Printer Friendly

Fossil records in the Lythraceae.

Introduction

The use of fossils to date the origins and divergence times of angiosperm taxa has recently become a universal and important component of plant molecular phyloge-netic studies. Timetrees and the information gained from accurate dating increasingly inform our understanding at many levels (Avise, 2009; Bell et al., 2010; Benton et al., 2009; Magall6n, 2010; Quental & Marshall, 2010). Beyond the initial patterns of divergence and relationships and the tracking of character evolution, robustly dated phylogenies are providing a means of: validating historical biogeographic scenarios (Renner & Meyer, 2001; Renner et al., 2001; Conti et al., 2002; Sytsma et al., 2004; Wang et al., 2009); answering questions about rates of speciation and extinction; and relating the relationship of species richness to past geological upheavals and climate changes (McElwain & Punyasena, 2007; Ricklefs, 2007).

Critical to the production of a dated phylogeny is a comprehensive record of credibly determined, accurately dated fossils to act as age constraint points in calibrating the rate of genomic change (Welch & Bromham, 2005). Reviews of the fossil records of individual flowering plant families are becoming more commonplace as phylogeneticists seek this information for dating purposes (e.g. Ickert-Bond et al., 2009 on Ephedra; Graham, 2009 on Rubiaceae; Martinez-Millan, 2010 on Asteridae; Bell et al., 2010 on all angiosperms; Jacques et al., 2011 on the Menispermaceae).

The Lythraceae have been incorporated in a number of molecular phylogenetic and dating analyses (e.g., Renner et al., 2001; Conti et al., 2002; Huang & Shi, 2002; Sytsma et al., 2004; Wang et al., 2009) but they have not yet been the focus of such an analysis. This study aims to identify the oldest reliably determined records for each extinct and extant genus attributed to the Lythraceae and to provide an up-dated, vetted summary of their fossil remains and stratigraphy for use in future dating analyses to elucidate times of origin, periods of greatest evolutionary activity, and to generate well-grounded historical biogeographic perspectives for the family and the major lineage in which it occurs.

The Lythraceae are a medium sized family of 28 genera and ca. 600 species. They constitute one of nine families of the Myrtales and together with the Combretaceae and sister family Onagraceae form a major lineage of the order (Webb, 1967; Sytsma et al., 2004; Graham, 2007, 2010b; Graham et al., 2011). The genera are equally divided between the Old and New World, primarily in the tropics and subtropics; only one genus, Lythrum, is equally diversified on two continents, in North America and Eurasia. The majority (19 of 28) of the genera comprise only one or two species. The family classification in current use (Koehne, 1903) recognizes two highly artificial tribes and four subtribes and is being extensively reorganized to reflect relationships revealed by molecular phylogenetic analyses and enhanced morphological and chromosomal knowledge of the taxa.

A review of the fossil record of the Lythraceae was first published over 40 years ago, before the molecular phylogenetic era in botany (Graham & Graham 1971b). Since then, the taxonomic limits of the family have expanded as a result of molecular-based evidence by inclusion of four small, previously independent families:

Duabangaceae, Punicaceae, Sonneratiaceae, and Trapaceae (Graham et al., 2005). The latter two have particularly abundant fossil records. The oldest accepted fossils of the family in 1971 were several capsular fruits and seeds of generalized lythracean structure from the Eocene of southeastern England (Reid & Chandler, 1933; Chandler, 1957, 1960, 1961 a, b, 1962, 1963a, b, 1964). Seven living genera were recognized in the fossil record, the earliest being seeds of the swamp genus Decodon from the middle Eocene Barton and Upper Headon beds in southern England (Chandler, 1960), and from Eocene deposits in Russia (Eyde, 1972). Muller (198 l a) summarized fossil pollen records for the family and Collinson (1983) reviewed the important early Eocene London Clay Flora (Reid & Chandler, 1933) which includes a number of extinct genera attributed to the family. In recent years, reports of fossil Lythraceae have substantially increased both in number and in geographic and stratigraphic range. Significant new fossils of unquestioned relationship to the Lythraceae have come to light and more accurate geological ages are being assigned to fossil beds containing them. The records, old and newly published, are assembled here.

Material and Methods

Fossil Lythraceae are represented by silicified wood remains, leaf impressions, and fossilized pollen, seeds, and fruits. A relatively complete record is presented here of fossils considered related at the family or genus level to the Lythraceae. Where abundant fossils of a genus exist, as is the case for Florshuetzia and Sonneratia pollen in the Oligocene and Miocene of southeastern Asia, for wood of Lagerstroemia in the Miocene of India and eastern Asia, and for fruits and seeds of Trapa and Decodon in the Miocene and younger epochs in Europe, not all published occurrences are listed. Rather, exemplars are listed that cover the geographic and stratigraphic ranges of the taxa up to near modern times. Emphasis of this study is on determining the oldest well-documented and accurately identifed fossils of the family.

Assessments of the validity of identifications are based mainly on the quality of the published descriptions and/or illustrations and on the extent to which the diagnostic features are consistent with modern genera. LaMotte's Catalog (LaMotte, 1952) was used as an initial source for North American fossil plant references. Digital images of type material were acquired to resolve some questions; actual type specimens were not consulted. Comparative data used for checking fossil identifications included: for wood anatomy, Baas, 1986 and Baas & Zweypfenning, 1979; for leaf venation, X-rays by genus from the Natural History Museum, Smithsonian Institution; for pollen, Coz Campos, 1964; Graham & Graham, 1971b; Graham et al., 1985, 1987, 1990; Kim et al., 1994; for seeds and general morphology of modern genera, Graham, 2007 and pers. observ.

The fossils are classified according to one of three levels of reliability. The record is accepted, if the fossil corresponds closely to an extant member of the family or is characterized by a suite of lythracean characters. The record is unconfirmed if total information suggests a relationship but the information is insufficient to reasonably establish the identity. A large number of fossils fall into this category because the description is inadequate to secure the determination and/or the illustration lacks the the clarity or detail needed to assign the specimen to a genus or a previously described morphotype. The record is rejected when the information available from the description and/or illustration does not compare favorably with accepted family or generic parameters. A large geographical or time gap is not considered reason for rejection; too many recent finds have come from unexpected geographic areas or from geological strata much older than previously known.

A cautionary note with respect to acceptance of fossils as members of the Lythraceae is that morphological commonalities are shared among many members of the Myrtales. Misleading instances of homoplasy, especially in wood, leaf, and pollen characters of the Lythraceae, need to be considered when assigning relationships. The Myrtales is "wood anatomically a fairly closely knit assemblage" with the "high probability of parallel specialization lines in individual families" (van Vliet & Baas, 1984: 794). For example, septate fibers, in addition to their presence in the Lythraceae, appear in the Onagraceae, Oliniaceae, Psiloxylaceae, and parts of the Melastomataceae and Combretaceae; paratracheal parenchyma also appears in parts of the Myrtaceae, Combretaceae, Melastomataceae, and Crypteroniaceae; specialized rays with an emphasis on procumbent cells are common in the Melastomataceae and Onagraceae; and vascular tracheids also are found in the Combretaceae (van Vliet & Baas, 1984). Leaves of the Lythraceae share brochidodromous venation with other myrtacean genera (Hickey & Wolfe, 1975) and cannot always be identified by shape and primary venation patterns alone. Identification is more certain when the tertiary venation is present but it is not always visible on leaf impressions. Some Decodon leaves have been misidentified in the past as members of the Moraceae, Asclepiadaceae, and Myrtaceae (Kvacek & Sakala, 1999). Pollen structure also can share selected details with unrelated plant groups, as exemplified by similarities in aperture structure in Sonneratia in the Lythraceae and Parkinsonia in the Fabaceae (Muller, 1978: 285). The pollen in Myrtales, although diverse, exhibits a cohesive structural core (Patel et al., 1984) and grains of different genera may appear superficially similar until studied closely, preferably with the scanning electron microscope (SEM; Ferguson et al., 2007; Liu et al., 2008). Pollen of Cuphea subgenus Cuphea bears some similarity to members of the Myrtaceae (i.e. triangular outline, small size, syncolpate condition, and low exine sculpture) but the grains are distinctive at higher magnifications with SEM (Graham & Graham 197lb; Patel et al., 1984; Graham et al., 1985). Even the unique pollen of Trapa (Lythraceae) with its frilled crests bears a general resemblance to pollen of some species of Ludwigia (Onagraceae) (Patel et al., 1984). Seed shape is described here as it is commonly illustrated with the hilum and micropyle at the 'bottom' of the seed, although it is acknowledged that this is developmentally the apex of the anatropous ovule. Acceptance of fossil determinations in this review relies first on the kind and amount of information presented by authors, then ultimately on my informed subjective judgement as to whether the paleobotanical assessments are convincing.

Attempts are made to provide up-to-date, accurate fossil ages; many recent dates differ from those originally assigned. The most reliable ones are from sediments dated by radiometric techniques such as 40K/40Ar and 40Ar/39Ar. However, many fossil beds are assigned an age based on stratigraphic position or on the presence of other taxa for which an age has been generally accepted. The geologic time scale followed here is that of Ogg et al. (2008).

Of particular interest to this study is the age of the fossil-rich Deccan Intertrappean Beds of India from which many lythracean forms have been recovered. The Deccan Traps are one of the largest outpourings of lava in the world, estimated to have covered as much as 1.5 krn2 or nearly half of present India and to have produced a tremendous volume of basaltic lava that at places is more than 2,000 m thick (Pande, 2002; Cripps et al., 2005). Today, the Traps cover approximately 512,000 km2 of western and central India. Prior to the 1990's they were generally considered Eocene in age or to range in age from Late Cretaceous to Eocene (Prakash, 1960). More recently, many individual deposits have been accurately dated using radiometric methods and it is now clear that the Traps were the result of three major eruptions over a period from the Late Cretaceous (Maastrichtian) to the early Paleocene (Danian, 67.3-63.8 Ma; Singh & Kar, 2002; Cripps et al., 2005; Keller et al., 2009c). The successive major flows thus bridged the asteroid event at 65.5+0.3 Ma (Keller et al., 2009a, c; Hedges & Kumar, 2009), capturing in the intervening sedimentary deposits (the intertrappean beds) diverse organisms whose remains reflect the significant climatic, environmental, and evolutionary biological changes over that time period. Floras of the Maastrichtian deposits contain abundant algae, ferns, gymnosperms, and palms together with numerous now extinct plant genera and a diverse fauna. Floras of the early Paleocene differ from those in the Maastrichtian layers by the presence of abundant angiosperm pollen and other angiosperm remains, including the oldest flowers attributed to the Lythraceae.

The exact position of the K-T boundary is not easily determined over much of the Deccan. However, dramatic changes in fossilized planktic foraminifera in the intertrappean beds at Jhilmili, Chhindwara district, Madhya Pradesh are an exception that provides the most accurate position of the K-T boundary in the Deccan to date (Keller et al., 2009c). This is important to the history of the Lythraceae because sediments at Jhilmili dated as Danian (65.5-61.7 Ma; Keller et al., 2009a, b, c; Keller, pers. comm.) are very likely equivalent to the highly productive Lythraceae fossil site at nearby Mohgaon-Kalan. Lists of fossil taxa of the Deccan, including representatives of two living and two extinct genera of the Lythraceae, are found in reviews of the Paleogene (Bande, 1992) and Neogene (Guleria, 1992) floras of peninsular India.

A previously generated phylogenetic tree of the Lythraceae from Morris (2007) is utilized as a framework for displaying the earliest confirmed ages of the genera. Dating of the phylogeny is not part of this review. The tree employed is a majority rule Bayesian consensus of 9,000 trees from combined sequence data of the rDNA nuclear internal transcribed spacer (ITS), and chloroplast rbcL, the trnL-F spacer, trnK-matK region, and the atpB-rbcL intergenic spacer. It is the most complete phylogeny of the family to date, including all modem genera except Crenea and Diplusodon Pohl, and carries a high degree of confidence with well-supported branches in all but the base of the tree. The tree is slightlY simplified from Morris by reduction of Lythrum from a 100 % supported three-member clade to one species as place holder for the genus and by collapse of the Heimia-Rotala-Didiplis clade to the base of the tree because of low support values for the branch (posterior probability 55; maximum-liklihood support 60 %). Genus Cuphea is added to the tree based on earlier results (Graham et al., 2005). The tree is substantially congruent with earlier results in recognizing two major elades in the family. It differs in placing Decodon + Lythrum on a moderately-supported branch (Bayesian prior probability 78; maximum likihood support 74 %) as sister to the major clade Duabanga--Ammannia. The positions of Decodon and Lythrum in previous analyses were unstable, occurring either at the base of the phylogeny or as they appear in the Morris tree.

The fossil taxa, age, locality, published source, and assessment of credibility are summarized in the Appendix. Most taxa that are merely listed as part of a fossil flora without reference to the original source are excluded from the Appendix unless the location or age of the fossil is exceptional and therefore worthy of notice.

Fossil Genera Attributed to the Lythraeeae

Adenaria Kunth. Adenaria floribunda Kunth is a monospeeific genus ranging from Mexico to Argentina in primary evergreen and secondary forests. Fossil reports are limited to seeds from the Pliocene (ca 3 Ma) of the high plain of Bogota, Colombia that are described as similar to those of Adenaria (Wijninga, 1996a, b). Seed sample T356 from this study is anatomically close, but not identical, to seeds of A.floribunda. The sample compares favorably to Lythraceae seeds by presence of adjacent sclerotic endotestal-lignified exotegmal layers in the seed coat and by a large, nearly encircling mesotestal spongy region (Wijninga, 1996b: pl. 25, figs. 155-157c) such as occurs in some genera of the family. The seeds are variously shaped due to compressed packing in the fruit. Sister genus Pehria Sprague, with approximately the same distributional range and habitats in Central and South America as Adenaria, shares a similar seed structure but differs by a selerenchymatous, rather than a spongy, mesotesta (pers. observ.). The fossil seeds from Colombia are lythraceous and sufficiently similar to be accepted as having close affinity to Adenaria.

Alatospermum Chandler. This seed genus is known from a single occurrence in the early Eocene Lower Bagshot flora, southem England. The seed of Alatospermum lakense Chandler (1962) is small (ca. 0.75-1.55x0.85-1.8 mm), dorsal-ventrally compressed with thin wings of spongy tissue flanking or surrounding a central seed body. The seed body bears a conspicuous long oval/oboval germination valve similar to that found in extinct Decodon and Microdiptera seeds and a similarly positioned raphe, chalaza, and hilum. Tiffney (1981) suggested Alatospermum could be ancestral to the Decodon-Mierodiptera-Mneme complex; this is unlikely given new evidence of a much earlier, Late Cretaceous occurrence of Decodon in North America (Estrada-Ruiz et al., 2009). Chandler raises the question of whether Alatospermum resembles seeds of Pemphis (Chandler, 1962: 120) which also possess wings of spongy tissue. Additional comparative anatomical details of Pemphis are needed to assess the relationship. Current evidence supports the inclusion of Alatospermum in the Lythraceae as an extinct genus of swamps or other aquatic habitats like those in which modern Decodon occurs.

Ammannia L. Ammannia is a wetland genus of ca. 80 species with a world-wide, tropical to temperate distribution and is particularly diverse in Africa and southern Asia. As a result of molecular phylogenetie studies, genera Nesaea and Hionanthera have been subsumed in Ammannia (Graham et al., 2011; Graham & Gandhi, 2013). The oldest putative occurrence of Ammannia is the single fossil seed, Ammannia lakensis M. E. J. Chandler from the early Eocene Pipe-Clay series, southern England (Chandler, 1962, 1964). The seed was compared by Chandler to seeds of living A. japonica Miq. (synonym of A. multiflora Roxb.), but there is little similarity to seeds of that species or any otherAmmannia. The fossil is 1.25 x 0.5 mm, oblong, bilaterally compressed, with a longitudinal raphe asymmetrically placed on one side. A testal flange projects at the chalazal end and the seed surface is ornamented by rounded tubercles (visible from photos of the type collection kindly supplied by M. Collinson). The seeds of extant Ammannia are considerably smaller (ca. 0.3 x 0.3 mm in A. multiflora), obovoid, bilateral, concave-convex, non-tuberculate, with a central raphe and with a convex aerenchymatous float on the adaxial raphal side. The Eocene seed is rejected as representing Ammannia; the affinities ofA. lakensis are unknown.

Wijninga (1996b: pl. 25, figs. 159a, 159b) has compared a seed from the Pliocene/ late Pliocene of Colombia to Ammannia seed. The seed coat, however, is unlike any in Ammannia in having sclerenchymatous inner layers thickened on one side of the seed and mesotestal spongy extensions at each end of the seed. Other seeds from the same sample (Wijninga, 1996b: pl. 25, fig. 159c; pl. 27, figs. 170, 171) have a mesotestal layer filled with large cavities such as are unknown in Ammannia, and the multicellular sclerenchymatous endotesta typical of the Lythraceae is not present. The assignment to Ammannia is rejected.

Fossil pollen of Ammannia from Thailand is listed by Watanasak (1990) as one of several markers for the early to middle Miocene in Thailand. An original source, description, or illustration of the fossil on which the listing is based is not cited. Songtham et al. (2005) also list pollen of Ammannia from the late middle Miocene of Thailand. Sepulchre et al. (2010), who analyzed four pollen assemblages and numerous samples from three distinct areas in Thailand that included lignites, claystone, sandstone, and mudstone deposits of late Oligocene, middle Miocene, and late Miocene age, did not report Ammannia or other lythracean pollen types. Ammannia pollen can be distinguished from other Lythraceae pollen (Graham et al., 1985, 1987, 1990; Graham et al., 2011) and it is not unreasonable to expect its presence in wetland tropical and subtropical fossil floras of southeastern Asia where it occurs today. However, published confirming evidence of the record has not yet been located. Pollen of the genus is also listed as present in Holocene (ca. 6,000 year B.P.) lake sediments from Senegal where Ammannia occurs today (L6zine, 1988). At present, verified documentation for fossil Ammannia older than the Holocene remains to be found.

