte is about 45 µm in length, the two centrally placed (equatorial) megaspore nuclei migrate towards the chalazal pole, while the micropylar- and chalazal-most nuclei remain stationary. The final orientation of the four megaspores is a bipolar 1 + 3 arrangement in which a single nucleus is situated at the micropylar pole of the female gametophyte and three nuclei are clustered at the chalazal pole (Fig. 3G–I; see Video 5 in Supplementary data).
All four megaspores enter prophase of the first free-nuclear mitotic division of megagametogenesis simultaneously (Fig. 3G), but the divisions that occur at each pole of the female gametophyte are very different. At the micropylar pole, the single megaspore nucleus (Fig. 3G–I) initiates a free-nuclear mitotic division (Fig. 3J–L; see Video 6 in Supplementary data, online) and two nuclei are produced (Fig. 4A–C; see Video 7 in Supplementary data). At the chalazal pole, the three megaspore nuclei (Fig. 3G–I) contribute to a single metaphase plate (Fig. 3J–L) which ultimately produces two presumably triploid restitution nuclei (as opposed to six haploid nuclei; Fig. 4A–C). This results in a second four-nucleate stage of development in which two nuclei are situated at both the micropylar and chalazal poles (Fig. 4A–C). Each nucleus at the micropylar pole contains a single nucleolus, while the two nuclei at the chalazal pole typically have a speckled appearance that makes it difficult to discern individual nucleoli (Fig. 4A–C).
A final round of free-nuclear mitoses produces eight nuclei within the female gametophyte. No eight-nucleate syncytial female gametophytes were observed. It is therefore reasonable to conclude that cytokinesis immediately follows the last free-nuclear mitotic division of female gametophyte development. Following cytokinesis, the female gametophyte contains seven cells: three antipodals, a three-celled egg apparatus, and a binucleate central cell (Fig. 4D–F; see Video 8 in Supplementary data).
Initially, the three cells within the egg apparatus are cytologically and morphologically similar (Fig. 4D, E, G, H) and it is difficult to distinguish between the egg and synergid cells. Polar nuclei within the central cell are visually identical (Fig. 4D), contain a single nucleolus (Fig. 4D) and fuse to yield a secondary nucleus shortly after cytokinesis in the female gametophyte (Fig. 4G–I; see Video 9 in Supplementary data, online). Thus, female gametophyte development in P. peltatum is Fritillaria-type.
Following cellularization, synergids decrease in volume while the egg cell, antipodal cells and central cell become larger (Fig. 4G–I and 5A). Just before fertilization, synergid cells can no longer be recognized and the egg cell has a prominent cell wall with a large nucleus (Fig. 5A). Antipodal nuclei and the secondary nucleus are also larger than at the time of female gametophyte cellularization, and two nucleoli persist in the secondary nucleus (Fig. 5A).
Four female gametophytes in which fertilization appeared to have very recently occurred were observed (Fig. 5B). The zygote nucleus within these female gametophytes contained two nucleoli and the primary endosperm nucleus had three nucleoli (Fig. 5B). Two of these ovules were used to collect microspectrofluorometric data (see below), and both exhibited fluorescence values indicative of fertilized female gametophytes. The central cell remains largely vacuolate through the time of fertilization, and the primary endosperm nucleus rests against a lateral wall of the primary endosperm cell (Fig. 5B). The first division of the primary endosperm cell (Fig. 5C) produces two endosperm cells (Fig. 5D). Starch accumulates in nucellus cells immediately after fertilization.
The relative fluorescence of restitution nuclei, secondary nuclei, zygote nuclei and endosperm nuclei stained with DAPI, a DNA-binding fluorochrome, were quantified to determine DNA content. To calibrate the photometric measurements in P. peltatum zygote nuclei were used during prophase, which by definition have 4C content of DNA. Zygote nuclei had an average RFU value of 102·71 ± 9·36 (n = 5). Therefore, in P. peltatum, 4C content of DNA is equivalent to roughly 100 RFU, 2C content of DNA should correspond to about 50 RFU, and 1C content of DNA should be approx. 25 RFU.
