Evolutionary Transitions Among Dinosaurs: Examples from the Jurassic of China
© Springer Science+Business Media, LLC 2009
Received: 8 September 2008
Accepted: 2 April 2009
Published: 9 May 2009
Dinosaurs have captured the popular imagination more than any other extinct group of organisms and are therefore a powerful tool in teaching evolutionary biology. Most students are familiar with a wide variety of dinosaurs and the relative suddenness of their extinction, but few are aware of the tremendous longevity of their time on Earth and the richness of their fossil record. We first review some of the best-known groups of dinosaurs and discuss how their less-specialized relatives elucidate the path through which each evolved. We then discuss our recent discovery of Yinlong downsi, a distant relative of Triceratops, and other fossils from Jurassic deposits in China to exemplify how the continuing discovery of fossils is filling out the dinosaur family tree.
KeywordsDinosaurs Evolution Transitional fossils Mesozoic China Jurassic
Dinosaurs need no introduction, they are ubiquitous in popular media as symbols of the terrors of nature and how the mighty fall. But the richness and remarkable duration of the fossil record of dinosaurs are less widely appreciated, and the abundant fossil record of dinosaurs is an excellent example of how fossils provide evidence for evolutionary relationships and transitions. Familiar highly specialized animals such as Triceratops, Pachycephalosaurus, Stegosaurus, Ankylosaurus, Tyrannosaurus, Brachiosaurus, and Parasaurolophus are joined in the fossil record by hundreds of their relatives with less-specialized features (Weishampel et al. 2004). Together, they comprise a many-branched evolutionary tree that flourished for over 160 million years.
Dinosaurs are found almost exclusively in sedimentary rocks formed in terrestrial, rather than marine, environments, indicating that they rarely ventured out into the sea. These river and lake deposits formed layers, or strata, that are overlain by younger rocks and underlain by older ones. By studying how these and other layers of rocks and their entombed fossils are stacked, scientists called stratigraphers have been able to recognize the sequence in which dinosaurs and other fossils occurred during the Earth’s history (Prothero and Schwab 2003). This led to the erection in the early nineteenth century of a timescale with now-familiar names for different time periods, including the Mesozoic Era (popularly called the Age of Dinosaurs) and its three subdivisions, from oldest to youngest, the Triassic, Jurassic, and Cretaceous periods (for a fascinating account of debates in the early nineteenth century over how the time periods were recognized, see Rudwick 1985). However, these divisions reflected only the relative ages of these fossils, and it was not until the last 60 years that ways of estimating the age in years of rocks and fossils were developed. These involve the decay of radioactive elements, which are experimentally shown to occur at rates that take millions or billions of years to complete, depending on the element and isotope.
Radiometric dating comprises a well-developed set of techniques that are now capable of very high resolution (Macdougall 2008). Only relatively young fossils (<50,000 years old) can be dated directly, by Carbon-14 dating of bone or other carbonaceous fossils, so the dating of dinosaurs millions of years old relies upon dating of the rock layers above, below, and rarely within beds containing fossils. The radioactive decay of isotopes of several different elements (argon, potassium, and uranium) can be used to date dinosaur-bearing rocks because they all have half-lives (the statistically estimated time it would take half of a sample to decay) of the appropriate length, and the use of different isotopes can provide independent checks on any one date.
Radiometric dating estimates the age of a rock at the time that it crystallized, by comparing the amount of “daughter” decay product to the original amount of the radioactive “mother” isotope (calculated as the amount of decay product plus what remains of the mother isotope). Rock layers such as tuffs that formed when volcanic ash, including recently formed tiny crystals, was ejected from a volcano and blanketed the countryside, are ideal for dating. Many tuffs occur in dinosaur-bearing deposits or above or below them, so we have a good idea of how old most dinosaur fossils are (and can estimate all the others).
The oldest dinosaurs, from the Valley of the Moon in Argentina, are dated at 227.8 ± 0.3 million years (Rogers et al. 1993; a report of an older dinosaur from Madagascar is being revised), and the extinction event 65.95 ± 0.04 million years ago (Kuiper et al. 2008) apparently killed off all of the dinosaurs alive at that time while leaving their descendents, the birds. Dinosaurs were therefore on the Earth for at least 161 million years, and their descendents the birds (reference to Padian contribution) continued on for another 65.95 million years (and thankfully show few signs of leaving us). To put this in perspective, the oldest human fossils are six to seven million years old (Brunet et al. 2002), so the dinosaur fossil record is about 23 to 27 times longer than that of humans.
