scientific correspondence Fish with fingers?

ingers and toes were long thought to be Figure 1 Comparison of fish and tetra- Fnovelties associated with the invasion of pod pectoral appendages. a, Ventral land by tetrapods. The recent identification view of newly discovered right pec- of a variety of aquatic specializations in toral fin of a rhizodontid sarcoptery- some early tetrapods has provoked a debate gian fish (ANSP 20581); proximal8 on whether digits arose in primarily terres- humerus not exposed. Scale bar, 1,2 trial or aquatic animals . We recently dis- 2 cm. Note (1) the array of jointed and covered a pectoral fin of a lobe-finned fish branched preaxial radials; (2) com- (Fig. 1a, b) that is remarkably similar to mon proximo-distal termination of tetrapod limbs. This discovery reveals that ulnare and intermedium; (3) lack of major tetrapod novelties are also seen in the postaxial process on the ulnare. b, paddles of some closely related fish and Line drawing of ANSP 20581, proximal therefore need not have arisen to meet the humerus reconstructed. c, Pectoral fin demands of a terrestrial existence. of the osteolepiform sarcopterygian We discovered the new fin along a road- Eusthenopteron. d, Pectoral limb of side in north central Pennsylvania (Catskill the Late Devonian tetrapod Acan- Formation; Late Devonian; ~370 Myr BP). thostega. Shaded elements empha- Features of the shoulder girdle and fin rays size the similarity in pattern of fin clearly identify it as belonging to a rhi- radials and digits. H, Humerus; U, 3 zodontid sarcopterygian . The specimen ulna; r, radius; u, ulnare; i, intermedi- compares favourably with Sauripterus, um; b, c, d are not to scale. known from a single specimen discovered in 1840, in having a broadly flattened radius. The quality of preservation of the new fin permits comparisons with limbs that were impossible until now. The closest relatives of tetrapods, ‘osteo- lepiform’ and elpistostegalid sarcoptery- Phylogenetic comparisons with other suggests that digits could have evolved for gians4, have fins designed on a simple sarcopterygians are complicated by missing reasons other than bearing weight during bifurcate pattern and do not have structures data but current hypotheses place rhi- terrestrial locomotion. that can be readily compared with digits zodontids just outside a group containing Edward B. Daeschler (Fig. 1c). In contrast, the new fin contains ‘osteolepiformes’, elpistostegalids and tetra- Neil Shubin an array of eight distally facing and jointed pods4,5. This suggests two possibilities: fin- Vertebrate Biology, preaxial radials that are superficially similar gers are either primitive to stem tetrapods, Academy of Natural Sciences of Philadelphia, in number and configuration to the digits or digit-like structures evolved indepen- 1900 Benjamin Franklin Parkway, Philadelphia, of early tetrapods (Fig. 1d). In addition, six dently in a closely related group of Devon- Pennsylvania 19103, USA of these radials articulate with homologues ian fish. Either phylogenetic interpretation and Department of Biology, of the carpus (intermedium and ulnare) at a forces us to question the use of digits as a University of Pennsylvania, Philadelphia, common proximo-distal level. The radials key innovation associated with the origin of Pennsylvania 19104, USA of the new fin differ from tetrapod digits in tetrapods. 1. Clack, J. A. & Coates, M. I. Bull. Mus. Natl Hist. Nat., Paris 17, that they are flattened and encased by stiff This discovery reveals that some 359–372 (1995). unjointed dermal fin rays (Fig. 1a, b). Devonian fish acquired a truly mosaic fin 2. Ahlberg, P. E. & Milner, A. R. Nature 368, 507–514 (1994). Taken together, these characteristics sug- skeleton, possessing both an extensive and 3. Long, J. A. J. Vertebr. Paleontol. 9 (1), 1–17 (1989). 4. Cloutier, R. & Ahlberg, P. E. Interrelationships of Fishes, 445–479 gest that the digit-like structures were not limb-like endoskeleton and elaborate der- (Academic, New York, 1996) the primary load-bearing elements of the mal fin rays. The presence of digit-like 5. Schultze, H.-P. Origins of Higher Groups of Tetrapods, 29–67 distal portion of the rhizodont appendage. structures in the paddle of an aquatic fish (Comstock, Ithaca, 1991).

