Abstract
Background and Aims The carpel is the definitive structure of angiosperms, the origin of carpel is of great significance to the phylogenetic origin of angiosperms. Traditional view was that angiosperm carpels were derived from structures similar to macrosporophylls of pteridosperms or cycads, which bear ovules on the surfaces of foliar organs. In contrast, other views indicate that carpels are originated from the foliar appendage enclosing the ovular axis. One of the key differences between these two conflicting ideas lies in whether the “axial homologs” are involved in the evolution of carpel. The aim of this paper is to investigate whether the “axial homologs” remain in the carpel of extant angiosperms.
Methods Anaxagorea was used for organogenesis and comparative anatomical study due to its outstanding and unusually long carpel stipe. The continuous change of vascular bundles at carpel stipe in different developmental stages was described in detail.
Key Results Organogenesis study shows that the carpel stipe of Anaxagorea occurs at the early stage of carpel development. Vascular bundles at the base of Anaxagorea carpel are a set of discrete ring-arranged collateral bundles (RACBs), which branch out at the upper portion into two sets of RACBs below each ovule.
Conclusions The ring-arranged collateral bundles indicates a clear morphological evidence for the existence of the “axial homologs” in the carpel and thus support the idea that carpels originated from the integration of the ovular axis and foliar parts. This finding may promote reconsiderations regarding the origin of the carpel and facilitate a greater understanding of the phylogenetic relationship between angiosperms, gnetophytes and Cordaitales.
Introduction
Angiosperms—the flowering plants— make up much of the visible world of modern plants. Both the beautiful flowers and the delicious fruits are closely related to human life. The name “angiosperm” is derived from the Greek words angeion, meaning “container,” and sperma, meaning “seed.” Therefore, the carpel, a “seeds container”, is the definitive structure of angiosperms. The carpel is an angiosperm-specific female reproductive feature in flowers and is the basic unit of the gynoecium, which protectively surrounds the ovules by closure and sealing along their rims or flanks (Dunal, 1817; Robinson-Beers, 1992; Endress, 2015). The evolution of the carpel set angiosperms apart from other seed plants, which developed ovules that are exposed to the air. Since the time of Darwin, elucidating the origin of angiosperms and their evolutionary success represents a primary goal in plant science. Scientists have attempted to search for possible ancestors of angiosperms through phylogenetic analysis based on fossil, morphological, and molecular data. In these efforts, particular emphasis has been placed on assessing the ease with which the ovule-bearing structures of various seed plants could be transformed into a carpel.
Mainstream theories of flower origins provide differing explanations that could be divided into two competing groups based on the assumed organogenetic homologies of the carpel. The “phyllosporous origin” theory suggests that the carpel is a foliar homolog called “megasporophyll” and the ovule either originated directly from the “megasporophyll” or transferred to the “megasporophyll” ectopically (organs borne in an atypical location). Under this scenario, the flower arose from a shoot bearing a series of foliar homologs (e.g., Euanthial theory [Arber and Parkin, 1907; Eyde, 1975]; early version of “Anthophyte” theory [Melville, 1963; Stebbins, 1974; Retallack and Dilcher, 1981; Crane, 1985; Doyle and Donoghue, 1986]; the Mostly Male theory [Frohlich and Parker, 2000]). In contrast, the “stachyosporous origin” theory suggests that the carpel is integrated axial-foliar originated (originated from the integration of the ovular axis and foliar appendage). Various hypotheses supporting the “stachyosporous origin” theory assumed different ancestors for angiosperms. Some hypotheses suggest that the placenta in the carpel is homologous with a female short-shoot of gymnosperms with distally placed ovules, each with scales forming the outer integument. It follows then that the angiosperm carpel is the result of a trend towards the continual reduction in the number of scales and ovules and increased protection of the ovules by the surrounding organs (e.g., Pseudanthial theory [Wettstein, 1907, 1935; Eames, 1952]; Neo-Pseudanthial theory [Nixon et al., 1994; Hickey and Taylor, 1996]; the Unifying Theory [Wang, 2010, 2018]). Another hypothesis holds that the cupule wall of seed ferns provides a homolog for the outer integument of angiosperm ovule and that expansion and folding of the cupule-bearing axis became the progenitor of the carpel (e.g., Caytonialean hypothesis [Doyle, 1978, 2006, 2008]). Despite assumptions for different ancestors of angiosperms, “stachyosporous origin” theory suggests that flowers derived from a compound shoot, that is, from an axis that bears leaf homologs, which in turn produced axillary shoots that bear ovules or microsporangia and the additional leaf homologs. However, it is still difficult to reach an agreement on the origin of flowers regarding these views because of (1) there is a lack of intermediate stages between the extant angiosperm carpel and comparable reproductive structures in fossils, (2) the phylogenetic relationships are uncertain among angiosperms and other seed plant lineages, and most importantly (3) no axial ovules have been found in extant angiosperms from what we know at present.
