Fire-prone Rhamnaceae with South African affinities in Cretaceous Myanmar amber

The rapid Cretaceous diversification of flowering plants remains Darwin’s ‘abominable mystery’ despite numerous fossil flowers discovered in recent years. Wildfires were frequent in the Cretaceous and many such early flower fossils are represented by charcoalified fragments, lacking complete delicate structures and surface textures, making their similarity to living forms difficult to discern. Furthermore, scarcity of information about the ecology of early angiosperms makes it difficult to test hypotheses about the drivers of their diversification, including the role of fire in shaping flowering plant evolution. We report the discovery of two exquisitely preserved fossil flower species, one identical to the inflorescences of the extant crown-eudicot genus Phylica and the other recovered as a sister group to Phylica, both preserved as inclusions together with burned plant remains in Cretaceous amber from northern Myanmar (~99 million years ago). These specialized flower species, named Phylica piloburmensis sp. nov. and Eophylica priscastellata gen. et sp. nov., exhibit traits identical to those of modern taxa in fire-prone ecosystems such as the fynbos of South Africa, and provide evidence of fire adaptation in angiosperms. This paper reports the discovery of two fossils of Rhamnaceae preserved in amber that provide evidence of fire adaptation in Cretaceous angiosperms.

is dominated by gymnosperms, including the extinct Bennettitales ( Supplementary Fig. 4), accompanied by diverse angiosperms [23][24][25] . The exceptionally preserved flowers possess linear pubescent xeromorphic leaves (seemingly needle-like and tightly clustered with densely stellate or linear hairs) typical of vegetation adapted to seasonally dry and fire-prone conditions (Fig. 1), and burned plant remains also occur within the amber assemblages. Our results provide evidence of fire-adapted angiosperms from the Cretaceous, supporting the hypothesis that fire represented a biologically important selective pressure in some Cretaceous ecosystems. The fossils demonstrate that specialized floral adaptations to frequent burning arose early in angiosperm evolution and have remained more or less unchanged for over 99 Ma in Rhamnaceae. Extant Phylica is restricted to the South African Cape biodiversity hotspot and represents one of the most characteristic members of the endemic and fire-prone fynbos flora 26,27 . The finding of inflorescences related to, and even identical with, Phylica in a mid-Cretaceous Gondwanan deposit backdates the origin of a key component of the fynbos by ~35 Ma 28,29 and suggests that the fynbos flora may contain other elements typical of an ancient vegetation type that evolved in the fire-prone Cretaceous 30 .
Etymology. The generic name is a combination of the Greek words, Ēṓs (Ἠώς, meaning 'dawn') and the extant genus Phylica L. (itself derived from Greek: phyllikos, meaning 'leafy'). The gender of the name is feminine.
Generic diagnosis. Identical to Phylica except with 8 sepals (2 × 4-merous) (versus 5 in Phylica), lacking petals (present in most Phylica, although secondarily reversed independently in several crown species), and indumentum composed of stellate rhamnaceous hairs (simple in Phylica). Ovary inferior, fused to receptacle. Style columnar, simple. Fruit a typical capsule, obovoid, about 5.0 mm long, crowned with persistent base of calyx with a convex calyx area. A complete description is provided in Supplementary Note 2.
Etymology. The specific epithet is a combination of the Latin terms priscus (meaning 'ancient') and stellatus (meaning 'starry').
Eophylica priscastellata Shi (Fig. 4) Etymology. The specific epithet is a combination of the Greek term pîlos (πῖλος, meaning 'hair') and burmensis (itself a combination of Burma and the Latin suffix -ensis, denoting place).   32 . On the basis of presence or absence of petals, the number of sepals and the form of hairs of the indumentum, we assigned our fossils to two new species: Eophylica priscastellata gen. et sp. nov. (Figs. 2-3) and Phylica piloburmensis sp. nov. (Fig. 4 and Supplementary Fig. 6), the presence or absence of petals reflecting morphologies seen in extant Phylica 32 . A close relationship between Eophylica and Phylica is supported by comparative morphological and phylogenetic reconstructions based on flowers, fruits and pollen grains (Figs. 2-4, Supplementary Figs. 21-30 and Videos 2-3). E. priscastellata can be distinguished from P. piloburmensis on the basis of the number of sepals and the absence of petals. The second species is morphologically indistinguishable from extant Phylica, and our phylogenetic analyses recovered the species as robustly nested within the Phylica clade. Given the completeness of our observed character suite, afforded by the exceptional fidelity of amber preservation, that flower morphology is the greatest source of data for ascribing species to supraspecific taxa, and the robustness of our phylogenetic results, we have placed this species in the extant genus Phylica. This greatly extends the age of this clade and represents the first extant angiosperm genus documented from the Mesozoic. While this seems controversial, we believe the extraordinary nature of these fossils and the data compel such a conclusion. Extant species are exceptionally rare in Myanmar amber, but are known to occur, and we believe future data will corroborate our conclusions regarding the identity of P. piloburmensis. We provide formal descriptions of the new fossil taxa in Supplementary Note 2.
