Reinvestigating an enigmatic Late Cretaceous monocot: morphology, taxonomy, and biogeography of Viracarpon

Introduction

During the Cretaceous period (∼145–66 Ma) angiosperms (flowering plants) underwent a rapid evolutionary radiation, first appearing in the fossil record around 140 million years ago and rising to geographic and ecological dominance in terrestrial ecosystems by the end of the Cretaceous (McElwain, Willis & Lupia, 2005; Berendse & Scheffer, 2009; Herendeen et al., 2017). The origin and diversification of angiosperms has been studied intensively by evolutionary biologists (Crane, Friis & Pedersen, 1995) but many questions on the timing of evolutionary events and relationships between lineages remain unresolved, particularly for monocots, which comprise approximately one-fifth of extant angiosperm species (Friis, Pedersen & Crane, 2004). The fossil record is especially important for addressing these questions because fossils provide minimum ages for lineages, empirical data on past diversity, and evidence of character combinations not present in extant taxa (Crepet, Nixon & Gandolfo, 2004; Smith, 2013). Pollen and macrofossils indicate an Early Cretaceous origin for monocots and subsequent diversification during the Late Cretaceous (Friis, Pedersen & Crane, 2004; Crepet, 2008; Smith, 2013). However, owing to their sparse fossil record during this time, details of the Late Cretaceous diversification of monocots are poorly understood (Crepet, 2008) and new discoveries are essential for shedding light on the tempo and mode of monocot evolution.

The fossil floras of the Maastrichtian–Danian (∼72–61 Ma) Deccan Intertrappean Beds of India are an important source of data for understanding angiosperm evolution and historical biogeography. Hundreds angiosperm taxa have been documented, roughly half of which are monocots (Kapgate, 2009; Srivastava, 2011; Wheeler et al., 2017), providing a window into monocot diversity during the Late Cretaceous when India was geographically isolated in the southern hemisphere tropics (Ali & Aitchison, 2008; Chatterjee, Goswami & Scotese, 2013). Moreover, connections with Gondwanan landmasses during the Jurassic, millions of years of geographic isolation following continental breakup, and subsequent physical connection with Eurasia for most of the Cenozoic raise the possibility that India had an important role in the biogeographic history of plant groups. Possible scenarios include high endemism stemming from geographic isolation, or “out-of-India” dispersal of Gondwanan lineages into Eurasia (Morley, 2003; Karanth, 2006; Whatley & Bajpai, 2006; Manchester, Kapgate & Wen, 2013). The excellent preservation of many of these permineralized fossils facilitates resolution of key morphological characters and makes precise taxonomic identifications possible (Manchester, Kapgate & Wen, 2013; Manchester et al., 2016), which are necessary for answering broader questions of evolution and biogeography. However, despite over a century of studies on the flora some of the taxa have eluded attempts to determine precise systematic relationships. One of these enigmatic plants is Viracarpon, a monocot found abundantly in the Deccan Intertrappean Beds.

Viracarpon comprises compact infructescences bearing sessile, six-seeded fruits, known from at least eight separate localities of the Deccan Intertrappean Beds (Kapgate, 2009), as well as in underlying sediments of the Lameta Formation (Samant & Mohabey, 2009). “Mulberry-like fruits” were first mentioned by Hislop (1853) and the fossils were later formally described as a new monocot genus, Viracarpon, by Sahni (1934). Four species have been named: Viracarpon hexaspermum Sahni (1934), Viracarpon elongatum Sahni (1944), Viracarpon sahnii Chitaley, Shallom & Mehta (1969), and Viracarpon chitaleyi Patil (1972). The presence of scattered vascular bundles in the peduncle and trimerous floral architecture indicate affinities with monocotyledons, and relationships with the Araceae, Pandanaceae, Cyclanthaceae, and Scheuchzeriaceae have all been suggested (Sahni, 1944; Chitaley, 1954; Verma, 1958; Nambudiri & Tidwell, 1978). However, the complex morphology of the fruits and unusual combination of characters have led to considerable disagreement over synonymy of species, interpretation of morphology, and the familial placement of the genus. Moreover, the unusual morphology of Viracarpon and lack of clear taxonomic affinities suggest that it may provide an example of a genus that was endemic to India, reflecting the prolonged isolation of the subcontinent.

