Lua Programming Gems

edited by

Luiz Henrique de Figueiredo Waldemar Celes Roberto Ierusalimschy

Lua.org

Rio de Janeiro 2008
edited by Luiz Henrique de Figueiredo, Waldemar Celes, Roberto Ierusalimschy. ISBN 978-85-903798-4-3. Copyright c 2008 by the editors and individual contributors. All rights reserved. Book cover by Pedro de Mazza Cerqueira. Lua logo design by Alexandre Nako. A Typesetting by the editors using LTEX. Book web site: http://www.lua.org/gems/ Although the editors and the authors have used their best efforts in preparing this book, they assume no responsibility for errors or omissions, or for any damage that may result from the use of the information presented here. All product names mentioned in this book are trademarks of their respective owners.

Contents

Preface. Foreword, by Cameron Laird. Lua and Lightroom, by Mark Hamburg Contributors.

vii ix xi xiii

Programming Techniques
Lua Per-Thread Library Context. Doug Currie Lua Performance Tips. Roberto Ierusalimschy Vardump: The Power of Seeing Whats Behind. Tobias Sulzenbruck and Christoph Beckmann Serialization with Pluto. Ben Sunshine-Hill Abstractions for LuaSQL. Tomas Guisasola Gorham Boostrapping a Forth in 40 Lines of Lua Code. Eduardo Ochs Effecting Large-Scale Change (with little trauma) using Metatables. Sergio Alvares Maffra and Pedro Miller Rabinovitch 71

Design Techniques

MVC Web Development with Kepler. Andre Carregal and Yuri Takhteyev Filters, Sources, Sinks, and Pumps. Diego Nehab Lua as a Protocol Language. Patrick Rapin 109

CONTENTS

Lua Script Packaging. Han Zhao Objects, Lua-style. Reuben Thomas Exceptions in Lua. John Belmonte
III Algorithms and Data Structures
Word Ladders. Gavin Wraith Building Data Structures and Iterators in Lua. Luis Carvalho A Primer of Scientic Computing in Lua. Luis Carvalho Complex Structured Data Input. Julio M. Fernandez-Daz Lua Implementations of Common Data Structures Matthew M. Burke Tic-Tac-Toe and the Minimax Decision Algorithm. Rafael Savelli and Roberto de Beauclair Seixas. 211 239

IV Game Programming

Using Lua in Game and Tool Creation. Konstantin Sokharev and Vadim Groznov A Dynamic and Flexible Event System for Script-Driven Games Robert Oates Lua for Game Programming. Steve Gargolinski Designing an Efcient Lua Driven Game Scripting Engine. Nicolas Peri. 269 281
V Embedding and Extending
28 Enhanced Coroutines in Lua. Patrick Rapin Using Lua in Pascal. Jeremy Darling Porting Lua to a Microcontroller. Ralph Hempel Writing C/C++ Modules for Lua. Ralph Steggink and Wim Couwenberg Interpreted C Modules. Jerome Vuarand. 337

Preface

It gives us great pleasure to publish this collection of Lua gems. Not only does it record some of the existing wisdom and practice on how to program well in Lua, but it also reects the maturity of the Lua community. It is gratifying to see that Lua has motivated other people to learn it well and to share their knowledge with other users. In well-written articles that go much beyond the brief informal exchange of tips in the mailing list or the wiki, the authors share their mastery of all aspects of Lua programming, elementary and advanced. Producing this book has required several steps. In response to a call for contributions, we received over 70 abstracts, selected 43, and received full versions for 28 of these. The authors received our comments and suggestions to prepare the nal version of their articles. The whole process took two years, much longer than we had imagined. The selection of abstracts proved to be surprisingly difcult. Many potentially good submissions could not be accepted due to space limitations. Despite the long time it took and the amount of work it required (or because of it!), we are very happy to have this collection of articles on Lua contributed by members of our community. We trust the book was worth waiting for. We thank all the authors for their hard work on the articles and everyone that submitted abstracts in the rst phase. We also thank the whole Lua community for its friendliness and expertise. The active participation of our users has been to us a constant source of motivation for improving Lua. Finally, we give our warm thanks to Cameron Laird and Mark Hamburg for writing forewords to this book. Additional material and errata will appear in the book web site: http://www.lua.org/gems/ The Lua team Rio de Janeiro, November 2008

Foreword

by Cameron Laird
When I need a programming language thats as easy as possible to embed, I choose Lua. Lua isnt just supple, free, portable, and compact, though; its also powerful and to get the most out of it, Im glad I have Lua Programming Gems. I need to explain that I mean something specic by that. Most of my reading is on the Net: I look up references, I read tutorials for unfamiliar material, I moderate a half-dozen Wikis, and I chat about specic techniques with colleagues on-line. My consumption of books has nose-dived. Lua Programming Gems is a book worth reading, though: its individual chapters get across ideas that simply arent explained anywhere else. Lua Programming Gems emphasizes practicality in a way I like. While the six authors in Part III certainly employ classroom concepts correctly in their examinations of Algorithms and Data Structures, they do it all with working Lua code. The same pattern is apparent throughout Lua Programming Gems: its lled with ideas likely to help me in my next programming project. If youre new to Lua, you might be anxious about what youll nd. You can see that Lua offers denite advantages, but how hard is it to pick up whats undeniably a minority language? Lua Programming Gems will ease your concerns: the authors write clearly, modestly, and even deftly. The very rst chapter, for example, tackles difcult material, including dynamically-allocated per-thread storage. The tone is consistent throughout the book. Rather than show off their expertise or indulge in private jokes, habits common for authors from other domains, the Lua Programming Gems authors focus on the specic details and examples that best teach their chosen topics. They make it inviting to dig deeper in Lua than you might do on your own. Among the highlights of Lua Programming Gems for me: Part IV gives insight into Game Programming, an area where Ill probably never work, although many of the techniques apply more broadly; Part V on Embedding and Extending is crucial for much of the programming I like; and Chapter 13, Exceptions in Lua, is a particular interest of mine.

FOREWORD

Do you want to program well in Lua? The Lua team set that as a goal when it rst announced its plans for Lua Programming Gems. The nal result fullls that goal; youll like it.

