sign in sign up
<<
Home , Anapsid

scientific correspondence

cu v, (Fig. 2). Retinal analogues (all-trans-retinal, Q) cu .c cu (/) e- E ·a, n-hexenal, 13-apo-12'-carotenal, and 2E,4E• Q) U) Q) 0 cu Q) cu E C' "O cu :§ 2 § E .§ octadienal) could all be incorporated to Uncertain cu ~~ "O ::, l!! ~ cu .~ ,2 e Q. .§ ·c ai ::, cu U) CJ ·c cu (/) ::, restore phototactic sens1tlv1ty of the 9 ~ li (/) (/) e 0 CJ Q. Q) £ ·a, 0 "O ::, relationships Cl) cu ~ 0 0 "O C E a Q) i::, C .c cu bleached zoospores, showing that the loss of >, ~ C ~ Q. ::, ·c. (/) Q) -- >- ro cu [I? .!JI t' Q) .!JI 0 sensitivity was due to the loss of native (/) 0 (/) Cl. 0 <( (.) §?. <( _J , o_" w chromophore. Turtles have generally been regarded as Many retinal analogues10 have been basal because they apparently retain bound in vitro to bovine opsin and the primitive anapsid pattern (no tem­ bacterio-opsin, and in vivo to the opsin of poral fenestrae) unlike other extant reptiles the green alga, Chlamydomonas reinhardtii. which are (two temporal fenestrae These studies have revealed the chemistry of on each side). However, the phylogenetic binding, the electric field and charge distri­ relationships between and other bution within the binding site, and the is not well understood. Despite shape of the site. Most retinal analogues some disagreement over the details, recent shift the action spectrum away from that of phylogenetic analyses 1- 3 seemed to be reach­ the native pigment and also from the ing a consensus that the turtles are the only Figure 1 inter-relationships based on the absorption spectrum of the unbound ana­ surviving members of the , and data of ref. 4 showing the uncer1ain position of tur• logues, demonstrating their incorporation thus only very distantly related to other tles. 1, Tu r11des as iapsids in unconstrained most and function. extant reptiles. In contrast, by considering parsimonious trees, requiring 770 steps'. 2, Tur1les We compared retinal analogues bound more morphological characters and more as parareptiles as posited in other studies1-3, requir• in zoospores with the native chromophore taxa than in previous studies, Rieppel and ing only three extra steps. that peaks at 536 nm. The analogue n-hex­ deBraga4 propose that turtles are crown­ enaI7 blueshifts to 339±32 nm, 13-apo-12' - group . We have tested their claim placodonts, at least some of which must carotenal8 redshifts to 626±18 nm, and that this parsimony analysis "robustly sup­ have extensive convergent similarities to 2E,4E-octadienal blueshifts to 439±11 nm ports the diapsid affinities of turtles''. turtles. Thus, barring the discovery of new (Fig. 2). The sensitivities and spectral We reanalysed the data from ref. 4 with and intermediate fossil forms, it seems shifts are comparable with their incorpo­ the PAUP5 software. Our unconstrained unlikely that the available morphological ration in C. reinhardtii, suggesting simil­ analysis replicated the results of ref. 4, yield­ data will prove sufficient to resolve the rela­ arities in the binding sites and the ing two most parsimonious trees with tionships of turtles convincingly. interaction of the chromophores with their lengths of 770 steps in which the turtles are The alternative hypotheses considered protein environments. The activity of the sister group of Placodus and Eosauro­ here (Fig. 1) differ in the relationships these analogues and the corresponding pterygia (position 1, Fig. 1). Remarkably, posited among the extant taxa, with either shifts of their action spectra show that however, the shortest trees produced using turtles more closely related to lepidosaurs these zoospores use a rhodopsin to track topological constraints to force the turtles () than to (crocodilians) light for phototaxis. to lie within the Parareptilia required only or archosaurs and lepidosaurs more closely The use of rhodopsins in phototaxis three additional steps. related to each other than to turtles. They by both green algae and fungal zoospores That such a small increase in tree length are thus amenable to testing