original finding. Scientists will thus once gramme of the Missouri Botanical Garden Takhtajania perrieri again have a ‘living fossil’ to study and and the World Wide Fund for Nature, col- place in the context of both extinct and lected a flowering tree in the Anjaha- rediscovered other extant primitive angiosperms. naribe-Sud Special Reserve southwest of At the suggestion of Peter H. Raven, Andapa. One of us (G.E.S.) identified the Spectacular finds of early Cretaceous fossil beginning in 1974 with the late Alwyn specimens as Takhtajania. The new locality flowers during the past decade have fuelled Gentry, Missouri Botanical Garden is 150 km to the east of the original collec- a resurgence of research into the origin botanists and their Malagasy colleagues tion. A large population of over 250 adults of flowering plants1, the “abominable mys- searched in vain at the Manongarivo Spe- has been found at the new site, where new tery” that so captivated Darwin. Now cial Reserve in northwestern Madagascar, research material has already been collect- Takhtajania perrieri (Capuron) Baranova where French botanist Henri Perrier de la ed: wood samples, leaves in silica gel for & J.-F. Leroy, the only extant Africa/ Bâthie had collected the only specimen in DNA extraction, flowers and fruit in spirit. Madagascar representative of the family 1909. In 1994, the Malagasy plant collector The Winteraceae, with about 65 species Winteraceae, has been rediscovered in Fanja Rasoavimbahoaka, performing a which exhibit a relict Gondwanan distrib- northeastern Madagascar, 85 years after its botanical inventory as part of a joint pro- ution, are thought to be among the oldest NATURE | VOL 391 | 8 JANUARY 1997 Nature © Macmillan Publishers Ltd 1998 133 scientific correspondence families of flowering plants (along with 1273–1275 (1983). show that the somatosensory cortical repre- Chloranthaceae) on the basis of fossil 3. Doyle, J. A., Hotton, C. L. & Ward, J. V. Am. J. Bot. 77, sentation of the left hand of string players, 1544–1557; 1558–1568 (1990). pollen types assigned to modern families. 4. Coetzee, J. A. & Muller, J. Ann. Missouri Bot. Gard. 71, which engages in the complex and The Cretaceous fossil pollen genus Walk- 1088–1099 (1984). demanding task of fingering the strings, is eripollis from the late Barremian – early 5. Praglowski, J. in World Pollen and Spore Flora (ed. Nilsson, expanded1. To test the hypothesis that S. T.) 8, 1–36 (Almqvist & Wiksell, Stockholm, 1979). Aptian of Gabon and the late Aptian – 6. Suh, Y., Thien, L. B., Reeve, H. E. & Zimmer, E. A. Am. J. Bot. fusion of cortical representations is caused 2,3 7–9 early Albian of Israel is winteraceous . 80, 1042–1055 (1993). by increased simultaneous stimulation , Winteraceae persisted in continental 7. Vink, W. Taxon 37, 691–698 (1988). we tested the somatotopic representations Africa until at least the lower Miocene in 8. Capuron, R. Adansonia (new series) 3, 373–378 (1963). of four blind Braille readers who used three 9. Leroy, J.-F. Adansonia (new series) 17, 385–395 (1978). 8 the southwestern Cape, represented there 10.Deroin, T. & Leroy, J.-F. C.R. Acad. Sci., sér. 3 316, 725–729 fingers (digits 2–4) of both hands simulta- by two different pollen types associated (1993). neously for reading, and were instructors of with modern-day Bubbia and Tasmannia 11.Takhtajan, A. Diversity and Classification of Flowering Plants the method typically engaged in this prac- (sometimes considered Drimys section (Columbia Univ. Press, New York, 1997). tice for several hours a day. We also tested Tasmannia), both now restricted to Aus- six Braille readers who employed one finger tralo-malesia4. The pollen of Takhtajania, for reading, and five sighted non-Braille- which is the largest reported in the family, reading subjects. Magnetic source imaging has been compared to that of Drimys sensu Changed perceptions in was used to determine the centre of cortical stricto, which is distributed only in the Braille readers magnetic responsivity to light tactile stimu- New World5. lation of the finger tips (left and right D1, Recent molecular phylogenetic studies D2 and D5) and right and left lower lip1. utilizing