As with the fossil record in general (Kidwell and Holland 2002), the record of dinosaur fossils is not consistent through time because the deposition of sedimentary rocks in the environments in which dinosaurs lived varied at different times in the Earth’s history. This depended both on local conditions and on global phenomena such as plate tectonics and sea level changes, as the mountain building that fueled sedimentation waxed and waned. The incomplete sampling of ancient species by the sedimentary record, coupled with the episodic, greatly incomplete record that was left of those sediments, could not possibly produce a series of forms showing all stages of continuous change from one species to another (whether or not absolutely complete continuity in morphology ever existed is an open question). But some examples of small changes between very similar species over time exist, such as the three successive species of the horned dinosaur Chasmosaurus over approximately 2.5 million years within the Dinosaur Park Formation of Canada (Holmes et al. 2001). More to the point, the periodic sampling of fossils over millions of years is sufficient to provide documentation of the sequential acquisition of specialized features in all of the major dinosaur lineages.
Fossiliferous rocks are more common near the end of the dinosaur’s record, at the close of the Cretaceous Period, than in the Jurassic and Triassic, so we know much more about the dinosaurs that lived shortly before they went extinct than those from earlier in their evolution. Nevertheless, the early fossil record of dinosaurs is reasonably well known and includes many complete or nearly complete skeletons. These early fossils are important because older fossils are more likely to include the most basal members of evolutionary lineages, lacking most of the specializations of later forms (we use basal here to refer to members of a group that diverged earlier than other members). These early dinosaurs are generally smaller than their later brethren and have only a few of the features distinguishing the lineage to which they belong. The search for early dinosaurs is among the “hottest” areas of dinosaur research because such dinosaurs are not as common as later ones and hold great potential for elucidating the dinosaurs’ early diversification into major evolutionary branches.
A tally of dinosaurs published 3 years ago (Wang and Dodson 2006) recognized 527 genera, the vast majority with only a single species. (Dinosaur paleontologists have a partiality to naming new genera even when a new species is closely related to an existing genus, something our entomological colleagues find mildly disturbing.) For each of these genera, a fossil exists with features that, theoretically, clearly differentiate it from every other known dinosaur genus. For some of these genera, skeletons of many individuals are known (e.g., Protoceratops, Psittacosaurus, Allosaurus, Coelophysis, and Iguanodon), sometimes in the hundreds, and one remarkable bone bed in Alberta, Canada, preserves bones from thousands of individuals of Centrosaurus, a one-horned relative of Triceratops. Wang and Dodson use the known number of dinosaurs and the rate at which new genera are discovered to estimate that 1,850 genera of dinosaurs had existed, in which case less than a third have been discovered.
Evolutionary relationships are evident through the features shared by groups of organisms (Gregory 2008). Just as a very high percentage of human DNA is identical to that of our closest relatives, the chimpanzees, closely related species of all kinds share similar DNA, anatomy, and physiology. One of the main reasons the theory of evolution is accepted is that it provides an explanation for why similar features are found in different organisms. To cite just one example, the presence in all land-living vertebrates of four limbs with digits (reference to Shubin contribution) is explained by the close evolutionary relationships of all tetrapods to each other, as members of the same lineages. Supposed alternative theories such as Intelligent Design are mute in explaining these rampant similarities.
Fossil vertebrates—those ancient animals with a bony skeleton—typically preserve only the bones or in rare cases soft tissues such as skin, cartilage, and feathers. DNA is not reliably known from dinosaurs, and recent reports of intact protein (Schweitzer et al. 2007) are intriguing but controversial (Pevzner et al. 2008). Anatomical features of bones such as the peculiar anatomy of the hip that distinguishes all dinosaurs have proven to be extremely useful in understanding dinosaur relationships and show the same hierarchical pattern of nested sets of features that Darwin recognized as reflecting evolutionary relationships. For example, they show that dinosaurs are divided into three distinctive groups—the herbivorous ornithischians such as Triceratops and duck-bills, the huge sauropodomorphs such as Diplodocus, and the generally carnivorous theropods such as Tyrannosaurus rex—and that the sauropodomorphs and theropods are closer to each other than to ornithischians.
