Enigmatic Phylogeny of Skuas: An Alternative Hypothesis Michael J. Braun; Robb T. Brumfield ST?R ? Proceedings: Biological Sciences, Vol. 265, No. 1400. (Jun. 7, 1998), pp. 995-999. Stable URL: http://links.jstor.org/sici?sici=0962-8452%2819980607%29265%3A1400%3C995%3AEPOSAA%3E2.0.CO%3B2-C Proceedings: Biological Sciences is currently published by The Royal Society. Your use of the JSTOR archive indicates your acceptance of JSTOR's Terms and Conditions of Use, available at http://www.jstor.org/about/terms.html. JSTOR's Terms and Conditions of Use provides, in part, that unless you have obtained prior permission, you may not download an entire issue of a journal or multiple copies of articles, and you may use content in the JSTOR archive only for your personal, non-commercial use. Please contact the publisher regarding any further use of this work. Publisher contact information may be obtained at http://www.jstor.org/journals/rsl.html. Each copy of any part of a JSTOR transmission must contain the same copyright notice that appears on the screen or printed page of such transmission. The JSTOR Archive is a trusted digital repository providing for long-term preservation and access to leading academic journals and scholarly literature from around the world. The Archive is supported by libraries, scholarly societies, publishers, and foundations. It is an initiative of JSTOR, a not-for-profit organization with a mission to help the scholarly community take advantage of advances in technology. For more information regarding JSTOR, please contact support@jstor.org. http://www.jstor.org Tue Aug 28 10:06:04 2007 'Th&'Th^al Socie?/ Enigmatic phylogeny of situas: an alternative hypothesis Michael J. Braun" and Robb T. Brumfield Laboratory of Molecular Systematics, National Museum of Natural History, Smithsonian Institution MRC 534, Washington, DC 20560, USA, and Department of ^oology. University of Maryland, College Park, MD 20742, USA Last year, Cohen et al. presented molecular data suggesting the surprising result that both currently recog- nized genera of skuas, Stercorarius and Catharacta (Aves: Stercorariidae), are not monophyletic. However, the most enigmatic conclusion from their analysis, that S. pomarinus is sister to C. skua, rests solely on mtDNA sequence data. When the mtDNA data are analysed in a maximum likelihood framework that accounts for variation in evolutionary rates, Catharacta monophyly cannot be rejected. None of the best trees that can be derived from two nuclear data sets of Cohen et al. support the controversial/)oman'KMi-C. skua node. We propose an alternative hypothesis, that pomarinus is sister to a monophyletic Catharacta, as the best explana- tion of the available molecular, morphological, and behavioural evidence. Keywords: Stercorariidae; skua; phylogeny; lineage sorting; mtDNA; avian systematics 1. INTRODUCTION Apparent conflicts between molecules and morphology in systematics have become almost commonplace (Baiter 1997), sparking heated debate in many cases (e.g. Poe 1996; Lee et al. 1997; Sullivan & Swofltord 1997). This debate is healthy because it stimulates additional research and new analyses that are generating fresh insights in a broad array of evolutionary disciplines. A recent case in point involves the skuas, a distinctive family of predatory seabirds. Cohen et al. (1997) present molecular data bearing on skua phylogeny that pose a striking enigma. The data indicate that one of the three Stercorarius skuas, S. pomarinus (the Pomarine skua), falls within the clade comprising the Catharacta skuas, rendering both genera non-monophyletic (figure la). Yet these two genera are (i) quite different in outward appearance, (ii) defined by putative morphological synapomorphies (Brooke 1978; Furness 1987), and (iii) have been maintained as distinct genera in the over- whelming majority of modern treatises (e.g. AOU 1983). Cohen et al. (1997) suggest three possible explanations for the remarkable discordance between molecules and morphology in skuas, but recognize that all three explana- tions are 'far-fetched'. Our purpose here is to explore this apparent paradox, and to propose a fourth hypothesis. We believe this fourth explanation best accounts for the available data, which now include plumage and skeletal morphology, behaviour, ectoparasitic lice, and a variety of molecular evidence. This