Lua and Lightroom

Mark Hamburg
Founder, Adobe Photoshop Lightroom
When we started work on the project that would become Adobe Photoshop Lightroom, we knew we wanted to make scriptability an important part of our story, so early on we reviewed the usual suspects. What drew us to Lua was its combination of simplicity, power, ease of embedding, and relatively highperformance. Having a straightforward license helped too when it came time to talk to Adobes lawyers. Personally, as an old Scheme fan, I was drawn to its rst-class closure support. I also found the coroutine system intriguing. The relative minimalism also resonated with a back-to-basics attitude that had us weaning ourselves away from intensive C++ usage and back toward C. Still, it was hard to position Lua as anything other than an obscure choice. We could cite heavy use in the games community and we had set out with a mission of learning something from game developers, but if asked what materials one could turn to learn Lua or where we would nd experienced Lua programmers, the answers were limited. For the former, we had the well-written reference manual, some good material on the Lua users wiki, and an intelligent forum on the Lua mailing list. This was good material, but there wasnt a lot of it. For the latter question, our answer was essentially Any programmer worth hiring ought to be able to learn Lua quickly. This was a situation we were prepared to deal with and the arrival of Programming in Lua certainly helped, but it was easy to understand why it might be off putting to someone looking in from the outside. Why this matters is that along with Luas simplicity come some issues that make people with backgrounds in other languages stumble. The beauty of a small core is that there is a real opportunity for mastery. This is one of Cs great strengths as well. That small core, however, comes at a price. For example, Lua has no syntax for exception handling. C doesnt either but having one seems almost required in modern languages. Lua has a syntax for object-oriented

LUA AND LIGHTROOM

message sends, but the actual implementation of an object system or class system is a roll-your-own affair in Lua. As a result, one sees such issues raised repeatedly on the Lua users list as people new to the language start using it and ? then ask But what about Programming in Lua provides answers to some of these questions but those answers are necessarily terse. Lua Programming Gems dives deeper on these issues and many more. It shows ways to deal with threading an issue we went through a few iterations on in Lightroom and gives extended examples of how to hook Lua into your application. You may not always like the answers. For example, the object system presented here is quite minimalist. In the spirit of Lua, however, you remain free to roll your own using or not using the ideas presented here. The value comes in seeing the well worked examples together with a discussion of motivations and comparisons to other approaches. If this book had been around during Lightrooms development, we probably would have happily adopted some of the techniques it presents while simply taking inspiration from others. As it was, we largely had to nd our own way and while that was rewarding in itself, the Lua community and particularly new Lua users can be happy to now have a eld guide that maps out some of the trails. The broader lesson from Lightroom that I would like to leave Lua users with is that you should let it pervade your work. Lua is sometimes described as being a language for gluing pieces together, but as we discovered that glue can extend quite deep. We started out looking for a scripting language for a native code application. Then we started thinking it would be nice to allow Lua to exist as a peer to native code. In the end, we ended up with a system where native code provides the foundation, but it is effectively a second-class citizen in the application as a whole. Large portions of Lightroom ended up getting written in Lua including the object-relational mapping layer for the database and the layout system for views. Lua denes the structure of the application and its extensibility mechanisms. As a result, we had an application that was smaller by far than some of its competitors, easy to change, largely cross-platform in its implementation, and suffered essentially no compile-link cycle. The reason things work out this way is that Lua is both very expressive compared to most native languages and sufciently efcient that you can let it do a lot more of the work than one might be tempted to in other scripting languages. At the same time, the boundary between native code and Lua is sufciently clean and efcient that when we needed to do things in native code, it wasnt a huge burden to expose that functionality to Lua nor to access functionality written in Lua. So, my advice to Lua users and potential users is to think seriously about how widely you can let Lua spread through your work, be grateful for books like this one and Programming in Lua and be even more grateful for the work that the Lua team has done and their generosity in sharing it with the world.

Contributors

Following the Brazilian tradition, the contributors are listed in alphabetical order of rst name.
Andr Carregal was introduced to Lua in 1994 during his MSc in Computer e
Science, which was supervised by Roberto Ierusalimschy. He has been working with web development using Lua since 1996. He currently coordinates the Kepler project and the LuaForge site while working as a consultant for Luarelated projects.
Ben Sunshine-Hill is a PhD student at the University of Pennsylvania,
studying computer graphics. He did his undergraduate studies at the University of Southern California and received an MSc from the University of California, Los Angeles. He has been a game developer at several mobile and mainstream game development studios, and has previously published work on real-time rendering methods.
Diego Nehab was introduced to Lua in 1996, while working for Tecgraf in
PUC-Rio. Over the years, he has been involved in a variety of Lua-related projects, including the IupLua, CDLua, IMLua, and LuaSQL libraries. He is best known as the author of the LuaThreads and LuaSocket libraries. Diego received a BEng in Computer Engineering and an MSc in Programming Languages from PUC-Rio, under the supervision of Roberto Ierusalimschy. He later received an MSc and a PhD in Computer Graphics from Princeton University. His research now focuses on high-quality shape acquisition and on real-time rendering techniques.
Doug Currie develops award-winning medical devices with Sunrise Labs,
Inc. in Auburn, New Hampshire, USA. Over a thirty-year career, Doug has led electronics, mechanical, and software teams developing high-tech products with particular emphasis on reliability and adaptability. Some of these products, based on a massively parallel computing architecture Doug invented, are used in national transportation and world-class manufacturing operations. With a

CONTRIBUTORS

special interest in little languages, Doug has also contributed technically to open source projects such as Moscow ML, Hibernate, Gambit Scheme, and SICStus Prolog. Doug holds an S.B. degree in Electrical Engineering and Computer Science from the Massachusetts Institute of Technology.
Eduardo Ochs is a mathematician, or sort of; his interest on simplication

