Molecular Phylogenetics and Evolution Vol. 11, No. 1, February, pp. 84–94, 1999 Article ID mpev.1998.0555, available online at http://www.idealibrary.com on

Molecular Evolutionary Relationships in the Avian Genus Anthus (: ) Gary Voelker1 Burke Museum and Department of Zoology, Box 353010, University of Washington, Seattle, Washington 98195

Received December 1, 1997; revised April 8, 1998

habitats, and this diversity along with distributional Nucleotide sequences for 1035 bp of the mitochon- differences have contributed to considerable variation drial cytochrome b gene were used to determine the in a number of life history traits such as migratory molecular evolutionary relationships of species in the behavior, clutch size, and molt pattern (Hall, 1961; cosmopolitan avian genus Anthus. Phylogenetic analy- Clancey, 1990; Voelker, unpublished data). Yet these sis of these mtDNA sequences supported four major differences in life history have received little attention clades within the genus: (1) the small-bodied African from a phylogenetic standpoint. pipits, (2) a largely Palearctic clade, (3) a largely South The striking similarity in morphology and plumage American clade, and (4) an African–Eurasian–Austra- across Anthus has historically been a barrier to resolv- lian clade. Anthus hellmayri, A. correndera, and A. rubescens are shown to be paraphyletic. The possibil- ing relationships within the genus (Ridgway, 1904; ity of paraphyly within A. similis is instead inferred to Hall, 1961; Clancey, 1990), although the similarity has be the discovery of a new species and supported by not precluded the description of numerous subspecies reference to the museum voucher specimen. Sequence (e.g., Hellmayr, 1935; Clancey, 1990 and references divergence suggests a Pliocene/Miocene origin for the therein). Previous works have speculated to a limited genus. Although Anthus cytochrome b is found not to extent on the relationships of a few species (e.g., Hall, be behaving in a clocklike fashion across all taxa, 1961; Clancey, 1990), relying primarily on limited speciation during the Pleistocene epoch can be reason- variation across morphological characters. A good gen- ably inferred for the 66% of sister pairs that are eral description of an Anthus species would be: plum- diverging in a clocklike manner. Base compositions at age color generally brown or some combination of each codon position are similar to those found across a brown, white, and black, with black ventral streaking, growing number of avian lineages. The resulting phylo- a long hindclaw, and weighing between 18 and 30 g. A genetic hypothesis is compared to previous hypoth- few species do have red (e.g., roseatus Rosy ), green eses of Anthus relationships, all of which deal with (e.g., hodgsoni Olive-backed Pipit), or yellow color relationships of a particular species or a particular patches (e.g., crenatus Yellow-tufted Pipit), and the species complex; roughly half of these previous hypoth- body mass of several species are smaller or larger, but eses are supported. ௠ 1999 Academic Press deviations from this general description are limited; thus, reliable means to determine relationships have INTRODUCTION been lacking. Only recently have attempts been made to unravel the phylogenetic relationships of any pipits With approximately 40 species, the avian genus using ecological and molecular differences (Zink et al., Anthus (pipits; Motacillidae) is one of the more speciose 1995; Arctander et al., 1996; Foggo et al., 1997). For genera of oscine . Anthus is globally distrib- example, Nazarenko (1978) defined habitat differences uted, being found on every continent except Antarctica between two species within the ‘‘’’ complex (Sibley and Monroe, 1990). Individual species distribu- (spinoletta (Water Pipit) versus rubescens (American tions range from island endemics (e.g., antarcticus on Pipit)), while Knox (1988) summarized ecological and South Georgia Island) to continental endemics (e.g., behavioral differences supporting recognition of three lineiventris from Africa) to intercontinental migrants species within the same species complex (spinoletta, (e.g., trivialis Eurasia to Africa) to species breeding on rubescens, and petrosus (Rock Pipit)). Using molecular multiple continents (e.g., cervinus in Eurasia and North data, Zink et al. (1995) found a high level of divergence America). There is substantial diversity in breeding between Siberian and American forms of rubescens, Foggo et al. (1997) found large differences between 1 Present address: Barrick Museum, Box 454012, University of populations of novaeseelandiae (‘‘New Zealand’’ Pipit) Nevada Las Vegas, Las Vegas, NV 89154. from New Zealand and nearby islands, and Arctander