The clustered, needle-like dry leaves are typical of fire-adapted angiosperms ( Fig. 1) 26,30 . Taxonomically, needle-like leaves clustered around the stems of a small shrub ( Fig. 1 26 ) and some non-fynbos clades (for example, Brunia and Berzelia (Bruniaceae)) also have narrow, clustered leaves. However, species of Phylica differ in having one or more of the following 32 : leaves at the tip of the stem form 'pseudanthium head' unique to the genus (Fig. 1); leaves and flowers are covered by dense hairs; unarmed but villous-pubescent, branched twigs; the absence of stipules, and leaves with only one main vein . Importantly, the typical pseudanthium head in these fossils is distinctive within angiosperms 32 . It is exclusive to some extant species of Phylica (for example, P. pubescens and P. plumosa; Supplementary Fig. 10e-j) 32,35 , where it represents an intermediate developmental stage from vegetative growth to the flowering phase ( Fig. 1, Supplementary Figs. 7-20 and Video 1) 32 . The head comprises many tightly clustered scale-like bracts, with mature leaves or bracts (below the true flower) developing later. Despite having a similar 'head' shape, the pseudanthium bracts are morphologically distinct from those of the true flower (Fig. 2m) 32 .
True scale-like bracts; the flowers are semi-opened; sepals are enlarged and triangular; sepals have a prominent median vein; hypanthia are turbinate in shape; the petals are either small and alternate with the sepals or are completely absent; stamens are opposite the petals; the disc is intrastaminal, thin to more or less fleshy; the ovary is below the disc and completely fused to the receptacle; the style is columnar, simple, shorter than the calyx tube; and the stigma is slightly lobed. Specially, as in extant Phylica 32 . Although capsule-like fruits are common in Rhamnaceae, Phylica fruit morphologies are shaped by their specific flower characters (such as the lengthened calyx tube and dense hairs). Differing from other capsules that lack sepal elements and usually have an exposed mature pistil, fruits of Phylica and Eophylica are crowned by the persistent calyx (or sepal) that includes a hidden pistil (Figs. 3f-h and 4k,l). The pericarp develops from the base of the persistent calyx, and in the fossil fruit appears hard, ribbed ( Fig. 3f and Supplementary Fig. 27), or covered with thick hairs originating from the flower ( Fig. 4k and Supplementary Fig. 28). As in extant species 32 , fossil fruits of Eophylica and P. piloburmensis are notably larger than the flower (≥5 mm wide) (Supplementary Figs. 27 and 28). Similar fruits occur in the genus Rosa in Rosaceae, but those fruits have the achenes inside, rather than having septal polycarpels as in our fossils and extant Phylica ( Fig. 4m and Supplementary  Fig. 27m,n). As in extant Phylica 32 , petals are either present or absent in our fossil flowers, with E. priscastellata lacking petals (Fig. 3a,b) and P. piloburmensis possessing petals (Fig. 4a). with the Rhamnaceae is further supported by the following features 37 : (1) medium-sized with polar axes of 21-24 μm and equatorial diameters of 18-21 μm (P/E ratio: 1.2); (2) colpi long, narrow and with blunt ends; and (3) the middle of the aperture lalongate, with two ends connected to a thinned part of the exine to form an H-shape (Fig. 2q-v and Supplementary Fig. 29).
Our fossils also preserve several other features whose combination is diagnostic of the family Rhamnaceae 38 . As in our fossils, extant Phylica is a morphologically divergent clade of the Rhamnaceae that exhibits only some plesiomorphic characters seen in the family 32 . Although a lengthened cyathiform calyx tube and enfolding sepals and petals are important flower characteristics of living Phylica, they obscure important internal Rhamnaceae characters (that is, 4-or 5-parted with a single whorl of stamens opposite the petals). We used micro-XCT scanning to show that our fossils preserve simple leaves, small flowers with 5(4)-merous sepals (5 in P. piloburmensis and 8 (2 × 4-merous) in E. priscastellata), which are valvate in the bud (Figs. 2b,l and 4e-g); 5 or 8 (2 × 4-merous; E. priscastellata) stamens that alternate with the sepals and are opposite the petals and anthers, which are tightly enfolded by the hooded petal apices (if petals are present; P. piloburmensis) ( Fig. 4a and Supplementary Figs. 21n, 22l,m and 23k); and the presence of an intrastaminal disc (Fig. 2b) 39 . These features, combined with crowding leaves (Fig. 1), are characteristic of Rhamnaceae 39 .
Phylogenetic analysis integrating morphological and molecular partitions (Supplementary Notes 6 and 7) confirmed crown-group membership of both fossil species ( Fig. 5a and Supplementary  Fig. 30). Although one major character (the presence of stellate hairs) makes the fossil E. priscastellata different from all extant species of Phylica (Supplementary Table 1), our new fossil species exhibit other features common to extant Phylica. In E. priscastellata, the leaves and twigs are similar to those of extant P. pubescens or P. ambigua ( Supplementary Fig. 10), but the presence of stellate hairs and absence of petals distinguish it. Its fruits lack a covered indumentum as in extant P. parviflora ( Supplementary Fig. 27m,n). Moreover, the leaves, flower and fruit characteristics of P. piloburmensis are also present in many other extant species of Phylica (represented by extant P. axillaris; Fig. 4c,h-j,m,n) 32 . All in all, our fossils provide evidence of a remarkable case of morphological, and probably also ecological, conservatism within a crown-eudicot clade since the mid-Cretaceous.