We reinvestigated Viracarpon using light microscopy and X-ray micro-computed tomography (μCT) to reconstruct the complex morphology of fruits and infructescences. The reconstructions help to resolve conflicting interpretations among earlier researchers and reveal new characters necessary for determining systematic affinities. Based on these results, we provide here a revised taxonomy and emended descriptions of the species. Further, the hypothesis that Viracarpon is endemic to India is rejected based on our reevaluation of Coahuilacarpon phytolaccoides infructescences from the late Campanian of Mexico. Viracarpon provides new data on the Cretaceous radiation of monocots and its reevaluation represents an important step toward using these data to address broader questions on angiosperm evolution.

Materials and Methods

The Deccan volcanic province (DVP) comprises a sequence of continental flood basalts (traps) formed during the late Maastrichtian to early Danian (67–64 Ma, chrons 30N–29N; Hooper, Widdowson & Kelley, 2010; Renne et al., 2015; Schoene et al., 2015), which crops out across central and western peninsular India. Intertrappean sedimentary layers, deposited during quiescent intervals between volcanic eruptions, contain abundant three-dimensionally preserved plants with cellular details documented from approximately 57 different exposures (Kapgate, 2009). Although the precise ages of many fossil localities are poorly constrained, most can be broadly considered either late Maastrichtian or early Danian depending on their location within the DVP and stratigraphic continuity with dated outcrops (Srivastava & Guleria, 2006; Smith et al., 2016).

Specimens of Viracarpon have been recovered from eight different localities: Mohgaonkalan, Keria, Phutala Tank, Mahurzari, Dongargaon, Takli, Singpur, and Marai Patan (Sahni, 1934; Chitaley, Shallom & Mehta, 1969; Nambudiri & Tidwell, 1978; Kapgate, 2009; this study). Within the state of Maharashtra, the Phutala Tank, Mahurzari, and Takli localities are part of the Nagpur district, while Dongargaon and Marai Patan are in the Chandrapur district. These localities are within the northeastern edge of the Deccan main plateau. Mohgaonkalan, Keria, and Singpur are in the Chhindwara district of Madhya Pradesh, and are in the southwestern portion of the Mandla lobe of the DVP. Most localities are found in or around small villages for which they are named, and more precise directions and descriptions are given by Kapgate (2009). The intertrappean beds at Takli, the type locality for Viracarpon, were located within the city of Nagpur but have since been completely excavated by building construction and are no longer accessible.

Chronostratigraphic investigations of the Phutala Tank, Mahurzari, Dongargaon, Takli, and Marai Patan localities have not been conducted to determine their age as either late Maastrichtian or early Danian. However, the presence of palynofloras containing Maastrichtian indicator taxa at the Singpur locality (Samant, Mohabey & Kapgate, 2008), as well as Maastrichtian vertebrates at the Naskal locality (ca. 100 km southwest of Nagpur; Rage, Prasad & Bajpai, 2004), suggest that a late Maastrichtian age for these localities is likely. The Mohgaonkalan locality is considered late Maastrichtian in age, based on evidence from palynology, vertebrate fossils, and magnetostratigraphy indicating deposition during chron 30N (Samant & Mohabey, 2009). Keria is part of the same intertrappean as Mohgaonkalan, and is also considered late Maastrichtian (B. Samant, 2017, personal communication). All current information on the age of the paleobotanical localities indicate that the Indian Viracarpon fossils are most likely late Maastrichtian in age.