Chitaleypushpam Paradkar. The extinct flower species Chitaleypushpam mohgaoense Paradkar (1973) from the Deccan Intertrappean Beds at Mohgaon Kalan, India most closely resembles Sahnianthus Shukla, another extinct flower genus from the same locality with a confirmed relationship to the Lythraceae. Chitaleypushpam differs from the better known Sahnianthus by lack of a stipitate ovary, presence of a short style and five-loculed rather than six to eight-loculed ovary, absence of a nectary, and stamens with short filaments of equal length. Pollen in the unopened anthers of Chitaleypushpam resembles but is slightly smaller (12-15 x 19-20 [micro]m) than similar pollen of Sahnianthus (22.5 [micro]m) and of modern Woodfordia (16-22 x 12-16 [micro]m). Given the abundant pollen present in specimens of Chitaleypushpam, an SEM study detailing exine sculpture and pore and colpi characters could help clarify its relationships. Older stages of C. mohgaoense with mature fruits and seeds are unknown. A second species of the genus from the Mohgaon Kalan beds (Kokate et al., 2005) is a nomen nudum. Paradkar (1973) suggested Chitaleypushpam and Sahnianthus might constitute an extinct family ancestral to the Lythraceae. Based on current information, Chitaleypushpam is a tentative but unconfirmed extinct member of the Lythraceae.

Cranmeria Reid & Chandler. This extinct genus of fruits and seeds from the London Clay flora was cited in the protologue as doubtfully related to the Lythraceae. Cranmeria trilocularis Reid and Chandler (1933; Chandler, 1961b), known only from the Sheppey locality, is a three-loculed berry, 17.5-18.5 mm, with 12-15 rows of seeds per locule attached to an axile placenta. Bifurcated rays project into each locule from the placenta. The seeds are ovoid-lanceolate, 1.75 mm, anatropous, with a tuberculed, hexagonal-celled exotesta. The generalized features of the multiloculate ovary with axile placentation suggest Lythraceae, but the rest of the characters rule against the relationship. It is excluded from the Lythraceae.

Crenea Aubl. Crenea is a modern genus of two species ofsubshrubs to small trees in South America. Crenea patentinervis (Koehne) Standley occurs in coastal Colombia and C. maritima Aublet extends from Venezuela along the coast of northeastern South America and southward to Bahia, Brazil in mangroves, on coastal mudflats and beaches, and along estuaries (Lourteig, 1986). The fossil record is based exclusively on pollen of the form genus Verrutricolporites van der Hammen and Wijmstra (1964). Verrutricolporites rotundiporus van der Hammen & Wijmstra (Fig. la) has been interpreted as either the probable precursor of modern Crenea (Fig. lb, c; Germeraad et al., 1968: 334) or as lythraceous with uncertain affinity to Crenea (Muller, 198 la, b). The uncertainty is due to the failure to consistently verify the presence of diagnostic pseudocolpi on the type material (A. Graham, pers. comm.). There is a tendency in fossilization for the grains to fold, obscuring the pseudocolpi and/or pores, and in modern Crenea for the pseudocolpi to vary in development.

Modern Crenea pollen is defined by the presence of six pseudocolpi that are shorter than the colpi and vary from lightly developed to prominent, by a perfectly circular pore at the midpoint of the colpus, and by an exine that ranges from coarsely verrucate or scabrate to nearly psilate (Fig. lb, c; Graham et al., 1985). The pollen of Crenea maritima typically has distinct pseudocolpi and a somewhat verrucate exine, whereas pollen of Crenea patentinervis is more weakly pseudocolpate and the exine more coarsely verrucate (Graham et al., 1985; Lourteig, 1986). Verrutricolporites rotundiporus is accepted as the pollen of Crenea based on the close similarities in pollen shape and exine sculpture. The fossil pollen comes from the same areas of coastal South America where Crenea grows today.

The type of Verrutricolporites rotundiporus is from the Oligocene/Miocene of Guyana (van der Hammen & Wijmstra, 1964). The pollen is common in coastal Miocene deposits of Trinidad where it is associated with pollen of the mangrove genus Rhizophora (Germeraad et al., 1968; Kumar, 1981). It is present in Oligocene/Miocene strata in Colombia and is a zonal marker for the early Miocene in Venezuela where it is "widely distributed in all upper Tertiary sedimentary deposits" (Lorente, 1986: 204). Regali et al. (1974) recovered abundant V. rotundiporus from Oligocene/ Miocene continental shelf deposits in waters off Amapa, Para, Maranhao, Sergipe, and Rio de Janeiro, Brazil, farther south than modem Crenea presently occurs.

Three pollen grains from the middle Eocene of Florida (Jarzen & Dilcher, 2006) are similar to the holotype of Verrutricolporites rotundiporus in shape and in type of exine sculpture but the expected sharply outlined circular pores and pseudocolpi that would verify the determination are not visible due to folding. Preservation of these fossils was in a shallow nearshore marine environment compamble to the modem Crenea habitat. The identification of the three grains as V. rotundiporus is unconfirmed but quite possible.

Two pollen grains determined as Verrutricolporites sp. have been recovered from the middle to late Paleocene (ca. 60-59 Ma) Cerrejon Formation in northem Colombia (Jaramillo et al., 2007). The depositional site is estuarine or on a fluvial coastal plain, habitats typical for extant Crenea. Unfortunately, preservation of the grains prevents viewing the pseudocolpi and the pores that are critical for confirming identity. The very coarsely verrucate exine is extremely coarse even for this notably variable pollen type. The report is unconfirmed at present.

Verrutricolporites rotundiporus also has been reported from late Oligocene of Africa and late Miocene of Nigeria (Germeraad et al., 1968; Legoux, 1978), and in Cameroon in deposits extending from the Miocene to the middle Pliocene (Salard-Cheboldaeff, 1981). Muller (1981a) suggested Verrutricolporites might also be in late Eocene deposits of western Malesia. The presence of Verrutricolporites outside of the Caribbean should be viewed with caution because determinations of fossil pollen as Verrutricolporites in Africa and as Crenea in southeastern Asia and the Pacific are complieated by the presence of a complex of very similar, non-pseudocolpate pollen types referred to Florshuetzia, a pollen form genus proposed by Morley as a probable precursor to the living Asian genera Sonneratia and Lagerstroemia among other genera (see details under Florshuetzia and Sonneratia). Morley found some Verrutricolporites identical to some variants of Florshuetzia semilobata Germeraad et al. (Morley, 2000:216-220). The hypothesis that Verrutricolporites in Africa is more likely pollen of the Florshuetzia/Sonneratia lineage than related to Crenea is supported by the presence of fossil wood of Sonneratia in the Eocene of northern Africa (see Sonneratia). Crenea has not yet been included in a molecular phylogenetic study and consequently the modern relationships of the genus, whether near Sonneratia, for example, are unknown.

The oldest occurence of Crenea-type Verrutricolporites rotundiporus pollen is from the late Eocene of South America (Germeraad et al., 1968: fig. 7, 15). This is accepted, with some uncertainty, as the time of origin of modern Crenea.

A second species, Verrutricolporites pachydermis Sun & Kong is a thick-walled pollen type from the Paleocene of southern China (Sun et al., 1980; Liu & Yang, 1999) and Tibet (Li et al., 2008). It appears from illustrations to be pollen of Lagerstroemia but further comparisons need to be made (Liu et al., 2008; pers. observ.).

Cuphea P. Browne. Cuphea, the largest genus of the Lythraceae with ca. 240-250 herbaceous mesophyllous species, is endemic to the New World, with centers of diversity in Mexico and Brazil. The genus is adapted to a wide variety of habitats, including savannas, rocky cerrados, swamps, and river margins. There is a substantial fossil record based on Cuphea pollen and as the pollen form genus and species Striasyncolpites zwaardii Germeraad et al. (1968). Modem Cuphea pollen is among the most distinctive and diverse in the family (Coz Campos, 1964; Patel et al., 1984; Graham & Graham, 1971b; Graham et al., 1985). The most common fossil and modem forms are oblate, trisyncolporate with slightly to prominently protruding pores and a finely to coarsely striate exine. The extensive diversification of the modem pollen in shape, size, number of pores, degree of pore protrusion, colpi fusion, and exine sculpture is, to some extent, also apparent in the fossil record (Fig. 2a-e). Fossil pollen can be matched to modem species groups and to some taxonomic sections of the genus.

The oldest reliably identified occurrences of Cuphea pollen are from Chiapas, Mexico (Fig. 2a; Palacios & Rzedowski, 1993) with an unsecured age of early to middle Miocene (ca. 14 Ma), and from the late middle Miocene ofTrinidad (Germeraad et al., 1968). The Chiapas grain has slightly protruding potes and a finely striated exine, similar to living Cuphea cyanea DC, a species of Cuphea section Diploptychia Koehne that occurs today in Chiapas (cf. Graham, 1998b: fig. 12). Pollen of Striasyncolpites zwaardii, from the late middle Miocene to Recent of Trinidad (Fig. 2d; Germeraad et al., 1968) is similar to the Chiapas fossil bur more triangular in outline. Its affinity is also with section Diploptychia rather than near the North American species C. aequipetala Cav. (sect. Heterodon Koehne) with which it was originally compared. Striasyncolpites zwaardii is also recorded from Miocene to Pliocene sediments of the northeastern Brazilian continental shelf(Regali et al., 1974).

A Cuphea pollen grain from late Miocene deposits in Alabama is a type found in Cuphea sections Euandra Koehne and Brachyandra Koehne, and is close to modem C carthagenensis (Jacq.) Macbr., C pseudosilene Griseb., and C sperrnacoce A. St. Hil. (Fig. 2b; Graham & Graham, 1971b; Graham, 1998a). The grain comes from deposits underlying the Pliocene Citronelle Formation that are probably equivalent to the Miocene Pascagoula clay (Isphording & Lamb, 1971; E. Leopold, pers. comm., 2011). Another type, similar to C Jerrisiae Bacig. (sect. Heterodon) occurs in middle Pliocene deposits in Veracruz, Mexico (Fig. 2c; Graham, 1976; Akers, 1979; Graham, 1988: fig. 14).

Increased diversification in pollen morphology and more frequent occurrences of fossil Cuphea pollen are apparent in Pleistocene and Holocene deposits in Mexico and South America (Graham & Graham, 1971a: figs. 1-9). Pollen from the middle Holocene of southern Brazil (Fig. 2e; Behling, 1997) and from the Holocene (Macedo et al., 2009) and the Quatemary of Rio Grande do Sul, Brazil (Scherer & Lorscheitter, 2009) are similar to living C. carunculata Koehne and C urbaniana Koehne (sect. Euandra) of the region. Pleistocene-Holocene pollen from Colombia compares well with C. dipetala (L.f.) Koehne (sect. Diploptychia; Hooghiemstra, 1984; Van der Hammen & Gonzalez, 1960). A number of studies list fossil Cuphea from Mexico (Ramirez-Arriaga et al., 2008, Miocene-Pliocene; Straka & Ohngemach, 1989, Pleistocene); from Cuba (Moncada et al., 1990-1991, Pleistocene); from Brazil (Whitney et al., 2011, Pleistocene and Holocene; Barberi et al., 2000, Pleistocene & Holocene; De Oliveira et al., 1999, Holocene); from Colombia (Berrio et al., 2002, Quatemary; Wymstra & van der Hammen, 1966, Holocene); and from the Galapagos (van Leeuwen et al., 2008, Holocene).

Oligocene pollen from the Isle of Wight, England identified as Cuphea (Machin, 1971) is incorreetly assigned (Graham & Graham, 1971a). Seven species of the pollen form genus Lythraites Yu et al. (Song et al., 2004) from late Cretaceous to Tertiary palynofloms in two Chinese provinces have been referred to Cuphea. The grains are neither Cuphea nor lythraeean. Pollen of Syncolporites subtilis Boltenhagen (1976) from the late Cretaceous (Coniacian) of Gabon was compared to Cuphea by Muller (1981a) who rejected the relationship.

Leaf impressions of Cuphea antiqua Britton (Britton, 1892; Berry, 1917, 1939a) from the Miocene or Pliocene of Potosi, Bolivia (Graham, 2010a: 391) could represent Cuphea or Diplusodon in the Lythraceae, or possibly some other member of the Myrtales. As was advised for the Myrtaceae (Manchester et al., 1998), caution is needed when relating isolated fossil leaves to modem genera of the Lythraceae because of the variability in structure within a genus and the numerous similarities in pattern shared with other genera or families. The brochidodromous venation on which the identification of C antiqua depends appears throughout the order in a number of genera that occur in Andean Bolivia today (Hickey & Wolfe, 1975); acceptance of the fossil leaf as Cuphea, therefore, is possible but unconfirmed.

In summary, Cuphea is well represented in the pollen fossil record beginning in the early Miocene/middle Miocene and it diversified extensively from the Pleistocene to the present exclusively in the New World. The variety of palynomorphs and the phylogenetic history of the genus are indicative of relatively recent diversification of the genus on both the North and South American continents (Barber et al., 2010).

Decodon J. F. Gmel. is a monospecific genus of eastern North America with D. verticillatus (L.) Ell. occurring in fresh water swamps and along lake margins. It has one of the most extensive fossil records in the Lythraceae and is especially well represented by seeds but also by leaves, stems, roots, and pollen. No fossil Decodon flowers are known. Seed shape in living D. verticillatus varies widely depending on the amount of crowding and compression of maturing seeds within the capsule. The extensive variation in seed shape within the living species makes comparisons with fossil species and their species delineations difficult. The number of fossil seed species is unquestionably overdescribed.

Until recently, the oldest known Decodon fossils in continental North America were fruits and seeds from the middle Eocene Princeton Chert, British Columbia (Decodon allenbyensis Cevallos-Ferriz & Stockey, 1988; Cevallos-Ferriz et al., 1991) and the middle Eocene Clamo Formation Nut Beds, Oregon (Decodon sp., Manchester, 1994) and leaf fragments from the early Oligocene Bridge Creek Flora, Oregon (Decodon brownii Meyer & Manchester, 1997). In 2009, the record of the oldest occurrences was over-turned by the discovery of three Decodon seeds in the Late Cretaceous Cerro dei Pueblo Formation, Coahuila, northem Mexico (Rodriguez-de la Rosa et al., 1998; Estrada-Ruiz et al., 2009). Named Decodon tiffneyi Estrada-Ruiz et al., the seeds compare favorably to those of modem D. verticillatus (Fig. 3; Table 2). This discovery extends the earliest record of Decodon back to the Late Cretaceous (late Campanian, 73.5 Ma; Eberth et al., 2004) and the past geographic range southward from northwestern North America to northern Central Mexico. Decodon tiffneyi is the second oldest known fossil occurrence of the Lythraceae, after the recent discovery of Lythrum/ Peplis pollen in the early Campanian (82-81 Ma) Eagle Formation of northwestern Wyoming, USA (Grimsson et al., 2011; see Lythrum).

In North America, Deeodon was present in the high northern latitudes of the warm Arctic/Subaretie by the late early Miocene. A review of Neogene macrofossils from North America includes records of Miocene Decodon from Alaska and numerous Decodon seeds from the Northwest Territories (Banks Island, Prince Patrick Island, and Ellesmere Island) at latitudes between 66[degrees] and 76[degrees] N (Matthews & Ovenden, 1990). Tentative age correlations place the oldest Decodon seeds from the Northwest Territories in early Miocene (18 Ma) in the Mary Sachs gravel at South Banks Island and West River (Horton River). The Mary Sachs Gravel flora is one of the most diverse North American Arctic floras comprising many of the same genera that occur in fossil floras of eastern Siberia, including Microdiptera, an extinct seed genus related to Decodon (Matthews & Ovenden, 1990: 375). Decodon is found at slightly higher latitudes and more easterly longitudes in Pliocene deposits of North Banks Island (5 Ma) and Prince Patrick Island, and in the middle Pliocene (3 Ma) of Ellesmere Island at ca. 78.5[degrees]N, 82.5[degrees] W. In the northeastern United States, Decodon was not found in the early Miocene Brandon lignite, although Microdiptera is present (Haggard & Tiffney, 1997; Tiffney, 1981). Decodon pollen is recorded in the late Miocene at high Atlantic-European latitudes in Iceland (Denk et al., 2011).