According to the above-mentioned hypotheses, the current debate focuses on whether or not the “axial homologs” existence in the carpel is of great significance to the origin of the carpel. Following the logic that there is no doubt the carpel wall is derived from “foliar homologs”, as long as the “axial homologs” in the carpel are determined, the carpel should be axial-foliar originated as the “stachyosporous origin” theory suggested. Recently, some vascular anatomy studies (e.g., Actinidia [Guo et al., 2013]; Magnolia [Liu et al., 2014]; Dianthus [Guo et al., 2017]; Michelia [Zhang et al., 2017]) proposed that the vascular system of placenta is independent to that of the ovary wall in the angiosperm carpel. Vascular bundles appear amphicribral (an amphicribral bundle has its xylem surrounded by the phloem) in the placenta, whereas they are collateral (a collateral bundle has adaxial xylem and abaxial phloem) in the ovary wall. Collateral bundles are typical arrangements of vascular bundles in angiosperms (Evert and Eichhorn, 2011). In different organs, the distribution of the collateral bundles is dissimilar. On the cross section of the stem in most angiosperms (other than monocots), discrete collateral bundles form a single ring (hereinafter abbreviated as ring-arranged collateral bundles, RACBs) around the pith. In leaves (other than monocots), the largest collateral bundle often extends along the long axis of the leaf as a midvein, from which smaller veins diverge laterally (Metcalfe and Chalk, 1979; Evert, 2006; Beck, 2010; McKown and Dengler, 2010). Comparatively, amphicribral bundles are frequently seen in small branches of early land plants, in monocots, or in young branches of dicots as simplification and combination of the RACBs (Fahn, 1990). The fact that vascular bundles of the placenta are independent from the ovary wall, and the placenta bundles are amphicribral might provide possibility for the “stachyosporous origin” of the carpel. However, the combination of these two lines of evidence are still insufficient because of the flexibility of vascular bundles, which brings some controversy (Endress, 2019). We believe it is possible to find a more appropriate material from which extant angiosperms show clearer evidence for the existence of “axial homologs” via anatomical study.
In this study, the carpel of extant angiosperm Anaxagorea (Annonaceae) was selected for organogenesis and vascular anatomic examination, aimed at finding dependable “axial homologs” as evidence for the integrated axial-foliar origin of the carpel. Annonaceae are one of the largest families of Magnoliales, which is one of the most important lineages in the early radiation of angiosperms (Sauquet et al., 2003). Anaxagorea is phylogenetically the most basal genus in Annonaceae (Doyle and le Thomas, 1996; Doyle et al., 2004), which live in dimly-lit riparian zones in forest slope understory habitats. In this genus, plants have simple leaves that are alternately arranged in two rows along the stems and the flowers usually have whorled phyllotaxis. Gynoecium are apocarpous (free carpels) throughout the life history (Deroin, 1988). Each carpel of Anaxagorea has a notable long stipe (Endress and Armstrong, 2011), which is one of the important characteristics of the genus. Our investigation is based on the considerations that (1) morphological features shared by Anaxagorea are relatively primitive in angiosperms, (2) the presence of the notable carpel stipe makes it possible to determine whether there are “axial homologs” in the carpel through tissue sectioning.
Materials And Methods
Study species and sites
Flower samples of Anaxagorea luzonensis and Anaxagorea javanica were collected in July 2017 at the Diaoluo Mountain (Hainan, China) and Xishuangbannan Tropical Botanical Garden (Yunnan, China), respectively. The former usually has two to four carpels, whereas the latter has a multicarpellate gynoecium.
SEM and paraffin slices
Flower samples were collected at different floral stages (from early bud to young fruit). Gynoeciums were isolated from the other parts of the flower and preserved in formalin–acetic acid–70% alcohol (5:5:90, v/v). The fixed material was dehydrated through a series of alcohol solutions ranging from 50% to 100%. To delineate the structure and development of carpel, some carpels were removed from the gynoeciums and passed through an iso-pentanol acetate series (SCR, Shanghai, China), and then critically point-dried, sputter-coated with gold, and were observed and photographed using a scanning electron microscope (Tescan VEGA-3-LMU, Brno, Czech). To illustrate the vasculature of carpel, some carpels were embedded in paraffin, sectioned serially into sections of 10-to 12-μm thickness, and then stained with Safranin O and Fast Green. Complete transverse and longitudinal series were examined and photographed under a light microscope (Olympus BX-43-U, Tokyo, Japan) and bright field optics.