Much of paleontological research is devoted to understanding the evolutionary relationships of the species represented in the fossil record. A typical study involves the assembly of a table (data matrix) in which a long list of features are marked as either present or absent for each of the species thought to be closely related to one another based upon previous research or preliminary analysis. For example, a report of a new stegosaur might present a table showing the presence or absence in each known stegosaur species of those features that differ among them. These data are then analyzed using computer algorithms that find the set of evolutionary relationships that most simply explains the features shared by groups of stegosaurs. The result is an estimate of the evolutionary tree of stegosaurs.
The Humble Origins of Some Distinctive Dinosaurs
The most memorable dinosaurs are those with weird and wonderful features, such as hollow crests, huge horns, giant claws, or bony plates in their skin. These tend to occur in the Late Jurassic and Cretaceous, in the latter half of the Age of Dinosaurs. Many examples of evolutionary transitions are available in the dinosaur fossil record, but here we present some of the more dramatic ones (for more detailed reviews, see Sereno 1999 and Weishampel et al. 2004). This is a greatly simplified account since dinosaur anatomy is complex and comprises hundreds of different features, and dinosaur evolution was of course not a simple march in a single direction but a many-branched tree with many nuances.
The hadrosaurs of the Late Cretaceous are preceded in the fossil record by a long series of forms that show the incremental acquisition of their specializations, and together with hadrosaurs these forms comprise the group Ornithopoda. The genus Iguanodon of the Early Cretaceous (Fig. 1c, d) is represented by many skeletons and several species, some of which are more closely related to hadrosaurs than are other Iguanodon species (for this reason, most species may not legitimately belong within the genus, and several genera have recently been named to recognize this; Paul 2008). Iguanodon have teeth that are closely packed to form a grinding surface, but there are many fewer teeth than in hadrosaurs. The bill is present, but it is smaller and narrower than that of hadrosaurs. Less specialized still are genera such as Hypsilophodon of the Early Cretaceous (Fig. 1e), which lacks a bill altogether and has teeth that are not closely packed.
The specialized ceratopsians with horns and frills—the ceratopsids—are only known in the later part of the Late Cretaceous, and more basal ceratopsians mainly from earlier deposits demonstrate the sequence in which these features evolved. Protoceratopsids, which occur in huge numbers in sand dune deposits of the Gobi Desert, have a frill and an epijugal bone and walked on four legs but lack horns (Fig. 2g). Psittacosaurus (Fig. 2h), from the Early Cretaceous, also lacked horns and had only a small frill and a flared cheek without the epijugal bone, and small forelimbs indicate that it walked primarily on its hind legs. The number of teeth and their close packing are reduced in these more basal species.
Sauropods were herbivorous (as indicated by the shape of their teeth), walked on all four limbs and had relatively long necks, and their skeletons have many features related to their large size, such as broad hind feet that spread the weight. Their closest relatives are often called prosauropods and comprise a series of forms that sequentially gained sauropod features. These Late Triassic and Early Jurassic forms include Plateosaurus (Fig. 3b), which had a long neck and plant-eating teeth but walked on its hind legs, and Thecodontosaurus, which had a relatively short neck, walked on its hind legs, and had teeth indicating a more omnivorous diet. Most prosauropods were distinctly smaller than sauropods but were still relatively large, but the oldest and most basal form, Saturnalia, was only about 5 feet (1.5 meters) long.
Ankylosaurs and Stegosaurs
The specialized features of stegosaurs, ankylosaurs, and of the entire group Thyreophora were sequentially acquired through a series of more basal forms. The basal stegosaur Huayangosaurus (Fig. 5c) has a wider and taller skull than other stegosaurs, among many other primitive features, and basal ankylosaurs like Cedarpelta and Minmi (Fig. 4c) lack most of the bumps and processes that make the skull of ankylosaurs so strange. Scelidosaurus (Fig. 5d), Emausaurus, and Scutellosaurus from the Early Jurassic are more basal than both ankylosaurs and stegosaurs but like them have bony plates in their skin, though they are simpler. These three forms share a few other features with stegosaurs and ankylosaurs, such as a relatively long body and pelvis, but are decidedly more basal, and the limbs of Scutellosaurus indicate that it walked mainly on its hind legs.