According to previously published data and the authors' description of female gametophyte development, secondary nuclei in mature female gametophytes of P. peltatum are derived from a presumably haploid polar nucleus from the micropylar pole of the female gametophyte and triploid restitution nucleus from the chalazal pole of the female gametophyte. Thus, the secondary nucleus should be tetraploid. Secondary nuclei in interphase averaged 108·24 ± 19·99 RFU (n = 7), suggesting that these nuclei are either diploid and in G2 of the cell cycle, or tetraploid and in G1. A t-test of zygote and secondary nuclei RFU values demonstrated that the difference between these means was not significant (P = 0·54; t-test). Mean relative fluorescence of metaphase plates in dividing nucellus cells, which should also contain 4C DNA content, was 97·04 ± 11·00 RFU (n = 24), and is also not significantly different from prophase zygote (P = 0·27; t-test) and secondary nuclei (P = 0·20; t-test) relative fluorescence values.
If the secondary nucleus is tetraploid, once fertilized, it should become a pentaploid primary endosperm nucleus, and contain 10C DNA during prophase. Mean relative fluorescence of primary endosperm nuclei in prophase was 240·89 ± 38·80 RFU (n = 4). This is almost exactly what would be expected if 1C DNA content is equal to 25 RFU (10C DNA content would correspond to 250C). Therefore, primary endosperm nuclei in Piper peltatum contain 10C content of DNA in prophase, the endosperm is pentaploid, and the secondary nuclei observed in the preparations are tetraploid and situated in the G1 phase of the cell cycle.
The first reports of female gametophyte development in Piper depict a tetrasporic Adoxa-type ontogeny (Fig. 1C) that does not form restitution nuclei (Johnson, 1902, 1910; Palm, 1915). However, the eventual discovery of restitution nuclei and the Fritillaria pattern of female gametophyte development in Liliaceae (Bambacioni, 1928) led some embryologists (Schnarf, 1931, 1936; Maheshwari, 1937) to view previous reports of Adoxa-type female gametophyte development in Piper with a certain degree of scepticism. The key ontogenetic difference between Adoxa-type (Fig. 1C) and Fritillaria-type (Fig. 1G) development is the presence or absence of restitution nuclei at the chalazal pole of the female gametophyte, and all subsequent embryological investigations of Piper species have consistently documented Fritillaria-type female gametophyte development (Maheshwari and Gangulee, 1942; Joshi, 1944; Swamy, 1944; Maugini, 1953; Murty, 1959; Yoshida, 1960; Kanta, 1962; Nikiticheva, 1981; Prakash and Kin, 1982; Prakash et al., 1994). The present data largely agree with these reports; however, important aspects of the megasporogenesis to megagametogenesis transition have been uncovered that serve as the basis for a new understanding of female gametophyte developmental evolution in Piperaceae.
Previous descriptions of nuclear positioning events between meiosis II and the first free-nuclear mitotic division of megagametogenesis in Piper are both three-dimensionally and ontogenetically ambiguous. The present data indicate that immediately following meiosis II, megaspore nuclei in P. peltatum are found in a tetrapolar tetrahedral arrangement (Fig. 3A–C) and, subsequently, migrate to create a tetrapolar cruciate (planar) configuration (Fig. 3D–F). This second four-nucleate ontogenetic stage is followed by migration of the two equatorially placed nuclei (midway along the micropylar–chalazal axis) to the chalazal pole of the female gametophyte (Fig. 3G–I), resulting in a third four-nucleate stage, a bipolar 1 + 3 arrangement.
The tetrapolar tetrahedral and tetrapolar cruciate arrangements of nuclei observed in P. peltatum are similar to the tetrapolar arrangements described in previous investigations (Maheshwari and Gangulee, 1942; Joshi, 1944; Swamy, 1944; Murty, 1959; Yoshida, 1960). There is also reason to believe that the linear megaspore arrangements described by earlier embryologists (Swamy, 1944; Murty, 1959; Kanta, 1962) are actually tetrapolar cruciate arrangements of nuclei that were observed from an angle similar to that shown in Fig. 2F. It was observed that tetrapolar cruciate and even tetrapolar tetrahedral megaspore nuclei could appear linear in serial sections even without the benefit of three-dimensional models.