84 1055-7903/99 $30.00 Copyright ௠ 1999 by Academic Press All rights of reproduction in any form reserved. Anthus SYSTEMATICS 85 et al. (1996) addressed the relationships of berthelotii considered here was amplified via PCR as a single unit (which is endemic to the Canary and Madeira Islands). using primers L14841 (Kocher et al., 1989) and H16065 While each of these studies has focused on a limited, (Helm-Bychowski and Cracraft, 1993). These two prim- and in some cases historically contentious, problem of ers were also used in various combinations along with pipit relationships, a more general understanding of L15114, L15299, L15609, H15547 (Edwards et al., pipit systematics is not available. 1991), and H15299 (Hackett, 1996) to amplify DNA In this paper, a segment of mitochondrial cytochrome from several taxa in overlapping segments. For sequenc- b sequence was obtained from 41 Anthus individuals ing, I used the above primers, H15915 (Edwards representing 35 previously recognized species, as well and Wilson, 1990), and the following primers de- as from 5 previously determined outgroup taxa. I use signed specifically from Anthus sequences (numbers these data to generate phylogenies, compare these correspond to 3Ј position in domestic fowl sequence results to previous studies, and provide a framework (Dejardins and Morais, 1990) ): L15086 (5Ј-CTCTG- for future studies of life history strategies and biogeog- TAGCTCACATATGCC-3Ј), L15376 (5Ј-CTAGCAGAAT- raphy across the genus. GAGCCTGAGG-3Ј), L15616 (5Ј-GTTGCCCTAACCC- TATTCTC-3Ј), L15811 (5Ј-CCCCTACTCCA-CACATCA- Ј Ј Ј MATERIALS AND METHODS AA-3 ), H15345 (5 -GTAATAACGGTAGCTCCTCA-3 ), H15671 (5Ј-GGTGTGAAGTTTTCTGGGTC-3Ј), H15853 I sequenced 1035 bp of the mitochondrial cytochrome (5Ј-GGCGGAAGGTTATTGATC-3Ј). I also used L15350 b gene from the 35 Anthus species available in genetic (5Ј-TTACAAACCTATTCTCAGC-3Ј), designed by J. resources collections. This number represents over 80% Klicka. of the species recognized by Sibley and Monroe (1990) Fragments were amplified in 50- or 100-µL PCRs; and includes the recently described longicaudatus amplification conditions were 30 s at 94°C, 30 s at 50°C, (Liversidge, 1996). Multiple specimens from 6 species and 30 s at 72°C, for 35 cycles. The exception was A. were included to provide a preliminary insight into sokokensis which was annealed at 61°C due to multiple species monophyly in cases where the species has a bands at lower temperatures. Amplified products were disjunct breeding range (similis, hellmayri, correndera) prepared as templates for automated sequencing by or has been previously postulated to be polytypic based purification and concentration in up to 22-µL of water on other evidence (spinoletta/rubescens/petrosus com- after three passes through Ultrafree-MC filters (Milli- plex). I did not include the Yellow-breasted Pipit (chlo- pore) by centrifugation. Two microliters of the purified ris), a species which has been placed in the genus and concentrated PCR product were used as a template Hemimacronyx (Cooper, 1985; Voelker and Edwards, in a 10-µL DyeDeoxy Cycle Sequencing reaction (ABI), 1998). As outgroups, I use three members of the genus along with one of the above primers, as per manufactur- Motacilla (confamilial with Anthus). I also include an er’s instructions (Perkin–Elmer). After cycle sequenc- accentor (Prunellidae) and an Old World sparrow (Pas- ing, products were placed on coarse-grained Sephadex seridae) suggested as close relatives of pipits and columns and cleaned of excess nucleotides via centrifu- based on molecular evidence (Sibley and Ahl- gal passage through the columns (1500 rpm for 7 min). quist, 1981, 1990; Voelker and Edwards, 1998). Both light and heavy strands of the entire 1035-bp fragment considered were sequenced using flourescent Cytochrome b Isolation, Amplification, and Sequencing dye-deoxy chemistry on an ABI Model 373 automated From 1993 to 1995 I collected Anthus specimens in sequencer. Sequences were aligned unambiguously by Russia, Kazakhstan, South Africa, and Argentina. All eye using Genetic Data Environment (developed and specimens were prepared in the field as either skin/wing/ maintained by S. Smith, with compilation of programs tissue or skeleton/wing/tissue combinations; the tissue by various authors; available free from ftp.bio.indiana. samples were preserved in liquid nitrogen. Frozen edu, in molbio/unix/GDE) and have been submitted to tissues were available for all University of Washington GenBank (pending). Burke Museum specimens (except A. roseatus and A. spragueii) (Table 1), and mtDNA was isolated from Phylogenetic Analysis and Test of Molecular Clock these specimens via a cesium chloride gradient (Dowl- All analyses were performed using PAUP* (written ing et al., 1990), which minimizes the chance of amplify- by David L. Swofford; results published with permis- ing nuclear mitochondrial sequences. Tissues or blood sion), and were rooted to Passeridae. In general, I samples of all specimens from other institutions (Ta- followed the methods described in Voelker and Ed- ble 1) had been stored in buffer, and total genomic DNA wards (1998) for obtaining an estimate of phylogeny, was obtained by Chelex extraction (Walsh et al., 1991), except that all data parameters were estimated in with the exception of A. sokokensis which was prepared PAUP*. Briefly, I generated an initial neighbor-joining via phenol/chloroform extraction and ethanol precipita- (NJ; Saitou and Nei, 1987) tree in PAUP* using the tion. HKY85 model of DNA evolution which allows for two For most samples, the segment of cytochrome b substitution rates (transitions vs transversions) as well 86 GARY VOELKER