Until recently, tyrannosaurids were only known with certainty from the Late Cretaceous. However, recent discoveries have contributed greatly to our understanding of early tyrannosaurid evolution. In particular, the discovery of Dilong (Fig. 7b) in the Early Cretaceous of China (Xu et al. 2004) showed not only that tyrannosaurs began as relatively small animals with long arms but that they were covered with short filamentous feathers (“protofeathers”). This was not entirely surprising, as it was known that tyrannosaurs are closely related to birds, though not as close as are some other theropods. Feathers probably occurred in other tyrannosaurs, but only rarely are conditions in the fossil record ideal for preserving soft tissues such as these.
The first of these discoveries was Beipiaosaurus, from the Early Cretaceous of China (Xu et al. 1999), which showed that basal therizinosaurs had a three-toed foot like theropods, and the four-toed prosauropod-like foot evolved later in therizinosaurs independent of prosauropods. Beipiaosaurus also was preserved with filamentous feathers, like those of the tyrannosaur Dilong. More recently, an even more basal therizinosaur was discovered in the Early Cretaceous of Utah, Falcarius (Kirkland et al. 2005), which mainly has primitive features but also a few distinctive therizinosaur features (Fig. 8b). It has leaf-like teeth and a broad pelvis like other therizinosaurs, interpreted to be indicative of an herbivorous diet (the broad pelvis is thought to indicate a relatively large stomach, as in living herbivorous mammals). But unlike all other therizinosaurs Falcarius lacks a distinctive ridge on its lower jaw; its pelvis is not as specialized, and its foot is even more like that of other theropods than is that of Beipiaosaurus.
Our Search for Early Dinosaurs in the Gobi Desert of China
The search for new dinosaurs is an ongoing effort involving hundreds of paleontologists. Our own recent efforts targeted fossil deposits in northwestern China, in the Junggar Basin west of Mongolia. This region is in the westernmost part of the Gobi Desert, a desert renowned for its rich fossil beds. We chose the Junggar Basin because earlier exploration of the sediments here by Chinese and Canadian paleontologists discovered intriguing fossils from a poorly known time period critical to understanding the evolution of dinosaurs.
After the origin of dinosaurs in the Triassic Period, they diversified and first reached immense sizes in the Jurassic Period. The dinosaurs of the Early Jurassic are large by modern standards (some were larger than elephants) but relatively small compared to the giants of the Late Jurassic and Cretaceous. Although the three major dinosaur groups (ornithischians, sauropodomorphs, and theropods) were already present, many of the distinctive lineages within each did not appear until later. Surprisingly, Early Jurassic dinosaurs from around the world are very similar, a reflection of the geographic continuity between all of the continents, which were then part of the single supercontinent of Pangaea. The diversification and enlargement of dinosaurs later in the Jurassic corresponded with the earliest splitting of Pangaea, marked by the opening of the Atlantic Ocean, which ultimately resulted in the continental configurations we know today.
One of the most poorly known time periods in the history of dinosaurs is the middle part of the Jurassic. It was during this period that many distinctive groups of dinosaurs, including the first gigantic ones, appeared. Only three areas in the world are known to have fossil beds with relatively abundant dinosaur skeletons at this time, and two of them are in China (the other is in the Patagonian region of Argentina). The first to be discovered is in the southern Chinese province of Szechuan, near the city of Zigong. Rich fossil beds densely packed with dinosaur skeletons occur here, and there is a spectacular display of in situ skeletons at a museum that should be on every dinosaur fan’s agenda should they tour China.
We chose to concentrate our efforts in another area of China, the Shishugou Formation in the Xinjiang Autonomous Region near where the ancient Silk Road once passed. Based upon previous discoveries there, such as the theropod dinosaur Sinraptor and the sauropod dinosaur Bellusaurus, we reasoned that an immense area that had not been explored must preserve many more fossils. For six summers, from 2001 to 2006, we assembled a large group of scientists, technicians, students, and local workers to camp out in the badlands and comb the area for fossils.
Wucaiwan proved to be a treasure trove of fossils, and curiously they are predominately small, with very few large fossils. This is in many ways fortunate because smaller fossils are easier to excavate than huge skeletons, and a greater number and diversity can be collected in a short time. Huge fossils are preserved at Wucaiwan, such as the occasional sauropod vertebra or leg bone we encountered, but for reasons having to do with how the sediments were being deposited we found no intact large skeletons.
