Opinion TRENDS in Plant Science Vol.8 No.8 August 2003 369 Archaefructus – angiosperm precursor or specialized early angiosperm? Else Marie Friis1, James A. Doyle2, Peter K. Endress3 and Qin Leng4 1Department of Palaeobotany, Swedish Museum of Natural History, Box 50007, SE-104 05 Stockholm, Sweden 2Section of Evolution and Ecology, University of California, Davis, CA 95616, USA 3Institute of Systematic Botany, University of Zurich, 8008 Zurich, Switzerland 4Nanjing Institute of Geology and Palaeontology, CAS, 39 East Beijing Road, 210008 Nanjing, The People’s Republic of China With molecular analyses indicating that angiosperms Ovulate and microsporangiate organs are not closely related to any other extant seed plant Reproductive axes are elongate with ovulate organs borne group, information from fossils might provide the only distally and microsporangiate organs borne proximally. basis for reconstructing their origin. Therefore the Some specimens apparently have unisexual (ovulate) axes. description of a new Early Cretaceous angiosperm, The organs are crowded in young specimens (Fig. 1a) but Archaefructus, placed as the sister of all extant angio- spaced out in more mature specimens (Fig. 1b). Ovulate sperms, has created much debate and optimism. organs are pedicellate and typically borne in pairs on each However, we question both the interpretation and the pedicel (Fig. 2). They are elongate with an acuminate apex analysis of Archaefructus, concluding that it might be a and enclose few (A. liaoningensis) or many (A. sinensis) crown-group angiosperm specialized for aquatic habit ovules or seeds. rather than a more primitive relative. Although ovulate organs of Archaefructus have not been shown to have all the features of carpels, in other The family Archaefructaceae [1] was established to respects they are more like angiosperm carpels than they accommodate two extinct species, Archaefructus liaonin- are like the ovulate organs of other seed plants with gensis [2] and Archaefructus sinensis [1], from the Yixian enclosed ovules (Box 2). However, it is not clear that they Formation in northeastern China. In the initial report [2], were conduplicate (i.e. like a leaf folded lengthwise and A. liaoningensis was presented as the oldest known sealed along its margins), or that the fruits were follicles, angiosperm because the beds were dated as Late Jurassic. as described by Sun et al. [1,2]. Follicles open long- However, radiometric dating strongly supports a mid- itudinally along the ventral side. No such open fruits and Early Cretaceous age for the Yixian Formation [3], close to the age of the first records of angiosperms in other areas (Box 1). Box 1. Age and palaeoenvironment of Archaefructus The Archaefructaceae were defined as including small The Archaefructus fossils were all collected in the Yixian Formation, herbaceous water plants with axillary branches terminat- outcropping in the western part of Liaoning Province in northeastern ing in reproductive organs, and with strongly dissected China. The Yixian Formation is part of the Jehol Group, which has yielded a wealth of exquisitely preserved fossil animals, most leaves. Reproductive axes were assumed to be exposed notably feathered theropod dinosaurs and a diversity of early above water level and were described as solitary flowers birds. The sediments were deposited in a low-energy lacustrine without perianth, with numerous stamens borne in environment in a seasonally semiarid climate and are intercalated helically arranged pairs and numerous conduplicate with volcanic tuffs and basalts (the environment and biota are reviewed in Ref. [3]). Although the faunal elements of the Jehol biota carpels borne in a helical or whorled arrangement and exhibit exceptional preservation, the plant fossils rarely have cellular maturing into multi-seeded follicles. details intact, and tissues are often replaced or infilled with pyrite Ge Sun et al. [1] evaluated the phylogenetic position of framboids and microcrystallines [33]. However, some fossil plants Archaefructus by including its morphological characters in have been preserved with many of the organs (roots, stems, leaves) still in organic connection. This is unusual for Cretaceous plant a combined three-gene molecular and morphological fossils and provides unique information about growth form in early analysis, which placed Archaefructus as the sister to angiosperms. The only contemporaneous flora with whole-plant extant angiosperms. We will discuss the characters of preservation is the Crato flora from the Early Cretaceous of Brazil, Archaefructus in detail and question aspects of their deposited in a similar low-energy lake system [34]. The thickness of the Yixian Formation reaches 4700 m, with plant interpretation. Our