explanation has the further advantage that it is readily testable. Cohen et al. (1997) present five data sets, four of which are molecular, the fifth of which is based on ectoparasitic lice. The data are consistent in many ways, and demon- strate convincingly that pomarinus is more closely related 'Author for correspondence (braun@onyx. si . edu). Proc. R. Soc. Land. B (1998) 265, 995-999 Received 26 January 1998 Accepted 3 March 1998 to Catharacta than to other Stercorarius. While this is surprising in the light of the general resemblance of pomarinus to other Stercorarius, this result does not actually contradict the morphological data. As Cohen et al. point out, the Stercorarius morphotype appears to be ancestral in the Stercorariidae; therefore, it does not provide cladistic information supporting monophyly of Stercorarius. The single putative morphological synapomorphy linking pomarinus with Stercorarius, barred juvenile plumage (Brooke 1978), may instead be a symplesiomorphy in the family that has been lost in Catharacta. What is more difficult to accept is the idea that Catharacta is also not monophyletic. The phylogeny presented by Cohen et al. places pomarinus sister to C skua (Great skua), to the exclusion of the other five Catharacta species (figure la). This aspect of their phylogeny forces the authors to propose remarkable convergence in the origins of either the Catharacta or Stercorarius morphology (their hypotheses (a) and [b)) or a bizarre inter-generic hybridization event resulting in a stable hybrid species (pomarinus) that is Catharacta-\i\it in its mitochondrial and nuclear genomes but Stercorarius-like in its external appear- ance (their hypothesis (c)). It is this node which causes the bulk of the enigma they face. Although Cohen et al. discuss five data sets, the phylogeny they present is actually derived from only one, the mitochondrial DNA (mtDNA) sequences. The mtDNA is a single, clonally inherited genetic unit. Any phylogeny based upon it must be regarded as the phylogeny of a single gene, which may or may not accurately track the species phylogeny (Nei 1987, p. 288; Maddison 1997). One set of conditions under which a gene phylogeny is likely to diSer from the species phylogeny is when a series of speciation events occurs with insufficient time between them for the gene lineages to reach reciprocal monophyly. In this situation, random fixation of gene lineages in the daughter species can result in a gene tree that is incongruent with the 995 ? 1998 us Government 996 M. J. Braun and R. T. Brumfield Reanalysis of skua phytogeny (a) ri 5 other Catharacta IS. pomarinus IC. skua =1S. ?ongicaudus z? S. parasltlcus ib) all Catharacta ] S. pomarinus 1 S. Iongicaudus ^ S. paras?ticus Figure 1. Hypotheses of phylo- genetic relationship among skuas. Heavy black branches are those on which a Catharacta- like plumage morphology can be inferred. Open branches are those on which a Stercorarius- like plumage morphology can be inferred. Branch lengths are not proportional to evolu- tionary change, [a] Major features of the topology proposed by Cohen etal. (1997) based on mtDNA sequences. Either the Stercorarius or the Catharacta morphology must have evolved twice on this tree. (?) Alternative topology proposed herein. Each morphology need only evolve once on this tree. species tree. The small divergences among Catharacta species a.n? pomarinus in all four genetic data sets (see Cohen 1997; note branch lengths in figure 2 herein) indicate that specia- tion has been rapid and relatively recent in skuas. Thus, it seems plausible that reciprocal monophyly of mtDNA may not have been reached in some lineages of the group, and that the mtDNA tree, however well-resolved it might be, simply is not the same as the species tree. We propose an alternative hypothesis of relationship that accommodates both the morphology and the data of Cohen et al. (figure le). In this hypothesis, pomarinus is sister to a monophyletic genus Catharacta, reflecting the close relation- ship so strongly indicated by a wealth of molecular evidence. The pomarinus-C skua node of figure la is presumed incorrect, either due to the gene tree-species tree problem discussed above or to inadequate resolution of the mtDNA tree (see below). Although Stercorarius is not monophyletic, figure \b only requires each morphotype to evolve