One of the first things we did was to collect samples of tuffs for dating. Fortunately, tuffs occur at several levels in the formation, so we could bracket most of the fossil occurrences between these dated tuffs. Radiometric dating of these tuffs showed that the rocks forming the Shishugou Formation at Wucaiwan were first deposited near the end of the Middle Jurassic and continued into the early part of the Late Jurassic about 159 million years ago.
We discovered at least seven new species of dinosaurs at Wucaiwan, mostly theropods. This is odd because theropods, which were generally carnivores, are relatively rare in the fossils record, as are carnivores in modern ecosystems. One of the reasons for the abundance of theropods at Wucaiwan is that many of the theropods we collected had been trapped in mud pits, small areas of watery sticky mud that the theropods were unable to extricate themselves from. These “death traps” were the subject of a TV documentary by National Geographic and an article in National Geographic Magazine (Gwin 2008). The skeletons in these pits are predominately those of theropods, a highly unusual occurrence. We surmise that the theropods’ long gracile legs and, in some cases, weak arms, made them more vulnerable than other animals to the sticky mud.
One of the most important and revealing discoveries we made at Wucaiwan was the ceratopsian Yinlong (from the Chinese words for “hidden dragon,” a homage to the movie filmed nearby). Yinlong (Fig. 10c) is clearly a ceratopsian because it shares several distinctive features with them including a bone at the end of its nose found only in ceratopsians (the rostral bone). But it is extremely primitive, more so than any other known ceratopsian, as might be expected since it is also the oldest known ceratopsian. Its limbs indicate that it walked on its hind limbs rather than on all four legs; there is almost no frill behind its head; there is little flaring of the cheek; the teeth are not compressed together, and the rostral bone is much smaller than in all other ceratopsians. More surprising is that, in addition to the features it shares with ceratopsians, it also possesses a few features that previously were known only in pachycephalosaurs.
Paleontologists suspected that ceratopsians and pachycephalosaurs were related, so they were placed together in a group called Marginocephalia, named for the presence in both of a bony outgrowth around the back margin of the skull. What Yinlong shows is that early ceratopsians shared several features with pachycephalosaurs that were retained from the common ancestor of both groups. The skull of pachycephalosaurs is distinguished mainly by bony outgrowths along the edge and the top of the skull. Some of these are seen in Yinlong, in which the edges around the top of the skull, behind the eye sockets, and on the lower jaw have rough areas in the same positions as the bony outgrowths of pachycephalosaurs. These and a few other features considerably strengthen the evidence that ceratopsians and pachycephalosaurs, two of the most specialized groups of dinosaurs, shared a common ancestor in the Jurassic.
The Wucaiwan fossils continue to shed light on the early evolution of dinosaurs, and many we collected have not yet been studied. Already, they include primitive members of three of the major groups of dinosaurs that flourished near the end of the Cretaceous—the ceratopsians, duck-bills, and tyrannosaurs—at least two of which are the oldest representatives of their groups. Ongoing studies of these fossils promise to shed light on the early evolution of several groups of dinosaurs, especially the theropods closely related to birds.
Support for this project was provided by the National Geographic Society, the National Natural Science Foundation of China, and the US National Science Foundation.
Images provided by permission of G.S. Paul (1b,c, 3a, 5d, 7a, 8a,b), K. Carpenter (4b), American Museum of Natural History (2g,h), D. Nicholls (1a), D. Norman (1d), T. Maryanska and Palaeontologia Polonica (6), P. Sereno and J. Vert. Paleo. (5b,c), S. Sampson and J. Vert. Paleo. (2b,d), R. Molnar and the Queensland Museum (4c), Yale Peabody Museum (2a,c,e,f), P. Galton (1e, 3b), Nature (7b, 10a,b), and Royal Soc. London (10c).