The present three-dimensional reconstructions confirm the findings of earlier researchers, and add a degree of descriptive structural and ontogenetic precision unavailable in previous investigations. Embryological data regarding later stage developmental aspects of female gametophyte ontogeny in Piper are consistent (Maheshwari and Gangulee, 1942; Joshi, 1944; Swamy, 1944; Maugini, 1953; Murty, 1959; Yoshida, 1960; Kanta, 1962; Nikiticheva, 1981; Prakash and Kin, 1982; Prakash et al., 1994), and with the ambiguity in nuclear positioning at the four-nucleate stages resolved, it is likely that the three-dimensional reconstructions for P. peltatum typify development for Piper as a whole.
Monosporic (Fig. 1A), bisporic (Fig. 1B) and tetrasporic (Fig. 1C–G) patterns of female gametophyte development in angiosperms have distinct patterns of meiotic wall formation that determine how many nuclei are ultimately contributed to the functional megaspore cell. Monosporic female gametophytes form walls after meiosis I and meiosis II, and produce four uninucleate megaspore cells (Fig. 1A). One of these megaspores (usually the chalazal-most cell) will go on to produce the mature female gametophyte (Fig. 1A; Maheshwari, 1950; Davis, 1966; Johri et al., 1992). Bisporic female gametophyte ontogenies form a wall after meiosis I but not after meiosis II (Fig. 1B). Lack of wall formation after meiosis II results in the placement of two nuclei within the functional megaspore cell, and at maturity bisporic female gametophytes contain the mitotic derivatives of two megaspore nuclei (Fig. 1B). Tetrasporic female gametophyte development results from the absence of wall formation during meiosis I and II of megasporogenesis (Fig. 1C–G), as is shown here in Piper (Figs 2 and and33).
Previous reports of megasporogenesis in Piperales (Table 1) were combined with published phylogenetic analyses to infer evolutionary transitions in patterns of megasporogenesis. Unweighted parsimony analyses (Fig. 6) and maximum likelihood analyses (see Fig. S1 in Supplementary data, available online) of wall formation associated with megasporogenesis indicate that the common ancestor of Saururaceae + Piperaceae initiated wall formation after meiosis I (Fig. 6A), but did not initiate wall formation after meiosis II (Fig. 6B). These patterns of meiotic wall formation are consistent with a bisporic pattern of megasporogenesis (Fig. 1B) in the common ancestor of Saururaceae + Piperaceae (Fig. 6C). These results are interesting because they indicate that Polygonum-type female gametophyte development in Houttuynia (Saururaceae) may represent an evolutionary reversal to monospory from a bisporic ancestor (Fig. 6C). An alternative, albeit less parsimonious explanation, is that the common ancestor of Saururaceae + Piperaceae was monosporic, and that bispory in Saururaceae has independently evolved at least twice, in Anemopsis (Saururaceae) and Saururus (Saururaceae). Although current empirical data indicate that the common ancestor of Saururaceae + Piperaceae was bisporic (Fig. 6), future embryological studies of Gymnotheca (Saururaceae) and Verhuellia (Piperaceae) could (depending on what is discovered) provide support for monospory at this node of the Piperales phylogeny.
Unweighted parsimony (Fig. 6) and maximum likelihood analyses (see Fig. S1 in Supplementary data, available online) also indicate that the common ancestor of Piperaceae minus Verhuellia did not initiate wall formation after meiosis I or II (Fig. 6A, B). These data indicate that a tetrasporic pattern of female gametophyte development was present in the common ancestor of Piperaceae minus Verhuellia (Fig. 6C), and are in agreement with previous embryological investigations and comparative surveys in Piperaceae (e.g. Arias and Williams, 2008).