once (contrary to hypotheses (a) and (b) of Cohen et ai; see figure la). The resemblance o? pomarinus to other Stercorarius is explained by assuming that pomarinus has retained the ancestral Stercorarius morphotype. This hypothesis also accounts for the similarity oipomarinus to Catharacta observed in a phenetic study of 50 skeletal characters (Schnell 1970), as well as behavioural similari- ties in calls and displays (Andersson 1973). In fact, Andersson seems to have proposed essentially the same hypothesis of relationship entailed in figure l? when he wrote 'the most likely explanation for this [behavioural similarity] seems to be that the Pomarine and Great skuas diverged from each other at a time when the predecessor of the two smaller species had already branched from the common skua ancestor' (Andersson 1973, p. 14). Although he did not explicitly state that he considered Catharacta monophyletic, Andersson treated the South Polar skua (C. s. maccormicki) as a subspecies of the Great skua (C s. skua) in the same section, so the idea seems implicit. 2. METHODS, RESULTS AND DISCUSSION (a) Mitochondrial data Given the two competing hypotheses of relationship contained in figures \a and \b, it is natural to ask whether the mtDNA sequence data support one hypothesis significantly more than Proc. R. Soc. Land. B (1998) Reanalysis qfskuaphylogeny M. J. Braun and R. T. Brumfield 997 Larus canus 0.1 ENSS 62 C. ant?rctica C. chilensis C. hamiltoni C. lonnbergi C. maccormicki r C. skua S. pomarinus 58 99 S. longicaudus S. parasiticus Figure 2. Maximum-likelihood estimate of skua phylogeny based on mtDNA sequence data of Cohen et al. (1997). Numbers refer to percentage of 100 unconstrained likelihood bootstrap replicates in which each node occurs. Only values greater than 50% are shown. Branch lengths are proportional to the expected number of substitutions per site (ENSS). Likelihood parameters for the general time reversible (GTR) substitution model (Lanave et al. 1984) with among-site rate heterogeneity following a gamma distribution (F; Yang 1993) were estimated separately on the cytochrome b (cyt b) and 12S ribosomal RNA (12S) data. The GTR+F likelihood model was used because it fit the data significantly better based on likelihood ratio tests (Goldman 1993) than other models available in PAUP* (e.g. GTR+F versus HKY85+F, x =21.02, d.(. = 5,p<0.05). Because the estimated parameters (i.e. rate matrix and gamma-shape parameter) were very similar for both genes, cyt b and 12S sequences were combined. Using all 1415 bp of sequence data, likelihood parameters for the GTR+F model were estimated using the successive approximations approach suggested by Swofford et al. (1996, p. 445), beginning with a neighbour-joining tree (Saitou & Nei 1987) constructed from HKY85 (Hasegawa et al. 1995) genetic distances assuming no among- site rate heterogeneity. The parameter estimates used were rate matrix: A-C = 2.329 X 10?, A-G= 1.826 x 10^ A-T = 8.528 x 10^, C-G = 7.613 x lO'^, C-T = 3.455x 10', G-T=l; gamma-shape parameter, a = 0.0974665. the other. The advantages of model-based maximum-likelihood methods of phylogenetic analysis in this regard have been discussed (see Swofford et al. 1996; Huelsenbeck & Rannala 1997 and references therein). To assess whether the mitochondrial sequence data are significantly more likely under the topology of figure la than \h, we performed a Kishino?Hasegawa test (KH test; Kishino & Hasegawa 1989) on maximum-likelihood trees with these topologies. Unless otherwise specified, all phylogenetic analyses were performed with PAUP* v. 4.0 (Swofford 1998). An unconstrained branch-and-bound search found a most likely tree (log-likelihood score =? 3259.52; figure 2) identical in topology to the maximum-parsimony tree of Cohen et al. Using the same model and parameter estimates, a branch-and-bound search in which Catharacta was constrained to be monophyletic also found a most likely tree (score = ? 