The great diversity observed in morphology, behaviour, breeding systems and other ecological characters in shorebirds (Charadriiformes) provides tests of evolutionary hypotheses of life-history traits (e.g. Myers 1981; Whitfield & Tomkovich 1996; Barbosa & Moreno 1999). A well-supported phylogeny is a prerequisite for comparative studies because it can be used to understand how and when adaptive changes took place. For instance, by mapping morphological characters onto a well-supported phylogeny of the shanks derived from nuclear and mitochondrial DNA gene sequences, Pereira & Baker (2005) identified many characters that have retained ancestral states or have evolved in parallel. Although the phylogenetic relationships within shorebirds are well established at the family level based on nuclear and/or mitochondrial sequences (Ericson et al. 2003; Paton et al. 2003; Paton & Baker 2006), they are uncertain or unknown for many groups of species and genera, including enigmatic taxa such as the ibisbill (Ibidorhyncha struthersii), thick-knees (Burhinidae) and sheathbills (Chionidae). Hence, most of the evolution of life-history traits among shorebirds has yet to be mapped onto a well-structured phylogenetic framework.

To overcome this problem, Thomas et al. (2004) used the matrix representation with parsimony method (also known as supertree method) to propose the most inclusive phylogenetic hypothesis for extant shorebirds to date. However, their framework may seriously compromise interpretations of evolutionary histories of shorebirds because the supertree method lacks measures of nodal support and is known to be highly prone to biases depending on which source trees are included in the analysis, and by treating well-supported and poorly supported source trees as equally likely. Additionally, the supertree presented in Thomas et al. (2004) falls short in providing good resolution for many congeneric species because not enough source trees or none at all exist for these groups.

DNA sequencing is the logical approach to gather enough data and establish well-resolved phylogenetic hypotheses that better represent the evolutionary history of the Tree of Life. Here, we provide a comprehensive phylogeny and divergence times for shorebird genera based on DNA sequences of four genes. Our study provides a scaffold for future studies aimed at resolving phylogenetic relationships among species of shorebirds as well as a temporal framework for the evolution of life-history traits among shorebirds.

We sampled 90 out of 96 putative genera of Charadriiformes including two species of the enigmatic Turnix for DNA amplification and sequencing of the small ribosomal subunit (12S rDNA), NADH dehydrogenase subunit 2 (ND2), cytochrome b (cyt b) and recombination-activating protein (RAG-1) gene (species, primers and PCR conditions given in the electronic supplementary material). Gene fragments were concatenated in an alignment of 5198bp, including gaps. All sequences obtained in this study were deposited in GenBank (table 1 in electronic supplementary material).

Tree inference was performed by a Metropolis-coupled Markov chain Monte Carlo Bayesian approach by running two simultaneous independent runs, each with one cold and five heated chains as implemented in MrBayes v. 3.1.2 (Ronquist & Huelsenbeck 2003). Divergence times were estimated using a Bayesian approach that accounts for uncertainties in branch lengths for individual gene partitions, rates of evolution and time constraints (Thorne & Kishino 2002). For details, see the electronic supplementary material (figures 1 and 2).

Figure 1 Bayesian tree for Charadriiformes genera. Letters A to P indicate nodes for which fossil or molecular time constraints were used to estimate divergence times (see table 2 in electronic supplementary material). Numbers at nodes are posterior probabilities (more ...)

Figure 2 Bayesian chronogram for Charadriiformes genera. The Cretaceous/Paleocene boundary is marked by a grey bar. Nodes are labelled A, C, L and S for Avian outgroups, Charadrii, Lari and Scolopaci divergences, respectively.