Peperomia-, Penaea-, Drusa- and Fritillaria-type patterns of female gametophyte development have been documented in Piperaceae (Fig. 1 and Table 1); however, the specific developmental modifications that led to evolutionary transitions between each of these tetrasporic ontogenies have never been examined. Both long-standing theoretical predictions (Schnarf, 1936; Maheshwari, 1950; Favre-Duchartre, 1977; Battaglia, 1989; Haig, 1990) and recent comparative analyses (Friedman and Williams, 2003, 2004; Friedman et al., 2008) recognize that all angiosperm female gametophyte ontogenies appear to be structured around the iterative production of nuclear ‘tetrads’, and have speculated that nuclear positioning events early in development may be essential to the generation of downstream developmental and structural diversity. Yet, the specific role that nuclear migration events have played in the radiation of female gametophyte developmental patterns in Piperaceae has never been examined.
The present data document the presence of four distinct four-nucleate ontogenetic stages associated with nuclear migration and division at the end of megasporogenesis in Piper (Figs 22–4). To understand the role these nuclear positioning events have played in female gametophyte developmental evolution, the embryological literature on Piperaceae (Table 1) was surveyed and it was determined which of the ontogenetic stages observed in Fritillaria-type female gametophytes of P. peltatum (Figs 22–4) are present in Peperomia-, Penaea- and Drusa-type female gametophyte ontogenies found in other members of Piperaceae (Fig. 7 and Table 1).
Review of the embryological literature indicates that megaspore nuclei in all female gametophyte ontogenies in Piperaceae conform to a tetrapolar tetrahedral arrangement immediately after meiosis II (Fig. 7). However, only Peperomia-type female gametophytes initiate megagametogenesis directly from a tetrapolar tetrahedral nuclear configuration (Fig. 7). Penaea-, Drusa- and Fritillaria-type female gametophytes transition from a tetrapolar tetrahedral to a tetrapolar cruciate arrangement of megaspore nuclei (Fig. 7). Penaea-type female gametophytes then initiate free-nuclear divisions associated with megasporogenesis (Fig. 7), while megaspore nuclei in Drusa- and Fritillaria-type female gametophytes undergo nuclear migration to create a bipolar 1 + 3 arrangement in which a single nucleus is situated at the micropylar pole of the female gametophyte and three megaspore nuclei are positioned at the chalazal pole (Fig. 7). Drusa-type female gametophytes initiate megagametogenesis from this bipolar arrangement to produce 16 haploid nuclei (four at the micropylar pole and 12 at the chalazal pole), while Fritillaria-type female gametophytes transition through one final four-nucleate stage in which two haploid nuclei are produced at the micropylar pole and two triploid restitution nuclei are produced from three haploid nuclei at the chalazal pole (Fig. 7) as is shown here for P. peltatum (Figs 3 and and44).
This comparative developmental survey makes it apparent that in Piperaceae, female gametophytes initiate megagametogenesis from four-nucleate tetrapolar tetrahedral, tetrapolar cruciate, bipolar 1 + 3 or bipolar 2 + 2 nuclear configurations (Fig. 7). Tetrapolar tetrahedral nuclear placement correlates with Peperomia-type female gametophytes, tetrapolar cruciate nuclear placement correlates with Penaea-type female gametophytes, bipolar 1 + 3 nuclear orientation correlates with Drusa-type female gametophytes, and bipolar 2 + 2 nuclear configuration correlates with Fritillaria-type female gametophytes (Fig. 7).
All female gametophyte ontogenies in Piperaceae include a stage in which megaspore nuclei are tetrahedrally arranged, but tetrapolar cruciate and bipolar 1 + 3 nuclear configurations arise through additional nuclear migration events linked to specific female gametophyte developmental trajectories (Fig. 7). Drusa- and Fritillaria-type female gametophyte ontogenies transition through three distinct four-nucleate ontogenetic stages, but only Fritillaria-type female gametophytes initiate a fourth four-nucleate stage of development as a consequence of the formation of restitution nuclei. Thus, patterns of nuclear migration and division at the four-nucleate stage of development in Piperaceae are predictive of the structure of the female gametophyte at maturity.