Specimens of C. phytolaccoides were loaned by Dr. Sergio Cevallos-Ferriz (Universidad Nacional Autónoma de México (UNAM)). They were originally collected from the Cerro del Pueblo Formation in the southwestern region of the state of Coahuila in Mexico (Cevallos-Ferriz, Estrada-Ruiz & Perez-Hernandez, 2008). The Cerro del Pueblo Formation is late Campanian in age (72.3–71.3 Ma) based on ammonite biostratigraphy and magnetostratigraphy (Kirkland et al., 2000; Cevallos-Ferriz, Estrada-Ruiz & Perez-Hernandez, 2008). Specimens are curated at the Colección Nacional de Paleontología, Instituto de Geología, UNAM, in Mexico City, Mexico.

Viracarpon specimens examined in this study included petrographic thin sections, peels (cellulose acetate or butyl acetate), and intact specimens curated at the Florida Museum of Natural History (FLMNH), University of Kansas Biodiversity Institute (KU), Cleveland Museum of Natural History (CMNH), Natural History Museum London (NHMUK), and the Birbal Sahni Institute of Paleobotany (BSIP) in Lucknow, India. Figured specimens include those accessioned at NHMUK (V35059, V35038, V35036), CMNH (P24742, P3771, slide 4305), FLMNH (UF19442-70248), and BSIP (9349, 9351). NHMUK specimens originate from the Takli locality. Figured specimens from BSIP and CMNH were collected from Mohgaonkalan. One specimen of V. elongatum (UF19442-70248) was collected from the Marai Patan locality. Viracarpon fossils comprise whole infructescences permineralized in chert and preserved to the tissue and sometimes cellular level, with little or no compression of organs. The completeness of specimens and preservation of fine structures such as spines and trichomes indicates that most infructescences experienced relatively little transport prior to fossilization. Specimens of Coahuilacarpon comprise whole infructescences preserved by iron silicate permineralization (Rodríguez-de la Rosa & Cevallos-Ferriz, 1994), and exhibit varying degrees of compression. However, the specimen figured here is more or less uncompressed.

In order to accurately reconstruct the three-dimensional (3D) morphology of Viracarpon and non-destructively study valuable museum specimens, many of the fossils were studied using μCT (Smith et al., 2009; Collinson et al., 2012, 2016). In a μCT scan, X-rays are emitted by a microfocus X-ray tube and pass through a rotating sample to an X-ray detector, which takes a series of two-dimensional (2D) radiographic projection images as the sample rotates. Denser materials absorb more of the X-ray photons, allowing fewer of them to reach the detector. In reconstructed μCT scans, for which the internal structural information has been computed from the projection images, denser materials are represented by lighter gray values while less dense ones are darker. All figured specimens, except those from NHMUK, were scanned at Nikon Metrology (Brighton, MI, USA) and the University of Michigan CTEES facility with a Nikon XT H 225ST industrial μCT system using a tungsten reflection target. Specimens curated at NHMUK were scanned using a Nikon HMX ST225 μCT system, at the NHMUK Imaging and Analysis Centre. Depending on the specimen, scans were set at 100–180 kV, 120–300 μA, and used 0–2 mm of copper filter, which reduces strong artifacts in reconstructed images by suppressing lower energy X-rays. Pixel size varied from ca. 14–20 μm. All scans were acquired using Inspect-X and reconstructed using CT Pro 3D (Nikon Metrology, USA), which uses a FDK (Feldkamp–Davis–Kress) type algorithm. This software takes the 2D projection images acquired by the X-ray detector and generates a 3D image represented by gray values that are distributed in a volumetric space. Reconstructed datasets were analyzed and 3D models were produced using Avizo 9 Lite 3D software (FEI, Hillsboro, OR, USA). We refer to sections obtained from the reconstructed μCT data as “digital sections.” Volume renderings were produced using both the “volren” and “volume rendering” modules in Avizo, while 3D models were done by manual segmentation of μCT scans. 3D models were converted to U3D format using MeshLab (http://meshlab.sourceforge.net/) for generating 3D PDF files using Adobe Acrobat Pro IX (Adobe, San Jose, CA, USA; Fig. S1). Figures were constructed using CorelPhoto-Paint X8 and CorelDraw X8, and supplemental videos produced using Corel VideoStudio X10 (Corel Corporation, Ottawa, ON, Canada).