TABLE 1

Species, Specimen Numbers,a,b and Collecting Localities for Anthus Specimens Examined

Species Specimen number Locality cinnamomeus UWBM 52816 South Africa: E. Cape Prov., Umtata, 90 km N, 50 km W richardi UWBM 51732 Russia: Irkutskaya OblastЈ,UstЈ-Ordynskiy, 3 km S, 4 km W rufulus FMNH 358348 Philippines: Sibuyan Island; Tampayan. Magdiwang, 1 km S, 1.25 km E novaeseelandiae UWBM 57589 Australia: New South Wales, near Mungo National Park hoeschi UWBM 53185 South Africa: E. Cape Prov., Umtata, 90 km N, 50 km E leucophrys UWBM 53171 South Africa: KwaZulu/Natal Prov., Melmoth, 2 km N, 4 km E vaalensis UWBM 53217 South Africa: PWV Prov., 25°28ЈS 28°30ЈE melindae NMK 639 10821 Kenya: Malindi campestris COP 01077 Denmark: Jylland, Skagen godlewskii UWBM 46342 Russia: Buryatia Republic, Ulan-Ude, 30 km S, 55 km W berthelotii COP 03169 Islas Canarias, Ihlas Selvagens similis (2) UWBM 52758 South Africa: Orange Free State, Springfontein, 26 km S, 26 km W MM B2548 South Africa: Cape Prov., Kimberley nyassae K. Hustler #5 Zambia: Choma brachyurus UWBM 52901 South Africa: KwaZulu/Natal Prov., Melmoth, 2 km N, 4 km E caffer UWBM 52810 South Africa: KwaZulu/Natal Prov., Melmoth, 2 km N, 4 km E trivialis UWBM 46595 Russia: Gorno-Altay Republic, Gorno-Altaysk, 95 km S, 20 km W hodgsoni UWBM 46624 Russia: Buryatia Republic, Ust-Barguzin, 15 km S, 30 km W gustavi UWBM 43951 Russia: Chukotka Republic, Anadyr, 68 km S, 36 km E pratensis UWBM 49705 Russia: Murmanskaya OblastЈ, Teriberka, 20 km S, 10 km W cervinus UWBM 49692 Russia: Murmanskaya OblastЈ, Teriberka, 20 km S, 10 km W roseatus UWBM 51485 : Rupchet, Betrabati, 11 km N, 7 km W spinoletta blakistoni (2) UWBM 46609 Russia: Buryatia Republic, Barguzin, 30 km N, 17 km E UWBM 51827 Russia: Irkutskaya OblastЈ, Slyudyanka, 15 km S, 8 km W petrosus COP 01104 Denmark: Jylland, Blaavand rubescens alticola UWBM 53403 Colorado: Gunnison Co., 0.5 mi N of Cottonwood Pass rubescens japonicus UWBM 47533 Russia: Khabarovskiy Kray, Badzhal’skiy Khrebet sylvanus AMNH 831297 Nepal: Kipsung, Betribati, 4 km N, 4 km W spragueii ANS 183670 Missouri: Buchanan Co., St. Joseph, Rosecrans Airport furcatus UWBM 54556 Argentina: Prov. Tucuma´n, San Miguel de Tucuma´n,10kmS,45kmW hellmayri hellmayri UWBM 54528 Argentina: Prov. Tucuma´n, San Miguel de Tucuma´n, 7 km N, 60 km W hellmayri brasiliensis UWBM 54574 Argentina: Prov. Corrientes, Corrientes, 55 km S, 5 km E lutescens UWBM 54500 Argentina: Prov. Tucuma´n, San Miguel de Tucuma´n,20kmN correndera correndera UWBM 54553 Argentina: Prov. Tucuma´n, San Miguel de Tucuma´n,10kmS,45kmW correndera catamarcae UWBM 54511 Argentina: Prov. Tucuma´n, San Miguel de Tucuma´n,20kmN bogotensis UWBM 54521 Argentina: Prov. Tucuma´n, San Miguel de Tucuma´n, 5 km N, 50 km W antarcticus BAS 069 South Georgia, Island, Freshwater Bay sokokensis COP 01098 Kenya: Arabuko-Sokoke, 3.30° S, 39.50° E crenatus UWBM 52780 South Africa: Orange Free State, Harrismith, 50 km N, 30 km E lineiventris UWBM 52809 South Africa: KwaZulu/Natal Prov., Melmoth longicaudatus MM B2550 South Africa: Cape Prov., Kimberley