In Europe, Decodon is listed from late Paleocene deposits in southern England (Collinson, 1986) but the determination is now uncertain (Collinson, pers. eomm., 2011). Decodon-like seeds reported from the Eocene at the Paleocene-Eocene boundary (ta. 55 Ma) in Belgium (Fairon-Demaret & Smith, 2002) are too poorly preserved for positive identification. Verified D. gibbosus E. M. Reid is described from the late Eocene Barton Beds (Chandler, 1960) in England and D. vectensis Chandler from the Oligocene Upper Headon Beds on the Isle of Wight (Chandler, 1963a). Five species are reeognized from Oligocene deposits world-wide: D. brownei from the early Oligoeene of Oregon (Meyer & Manchester, 1997), D. sibiricus Dorofeev and D. tavdensis Dorofeev from the Oligocene/Miocene of Russia (Dorofeev, 1977), and D. vectensis (Chandler, 1963a) from England and D. gibbosus (Mai & Walther, 1991) from Germany. The genus subsequently became more common and more diverse throughout the Miocene and Pliocene in Europe and Asia (Dorofeev, 1960, 1977; Tiffney, 1981; Mai, 1985, 1995; van der Burgh & Zetter, 1998; Grimsson et al., 2012). Tiffney (1981) has provided a substantial review of the fossil history of the genus in a study of the early Miocene Brandon Lignite flora of Vermont (Tiffney, 1981, 1994). He lists 13 species of fossil Decodon seeds and Grimsson et al. (2012) bring the world record up-to-date, recording 21 species. Species in Europe and Asia based on seeds are described from: Oligoeene and Miocene fossil sites in Russia (Dorofeev, t963, 1977); the Miocene of Denmark, Germany, Italy, Poland; and the Plioeene/Pleistocene of Germany and Italy (Szafer, 1961; Friis, 1975, 1979, 1985; Girotti et al., 2003; van der Burgh, 1987; Mai & Walther, 1978). Additional reports or listings of Miocene to Pleistocene Decodon seeds from Eurasia are compiled in Dorofeev (1977), Mai (1995, 2001), and Grimsson et al. (2012) and are included in the online collaborative site, Paleobiology Database (http://www.paleodb.org.).

Three morphologically-based sections of fossil seed species in Decodon were recognized by Dorofeev (1977): sect. Vectensis ranged from the Oligocene to the middle Miocene; sect. Gibbosus from the late Eocene to the end of the Miocene; and sect. Globosus from the Oligocene to the early Pleistocene of Europe (Mal, 1964). The latter section was widely represented in Eurasia in the Miocene but the only member surviving to the present is D. verticillatus, now restricted to eastern North America. Mai regarded D. gibbosus E. M. Reid and D. globosus E. M. Reid as extremes of the same species (Mai & Walther, 1978), an idea supported by Tiffney (1981). The Late Cretaceous seeds of Decodon tiffneyi are considered morphologically near D. gibbosus but more strictly obpyramidal with a shorter germination valve and a well-developed flexure. They are also said to "compare closely to" D. verticillatus (Estrada-Ruiz et al., 2009: 285). The seed anatomy of D. tiffneyi is unknown.

Although there are abundant seed records of Decodon, fossil pollen of the genus has not been widely reported. It is recognized in the middle Eocene Princeton Chert, British Columbia, in late Miocene deposits in Iceland where it was part of a once widespread, now restricted or partly replaced northern hemisphere flora (Grimsson et al., 2012; Denk et al., 2011), and in the lower Pliocene in Germany where it occurs with Decodon seeds (van der Burgh & Zetter, 1998). Late Quaternary (Holocene) pollen of D. verticillatus is recorded from the southeastern United States in Tennessee and Alabama (Delcourt, 1979, 1980).

The pollen form genus Lythraceaepollenites from Germany and Poland is accepted as belonging to the Lythraceae but the relationship to Decodon of Lythraceaepollenties decodonensis Stuchlik from the Miocene of Poland remains to be confirmed, although it is likely. The species L. bavaricus Thiele-Pfeiffer (1980) is accepted as lythracean. The species L. minimus Thiele-Pfeiffer (1988) has not been reviewed. Modem Decodon pollen can be difficult to separate from that of Lagerstroemia without benefit of SEM observations (see comments under Lagerstroemia) but the historical ranges of the two genera appear to be distinct. A Pliocene pollen grain from Germany (Menke, 1976) attributed to "cf. Lagerstroemia" may represent Decodon (Muller, 1981 a).

The record of Decodon based on attached and detached leaves includes: D. allenbyensis, middle Eocene, British Columbia (Cevallos-Ferriz & Stockey, 1988); D. brownii, Oligocene, Oregon, Meyer and Manchester (1997); D. gibbosus, early Miocene, Czech Republic (Kvacek & Sakala, 1999); D. alaskana, Miocene, Alaska, (Wolfe & Tanai, 1980); and Decodon sp., late Miocene, Austria (Kovar-Eder et al., 2002, listed). Decodon leaves are possibly also represented in the Miocene of California (Condit, 1938; Wolfe & Tanai, 1980).

Abundant remains from two fossil sites allow for nearly complete plant reconstruction of two fossil species of Decodon. Decodon allenbyensis from the middle Eocene, British Columbia is described from associated roots, stems, leaves, fruits and seeds (Cevallos-Ferriz & Stockey, 1988; Little & Stockey, 2003, 2006). Decodon gibbosus Nikitin emend. Kva6ek & Sakala from the early Miocene of the Czech Republic (Kvacek & Sakala, 1999) is described from twigs with attached leaves, fruits and seeds.

The nearly global range of Decodon in the northern latitudes during the Miocene became severely restricted as climates deteriorated in the Pliocene and Pleistocene. A report from the middle Pleistocene in the Netherlands cited by Grimsson et al. (2012) represents the youngest record of the genus in western Europe. Mai considered the genus extinct in Eurasia by the early Pleistocene (cf. Mai, 1964; age of the Tegelen fossils is Pleistocene, not Pliocene). There appear to be no records of Decodon in Asia in the Quatemary. Decodon successfully retreated southward during glacial periods of the Pleistocene only in eastem North America where the north-south trending Appalachian Mountains presented no barrier to migration. Southward migration in Western Europe to warmer climates was prevented by the east-west orientation of the Alps and the Pyrenees. Decodon verticillatus persists in Eastem North America to the present, occurring from southern Canada to Florida in the east and from Minnesota to Texas in the west (http://plants.usda.gov). See Grimsson et al. (2012) for a an extensive table and map of Decodon collections arranged by age.

Matsumoto et al. (1997) have proposed a biogeographic scenario for the spread of Decodon in the Eocene from Western North America into Western Europe by eastward dispersal across North America and over the Thulian-Greenland and DeGeer route land bridges. Decodon seeds are recorded in the late Eocene in southern England and on the continent in Germany (Chandler, 1960; Mai & Walther, 1978, 1985), suggesting that dispersal from North America was accomplished, possibly by the southern Thulian route, as Matsumoto suggests. At high northern latitudes, fossil seeds of Decodon are not recorded in eastern North America until the early Miocene (Matthews & Ovenden, 1990). In Asia, in southern Siberia, seeds of the genus appear in the Oligocene following closure by early Oligocene of the Turgai Strait which had earlier separated western and eastem Eurasia in northern latitudes (R6gl, 1999). By late Miocene seeds and some pollen are widely reported across Western Europe and Iceland and are found in Central Asia and Siberia (Grimsson et al., 2012). At the high latitudes in the late Miocene, Decodon could have been dispersing from North America via the more northerly DeGeer route into Europe and could have come as well at that time into Western Europe from Western Asia. In the Far East, Decodon is recorded from the Miocene in northern Siberia, northeastern China, and Japan (Grimsson et al., 2012), a distribution seemingly derived from Miocene introductions from western North America (Alaska) that were subsequently halted by the decline in temperatures of the mid-late Miocene cooling period (ca. 15-5 Ma) and that extirpated the genus in the area by the Pliocene. There is a gap in the Decodon fossil record in northern Russia between ca. 90o-125[degrees] E longitude which, if representing actual absence, would support bi-directional introduction of Decodon eastward into northern Europe from northern Atlantic sources beginning in the Eocene and westward into the Far East from North America in the Miocene. Matsumoto et al. (1997) describe the late middle Miocene Japanese species Decodon mosanruensis as more similar to the Eocene D. allenbyensis of British Columbia than to the Miocene Decodon species of Eastem Siberia, implying that Decodon had migrated westward across the Beringian land bridge by middle Miocene, as well as eastward into Europe.

The late Campanian fossil of Decodon tiffneyi in northern Mexico demonstrated that the genus was considerably older than previously known. Estrada-Ruiz et al. (2009) suggested that the genus originated in south-central North America based on this find. The early presence of Decodon in the lower latitudes of western North America does provide an anchor for a hypothesized extensive northward migration of the genus in the warm Paleocene/Eocene along the north Pacific coast and later eastward along the Bering land bridge to eastem Asia. Given the latest finds of fossil Lythraceae which are even older than D. tiffneyi (see Lythrum/Peplis and possibly Sonneratia turonicum), it is clear that the family had extensively diversified and was widespread much earlier than previously recognized. These recent findings demonstrate that an understanding of the origins, radiation and diversification of Decodon and of all the Lythraceae is still very much a work in progress.

A selection of fossil Decodon records published since 1981 is provided in the Appendix. The entries are supported by descriptions and/or illustrations that verify their identity, with a few exceptions where a listing only was included because of the unique age or geographic site of the fossil. An extensive listing that is more comprehensive for the Miocene of Europe and Western Asia than is cited here is included in the study by Grimsson et al. (2012).

Duabanga Buch.-Ham. This is a genus of two or three species of primary rain forest or early successional trees in southeastern Asia and the southern Pacific. Duabanga grandiflora Walp. extends from India to Myanmar, Yunnan, China, and Indochina. Duabanga moluccana B1. occurs from Java and Borneo to the Philippine Islands and south to New Guinea. A third species, Duabanga taylorii Jayaweera, is known only as a cultivated tree presumed once native to Java (Jayaweera, 1967). Duabanga, Lagerstroemia, and Sonneratia are the three Asian tree genera of the Lythraceae. Ali have fossil records in Asia based on wood remains and can be distinguished from one another by a suite of wood anatomical features. Major wood anatomical character states that separate them are listed in Table 1.

The characters diagnostic of Duabanga wood are: diffuse, predominantly solitary vessels (multiples occasionally up to 3) that are fewer, larger, and longer than in Sonneratia; abundant axial paratracheal aliform parenchyma; commonly biseriate, strongly heterogeneous rays with erect and square to procumbent cells with crystals; and non-septate fibers. Fossil Duabanga wood is referred to the form genus Duabangoxylon Prakash and Awasthi (1970). Duabangoxylon tertiarum Prakash and Awasthi (1970) and D. indicum (Navale) Awasthi (Awasthi & Prasad, 1987) are described from the Miocene and the Miocene/Pliocene of southeastern India. Duabanga tertiarum is also known from the Tertiary of Java (Kramer, 1974). The species Duabangoxylon bifarium Gottwald (1994) appears to be a minor variant of D. tertiarum. It is described from silicified material from an unknown source on erosion river terraces in the Chindwin Basin, Myanmar with an age "in ali probability" of middle Miocene. It is said to differ from D. tertiarum which occurs in the same deposit by biseriate rays, some isolated septate fibers, and larger vessel pits. Duabangoxylon pakistanicum Ahmed et al. (1991) is described from the Pleistocene to Subrecent of Pakistan. Vessel, wood parenchyma, and ray attributes are either the same as or overlap those of D. tertiarum. The earliest confirmed fossil records of Duabanga are woods of Duabangoxylon tertiarum and D. indicum from the middle Miocene of India and D. tertiarum from the Miocene of Myanmar.

The fossil fruit Duabangocarpon deccanii Kadoo and Kolhe (2002) is a multilocular capsular fruit from the Maastrichtian-Danian Deecan Intertrappean Beds at Mohgaon Kalan. It has some generalized characters of Lythraceae fruits, i.e. possible enclosure in a persistent floral tube, a superior ovary with axile placentation, eight locules, loculicidal dehiscence, and possibly two rows of seeds (described as two seeds in each locule, but serial diagrams suggest the condition may be two or more rows of seeds per locule), and an undifferentiated seed coat. Unlike Lythraceae fruits, the sepals of Duabangocarpon are unequal and imbricate, not valvate, and the seed is contradictorily described as having endosperm and as exalbuminous. Lythraceae seeds lack endosperm at maturity. Duabangocarpon deccanii is neither related to Duabanga nor to the Lythraceae. The genus is part of a diverse group in the Deccan of fossil multilocular capsules with axile placentas and numerous small seeds. The near familial relationships of most are tentatively assigned to select angiosperm families but cannot be associated with certainty to any modem genus. They represent extinct members of the rich, early angiosperm flora preserved in the Deccan Intertrappean Beds at the beginning of the Tertiary.

Leaves with some Duabanga-like features from the middle Eocene Princeton Chert, British Columbia are assigned to Decodon allenbyensis based on the extensive associated context of stems, roots, fruits, and seeds of that species (Little et al., 2004). The presence of abaxial epidermal mamilliform papillae on the fossil leaves suggested a relationship to Duabanga (Keating, 1984), whereas the absence of leaf sclereids, diagnostic in the Lythraceae only of Duabanga and Sonneratia leaves (Bannerjee & Rao, 1975; Rao & Das, 1979), mies against it.

Enigmocarpon Sahni. Hundreds of silicified fossil fruits from the Deccan Intertrappean Beds of India at Mohgaon Kalan, Madhya Pradesh, India represent multiloculate capsules of the extinct Enigmocarpon parijai Sahni (Sahni, 1934, 1943; Sahni & Rode, 1937). The fruits are associated in the same cherts with Sahnianthus flowers (Fig. 4; Shukla, 1944). Sahnianthus is the oldest known lythracean type flower, with a probable age of early Paleocene (Danian; see Sahnianthus).

Enigmocarpon is a robust, thick-walled capsule, relatively large at 11-18x 13-14 mm compared to all other fossil Lythraceae fruits except Shirleya (10x 11.5-12.5 mm). The fruit has axile placentation, seven to 12 locules, loculicidal dehiscence, septa attached to the valves at dehiseence, and ca. six to eight seeds per locule alternating in two rows (Fig. 4c). The seeds, which are remarkably well preserved, are anatropous, ca. obpyramidal, compressed in the locules, 1.5-3 mm long, and have a seed coat largely composed ofa dense, spongy mesotesta of thin-walled cells. Interior to the mesotesta the seed cavity is "lined with very narrow elongated thick-walled cells" (Sahni, 1943: 62), suggesting the presence of the longitudinally elongated tracheoid exotegmen diagnostic of seeds of the Lythraceae, Onagraceae, and Combretaceae (Comer, 1976: 176). The raphe is large and expanded at the chalazal end of the seed, a hypostase is present, endosperm is apparently free nuclear (Biradar and Mahabale, 1976), and the cotyledons of the embryo are straight. These features are part of the character suite that defines most Lythraceae seeds. Additional characteristics of the capsule are provided by Sahni (1943) and the embryology is described by Mahabale and Deshpande (1957) and Biradar and Mahabale (1976). A second species, E. sahnii Chitaley and Kate (1977) is only slightly smaller ( 11 vs 16 mm long) and more orbicular. The fruits have been restudied together with Sahnianthus by Manchester and Kapgate (2011) who provide new insight and excellent illustrations of the genera.

Sahni regarded Enigmocarpon as unquestionably belonging to the Lythraceae and considered Decodon and Heimia the "only genera admitting of a serious comparison" (Sahni, 1943: 82). The seeds of Enigmocarpon share with extant Decodon, and also with Lawsonia and Pemphis, an obtrigonal shape and a spongy mesotesta that buoys the seeds in water. Decodon seeds differ in having a germination valve and they lack the exotestal papillae of Enigmocarpon seeds. The papillae, apparently formed by the external wall of the exotestal cells, are not homologous to inverted trichomes of the seed coat that occur in 14 genera of the Lythraceae. Phylogenetically, Decodon, Lawsonia, and Pemphis are distantly related (Graham et al., 2005). Mesotestal float tissue evolved independently in these genera and cannot be invoked as support for a close relationship to Enigmocarpon. Seeds of the American genus Heimia are not at all comparable to Enigmocarpon seeds, having a relatively simple seed coat with a one to three-layered mesotesta of thin-walled cells and no papillae (pers. observ.).

Enigmocarpon has also been related to Sonneratia (Mahabale & Deshpande, 1957; Mehrotra, 2003). The six-loculate to multiloculate capsule of Sonneratia and its Afro-Asian geographic distribution support this idea. However, embryos of Sonneratia, unlike those of Enigmocarpon, have contorted cotyledons (Comer, 1976). Fossil seeds described as Lawsonia-like from the middle Eocene of British Columbia are within the same size range as Enigmocarpon, but lack the lythracean narrow, elongated, fibrous exotegmen (Cevallos-Ferriz & Stockey, 1988). A compafison of seleeted seeds from extinct and modem genera of the Lythraceae similar to Enigmocarpon is made in Table 2.

Shukla (1947) first reported the co-occurrence of intermediate stages between the young ovary of Sahnianthus and the mature fruit of Enigmocarpon. Recent reinvestigations support this connection (Manehester & Kapgate, 2011). Sahnianthus and Enigmocarpon assuredly represent parts of the same plant type given the similarity in morphology and the developmental continuum of fossil remains. The closest modem descendents of the genera appear to be in a clade nearKoehneria and Woodfordia, genera that share with Enigmocarpon and Sahnianthus a stipitate superior ovary, multicellular non-secreting trichomes, and similar, non-pseudocolpate pollen. Koehneria and Woodfordia, however, produce anatomically simpler seeds that lack a spongy mesotesta.