internal transcribed spacer (ITS) The mature mammalian nervous system There was a substantial enlargement of sequences of ribosomal DNA aimed at has a striking capacity for plastic remodel- the hand representation in the three-finger resolving genetic relationships within ling in response to environmental changes, Braille readers compared with the other Winteraceae6 largely corroborated pro- but little is known about the perceptual and two groups (three-finger readers, 14 mm; posed relationships based on morphologi- behavioural relevance of this phenomenon. one-finger readers, 8 mm; sighted, 7 mm; 7 ǃ < cal features , but also clearly illustrate how Using magnetic source imaging we show F2,12 16.6, P 0.001). The three-finger widely divergent Tasmannia is from that the cortical somatosensory representa- readers also differed from sighted subjects Drimys. Preliminary analysis of the tion of the fingers is topographically disor- in the topographical arrangements of the Takhtajania ITS sequence indicates a basal dered in blind Braille readers who use three finger representations along the postcentral position for the genus within Winteraceae fingers on both hands to read. In addition, gyrus (Fig. 1). In all sighted subjects the (Elizabeth Zimmer, personal communica- they frequently misperceive which of these expected homuncular pattern was tion). fingers is being touched. This correlation is observed; D1 being inferior (lateral), D2 Takhtajania perrieri was originally suggestive of a functional role for cortical being more medial, and D5 being the most described in 1963 by the French forest reorganization in the perceptual experience superior (medial). In each of the three-fin- botanist Réné Capuron in the genus Bub- of these individuals. ger Braille readers, this pattern was changed bia, known from Australia, New Caledonia The somatosensory cortical representa- in one or both hemispheres. Only one of and the Lord Howe Islands8. On the basis tion of fingers expands when the amount of the one-finger Braille readers’ cortical of several anomalous features unique sensory input arriving in that part of the topography of finger representations was within Winteraceae, including a putative brain is increased1–6. Indeed, we have previ- disordered. These are significant differences bicarpellate ovary and a primitive arrange- ously used magnetic source imaging to (P< 0.005, Fisher’s exact test, 2-tailed), ment of undifferentiated cells surrounding the stomata, Leroy and Baranova in 1978 created a new genus and subfamily to Left hemisphere Right hemisphere accommodate the species, naming it in 3-finger 1-finger Sighted 3-finger 1-finger Sighted honour of the Russian plant systematist M M Armen Takhtajan9. Recent anatomical 1. 5 M studies utilizing the type material support 1. 0 M M the hypothesis of a compound bicarpellate M M 10 0.5 X ovary . In St Louis in June 1997, while cel- -axis in cm) X z M ebrating his 87th birthday and the publica- 0.0 tion of his latest synthesis on the -0.5 classification of flowering plants11, Takhta- X X X X jan was presented with a specimen of -1.0 Takhtajania for the Komarov Botanical Distance to D1 ( -1.5 Institute in St Petersburg. D5 George E. Schatz, D2 Porter P. Lowry II D1 Missouri Botanical Garden, P.O. Box 299, St. Louis, D2–D1 D5–D1 Missouri 63166, USA e-mail: [email protected] Annick Ramisamihantanirina Missouri Botanical Garden, B.P. 3391, 101 Figure 1 Distances between centres of cortical responsivity: D1, thumb; D2, index finger; D5, little finger. Dis- Antanananarivo, Madagascar tances are displayed with reference to the D1 location, which is graphed at the zero-line. The inset shows the centre of responsivity for D1, D2 and D5 superimposed on the brain of a sighted subject. X, Hemispheres 1. Friis, E. M., Pedersen, K. R. & Crane, P. R. Pl. Syst. Evol. (Suppl.), 8, 31–49 (1994). in which the homuncular organization of the cortical representations of the fingers was disordered; M, corti- 2. Walker, J. W., Brenner, G. J. & Walker, A. G. Science 220, cal representation of hands in which the location of tactile stimulation was mislocalized. 134 Nature © Macmillan Publishers Ltd 1998 NATURE | VOL 391 | 8 JANUARY 1998