Data describing these ontogenetic stages in four-nucleate female gametophytes of Piperaceae (Fig. 7) were combined with published molecular phylogenetic analyses to reconstruct ancestral female gametophyte nuclear positioning events (Fig. 8). These data were then used to infer the evolution of female gametophyte ontogenies throughout Piperaceae (Fig. 9). Embryological data regarding female gametophyte development in Verhuellia do not exist (Table 1), making it impossible to resolve character states in the common ancestor of Piperaceae (Fig. 8). However, ancestral state reconstruction can be resolved in the common ancestor of Piperaceae minus Verhuellia (Fig. 8).
Unweighted parsimony analyses (Fig. 8) and maximum likelihood analyses (see Fig. S2 in Supplementary data, available online) indicate that the common ancestor of Piperaceae minus Verhuellia had a tetrasporic female gametophyte (Fig. 6C) with megaspore nuclei that were initially tetrapolar tetrahedral (Fig. 8A), and then passed through tetrapolar cruciate (Fig. 8B) and bipolar 1 + 3 nuclear configurations (Fig. 8C). Evolutionary analysis clearly indicates that the common ancestor of Piperaceae minus Verhuellia did not form restitution nuclei (Fig. 8D). In total, these developmental characteristics reveal that Drusa-type female gametophyte development (Fig. 7) was present in the common ancestor of Piperaceae minus Verhuellia (Fig. 8), and indicate that Peperomia-, Penaea- and Fritillaria-type ontogenies are evolutionarily derived in this clade (Fig. 9).
It is important to note that female gametophyte development in Manekia has been reported to be variable: female gametophyte development is Drusa-type 86 % of the time, and Penaea-type 14 % of the time. Arias and Williams (2008) conclude that female gametophyte development in Manekia is ancestrally Drusa-type. If this is correct, then unweighted parsimony and maximum likelihood analyses indicate that the common ancestor of Piperaceae minus Verhuellia had a bipolar 1 + 3 configuration of megaspore nuclei (Fig. 7C; see Fig. S2 in Supplementary data, available online). However, if Manekia is ancestrally Penaea-type and nuclear migration to a 1 + 3 arrangement is not plesiomorphic, parsimony and maximum likelihood analyses are less revealing and the evolution of nuclear migration leading to a bipolar 1 + 3 arrangement is equivocal at all nodes of Piperaceae (see Fig. S3 in Supplementary data).
Drusa-type female gametophytes transition through three four-nucleate ontogenetic stages of development (Fig. 7). If this ontogeny is ancestral in the clade Piperaceae minus Verhuellia (Fig. 9), then the evolutionary loss and gain of nuclear positioning events at the four-nucleate stage of development has led to the origin of derived female gametophyte developmental patterns (Fig. 10).
In Peperomia the complete loss of nuclear migration events that would normally lead to bipolar 1 + 3 and tetrapolar cruciate nuclear arrangements in Drusa-type female gametophytes has given rise to Peperomia-type female gametophyte development (Fig. 10). In Piper, developmental alterations associated with the formation of restitution nuclei are appended onto the ancestral Drusa-type developmental framework (Fig. 7), and result in the formation of a fourth four-nucleate ontogenetic stage and a Fritillaria-type female gametophyte. Patterns of cytokinesis in Fritillaria-type female gametophytes are similar to those of Drusa-type female gametophytes (a single nucleus from each pole of the syncytium becomes partitioned into the common cytoplasmic space of the central cell), and developmental modifications leading to the formation of restitution nuclei appear to be the only factors that distinguish Fritillaria- from Drusa-type female gametophyte ontogenies (Fig. 9).
Thus, Drusa-type female gametophytes in Zippelia and Manekia have been inherited from the common ancestor of Piperaceae minus Verhuellia, while Peperomia- and Fritillaria-type female gametophytes in Peperomia and Piper, respectively (Table 1), have evolved through either losses or gains of four distinct four-nucleate ontogenetic stages that have been identified at the end of megasporogenesis (Figs 6, ,88 and and9).9). Interestingly, this evolutionary developmental scenario predicts that the occasional production of Penaea-type female gametophytes in Manekia is likely to be the consequence of the loss of the nuclear migration event that normally leads to a bipolar 1 + 3 arrangement of megaspore nuclei (Fig. 10), a prediction that is in agreement with ideas initially put forth by Arias and Williams (2008).