The Scolopaci comprises two major clades of genera, one containing the Scolopacidae (sandpipers, snipes and allies) and the other containing the Jacanidae and Rostratulidae (Jacana and allies and painted snipes Nyctocryphes and Rostratula, respectively) as a sister group to the Pedionomidae plus the Thinocoridae (Plains Wanderer Pedionomus and seedsnipes Thinocorus, respectively). Within the Scolopacidae, the calidridine sandpipers (Calidris and allies) plus the turnstones (Arenaria) are sister to the shanks (Tringa and allies) and phalaropes (Phalaropus). Clades formed by the snipes and woodcock (Gallinago, Coencorypha, Scolopax) and dowitchers (Limnodromus) plus jacksnipe (Lymnocryptes), then the godwits (Limosa), and finally the curlews (Numenius) plus Upland sandpiper (Bartramia) branch off in a pectinate sequence identical to that in the nuclear RAG-1 tree of Paton et al. (2003). Within the jacanas, we found the same phylogenetic relationships among geographically disjunct sister pairs of genera reported by Whittingham et al. (2000). The predominantly Australian Irediparra is sister to the African Microparra and the Asian Hydrophasianus is sister to New World Jacana, consistent with their hypothesis of extinction of intervening forms to explain this biogeographic pattern.

In the Charadrii, the typical plovers (Charadrius and allies) are in one clade with the notable exception that Pluvialis groups with the oystercatchers (Haematopus) and ibisbill (Ibidorhyncha), stilts (Himantopus, Cladorhynchus) and avocets (Recurvirostra), thereby rendering typical plovers paraphyletic. The lapwings (Vanellus) are sister to three Australasian genera of plovers (Anarhynchus, Peltohyas, Erythrogonys). Additionally, the enigmatic Egyptian plover (Pluvianus) is sister to these clades. Thick-knees (Burhinus, Esacus) form a clade with sheathbills (Chionis) and the Magellanic Plover (Pluvianellus) which are the basal sister group to the rest of the Charadrii, as in previous DNA trees (Ericson et al. 2003; Paton et al. 2003; Paton & Baker 2006). Finally, the phylogenetic placement of the sandgrouse (Pterocles) remains unresolved, but they do not appear to be shorebirds.

While it is tempting to suggest systematic revisions of the Charadriiformes based on our results, caution is warranted because anomalous gene trees that differ from the species tree can be common when sequence data are concatenated (Degnan & Rosenberg 2006). This is especially likely to be a problem when recent branches are long and deeper branches in the species tree are short, as occurs in parts of our tree including the branch to Pluvialis. We therefore advocate prudence until clade markers such as CR1 retroposons can be found to check controversial phylogenetic placements of genera.

Moreover, we added 14 other fossil constraints within the Charadriiformes owing to our much larger taxon sampling than in previous studies. With these constraints, the three suborders were estimated to have originated in the late Cretaceous between 79 and 102Mya (table 3 in electronic supplementary material). Fourteen ancestors of extant lineages apparently pre-date the K/T boundary 65Mya, as follows: terns plus gulls plus noddies; alcids plus skuas; coursers; pratincoles; buttonquails; sandpipers and allies plus curlews; jacanas plus painted snipes; seedsnipes plus plains wanderer; traditional plovers; oystercatchers and allies including ibisbill; Pluvialis; Egyptian plover; sheathbills and Magellanic plover; and thick-knees.

The survival of so many lineages across the K/T boundary suggests that many more shorebird lineages diversified well before the K/T boundary than assumed by the long fuse model for the evolution of shorebirds (Feduccia 2003). However, diversification of genera within the three suborders predominantly post-dates the K/T boundary. As in many ordinal clades of birds (Pereira et al. 2002; Baker et al. 2006; Tavares et al. 2006), these radiations coincide with the warming in the Eocene when ecosystems were highly productive (Jaramillo et al. 2006). We conclude that shorebirds originated earlier than was estimated previously, and they apparently were not affected by the bolide impact 65Mya. Finally, the survival of ancient lineages of shorebirds since the late Cretaceous (Cooper & Penny 1997) suggests a possible reason why a number of oddball taxa were so difficult to place phylogenetically, as they are the ancient relicts of a much earlier radiation.

A.J.B. was supported by an operating grant from the Natural Sciences and Engineering Research Council of Canada, the Royal Ontario Museum Foundation and the National Science Foundation (Assembling the Tree of Life (AToL) Program—EF-0228693). We are grateful to R. T. Brumfield and D. Dittmann at the LSU Museum of Natural Science Collection of Genetic Resources, USA and Jon Fjeldså at the Zoological Museum of the University of Copenhagen, Denmark for granting tissue loans for this research.