Discussion
Early origin of fire-adapted angiosperms in the Cretaceous. The mid-Cretaceous (Albian-Turonian; 112-90 Ma) was exceptionally warm, with some of the highest temperatures in Phanerozoic geological history 40,41 and a relatively shallow temperature gradient between the poles and the tropics 42 . Elevated atmospheric oxygen levels in the Cretaceous (23-29% compared with 21% at present [43][44][45][46] ) and abundance of charred plant fossil remains in the rock record [47][48][49][50][51] suggest that fires were frequent during this period 14,[52][53][54][55][56][57][58] . Some hypotheses of angiosperm diversification highlight that Cretaceous fire regimes may have opened land up for the first flowering plants, whose rapid colonization of regeneration gaps may have provided an advantage over the then-dominant gymnosperms that are intrinsically slower growing 13,14,59 . By providing a new source of fuel, angiosperms may have also substantially altered the existing fire regimes, to the disadvantage of gymnosperms 13,15,60 . There is abundant evidence that fires burnt vegetation containing angiosperms in the Cretaceous 13 , the most remarkable examples being three-dimensionally preserved charcoalified flowers, fruits, seeds and other organs of angiosperms co-occurring with burned remains of gymnosperms and free-sporing plants 10,11,61 . Moreover, previous molecular clock studies have inferred a Cretaceous origin of fire-proneness in some gymnosperm (Pinaceae 62,63 ) and angiosperm (Proteaceae, Myrtaceae, Haemodoraceae and Restionaceae 54,64-66 ) lineages. However, direct palaeontological evidence of fire-prone traits are scarce, primarily because such traits evidently have low preservation potential 67,68 .
Fossil Eophylica and Phylica inflorescences preserve unambiguous morphological hallmarks of fire-adapted angiosperms. Many fire-prone plants are 'drought resisters' 34,69 and Eophylica-Phylica belong to this group. Numerous Cape species (such as species in the families Ericaceae and Proteaceae) that are drought resisters are able to produce new growth after fires, from buds on buried lignotubers at the base of burned stems 69 . Phylica are small shrubs whose clustered, needle-like dry leaves are typical of fire-prone vegetation; these characters in turn ensure relatively rapid renewal after fire and confer resistance to severe drought 70,71 . While some extant members of the genus Phylica typically do not survive wildfires, which may in itself have an adaptive value in environments that burn frequently 72 , their seeds accumulate in the soil and germinate after fire 73,74, . Their relatively small and rounded seeds are easily incorporated into the soil and reach greater depths than larger seeds, and is linked to their ability to form long-term persistent seed banks 75,76 . Our discoveries of burned plant remains in associated amber pieces further provide direct evidence for a community living in a fire-prone environment ( Supplementary Fig. 5).
An early Gondwanan origin. Fossil Eophylica flowers are distinct from those of any angiosperms that grow in the subtropics of northern Myanmar today. Phylica is endemic to the fire-prone fynbos flora of the South African Cape. The fynbos represents one of the most remarkable global floral biodiversity 'hotspots' , characterized by unique species richness and endemism, with over 9,000 recorded plant species restricted to a small geographic area (±90,000 km 2 ), of which almost 70% are endemic 77 . Much of this diversity is due to a handful of speciose clades (the 'Cape floral clades') that, including Phylica 31,78,79 , have traditionally been regarded as having radiated within the Cape 31,78,79 . Geological studies and fossil evidence [80][81][82][83][84] indicate a Gondwanan origin for the diverse biota in Myanmar ambers, but two alternative pathways for its transport to Myanmar, either via the West Burma Block or the Indian Plate, were possible 80 . Although palaeomagnetic studies indicate that the West Burma Block rifted from north-western Australia between the Late Triassic and Late Jurassic and was an isolated landmass in the Tethys Ocean in the mid-Cretaceous 85,86 , our result indicates that the biota may equally well have drifted northward on the Indian Plate after it separated from Madagascar and southern Africa at approximately 88-90 Ma 64,80,87 , just after the formation of the Myanmar amber (110-99 Ma; Fig. 5). Thus, when the Myanmar ambers were forming, the ancestor of Phylica had an ancient distribution within the united Gondwana and was possibly widespread at middle southern latitudes, including parts of Africa. Its survival today at a similar latitudinal band, but under a cooler global climate, suggests that it may be close to the edge of its natural climatic tolerance envelope (Fig. 5).