Eotrapa Miki. Eotrapa (Miki, 1961, 1967) is a genus of fossil flowers from middle Miocene deposits (Makinouchi, 1985; Momohara, 1994, 1997) at two localities in Central Hondo, Japan. The flowers were initially described as Lythrum tetrasepalum Miki (1959; Kovar-Eder et al., 2002) and proposed as ancestral to the water chesnut genus Trapa by way of a putative intermediate form, the extinct genus Hemitrapa Miki (1959). In 1961, Miki decided the flowers were not Lythrum although they were still on the evolutionary trajectory to Trapa and he transferred them to a new genus, Eotrapa. The distinctive features of Eotrapa not present in Lythrum or any other member of the Lythraceae were two 'stipules' or 'frontal appendages' on each sepal. These Miki (1961) hypothesized coalesced through time to form the characteristic spines or horns of Hemitrapa and Trapa. Following study of additional materials, Miki (1967, 1968) found five flagella-like appendages on the flower of Eotrapa and no frontal appendages, and discovered the fruits contained two or three orbicular seeds with a germination lid. He decided that Eotrapa was not related to Hemitrapa or Trapa but was closest to the modem aquatic genus Trapella (Plantaginaceae, formerly Trapellaceae, Stevens, 2001), fossils of which occur in abundance in the same deposits. Living species of Trapella have either three or five long appendages; fossils of Trapella from the Miocene of western Siberia and Pliocene of Europe often have three appendages (Tralau, 1964). The overall impression from photographs of the fossils (Miki, 1967) supports Miki in the decision to equate Eotrapa with Trapella. The taxonomic placement of Eotrapa might be finally settled by SEM comparisons of the seed with possible relatives. As presently understood, the affinities of Eotrapa are not with any member of the Lythraceae.

Florshuetzia Germeraad et al. (1968) is the pollen form genus of several species known mainly from abundant, varied fossil grains from the tropics of Southeast Asia, and also from the Paleocene of southern France (Fig. 5). According to Muller and Morley (Germeraad et al., 1968; Muller, 1978, 1984; Morley, 2000: cf. 219, fig. 9.37), Florshuetzia is at the base of a complex of emerging pollen types that transitioned to pollen forms of Lagerstroemia, Sonneratia, possibly Trapa and other genera, and to modern species of Sonneratia. The oldest occurrence of accepted Florshuetzia pollen is from the late Paleocene Thanetian (58.7-55.8 Ma) of south central France which was then at the southern border of the Tethys (Fig. 5b; Gruas-Cavagnetto et al., 1988; Plaziat et al., 2001). I have not been able to determine the species of this fossil published as Sonneratiaceae with any certainty based on the light microscope photographs of isolated grains but the pollen appears to be an early form of Florshuetzia lacking the distinctive polar caps, prominent mesocolpal ridges, and the well-defined verrucate equatorial belt of later Florshuetzia species.

Florshuetzia is abundant and structurally highly varied in deposits in southeastern Asia beginning in the Eocene (Muller, 1981b; Guleria et al., 1996; Morley, 2000). The oldest are morphologically more diverse than later ones. The earliest species in Asia is F. trilobata Germeraad et al., the species considered by Muller and Morley to be precursor to both Sonneratia and Lagerstroemia (Germeraad et al., 1968; Morley, 2000). Florshuetzia trilobata (Fig. 5a) is a subprolate, thick-walled, triporate grain with colpoid grooves and meridional thickenings. The pores are circular and distinct, and the exine is psilate or covered with broken separate verrucae. Modem Sonneratia and Lagerstroemia pollen are typically tricolporate and may be three-pseudocolpate (Sonneratia) or six-pseudocolpate (Lagerstroemia) although the prominence of the pseudocolpi varies, as does the exine sculpture in the pollen of both genera. Variants of Florshuetzia approach modem Sonneratia and Lagerstroemia pollen in the development of the colpi and exine sculpture. Early Florshuetzia has been confused with some variants of Verrutricolporites (see Crenea) in which the colpi of the Verrutricolporites grains are so indistinct as to appear missing or the typically verrucate exine is reduced to nearly psilate, thus approaching Florshuetzia, except in its smaller size. Morley has speculated that Verrutricolporites, Lagerstroemia, and Sonneratia share an early history with Florshuetzia (Morley, 2000: fig. 9.37A).

Florschuetzia trilobam occurs with Lagerstroemia-like pollen in the middle Eocene Nanggulan Formation of Central Java (Morley, 2000). It appears in deposits of late Eocene age in Assam (Handique, 1993) and Kalimantan (Morley, 2000), in the Oligocene and Miocene of Borneo (Muller, 1981a), and in the mid-Tertiary of the Red Sea and Nile Delta (Legoux, 1978 fide Morley, 2000). It is not present in the Late Cretaceous to Eocene formations of northwest Borneo (Muller, 1968). Florsehuetzia trilobata is a component oflacustrine fresh-water swamp sediments from the Oligocene in Vietnam, Thailand, and Cambodia, and from the early Miocene of Malaysia and Indonesia. Only later in the Miocene does the genus occur with brackish and salt-water mangrove vegetation, indicating that Florshuetzia was initially wide-spread in fresh water in southeast Asia and not restricted to brackish waters or mangrove vegetation as the probable derivative genus Sonneratia is today (Morley, 2000). Muller found that Florschuetzia by the Miocene coexisted with extinct Sonneratia pollen types and speculated that it was ancestral to and was eventually replaced by the modern species of Sonneratia in mangrove habitats (Muller, 1978, 198 lb, 1984).

Muller (1984: 437) provided a diagram comparing the distribution of four species of Florshuetzia through time (Fig. 6). Following the Eocene appearance of F. trilobata in Java, E semilobata Germeraad et al. occured at the base of the Miocene in Borneo (Muller 1981 a), F. levipoli Germeraad et al. appeared shortly after in the early Miocene, and F. meridionalis Germeraad et al. in the middle Miocene. Transitional types were also present. Florschuetzia trilobata became extinct by middle Miocene. Florshuetzia levipoli (Fig. 5c) was considered equivalent to pollen of modern Sonneratia caseolaris, and F. meridionalis (Fig. 5d) equivalent to modem S. alba J.Smith, Florshuetzia trilobata, F. semilobata, and F. levipoli are listed as coexisting in the Miocene of Thailand (Watanasak, 1990). Florshuetzia levipoli and F. meridionalis co-occurred in middle Miocene (ca. 16 Ma) of Borneo (Muller, 1978, 1981a, 1984) and in the late Miocene to Recent of Papua, New Guinea (Khan, 1974), and E reticulata Sohma is described from the Quatemary of Sulawesi (Sohma, 1973). The SEM images of F. claricolpata Yamanoi from the middle Miocene of Japan (Yamanoi, 1984) closely resemble modem Lagerstroemia pollen (e.g. see Kim et al., 1994: fig. 4A). Other Florshuetzia ranging from middle Miocene to the present have been referred to by informal names.

Morley (2000: 218), speculating on the earliest evolutionary relationships and historical dispersai pattems among Florschuetzia and modem pollen of Sonneratia, Lagerstroemia, Duabanga, and Trapa, envisioned a common ancestral form that originated in the Western Tethys region. He hypothesized that the ancestor of the four modem genera initially diversified as two lineages, one of which remained in the Tethyan Old World and a second that dispersed to South America, later dispersing back to West Africa. The speculative diagram presented by Morley (2000: 219, fig. 9.37) clearly illustrates the difficulties in resolving early relationships among the several genera based on the complex fossil pollen record. Although the account is theoretical, it is noted that the four modem Asian genera of the Lythraceae are monophyletic in molecular phylogenetic studies with Sonneratia + Trapa strongly supported as sister to Lagerstroemia + Duabanga (Graham et al., 2005), lending credence to early connections among them suggested by commonalities in the fossil pollen.

Heimia Link. This is a New World shrub genus of stream margins and other wet places with three morphologically similar species. The most widely distributed, H. salicifolia Link, ranges from Texas to Argentina whereas the others have narrow ranges, H. montana (Griseb.) Lillo in southern Bolivia and Argentina and H. apetala (Spreng.) S.A. Graham & Gandhi in southern Brazil. A single fossil flower, Antholithus heimiaformis Berry from the Miocene of Cuba was described as either lauraceous or "some such flower as in the genus Heimia" (Berry 1939b: 133). From the illustration, the flower is not Heimia. It has large sepals (petals fide Berry) with a venation pattern unlike that in Heimia and lacks the conspicuous long epicalyx segments diagnostic of the genus. Heimia is unknown in the fossil record. It is also not part of the modem Cuban flora.

Hemitrapa Miki. Hemitrapa is an extinct genus once distributed in Eurasia and North America and related to, possibly ancestral to, the genus Trapa. A detailed discussion of this genus is presented under Trapa.

Lafoensia Vand. A New World genus of five or six species of shrubs to tall trees, Lafoensia is native to South American savannas and cerrados, with one species, L. punicifolia DC. extending north as far as southern Mexico (Lourteig, 1985). Lafoensia has a limited and recent fossil record. Pollen grains from the Holocene of Panama are figured by Bartlett and Barghoorn (1973) and the genus is listed from the Pleistocene and Holocene of central Brazil (Barberi et al., 2000).

Lagerstroemia L. Lagerstroemia comprises ca. 56 species of trees and shrubs of lowland swamps, and montane and secondary forests. Distribution of the genus extends across southern Asia from India, Myanmar and southern China to Indonesia, eastward to northern Australia and north to Japan (Furtado & Srisuko, 1969). Lagerstroemia together with tree genera Sonneratia and Duabanga and the herbaceous aquatic Trapa form a subclade of one of the two major clades of the Lythraceae. Phylogenetically, Lagerstroemia is strongly supported as sister to Duabanga based on multi-gene data (Graham et al., 2005; Morris, 2007).

The fossil record consists of leaf impressions, wood, pollen, and possibly fossil fruits. The two oldest leaf impressions attributed to Lagerstroemia are L. patelii Lakhanpal and Guleria (1981, 1984) from the well-known Panandhro Lignite Mine at Kutch, western India which is early Eocene (Lakhanpal et al., 1984) or late Paleocene/Thanetian in age (Biswas, 1992), and Lagerstroemia sp. from an intertrappean bed at Bharatwada, Nagpur, India that is Paleocene in age based on probable coeval deposits at Mohgaon Kalan in Chhindwara District (Trivedi, 1956; Keller et al., 2009a, c; Keller, pers. comm.). The Panandhro Mine specimen lacks the apex and petiole but leaf size and shape, clear secondary brochidromous venation and orientation of the secondary veins are within the variation of modern L. speciosa (L.) Pers. Lagerstroemia patelii is also recorded from a middle Miocene to Pliocene Siwalik fossil flora of the Himalayan foot-hills near Darjeeling, India (Antal & Awasthi, 1993). The two Nagpur impressions are described as having the venation pattern of Lagerstroemia and resembling those of L. indica L. However, the tertiary venation pattern that could definitely relate the fossil to Lagerstroemia is not visible and the identity is unconfirmed.

Two minimally different leaf species, L. mioparviflora Dwivedi et al. and L. eomicrocarpa Dwivedi et al. from the middle Miocene of the Siwalik Formation in western Nepal compare favorably with Lagerstroemia (Dwivedi et al., 2006). Other leaf impressions from middle Miocene deposits of the same formation at Kathgodam are unconfirmed as Lythraceae (Prasad, 1994b). Still other collections said to resemble extant L. indica (Shukla, 1950a; Trivedi, 1956; Lakhanpal & Dayal, 1966) were subsequently considered doubtfully Lagerstroemia by Prasad (1994b) or are rejected here for lack of adequate information. Fossil leaves of L. neyveliensis Agarwal from the Miocene of Tamil Nadu, India (Agarwal, 2002) were rejected as Lagerstroemia by Prasad et al. (2004). Lagerstroemia jamraniensis Prasad et al. (2004), middle Miocene in age from the Kathgodam area, appears to match living L. speciosa. Lagerstroemia siwalica Prasad (1994a) is described from the middle Miocene of Nepal. The number of putative fossil Lagerstroemia leaves is substantial and not all collections have been critically reviewed but the several verified as Lagerstroemia suggest that the genus was common and somewhat diverse in the wet subtropical forests of the Indian subcontinent in the middle Miocene.

Elsewhere, Lagerstroemia imamurae Tanai and Uemura (1991) from Honshu, Japan is described from a late Oligocene leaf impression (age determined by paleontological associations). By description (the accompanying illustration lacking details), it is closely similar to leaves ofL. speciosa (L.) Pers., having distinctive more or less quadrangular areoles filled with branching veinlets present between the secondary veins. The Pleistocene of Japan has also yielded Lagerstroemia leaves and pollen (listed only) (Iwauchi & Hase, 1987). The identification of an Oligocene leaf from Romania (Staub, 1887) as Lagerstroemia is erroneous (Liu et al., 2008).

Wood of Lagerstroemia is well represented in the fossil record as the form genus Lagerstroemioxylon Madler (1939). At least 16 species have been described and 10 are compared anatomically by Cheng et al. (2007). The modem wood anatomy of this species-rich genus is the most diverse among the three tree genera of the Lythraceae in Asia but can generally be distinguished from wood of Sonneratia and Duabanga by a suite of characteristics (Table 1; Baas & Zweypfenning, 1979). Occurrences of Lagerstroemioxylon are confirmed from the middle Miocene into the Pleistocene in India and Japan, the Miocene/Pliocene of Myanmar, the Pliocene of China, and the Quatemary of Sumatra (see Appendix). Lagerstroemioxylon eoflosreginum Prakash and Tripathi (1970) is unquestionably equivalent to modem Lagerstroemia and is most similar to modern L. speciosa. The wood is recorded from the Tertiary of Sumatra (Kramer, 1974) and widely encountered in India at several localities of Miocene age (Prakash & Tripathi, 1970; see also lists in Mehrotra et al., 2005; Srivastava & Guleria, 2006). Six species of Lagerstroemioxylon from the Bikaner District, Rajasthan, India (Harsh and Sharma, 1995) are insufficiently treated to verify their relationship to Lagerstroemia. Lagerstroemioxylon reported from the middle Miocene of Romania (Iamandei et al., 1998) is rejected because the fossil is anatomically anomalous in the Lythraceae due to the presence of secretory cells on the margin of the mostly heterogeneous rays, unlike the mainly homogeneous, non secretory-celled rays in Lagerstroemia.

Lagerstroemia also has an extensive fossil pollen record, although many of the reports cannot be confirmed due to inadequate descriptions and illustrations, or because the genus was merely listed in a pollen flora. The modern pollen is typically prolate, tricolporate with pronounced meridional ridges, six-pseudocolpate with a near psilate to verrucate or mgulate exine. Variation occurs in presence or absence of pseudocolpi, degree of development of meridional ridges, and in exine sculpture pattern. Some variation is related to the pollen dimorphism which occurs in many species (Kim et al., 1994) and is apparent in fossil material as well (e.g., Fujiki et al., 2001). Pollen from the typically more numerous antepetalous anthers is on average larger, thicker-walled with more distinct pseudocolpi and exine sculpture, and with a more perforated tectum than pollen from the antesepalous anthers (Muller, 1981b; Kim et al., 1994: Fujiki et al., 2001). Under light microscopy, Lagerstroemia pollen that has only faintly expressed pseudocolpi and a low-relief exine is difficult to distinguish from Decodon pollen. Ferguson et al. (2007) and Liu et al. (2008) have shown the advantage of using SEM to accurately identify pollen of Lagerstroemia and other taxa.

Among the earliest pollen said to be related to Lagerstroemia is a Lagerstroemia--type grain from the middle Eocene Nanggulan Formation of Java that is possibly equivalent to early forms of Florshuetzia. The type is hypothesized by Morley (2000) as ancestral to modern Lagerstroemia pollen and possibly a precursor to pollen of both Lagerstroemia and Sonneratia. Pollen comparable to that of extant Lagerstroemia is listed from the middle Eocene to early Miocene of central Sumatra and southern Thailand in both freshwater and brackish water deposits (Morley, 2000). Lagerstroemia pollen is confirmed from early/middle Miocene and possibly Oligocene deposits in northern Thailand (Songtham et al., 2005). It is confirmed from the Miocene in China and confirmed or listed from multiple localities of Miocene to Holocene age in China (Liu et al., 2008) and from Japan and Taiwan (Chung & Huang, 1972; Yamanoi, 1984 as Florshuetzia; Miyoshi et al., 1999; Fujiki et al., 2001). A number of Pleistocene-Holocene Lagerstroemia fossil pollen records from Japan are listed in the database, "Plant Fossil Record, Vers. 2.2, International Organization of Paleobotany" (1997). Living Lagerstroemia is represented today in Japan on the Ryukyu Islands including Kyushu, the northernmost locality of the genus, by L. subcostata Koehne var. subcostam and var. fauriei (Koehne) Yahara (Ohashi, 1999).

A number of fossil pollen reports are unconfirmed or rejected. The identity of pollen attributed to Lagerstroemia from the early Miocene of Japan (Yamanoi, 1978) was not accepted by Muller (1981a) following additional information supplied by Yamanoi. Pollen illustrated from the early middle Miocene of Korea (Yamanoi, 1992a) possibly relates to Lagerstroemia but the illustration lacks details to confirm the generic identification. Pollen identified as Lagerstroemia from the Pliocene of Schleswig-Holstein, northern Germany is most likely that of Decodon (Menke, 1976). Three pollen form genera from China have been equated with Lagerstroemia or are considered Lagerstroemia-like: Verrutricolporites pachydermis Sun et al. (1980) from the Oligocene of China; Rugulitriporites cf. vestibulipori Muller (Zhang & Zhan, 1991), and Planotricolporites lagerstromiaformis (Zheng) Song (Song et al., 1999). Verrutricolporites pachydermis bears some resemblance to Lagerstroemia but the identity is not confirmed. Rugulitriporites cf. vestibulipori (Muller, 1968) is fossil pollen of unknown affinity from the late Cretaceous of northwest Borneo and is unlike that of Lagerstroemia. Liu et al. (2008) indicate that the first two taxa were transferred as synonyms to Planotricolporites lagerstromiaformis in Song et al. (2004) but I did not fmd the transfer there or in Song et al. (1999). Planotrieolporites lagerstromiaformis from the Miocene of China is cited as approaching modern pollen of L. subcostata (Song et al., 2004) and similar to pollen from the Miocene of Korea (Yamanoi, 1992a). These relationships could not be confirmed from published evidence.