a See Acknowledgments for tissue sources. b All numbers refer to voucher specimens except COP and BAS numbers. as unequal base frequencies (Hasegawa et al., 1985). the possibility of achieving only local optima. I then On this initial tree, ␣ (among-site rate variation) and ␬ reestimated ␣ and ␬ on this tree, conducted a new (transition bias) parameters (see Wakeley, 1993; Wake- heuristic ML search, and repeated this process for a ley, 1996; Yang, 1996a) were determined simulta- third time. On the final analysis, over 25,000 rearrange- neously for each codon position and for the entire data ments were performed; again the log-likelihood value set. did not change after the first few thousand rearrange- I used these parameters to employ two different ments, and the tree topology had not changed from the strategies for maximum likelihood (ML) analyses. First, previous heuristic search. Second, I conducted ML I used the overall ␣ and ␬ values estimated on the analyses using the initial NJ as the starting tree. I initial NJ tree to conduct a heuristic ML search, using again used TBR, and followed the methodology de- the TBR swapping option. This search was stopped scribed above, but always used the new topology in after 4966 rearrangements were performed; the log- subsequent analyses. This process was repeated until likelihood value had not decreased for over 3000 rear- the tree topology stabilized. Due to the large number of rangements, although this analysis still leaves open taxa involved, it was logistically impractical to run an Anthus SYSTEMATICS 87

TABLE 2

Overall and Codon Specific Dynamics of the Cytochrome b Gene Estimated on the ML Tree Both Across All Taxa, and with Anthus Alone

Dynamic First position Second position Third position All positions

Mean base compositiona All taxa (24, 24, 22, 30) (13, 20, 42, 25) (3, 39, 10, 48) (13, 28, 25, 34) Within Anthus (24, 24, 22, 30) (13, 20, 42, 25) (3, 38, 10, 49) (13, 27, 25, 35) ␣ All taxa 0.351 0.289 1.952 0.360 Within Anthus 0.355 0.255 1.890 0.359 ␬ All taxa 2.932 0.914 22.734 5.309 Within Anthus 2.845 1.359 24.239 5.507

a Base composition is presented as (%G, %A, %T, %C) averaged over all sequences.

ML analysis until termination. However, these method- rable to most other empirical values found across ologies do result in good estimates of phylogeny (J. diverse avian lineages; thus, compositional bias is Felsenstein, pers. comm, J. Huelsenbeck, pers. comm.). similar as well. Little bias in nucleotide composition is To test whether a molecular clock is operating within found at first positions, whereas second positions show Anthus, I enforced a molecular clock on the ML phylog- a high percentage of T’s. Third codon positions show an eny using PAUP*, and compared this tree to the ML extreme lack of G and to a lesser extent T nucleotides, a tree (i.e., comparison of branch lengths) via the Kishino– finding that is consistent among avian and mammalian Hasegawa (1989) test. I also used the two-cluster test lineages (e.g., Irwin, 1991; Nunn and Cracraft, 1996; (Takezaki et al., 1995), which examines whether a Voelker and Edwards, 1998). Uncorrected sequence molecular clock is operating along two daughter lin- divergence ranged from 8.88 to 19.23% between out- eages created by an interior node of a tree. I applied groups and Anthus. Sequence divergence within An- this test, using the ML topology and Tajima–Nei dis- thus ranged from 2.51 to 17.48%. A small number (six) tance, to determine whether sister taxa inferred to be of the sequence divergence comparisons between sister the result of Pleistocene speciation (via the now tradi- taxa would be consistent with Pleistocene speciation tional but noncritical criterion of percentage sequence events (Table 3). Note that for the two comparisons divergence) have in fact diverged in a clocklike fashion. which represent within species comparisons, sequence divergence is substantially higher than might be ex- RESULTS pected for conspecifics (Table 3). As expected for a protein coding gene, substantial Cytochrome b Dynamics rate heterogeneity was present across codon positions I aligned 1035 bp without insertions or deletions. (Table 2). Both across all taxa and within Anthus, the ␣ Because mtDNA was isolated for many specimens and low values at first and second codon positions sug- the entire fragment of cytochrome b was typically gests that most sites have either a low substitution rate isolated as a single unit for amplification, nuclear or are invariable, whereas the higher values for third copies are less likely for these taxa, but have a greater positions suggests that most sites have intermediate chance of amplification from those specimens whose substitution rates with only a few sites having very low tissue was preserved in buffer (as mtDNA could not be or very high rates (Sullivan et al., 1995; Yang, 1996a). isolated from these samples). However, the lack of Third positions were also found to have a much higher ␬ alignment problems and stop codons was evidence that transitional bias ( ) compared to first and second codon the genes amplified were of mitochondrial origin. positions. For the 1035 nucleotide sites 550 were variable, of which only 386 would be considered informative under Phylogenetic Analysis and Molecular Clock parsimony criteria. This range of variability was found Both strategies of maximum likelihood analysis re- both across all taxa considered, and within Anthus sulted in the same tree topology (Fig. 1). I rely on this alone. Of variable sites, 161 were at first, 81 at second, ML tree as a better estimate of phylogeny for several and 308 were at third codon positions. Of the parsi- reasons. First, there have been an increasing number of mony informative sites, 95 were at first, 24 were at studies in which cytochrome b data have been unable to second, and 267 were at third codon positions. provide reasonable levels of support for relationships Mean base composition is similar across all taxa and under certain phylogenetic algorithms, especially when within Anthus (Table 2). These mean values are compa- rapid cladogenesis is inferred (e.g., Lara et al., 1996; 88 GARY VOELKER

