An Anatomically Preserved Cone-like Flower from the Lower Cretaceous of China

Although diverse fossil angiosperms (including their reproductive organs) have been reported from the Early Cretaceous, few of them are well-documented due to poor preservation and limited technologies available to apply. For example, paraffin sectioning, a routine technology applied to reveal the anatomical details of extant plants, was hitherto at most rarely applied to fossil plants. This undermines the comparability between the outcomes of studies on fossil and extant plants, and makes our understanding on plants incomplete and biased. Here, we applied paraffin sectioning technology, in addition to light microscopy, SEM, and TEM, to document a fossil reproductive organ, Xilinia gen. nov., from the Early Cretaceous in Inner Mongolia, China. The anatomical details of this new reproductive organ were documented. Xilinia bears a remarkable resemblance to conifer cones, although its ovules are enclosed in carpels. The paradoxical cone-like morphology of Xilinia appears to represent a transitional snapshot of plant evolution that is absent in extant plants.


Introduction
Angiosperms are by far the most important plant group for humans, as they provide most of the necessary materials for the origin and survival of humans. Naturally, angiosperms are one of the foci for botanical studies. Among all the questions concerning angiosperms, the origin of angiosperms is a core question of key significance since answers for many other questions are hinged upon the answer to this question. Previously, confronting hypotheses were raised to account for the origin and radiation of angiosperms in the Cretaceous, for example, Euanthial theory vs. Pseudoanthial theory, and the phyllosporous school vs. the stachyosporous school. Although Euanthial theory used to be of great influence in the 20th century due to the pioneering work of Arber and Parkin [1], now both the Euanthial and Pseudoanthial theories are out of phase due to the lack of fossil evidence favoring either of them, while the previously confronting phyllosporous and stachyosporous schools seem to have been reconciled by the recently advanced Unifying Theory, which is mainly based on fossil evidence [2]. In the past three decades, paleobotanists have uncovered many interesting discoveries of early angiosperms from the Early Cretaceous (including Chaoyangia [3], Archaefructus [4][5][6][7], Sinocarpus [8,9], Callianthus [10,11], Liaoningfructus [12], Sinoherba [13], and Varifructus [14]), which shed new light on this question [2][3][4][5][6][7][8][9][10][11][12][13][14][15][16][17][18][19]. However, from which group angiosperms were derived remains an unanswered question [20][21][22][23][24], partially due to a lack of anatomically preserved fossils of related plant reproductive organs that can be compared with extant plants. Here, we report a fossil female flower, Xilinia shengliensis gen. et sp. nov, which demonstrates  [26]. (d) Stratum column of the Shengli coal mine showing the horizon yielding our fossil (asterisk). Modified from Shen et al. [26].  Description: Only one specimen (holotype) was collected for our study. The parts of distinct morphologies aggregated in our fossil eliminate the possibility of its being an inflorescence, which is expected to have flowers of relatively uniform morphology. The flower is female, pedicellate, with perianth, up to 23 mm long and 14 mm wide (Figures 2a,b  and 3a). The pedicel is about 3.5 mm in diameter, 11 mm long (Figure 2a-f). The perianth is composed of numerous linear perianth elements that are rhomboidal in cross-view, leaving rhomboidal scars on the axis, 1 mm wide, at least 22 mm long, with simple pits on the walls of tracheids (Figures 2a-g and 4a,b). The gynoecium is ovoid, about 8 mm wide and 9 mm long, with numerous carpels helically arranged on the flower axis (Figures 2a,b, 3a and 5a,f). The axis is about 3.5 mm wide in the bottom, tapering distally, with numerous stubs left by fallen carpels on it (Figures 2a,b, 4e,h and 6a,b). Anatomically, the axis is composed of a central pith surrounded by secondary xylem, phloem, and epidermis (Figure 6a-i). The pith is parenchymatic, about 1.1 mm in diameter (Figure 6a,f-h). Cells in the pith are more or less isodiametric in cross-view, 17-25 × 19-39 µm, elongated longitudinally, 49-100 µm long, some with simple pits on their cell walls and organic infillings in the lumina (Figure 6a,c,f-h). There are two