Raw μCT data (projection files, volumes, and image stacks) for figured specimens curated at NHMUK are archived by the museum Imaging and Analysis Centre (IAC). Archived data can be accessed by contacting staff at IAC (http://www.nhm.ac.uk/our-science/departments-and-staff/core-research-labs/imaging-and-analysis-centre.html). Additional video files of figured specimens V35059 and V35036 are available in the Supplemental Information (Videos S3–S6). Raw μCT data (tiff stacks) for figured CMNH, FLMNH, and UNAM specimens are available publicly on MorphoSource (Duke University; https://www.morphosource.org/Detail/ProjectDetail/Show/project_id/430) under project number P430 (title: Viracarpon, an enigmatic monocot from the Late Cretaceous). Additional video files and surfaces (.ply format) for these specimens are also available in the MorphoSource project.

Results

• Systematic paleobotany

• Division—Magnoliophyta

• Class—Liliopsida

• Family—incertae sedis

• Genus—Viracarpon Sahni, Proc. 21st Ind. Sci. Cong. 21: 318. 1934.

Emended generic diagnosis—Infructescence compact, subglobose to elongate; fruits densely inserted, sessile, arranged helically or in whorls; peduncles slender, with scattered vascular bundles; gynoecia composed of six single-seeded carpels, semicarpous; ovaries fused centrally, lateral walls of adjacent carpels free or partially fused, styles free; each ovary with prominent central axis; six vascular bundles per fruit in central axis, each bundle opposite of and extending over top of locule distally; placentation apical, each seed vascularized by single bundle entering from distal end of locule.

• Type Species—Viracarpon hexaspermum Sahni emend. Matsunaga, S.Y. Smith, & Manchester.

• Basionym: Viracarpon hexaspermum Sahni, Proc. 21st Ind. Sci. Cong. 21: 318. 1934.

• Synonyms: Viracarpon sahnii Chitaley, Shallom & Mehta, J. Sen Mem. Vol. Bot. Soc. Bengal: 333. 1969.

• Holotype: V35038 (NHMUK; Fig. 1D), Sahni, 1944, Figs. 25–28.

• Other specimens studied: V35059 (NHMUK; Figs. 1B, 1C, 1E, 2C and 3D–3I), V35056 (Figs. 2A and 2B), V35038 (Figs. 2D and 2E), CMNH P24742 (Figs. 1I, 1H, 2F and 3A–3C).

• Type locality, stratigraphy, and age: Takli—Deccan Intertrappean Beds, India, Late Maastrichtian–early Danian.

• Other occurrences: Mohgaonkalan, Mahurzari, Phutala Tank, Singpur, Marai Patan—Deccan Intertrappean Beds, India, late Maastrichtian–early Danian.

Emended specific diagnosis—Infructescences globose to ellipsoid, up to 3.0 cm long and 2.0 cm wide, fruits helically arranged; peduncle wider in region of fruit attachment, containing three to four cycles of vascular bundles; outer vascular bundles smaller than innermost ones; fruits (excluding styles and perianth elements) up to 5.0 mm wide and 5.0 mm long, size variable between specimens; lateral walls of adjacent carpels in ovaries free; styles persistent, spine-like, forming prominent ridges on distal surface of fruit, fused to perianth basally; six channels extending down from top of fruit between adjacent carpels; six inner perianth elements, persistent, fused to ovary and style bases, tapering into sinuous spine-like structures; adaxial surface of styles and perianth densely trichomatous; fruit vascularized by six bundles extending from the base of the fruit into the styles.

Description:

Infructescence structure: Infructescences of V. hexaspermum are generally ca. 1.0–2.0 cm wide and up to 3.0 cm tall (Figs. 1A–1D). The peduncle is slender, 3.0 mm wide, and is expanded apically to about 6.0 mm wide where fruits are attached (Figs. 1B and 1C). Fruits are sessile and exhibit a compact helical arrangement, with little or no space between adjacent fruits, forming ca. 8–12 orthostichies (Figs. 1A–1H). Reduced or abortive fruits are present near the base and at the very apex of infructescences (Fig. 1A, arrows).

