A new Oligocene-Miocene tree from Panama and historical Anacardium migration patterns

Migration of Boreotropical megathermal taxa during the Oligocene and Miocene played a key role in assembling diversity in tropical regions. Despite scattered fossil reports, the cashew genus Anacardium offers an excellent example of such migration. The fossil woods described here come from localities in Veraguas, Panama mapped as Oligocene-Miocene. We studied, described, and identified two well-preserved specimens using wood anatomical characteristics and completed extensive comparisons between fossil and extant material. The studied fossil woods share several diagnostic features with the modern Anacardium genus, including large solitary vessels, large intervessel-pitting, a simple vessel-ray pitting pattern, and mostly 1–3 seriate rays with large rhomboidal solitary crystals. We propose a new fossil species named Anacardium gassonii sp. nov., that adds an essential piece to the understanding of the historical biogeography of the genus. In addition, our findings confirm previous interpretations of this species’ migration from Europe to North America and its crossing through Panama, leading to subsequent diversification in South America. This discovery provides an important link to the historical migration patterns of the genus, supporting the notion of an Eocene migration to the Neotropics via Boreotropical bridges, as well as an Oligocene-Miocene crossing of Central America followed by diversification in South America.


Introduction
The Anacardiaceae family has approximately 80 genera and 900 species, represented by trees, shrubs, and some woody climbers; the family is widely distributed in tropical and subtropical areas as well as in warm-temperate regions [1]. The extensive fossil records of Anacardiaceae worldwide make this family an excellent example for biogeographical studies. There are

Sampling
Samples of MUPAN-STRI 44071 and MUPAN-STRI 44051 (Fig 1) were donated by Mr. Carlos Sandoval, a local farmer who has provided several samples for analysis. Both specimens were accessioned in the Smithsonian Tropical Research Institute repository, Panama (https:// biogeodb.stri.si.edu/jaramillosdb/web/fossils/). We highlight that MUPAN-STRI 44071 was a remarkably large specimen, with a perimeter of~2.5 m and a preserved length of nearly 7 m. The trunk was found in Los Boquerones, Veraguas, where we also collected the hand sized STRI 44051 specimen (latitude 08˚13' 46.4" N: longitude 80˚51' 45.1" W). Unfortunately, we do not have measurements for the total preserved length of the original trunk (Figs 1A and 2B) because a few pieces were extracted and sold (personal communication, Mr. Carlos Sandoval, 2019).

Geological setting
The wood specimens described here come from gullies in Boquerones, Veraguas, Panama. The radiometric ages of the geological units exposed in this area are unknown; however, the units have been mapped as Oligocene-Miocene [15][16][17]. In a recent visit to Los Boquerones, we did not observe much exposure to the related geologic unit. We have explored the surrounding areas identified as part of the Miocene Santiago Formation, but we restrain from inferring that the woods analyzed here are from the same formation. Further detailed geologic mapping is needed in this area to confirm these conclusions.

Fossil specimen preparation and identification
Petrographic thin sections of fossil material were prepared in transverse (TS), radial longitudinal (RLS), and tangential longitudinal (TLS) sections. Sections were ground to a thickness of 30 μm, mounted on glass slides using EpoFix resin, and coverslips were affixed with a UVcurable acrylates gel. The material was observed and imaged using an Olympus BX53 and an SC100 digital camera with a 10.5 Mpix CMOS sensor and a Zeiss AXIO Zoom V16; the material was then photographed with an AxioCam MRc5 camera.

Geographic distribution
The geographic distribution was plotted using QGIS software [23] and using free vector map data from Natural Earth [24]. For taxon occurrences, we included information from the location of A. gassonii (described here) and A. germanicum [14]. Modern distribution data were obtained from [25].

Nomenclature
The electronic version of this article in Portable Document Format (PDF) in a work with an ISSN or ISBN will represent a published work according to the International Code of Nomenclature for algae, fungi, and plants, and hence the new names contained in the electronic publication of a PLOS ONE article are effectively published under that Code from the electronic edition alone, so there is no longer any need to provide printed copies. The online version of this work is archived and available from the following digital repositories: PubMed Central, LOCKSS.