types of organic infillings, solid or spongy, in the cell lumina (Figure 6c,d). The xylem includes isolated primary xylem bundles and a secondary xylem cylinder (Figure 6f-h). The primary xylem is endarch and distributed along the margin of the pith (Figure 6f-h). The secondary xylem forms a ring around the pith, is about 0.7 mm thick, composed of tracheids, cavities, and rays, penetrated by carpel traces (Figure 6a,b,f). The tracheids are about 18 µm wide with simple and bordered pits (Figures 5i,j and 6c,d,g,h). The cavities frequently seen in the early secondary xylem are up to 265 µm long and 115 µm in diameter, probably lysigenous (Figure 6a,b,d-f,i). Rays are uniseriate, 3-8 cells high, and up to 72 µm high (Figure 6e). The phloem and epidermis of the flower axis were hardly discernable in the paraffin sections, probably due to the nitric acid processing, although the epidermis could be seen using SEM (Figure 4e,h). There was no trace of any androecium in the flower (Figure 2a,b). More than 60 carpels are helically arranged along the flower axis, and the traces of these carpels penetrate the xylem cylinder in the flower axis (Figures 2a,b, 3a and 6b,f,g). Carpel size and shape vary depending on their positions in the gynoecium: they are inverted triangular in adaxial and abaxial views, wedge-shaped in side view, 1.7-2.3 mm long, 0.7-1.06 mm wide, and 0.56-0.85 mm thick (Figures 3c, 4c,d,f,g and 5a,f). A carpel comprises an ovary wall and an anatropous ovule within (Figures 1c-h, 3c, 4c and 5f-h). The ovary wall is composed of longitudinally oriented hypodermis and epidermis (Figure 5a,b,d-f,h and Figure 6j). The epidermal cells are isodiametric in surface view, about 7-12 × 11-22 µm in surface view ( Figure 6j). The ovary wall may be up to 35 µm thick, with longitudinal striations on its inner surface (Figure 5d-h). There is no style, and the papillae are restricted to the distal portion of the carpel (Figure 5a-c). The ovule is anatropous, with its micropyle close to the flower axis (Figure 5f-h). The integument encloses nucellus, up to 83 µm thick, composed of radially arranged parenchymatic cells, easily dissolved in nitric acid (Figure 5f-h). The ovular membrane sac is about 1.8 mm long, 1 mm wide, thin, smooth-surfaced, amber in color, tapering distally, of longitudinally oriented cells, becoming thicker at the apex due to additional layer of the integument, enclosed by the integument (Figures 3c-e, 4c, 5f,h, 6k-m and 7a-d). Little content is seen within the ovular membrane; therefore, two layers of the membrane are tightly compressed against each other (Figure 7a-c). Each membrane is about 1.7 µm thick, including three distinct layers, namely, a 0.36 µm thick foot layer, a 0.86 µm thick columella layer, and a 0.66 µm thick tectum layer (Figure 7a-g). The columella layer includes sparse rod-formed vertical structures separated by wide space (Figure 7e-g). The tectum layer covers the columella layer with some stratification (Figure 7e-g). restricted to the distal portion of the carpel (Figure 5a-c). The ovule is anatropous, with its micropyle close to the flower axis (Figure 5f-h). The integument encloses nucellus, up to 83 μm thick, composed of radially arranged parenchymatic cells, easily dissolved in nitric acid (Figure 5f-h). The ovular membrane sac is about 1.8 mm long, 1 mm wide, thin, smooth-surfaced, amber in color, tapering distally, of longitudinally oriented cells, becoming thicker at the apex due to additional layer of the integument, enclosed by the integument (Figures 3c-e, 4c, 5f,h, 6k-m and 7a-d). Little content is seen within the ovular membrane; therefore, two layers of the membrane are tightly compressed against each other (Figure 7a-c). Each membrane is about 1.7 μm thick, including three distinct layers, namely, a 0.36 μm thick foot layer, a 0.86 μm thick columella layer, and a 0.66 μm thick tectum layer (Figure 7a-g). The columella layer includes sparse rod-formed vertical structures separated by wide space (Figure 7e-g). The tectum layer covers the columella layer with some stratification (Figure 7e-g).  Figure 2e showing the pedicel (arrow, pd) and physically connected surrounding perianth elements (pe). Scale bar = 2 mm. (g) Detailed view of the perianth element (pe) arrowed in Figure 2e showing the organic preserved tissue. Refer to Figure 4a,b. Scale bar = 1 mm.          • Etymology: shengli, for the name of the formation from which the fossil was collected. • Holotype specimen: 9222.