TABLE 3 branches tend to be near a terminal pair of taxa on the tree. Sister Taxa for which Percentage Sequence Diver- gence (Uncorrected) Is Approximately Consistent with Second, recent studies have observed problems in the Pleistocene Speciation reliability of NJ and MP under certain conditions. For NJ, accuracy in recovering trees has been shown to % Sequence Clocklike decrease both when the number of taxa included in- a,b c Sister taxa divergence MYBP divergence? CP(%) creases and when branch lengths are short (Strimmer correndera (1)/ant- and von Haeseler, 1996). Under MP, not only is an arcticus 2.7 1.0 Yes 11.92 explicit model of evolution lacking, but there is evi- rubescens (N.A.)/ru- dence that stringent and unrealistic assumptions about bescens (Asia) 3.3 1.3 Yes 73.72 rufulus/richardi 4.3 1.8 No 98.58 substitution processes exist (Yang, 1996b). Due to these similis (1)/melindae 4.8 2.05 Yes 12.72 shortcomings, Yang and Goldman (1996) suggest that spinoletta (1)/spino- parsimony analysis is ill-adapted to studying the evolu- letta (2) 3.5 1.4 No 99.70 tionary process. A further concern is that incorrect hellmayri (1)/bogo- tensis 4.4 1.85 Yes 25.86 placements of taxa can nonetheless provide substantial levels of bootstrap support under parsimony criteria Note. The two-cluster test (Takezaki et al., 1995) is used to test (Naylor and Brown, 1997). whether these sister taxa are in fact diverging in a clocklike manner. Finally, ML has been shown to outperform other a Time in millions of years, assuming 2% sequence divergence per million years. methods, is less sensitive to base composition inequali- b Corrected to account for amounts ancestral diversity inferred for ties and models of DNA substitution, accounts for several species (350,000 years; see text). multiple hits, and allows for the incorporation of criti- c CP is the confidence probability (1 Ϫ P value) for the two-cluster test, with larger values indicating increasing certainty of rejection of cal parameters estimated directly from the data being a molecular clock; values greater than 95% (P Ͻ 0.05) reject a considered (Huelsenbeck, 1995; Perna and Kocher, molecular clock outright. 1995; Scho¨niger and von Haeseler, 1995; Swofford et al., 1996; Sullivan and Swofford, 1997). The fully resolved ML tree, then, suggests that Mariaux and Braun, 1996; Nunn and Cracraft, 1996; Anthus can be divided into several major clades: (1) the Randi, 1996; Voelker and Edwards, 1998). To determine small-bodied African pipits (sokokensis, brachyurus, whether this phenomenon is apparent in this data set, I and caffer), (2) a largely South American radiation, (3) a also performed maximum parsimony (MP) and neigh- clade composed largely of Palearctic migrants, and (4) bor-joining (NJ) analyses. In MP all sites were initially an African–Asian–Australian clade consisting of gener- treated as unordered character states with equal weight. ally larger species (Fig. 1). Relationships for several of ␬ was incorporated (as the correlated transition: these groupings received reasonable levels of bootstrap transversion ratio) into subsequent analyses via step- support (Fig. 1) and are consistent with previously matrices applied to each codon position. I used heuris- postulated estimates of phylogenetic affinities (see be- tic search, random addition, with 10 replicates for each low). Within each of these clades, there are a number of weighting scheme. Cladistic signal was determined for smaller groupings of taxa that receive support from one each clade by bootstrapping (100 replicates; Felsen- or both of the bootstrapping analyses. stein, 1985). NJ was used to fit a tree to a matrix of This tree also suggests that a number of species are HKY85 distances; cladistic signal was determined for paraphyletic (hellmayri, similis, correndera, rubes- each clade via subsequent bootstrapping analysis (1000 cens). This is not surprising given the morphological replicates). similarity of pipits and the shortage of life history and Equally weighted MP analysis resulted in 32 equally other biological (e.g., behavioral) data that might con- parsimonious trees (2233 steps), found 8 other trees tribute evidence for recognizing subspecies as species. which were only 1 step longer, and did not resolve the The molecular clock-enforced ML tree was signifi- relationship of Anthus relative to Motacilla. MP and NJ cantly worse (-ln L ϭ 11980.82; P Ͻ 0.0001) than the bootstrap analyses failed to recover many nodes within nonclock ML tree; thus, Anthus cytochrome b is not Anthus at greater than 60% (see Fig. 1), a marginal behaving, across all taxa, in a clocklike fashion. This is level of support, as well as the node between Anthus evident in sister taxon comparisons (Table 3); the and Motacilla (but see Voelker and Edwards, 1998); results of the two-cluster test suggest that the cyto- these internodes are generally much shorter than those chrome b of just 66% (four of six comparisons) of the internodes receiving bootstrap support. This suggests sister taxa inferred to be the result of Pleistocene that bootstrapping has difficulty recovering short inter- speciation events by the criteria of percentage sequence nodes (rapid cladogenesis), which tends to reflect few divergence can with some measure of confidence be parsimony informative characters, unless those dated according to a molecular clock concept. Anthus SYSTEMATICS 89