Comparative remarks
We conducted several searches using the Inside Wood Database. The most restrictive search was as follows: wood diffuse-porous (5p), vessels in tangential bands absent (6a), vessels in diagonal and/or radial pattern absent (7a), vessels in dendritic pattern absent (8a), exclusively solitary vessels absent (9a), vessels in radial multiples of 4 or more common absent (10a), vessel clusters common absent (11a), simple perforation plates (13p), intervessel pits alternate (22p), intervessel pits large (27p), vessel-ray parenchyma pits with much reduced borders to simple: pits rounded or angular (31p), mean vessel tangential diameter > 200 μm (43p), axial parenchyma aliform (80p), exclusively uniseriate rays absent (96a), larger rays commonly > 10seriate absent (99a), all rays procumbent absent (104a), prismatic crystals present (136p), prismatic crystals in upright and/or square ray cells (137p), tree (189p) with 0 allowable mismatches. We obtained 23 results, all belonging to three families: Anacardiaceae, Moraceae and Urticaceae. We ruled out the Moraceae genera because of the common occurrence of laticifers. Although Streblus glaber do not show laticifers, this species can also be distinguished from the fossil sample because it possesses abundant sclerotic tyloses, banded parenchyma, and sheath cells. Our fossil wood specimen is distinct from the Urticaceae results based on the occurrence of marginal bands of parenchyma, sheath cells, and the stronger winged parenchyma pattern. The listed results included several Spondioideae genera; therefore, we completed a comparison with available information in the IWD and the literature, e.g., [18][19][20] and the IWD, which is compiled in Table 1.
We also revised comprehensive surveys of Anacardiaceae mostly of the old Continent, e.g., [26,27]. We included a few of these genera in Table 1. Genera such as Bouea and Gluta could be distinguished because of the occurrence of marginal bands of parenchyma, features absent   Semecarpus woods have a strong winged-aliform pattern and wider rays compared to A. gassonii. Also [28], reported that Rhus, Cotinus and Pistacia have oblique to dendritic latewood vessel distribution and narrow fibers. Finally, Mangifera sp. shows parenchyma bands not observed in the fossils.

Comparison with Anacardium species
We revised the IWD and literature [19,20] to further study the Anacardium genus and develop a more rigorous analysis of this fossil. A summary of quantitative features is presented in Table 2, where we note differences and similarities when compared to the species available in the literature.
A few qualitative differences were noted when we compared the fossils with several species of the genus. For example, two distinctive vessel size categories can be observed in A. corymbosum, A. humile, A. nanum, and A. occidentale. In addition, sclerotic tyloses occur in A. corymbosum, while woods of A. giganteum and A. parviflorum have only uniseriate rays. The feature that most strongly distinguishes this new fossil species from most of the

PLOS ONE
Anacardium is the abundance of silica bodies in the ray cells. The sole species that lacks silica bodies and has abundant large prismatic crystals, as in the case of the study fossil A. gassonii, is A. excelsum. The new fossil species described here shows a strong resemblance to Anacardium excelsum (Fig 4), sharing the following diagnostic features: the absence of growth rings; solitary vessels with a few radial multiples (Fig 4A and 4B); vessel outline oval to round ( Fig 4B); simple perforation plates; alternate, polygonal, and large intervessel pitting ( Fig 4C); vessel-ray parenchyma pits with highly reduced borders, round to horizontally elongated ( Fig 4D); mean tangential vessel diameter > 200 μm (Fig 4A and 4B); axial parenchyma lozenge-aliform, and vasicentric (Figs 3B and 4A); rays mostly 1-3 cells wide (Fig 4E); mean ray spacing 7-12 per mm ( Fig 4E); and composed and abundant large prismatic rhomboidal solitary crystals present in the procumbent and square ray cells (Fig 4F). In a few specimens of A. excelsum reported in the literature, the vessels are smaller, the inclination of the perforation

PLOS ONE
plates is higher, and parenchyma strands can be shorter. After a detailed comparison, we confidently assign this fossil wood to the Anacardium genus, and we highlight its similarities to the modern Anacardium excelsum species. Trees of this species can attain 2 m in trunk diameter and reach 40 m in height; today they are commonly present in Panama, mainly in the Pacific slope. These trees are especially abundant along streams, and a few large individuals can inhabit mature forests. The species adapts well to disturbed areas (Pérez and Condit, n.d.).
The discovery of these fossils supports the notion that the genus Anacardium was present in Central America since the Oligocene-Miocene, suggesting that A. gassonii could represent the ancestral species for modern neotropical Anacardium species. This assertion may be supported by future phylogenetic analysis for the genus.
Gasson. This wood has axial parenchyma apotracheal diffuse and scanty paratracheal and the vessel-ray pits are smaller and round, whereas the new fossil wood discussed here has vesselray pits tend to be horizontally elongated and slightly in palisade.