using molecu­ suggests that vision may have evolved from ( < 0.4%) is needed to include turtles within lar and other neontological data. Molecular the phototaxis of a unicellular ancestor. the Parareptilia is not consistent with robust studies have yet to provide a well supported Because they use the same type of photo­ diapsid affinities of turtles. Furthermore, resolution of amniote relationships9- 11, but, receptor, it is not surprising that some pairwise statistical comparisons of con­ simply by virtue of the potential wealth of phototactic chytridiomycetes can gather strained and unconstrained trees reinforce sequence data, may yet provide our best in the same places as the green algae that this lack of robustness. We compared the fit chance of distinguishing between the alter­ they parasitize11 • Most important, a uni­ of the characters to the alternative trees native phylogenetic placements of turtles. cellular non-photosynthetic model system using the null hypothesis that each character Until such a time, phylogeneticists must be for a rhodopsin photoreceptor is now is equally likely to support either tree. Using circumspect about the precise affinities of available. the Templeton- Felsenstein test6'7, as imple­ these enigmatic reptiles. Jureepan Saranak mented in PHYLIP8, the most parsimonious Mark Wilkinson, Joe Thorley Kenneth W. Foster trees in which turtles are parareptiles are not School of Biological Sciences, Physics Department, significantly less likely than those in which University ofB ristol, Bristol BSS 1 UG, UK Syracuse University, Syracuse, turtles are diapsids (P> 0.05). Application Michael J. Benton New York 13244- 1130, USA of the Wilcoxon signed ranks test6 supported Department of Geology, University of Bristol, this conclusion (P > 0.74, two tailed). Bristol B58 I Rf, UK I. Foster, K. W. et al. Na ture 311, 756-759 (I 984). 2. Wainright, P. 0 ., Hinkle, G., Sogin, M. L. & Stickel, S. K. Science It is interesting that relationships be­ I. Spencer, P. S. thesis, Univ. Bristol ( 1994 ). 260, 34{}-342 (1993 ). tween the other taxa are unaffected by the 2. Lee, M. S. Y Biol. Rev. 70, 459- 547 (I 995). 3. Robertson , J. A. Arch. .Mikrobiol. 85, 259- 266 (1 972). widely divergent positions of the turtles in 3. Laurin, lvL & Reisz, R. R. Zoo/. /. Linn. Soc. 113, 165- 223 4. Foster, K. W. & Smyth, R. D. Microbial. Rev. 44, 572---030 (1980). the unconstrained and constrained trees, (1995). 5. Dorwuod, D. W. & Powell, M. J. Protoplasma 112, 181-188 4. Rieppel, 0 . & de Braga, M. Nature 384, 453--455 ( 1996). (1982). indicating that the turtles are a particularly 5. Swofford, D. L. PAUP J.1. l (Smithsonian Inst., 1993). 6. Kazama, F. Y. & Schorinstein, K. L. Protoplasma 91. 143-156 problematic taxon to classify. Without 6. Templeton, A. R. Evolution 37. 22 1- 244 (1983 ). (1 977). doubt, the highly distinctive and divergent 7. Fclscnstein, J. Syst. Zoo/. 34, 152 - 162 (1985). 7. f oster, K. W., Saranak, J. & Dowbcn, P. A. J. Photod1em. 8. Felscnstein, J. PHYLIP 3.572c(Univ. Washington, 1993). Photobiol. B: Biol. 8, 385- 408 ( 1991). morphology of turtles is one major obstacle 9. Bishop, M. J. & Friday, A. E. in Molecules and Morphology in 8. Saranak, ). & Foster, K. W.]. Exp. Bot. 45, 505- 5 11 (1994). to inferring their phylogenetic relationships Evolution: Conflict or Compromise (ed. Patterson, C.) 123-139 9. Hcgemann, P., Hegernann, U. & Foster, K. W. Photochem. and evolutionary on gms with much (Cambridge Univ. Press, 1987). Photobiol. 48, 123- 128 (1988). certainty. Another obstacle is the limited 10.Ecrnisse, D. J. & Kluge, A. G. Mo/. Riol. Evol. 10, 11 70-1195 10. Nakanishi, K. & Crouch, R. Israel J. Chem. 35, 253- 272 (1995). (1 993). 11. Strasburger, E. ]enai.c,che Z. Nar. 12, 551- 625 ( 1878}. morphological data available for extinct 11. Fushitani, K., Higashiyama, K., Moriyama, E. N., lrnai, K. & 12. Fisher, S. K. et al. Metl1. Neurosci. 15, 3- 36 ( 1993). putative relatives such as and Hosokawa, K. Mol. Rio/. £vol. 13, 1039- 1043 (1 996).

466 NATURE I VO l.387 I 29 MAY 1997