FIG. 1. Maximum likelihood tree of Anthus relationships using estimated values of among-site rate heterogeneity and transitional bias. Log-likelihood score is Ϫ11881.16. Numbers above internodes are frequencies of occurrence (Ͼ60%) of different groups in 1000 neighbor- joining bootstrap replicates; values below internodes are frequencies (Ͼ60%) in 100 maximum parsimony bootstrap replicates. Vertically numbered bars delineate clades discussed in text: 1, African small-bodied pipits; 2, a largely South American radiation; 3, a clade composed mostly of Palearctic migrants; and 4, an African–Asian–Australian clade. 90 GARY VOELKER

DISCUSSION lineages in this genus are evolving in a clocklike fashion (Table 3); the two-cluster test was able to Cytochrome b Dynamics determine specifically which of the sister taxon compari- The differences in base composition found here are sons (which were possibly the result of Pleistocene strikingly similar to those found across a growing speciation) were nonclocklike (Table 3). number of distantly related avian lineages. These Yet, despite the inference of clocklike behavior for include oscine (Edwards et al., 1991; Hackett, 1996; some of these sister taxa comparisons, it is quite Nunn and Cracraft, 1996; Voelker and Edwards, 1998) unlikely that similar rates are in effect between these and suboscine (Edwards et al., 1991) taxa, as same sets of sister taxa. For example in Table 3, well as nonpasserine taxa such as cranes (Krajewski although the sisters correndera (1)/antarcticus and and King, 1996), woodpeckers (Cicero and Johnson, hellmayri (1)/bogotensis are both inferred to be diverg- 1995), galliformes (Randi, 1996), and Pelecaniformes ing according to a clock, the node which links these two (Friesen and Anderson, 1996). The similarity in base sets of sister taxa (Fig. 1) is not clocklike (Voelker, in composition across these disparate avian lineages indi- prep.). So, while the two daughters within each of the rectly suggests similarity in cytochrome b dynamics above sister taxa comparisons are diverging from each due to patterns of directional mutation pressure other in a clocklike fashion, those same two sets of sister taxa are not diverging from each other at the (Sueoka, 1992; Jermiin et al., 1995), an idea which has same rate. Despite this problem, rate heterogeneity is only recently and preliminarily been addressed in usually moderate when closely related sequences are (i.e., Griffiths, 1996; Krajewski and King, 1996). used (Takezaki et al., 1995). Therefore, though a time Short internodal distances at several deeper splits on can be inferred at all clocklike sister taxa nodes (and at the tree suggest a rapid divergence of pipits approach- other clocklike nodes deeper within the tree; Voelker, in ing a star phylogeny. Short internodal differences have prep.), the tenuous nature of rate homogeneity at all been attributed to rapid evolution and thus short time clocklike nodes should be recognized. intervals between the divergences of a number of higher level avian and mammalian relationships (Sib- Phylogenetic Relationships ley and Ahlquist, 1990; Bleiweiss et al., 1994; Lara et al., 1996; Sheldon and Gill, 1996; Voelker and Edwards, There are no previous hypotheses addressing the phylogenetic relationships within the entire genus to 1998). The idea of rapid cladogenesis can be tested with which the ML phylogeny can be compared. However, another gene region. the phylogenetic hypothesis presented here, which Assuming the commonly employed rate of avian includes most extant Anthus species, can be compared molecular sequence divergence of about 2% per million to studies that have provided evidence to support years (several independent correlations of sequence divergence rate fall near this value; i.e., Shields and relationships between particular species. Wilson, 1987; Tarr and Fleischer, 1993; Wood and Small-Bodied Pipits Krajewski, 1996), only a small fraction of the pairwise comparisons of sequence divergence within Anthus are Of the four major clades resulting from the ML consistent with Pleistocene speciation events (Table 3); analysis, only the African small-bodied pipit clade (A. most suggest Pliocene or Pliocene/Miocene origins. This sokokensis/A. brachyurus/A. caffer) has previously been result is consistent with molecular evidence which recognized by several authors (e.g., Hall, 1961; Hall either refutes many postulated Pleistocene passerine and Moreau, 