of proofs led him to Non-Standard Analysis, and from there he drifted to Logic, Type Theory and Categorical Semantics. In parallel with his normal academic life he has been a contributor to the GNU Project since 1999, and his main focus areas in Free Software are little languages and programmable textual interfaces. He keeps a big, messy homepage at http://angg.twu.net/; all the html pages in it are generated with BlogMe, another extensible little language built on top of Lua.
Gavin Wraith is Emeritus Reader in Mathematics, Sussex University, UK.
Joined Sussex University in 1963, retired in 1999. Founding chairman of the Sussex University Computer Science department in 1985.
J r me Vuarand is a young software engineer specialized in AI and working e o
in the video games industry. He discovered Lua while looking for an embeddable scripting language, just when Lua 5.1 was released, and he fell in love both for the language itself and its new package system. His initial motivation was to move away from legacy in-house script engines, but hes now using it as a general programming language in all his personal projects, from mobile robotics to modern game engines entirely written in Lua.
John Belmonte is a software engineer currently residing in New York City.
He happened upon Lua as a video game developer in 2000 and was among the rst to embed it into a home console title. Since then he has been active in the Lua community through chartering lua-users.org, participating in workshops, and contributing to the languages evolution.
Julio M. Fernandez-Daz has a PhD degree in Mining Engineering (1989).
He is Professor of Applied Physics at the University of Oviedo in Spain and researches mainly in the eld of atmospheric aerosols. His interests lie in developing physical and mathematical simulations on computers. His rst computer was an HP25 calculator in 1977. As a programmer, he uses Fortran, C, Lua, Tcl/Tk and Postscript, usually as part of his research.
Konstantin Sokharev professionally develops video games since 2001, successfully completed two RPG/RTS projects for PC, one for PocketPC/Palm. He currently holds a post of technical director at IceHill llc. developing Action-RTS title Empire Above All and several unannounced projects.
Luis Eduardo Ximenes Carvalho has a BSc (1997) in Civil Engineering
from the Federal University of Ceara (UFC), an MSc (2000) in Transportation
Engineering from the Federal University of Rio de Janeiro (UFRJ), and an MSc (2002) in Computer Science from UFC, all in Brazil. He is currently a PhD candidate in the Division of Applied Math at Brown University, where his research comprises applications of Bayesian statistics to computational biology. He also has interest in logistics and optimization, scientic computing, graph theory, and programming languages, especially Lua.

Matthew Burke is an Assistant Professor of Computer Science at The George
Washington University in Washington, D.C. Lua has replaced Forth as his favorite language in which to program while riding the subway, and he does so using whatever device is serving as his PDA du jour. He is also developing a curriculum for introductory Computer Science which uses Lua. When not programming, he likes to travel with his wife and son. He was the organizer of the Lua Workshop 2008.
Nicolas Peri is co-founder and technical director of the French company StoneTrip, creator of the 3D game development platform ShiVa. He is in charge, among other things, of the ShiVa scripting engine, which is based on Lua. Before that, he worked as engine developer for other gaming companies, including Kalisto Entertainment and UbiSoft Tiwak. Patrick Rapin studied at the Swiss Federal Institute of Technology at Lausanne (EPFL). He is now a software engineer working for Olivetti Engineering at Yverdon-les-Bains, developing printer rmware, image processing algorithms, and printer test tools.
Pedro Miller Rabinovitch , a PUC-Rio graduate, has worked with Lua at
Tecgraf and Cipher Technology, and is currently a game developer at Jagex.
Rafael Moreira Savelli graduated in Computer Engineering at PUC-Rio.
He worked for Tecgraf in PUC-Rio for over four years. He is now studying for an MSc at UFF and working in the Visgraf laboratory at IMPA.
Ralph Hempel is a Professional Engineer in Ontario, Canada and specializes
in designing embedded systems. After learning to program on an HP41C, he never lost his fascination with small languages and hacking consumer products. He wrote pbForth for the LEGO MINDSTORMS RCX and then ported Lua to the NXT. When hes not wrangling embedded systems, Ralph enjoys mountain biking in the summer, snowboarding in the winter, and ice hockey all year long.
Ralph Steggink joined Oc in 2001. With a degree in both chemistry and come
puter science, he now develops controller software for printers. Together with Wim Couwenberg he prototyped revolutionary concepts using Lua. These currently nd their way into several Oc products. He is an enthusiastic volleyball e player and trainer.
Reuben Thomas is a freelance singer and computer scientist living in London. He took a BA in Mathematics with Computer Science from Cambridge University, as well as a doctorate in virtual machines. These days his computing interests center on contributions to a multitude of open source projects, with particular emphasis on improving the quality of mature software, and on automatic document processing. He is mostly employed as a classical baritone.

Robert Oates is a professional game programmer specializing in gameplay
systems, articial intelligence, and machine learning.
Roberto de Beauclair Seixas works with Research and Development at
the Institute of Pure and Applied Mathematics (IMPA) in Rio de Janeiro, as member of the Vision and Computer Graphics Laboratory (Visgraf). He got his PhD in Computer Science at PUC-Rio, where he works with the Computer Graphics Technology Group (Tecgraf). From 1982 to 1998, he worked in the Computer Science Department at the National Laboratory for Scientic Computation (LNCC). His research interests include Scientic Visualization, Volume Rendering, Computer Graphics, High Performance Computing, Geometric Modeling, Military Warfare Simulations, GIS, and Medical Images.
Roberto Ierusalimschy is an Associate Professor at the Catholic University
in Rio de Janeiro. He is the leading architect of Lua and the author of the book Programming in Lua.
S rgio Alvares Maffra is a MSc and Computer Engineer from PUC-Rio. Hes e
been working with Lua at Tecgraf as a software developer for over a decade now.
Steve Gargolinski spent his early programming days hacking together small
games built with code snippets from a QuickBasic programming manual. He has since evolved into a professional game developer, working as a member of the technical teams that produced the Zoo Tycoon 2 series, Star Trek: Legacy, and the upcoming Empire Earth III. Steve is currently working for Blue Fang Games as an AI Programmer. His interests include baseball, abstract strategy, practical AI, and walking in the woods.
Tobias Sulzenbruck and Christoph Beckmann are bachelor students of
media systems at the Bauhaus-University of Weimar. Tobias elds of interests range from web development up to graphics programming. He has implemented a multi-agent system for simulating construction processes in Lua. Christoph is also interested in web development and is active in the research eld of computer-supported cooperative work.
Tomas Guisasola works with Lua since 1995 when he developed with Roberto
Ierusalimschy (his MSc advisor) the rst implementation of the hooks mechanism and the debug facilities. Since then he worked mainly with CGILua as the platform for some administrative systems at PUC-Rio and also contributed
with the Kepler team in the development of LuaLDAP, LuaXMLRPC, LuaSOAP, LuaDoc, and LuaSQL.
Vadim Groznov began programming at the age of fourteen, was involved
in database programming for a long time, and took part in the creation of a custom scripting language. He professionally develops video games since 2002. His extensive experience of system programming allowed for the design and realisation of complex architectural solutions for game tools at IceHill llc.