3262.40). A KH test indicates that this tree is not significantly worse than the most likely tree from the unconstrained search as an explanation of the data (7'=0.8883, ji) = 0.3745). One hundred unconstrained, likelihood bootstrap replicates using the above parameters (with heuristic search, addseq = as-is) also demonstrate less confidence in the enigrasLtic pomarinus-C. skua node than was found by Cohen et al. in an unweighted parsimony framework (58% versus 97%; figure 2). Tj understand why this node is less robust under maximum likelihood, it is important to note that the branch lengths in the pomarinus?Catharacta clade are all quite short. This results from the fact that the maximum-likelihood model accounts for the substantial variation among sites in evolutionary rates present in this data set (see figure 2 legend). Although we believe model-based likelihood methods are better suited for phylogenetic analysis of sequence data, in the interest of completeness we also performed KH and Templeton (1983) tests on optimal trees from unconstrained and constrained unweighted parsimony analyses analogous to those presented by Cohen et al. Our analysis of 102 informative sites resulted in a single most parsimonious tree (145 steps) identical in topology to that presented by Cohen et al. Discrepancies in the number of informative sites and tree length from that presented in Cohen et al. (109 informative sites, tree length 148) are due to an improve- ment in the ability of PAUP* to detect uninformative characters in certain situations involving polymorphic terminal taxa. This improvement does not affect their phylogenetic conclusions. A search in which Catharacta was constrained monophyletic resulted Proc. R. Soc. Land. B (1998) 998 M. J. Braun and R. T. Brumfield Reanalysis qfskuaphylogeny in a single most parsimonious tree (151 steps) that placed pomar- inus basal to the Catharacta clade. The KH test found the unconstrained tree to be better than the constrained tree (T=2.1591, /) = 0.0332). There was no significant difference as judged by theTempleton test ^ = 1.8904, n.s.). (b) Nuclear data Thus, maximum-likelihood methods indicate that the tree of figure \a is not a significantly better explanation of the mtDNA data than the tree of figure \b, and parsimony methods indicate that it is at best marginally so. However, even if there was a significant difference, the gene tree-species tree problem would remain. Whether a particular gene tree is congruent with the species tree can best be tested by independent estimates of the species tree based on other data (i.e. other genes, morphology, behaviour, etc.). Cohen et al. describe two nuclear data sets (allo- zymes, RAPDs) that can be used for this purpose. However, they do not present trees based on these data, concluding instead that the nuclear gene data strongly confirm the close relationship of pomarinus to Catharacta, but 'provide no critical evidence about the ancestry of the Pomarine skua . . . '. Reanalysis of the nuclear data sets reveals that the best trees that can be derived from either of them supports the close rela- tionship 0?pomarinus to Catharacta, but contradicts the pomarinus- C. skua node of the mtDNA tree. That this might be true can be seen by inspection of the distance matrices (Cohen 1997). The RAPD distance matrix indicates that C skua and C maccormicki are more similar to each other than either is to pomarinus i^kua and maccormicki axe the only Catharacta taxa in the RAPD data matrix). Minimum evolution (Kidd & Sgaramella-Zonta 1971) and FM (Fitch & Margoliash 1967) analyses confirm this simi- larity, yielding a tree with Catharacta monophyletic. Parsimony analysis of the RAPD data yields four most parsimonious trees, all of which have C. maccormicki sister to pomarinus. The shortest trees (90 steps) are only one step shorter than a tree in which Catharacta is monophyletic. Thus, neither distance nor parsimony trees derived from the RAPD data contain z. pomarinus-C. skua node. Using BIOSYS (Swofford &. Selander 1981), distance matrices (Cavalli-Sforza & Edwards 1967; Nei 1972) were calculated from the original 42-locus allozyme data set provided by A. Baker (including al?ele frequency data for the outgroups Larus fuscus, L. novaehollandiae and Sterna hirundo). In each case, C. skua and C. ant?rctica axe more similar to one another than either is to pomarinus (skua, ant?rctica and maccormicki are the only Catharacta taxa in the allozyme data matrix). In fact, one al?ele at the guanine deaminase (GDA) locus represents an unambiguous synapomorphy for C. skua and C ant?rctica. Given that the mtDNA data are also derived from a single genetic unit, we consider GDA by itself to be an important conflict for the mtDNA tree of figure \a. A frequency