1970; Clancey, 1990). This grouping was sister taxa speciation events (Klicka and Zink, 1997) or based largely on body size, wing formula, and hind claw strongly suggests that diversification among passerine shape compared to virtually all other Anthus species. groups is lower in the Pleistocene than in previous Within this clade, Hall and Moreau (1970) postulated periods (Zink and Slowinski, 1995; but see Avise and that sokokensis and caffer were most closely related Walker, 1998). This is true for Anthus even if interspe- based on closer body size proportions and dorsal color, cific divergence estimates are corrected for amounts of compared to brachyurus; this is not supported by the ancestral diversity suggested for other species (about ML phylogeny. 350,000 years; Moore, 1995; Edwards, 1997; but see Avise and Walker, 1998), although the general applica- South American Radiation bility of this value across taxa is as yet undetermined. Within the clade consisting almost entirely of a South However, it should be apparent that if a molecular American radiation, the ML phylogeny supports the clock is operating, rates of sequence divergence must be contention of Hall (1961) that correndera and antarcti- equal, or very nearly equal, along each daughter lin- cus are sister taxa, but does not support a superspecies eage above a given node on the tree. If this is not true, relationship consisting of these species and novaeseelan- then dates cannot be reliably assigned to such nodes diae (Hall and Moreau, 1970). The ML phylogeny and divergence estimates are incorrect. Both tests for a supports a sister relationship between lutescens and molecular clock for Anthus reject the possibility that all spragueii (endemic to North America), which conflicts Anthus SYSTEMATICS 91 with Hall’s (1961) conclusion that spragueii is a subspe- phylogeny, which places pratensis between rubescens cies of furcatus, while at the same time supporting her and petrosus/spinoletta. contention that it was in fact part of the otherwise Based on a recent analysis of mitochondrial DNA entirely South American radiation. restriction fragments, Zink et al. (1995) found that the Also within this South American clade, the ML eastern Siberian race of rubescens and a North Ameri- phylogeny suggests that species limits within hell- can race of rubescens were highly differentiated, at a mayri and correndera need further study. Representa- level suggesting species status for both (P ϭ 0.029; tive samples from two subspecies (for each species) Zink et al., 1995). Although only one specimen of each were not most closely related to each other. For hell- ‘‘species’’ was included in this study, they are roughly mayri, the two subspecies considered (A. h. hellmayri 3% sequence divergent (Table 3), which is equal to the and A. h. brasilianus) breed in allopatry, separated by restriction fragment difference found by Zink et al. most of northern Argentina and Paraguay (see Ridgely (1995). In addition to these genetic differences, these and Tudor, 1989). They are highly distinguishable two ‘‘species’’ have differentiated in migratory habits, ‘‘subspecies,’’ differing in size, coloration, and amount of each wintering on the same continent on which they ventral streaking and have slightly different song breed (Dement’ev and Gladkov, 1970; Cramp, 1988). (pers. obs.). The third hellmayri subspecies, A. h. Several authors (Hall and Moreau, 1970; Clancey, dabbenei (unavailable), also breeds in allopatry from 1990) have suggested a close relationship between the other subspecies, separated by most of central spinoletta (or the Water Pipit complex) and similis; this Argentina, and is the only subspecies of hellmayri that relationship is not supported by the ML phylogeny. is an austral migrant. Thus hellmayri as currently African–Asian–Australasian Radiation recognized may be a composite of up to three phylo- genetic species (Zink and McKitrick, 1995). A. novaeseelandiae has until recently been described The relationships of the two subspecies of correndera as a wide ranging (southern Africa to New Zealand) included here (nominate correndera and A. correndera superspecies, in which relationships were poorly under- catamarcae) suggest that this species is paraphyletic as stood. Glutz von Blotzheim and Bauer (1985) split the well. Morphological differences are less striking be- complex into five species based on allopatric popula- tween these taxa than between hellmayri