Fruit structure: Most fruits range from about 4.0–6.0 mm wide and 5.0 mm tall, excluding the spines attached to each fruit (Figs. 1E and 1F). The gynoecia are semicarpous, composed of six single-seeded carpels fused centrally and forming a thick axis around which the locules are arranged (Figs. 1E and 1G). This central fusion extends the entire length of the ovary. Lateral walls between adjacent carpels are not fused but are often closely appressed (Figs. 1E and 1G). Seeds are usually poorly preserved, represented by a very thin seed coat with a lenticular cap of tissue on the distal end, which may represent a hypostase (Figs. 2D–2F). No convincing embryo or endosperm tissues have been observed, and seed coat preservation is too poor to determine thickness or histology. The pericarp is thin, mostly sclerenchymatous, and the fruits are densely packed in the infructescence, indicating they were probably not very fleshy at maturity. Detailed anatomical descriptions were given by Chitaley (1954), but the cellular preservation of specimens studied here is generally poor and it is difficult to confirm the histological observations made by previous researchers. However, in some specimens grazing sections through the pericarp show that the innermost cell layer, which lines the locule, is comprised of elongate cells with sinuous walls, approximately 40 μm wide and 200 μm long (Fig. 2G). This anatomy was also documented by Chitaley (1954), who also noted a layer of fibers to the inside of the locular epidermis. The distal end of the fruit possesses a sclerenchymatous tissue that forms a stellate pattern of ridges, which radiate from a central depression. Each ridge forming a spine that extends distally from the outer edge of the fruit (Figs. 3B–3E; Video S1). The position of the spines above each locule, as well as patterns of vascularization (described below) suggest that these spines are persistent styles. No dehiscence structures, germination sutures, or opercula have been observed.

Vasculature: Transverse sections through the peduncle reveal three to four cycles of scattered vascular bundles arranged around a narrow central pith (Figs. 4A and 4B). Vascular bundles tend to be larger toward the center of the stem both below and within the zone of fruit attachment, and thus the smaller peripheral bundles are not vascular traces supplying individual fruits (Figs. 4A and 4B). Preserved vascular bundles are composed fibers that either surround the vascular tissues or form two caps adjacent the xylem and phloem. Xylem tracheary elements are sometimes preserved (Chitaley, Shallom & Mehta, 1969), but phloem anatomy was not preserved in our specimens and has not been documented by other authors. Individual fruits are vascularized by six bundles, which run through the central axis of the fruit, positioned opposite the locules (Figs. 4C and 4D). In digital μCT sections these bundles appear white or light gray, contrasting with the darker and less dense organic material of the preserved plant tissues (Fig. 4C). This probably results from infilling of the conducting tissues with the dense siliceous matrix in which the fossils are preserved. Similarly, in the vascular bundles of the peduncle the fiber caps are well-preserved, but the conducting tissues are not and exhibit gray values similar to those of the matrix (Fig. 4B). Each bundle in the fruit vascularizes a seed as well as the corresponding spine. The bundle extends up through the central axis of the fruit and curves laterally over the top of the locule where it bifurcates, with one strand immediately entering the locule wall and the other continuing apically into the spine (Figs. 4E–4K). This is most easily seen in μCT videos through successive digital slices that follow the trajectory of the bundles through the fruit (Figs. 4G–4K; Video S2). The strand that continues into the spine is wider than the original bundle (Figs. 4E and 4K), but resolution in μCT scans is insufficient to ascertain whether it composed of a single strand or of multiple closely spaced strands. The consistent vascularization of each locule wall at the distal end indicates apical placentation of ovules. This is congruent with the apical attachment of seeds within the locules observed by Chitaley (1954). Although nearly all monocots with apical placentation have orthotropous ovules (Dahlgren, Clifford & Yeo, 1985), preservation is insufficient to confirm this ovule orientation.