Wim Couwenberg holds a PhD in mathematics and is employed at the R&D
department of the European printing and document company Oc , based in The e Netherlands, where he organised the international Lua Workshop 2006. He has been using Lua in projects ranging from simple data processing scripts to entire networked applications.
Yuri Takhteyev is a doctoral student at the UC Berkeley School of Information studying the role of space in software development communities.
Han Zhao is a shareware programmer in Beijing, P.R. China. Before that he worked for a mobile-phone design house. Now he uses Lua and C++ for everyday programming: an isometric role-playing game engine, an action game, and a shareware product. He also maintains a bit-operation lib LuaBit on LuaForge.

NATURE|January 2009|doi:10.1038/nature07740

15 EVOLUTIONARY GEMS

A resource from Nature for those wishing to spread awareness of evidence for evolution by natural selection.
Henry Gee, Rory Howlett and Philip Campbell*
Most biologists take for granted the idea that all life evolved by natural selection over billions of years. They get on with researching and teaching in disciplines that rest squarely on that foundation, secure in the knowledge that natural selection is a fact, in the same way that the Earth orbits the Sun is a fact. Given that the concepts and realities of Darwinian evolution are still challenged, albeit rarely by biologists, a succinct briefing on why evolution by natural selection is an empirically validated principle is useful for people to have to hand. We offer here 15 examples published by Nature over the past decade or so to illustrate the breadth, depth and power of evolutionary thinking. We are happy to offer this resource freely and encourage its free dissemination.
Gems from the fossil record
15 Land-living ancestors of whales From water to land The origin of feathers The evolutionary history of teeth The origin of the vertebrate skeleton Natural selection in speciation Natural selection in lizards A case of co-evolution Differential dispersal in wild birds Selective survival in wild guppies Evolutionary history matters Darwins Galapagos finches Microevolution meets macroevolution Toxin resistance in snakes and clams Variation versus stability

Gems from habitats

Gems from molecular processes
*Henry Gee is a senior editor for Nature; Rory Howlett is a consultant editor for Nature; Philip Campbell is editor-in-chief of Nature.
www.nature.com/evolutiongems
2009 Macmillan Publishers Limited. All rights reserved
Land-living ancestors of whales
Fossils offer crucial clues for evolution, because they reveal the often remarkable forms of creatures long vanished from Earth. Some of them even document evolution in action, recording creatures moving from one environment to another. Whales, for example, are beautifully adapted to life in water, and have been for millions of years. But, like us, they are mammals. They breathe air, and give birth to and suckle live young. Yet there is good evidence that mammals originally evolved on land. If that is so, then the ancestors of whales must have taken to the water at some point. As it happens, we have numerous fossils from the first ten million years or so of whale evolution. These include several fossils of aquatic creatures such as Ambulocetus and Pakicetus, which have characteristics now seen only in whales especially in their ear anatomy but also have limbs like those of the land-living mammals from which they are clearly derived. Technically, these hybrid creatures were already whales. What was missing was the start of the story: the land-living creatures from which whales eventually evolved. Work published in 2007 might have pinpointed that group. Called raoellids, these now-extinct creatures would have looked like very small dogs, but were more closely related to even-toed ungulates the group that includes modern-day cows, sheep, deer, pigs and hippos. Molecular evidence had also suggested that whales and even-toed ungulates share a deep evolutionary connection. The detailed study, by Hans Thewissen at Northeastern Ohio Universities Colleges of Medicine and Pharmacy in Rootstown and his colleagues, shows that one raoellid, Indohyus, is similar to whales, but unlike other eventoed ungulates in the structure of its ears and teeth, the thickness of its bones and the chemical composition of its teeth. These indicators suggest that this raccoon-sized creature spent much of its time in water. Typical raoellids, however, had a diet nothing like those of whales, suggesting that the spur to take to the water may have been dietary change. This study demonstrates the existence of potential transition forms in the fossil record. Many other examples could have been highlighted, and there is every reason to think that many others await discovery, especially in groups that are well represented in the fossil record.
Reference Thewissen, J. G. M., Cooper, L. N., Clementz, M. T., Bajpai, S. & Tiwari, B. N. Nature 450, 11901194 (2007). Additional resources Thewissen, J. G. M., Williams, E. M., Roe, L. J. & Hussain, S. T. Nature 413, 277281 (2001). de Muizon, C. Nature 413, 259260 (2001). Novacek, M. J. Nature 368, 807 (1994). Zimmer, C. At The Waters Edge (Touchstone, 1999). Video of Thewissens research: www.nature.com/nature/videoarchive/ancientwhale Author website Hans Thewissen: www.neoucom.edu/DEPTS/ANAT/Thewissen

2 From water to land

The animals we are most familiar with are tetrapods they are vertebrates (they have backbones) and they live on land. That includes humans, almost all domestic animals and most of the wild ones that any child would recognize: mammals, birds, amphibians and reptiles. The vast majority of vertebrates, however, are not tetrapods, but fish. There are more kinds of fish, in fact, than all the species of tetrapods combined. Indeed, through the lens of evolution, tetra ods are just one branch of the fish family tree, the members of which just p happen to be adapted for life out of water. The first transition from water to land took place more than 360 million years ago. It was one of the most demanding such moves ever made in the history of life. How did fins become legs? And how did the transitional creatures cope with the formidable demands of land life, from a desiccating environment to the crushing burden of gravity? It used to be thought that the first landubbers were stranded fish that evolved to spend more and more time l ashore, returning to water to reproduce. Over the past 20 years, palaeontologists have uncovered fossils that have turned this idea upside down. The earliest tetrapods, such as Acanthostega from eastern Greenland around 365 million years ago, had fully formed legs, with toes, but retained internal gills that would soon have dried out in any long stint in air. Fish evolved legs long before they came on land. The earliest tetrapods did most of their evolving in the more forgiving aquatic environment. Coming ashore seems to have been the very last stage. Researchers suspect that the ancestors of tetrapods were creatures called elpistostegids. These very large, carnivorous, shallow-water fish would have looked and behaved much like alligators, or giant salamanders. They looked like tetrapods in many respects, except that they still had fins. Until recently, elpistostegids were known only from small fragments of fossils that were poorly preserved, so it has been hard to get a rounded picture of what they were like. In the past couple of years, several discoveries from Ellesmere Island in the Nunavut region of northern Canada have changed all that. In 2006, Edward Daeschler and his colleagues described spectacularly wellpreserved fossils of an elpistostegid known as Tiktaalik that allow us to build up a good picture of an aquatic predator with distinct similarities to tetrapods from its flexible neck, to its very limb-like fin structure. The discovery and painstaking analysis of Tiktaalik illuminates the stage before tetrapods evolved, and shows how the fossil record throws up surprises, albeit ones that are entirely compatible with evolutionary thinking.