Fossil fruits and seeds with a suggested relationship to Lagerstroemia are rare. Petrified fruits from the early Eocene Vastan Open Cast Lignite Mine in Gujarat state, India are named Lagerstroemia sahnii (Singh et al., 2010). There is no evidence on any of the five fruits illustrated of a persistent floral tube or sturdy pedicel, features present in modem Lagerstroemia. Internally, the axile placentation and six locules conform to Lagerstroemia morphology but these features are not limited to the genus or to the Lythraceae. Size of the fruits in the type description (13-24x48-50 mm) is at odds with the size of the fruits in the illustrations (ca. 27 x 10 mm). Characteristics that would support the identity of the fruits, such as the morphology, arrangement, and number of seeds, are not present. The identity of L. sahnii as a Lagerstroemia is unconfirmed. Middle Pleistocene fruits from the Kiyokawa Formation in Chiba Prefecture, Japan are comparable to L. subcostata var. fauriei and are confirmed as such on the basis of the published illustration which clearly shows the reflexed sepals at the base of the fruit and the elongated valves of the capsule (Momohara et al., 2006).

In other reports, capsules of Lagerstroemia lignitica (Menzel) Mai (1960, 1971) from the late Miocene and L. europaea Mai (1964) from the middle Miocene of Gennany have been transferred to Polyspora lignitica (Mai) Gregor (Theaceae; Gregor, 1978). The fossil fruit of the extinct genus Minsterocarpum alatum Reid & Chandler from the upper Eocene of England has been compared to Lagerstroemia (Reid & Chandler, 1933). The seeds, but not the fruit shape, are similar to those of Lagerstroemia (see Minsterocarpum for further details). The remarkably preserved fruits and seeds of the extinct genus Shirleya from the Miocene of Washington state represent the most credible close extinct relative of modem Lagerstroemia (see Shirleya for details).

In summary, the oldest confirmed record of Lagerstroemia is the late Paleocene/ Eocene (ca. 56 Ma) leaf impression from India of L. patelii described by Lakhanpal and Guleria (1981). The pollen record of Lagerstroemia possibly begins (as "comparable to" or Lagerstroemia-like) in the middle Eocene of Central Java (Morley, 2000). By middle Miocene diverse Lagerstroemia wood and pollen clearly document the presence of the modern genus from tropical and subtropical southern Asia north to Japan. The genus was present in Japan in the Pleistocene and persists there today represented by L. subcostata on the island of Kyushu.

Lawsonia L. The genus, with the single species L. inermis L., has been cultivated for centuries throughout Eurasia for its sweet-smelling flowers and especially for the leaves from which the cosmetic pigment henna is obtained. The fossil record is based on seed remains. Seed clusters initially described from the late Miocene of northwestern Germany as Carpolithus lawsonioides Menzel (1913) and ascribed to Lawsonia, were later related to Decodon by Kirchheimer (1957). Mai (1996), in newly collected material of C. lawsonioides found similarities to Lawsonia and, importantly, did not find the expected germination valve characteristic of Decodon seeds. Based on this information, he made the combination Lawsonia lawsonioides (Menzel) Mai. The seeds illustrated by Mai are somewhat abraded but the description includes morphological and anatomical details correct for Lawsonia; anatomical details are not illustrated. The obpyramidal seeds are attached to a central placenta in rows, have a thick testa with rounded comers, a sub-basal micropyle, spongy parenchyma within, and are 2-3 x 1.8-2.3 rum, all features compatible with seeds of modem Lawsonia. The fossils are less slender, thicker-walled, and less sharply pointed than modem Lawsonia seeds. Lawsonia lawsonioides is also listed flora the middle Miocene of the Meuro sequence of Saxony and Brandenburg in eastem Germany (Mai, 2001).

Seeds resembling Lawsonia in size and shape are reported from the middle Eocene Princeton Chert, British Columbia, Canada (Cevallos-Ferriz & Stockey, 1988) at the type locality of Decodon allenbyensis. The variation seen in shape of the fossil seeds due to crowding is commonly seen in living Lawsonia fruits and in seeds of several other genera of the Lythraceae. The typical sclerotic endotestal and exotegmic tracheoid layers of Lythraceae seeds are not seen. There is a general similarity of the fossils to Lawsonia but the relationship is unconfirmed at present. The listing of a fossil species of Lawsonia by von Ettingshausen (1879) remains a nomina nuda (Graham & Graham, 1971 a). The confirmed fossil record of Lawsonia is restricted to seeds described by Mal (1996, 2001) from the middle and late Miocene of Germany.

Lythraceaepollenites Thiele-Pfeiffer. This is a pollen form genus with two species from the Miocene of Germany and Poland. One species, L. decodonensis Stuchlik (1994), is possibly related to Decodon. Another, L. bavaricus Thiele-Pfeiffer, with distinct affinity to the Lythraceae, has not yet been associated with an extant genus (Thiele-Pfeiffer, 1980, Worobiec, 2009) anda third, L. minimus Thiele-Pfeiffer (1988) has not been reviewed.

Lythraeeoearpon Kapgate et al. A six-locular capsule with loculicidal dehiscence was described by Kapgate et al. (2003) from the well-known Deccan Intertrappean Beds of India at Mohgaon Kalan as Lythraceocarpon mohgaonse. The fruit wall is furrowed with six ridges and each locule contains an oval seed filled with endosperm. Although the name suggests ah affinity to the Lythraceae, neither the deeply furrowed fruit shape nor cellular endosperm is found in the family. The characteristics suggest a relationship to the Malvaceae.

Lythrum L. Lythrum is an herbaceous genus of ca. 35 species, occurring primarily in wet places in the north temperate zones of North America, Europe, and Asia. The relationship of Lythrum to Peplis is somewhat unsettled. Peplis was reduced to taxonomic synonomy in Lythrum based on morphological evidence (Webb, 1967, 1980). Phylogenetic analyses of molecular data support this position; P. portula is nested in Lythrum as sister to the Eurasian clade. However, the relationships of two other species previously recognized in Peplis are not known (Morris, 2007) and some differences in pollen morphology of Lythrum and Peplis are apparent. At present, the preponderance of evidence supports recognition of a single genus Lythrum. Peplis is briefly summarized later in this study.

The primary evidence of early Lythrum/Peplis is fossil pollen. The oldest known ocurrences of the genera are based on pollen of Lythrum elkensis Grimsson et al. and Peplis eaglensis Grimsson et al., both recently described from the Late Cretaceous early Campanian (82-81 Ma) Eagle Formation at Elk Basin, Wyoming, USA (Fig. 7; Grimsson et al., 2011; see also Peplis). They are the oldest representatives of the genera as well as the oldest credible remains of the Lythraceae world-wide, exceeding in age the late Campanian (ca. 73.5 Ma) record of Decodon tiffneyi seeds from deposits in northern Mexico (Estrada-Ruiz et al., 2009). Somewhat younger pollen is P. yakutiana Grimsson et al. (72-69 Ma) from western Central Siberia. The fossil determinations are supported by remarkably detailed SEM photographs that leave no question as to their identity. The authors regard the pollen types as sufficiently distinct to warrant recognition of two genera. In comparison to pollen of most living species of Lythrum, L. elkensis is at the small end of the size range with thicker walls anda more densely microrugulate to microechinate colpal and pseudocolpal membrane. Pollen of Peplis eaglensis is larger with thicker walls and longer pseudocolpi than found in modem species (Grimsson et al., 2011).

There is a large gap between the Campanian Lythrum/Peplis pollen and the next youngest pollen records of Lythrum which extend in Europe from the Miocene to the present rime. Fossil pollen conforming to the Lythrum type occurs in the middle Miocene, late Miocene, and Pliocene of Spain (Barron et al., 2010 and citations therein; Van Campo, 1989), in the Pliocene of Schleswig-Holstein (Menke, 1976), and as L. wilhelmii Grimsson et al. (2011) in the late Miocene of Austria. Lythrum pollen is listed as present in the Pliocene of Romania (,Ticleanu & Diaconia, 1997), the middle Pleistocene of France (Andrieu et al., 1997), the Pleistocene of Burundi, Central Africa (Bonnefille & Rioliet, 1988), and the Holocene of France (Bakels, 1995), Korea (Fujiki & Yasuda, 2004), Arizona (Davis & Shafer, 1992), and other Holocene localities. Pollen originally reported as cf. Lythrum from the Eocene to Oligocene of Cameroon under the name Heterocolpites verrucatus (SalardCheboldaeff, 1978) subsequently was attributed to Crenea or Rotala (SalardCheboldaeff, 1981). The minimal information provided about Heterocolpites could apply equally to pollen of other lythraceous genera or other families of the Myrtales (cf. Patel et al., 1984).

Fossil flowers originally described as Lythrum tetrasepalum Miki (1959) from the Miocene of Japan subsequently were transferred to a new genus, Eotrapa Miki (1961; see details under Eotrapa). Macrofossils (leaves?, type unspecified) of Lythrum salicaria are listed from the late Quatemary (10,000 B.P. to present) of western Iran (Wasylikowa, 1967).

Microdiptera Chandler. Microdiptera is an extinct seed genus widespread in the northern latitudes of Eurasia and North America in swamp or lake margin sediments of middle Eocene to middle Miocene age, and persisting into the Pliocene in Eurasia (Chandler, 1957, 1963a, b; Mal & Walther, 1978; Friis, 1980; Tiffney, 1981; Mai, 1985; Matthews & Ovenden, 1990; see Eyde, 1972 for citations from Russia and other Eurasian localities). Microdiptera is treated here as including Diclidocarya E. M. Reid, in part, and Mneme Eyde, extinct genera sometimes recognized as distinct from Microdiptera (Nikitin, 1929; Eyde, 1972; Mai & Walther, 1978; Friis, 1980). In the taxonomic history of Diclidocarya, two species including the type of the genus, D. globosus, were considered equivalent to living Decodon and transferred to Decodon as Decodon globosus (E. M. Reid) Nikitin and Decodon gibbosus (E.M. Reid) E. M. Reid. A third species, Diclidocarya menzelii E. M. Reid was judged sufficiently distinct to be retained in Diclidocarya (Reid in Nikitin, 1929). Eyde (1972) subsequently provided a new generic name, Mneme, for Diclidocarya menzelii which had been orphaned by the transfer of the type species of Diclidocarya to Decodon. Mneme was later subsumed into Microdiptera as subgenus Mneme (Eyde) Mai and Walther (1978). Friis (1980), studying Miocene seeds of this complex from Denmark, found a number of features that suggested Mneme could be consistently distinguished from Microdiptera and Decodon, including less developed wings, deep dorsal furrows, less protruding ventral seed body, and smaller cells composing the germination valve. Tiffney (1981) recognized Mneme and Microdiptera as separate, extinct genera related to Decodon and provided a key and descriptions to them.

Microdiptera (in the broadest sense) is morphologically and anatomically similar to Decodon, sharing the following characteristics: anatropous ovules, bitegmic seeds with inner lignified layers typical of the Lythraceae, a germination valve on one side of the seed, and extensive spongy tissue through which the raphe runs from near the hilum to the chalaza. When considered as three distinct genera, Microdiptera, Diclidocarya, and Mneme, one to three co-occur with Decodon in Miocene lacustrine habitats in Europe, Asia and North America (Friis, 1980; Mai, 1987, 1995, 2001; Matthews & Ovenden, 1990; KvaCek & Teodoridis, 2007).

Chandler (1957) considered Microdiptera distinct from and closely related to Decodon. The small seeds (ca. 1.25 x 1.26 mm; Tiffney, 1981) differ most conspicuously from equally small obpyramidal-shaped Decodon seeds (ca. 1.2 x 1 mm) by presence of internal spongy tissue which projects as thick wings on either side of the ventrally convex seed body and germination valve. The germination valve extends half to four-fifths the length of the seed and is ca. one-fourth to one-half the width. Two furrows flank the raphe on the flattened side of the seed.

The oldest known Microdiptera are from the middle Eocene Bournemouth Fresh Water Beds of England and from Eocene deposits in eastern Central Russia (Chandler, 1960; Eyde, 1972; West, 2010). Microdiptera occurs in at least tive localities of Eocene and Oligocene age in southern England (Chandler, 1963a, 1964), and also in the Oligocene and early Miocene of western Siberia (Dorofeev, 1977), the early Miocene of the Brandon Lignite in Vermont, USA (Tiffney, 1981, 1994), and the Northwest Territories (Matthews & Ovenden, 1990). It is found at many localities in the Miocene of Europe. The most common and widespread species is M. parva Chandler, reported from North America, Europe and Asia. Mai (1998: 59) lists M. parva from the "middle Eocene to middle Miocene in western and central Europe." The genus extends into the Pliocene of eastern Europe and Russia (Tiffney, 1981; Matthews & Ovenden, 1990). Because the seeds vary extensively in size, shape, and minor details, several species have been described. The number of species is certainly fewer than the number of names suggests. The Appendix cites a few representative species.

Mikia Kovar-Eder & Wojcicki. Mikia pellendorjensis Kovar-Eder & Wojcicki, from a late Miocene aquatic flora in Austria, is described as the floating leaves ofthe extinct fruit Hemitrapa based on its occurrence in the same locality and stratigraphic levels as fruits of Hemitrapa trapelloidea Miki and TrapapellendorJensis Wojcicki & Kovar-Eder (Kovar-Eder et al., 2002). The leaves are fan-shaped with a weakly dentate margin, palmate venation, anda non-inflated petiole. They are unlike leaves of extant Trapa which are triangular or rhombic in outline with an inflated petiole, distinct midvein, and coarse marginal teeth ending in a bi-fid apex; the latter is particularly diagnostic of the genus. In North America, a different extinct leaf genus, Quereuxia Krysht. with compound leaves, occurs in the same beds as Hemitrapa (e.g., Ward, 1885, 1887; Knowlton, 1898a,b). Mikia, with entire leaves, also differs from Quereuxia by being fan-shaped and more strongly toothed (cf. Kovar-Eder et al., 2002: figs. 6, 8-13 & Golovneva, 1991: figs 2-4). Sizes are similar (30-44x35-67 vs. 25-55x30-55 mm). The taxonomic affinities of Mikia and Quereuxia are unknown at present (see Querewcia for further details of the genus).

Minsteroearpum Reid & Chandler. The extinct Minsterocarpum alatum Reid and Chandler (1933) from early Eocene beds at Sheppey, England is a four to six-loculed capsule, 16-22 mm long, with axile placentation and numerous seeds arranged in two rows in each locule. The seeds, to 3.5 mm, from anatropous ovules, have a chalaza at the broad end of the seed and hilum at the narrow end. Spongy wings extend on either side of the central seed body. Intemally, a distinctive testal carbonaceous cell layer overlies a thin tegmen of elongate cells, suggesting the presente of sclerotic/tracheoid layers such as characterize seeds of the Lythraceae. Reid and Chandler (1933: 418) refer Minsterocarpum to Lagerstroemia. Modem Lagerstroemia has larger seeds and contorted embryos. Embryos are not preserved in the pyritized seeds of Minsterocarpum. The suite of fruit and seed characters of Minsterocarpum is lythraceous and the genus is tentatively accepted as an extinct member of the Lythraceae whose closest living relative is unknown.

Pachyspermum Reid & Chandler. Pachyspermum quinqueloculare Reid & Chandler (1933) from the early Eocene beds at Sheppey, England is based on a single 2.6 cm long, five-locular fruit that is possibly capsular with septicidal dehiscence. Numerous seeds from anatropous ovules cover the elongated axile placenta. The seeds are oblong, subcylindrical, 0.3-1.5 mm, and composed of a small seed body and a larger spongy lateral appendage that partly surrounds the seed body. The structure of the seed coat is typical of the Lythraceae with the inner cell layers of the testa sclerotic and with a striated tegmen layer of elongate cells. Pachyspermum, Minsterocarpum and, to some extent, the seeds of Microdiptera, morphologically resemble those of Pemphis (pers. observ.), an extant member of the family whose seeds have a similar spongy mesotesta, but are not yet fully known anatomically. Pachyspermum is appropriately assigned to the Lythraceae but the fruit is large and is not clearly similar to an extant genus. More detailed comparisons of seed structure within the family would provide a new set of characters on which to better compare seeds of the present genera with numerous early forms like Pachyspermum.

Paleolythrum Chandler. The genus consists of two extinct seed species, P. bournense Chandler (1960) from early Eocene to late Eocene Boumemouth deposits of southem coastal England and P. gailense Chandler (1963b) from the Pliocene of Pont de Gail, France. The relationship to the Lythraceae is moderately supported by the obovate, bifacial, convex-concave seed shape, by the small seed size (2-2.5 mm), and by the elongate cells of the inner layer of the seed coat (exotegmen?). The abraded seed of P. bournense displays testal cells in longitudinal ridges anda 0.3 mm wide marginal rim, features unknown in Lythrum seed. The seed outline of Paleolythrum is less elongate than is typical of modem Lythrum seeds. Paleolythrum is possibly lythracean, but in the absence of adequate anatomical details, a specific relationship to Lythrum is unconfirmed.