- Brunet M, et al. A new hominid from the Upper Miocene of Chad, Central Africa. Nature. 2002;418:145–51. doi:https://doi.org/10.1038/nature00879.View ArticleGoogle Scholar
- Gregory TR. Understanding evolutionary trees. Evo Edu Outreach. 2008;1:127–37.Google Scholar
- Gwin P. The real Jurassic park. Nat Geog. 2008;214:104–15.Google Scholar
- Holmes R, Forster C, Ryan M, Shepherd K. A new species of Chasmosaurus (Dinosauria: Ceratopsia) from the Dinosaur Park Formation of southern Alberta. Can J Earth Sci. 2001;38:1423–38. doi:https://doi.org/10.1139/cjes-38-10-1423.View ArticleGoogle Scholar
- Kidwell SM, Holland SM. The quality of the fossil record. Annu Rev Ecol Syst. 2002;33:561–88. doi:https://doi.org/10.1146/annurev.ecolsys.33.030602.152151.View ArticleGoogle Scholar
- Kirkland JI, Zanno LE, Sampson SD, Clark JM, DeBlieux DD. A primitive therizinosauroid dinosaur from the early Cretaceous of Utah. Nature. 2005;435:84–7. doi:https://doi.org/10.1038/nature03468.View ArticleGoogle Scholar
- Kuiper KF, Deino A, Hilgen FJ, Krijgsman W, Renne PR, Wijbrans JR. Synchronizing rock clocks of Earth history. Science. 2008;320:500–4. doi:https://doi.org/10.1126/science.1154339.View ArticleGoogle Scholar
- Macdougall D. Nature’s clocks: how scientists measure the age of almost everything. Berkeley: University of California Press; 2008. p. 271.Google Scholar
- Paul GS. A revised taxonomy of the iguanodont dinosaur genera and species. Cretac Res. 2008;29:192–216.View ArticleGoogle Scholar
- Pevzner PA, Kim S, Ng J. Comment on “protein sequences from Mastodon and Tyrannosaurus rex revealed by mass spectrometry”. Science. 2008;321(5892):1040. doi:https://doi.org/10.1126/science.1155006.View ArticleGoogle Scholar
- Prothero D, Schwab F. Sedimentary geology. 2nd ed. New York: Freeman; 2003. p. 600.Google Scholar
- Rogers RR, Swisher CC III, Sereno PC, Monetta AM, Forster CA, Martínez RN. The Ischigualasto tetrapod assemblage (Late Triassic, Argentina) and 40Ar/39Ar dating of dinosaur origins. Science. 1993;260:794–7. doi:https://doi.org/10.1126/science.260.5109.794.View ArticleGoogle Scholar
- Rudwick MJS. The great Devonian controversy. Chicago: University of Chicago Press; 1985. p. 494.View ArticleGoogle Scholar
- Schweitzer MH, Suo Z, Avci R, Asara JM, Allen MA, Arce FT, et al. Analyses of soft tissue from Tyrannosaurus rex suggest the presence of protein. Science. 2007;316:277–80. doi:https://doi.org/10.1126/science.1138709.View ArticleGoogle Scholar
- Sereno PC. The evolution of dinosaurs. Science. 1999;284:2137–47. doi:https://doi.org/10.1126/science.284.5423.2137.View ArticleGoogle Scholar
- Wang SC, Dodson P. Estimating the diversity of dinosaurs. Proc Natl Acad Sci U S A. 2006;103:13601–5. doi:https://doi.org/10.1073/pnas.0606028103.View ArticleGoogle Scholar
- Weishampel D, Dodson P, Osmólska H. The Dinosauria. 2nd ed. Berkeley: University of California Press; 2004. p. 861.View ArticleGoogle Scholar
- Xu X, Tang ZL, Wang XL. A therizinosauroid dinosaur with integumentary structures from China. Nature. 1999;399:350–4. doi:https://doi.org/10.1038/20670.View ArticleGoogle Scholar
- Xu X, Norell M, Kuang X, Wang X, Zhao Q, Jia C. Basal tyrannosauroids from China and evidence for protofeathers in tyrannosauroids. Nature. 2004;431:680–4. doi:https://doi.org/10.1038/nature02855.View ArticleGoogle Scholar
- Xu X, Clark JM, Forster CA, Norell MA, Erickson GM, Eberth DA, et al. A basal tyrannosauroid dinosaur from the Late Jurassic of China. Nature. 2006a;439:715–8. doi:https://doi.org/10.1038/nature04511.View ArticleGoogle Scholar
- Xu X, Forster CA, Clark JM, Mo J. A basal ceratopsian with transitional features from the Late Jurassic of northwestern China. Proc R Soc Lond B Biol Sci. 2006b;273:2135–40. doi:https://doi.org/10.1098/rspb.2006.3566.View ArticleGoogle Scholar