Our results demonstrate that a key element of the xeromorphic fynbos vegetation existed as long as 99 Ma ago, as well as the great antiquity of open fire-prone vegetation in Gondwana 64 . Given that southern Gondwana during the middle Cretaceous (Albian-Turonian; 112-90 Ma) was arid to semi-arid 88,89 , possibly possessing a similar climate as the South African Cape today, the fynbos could provide a recent proxy for understanding fire-prone plant communities in the Cretaceous 30 (Fig. 6). However, whether or not the fynbos biome itself dates back to the Cretaceous remains to be tested with the aid of further plant fossils. Miocene pollen assemblages from the southwestern Cape are noted to be comparable with those of Australian floras 90 , although some of these Gondwanan elements were lost after climatic changes at the end of the Miocene. There are no Cretaceous fossil records in the Cape, but offshore records from the West Coast exhibit abundant conifer pollen [91][92][93] . This could be a taphonomic effect due to the high productivity and wind dispersal of conifer pollen compared with the lower pollen productivity and more targeted dispersal of the insect-pollinated shrub and herb components typical of fynbos-type vegetation 94 . Notably, herb pollen declines in abundance moving offshore 94 . Inland records from kimberlite pipes show a mixed flora with some fynbos elements (Proteaceae, Ericaceae and Restionaceae) in the Palaeogene in a tropical to warm temperate environment 95 . From the southwestern Cape coast (where fynbos occurs today), fynbos elements were present (restios, ericas, proteas) during the Oligo-Miocene, mixed with palm and miombo floras, and interpreted as tropical to subtropical forest with lianas, vines, evergreen trees, palms and ferns [96][97][98] . Taken together, these lines of evidence point towards a much more complex history of the fynbos biome than previously thought.
The Myanmar amber tropical forest palaeoenvironment. The presence of abundant fire-prone vegetation in amber from northern Myanmar has important implications for interpreting the palaeoenvironment of this key Mesozoic Konservat-Lagerstätten. Myanmar amber harbours perhaps the most diverse Cretaceous amber biota known to date 99 , famous for its well-preserved vertebrate remains and abundant insect fossils. The Myanmar amber palaeoenvironment is most often reconstructed as a hot tropical forest 99 , with a near-equatorial climate 85 located in the vicinity of brackish water and the seashore 19,100,101 , perhaps akin to modern swamp forests 102 . Eophylica and Phylica fossils and associated burned plant remains suggest that the Myanmar amber forest may have been prone to seasonal fires, similar to some tropical peat swamp forests in more recent geological history 103 . Major fires and subsequent drought may have been associated with some amber deposits, such as the Cretaceous New Jersey amber that contains abundant charred remains of plants and insects 104 , and fire is a common cause of resin production in fire-prone vegetation today 105 . It is probable that seasonal fires may have also played a role in the production of Myanmar amber, one of the largest amber deposits in the world, since resin secretion can be triggered by wounding, including fire wounds 105,106 , as well as any stresses that impact water uptake (for example, insect infestation, drought, etc 107,108 .). Notably, burned plant remains are abundant in our ambers (Supplementary Fig. 5) and it is clear that these remains were charred prior to resin entrapment. Abundant seasonal fire may also partly explain the abundance of amber in the Cretaceous rock record, as fossiliferous ambers of older age are exceedingly rare and largely occur only as traces 109 .
Amber provides a complementary window of early angiosperm diversity. The unique floral architecture preserved in Eophylica and Phylica fossils is identical to that seen in extant members of the genus Phylica, predating molecular clock estimates for the genus by  at least 35 Myr 28,29 . This makes Eophylica-Phylica one of the best documented 'living fossil' clades in the angiosperm fossil record. Our study also demonstrates that amber provides a complementary window into the early evolution of flowering plants, by overcoming some biases inherent to compression and carbonified fossils that make up the majority of the early angiosperm fossil record 10,11 . First, most compression fossils are preserved primarily in wetlands, while amber palaeoenvironments range from warm forests to seashores 104 , capturing a different set of palaeoenvironments. Second, amber inclusions are often preserved with extreme fidelity, comparable to the inflorescences of modern plants. The discovery of further well-preserved angiosperm fossils in Cretaceous amber deposits may eventually backdate other crown angiosperm clades, helping to provide more calibration points for understanding the tempo of Cretaceous angiosperm radiation and ultimately contributing to reconciling the often-perceived incongruence between the angiosperm fossil record and molecular clock estimates 71,110,111 .