Palaeotrapa Goloneva. This is an extinct form genus of three species of unattached fruits found in association with Quereuxia leaves in Maastrichtian deposits in the Koryak Upland of far northeastem Russia (Golovneva, 1991, 1994, 2000). The fruits differ sufficiently from one another to suggest that more than one type of plant is represented. Ali have horns or spines on the fruit but the fruits are not similarly shaped. Palaeotrapa aculeata (Krysht.) Goloneva, in particular, differs from P. bicornata Goloneva and P. triangulata Goloneva in being elliptic rather than triangular in outline, having curved rather than erect spines and more numerous elongate ribs on the fruit surface. Palaeotrapa triangulata resembles the extinct genus Schenkiella Wojcicki & Kvacek, although the specimen is too poorly preserved to be certain of any relationship (Wojcicki & Kvacek, 2002a). The inflorescences of Paleotrapa are racemose, unlike the solitary flowers in Trapa. Paleotrapa is rejected as an extinct genus of the Lythraceae. Relationship to Quereuxia, or some other non-Lythraceae lineage is also undetermined due to the absence of structural information and any direct physical connection to associated remains.

Pemphis J. R. Forst. & G. Forst. Pemphis acidula J. R. Forst. & G. Forst. is the only species of this woody genus. It occurs along coastal shores and inlets from East Africa to the Marshall Islands extending northward to the Ryukyu Islands. Pemphis is represented in the fossil record by Pemphis-type pollen from late Eocene (ca. 3734 Ma) lacustrine beds in the Ebro Basin (Cavagnetto & Anadon, 1996) and as Tricolporopollenites sp., from the middle Oligocene (ca. 28 Ma) As Pontes Basin in northeastern Spain (Cavagnetto, 2002). The grains, especially those illustrated in Cavagnetto (2002: pl. 14, figs. 16, 17), are unmistakably equivalent to those of extant Pemphis acidula (cf. Graham et al., 1987: fig. 28). Pemphis can be expected as fossil pollen or seed in floras deposited along past tropical or subtropical marine shorelines or backwaters of Europe or Afro-Asia. Modern Pemphis seeds, like those of Decodon and Lawsonia, are well-adapted for aquatic dispersai by the buoyant spongy tissue surrounding the central seed body, and some of the large numbers of small seeds (ca. 3 mm) produced should preserve in aquatic fossil floras due to their thick seed coat.

Peplis L. Peplis is a Eurasian genus of three species that in the most recent taxonomic revision is regarded as congeneric with Lythrum (Webb, 1967, 1980). This placement is supported by molecular phylogenetic data in which Peplis portula is nested within a monophyletic Lythrum as sister to the Eurasian species of Lythrum (Morris, 2007). The molecular relationships of the other two species of Peplis are unknown. Pollen of Peplis can be morphologically distinguished from Lythrum pollen in SEM by the mostly perpendicular orientation of the exine striae to the colpi and by pseudocolpi wider than the colpi (Fig. 7). In Lythrum the striae are mostly parallel to the colpi and the pseudocolpi are narrower than the colpi (Graham et al., 1987). Grimsson et al. (2011) choose to recognize fossil Peplis as distinct from fosssil Lyhrum based on the differences in pollen morphology of two new fossil species. Peplis eaglensis is rare in the Late Cretaceous early Campanian (82-81 Ma) Eagle Formation of northwestern Wyoming, USA. It differs from extant Peplis by larger grain size, longer pseudocolpi, thicker pollen wall with a more densely sculptured exine, and more conspicuous arching of the sexine over the pores. Peplis eaglensis and Lythrum elkensis, from the same formation, are the oldest known credible occurrences of the Lythraceae. Variation in the size and exine thickness of P. eaglensis has suggested the possibility of heterostyly in the species (Grimsson et al., 2011). The extant species ofLythrum formerly classified as Peplis are homostylic; modern Lythrum includes homo-, di-, and tristylic species. A somewhat younger fossil species, P. yakutiana from the late Campanian to early Maastrichtian (72 -68 Ma) of western Central Siberia, Russia differs from P. eaglensis primarily by its greater range of sizes and by colpi that are wider, by striate rather than micro-rugulate sculpturing on the colpal and pseudocolpal membranes, and by broader striae more widely separated along the colpi. Peplis pollen is listed from the Pleistocene of Greece (Wijmstra, 1969). Whether as a congener of Lythrum or as an independent genus, P. eaglensis shares with L. elkensis the record for the world's oldest verified fossil occurrence of the Lythraceae.

Prototrapa Vassiljev. The genus was erected with three species by Vassiljev (1967) to accommodate long-appendaged fruits from the Aptian-Albian of southeastem Australia initially identified as remains of Hemitrapa (Douglas, 1963, 1994). Prototrapa differs from Hemitrapa by a much smaller fruit size (1-3 mm vs. 1620 mm), presence of two long (5-7 mm) non-barbed appendages, and much thinner fruit construction. Although Prototrapa is ca 40 million years older than Hemitrapa and the genera occur in different hemispheres, a relationship should not be excluded on that basis. The morphological differences, however, indicate Prototrapa is not related to Hemitrapa or Trapa (Tiffney, 1984) and its position among other early angiosperms remains unknown.

Pseudotrapa Z. Kvacek. A single species Pseudotrapa buzekii Z. Kvacek is described from fossil leaves of the early Miocene (ca. 19-18 Ma) Most Basin, Czech Republic (Kvacek et al., 2004). The leaves differ significantly from modern Trapa by a somewhat undulate rhomboidal outline, coarsely dentate margin, and teeth with a rounded apex. Modem Trapa leaves are more strongly rhomboidal with more numerous, sharp teeth with a bi-fid apex. Two other leaf fossils suggestive of Pseudotrapa appear under the name Trapa assmanniana (Goeppert) Gothan (syn. Populus assmanniana Goeppert, 1852). These are from the Oligocene of Kazakhstan (Kryshtofovich, 1956 fide Kvacek et al., 2004) and from the late Neogene of Sosnica, Poland (Goeppert, 1855). The published descriptions and illustrations of the two collections are inadequate to support assignment to Trapa and the original Goeppert material is missing. Some earlier authors accepted T. assmanniana as the leaf of the fossil fruit T. silesiaca because they were associated in the same deposit but, lacking an attachment, the relationship has not been proven. Kvacek et al. (2004) doubt a Pseudotrapa connection to Trapa. There is no evidence to relate Pseudotrapa to Trapa or to the Lythraceae.

Punica L. Punica in the former Punicaceae is now part of the Lythraceae based on relationships uncovered in molecular phylogenetic studies (Graham et al., 2005). A genus of two species of shrubs originally from semi-arid habitats of the middle East, Punica is generally thought to have originated on the Iranian Plateau. Punica granatum L., the cultivated pomegranate, is best known. The smaller-fruited species, Punica protopunica Balf., with unspecialized, regularly-arranged fruit locules, is endemic to Socotra in the Gulf of Aden.

The oldest fossil attributed to Punica is silicified wood of Punicoxylon eocenicum Prive-Gill from the middle Eocene of the Paris Basin (48.6-40.4 Ma; Prive-Gill, 1981). Although the wood anatomy of living Punica is well known (Shilkina, 1973; Bridgwater & Baas, 1978 and citations therein; van Vliet & Baas, 1984), identification of putative fossil wood of Punica is hindered by the numerous anatomical characters Punica shares with other Lythraceae and to a lesser extent with other myrtalean genera (Baas & Zweypfenning, 1979). Bridgwater and Baas (1978) in study of an extensive reference collection of wood anatomical samples of the Lythraceae, noted it was difficult to separate Punica wood from that of other genera in the family. They concluded that anatomically it was most similar to the American genera Ginoria Jacq. and Pehria but only slightly less so than to other Lythraceae. The wood anatomical description of Punicoxylon eocenicum is detailed and the character states and sizes in the fossil fall within the range of modern Punica with the exception that the height of the rays is at the high end of the modern range of the genus. Although there is no way to be completely certain of the identity, the presence of fruits, seeds, and leaves of the genus in late Eocene and Oligocene deposits of western Europe (see below) suggests the genus could well have been present in the middle Eocene in France. Dry vegetation types with sclerophyllous plants like Punica were beginning to expand in earnest in the middle and/or late Eocene in Europe and were established in southern Europe in the Oligocene (Collinson & Hooker, 2003). The most abundant fossil remains assigned to Punica are fruits and seeds. Remains of fruits are minimal, consisting merely of an axile placenta with numerous crowded, attached seeds. The seeds, initially recognized under the form name Carpolithus natans Nikitin, are present in late Eocene to late Pliocene floras in Europe, western Siberia, and Belarus (Nikitin, 1965; Mai & Walther, 1978, 1985; Mai, 2001). The oldest are from the late Eocene of Germany (Mai & Walther, 1985: 102).

Two fossil species of Punica seeds are recognized: Punica antiquorum (Heer) Mai, (syn. Punica natans [Nikitin] Gregor), common in the Oligocene anal Miocene of Europe (Mai & Walther, 1985: 101); and Punica tertiaria Gregor (1978) from the middle Miocene of Germany. The description and illustrations of the type of the basionym Cydonia? antiquorum Heer are minimally informative (Heer, 1869b: 99, pl. 30, figs. 36-40). The external morphology and the variation in seed shapes attributed to P. antiquorum and P. tertiaria are typical of the variation seen in modern Punica seeds devoid of the fleshy sarcotesta. Unfortunately, few anatomical details of the fossils are available. Initially, Dorofeev (1963) considered seeds of these species to belong either to the Onagraceae or the Lythraceae. Nikitin (1965) suggested Nyssaceae. Mai, who has made an extensive study of fossil Punica, accepts them as Punica (Mai & Walther, 1985). His decision is tentatively accepted here. Extensive records, illustrations, and collections of Punica seeds and Carpolithus natans from the Miocene of Germany are present in the Berlin Natural History Museum (records supplied by B. Mohr, pers. comm., 2010). The range of seed forms in Carpolithus natans suggests more than one plant type is represented.

Fossil leaves of Punica paleogranatum Kutuzkina (1974) are described from the late Miocene upper Sarmat flora of southwestern Russia. They are fragmentary and difficult to interpret from the published photographs, although a sparsely diagrammed leaf matches modern Punica leaves to the extent of the features presented. Entire, somewhat degraded leaves of P. paleogranatum are also recognized from the middle Miocene lower Sarmat flora in the Republic of Moldova (Shtephyrtza, 1989). These match Punica in size and venation pattern, and especially in the unusual thickened leaf tip which in modern Punica is an apical foliar nectary (Turner & Lersten, 1983). Palamarev (1989) lists Punica palaeogranatum from the early Oligocene of Bulgaria as a woody element of the sclerophyllous Mediterranean flora that was expanding there at that time. Other leaf impressions of Punica are listed from the middle Miocene (early Sarmatian) flora of Moldova (13.6-13.4 Ma; Givulescu, 1999; Iamandei et al., 2005); as part of an early Miocene mixed laurophyllous and arctotertiary flora from Lagenau, Germany (Mai, 1995: 390); and as a Mediterranean element of the late Pliocene from Central France (Mai, 1995: 378).

Punicites hesperidum Weber from the late Oligocene of Germany is a detached fossil flower resembling Punica. The type is lost but a 3.5 cm long fossil corresponding in shape to the type, though from a different collection, is briefly described and illustrated by Weyland (1948). It has a coriaceous-subcarnose funnel-shaped floral tube that is broadened below a median contraction and six concave lanceolate-acuminate sepals. Weyland was uncertain of the relationship to Punica, suggesting that it might be Cucubalites Goepp. (Caryophyllaceae). Leaves and unattached flower buds of Pliocene age from southern France were described as Punica planchonii Saporta and Marion (1876). These were regarded by Kutuzkina (1974) as more similar to Periploca L. (Asclepiadaceae) than to Punica but too fragmentary to be certain. The description and hand-drawn illustrations by Saporta & Marion match leaves of Punica only at the most simplified level. The pair of joined flower buds are elongate, oval in outline and 5-parted. Other informative details are lacking and Punica planchonii cannot be confirmed as Punica.

In summary, Punica is represented in the fossil record by seeds and infrequently by wood and leaves. The oldest fossil accepted as Punica is wood of Punicoxylon eocenicum from the middle Eocene (48.6-40.4 Ma) of the Paris Basin. Leaf impressions and seeds tentatively accepted as Punica from the Miocene of Europe indicate the genus was established in the middle Miocene as an element of the expanding sclerophyllous vegetation of that time. Fossil Punica remains, mostly seeds, are reported from Europe and western Asia (France, Spain, Germany, Poland, Bulgaria, Moldova, Georgia) in deposits ranging in age from the Eocene to the Pliocene.

Quereuxia Krysht. This is an extinct aquatic genus of floating rosettes of compound leaves that has been confused at rimes with Trapa and other Trapa-like fossils. Quereuxia and Trapa-like aquatic plants, present in the same type of aquatic habitats in northeastern Asia and northwestern North America in the Late Cretaceous and early Cenozoic, share a similar rosette arrangement of leaves but differ in a number of other vegetative characters. Floating Quereuxia leaves are compound with 3-7 or more obovate leaflets, unlike the floating simple, rhombic leaves of Trapa; the petioles are not inflated as in Trapa; and the teeth of the serrate leaf margin have a single mucronate apex, not the distinctive bifid apex of teeth on the leaf margin of Trapa. The submerged leaves of both genera are similarly filiform and branched.

Quereuxia is widely known in Eurasia from the Late Cretaceous Cenomanian to the end of the Paleogene (Lee & Li, 1959; Golovneva, 1991, 2000; Moiseeva, 2010) and in North America from the Late Cretaceous Campanian and Maastrichtian to the Eocene (e.g. Peppe, 2010). In North America, early paleobotanists consistently employed the synonym Trapa? microphylla Lesquereux (e.g., Lesquereux, 1878; Stanton & Knowlton, 1897; Ward, 1885, 1887; Penhallow, 1908; Berry, 1928, 1935; Hollick, 1930; Brown, 1962; Brown & Houldsworth, 1939; Dorf, 1942) for Quereuxia leaves from Late Cretaceous and Paleocene deposits in Alaska, western Canada, and western United States (McClammer & Crabtree, 1989). Trapago angulata (Newb.) McIver and Basinger (1993, 1999) was applied to the same leaf type occurring in the Maastrichtian of Alberta, Canada; the early to middle Paleocene deposits from southwestern Saskatchewan, Canada (McIver & Basinger, 1993); and the middle (or late?) Paleocene deposits of the eastern Arctic Archipelago (McIver & Basinger, 1999). Hickey (2001) revised the nomenclature of Quereuxia, recognizing Trapa? microphylla Lesquereux as a synonym of Quereuxia angulata (Newb.) Krysht. ex Baikovskaja. There has not been a comprehensive taxonomic study to determine if more than one plant type is represented among the several species of Quereuxia. At present, there is no accepted record of fruits attached to Quereuxia plants to help clarify the family relationship and the affinities of the genus remain unknown (Stockey & Rothwell, 1997 and citations therein). There is no evidence that Quereuxia is an extinct member of the Lythraceae.

Raoanthus Chitaley & Patel. Raoanthus intertrappea Chitaley and Patel (1975) is a petrified flower from the Deccan Intertrappean Beds at Mohgaon Kalan cited as possibly related to Sonneratia or the Lythraceae. The flower is zygomorphic and apetalous, with a single whorl of nine stamens, a partly inferior syncarpous ovary that is unilaterally adnate to the perianth and seven-loculed with the septa fused at the base of the ovary. Although some of these features are individually present in some genera of the Lythraceae, together they constitute a highly unusual assemblage for the family not occurring in any of the modem genera. As currently understood, Raoanthus is best considered an extinct genus of unknown familial affinity.

Rotala L. Rotala is a small herbaceous aquatic or amphibious genus of ca. 45 species distributed primarily in Afro-Asia. The greatest morphological diversity occurs in southern Asia and the highest number of species in Africa (Cook, 1979). The fossil record of the genus is scanty or non-existent. Rotala sp. is listed in subfossil records (ca. 14,000 years old) from northeast Thailand (Penny, 2001); the record remains to be verified. Pollen grains first determined as Verrutricolporites rotundiporus (Salard-Cheboldaeff, 1976) and Heterocolpites verrucatus van der Hammen (Salard-Cheboldaeff, 1978) from the Eocene and Oligocene of Cameroon were later related to Rotala or Crenea (Salard-Cheboldaeff, 1981). The grains are 6-pseudocolpate as in Rotala, but the pseudocolpi are less prominent, the grains are shorter and wider, and the exine is more coarse than in modern species of Rotala. Muller (1981a) rejected their relationship to Rotala and it is unlikely they are lythracean. Although occurrence of Rotala in the fossil record has not been demonstrated with certainty, fossil Rotala pollen can reasonably be expected in young swamp or lake sediments in Africa or southeastern Asia where the genus is common today.

Sahnianthus Shukla. Sahnianthus is one of few fossil flowers with a clear affinity to the Lythraceae. The mature fruiting stages of Sahnianthus are known as the form genus Enigmocarpon (Fig. 4; Sahni & Rode, 1937; Sahni, 1943; Shukla, 1944; see Enigmocarpon for details). The fossils of both genera are present in rocks formed from sediments of an ancient lagoon and weathered from cherts of the Deccan Intertrappean Beds at Moghaon Kalan, Madhya Pradesh, India. The stratigraphic placement of the fossil layer may be on either side of the K-T boundary but it is increasingly regarded as Danian in age (Cripps, pers. comm.). Sahnianthus is also found at nearby Paladon (Chitaley, 1955).

The reproductive characters indicative of lythracean affinity that are sought in Sahnianthus, Enigmocarpon, and other similar flower or fruit remains in the Deccan Intertrappeans are: actinomorphic, bisexual, perigynous floral tubes with valvate sepals equal to or shorter than the floral tubes; petals inserted on the inner margin of the floral tube and alternating with the sepals; stamens in two whorls, usually twice as many as the sepals (but multiplied in some genera), anthers dorsifixed, inflexed, dehiscing longitudinally; gynoecium syncarpous, with a single style and capitate stigma, ovary superior and mostly four to six-locular but also two-locular or multi-locular, placentation axile, ovules many in two rows in each locule or irregularly covering the placenta; fruit enclosed in a persistent floral tube, capsular, rarely a leathery berry, dehiscent or indehiscent; seeds numerous, typically small, with a decay-resistant double internal layer composed of one or more sclerotic testal layers and ah exotegmen of elongated tracheoid cells.