Perianth: Each fruit has six perianth elements that are fused to both the ovary wall and the bases of the persistent styles (Fig. 3). Above the ovary, these perianth elements are laminar for several millimeters before tapering into long, slender, spine-like structures that run parallel with the persistent styles (Figs. 3A and 3F–3I). In many specimens both the persistent style and the adaxial surface of the perianth are covered densely in long, sinuous trichomes (Figs. 1H and 1I). The absence of trichomes in some specimens could reflect either differential degradation or intraspecific variation.

Other structures: Two other notable features of the fruit were observed. The first is the presence of six hollow, tapering channels in the fruits positioned between the bases of each style where they diverge from surface of the fruit (Figs. 3B–3F). These channels run approximately one-third of the way down the length of the fruit and are aligned with, but not connected with, the gaps between the partially fused carpels (Figs. 2A–2C and 3F). The second feature consists of 12 depressions along the perimeter of the distal surface of each fruit (Figs. 3C and 3E). Each style is flanked by two such depressions. It is unclear what these two features represent. The position of the six channels suggests they could be the remnants of a septal nectary, which are common in some monocots (Rudall, 2002), but no anatomical information to support this is preserved. They may alternatively be additional gaps between carpels, possibly formed during fruit maturation. The 12 depressions may be formed incidentally by the fusion of the style and perianth in that region, but it is alternatively possible that they represent the attachment points of stamens. Chitaley (1954) described vascular bundles in the pericarp below these depressions. However, no direct evidence of stamens or pollen have been observed in any specimens, and preservation of specimens studied here is insufficient to confirm whether the depressions are vascularized.

• Species—Viracarpon elongatum Sahni ex Chitaley & Patil emend. Matsunaga, S.Y. Smith, & Manchester.

• Basionym: Viracarpon elongatum Chitaley & Patil, Botanique 2: 44–46. 1971.

• Synonyms: Shuklanthus superbum Verma, Jour. Palaeon. Soc. India 3: 196. 1958.

• Viracarpon chitaleyi Patil, Botanique 3: 25–26. 1972.

• Lectotype: Specimen No. 3 Moh/P-5 (Institute of Science Nagpur), Chitaley & Patil, 1971, Plate 1, Fig. 1. Lectotype and other figured specimens have been lost.

• Neotype: The neotype is designated here as specimen P3771 (CMNH; Fig. 5A, 5C, 5D, 5G and 5I).

• Additional material: FLMNH UF19442-70248 10 (Figs. 5B, 5E and 5H).

• Type locality, stratigraphy, and age: Mohgaonkalan, Deccan Intertrappean Beds, India, Late Maastrichtian.

• Other occurrences: Takli, Marai Patan—Deccan Intertrappean Beds, India, Late Maastrichtian–early Danian.

Emended specific diagnosis—Infructescences slender, elongate, up to 6.5 cm long and 0.5 cm wide; peduncle of uniform width throughout, with two to three cycles of vascular bundles; fruits in whorls of three, up to 2.5 mm wide and 2.0 mm long; lateral walls of adjacent carpels in ovary free; distal surface of fruit with six shallow ridges; perianth in two whorls, outer perianth whorl free, inner whorl partially fused to ovary.

Description:

Infructescence structure: V. elongatum forms slender, elongate, spike-like infructescences up to 5.0 mm wide and 6.5 cm long, with fruits forming five to six orthostichies (Figs. 5A and 5B). Unlike V. hexaspermum the peduncle is not expanded where fruits are attached and is instead a uniform diameter throughout, ca. 1.5 mm. Fruits are sessile and organized into alternating whorls of three (Fig. 5I), although one specimen was originally described as having helical arrangement (Patil, 1972).