References Daeschler, E. B., Shubin, N. H. & Jenkins, F A. Nature 440, 757763 (2006). Shubin, N. H., Daeschler, E. B., & Jenkins, F A. Nature 440, 764771 (2006). Additional resources Ahlberg, P. E. & Clack, J. A. Nature 440, 747749 (2006). Clack, J. Gaining Ground (Indiana Univ. Press, 2002) Shubin, N. Your Inner Fish (Allen Lane, 2008) Gee, H. Deep Time (Fourth Estate, 2000) Tiktaalik homepage: http://tiktaalik.uchicago.edu Author websites Edward Daeschler: http://www.ansp.org/research/biodiv/vert_paleo/staff.php Neil Shubin: http://pondside.uchicago.edu/oba/faculty/shubin_n.html

The origin of feathers

One of the objections to Charles Darwins theory of evolution was the lack of transitional forms in the fossil record forms that illustrated evolution in action, from one major group of animals to another. However, hardly a year after the publication of On the Origin of Species, an isolated feather was discovered in Late Jurassic (about 150 million years old) lithographic limestones of Solnhofen in Bavaria, followed in 1861 by the first fossil of Archaeopteryx, a creature with many primitive, reptilian features such as teeth and a long, bony tail but with wings and flight feathers, just like a bird. Although Archaeopteryx is commonly seen as the earliest known bird, many suspected that it was better seen as a dinosaur, albeit one with feathers. Thomas Henry Huxley, Darwins colleague and friend, discussed the possible evolutionary link between dinosaurs and birds, and palaeontologists speculated, if wildly, that dinosaurs with feathers might one day be found. In the 1980s, deposits from the early Cretaceous period (about 125 million years ago) in the Liaoning Province in northern China vindicated these speculations in the most dramatic fashion, with discoveries of primitive birds in abundance alongside dinosaurs with feathers, and feather-like plumage. Starting with the discovery of the small theropod Sinosauropteryx by Pei-ji Chen from Chinas Nanjing Institute of Geology and Palaeontology and his colleagues, a variety of feather-clad forms have been found. Many of these feathered dinosaurs could not possibly have flown, showing that feathers first evolved for reasons other than flight, possibly for sexual display or thermal insulation, for instance. In 2008, Fucheng Zhang and his colleagues from the Chinese Academy of Sciences in Beijing announced the bizarre creature Epidexipteryx, a small dinosaur clad in downy plumage, and sporting four long plumes from its tail. Palaeontologists are now beginning to think that their speculations werent nearly wild enough, and that feathers were indeed quite common in dinosaurs. The discovery of feathered dinosaurs not only vindicated the idea of transitional forms, but also showed that evolution has a way of coming up with a dazzling variety of solutions when we had no idea that there were even problems. Flight could have been no more than an additional opportunity that presented itself to creatures already clothed in feathers.

References Chen, P.-J., Dong, Z.-M. & Zhen, S.-N. Nature 391, 147152 (1998). Zhang, F., Zhou, Z., Xu, X., Wang, X. & Sullivan, C. Nature 455, 11051008 (2008). Additional resources Gee, H. (ed.) Rise of the Dragon (Univ. Chicago Press, 2001). Chiappe, L. Glorified dinosaurs (Wiley-Liss, 2007). Gee, H. & Rey, L. V. A Field Guide to Dinosaurs (Barrons Educational, 2003).
4 The evolutionary history of teeth
One motivation in the study of development is the discovery of mechanisms that guide evolutionary change. Kathryn Kavanagh at the University of Helsinki and her colleagues investigated just this by looking at the mechanisms behind the relative size and number of molar teeth in mice. The research, published in 2007, uncovered the pattern of gene expression that governs the development of teeth molars emerge from the front backwards, with each tooth smaller than the last. The beauty of the study lies in its application. Their model predicts the dentition patterns found in mouselike rodent species with various diets, providing an example of ecologically driven evolution along a developmentally favoured trajectory. In general, the work shows how the pattern of gene expression can be modified during evolution to produce adaptive changes in natural systems.
Reference Kavanagh, K. D., Evans, A. R. & Jernvall, J. Nature 449, 427432 (2007). Additional resources Polly, P. D. Nature 449, 413415 (2007). Evans, A. R., Wilson, G. P., Fortelius, M. & Jernvall, J. Nature 445, 7881 (2006). Kangas, A. T., Evans, A. R., Thesleff, I. & Jernvall, J. Nature 432, 211214 (2004). Jernvall, J. & Fortelius, M. Nature 417, 538540 (2002). Theodor, J. M. Nature 417, 498499 (2002). Author website Jukka Jernvall: http://www.biocenter.helsinki.fi/bi/evodevo
The origin of the vertebrate skeleton
We owe much of what makes us human to remarkable tissue, found only in embryos, called the neural crest. Neural-crest cells emerge in the developing spinal cord and migrate all over the body, effecting a remarkable series of transformations. Without the neural crest, we would not have most of the bones in our face and neck, or many of the features of our skin and sensory organs. The neural crest seems to be unique to vertebrates, and helps to explain why vertebrates have distinctive heads and faces. Untangling the evolutionary history of the neural crest is especially hard in fossil forms, as embryonic data are obviously absent. One key mystery, for example, is how much of the vertebrate skull is contributed by neuralcrest cells and how much comes from deeper layers of tissue. New techniques have allowed researchers to label and follow individual cells as embryos develop. They have revealed the boundaries of the bone derived from the neural crest, down to the single-cell level, in the neck and shoulders. Tissue derived from the neural crest anchors the head onto the front lining of the shoulder girdle, whereas the skeleton forming the back of the neck and shoulder grows from a deeper layer of tissue called the mesoderm. Such detailed mapping, in living animals, casts light on the evolution of structures in the heads and necks of animals long extinct, even without fossilized soft tissue such as skin and muscle. Skeletal similarities that result from a shared evolutionary history can be identified from muscle attachments. This allows the tracing of, for example, the location of the major shoulder bone of extinct land vertebrate ancestors, the cleithrum. This bone seems to survive as part of the shoulder blade (scapula) in living mammals. This kind of evolutionary scan may have immediate clinical relevance. The parts of the skeleton identified by Toshiyuki Matsuoka from the Wolfson Institute for Biomedical Research in London and his colleagues as being derived from the neural crest are specifically affected in several developmental disorders in humans, providing insights into their origins. Matsuokas study shows how a detailed analysis of the morphology of living animals, informed by evolutionary thinking, helps researchers to interpret fossilized and now-extinct forms.