The character suite of Sahnianthus falls within these generalized Lythraceae floral character states. Sahnianthus is a pedicellate, bisexual, tubular, perigynous, six to eight valvate-lobed flower ca. 7-10.5x1.8 mm (Fig. 4a-c). Sepals ca. 3 mm long make up nearly half the total length of the floral tube (Chitaley, 1955, 1964). Petals have not been seen in buds. Stamens number eight to twelve, are opposite the sepals and petals, and have introrse, dorsifixed anthers that dehisce by longitudinal slits. The gynoecium is superior, with a simple style and an enlarged, somewhat elongate stigma; the ovary is distinctly stipitate and free of an encircling nectary; six to eight locules each enclose 14 to 20 anatropous ovules in two rows on an axile placenta (Chitaley, 1955). Seeds are elongate and surrounded in part by spongy tissue (Manchester & Kapgate, 2011). The presence of a stipitate ovary (Fig. 4a, b) is unusual in the Lythraceae, occuring in seven other genera where it varies from scarcely to well developed. In the modern genera Woodfordia and Sonneratia the stipe is short and in Sonneratia it is encircled by a floral nectary (Tobe et al., 1998). The stipe is more prominent in the Madagascar endemic Koehneria S. Graham et al. and in the American genera Adenaria, Lafoensia, Pehria and Pleurophora D. Don. Two species have been described: S. parijai Shukla (1944) with one nectary scale and S. dinectrianum Shukla (1958) with two stalked nectaries. The validity of the these names needs to be established.

Initially, Sahnianthus flowers were considered possibly heterostylic and related to modern Decodon or, if monostylic, then most similar to the living American genus Heimia (Shukla, 1944). Chitaley (1955) determined that Sahnianthus was monostylic, the different filament and style lengths in the fossil material the result of differing maturity of the buds and flowers. No morphological evidence relates Sahnianthus specifically to Decodon or Heimia.

Relationships to the Asian genera Woodfordia (Shukla, 1944) and Sonneratia (Verma, 1950; Mahabale & Deshpande, 1957; Mehrotra, 2003) have also been suggested. Woodfordia differs from Sahnianthus by: a slightly zygomorphic floral tube ca. four to six times longer than the sepals, unlike the short tube with sepals nearly equalling the floral tube in Sahnianthus; by a two-locular ovary with ovules covering the placental surface, not restricted to two rows; and by a very short ovary stipe in contrast to the prominent stipe of Sahnianthus. Modern Woodfordia pollen (cf. Graham et al., 1990: figs. 10-12, 44, 45) is tricolporate, lacks pseudocolpi, has a coarsely granular scabrate exine (W. fruticosa S. Kurz) or a nearly glabrous one (W. uniflora Koehne), and is 16-22 [micro]m in diameter. Pollen of Sahnianthus is abundant in situ in several of the fossils but difficult to interpret due to collapse, folding and preservation (Manchester & Kapgate, 2011). It is approximately spherical, either triporate or tricolporate, with a psilate exine (Dwivedi & Shukla, 1958; Chitaley, 1951; Manchester & Kapgate, 2011) or possibly a coarsely granular one similar to the pollen exine of Woodfordia fruticosa (Shukla, 1944), and is 13-21 [micro]m in diameter. Sahnianthus pollen is most similar to pollen occuring in a clade of five modern genera of the Lythraceae that includes Woodfordia and Koehneria in the Old World and Adenaria, Lourtella S. Graham et al., and Pehria in the New World (Graham et al., 1990). Manchester and Kapgate (2011) have recently reexamined Sahnianthus flowers and Enigmocarpon fruits and provide new descriptions and excellent illustrations from the original and additional collections.

Sahnianthus flowers, fruit, and pollen are morphologically distant from those of modern Sonneratia. In Sonneratia the flowers are much larger and structurally much thicker (22-33 x 12-22 mm in mature bud) than those of Sahnianthus; the ovary is superior to partly inferior; stamens are many (up to 300 fide Duke & Jackes, 1987) and multiseriate; a saucer-shaped nectary surrounds a very short-stipitate ovary (Mahabale & Deshpande, 1957); and the fruit is ah indehiscent, thick-walled berry with a pulpy interior (Backer & van Steenis, 1954; Duke & Jackes, 1987). The pollen of Sonneratia is among the largest in the family (58-62 Px37-39 E [micro]m) and is tricolporate to triporate, with or without three faint pseudocolpi, with prominent mesocolpal ridges, protruding pores, and a scabrate to verrucate exine (Muller, 1978; Graham et al., 1990). Pollen cited as Triorites sp. from the cherts at Mohgaon Kalan has been suggested as belonging to Sahnianthus (Chitaley, 1951) but the generic description is insufficient to confirm the claim. Two fossil woods, also from Mohgaon Kalan, the root Sonneratiorhizos raoi Chitaley (1968) and the stem Sonneratioxylon preapetalum Awasthi (1969), have been proposed as parts of the same plant that produced Sahnianthus and Enigmocarpon (Chitaley, 1977). To date, no attachments have been found among fossils of the four genera. Chitaley informally combined the four fossil forms under the name Enigmocarpon parijai (Chitaley, 1977).

Sahnianthus and the associated fruit Enigmocarpon unquestionably fall within the limits of the Lythraceae. They are accepted as two organs of the same genus and as an extinct member of the family. Their origins date to the early Paleocene and they are without clear relationship to any living genus.

Sahnipushpam Sahnipushpam shuklai Verma (Shukla, 1947, 1950b; Prakash, 1955; Verma, 1956; Prakash & Jain, 1963; Ambwani et al., 2001; Kapgate et al., 2011) (syn. S. glandulosum Prakash) from Mohgaon Kalan is a poorly preserved flower/fruit initially assigned to the Sonneratiaceae (Verma, 1956) or said to be close to the Myrtaceae (Prakash, 1955, 1956). Based on pollen, Sahnipushpam was considered near the Araceae (Prakash & Jain, 1963) but the relationship was refuted (Mayo et al., 1997; Hesse & Zetter, 2007). A reinvestigation of the type material was made by Ambwani et al. (2001), and most recently a new study of additional material expanded information about the genus (Kapgate et al., 2011). As emended by Kapgate and co-workers, S. shuklai consists of small mostly pistillate and a few hermaphroditic flowers arranged on a slender spike. A quadrangular four-lobed, persistent floral tube includes a single stout style with peltate stigma. Four stamens arising from the base of the flower bear oblong or boat-shaped, monosulcate, finely reticulate pollen 25-30 [micro]m long. The ovary is superior with axile placentation and is 4-6-carpellate; each carpel is subdivided by a partial secondary septum. The fruit is interpreted as a single "ruminate-seeded cuboidal nut" that is lobed by intrustions of the septa, in opposition to earlier interpretations that the fruit contained two ovules per locule (Ambwani et al., 2001). The wall of the fruit is studded by scattered circular glands said to resemble oil cells. A position in the Lythraceae is supported only by characters common to many angiosperm familes, and is opposed by pollen, the glands on the fruit wall, by the partially partitioned locules, and by the single seeded nut (or by two seeds per locule). The dioecious condition is rare in the Lythraceae occuring only in Capuronia benoistii (Leandri) P. E. Berry from Madagascar. Based on morphological evidence Sahnipushpam shuklai is rejected as related to any member of the Lythraceae. Kapgate et al. (2011: 216) concluded that Sahnipushpam is "likely most closely similar to the Araceae" but "the possibility remains that Sahnipushpam belongs to an extinct family with few if any extant relatives." The flowers are associated at Mohgaon Kalan with wood of Dryoxylon mohgaoense Rode, a genus said to be related to the Myrtaceae (Prakash, 1956).

Schenkiella Wojcieki & Kvacek. Specimens of this genus of fossil fruits were initially described as Trapa eredneri Schenk (1877). Sehenkiella credneri (Schenk) Wojcicki & Kvacek, from the early Miocene of central Europe, is a sharply obtrigonal, three-homed fruit, 10-15 x 8-14 mm, trowel-shaped in outline, with a finely alveolate porous surface (the endocarp). Wojcicki and Kvacek (2002a) reject the relationship of Schenkiella to Trapa, citing the absence of a comparable porous surface in Trapa. They have also compared Schenkiella to Paleotrapa triangulata based on similar shape and found the impressions of Paleotrapa triangulata too poorly preserved to determine whether the genera belong to the same evolutionary lineage. The relationship of Schenkiella to Trapa or to the Lythraceae is rejected.

Shirleya Pigg & DeVore. This extinct North American genus from the middle Miocene Yakima Canyon flora of Washington state is represented by fruits and seeds of Shirleya grahamae Pigg and DeVore (2005; Pigg & Wehr, 2002). The genus is based on silicified capsules with enclosed permineralized seeds exhibiting an exceptional wealth of morphological and anatomical detail. The fruits, from a superior ovary enclosed within a persistent floral tube, strongly suggest those of the Asian tree genus Lagerstroemia, even displaying the contorted seed cotyledons characteristic of Lagerstroemia. The fruit size at 10 x 11.5-12.3 mm, five to seven locules, loculicidal dehiscence, and winged seeds attached to an axile placenta are typical of Lagerstroemia fruits. Shirleya seeds differ from those of modern Lagerstroemia by their pendulous position on the placenta. Lagerstroemia seeds are held erect on the axile placenta. Lagerstroemia is unknown as a fossil in North America. Shirleya is accepted as an extinct member of the Lythraceae and closely related in the family to modern Lagerstroemia.

Sonneratia L. f. Living Sonneratia comprises ca. seven species that are frontal members of mangrove swamps or tidal inlets and bays from coastal tropical East Africa to Indo-Malaysia, southern China, New Guinea, Australia, and islands of the Western Pacific. The genus has an extensive fossil record based on wood and pollen.

Fossil wood of the form genus Sonneratioxylon Hofmann (1952) is reported from the Late Cretaceous and throughout the Cenozoic in Europe, Libya, India, Southeast and Central Asia, southeastern China and Japan. The wood anatomy of the modern genus and of several confirmed fossil woods of Sonneratia is well known, so the comparative information needed for identification is readily available (Pearson & Brown, 1932; Gamble, 1972; Baas & Zweypfenning, 1979; van Vliet & Baas, 1984; Rao et al., 1987; Mehrotra, 1988; Deng et al., 2004).

The earliest putative fossils of Sonneratia are two pieces of wood found in a complex red and mixed-colored stratified sandstone and in unlayered fine red soil in the Dzhetymtau Mountains, northern Kyzylkum, Uzbekistan (near Basbulak fide F. Khasanov, TASH, pers. comm., 2011). They are Turonian in age (93.5-89.3 Ma) by association with dated faunal and microfaunal fossils. The Kyzylkum region is paleontologically known for a diversity of dinosaur fossils and, consequently, deposits have been extensively studied and dated (H.-D. Sues, Smithsonian Institution, pers. coram., 2010). The wood, named Sonneratioxylon turonicum Shelomentseva (1992), if correctly determined as belonging to Sonneratia, would be the oldest known fossil of the Lythraceae, ca. 7 Ma older than the New World early Campanian pollen of Lythrum/ Peplis (Grimsson et al., 2011) and would place the earliest occurrence of the family in the Old World, in the Tethys Sea region of southern Central Asia.

Sonneratioxylon turonicum shares the wood anatomical features of vessels with simple perforations, uniseriate rays, and alternate pitting with at least eleven angiosperm fanilies (Shelomentseva, 1992). Within the Lythraceae, modern Sonneratia wood is similar to that of Duabanga, Lagerstroemia, and Punica, but is distinguished from them by the total absence of axial parenchyma, by the presence of homogeneous to weakly heterogeneous rays of mostly squared and procumbent cells, and by the presence of at least some septate fibers. Sonneratioxylon turonicum appears to match these characteristics but verification of the original material needs to be made because it is difficult to be certain of the identity from the published information alone due to the poor quality of the images available for study and some degradation of the sample. Table 1 compares characters of S. turonicum with other species of fossil Sonneratioxylon and with modern woods of Sonneratia, Duabanga, Lagerstroemia, and Punica. A cautionary note is that modern Sonneratia wood can also be similar to some woods in the Melastomataceae, both having vessels solitary or in radial chains of two to four, rays homogeneous tO weakly heterogeneous, and fibers in part septate (Melastomoideae, van Vliet & Baas, 1984). Modern Sonneratia wood typically contains large rhomboidal crystals in the rays. These are absent in many Sonneratioxylon fossils including S. turonicum, and are generally presumed lost during preservation.

The type material and original anatomical preparations of S. turonicum were transferred from the initial depository to a different facility at the Institute of Botany in Tashkent (TASH) and have not been curated since the change. Recent attempts to find the collection were unsuccessful, but it is hoped that efforts will continue. The present location of the author of the species is unknown to the staff at TASH (F Khasanov, pers. comm.) and at LE (Komarov Botanical Institute; T. Shulkina, MO, pers. comm.). For now, this intriguing record is unconfirmed but should be kept in mind when proposing hypotheses on the origin and radiation of the family.

The oldest confirmed occurrences of Sonneratioxylon are early Paleocene (Danian) in age, and come from plants that were rafting northward from Africa on the Indian subcontinent while it was still south of the equator. A number of species of Sonneratioxylon have been described from Danian levels (67.3-63.8 Ma) of the Deccan Intertrappean flora and from younger Paleocene deposits elsewhere in India (Srivastava & Guleria, 2006; see Appendix). Other species are known from the Miocene-Pliocene Cuddalore sandstone deposits near Pondichery, India. Sonneratioxylon preapetalum Awasthi (1969) (corrected from the orthographically incorrect S. preapetala) is most commonly reported. It is extensively distributed from the Paleocene to the early Miocene of peninsular India (Guleria, 1992; Srivastava, 2008), from the Miocene of Thailand (Vozenin-Serra et al., 1989), and from the Tertiary (upper Tertiary?) of Sumatra (Kramer, 1974; see Appendix for synonyms of S. preapetalum). Sonneratia kyushuensis Srivastava and Suzuki (2001) from the early Oligocene of the Kyushu Islands, Japan, is minimally different from Sonneratioxylon preapetalum and differs from modern Sonneratia wood by slightly larger vessels and the apparent absence of crystals in the rays (Fig. 8). The taxonomy of Sonneratioxylon would benefit from a more thorough understanding of the variation in wood anatomy of the living species of Sonneratia. The wood of Dryoxylon mohgaoense Rode (1936; syn. Parajugloxylon mohgaonensis) from the Deccan Traps is rejected as related to Sonneratia (Prakash, 1956).

Beyond India and Asia, Sonneratioxylon aubrevevillei Louvet (1970; Boureau et al., 1983) is described from the northern margin of the African continent in middle Eocene coastal mangrove sediments in Libya and indicates an extended distribution of Sonneratia beyond Asia in the Paleogene. The description of the type of Sonneratioxylon Hofmann (1952) from the late Oligocene of Austria corresponds to modern Sonneratia wood anatomy only at a very general level, lacks quantitative details and, contrary to wood of modern Sonneratia, includes the presence of scanty paratracheal parenchyma. A comparison of Sonneratia wood anatomical features in a phylogenetic context (e.g. Shi et al., 2000), could be useful in revealing patterns of evolutionary change related to adaptation to mangrove habitats if based on reliable fossil records over the full stratigraphic and geographic range of the genus.

The fossil pollen record of Sonneratia is one of most intensively researched fossil studies in the Lythraceae. It is an exceptional example of how fossil pollen has been employed to reconstruct evolutionary changes in plant taxa through time (Muller, 1984; Morley, 2000; Plaziat et al., 2001). A precursor to pollen of modern Sonneratia and Lagerstroemia was first described from the Oligocene of Borneo as the extinct pollen genus and species Florshuetzia trilobata by Germeraad et al. (1968). Florshuetzia is abundant and diverse in deposits in southeastern Asia and its extensive variation led Muller and Morley to study the morphological changes through time (Muller, 1969, 1978, 1981a, b, 1984; Morley, 2000; see Florshuetzia for further details). They determined that Florshuetzia had changed in in size, colpal development, and sculpture pattern, evolving to the modern pollen type recognized as Sonneratia (Muller, 1969: 225, 1978; Patel et al., 1984; Graham et al., 1990; Morley, 2000).

The first fossil pollen directly referable to modern Sonneratia is Florschuetzia levipoli from the base of the early Miocene of Borneo (ca. 19 Ma) which "agrees in all essential characteristics with the pollen of [modern] Sonneratia caseolaris" and marks the beginning of Sonneratia as an obligate mangrove genus (Germeraad et al., 1968: 308, fig. 5; note that a verrucate equatorial belt in these fossils is interpreted as equivalent to a colpal field in modern Sonneratia). This pollen type is followed in the middle Miocene by fossil pollen referable to modern S. alba (F. meridionalis) and subsequently in the Pliocene by the appearance of pollen of other modern species (Morley, 2000). Sonneratia caseolaris is known for a lower tolerance to salinity than the highly tolerant Sonneratia alba (Rugmai et al., 2008), perhaps reflecting the early evolution of S. caseolaris in brackish waters. A molecular-based phylogenetic analysis which found S. caseolaris sister to the rest of the genus can be interpreted as supporting the greater age of the species suggested by the pollen evidence (Shi et al., 2000; S. paracaseolaris Ko et al., shown as sister to S. caseolaris in the study Shi et al., is a hybrid of S. caseolaris x S. alba).