Fruit structure: Fruits are relatively small, measuring 2.0–2.2 mm in diameter and 1.3–1.5 mm tall, but exhibit a similar gynoecial morphology consisting of six single-seeded carpels fused centrally and with the lateral carpel walls free (Fig. 5E). Seeds, when preserved, have a thin seed coat, show no evidence of the distal lenticular structure seen in the seeds of V. hexaspermum, and are often colonized by septate fungal hyphae (Fig. 5F). Preservation is insufficient to determine seed coat thickness or histology. No projecting structures are seen above the ovary such as stylar spines or perianth elements (Fig. 5C). However, the upper surface of each fruit has six raised ridges positioned medially on top of each carpel, four of which sit slightly higher than the other two. A shallow depression is present in the center of the ridges, directly above the axis of the fruit (Fig. 5G and 5H). The pericarp is thin, approximately two to three cell layers thick where it can be distinguished from the inner perianth, and sclerenchymatous (Figs. 5E and 5F). No clear evidence of germination sutures or dehiscence structures have been observed.

Vasculature: The peduncle contains usually two to three cycles of vascular bundles arranged irregularly in the cortex, which is parenchymatous (Figs. 5I and 5J). Vascular bundles are collateral, with fibers forming caps to the inside and outside of the bundle, or sometimes surrounding the bundle completely. Xylem tracheary elements are preserved in some specimens, consisting of ∼8–16 cells measuring 10–25 μm in diameter. Phloem is not preserved (Fig. 5J). Each fruit is vascularized by six bundles that run through the central axis and curve over the top of the locules, corresponding to each ridge on the surface of the fruit (Fig. 5E). However, it is not clear whether the bundle vascularizes the seed as it does in V. hexaspermum and Viracarpon phytolaccoides.

Perianth: V. elongatum possesses an outer perianth whorl of six elements, free from and arranged opposite each carpel (Figs. 5C and 5D). Additionally, at the distal end of each carpel is an outer layer of sclerenchymatous tissue that is free from the ovary and the ridges on top of each fruit (Figs. 5C–5H). This layer may be derived from an inner perianth whorl partially fused to the ovary wall like that of V. hexaspermum, as has been suggested by previous investigators (Patil, 1972).

• Species—Viracarpon phytolaccoides (Cevallos-Ferriz, Estrada-Ruiz & Perez-Hernandez) Matsunaga, S.Y. Smith, & Manchester comb. nov.

• Basionym: Coahuilacarpon phytolaccoides Cevallos-Ferriz, Estrada-Ruiz & Perez-Hernandez, Am. J. Bot. 95: 79. 2008.

• Holotype: IGM-PB 1244 (Universidad Nacional Autonoma de Mexico).

• Other specimens studied: One additional specimen was studied by μCT (IGM-PB 9884, Fig. 6).

• Stratigraphy and age: Cerro del Pueblo Formation, late Campanian (72.3–71.3 Ma).

• Emended specific diagnosis—Infructescences up to 4.5 cm long and 2.5 cm wide, with helically arranged fruits; peduncle wider in region of fruit attachment, with scattered vascular bundles in several cycles; outer vascular bundles smaller than innermost ones; fruits up to 7.0 mm wide and 8.0 mm long; vascular bundles of fruit trifurcating above locule, one bundle entering seed, two bundles entering style.

Description:

Infructescence structure: V. phytolaccoides forms compact infructescences, 2.0–4.4 cm long by 1.5–2.3 cm wide (Figs. 6A–6D; Cevallos-Ferriz, Estrada-Ruiz & Perez-Hernandez, 2008), oblong to ellipsoidal in shape, and bearing helically arranged fruits in usually eight orthostichies (Fig. 6D). The infructescence axis is approximately 6 mm in diameter where fruits are attached and tapers to ∼2.5 mm at the proximal end, just below fruit attachment (Fig. 6C).