conclusions are based on published and animal fossils occurring throughout. The age of the Yixian descriptions and illustrations [1,2,4] and our own studies Formation has been much debated [35] but there is now strong of additional specimens housed in the Institute of evidence from radiometric dating that it is Early Cretaceous [3,36–38]. Vertebrate Paleontology and Paleoanthropology, CAS, The oldest Archaefructus is from the Jianshangou Bed, which is in the lower part of the Yixian Formation and dated radiometrically to Beijing, China. ,125 million years, corresponding to the Barremian stage [3], close to the age of the first well-known angiosperm floras Corresponding author: Else Marie Friis ([email protected]). elsewhere [31,39]. http://plants.trends.com 1360-1385/03/$ - see front matter q 2003 Elsevier Ltd. All rights reserved. doi:10.1016/S1360-1385(03)00161-4 370 Opinion TRENDS in Plant Science Vol.8 No.8 August 2003 Box 2. Reproductive structures of Caytonia and angiosperms Several non-angiospermous Mesozoic seed plants resemble Archae- fructus and extant angiosperms in having enclosed ovules, but their enclosing structures clearly differ from carpels or are poorly under- stood. The best examples are Caytonia (Caytoniales, a ‘Mesozoic pteridosperm’ [16]), Irania (related to Czekanowskiales? [40]), and Leptostrobus (Czekanowskiales [41]). Caytoniales were first described by Hamshaw Thomas [42] from Yorkshire (UK) as Jurassic angiosperms, but later Tom Harris [16] documented that Caytonia had gymnospermous reproduction, with pollen germinating inside the micropyles of the ovules. Subsequently, the homology of the ovule-bearing structures (‘cupules’) with carpels was questioned. The cupules appear to be borne in two rows on a bilateral axis, like leaflets on the rachis of a compound leaf, unlike carpels, which are borne like leaves on a radial stem. Caytonia cupules are circinately curved toward the axis; their plane of symmetry is perpendicular to the axis, and ovules form an arc perpendicular to this plane of symmetry. By contrast, the ovulate structures of Archaefructus are elongate and acuminate, their plane of symmetry is parallel to the axis, and ovules are attached parallel to the plane of symmetry – typical features of carpels. Leptostrobus also has ovulate organs arranged along an axis. They consist of two valves, but each valve seems to differ from a carpel in its symmetry and in bearing ovules in an arc perpendicular to its plane of symmetry [41]. Irania also has paired ovulate organs borne along an axis [40], but the detailed structure of the organs is unknown and a more detailed comparison is not possible. Fig. 1. Line drawings showing details of the reproductive organs of Archaefructus The ovules of Caytonia have only one integument, rather than two, sp. (a) and Archaefructus liaoningensis (b,c) from the Early Cretaceous Yixian For- the ancestral state in angiosperms [17]. Sun et al. [2] described ovules mation of northeastern China. (a) Immature reproductive axis with densely spaced of Archaefructus as bitegmic and anatropous (with two integuments ovulate organs (carpels). (b) Mature reproductive axes with carpels above (ovulate and a reflexed orientation), but unfortunately their structure is zone) and microsporangiate organs (stamens) below (staminate zone). (c) Basal part of a mature reproductive axis showing clustering of the stamens in the stami- uncertain. Even in well-preserved specimens (B0112, Fig. 2) we failed nate zone; a few carpels of the ovulate zone are also shown. The paired arrange- to establish the number of integuments, nature of internal structures, ment of both the carpels and stamens is evident in several places (arrow heads). or orientation and attachment of ovules. Scale bar ¼ 1 cm. (a) Redrawn from Plate 32, Fig. 6, in Ref. [4]; (b) redrawn from Because they both have four microsporangia, stamens of Plate 32, Fig. 3 in Ref [4]; (c) redrawn from Plate 31, Fig. 1, in Ref [4]. angiosperms have often been compared with the synangia of Caytoniales (Caytonanthus) [42]. These are borne in groups on lateral stalks along an axis. The stalks often branch several times, but occasionally there are only two or three synangia grouped closely on an unbranched stalk, suggesting the arrangement in Archaefructus. However, Caytonanthus synangia are radial structures with four equally spaced microsporangia, rather than bilateral structures with (a) (b) two pairs of sporangia on either side of a sterile connective, like angiosperm stamens. Ovulate zone Caytonia has figured in discussions of angiosperm origin, but with the cupules interpreted as homologous with the bitegmic ovules of angiosperms, not with the carpels [12,14,43–46]. no dehiscence line were observed. Even in living material