parsimony analysis of the allozyme data was performed by converting the BIOSYS file to a FREQPARS file using the FORTRAN program BI02FREQ, (Swofford & Berlocher 1987; program available via anonymous FTP at onyx, si . edu). The FREQPARS file was imported into PAUP* and a branch-and-bound search was performed (for details of the FREQPARS analysis utilizing PAUP*, see Berlocher & Swofford 1997). The three most parsimonious FRE??ARS trees (49.916 steps) have C skua sister to C. ant?rctica. A tree with Catharacta constrained monophyletic is only 1.1 steps longer (a FREQPARS step is equivalent to an allelic frequency change of 0.5). UPGMA (unweighted pair group method), neighbour-joining, and minimum evolution trees were also constructed using the chord distance (Cavalli-Sforza & Edwards 1967); all of these also have C. skua sister to C. ant?rctica. 3. CONCLUSIONS (a) Phytogeny In summary, none of the best trees derived from the nuclear data sets includes the pomarinus-C. skua node found in the mtDNA tree, and near optimal trees exist in each case in which Catharacta is monophyletic. While neither nuclear data set can be used to exclude confidently one or other of the topologies in question, the same can probably be said of the mtDNA data as explored above. Given the conflict between data sets and the gene tree- species tree problem, a more conservative interpretation of the total genetic evidence (mitochondrial and nuclear) would be that it supports a clade composed of pomarinus plus all Catharacta, but no further resolution within that clade is yet possible. When the several apparent morpho- logical synapomorphies that unite Catharacta are considered (Furness 1987), it seems more reasonable to suppose that the group is monophyletic (figure lb). Distinguishing between these two hypotheses of rela- tionship (figures la and le) should be straightforward. Any independent estimate of the species tree that resolves the nodes within the pomarinus-Catharacta clade will corroborate one of these hypotheses (or one of the other possible resolutions within the clade), and refute the rest. In principle, it would be possible to make such an estimate from morphological or behavioural data, if such infor- mation exists for all the forms of Catharacta. However, given that genetic samples from all named forms of Catharacta and Stercorarius are in hand in several labora- tories, it seems most likely that independent estimates of the species tree will come from DNA sequences of nuclear genes. (b) Taxonomy The phylogeny in figure lb makes Stercorarius non- monophyletic, while that of figure la renders both Stercorarius and Catharacta non-monophyletic. In either case, a different generic treatment of the group is required. One possibility is to treat all skuas in a single genus, Stercorarius, as recommended by Hartert (1912), Moynihan (1959) and Andersson (1973). However, if the topology of figure lb proves correct, it would also be reasonable to retain Stercorarius and Catharacta, and place pomarinus in a separate genus. This treatment would have the advantage of recognizing the morphological distinctiveness that separates pomarinus from Catharacta, while highlighting the true phylogenetic structure of the group. The disadvantage of this arrangement is that it would erect separate genera for two groups (pomarinus and Catharacta) which, based on their genetic similarity, probably share a more recent common ancestor than do the two remaining species of Stercorarius. Accumulation of genetic and palaeontological data may make recency of common ancestry a desirable metric of categorical rank in future. At present, there is only a loose correlation between the two in avian taxonomy (e.g. Brumfield et al. 1997). The dual goals of recognizing phylogeny and morphological similarity take priority in current usage, and this second treatment is the Proc. R. Soc. Land. B (1998) Reanalysisofskuaphylogeny M. J. Braun and R. T. Brumfield 999 one we favour if figure \b is the correct phylogeny. The generic name with priority for pomarinus under this scenario seems to be Coprotheres Reichenbach 1850. We thank A. J. Baker, B. L. Cohen, P. Harvey, A. J. Helbig, J. P. Huelsenbeck, S. Steppan, and D. L. Swofford for helpful discussion and criticism of the manuscript. R. C. Banks advised us on the historical taxonomy and nomenclature of the Stercorariidae. REFERENCES AOU 1983 Check-list of North American birds, 6th edn. Washington, DC: American Ornithologists' Union. Andersson, M. 1973 Behaviour of the Pomarine Skua, Stercorarius pomarinus T?mm. with comparative remarks on Stercorariinae. Omis Scand. 