Reference Matsuoka, T. et al. Nature 436, 347355 (2005). Author website Georgy Koentges: http://www2.warwick.ac.uk/fac/sci/systemsbiology
Natural selection in speciation
Evolutionary theory predicts that divergent natural selection will often have a key role in speciation. Working with sticklebacks (Gasterosteus aculeatus), Jeffrey McKinnon at the University of Wisconsin in Whitewater and his colleagues reported in 2004 that reproductive isolation can evolve as a by-product of selection on body size. This work provides a link between the build-up of reproductive isolation and the divergence of an ecologically important trait. The study was done on an extraordinary geographical scale, involving mating trials between fish taken in Alaska, British Columbia, Iceland, the United Kingdom, Norway and Japan. It was underpinned by molecular genetic analyses that provided firm evidence that fish that have adapted to living in streams had evolved repeatedly from marine ancestors, or from fish that live in the ocean but return to fresh water to spawn. Such migratory populations in the study had larger bodies on average than did those living in streams. Individuals tended to mate with fish of a similar size, which accounts well for the reproductive isolation between different stream ecotypes and their close, seafaring neighbours. Taking into account the evolutionary relationships, a comparison of the various types of stickleback, whether stream or marine, strongly supports the view that adaptation to different environments brings about reproductive isolation. The researchers experiments also confirmed the connection between size divergence and the build-up of reproductive isolation although traits other than size also contribute to reproductive isolation to some extent.
Reference McKinnon, J. S. et al. Nature 429, 294298 (2004). Additional resources Gillespie, R. G. & Emerson, B. C. Nature 446, 386387 (2007). Kocher, T. D. Nature 435, 2930 (2005). Emerson, B. C. & Kolm, N. Nature 434, 10151017 (2005). Author websites Jeffrey McKinnon: http://facstaff.uww.edu/mckinnoj/mckinnon.html David Kingsley: http://kingsley.stanford.edu Dolph Schluter: http://www.zoology.ubc.ca/~schluter
Natural selection in lizards
A popular evolutionary hypothesis is that behavioural shifts in new environments negate the effects of natural selection. But work by Harvard Universitys Jonathan Losos and his colleagues in 2003 lends little support to this theory. The researchers introduced the large ground-dwelling predatory lizard Leiocephalus carinatus to six small islands in the Bahamas, with six other islands serving as controls. They found that the lizards prey, a smaller lizard called Anolis sagrei, spent more time higher up in the vegetation on islands occupied by the larger predator than they did on the islands where L. carinatus was absent. But mortality in A. sagrei was still much higher on the experimental islands than on control islands. The presence of the larger predator selected for longer-legged male A. sagrei lizards, which can run faster, and also favoured larger females, which are both faster and harder to subdue and ingest. The researchers did not detect any selection on size in males; they suggested that the larger males may have been more vulnerable because of their conspicuous territorial behaviour. The study shows how the introduction of a predator can cause individuals of a prey species to change their behaviour so as to reduce the risk of predation, but also cause an evolutionary response at the level of the population that differs between the sexes according to their ecology.

Reference Losos, J. B., Schoener, T. W. & Spiller, D. A. Nature 432, 505508 (2004). Additional resources Butler, M. A., Sawyer, S. A. & Losos, J. B. Nature 447, 202205 (2007). Kolbe, J. J. et al. Nature 431, 177181 (2004). Calsbeek, R. & Smith, T. B. Nature 426, 552555 (2003). Losos, J. B. et al. Nature 424, 542545 (2003). Author website Jonathan Losos: http://www.oeb.harvard.edu/faculty/losos/jblosos

A case of co-evolution

Species evolve together, and in competition. Predators evolve ever deadlier weapons and skills to catch prey, which, as a result of Darwins canonical struggle for existence, become better at escaping them, and so the arms race continues. In 1973, evolutionary biologist Leigh Van Valen likened this to the Red Queens comment to Alice in Lewis Carrolls Through the Looking Glass, it takes all the running you can do, to keep in the same place. If you want to get somewhere else, you must run at least twice as fast as that! The Red Queen hypothesis of co-evolution was born. A problem with studying Red-Queen dynamics is that they can be seen only in the eternal present. Discovering their history is problematic, because evolution has generally obliterated all earlier stages. Happily, Ellen Decaestecker from the Catholic University of Leuven in Belgium and her colleagues discovered a remarkable exception, in the co-evolutionary arms race between water fleas (Daphnia) and the microscopic parasites that infest them; their research was published in 2007. As the water fleas become better at evading parasitism, the parasites become better at infecting them. Both prey and predator in this system can persist in dormant stages for many years in the mud at the bottom of the lake they share. The sediments of the lake can be dated to the year they were formed, and the buried predators and prey can be revived. Thus, their interactions can be tested, against one another, and against predators or prey from their relative pasts and futures. Confirming theoretical expectations, the parasite adapted to its host over a period of only a few years. Its infectivity at any given time changed little, but its virulence and fitness rose steadily matched at each stage by the ability of the water fleas to resist them. This study provides an elegant example in which a high-resolution historical record of the co-evolutionary process has provided an affirmation of evolutionary theory, showing that the interaction of parasites and their hosts is not set in time but is instead the result of a dynamic arms race of adaptation and counter-adaptation, driven by natural selection, from generation to generation.

Reference Decaestecker, E. et al. Nature 450, 870873 (2007). Additional resources The Red Queen Hypothesis: http://en.wikipedia.org/wiki/Red_Queen Van Valen, L. Evol. Theory 1, 130 (1973). Author website Ellen Decaestecker: http://bio.kuleuven.be/de/dea/people_detail.php?pass_id=u0003403
Differential dispersal in wild birds
Gene flow caused, for instance, by migration, can disrupt adaptation to local conditions and oppose evolutionary differentiation within and between populations. Indeed, classical population genetics theory suggests that the more that local populations migrate and interbreed, the more genetically similar they will be. This concept seems to accord with common sense, and it assumes that gene flow is a random process, like diffusion. But non-random dispersal can actually favour local adaptation and evolutionary differentiation, as Ben Sheldon of the Edward Grey Institute of Field Ornithology in Oxford, UK, and his colleagues reported in 2005. Their work was part of a multi-decade study of the great tits (Parus major) that inhabit a wood in Oxfordshire, UK. The researchers found that the amount and type of genetic variation in nestling weight in this songbird differs from one part of the wood to another. This pattern of variation leads to varying responses to selection in different parts of the wood, leading to local adaptation. The effect is reinforced by non-random dispersal; individual birds select and breed in different habitats in a way that increases their fitness. The authors conclude that when gene flow is not homogeneous, evolutionary differentiation can be rapid and can occur over surprisingly small spatial scales. In another study of great tits on the island of Vlieland in the Netherlands, published in the same issue of Nature, Erik Postma and Arie van Noordwijk from the Netherlands Institute of Ecology in Heteren found that gene flow, mediated by non-random dispersal, maintains a large genetic difference in clutch size at a small spatial scale, again illustrating, as these scientists put it, the large effect of immigration on the evolution of local adaptations and on genetic population structure.