Fruit structure: Fruits of V. phytolaccoides are up to 7.5 mm wide distally, narrowing slightly toward the base, and 7.0–8.0 mm tall (Fig. 6E and 6F). Measurements are based on the most complete and uncompressed specimens, which tended to be among the largest in the size spectrum. Many smaller specimens were too compressed to yield reliable measurements, but may have smaller fruits. The gynoecium comprises six carpels, each containing a single-seeded locule, which are arranged around a central axis (Figs. 6D–6I). Ovary fusion is somewhat ambiguous, but they appear to be completely fused for most of their length and free distally, as indicated by gaps and indentations between adjacent carpels (Fig. 6I). However, the preservation makes it difficult to determine unequivocally whether proximal regions of the fruit are truly fused or just very closely appressed. In μCT sections the pericarp is composed of an inner darker and less dense region, and an outer denser region on the distal surface of the fruit (Figs. 6E and 6F). These density differences may reflect histological differences between two tissues of the pericarp. Cevallos-Ferriz, Estrada-Ruiz & Perez-Hernandez (2008) documented isodiametric cells in the pericarp, but preservation was too incomplete overall to determine other details of pericarp anatomy. Above each locule, the darker inner tissue forms a thin tapering structure that projects through the lighter outer tissue layer (Fig. 6E, arrow). While its position suggests the structure is probably the base of a style, available specimens do not retain any of the rock matrix surrounding the infructescence in which projecting styles or perianth would be preserved. Germination sutures and dehiscence structures were not observed.

Vasculature: Vascular bundles in the infructescence axis are often poorly preserved, but some specimens show three or four cycles of vascular bundles present in the cortex, surrounding a narrow central pith (Fig. 6G). The outermost vascular bundles appear smaller than the inner ones (Fig. 6G). Individual fruits are vascularized by six bundles that run through the axis of the fruit, opposite the carpels (Fig. 6H). At the distal end of the fruit, near the top of the locules, each vascular bundle trifurcates (Fig. 6I, arrowheads): the middle strand enters the distal end of the locule wall (Fig. 6F), and the other two strands extend over the top of the locule and out of the rock matrix (Fig. 6E). The vascularization of the locule is characteristic of apical placentation of ovules within the locule, consistent with the observations of Cevallos-Ferriz, Estrada-Ruiz & Perez-Hernandez (2008).

Perianth: Perianth elements in V. phytolaccoides specimens are poorly preserved, making it difficult to resolve the precise number and organization relative to the carpels. Transverse sections show an outer tissue layer that may or may not be fused to the pericarp (Fig. 6I), which can sometimes be seen in longitudinal sections extending all the way to the edge of the rock specimen (Fig. 6E). In longitudinal sections, small flanges can be seen at the base of the fruit below the ovary (Figs. 6C and 6F), possibly corresponding to an outer perianth whorl.

Discussion

Conclusion

Fossils of Viracarpon were first collected over a century ago by Hislop (1853). Since then, understanding the morphology of this genus, particularly V. hexaspermum, has been challenging and its taxonomic affinities remain elusive. In this study we undertook a morphological reinvestigation of Viracarpon by examining type material and studying museum specimens using X-ray μCT. Three-dimensional reconstructions of the fruits and visualization of specimens in multiple planes of section allowed us to clarify conflicting interpretations of fruit morphology and identify additional characters. These characters will be useful in refining the potential systematic affinities in future phylogenetic analyses. Comparisons of our reconstructions and observations with those of previous researchers revealed that the reconstruction of Chitaley (1954), the first to be published, was accurate with respect to the overall morphology of the fruits, capturing aspects of the complex external morphology resolved here using μCT. Further, we tested the hypothesis that Viracarpon was endemic to India by comparing it to C. phytolaccoides from the late Campanian of Mexico. Morphological similarities and the presence of diagnostic characters of Viracarpon indicate that Coahuilacarpon is a junior synonym of Viracarpon, leading to the recognition of V. phytolaccoides comb. nov. The geographic disjunction between these two occurrences presents an intriguing biogeographic conundrum and indicates that, rather than being endemic to India, the genus was once widespread. Viracarpon provides additional evidence for characterizing the largely invisible Cretaceous radiation of monocots. Its future inclusion in phylogenetic analyses will hopefully aid in resolving its systematic placement and elucidate evolutionary patterns and processes underlying the diversification of monocot angiosperms.

Supplemental Information