Fossil record of Anacardium and comparison with other Anacardium fossils
The fossil record of Anacardium is incomplete. To date, the oldest fossil of the genus was recovered from the Eocene Messel flora in Germany and consists of fruits with attached hypocarps [14]. In their work, Manchester et al. [14] provide a comprehensive survey of other Anacardium fossil reports, which include several permineralized fruits from Colombia, Ecuador, and Peru. They concluded that the most reliable report is a fossil fruit found in Messel. Other fossils related to Anacardium include pollen grains from the Middle Miocene Salto de Tequendama, Colombia [32] and others from Malaysia, which were reported as "Tertiary" age [33].
Regarding fossil woods, only two have been discovered in localities of Peru. Pons and De Franceschi [34] described a wood specimen from the Middle Miocene Pebas Formation and suggested one of the studied woods resembled Anacardium. Unfortunately, the description of this specimen is neither detailed nor illustrated, and the authors did not describe the type of vessel-ray parenchyma pits, a trait that is key in the comparison of this genus. [8] described a new fossil species of Anacardium (A. incahuasi) from the Early Eocene of the Fossil Forest Piedra Chamana in Peru. This wood shares several features with Anacardium, including the absence of growth rings, vessel density, simple perforation plates, vessel-ray parenchyma pits with reduced borders, paratracheal parenchyma, and the presence of large prismatic crystals. Although we agree that A. incahuasi shares several characters with members of the Anacardiaceae family, the vessel-ray pitting does not resemble the patterns observed in the cashew genus. We note that the patterns observed in A. incahuasi are similar to those of a new fossil genus, Llanodelacruzoxylon sandovalii Rodríguez-Reyes, Estrada-Ruiz et Gasson [12], which was discovered in the same Santiago Formation as the Anacardium described here. This new fossil wood collected in Los Boquerones, Veraguas, Panama has a distinct set of features that match the Anacardium genus more closely. Therefore, we conclude that it can be confidently identified as a new fossil Anacardium species of, named A. gassonii Rodríguez-Reyes, Estrada-Ruiz et Terrazas.

Biogeographical significance of this new fossil Anacardium species
Anacardiaceae is a cosmopolitan plant family currently found in temperate, seasonally dry tropical forests and tropical wet forest regions [35]. Anacardiaceae originated in South East Asia during the Upper Cretaceous [9,35]. By the Paleocene, the family diversified in Southeast Asia and expanded its geographic range to sub-Saharan Africa; later, Anacardiaceae colonized South America. The most recent studies suggest that Anacardiaceae dispersed into North America, Oceania, and Madagascar, with some ancestors from tropical wet climatic niches expanding into tropical dry as well as temperate climatic regions [9].
Given the distribution of extant Anacardiaceae, long distance dispersal and vicariance events appear to be the most likely explanation for the modern distribution of the family [9,36,37]. Currently, Anacardiaceae is not found in extreme cold regions in the Northern Hemisphere; however, during the Paleocene Eocene Thermal Maximum (PETM) many tropical species spread throughout those regions [9,38,39].
Within Anacardiaceae, the genus Anacardium is restricted to the Neotropical region. This location is contrary to its sister group Fegimanra Pierre, a paleotropical genus that is restricted to the West African coast, and other phylogenetically close genera (e.g., Bouea Meisn., Mangifera L., Swintonia Griff. and Gluta L.) also native to paleotropical continents [24,35]. Three Anacardium species are found in South American savannas, whereas the rest of the species occur in humid forests [25]. Anacardium excelsum is the only species within Anacardium that occurs in the northern region of the Andes and in Central America. The current distribution of this species, together with the anatomical and geographic proximity to A. gassonii, may indicate a relict occurrence of the genus in those regions prior to the closure of the Central America Seaway (Fig 5).
The earliest reliable evidence of the cashew genus is a fossil fruit named Anacardium germanicum from the Eocene of Germany [14]. This fossil shows that Anacardium was widespread in the Eocene and suggests a paleotropical origin, as well as for other taxa in the family. The occurrence of A. germanicum in Messel also supports the hypothesis of a Boreotropical route from Eurasia to North America during warm climatic intervals and the subsequent colonization of tropical areas in Central and South America [14].
Past migrations that were facilitated by changes in continental arrangement and climatic changes in temperate regions from the Northern Hemisphere (Boreotropical vegetation) are well known in other genera. For example, Rhus L. (Anacardiaceae) [40] and Staphylea L. (Staphyleaceae) [41] have disjunct distributions between Asia, Europe, and the Americas, while Searsia F.A. Barkley (Anacardiaceae) [42] is found in southern Africa and Asia.
The discovery of A. gassonii confirms that the genus was present in the neotropics during the Oligocene-Miocene. This new fossil species, the distribution of extant Anacardium species, and the last proposed phylogenetic dating [9] support previous conclusions regarding migration routes and allow us to infer that the genus crossed the Central American Seaway (CAS) prior to its final closure (Fig 5) [17].

Conclusions
Anacardium gassonii shares most features with Anacardium excelsum, including the presence of large vessels, a lozenge-aliform parenchyma pattern, large and polygonal intervessel pitting, simple vessel-ray pitting slightly in palisade, and abundant large prismatic crystals in upright and square ray cells. The identification of this new species strongly supports the occurrence of the Anacardium genus in Central America during the Oligocene-Miocene and adds to previous conclusions regarding the rainforests that dominated this region prior to the closure of the Panama Isthmus. This discovery also adds an important link to the historical migration patterns of the genus, supporting the idea of an Eocene migration to the Neotropics via Boreotropical bridges, and subsequent crossing of the CAS during the Oligocene-Miocene leading to diversification in South America.