Results
Carpel organogenesis
In A. luzonensis and A. javanica, carpels are free from each other completely (Figs. 1G, 2A). Each carpel consists of a plicate zone (Figs. 1G, 3B-H), a very short ascidiate zone (Fig. 2C, here we definite ascidiate zone as a carpel region bearing ovules while having no ventral slits), and a conspicuous long stipe (Figs. 1G, 3I-M). Carpel primordia are approximately hemispherically initiated (Fig. 1A). Carpel stipe forms early at the basal part in the early stage of carpel development (Figs. 1C, J), remains elongate, takes up roughly a quarter length of the carpel at anthesis (Fig. 1G), and continues to elongate during the fruit-stage. Continuous growth of the flanks on the ventral side of the young carpel raises the early closure. The closure does not extend to the bottom of the carpel (Fig. 1D). Subsequently, the dorsal part of each carpel notably thickens and the stigma forms (Figs. 1E, F). At anthesis, carpels are widest at the basal part, with an arch on abaxial side (Fig. 1G). Each carpel has two lateral ovules with the placentae at the base of the ovary (Figs. 2F, M, 3A).
Carpel vasculature
Main vascular bundles of Anaxagorea carpel are distinct as a dorsal bundle, a pair of ventral bundles, a pair of ovule bundles, and several lateral bundles (Figs. 2F, M, 3A, F). Vascular system here are described from the base up to the carpel for easier comprehension of vascular transformation according to the schematic (Fig. 3). At the base of each carpel stipe, discrete collateral bundles form a ring (RACBs) around the pith-like cell groups (Figs. 2A, I, 3M). At the slightly upper part of the carpel stipe, those discrete bundles tend to join adjacent bundles in phloem (Fig. 3L), subsequently separate, and finally are gathered into three groups: a dorsal bundle and two sets of lateral bundle complexes (LBCs, Fig. 3K). Each set of LBCs is assembled into an amphicribral bundle in the young carpel (Fig. 2B). While in the mature carpel, each set of LBCs is assembled as a set of RACBs, and thus there are two sets of RACBs in one carpel (Figs. 2E, K, 3J). Below each placenta, each set of RACBs (or amphicribral bundle) is transformed to a set of “C”-shaped bundle complexes (CBCs), from which several collateral bundles separate from each set of CBCs, and merge with each other to form an amphicribral bundle, which is kept separate from other bundles (Figs. 2C, L, 3G–I). Each set of CBCs allocates an ovule bundle into each ovule and other bundles into the carpel wall, where the ventral bundles are also separated (Figs. 2D, F, M, 3E–F). On the cross section of the ovary locule, both ventral bundles and ovule bundles are amphicribral (Figs. 2G, H). In the carpel wall, lateral bundles present as pinnate arranged collateral bundles, which are connected to form meshy complexes by poorly-differentiated strands (Figs. 2M, 3D, E). The dorsal bundle is also connected with lateral bundles by poorly-differentiated strands (Figs. 2D, 3F). Upon the ovule locule, bundles in the carpel wall join each other with xylem and phloem and finally diminish (Figs. 3B, C).
Discussion
Morphological analysis
Organogenesis study shows that the carpel stipe of Anaxagorea occurs at the early stage of carpel development and continues to elongate with the development of the carpel. It is different from the mericarp stipes of other Annonaceae plants, the latter of which are not obvious until the fruit stage, and partly comes from the elongated receptacle (e.g., Annona, Cananga, Uvari, Xylopia [Deroin, 1999]). The carpel stipe in Anaxagorea provides a buffer space for the drastic changes in structure of the vascular bundle at the base of carpel, whereas in most angiosperms with apocarpy, the base of the carpel is very close to the placenta and the vascular bundle in this part tends to be simplified (e.g., Michelia [Tucker, 1961]; Sagittaria [Kaul, 1967]; Drimys [Tucker, 1975]; Illicium [Robertson and Tucker, 1979]; Brasenia [Endress, 2005]). The RACBs at the basal cross sections act in accordance with a typical form of the stem vascular anatomy of most angiosperms other than monocots (e.g., Caprifoliaceae; Leguminosae; Ulmaceae). In the distal region, a single collateral bundle separated from the basal RACBs, showed foliar homology. The remainder gathered into two sets of RACBs below each placenta, which are especially obvious in the more mature carpel, and showed axial homology. As we know, the vascular distribution in the leaves are lateral organized in most seed plants (e.g., Pinaceae; Gnetaceae; Magnoliaceae; Gramineae; Brassicaceae). Even in the branching veins in Ginkgo, each of the veins is a single collateral bundle. The topological structure of the lateral organized bundles is equivalent to a curve, and the curved surface can only form a one-way tunnel (like macaroni pasta) by coiling. However, the branching RACBs (one set of RACBs branch into two sets of RACBs) in the carpel of Anaxagorea are equivalent to a three-way tunnel. The branching of RACBs in each carpel seems to be reliable evidence for the existence of “axial homologs” in the carpel because the foliar homologs (lateral organs) cannot form such a topology by coiling transformation.