Fritillaria-type female gametophyte development has evolved at least six times outside of Piperaceae (Joshi, 1938; Maheshwari, 1946; Davis, 1966; Johri et al., 1992), and a brief review of this literature provides some interesting parallels with the conclusions presented in the present study. In Tamaricaceae, Fritillaria-type Myricaria (Schnarf, 1931; Battaglia, 1943) is sister to Tamarix (Lledo et al., 1998; Cuenoud et al., 2002), which produces Adoxa-, Drusa- and Fritillaria-type female gametophytes (Joshi and Kajale, 1936; Sharma, 1939, 1940; Johri and Kak, 1954). Euphorbia (Euphorbiaceae) contains members with Fritillaria-type female gametophytes (Carano, 1925, 1926; Kapil, 1961) and is within the larger clade Euphorbioideae, which is sister to Acalyphoideae (Tokuoka, 2007). Acalyphoideae is exclusively tetrasporic and contains several members with Penaea-type female gametophytes (Tateishi, 1927; Swamy and Balakrishna, 1946; Banerji, 1949; Thathachar, 1952; Johri and Kapil, 1953; Kajale and Murthy, 1954; Mukherjee, 1958; Kapil, 1960). Penaea-type female gametophyte development is also reported within Euphorbia itself, in E. procera (Modilewski, 1909, 1910). Liliaceae is largely Fritillaria-type (Bambacioni, 1928; Bambacioni and Giombini, 1930; Bambacioni-Mezzetti, 1932; Cooper, 1934, 1935; Westergard, 1936; Fagerlind, 1938), but contains members with Drusa- and Adoxa-type female gametophytes in the genus Tulipa (Romanov, 1936, 1938, 1957). Fritillaria-type female gametophyte development is reported in Plumbaginaceae in Limonium (Dahlgren, 1916) and Armeria (Dahlgren, 1916). Armeria and Limonium share a recent common ancestor (Lledo et al., 1998; Cuenoud et al., 2002) and Limonium also contains members that produce Penaea-type and Adoxa-type female gametophytes (Fagerlind, 1938; Davis, 1966).
Thus, five of the seven independent origins of Fritillaria-type female gametophyte development in angiosperms have occurred in clades that are associated with Drusa- and/or Penaea-type female gametophyte ontogenies. The evolutionary developmental hypotheses presented in this study predict that both of these developmental patterns would be antecedent to and/or closely allied with transitions to Fritillaria-type female gametophyte development (Figs 7 and and10).10). The Adoxa-type ontogeny is also commonly associated with clades that include members with Fritillaria-type female gametophytes. An interesting goal for future studies would be to characterize nuclear migration and positioning events in syncytial stages of Adoxa-type female gametophyte development, and determine which (if any) of the ontogenetic modifications that are described in Peperomia-, Penaea-, Drusa- and Fritillaria-type female gametophytes also occur in Adoxa-type ontogenies. Provided that future embryological investigations include detailed information about nuclear migration and positioning events throughout female gametophyte development, the generality of these hypotheses can be tested in other angiosperm lineages, and used to understand the developmental basis of evolutionary transitions to Fritillaria-type female gametophyte development throughout angiosperms.
Supplementary data are available online at www.aob.oxfordjournals.org and consists of the following files. Nine videos are provided of the three-dimensional models that are presented in the text (Figs 2, ,33 and and4):4): in each video, a single three-dimensional model rotates by 360° on the y-axis, and then again by 360° on the x-axis. The colour-coding scheme used for cellular components within each model is identical to the corresponding figure in the text. A file is also available containing the following three figures: maximum likelihood analyses of wall formation associated with megasporogenesis (Fig. S1), nuclear migration (Fig. S2) and nuclear division events (Fig. S3).
We thank Joe Williams for helpful discussions in preparing this manuscript, and the very constructive comments of two anonymous reviewers. This study was sponsored by a grant from the National Science Foundation to W.E.F. (IOB-0446191).