Methods
Fossil provenance and ethical statement. The material described herein originates from two amber mines, Tanaing and Hkamti, located in the Hukawng Valley, Kachin State in northern Myanmar 17 . The mines are introduced in Supplementary Note 1. The amber has been radiometrically dated to the earliest Cenomanian, ~99 Ma, and was not produced earlier than the late Albian 18,19 . Our study was initiated in 2015 and all amber specimens were acquired from local sellers before December 2016, prior to the escalation of the humanitarian crisis in the region  Fig. 6). Of the 21 pieces of amber, 19 originate from the 'Tanaing' amber mine in Hukawng Valley, except for QUST-AM20511 and QUST-AM32414, which originate from the nearby 'Hkamti' mine (further shown in Supplementary Figs. 1-3 and 6).
Photography. Fossils were photographed with a digital camera (Fujifilm GFX 50 R with Laowa C11625 2.6 X, Cambo Actar 105 hr or Mitutoyo 5-10 X lens) fitted to a macro rail (Cognisys). For every photograph, 30 to 200 images were stacked with Combine ZP and Photoshop CS4. Some specimens were also photographed using Leica DVM6 and M205FA microscopes (Leica).
To examine internal structures of the flowers, fossils were scanned with Xradia Versa Micro-XCT 620 (Carl Zeiss X-ray Microscopy) housed in the Advanced Materials Research Institute of Yangtze Delta. All specimens were scanned with the same beam energy of 40 kV, 3 W and the LE1 filter, but with different exposure times and pixel sizes that depended on the sample size and condition. Clear scanned images were obtained for 20 of the 22 fossil specimens, but the other 2 samples lacked sufficient contrast for scanning. The obtained image stacks were reconstructed with Dragonfly (ORS). Final figures were prepared with Photoshop CS5 and Illustrator CS5 (Adobe).
Character scoring and phylogenetic analysis. Morphological characters scored from leaf, habit, flower, fruit and pollen were compiled from previous studies 35,36,112 and are described in Supplementary Note 6. The combined phylogenetic tree was reconstructed on the basis of morphological characters and molecular data following the method of Wilf et al. 113 ; the procedure is further described in Supplementary Note 7.
Statistics and reproducibility. During the photographing, micrographs for each fossil were taken repeatedly (5 to 30 repeats) to ensure consistency, and the detailed micrographs for each fossil are exhibited in Supplementary Figs. 5-29.
Reporting Summary. Further information on research design is available in the Nature Research Reporting Summary linked to this article.