In the carpel of Anaxagorea, amphicribral bundles were observed in the young carpel stipe and in the funiculus of the mature ovule as simplification and combination of the RACBs. The fact that the amphicribral bundles are widespread in the carpel and funiculus of angiosperms (e.g., Papaver [Kapoor, 1973], Psoraleae [Lersten and Don, 1966], Drimys [Tucker, 1975], Nicotiana [Dave et al., 1981], Whytockia [Wang and Pan, 1998], Pachysandra [Von Balthazar and Endress, 2002], Actinidia [Guo et al., 2013], Magnolia [Liu et al., 2014]; Dianthus [Guo et al., 2017]; Michelia [Zhang et al., 2017]) would be a reasonable result if the ovule-placenta in the carpel is originated from the ovular axis. In addition, we should notice that a carpel has developed as a functional entity, and there is no axis or axillary shoot in the carpel. The words “axial homologs” we use are mainly to emphasize the possible phylogenetic origination of the carpel.
Homology Comparison
The existence of “axial homologs” in carpels might facilitate understanding of the phylogenetic origin of angiosperms by homology comparison. In outgroups, the female reproductive shoots, which are similar to the branching RACBs-placenta complex in the carpel of Anaxagorea, can be found in Ephedra and the extinct Cordaitales-like plants (Fig. 4).
Ephedra is an extant gymnosperm representing gnetophytes. Neo-Pseudanthial theory suggested that the angiosperm carpel arose from the foreshortening and integration of the female reproductive shoot of Ephedra (Hickey and Taylor, 1996). Our results suggest that the carpel vasculature of Anaxagorea has branching RACBs and is comparable with the female reproductive shoot of Ephedra. If the branching RACBs are the plesiomorphy of the common ancestor of angiosperms and Ephedra, then it is more plausible that the carpels have stronger homology with the female reproductive shoot than the cupule-bearing axis in Caytonia (Doyle, 1978, 2008) because the cupule-bearing axis of Caytonia does not bear the secondary shoot. Despite the phylogenetic hypothesis that Ephedra are the closest living relatives of angiosperms in the Neo-Pseudanthial theory, this postition has currently been refuted (Hajibabaei et al., 2006; Wickett et al., 2014; Wan et al., 2018). However, we should be careful not to discard the homology comparison as an independent component of the theory.
Other views have provided a relatively moderate explanation that angiosperms, conifers, and gnetophytes (Ephedra, Gnetum, Welwitschia) are derived from a common ancestor in Cordaitales (Eames, 1952; Bold, 1973; Hickey and Taylor, 1996; Doyle, 2008; Wang, 2010, 2018). The Unifying Theory (Wang, 2010, 2018) suggests that angiosperm carpels are derived from the sterile bracts (formed carpel wall) attached to the reproductive secondary shoot (formed ovule/placenta) of Cordaitales. Taking Cordaitales as the common ancestor of angiosperms and other gymnosperms may resolve the contradiction between the similar morphological characters and the distant phylogenetic relationship between angiosperms and gnetophytes. The branching RACBs of Anaxagorea carpels suggest that the carpel might comparable to a compound shoot and provide an understanding of how the bract-bracteole-terminal ovule system in Cordaitales-like gymnosperms evolved to an angiosperm carpel.
Our results provide more reliable evidence for the existence of “axial homologs” in the carpel of extant angiosperms. An “axial homolog” should not appear in the “megasporophyll”. However, if that is true, angiosperm carpels should be integrated axial-foliar originated. Nature follows the principle that selecting a simple mechanism or module as building blocks for a complex system and then using that module repeatedly in other systems. The presentation of “axial homologs” in carpel seems the embodiment of this rule.
Funding
This work was supported by the National Natural Science Foundation of China [grant number 31270282, 31970250].
Acknowledgements
We thank Chun-Hui Wang, Shi-Rui Gan, and Yan-Lian Qiu for their help in searching for the target species in the field, to Qing-Long Wang for his help to take care of the transplanting materials in Hainan province, and to Lan-Jie Huang and Ke Li for their advice on writing.