References Garant, D., Kruuk, L. E. B., Wilkin, T. A., McCleery, R. H. & Sheldon, B. C. Nature 433, 6065 (2005). Postma, E. & van Noordwijk, A. J. Nature 433, 65-68 (2005). Additional resource Coltman, D. W. Nature 433, 2324 (2005). Author websites Ben Sheldon: http://www.zoo.ox.ac.uk/egi/people/faculty/ben_sheldon.htm Erik Postma: http://www.nioo.knaw.nl/ppages/epostma Arie van Noordwijk: http://www.nioo.knaw.nl/PPAGES/avannoordwijk David Coltman: http://www.biology.ualberta.ca/faculty/david_coltman
Selective survival in wild guppies
Natural selection favours traits that increase fitness. Over time, such selection might be expected to exhaust genetic variation by driving advantageous genetic variants to fixation at the expense of less advantageous or deleterious variants. In fact, natural populations often show large amounts of genetic variation. So how is it maintained? An example is the genetic polymorphism seen in the colour patterns of male guppies (Poecilia reticulata). As reported in 2006, Kimberly Hughes from the University of Illinois at Urbana-Champaign and her colleagues manipulated the frequencies of males with different colour patterns in three wild guppy populations in Trinidad. They showed that rare variants have much higher survival rates than more common ones. In essence, variants are favoured when rare, and selected against when common. Such frequency-dependent survival, in which selection favours rare types, has been implicated in the maintenance of molecular, morphological and health-related polymorphisms in humans and other mammals.
Reference Olendorf, R. et al. Nature 441, 633636 (2006). Additional resource Foerster, K. et al. Nature 447, 11071110 (2007). Author websites Kimberly Hughes: http://www.bio.fsu.edu/faculty-hughes.php Anne Houde: http://www.lakeforest.edu/academics/faculty/houde David Reznick: http://www.biology.ucr.edu/people/faculty/Reznick.html
Evolutionary history matters
Evolution is often thought to be about finding optimal solutions to the problems that life throws up. But natural selection can only work with the materials at hand materials that are themselves the results of many millions of years of evolutionary history. It never starts with a blank slate. If that were the case, then tetrapods faced with the task of moving on land would not have had their fins transform into legs; they might perhaps have evolved wheels. A real-life case of the ingenuity of adaptation concerns a moray eel (Muraena retifera), a long, snake-like reef predator. Historically, bony fish use suction to catch their prey. A fish approaching food opens its mouth wide to create a large cavity into which prey and water flood. As the excess water leaves through the gills, the fish sucks the prey down into its throat and pharyngeal jaws, a second set of jaws and teeth derived from the skeleton that supports the gills. But morays have a problem because of their elongated, narrow shape. Even with their jaws agape, their mouth cavity is too small to generate enough suction to carry prey to their pharyngeal jaws. The solution to this conundrum was documented in 2007. Through careful observation and X-ray cinematography, Rita Mehta and Peter Wainwright from the University of California, Davis, discovered evolutions breathtaking solution. Rather than prey coming to the pharyngeal jaws, the pharyngeal jaws move forwards into the mouth cavity, trapping the prey and dragging it backwards. This, the researchers say, is the first described case of a vertebrate using a second set of jaws to both restrain and transport prey, and is the only known alternative to the hydraulic prey transport reported in most bony fish a major innovation that could have contributed to the success of moray eels as predators. The mechanics of the morays pharyngeal jaws are reminiscent of the ratchet mechanisms used by snakes also long, thin and highly predatory creatures. This is an instance of convergence, the evolutionary phenomenon in which distantly related creatures evolve similar solutions to common problems. This study demonstrates the contingent nature of evolution; as a process it does not have the luxury of designing from scratch.

Reference Mehta, R. S. & Wainwright, P. C. Nature 449, 7982 (2007). Additional resource Westneat, M. W. Nature 449, 3334 (2007). Author websites Rita Mehta: http://www.eve.ucdavis.edu/~wainwrightlab/rsmehta/index.html Peter Wainwright: http://www.eve.ucdavis.edu/~wainwrightlab
12 Darwins Galapagos finches
When Charles Darwin visited the Galapagos Islands, he recorded the presence of several species of finch that all looked very similar except for their beaks. Ground finches have deep and wide beaks; cactus finches have long, pointed beaks; and warbler finches have slender, pointed beaks, reflecting differences in their respective diets. Darwin speculated that all the finches had a common ancestor that had migrated to the islands. Close relatives of the Galapagos finches are known from the South American mainland, and the case of Darwins finches has since become the classic example of how natural selection has led to the evolution of a variety of forms adapted to different ecological niches from a common ancestral species termed adaptive radiation. This idea has since been reinforced by data showing that even small differences in the depth, width or length of the beak can have major consequences for the overall fitness of birds. To find out what genetic mechanisms underlie the changes in beak shape that mark each species, Harvard Universitys Arhat Abzhanov and his colleagues examined numerous genes that are switched on in the developing beaks of finch chicks; their study was published in 2006. The researchers discovered that shape differences coincide with differing expression of the gene for calmodulin, a molecule involved in calcium signalling that is vital in many aspects of development and metabolism. Calmodulin is expressed more strongly in the long and pointed beaks of cactus finches than in the more robust beaks of other species. Artificially boosting the expression of calmodulin in the embryonic tissues that give rise to the beak causes an elongation of the upper beak, similar to that seen in cactus finches. The results show that at least some of the variation in beak shape in Darwins finches is likely to be related to variation in calmodulin activity, and implicates calmodulin in the development of craniofacial skeletal structures more generally. The study shows how biologists are going beyond the mere documentation of evolutionary change to identify the underlying molecular mechanisms.