Data availability
The  Corresponding author(s): Shuo Wang Last updated by author(s): December 12, 2021 Reporting Summary Nature Portfolio wishes to improve the reproducibility of the work that we publish. This form provides structure for consistency and transparency in reporting. For further information on Nature Portfolio policies, see our Editorial Policies and the Editorial Policy Checklist.

Statistics
For all statistical analyses, confirm that the following items are present in the figure legend, table legend, main text, or Methods section.

n/a Confirmed
The exact sample size (n) for each experimental group/condition, given as a discrete number and unit of measurement A statement on whether measurements were taken from distinct samples or whether the same sample was measured repeatedly The statistical test(s) used AND whether they are one-or two-sided Only common tests should be described solely by name; describe more complex techniques in the Methods section.
A description of all covariates tested A description of any assumptions or corrections, such as tests of normality and adjustment for multiple comparisons A full description of the statistical parameters including central tendency (e.g. means) or other basic estimates (e.g. regression coefficient) AND variation (e.g. standard deviation) or associated estimates of uncertainty (e.g. confidence intervals) For null hypothesis testing, the test statistic (e.g. F, t, r) with confidence intervals, effect sizes, degrees of freedom and P value noted

Software and code
Policy information about availability of computer code Data collection CT data were reconstructed with Dragonfly Pro 4.1.

Data analysis
Phylogenetic analyses were performed using TNT 1.1 and RAxML 8.2.
For manuscripts utilizing custom algorithms or software that are central to the research but not yet described in published literature, software must be made available to editors and reviewers. We strongly encourage code deposition in a community repository (e.g. GitHub). See the Nature Portfolio guidelines for submitting code & software for further information.

Data
Policy information about availability of data All manuscripts must include a data availability statement. This statement should provide the following information, where applicable: -Accession codes, unique identifiers, or web links for publicly available datasets -A description of any restrictions on data availability -For clinical datasets or third party data, please ensure that the statement adheres to our policy The Field-specific reporting Please select the one below that is the best fit for your research. If you are not sure, read the appropriate sections before making your selection.

Life sciences Behavioural & social sciences Ecological, evolutionary & environmental sciences
For a reference copy of the document with all sections, see nature.com/documents/nr-reporting-summary-flat.pdf

Ecological, evolutionary & environmental sciences study design
All studies must disclose on these points even when the disclosure is negative.

Study description
Description of two fossil species from 21 amber pieces from Myanmar.

Research sample
A total of 21 amber pieces related to two fossil species were found among all our amber collection.

Sampling strategy
The fossil species were quite rare in the amber collection. Since we used all of them in the study, so no statistical methods were used to determine sample size.

Data collection
These fossils were photographed with a digital camera (Fujifilm GFX 50 R with Laowa C11625 2.6 X, Cambo Actar 105 hr or Mitutoyo 5-10 X lens). Some specimens were also supplementarily photographed using a microscope Leica DVM6 and M205FA (Leica AG, Heerbrugg, Switzerland). Chao Shi recorded the data.
Timing and spatial scale Our study was initiated in 2015 and all amber specimens were acquired from local sellers before December 2016.

Data exclusions
No data were excluded.

Reproducibility
Every micrographs were taken repeatedly (5 to 30 repeats) to ensure consistency.

Randomization
All fossil specimens were selected (not randomly selected) for our study.

Blinding
The fossils were used for morphological description but not for experimental test, thus blinding is not applicable to our study.
Did the study involve field work?

Yes No
Reporting for specific materials, systems and methods We require information from authors about some types of materials, experimental systems and methods used in many studies. Here, indicate whether each material, system or method listed is relevant to your study. If you are not sure if a list item applies to your research, read the appropriate section before selecting a response.