4, 1-16. Baiter, M. 1997 Morphologists learn to live with molecular upstarts. Science 276, 1032-1034. Berlocher, S. H. & Swofford, D. L. 1997 Searching for phyloge- netic trees under the frequency parsimony criterion: an approximation using generalized parsimony. Syst. Biol. 46, 211-215. Brooke, R. K. 1978 The Catharacta skuas (Aves: Laridae) occur- ring in South African waters. Durban Mus. Nov. 11, 295-308. Brumfield, R. T, Swofford, D. L. & Braun, M. J. 1997 Evolutionary relationships among the pot?os (Nyctibiidae) based on isozymes. Ornithol. Monogr. 48, 129-145. Cavalli-Sforza, L. L. & Edwards, A. W F. 1967 Phylogenetic analysis: models and estimation procedures. Evolution 32, 550-570. Cohen, B. L. 1997 Enigmatic phylogeny of skuas (Aves: Stercorariidae) (This is available at http://www.ibls.- gla.ac.uk/ibls/staff/bl-cohen/skuas.html. (October 22, 1997.) Cohen, B. L. (and 15 others) 1997 Enigmatic phylogeny of skuas (Aves: Stercorariidae). Proc. R. Soc. Loud B264, 181-190. Fitch, W. M. SL Margoliash, E. 1967 Construction of phylogenetic trees. Science 155, 279-284. Furness, R. W 1987 The skuas. Staffordshire, England: T & A. D. Poyser Ltd. Goldman, N. 1993 Statistical tests of models of DNA substitution. J. Molec. Evol. 36, 182-198. Hartert, E. 1912 Die Vogel derpal?arktischen Fauna. Berlin. Hasegawa, M., Kishino, H. &, Yano, T 1985 Dating of the human-ape splitting by a molecular clock of mitochondrial DNA. J. Molec. Evol. 21, 160-174. Huelsenbeck, J. P. & Rannala, B. 1997 Phylogenetic methods come of age: testing hypotheses in an evolutionary context. Science 276, 227. Kidd, K. K. & Sgaramella-Zonta, L. A. 1971 Phylogenetic analysis: concepts and methods. Am. J. Hum. Genetics 23, 235-252. Kishino, H. & Hasegawa, M. 1989 Evaluation of the maximum- likelihood estimate of the evolutionary tree topologies from DNA sequence data, and the branching order in Hominoidea. J. Molec. Evol. 29, 170-179. Lanave, C, Preparata, G., Saccone, C. &L Serio, G. 1984 A new method for calculating evolutionary substitution rates. J. Molec. Evol. 20, 86-93. Lee, K., Feinstein, J. & Cracraft, J. 1997 The phylogeny of ratite birds: resolving conflicts between molecular and morpholo- gical data sets. In Avian molecular evolution and systematics (ed. D. Mindell), pp. 173-208. San Diego, CA: Academic Press. Maddison, W P. 1997 Gene trees in species trees. Syst. Biol. 46, 523-536. Moynihan, M. 1959 A revision of the family Laridae (Aves). Am. Mus. Novitates 1928, 1-42. Nei, M. 1972 Genetic distance between populations. Am. Nat. 106, 283-292. Nei, M. 1987 Molecular evolutionary genetics. New 'Vbrk: Columbia University Press. Poe, S. 1996 Data set incongruence and the phylogeny of the crocodilians. Syst. Biol. 45, 393-414. Saitou, N. %L Nei, M. 1987 The neighbor-joining method: a new method for reconstructing phylogenetic trees. Molec. Biol. Evol. 4, 406-425. Schnell, G. D. 1970 A phenetic study of the suborder Lari (Aves), parts I-n. Syst. Zool 19, 35-57, 264-302. Sullivan, J. & Swofford, D. L. 1997 Are guinea-pigs rodents? The importance of adequate models in molecular phylogenetics. J. Mamm. Evol. 4, 77-86. Swofford, D. L. 1998 PAUP*, Beta test versions 56, 57, 59. Sunderland, MA: Sinauer Associates. Swofford, D. L. & Berlocher, S. H. 1987 Inferring evolutionary trees from gene frequencies under the principle of maximum parsimony. Syst. ^ool. 36, 293-325. Swoflbrd, D. L., Olsen, G. J.,Wadden, P J. & HiUis, D. M. 1996 Phylogenetic inference. In Molecular systematics (eds D. M. Hillis, C. Moritz & B. K. Mable), pp. 407-514. Sunderland, MA: Sinauer Associates. Swoflbrd, D. L. & Selander, R. B. 1981 BIOSYS-1: a FORTRAN program for the comprehensive analysis of electrophoretic data in population genetics and systematics. J. Hered. 72, 281-283. Templeton, A. R. 1983 Phylogenetic inference from restriction endonuclease cleavage site maps with particular reference to the humans and apes. Evolution 37, 221-244. Yang, Z. 1993 Maximum likelihood estimation of phylogeny from DNA sequences when substitution rates differ over sites. Molec. Biol. Evol. 10, 1396-1401. Proc R. Soc. Land. B (1998)