Reference Abzhanov, A. et al. Nature 442, 563567 (2006). Author websites Clifford Tabin: http://www.hms.harvard.edu/dms/bbs/fac/tabin.html Peter Grant: http://www.eeb.princeton.edu/FACULTY/Grant_P/grantPeter.html
Microevolution meets macroevolution
Darwin conceived of evolutionary change as happening in infinitesimally small steps. He called these insensible gradations, which, if extrapolated over long periods of time, would result in wholesale changes of form and function. There is a mountain of evidence for such small changes, called microevolution the evolution of drug resistance, for instance, is just one of many documented examples. We can infer from the fossil record that larger species-to-species changes, or macroevolution, also occur, but they are naturally harder to observe in action. That said, the mechanisms of macroevolution can be seen in the here-and-now, in the architecture of genes. Sometimes genes involved in the day-to-day lives of organisms are connected to, or are even the same as, those that govern major features of animal shape and development. So everyday evolution can have large effects. Sean Carroll from the Howard Hughes Medical Institute in Chevy Chase, Maryland, and his colleagues looked at a molecular mechanism that contributes to the gain of a single spot on the wings of male flies of the species Drosophila biarmipes; they reported their findings in 2005. The researchers showed that the evolution of this spot is connected with modifications of an ancestral regulatory element of a gene involved in pigmentation. This regulatory element has, over time, acquired binding sites for transcription factors that are ancient components of wing development. One of the transcription factors that binds specifically to the regulatory element of the yellow gene is encoded by engrailed, a gene fundamental to development as a whole. This shows that a gene involved in one process can be co-opted for another, in principle driving macroevolutionary change.
Reference Gompel, N., Prudhomme, B., Wittkopp, P. J., Kassner, V. A. & Carroll, S. B. Nature 433, 481487 (2005). Additional resources Hendry, A. P. Nature 451, 779780 (2008). Prudhomme, B. et al. Nature 440, 10501053 (2006). Author website Sean Carroll: http://www.hhmi.org/research/investigators/carroll_bio.html
Toxin resistance in snakes and clams
Biologists are increasingly coming to understand the molecular mechanisms that underlie adaptive evolutionary change. In some populations of the newt Taricha granulosa, for example, individuals accumulate the nerve poison tetrodotoxin in their skin, apparently as a defence against garter snakes (Thamnophis sirtalis). Garter snakes that prey on the newts that produce tetrodotoxin have evolved resistance to the toxin. Through painstaking work, Shana Geffeney at the Stanford School of Medicine in California and her colleagues uncovered the underlying mechanism; their study was published in 2005. Variation in the level of resistance of garter snakes to their newt prey can be traced to molecular changes that affect the binding of tetrodotoxin to a particular sodium channel. Similar selection for toxin resistance apparently occurs in softshell clams (Mya arenaria) in areas of the North American Atlantic coast, as reported by Monica Bricelj at the Institute for Marine Biosciences in Nova Scotia, Canada, and her colleagues in the same issue of Nature. The algae that produce red tides generate saxitoxin the cause of paralytic shellfish poisoning in humans. Clams are exposed to the toxin when they ingest the algae. Clams from areas subject to recurrent red tides are relatively resistant to the toxin and accumulate it in their tissues. Clams from unaffected areas have not evolved such resistance. Resistance to the toxin in the exposed populations is correlated with a single mutation in the gene that encodes a sodium channel, at a site already implicated in the binding of saxitoxin. It seems likely, therefore, that the saxitoxin acts as a potent selective agent in the clams and leads to genetic adaptation. These two studies show how similar selective pressures can lead to similar adaptive responses even in very different taxa.

References Geffeney, S. L., Fujimoto, E., Brodie, E. D., Brodie, E. D. Jr, & Ruben, P. C. Nature 434, 759763 ( 2005). Bricelj, V. M. et al. Nature 434, 763767 (2005). Additional resources Mitchell-Olds, T. & Schmitt, J. Nature 441, 947952 (2006). Bradshaw, H. D. & Schemske, D. W. Nature 426, 176178 (2003). Coltman, D. W., ODonoghue, P, Jorgenson, J. T., Hogg, J. T. Strobeck, C. & Festa-Bianchet, M. Nature 426, 655658 (2003). Harper Jr, G. R. & Pfennig, D. W. Nature 451, 11031106 (2008). Ellegren, H. & Sheldon, B. Nature 452, 169175 (2008). Author websites Shana Geffeney: http://wormsense.stanford.edu/people.html Monica Bricelj: http://marine.biology.dal.ca/Faculty_Members/Bricelj,_Monica.php
15 Variation versus stability
Species can remain mostly unchanged for millions of years, long enough for us to pick up their traces in the fossil record. But they change, too, and often very suddenly. This has led some to wonder whether species usually those developing along specific tracks store the potential for sudden change under the hood, unleashing a flood of otherwise hidden variation at times of environmental stress variation on which selection can act. This idea of such evolutionary capacitance was first mooted by Suzanne Rutherford and Susan Lindquist in startling experiments on fruitflies. Their idea was that key proteins involved in the regulation of developmental processes are chaperoned by a protein called Hsp90 that is produced more at times of stress. On occasion, Hsp90 is overwhelmed by other processes and the proteins it normally regulates are left to run free, producing a welter of otherwise hidden variation. Aviv Bergman from the Albert Einstein College of Medicine in New York and Mark Siegal at New York University explored whether evolutionary capacitance is particular to Hsp90 or found more generally; their study was published in 2003. They used numerical simulations of complex gene networks and genomewide expression data from yeast strains in which single genes had been deleted. They showed that most, and perhaps all, genes hold variation in reserve that is released only when they are functionally compromised. In other words, it looks as if evolutionary capacitance might go wider and deeper than Hsp90.

Reference Bergman, A. & Siegal, M. L. Nature 424, 549552 (2003). Additional resources Stearns, S. C. Nature 424, 501504 (2003). Rutherford, S. L. & Lindquist, S. Nature 396, 336342 (1998). Author websites Mark Siegal: http://www.nyu.edu/fas/biology/faculty/siegal/index.html Aviv Bergman: http://www.bergmanlab.org Susan Lindquist: http://www.wi.mit.edu/research/faculty/lindquist.html Suzanne Rutherford: http://depts.washington.edu/mcb/facultyinfo.php?id=142 Stephen Stearns: http://www.yale.edu/eeb/stearns
FOR MORE ON EVOLUTION AND DARWIN Visit www.nature.com/darwin

IMAGE: Jonathan Williams

www.nature.com/nature