The work of Muller and Morley emphasizes the lengthy presence of Florshuetzia and subsequently Sonneratia in southeastern Asia, but does not necessarily support a southeastern Asian origin of the genus. The earliest occurrence of confirmed Florshuetzia pollen anywhere is from late Paleocene deposits in south central France (58.7-55.8 Ma; Gruas-Cavagnetto et al., 1988; Plaziat et al., 2001). The pollen is younger than the earliest known fossil wood attributed to Sonneratia from the Deccan Intertrappean Beds. (See Florshuetzia for further details on the variation and distribution of the pollen in southeastern Asia and its relationship to other lythracean pollen genera.)

Several fossil pollen taxa are rejected as unrelated to Sonneratia (Muller, 1978, 1981a). Sonneratioipollis bellus Venkatachala and Kar (1968) is described with a psilate exine and from the illustrations lacks colpi, pseudocolpi, and the protruding pores typical of Sonneratia. Jugopollis tetraporites Venkatachala and Rawat (1972), Florschuetzia minutus Rawat et al. (1977), and F. cf. meridionalis (Rawat et al., 1977) were also rejected as lythracean by Muller.

A relatively faint leaf impression, Sonneratia meghalayensis Ambwani (1991), from the Paleocene at Meghalaya, India has been attributed to Sonneratia. However, it has an obtuse leaf apex and craspedodromous venation unlike leaves of extant Sonneratia in which the leaf apex is typically mucronate and thickened and venation is brochidodromous. The intramarginal vein is not seen, but it is not always visible on all leaves or in all species of modern Sonneratia (Duke & Jackes, 1987; S. Graham, pers. observ.). Sonneratia meghalayensis cannot be assigned with certainty to Sonneratia. Fossilized roots called Sonneratiorhizos raoi Chitaley (1968) have also been reported from the Paleocene of India; their relationship to Sonneratia was doubted by Kramer (1974) and the relationship is rejected here based on the presence of non-sonneratian type xylem with diffuse parenchyma. The relationship of Sonneratia to the fossil flowers and fruits of the extinct genera Sahnianthus and Enigmocarpon was rejected by Muller (1978) and present research supports that conclusion (see Sahnianthus).

In summary, the earliest well-documented occurrences of Sonneratia are wood of Sonneratioxylon preapetalum Awasthi (1969) from the early Paleocene of India (Danian, 67.3-63.8 Ma); Sonneratia-like pollen of Florshuetzia sp. from the late Paleocene of France (Thanetian, 58.7-55.8 Ma) (Gruas-Cavagnetto et al., 1988); and wood of S. aubrevevillei Louvet (1970) from the middle Eocene of Libya (48.6-40.4 Ma). Fossil pollen is first associated with Sonneratia in the middle Eocene deposits of southeastern Asia, specifically with the earliest occurrence of Florshuetzia trilobata in Central Java, a type Muller (1981a) viewed as not quite fully Sonneratia. The earliest pollen directly attributable to living species of Sonneratia is that of S. caseolaris (F. levipoli) from the base of the early Miocene of Borneo (ca. 19 Ma) followed by S. alba pollen (F. meridionalis) from the middle Miocene of Borneo (Muller, 1984; Morley, 2000). The Miocene record of S. caseolaris pollen in Borneo marks the beginning of modern Sonneratia as an obligate mangrove genus, earlier Florshuetzia trilobata being described from fresh-water and saline deposits. The unconfirmed Turonian record of fossil wood of Sonneratioxylon turonicum in Central Asia at present cannot be confirmed as Sonneratia.

Tamesicarpum Reid & Chandler. This is a monospecific genus of incompletely known fruits from the London Clay flora. It is doubtfully related to the Lythraceae (Reid & Chandler, 1933). Tamesicarpum polyspermum Reid and Chandler from Divisions A2, D & E, and from Hampshire Basin B1 & B2, consists of casts of three- to six-loculed fruits with evidence of loculicidal dehiscence and casts of tightly packed seeds attached to an axile placenta. The tegmen of the seeds differs from the tegmic longitudinal arrangment of narrow tracheoids typical of Lythraceae seeds. As Reid and Chandler concluded, the totality of evidence is insufficient to confirm a relationship to the Lythraceae.

Trapa L. and the related extinct genus Hemitrapa Miki. Trapa and the extinct genus Hemitrapa are aquatic plants with extensive fossil fruit and pollen records in Eurasia and North America. The fruits of each genus at their morphological extremes are unique but they vary to the extent that some specimens closely approach a common form that makes taxonomic separation difficult (Vassiljev, 1960; KovarEder et al., 2002, 2005; Wojcicki & Kvacek, 2003). The difficulties are apparent when fruits are compared among the numerous species described (see Miki, 1952a: fig. 2; Gregor, 1982a: fig. 3; and Mai, 1985: fig. 11). Trapa and Hemitrapa are discussed here under a single heading. They are listed separately and cross-referenced in the Appendix.

The distinctive horned fruits of Trapa, commonly known as water chesnut or water caltrop, are among the most intensively studied and most easily recognized fossils in the Lythraceae (Fig. 9g). Trapa is frequent in Miocene swamp deposits in Europe and Asia and persists today in quiet lakes, ponds and slow rivers in northern, central, and southeastern Europe, and in tropical to temperate Asia and Africa. Trapa is a major invasive in parts of eastern North America and Australia. The edible starchy seeds are consumed in China and India although they are less used than the common water chestnut of Chinese cooking which is a corm of the sedge Eleocharis dulcis (Burm. f.) Henschel.

Trapa is characterized by floating rosettes of coarsely toothed, rhombic leaves with inflated petioles, leaf margins with double-mucronate tooth apices, and indurated two or four-horned fruits. The modern relationships of the genus have been difficult to establish due to the accumulation of extensive morphological and embryological autapomorphies. The genus has been classified as the monogeneric family Trapaceae, or under the the superfluous illegitimate name Hydrocaryaceae, or has been erroneously included in the unrelated Plantaginaceae (formerly Trapellaceae). Some affinity to the Onagraceae or Lythraceae has long been recognized. Following results of multi-gene phylogenetic analyses modern Trapa is now accepted as unquestionably derived within the Lythraceae where it is strongly supported as sister to Sonneratia (Graham et al., 2005).

Trapa fruits are heavily indurated floral tubes enclosing an inferior ovary. They almost always have a prominent basal abscission scar from a non-persistent peduncle. Typically there are two (to four) strongly recurved to ascending horns (indurated sepals) with thick bases. The horns bear coarse harpoon-like barbs toward the apex. The upper portion of the ovary forms a cylindrical, sunken neck. Variation in size, shape, number and position of the horns, and tubercule-like projections on the fruit surface have been used to define extinct and modern species. Eighty-one modern species and innumerable varieties and forms have been named (International Plant Names Index; http://www.ipni.org). Far fewer extant species are accepted worldwide, often just three, or only one polymorphic species with a number of varieties (Verdcourt, 1998).

Unquestioned modern Trapa fruit forms begin in the middle Miocene. Collections of the fossil fruits are extensive and have been studied intensively by paleobotanists in Europe, Russia, and Japan. Extensive morphological variability was present in fossil Trapa fruits in the Miocene of Europe, Asia, and western North America and the larger number of species described from the middle and late Miocene suggests that morphological and geographical diversification of the genus was rapid during that time. Miki (1952b) described 16 fossil species of Trapa fruits from Japan. At least 16 species are recorded for the Miocene and Pliocene of Europe (Kovar-Eder et al., 2005; Wojcicki & Velitzelos, 2007). Ticleanu and Diaconia (1997) recorded five species from Pliocene of the coal basin at Olteni, Romania. Diversity of fruit types diminishes considerably in European Pliocene and Pleistocene deposits (Gregor, 1982b; Mai, 1985). The genus was present in the Holocene of England until 4,200 B.P. (Flenley et al., 1975; Schofield & Bunting, 2005), was abundant as far north as southwestern Finland in nutrient-rich lakes ca. 2,500 years ago (Korhola & Tikkanen, 1997), and lingered on as far north as southern Sweden until 1915 (Jonsell & Karlsson, 2010). It continues to decline in continental Europe where it is now a protected species.

Trapa is positively documented from the middle/late Miocene in China (Wojcicki et al., 1999) and was possibly present in the middle Miocene of Japan (Tanai & Suzuki, 1963). The present distribution of Trapa includes the African continent (Verdcourt, 1998).

Hemitrapa, an extinct genus of fossil fruits and pollen was once widespread in swamps and rivers across the northern latitudes of Europe, Asia, and North America. Fruits consistent with H. trapelloidea, the type species of the genus from the Miocene of Japan, are bowl-shaped with a protruding half-inferior ovary and are ca. 18-20 x 6-9 mm (Fig. 9a: Miki, 1941, 1952a, 1959). They taper basally to a slender persistent peduncle. The margin of the persistent floral tube bears slender, thread-like appendages (Fig. 9a) with terminal, densely brushy hairs. The appendages extend from about mid-length of the fruit, are typically ascending, and alternate with four smaller spine-like appendages (Miki, 1941, 1959). The number of appendages (arms, horns, spines) is usually two or four, as in Trapa. The appendages are most likely sepals and the smaller spines may represent segments of a lythracean epicalyx or may be merely tubercles. A thick, conical, striated protrusion, the exserted portion of the maturing fruit, emerges from mid-level of the floral tube tapering distally to form a neck which is not sunken as it is in Trapa (Miki, 1952a: fig. 2). The entire structure, i.e. the persistent floral tube plus the partially enclosed fruit, is thinner and less woody than in extinct and extant Trapa.

The separation of Hemitrapa and Trapa is made difficult by variations in fossil fruit shape, width, degree of taper to the fruit base, and the degree of development of the spine base. Miki transferred four species previously classified as fossil Trapa to Hemitrapa: the Asian H. yokoyamae (Nath.) Miki, H. hokkaidoensis (Okutsu) Miki, and H. sachalinensis (Okutsu) Miki; and the Asian/North American species H. borealis (Heer) Miki (Nathorst, 1888; Okutsu, 1939; Miki, 1948, 1952a). Hemitrapa was reduced to a subgenus of Trapa by Vassiljev (1960) who found no constant features in the fruits to separate the two genera. Gregor (1982a) treated Hemitrapa as distinct from Trapa and informally recognized two sections of Hemitrapa based on differences in the spines or horns.

A few other fossil species of Trapa have been transferred to Hemitrapa; alternatively, a few Hemitrapa are now recognized as Trapa. The species T. fritzlariensis Wojcicki and Wilde (2001), based on a single well preserved fruit from the Pliocene of Germany, has several features morphologically intermediate between Trapa and Hemitrapa. It is similar to the common European Hemitrapa heissigii Gregor (Fig. 9d; Gregor, 1982a) which is a stratigraphic marker for the middle Miocene in fossil deposits at Bayern, Germany. It is also similar to some collections from North America of T. americana Knowlton that are difficult to place to genus (e.g., see illustrations in Brown, 1935, 1937). Trapa silesiaca Goepp., one of the most common Trapa in the late Miocene of Europe (Kovar-Eder et al., 2005), was frequently confused in the past with Hemitrapa (Gregor, 1982a, Wojcicki & Zastawniak, 1998; Wojcicki & Kvacek, 2002b). Differences in compression during fossilization and the quality of preservation further complicate recognition. Kovar-Eder et al. (2002) believe Hemitrapa pseudoborealis Budantzev (1960) from Tertiary deposits on the southeastern coast of Lake Baikal most likely is a Trapa fruit compressed in an atypical plane.

To my knowledge the presence of Hemitrapa generally has not been recognized in North America. Wolfe and Tanai (1980) pointed out the transfer of T. borealis by Miki (1952a) to Hemitrapa (cited by Wolfe & Tanai as a member of the Trapellaceae). However, Hemitrapa borealis has not been cited in publications by North American paleobotanists, all specimens being referred to T. borealis.

Both Hemitrapa and modern forms of Trapa occur in fossil deposits in western North America. Six species of fossil Trapa have been named from North America: T. borealis Heer, T. americana Knowlton,, T.? occidentalis Knowlton, T. prenatans Dorf, T. alabamensis Berry and T. wilcoxensis Berry. An additional species by Knowlton appears to be a nomen nudum and has not been taken up in the literature. The first species to be described in North America was Trapa borealis (Heer, 1869a; Hollick, 1936) from abundant fruits in an early to middle Miocene deposit at Cook Inlet, Kenai Peninsula, Alaska. (Some Arctic floras described as Miocene by Heer were later found to be Paleocene in age; the age of the deposits containing T. borealis were verified as Miocene by Wolfe et al., 1966). Heer's T. borealis is clearly different from typical Miocene Trapa fossils (Fig. 9b; Heer, 1869a) and corresponds to Hemitrapa. Additional collections of the same Hemitrapa fruit type have since been recovered from the same area at Port Graham, Alaska (Knowlton, 1894; Wolfe et al., 1966; Wolfe & Tanai, 1980; Dickinson, 1995).

Hemitrapa borealis (as Trapa borealis) was also identified from the Paleocene of southernmost Alberta and Saskatchewan (Badlands west of Wood Mountain), Canada based on "a few obscure prints [that] seem to indicate a species of this genus which may be identical with the above species described by Heer from Alaska" (Dawson, 1875: 330; 1886: see incomplete fruit sketch, fig. 19). Trapa borealis is listed from late Paleocene/early Eocene Porcupine Creek and Red Deer River sediments in Alberta, Canada and from the Eocene or late Paleocene of "Spitzbergen, Alaska, Greenland, etc." by Penhallow (1907, 1908). Remains from the early/middle Miocene Weaverville Formation California were identified by MacGinitie (1937) as Trapa borealis. Examination of several detailed digital images of the Weaverville collection, supplied to the author by D. Erwin, Palaeontology Museum, University of California, fails to establish the identity. The difficulty lies in the incomplete exposure of the fruits and their irregular compression shapes and orientation. What may be two appendages with thick, somewhat ascending bases more like Trapa than Hemitrapa are visible. The width between horns is about 17 mm. No peduncles are apparent, also suggesting Trapa. The fruits may be those of Trapa americana.

Of the Trapa species named for North America, Trapa americana is the most commonly found. Described from the Miocene Payette Formation, Idaho, it is a relatively small (17 mm between horns, 20 mm long including horns), two-horned, wedge-shaped fruit with horizontal or ascending horns and a striate, central prominence (Fig. 9c). Knowlton (1898a: 734) considered it "quite different" from T. borealis, although "quite like" one of the T. borealis specimens illustrated by Heer, (Fig 9b; Heer, 1869a: fig. 11). Trapa? occidentalis Knowlton, also from the Payette Formation, was placed in synonomy of T. americana (Berry, 1928) and the name has not been used in later works. Occurrences of Trapa americana are recorded from middle/late Miocene floras of the Payette Formation, Idaho (Knowlton, 1898a,b; Brown, 1937), the Esmeralda Formation, Nevada (Berry, 1928); the Latah Formation, Idaho (Brown, 1935); and the Stinking Water Flora, Oregon (Chaney & Axelrod, 1959). Some of the collections determined as Trapa americana appear more like Hemitrapa than Trapa (e.g., see illustrations of the peduncled specimens of Brown, 1935, 1937). A fruit identified from a middle Miocene flora at Juliaetta, Idaho, as Trapa (Pigg & Wehr, 2002) is very similar to the European fossil Hemitrapa heissigii (cf. Fig. 9d & e). In contrast, several robust specimens of T. americana from the Miocene of Oregon (Fig. 9f; Chaney & Axelrod, 1959) are typical of modern Trapa fruits (Fig. 9g). Abundant remains of T. americana with robust appendages are also found in the late Pliocene Sonoma Flora at Neer's Hill, California (Axelrod, 1944). Trapa prenatans Dorf (1936), described as a two-horned species from Pliocene deposits in southwestern Idaho, was placed in synonymy of T. americana by Brown (1937). This fossil has the sunken, somewhat cylindrical projection expected in Trapa. Axelrod (1944), who examined types of T. americana, T.? occidentalis, and T. prenatans, interpreted all as 4-horned fruits of Trapa americana. The relationship of Hemitrapa and Trapa fruits and the identity of the North American collections require more detailed study and comparisons with the European and Asian fossils. Examination of any fossil pollen that may be present at the same localities may aid taxonomic interpretations (see below).

In eastern North America fossils of Hemitrapa are unknown whereas two fossil species of Trapa fruits have been described, T. wilcoxensis and T. alabamensis. The oldest is T. wilcoxensis Berry from the Eocene of Tennessee (Berry, 1914a, b, 1916a; Parks, 1975). Three fossils of the species were very simply diagrammed by Berry in the protologue. The three syntype specimens are too poorly preserved and faint to confirm the presente of Trapa or to even locate the specific fossils on the rock matrix with any certainty (pers. observ. of several excellent digital images provided by the Department of Paleontology, Natural History Museum, Smithsonian Institution).
COPYRIGHT 2013 New York Botanical Garden
No portion of this article can be reproduced without the express written permission from the copyright holder.
Copyright 2013 Gale, Cengage Learning. All rights reserved.

Article Details
Printer friendly Cite/link Email Feedback
Title Annotation:p. 48-97
Author:Graham, Shirley A.
Publication:The Botanical Review
Article Type:Report
Geographic Code:1USA
Date:Mar 1, 2013
Words:22897
Previous Article:A review of the fossil record of the genus Itea (Iteaceae, Saxifragales) with comments on its historical biogeography.
Next Article:Fossil records in the Lythraceae.
Topics:

Terms of use | Privacy policy | Copyright © 2020 Farlex, Inc. | Feedback | For webmasters