Discussion
A strict and sufficient criterion for angiosperms is angio-ovuly, that is, the ovules being enclosed before pollination [2,33]. All plants with their ovules enclosed before pollination are unexceptionally angiosperms. The presence of an ovular membrane with little content in Xilinia suggests that the original content lacked fossilizable materials before fossilization. Considering that the very delicate parenchyma of the integument has been preserved perfectly in the same fossil (Figure 5f-h), such a lack of preserved material within the ovular membrane sac implies that the ovules of Xilinia were still premature, lacking cellularized content when fossilized. Therefore, Xilinia was in its pre-pollination stage when fossilized. The ovule is inside the ovary wall that is integral except for physical cracks caused by desiccation (Figures 2c,d,f,g and 5a,f,g), suggesting that the ovules of Xilinia are fully enclosed by their ovary walls. This enclosure of ovules before pollination is in line with the occurrence of papillae on the carpel tip (Figure 5a-c), which may function as a stigma during the pollination. All the above information collectively points to the fact that Xilinia is an angiosperm, and allows for us to reconstruct it as shown in Figure 8.

Discussion
A strict and sufficient criterion for angiosperms is angio-ovuly, that is, the ovules being enclosed before pollination [2,33]. All plants with their ovules enclosed before pollination are unexceptionally angiosperms. The presence of an ovular membrane with little content in Xilinia suggests that the original content lacked fossilizable materials before fossilization. Considering that the very delicate parenchyma of the integument has been preserved perfectly in the same fossil (Figure 5f-h), such a lack of preserved material within the ovular membrane sac implies that the ovules of Xilinia were still premature, lacking cellularized content when fossilized. Therefore, Xilinia was in its pre-pollination stage when fossilized. The ovule is inside the ovary wall that is integral except for physical cracks caused by desiccation (Figures 2c,d,f,g and 5a,f,g), suggesting that the ovules of Xilinia are fully enclosed by their ovary walls. This enclosure of ovules before pollination is in line with the occurrence of papillae on the carpel tip (Figure 5a-c), which may function as a stigma during the pollination. All the above information collectively points to the fact that Xilinia is an angiosperm, and allows for us to reconstruct it as shown in Figure  8.  Although it is an angiosperm, Xilinia has several features frequently seen in gymnosperms and unexpected for typical angiosperms, including ovular membrane with columellate stratification, lack of a typical perianth, unisexuality, and bordered pits. We discuss these characters and their implications.
The helical arrangement of the carpels in Xilinia is not only like that of carpels in Magnoliales [55], but also like that of lateral cone appendages in Cycadales, Bennettitales, and Coniferales [56][57][58]. Such a feature has been taken as a plesiomorph in angiosperms [59].
Dioecism is a feature frequently seen in gymnosperms, but only relatively rarely seen in angiosperms [56,58,59,63,64]. In this respect, Xilinia appears to be more similar to cones in conifers rather than a typically bisexual flower. This comparison is further underscored by the needle-like morphology of the perianth elements in Xilinia.
The vessel element in xylem is frequently used as an anatomical feature to identify angiosperms, although it occurs in some living Gnetales [65] and fossil Giantopteridales [66,67]. There are cavities in the early secondary xylem in the flower axis of Xilinia. These cavities are up to 265 µm long and 115 µm in diameter, far beyond the dimensions of a regular tracheid and more comparable to a vessel element (Figure 6a,b,e,i). The possibility of them being resin ducts is slim, as resin content has been seen in the same axis, but is lacking in these cavities. Since we did not identify a perforation plate characteristic of vessel elements, and they appeared to not be connected to each other, we are not sure whether they represent vessel elements. If they do represent vessel elements, they might stand for the initial development of vessel elements in which these isolated cavities were yet to develop through lysigeny into a full vessel. Otherwise, the function of such cavities is mysterious.
The presences of simple pits (frequently seen in angiosperms) in the vascular bundle of the perianth element and bordered pits (frequently seen in gymnosperms, but only rarely seen in Monocots, e.g., Dracaena [68]) in the flower axis of Xilinia suggests an affinity swaying between angiosperms and gymnosperms, in line with the cone-like morphology and organization of Xilinia. This explains why we hesitated on placing Xilinia in a specific order or family in angiosperms. In short, Xilinia seems to have not completed its transition from gymnosperms to angiosperms; thus, it bridges these otherwise distinctly separated groups.

Conclusions
Xilinia is a new reproductive organ of a fossil plant from the late Albian, Early Cretaceous. The fossil's anatomical details were well-revealed and -documented due to the application of paraffin sectioning, which was rarely applied on fossil plants before. Xilinia is unique in that it bears a remarkable resemblance to conifer cones, although its ovules are enclosed in carpels. Its cone-like flower morphology appears to represent a transitional snapshot of plant evolution that is now missing in extant plants.