subspecies. I tions or allopatric groups of subspecies. All African am aware of no other differences reported in the subspecies are now considered races of cinnamomeus; literature which might have supported earlier specula- richardi comprises the migratory Asian populations, tion about species limits, with the exception of a range rufulus the sedentary races from south and southeast disjunction between northern subspecies (catamarcae Asia, australis all races from Australia and New Guinea, and A. correndera calcaratus) and southern subspecies and novaeseelandiae those populations from New Zea- in northwest Argentina; this region in fact divides the land; most current treatments generally recognize these ranges of many species (see Ridgely and Tudor, 1989) divisions (e.g., Clancey, 1985; Sibley and Monroe, 1990), and is part of an area of endemism for still others but retain australis as a subspecies under novaeseelan- (Cracraft, 1985). diae (see Christidis and Boles, 1994). The ML phylo- geny does not support a close relationship among these Palearctic Migrants species: while richardi and rufulus are sister species, novaeseelandiae and cinnamomeus are both more closely Within the clade of largely Palearctic migrants, the related to other species (Fig. 1). Recent molecular work relationships within the ‘‘Water Pipit complex’’ have suggests that several populations of novaeseelandiae received much attention (see reviews in Knox, 1988; (sensu Glutz von Blotzheim and Bauer, 1985) from Arctander et al., 1996). Initially considered one Holarc- small islands near New Zealand warrant species level tic breeding species, spinoletta, the complex was split recognition (Foggo et al., 1997). into three species consisting of petrosus (coastal Eu- Morphologically, the recently described longicauda- rope), spinoletta (Eurasia to eastern Siberia), and rube- tus is closest to vaalensis (Liversidge, 1996); a close scens (eastern Siberia and North America). Species relationship between these species is also supported by status for petrosus was based on allopatric breeding the ML phylogeny. The level of sequence divergence and different ecological associations compared to spino- (5.2%) is suggestive of the species level distinction letta (Knox, 1988). Differences in ecological associa- supported by morphological and vocal and foraging tions compared to spinoletta (Nazarenko, 1978) also behavior differences (Liversidge, 1996; Pers. obs.). prompted the recognition of rubescens as a distinct In a recent study addressing the phylogenetic rela- species. Based on differences in morphology and vocal- tionships of Berthelot’s Pipit (berthelotii), Alstro¨m and izations, Alstro¨m and Mild (1987) supported splitting Mild (1993) presented data on morphology, vocaliza- rubescens from spinoletta and suggested that spinoletta tions, and behavior which suggested a close relation- and petrosus were closer to each other than either was ship with campestris. Arctander et al. (1996) found to rubescens. This suggestion is supported by the ML unambiguous molecular evidence supporting berthe- 92 GARY VOELKER lotii and campestris as sister taxa compared to six other history information for many Anthus taxa has resulted Anthus taxa. The relationship between these two taxa in few well-supported or agreed upon species relation- is supported by the ML tree. However, the close relation- ships, which in turn has contributed to both paraphy- ship suggested by Arctander et al. (1996) between these letic species designations and several cryptic species two taxa and novaeseelandiae (A. n. lacuum (ϭcinna- being unrecognized until recently (Liversidge, 1996; momeus)) based on molecular evidence and juvenile this study); it is not unlikely that other Anthus species plumage is not. Instead the ML phylogeny places will prove paraphyletic and that other cryptic species berthelotii/campestris as sister to leucophrys, which await discovery. also conflicts with the highly speculative suggestion of Beyond being an hypothesis of pipit relationships, Bourne (1995) that they are closely related to similis. the ML topology presented here provides a framework The placement of leucophrys on the ML tree conflicts for testing evolutionary hypotheses for this widespread with speculation that it is part of a superspecies group. While the addition of the few species not in- consisting of leucophrys, vaalensis, and pallidiventris cluded here may change relationships to some degree, (the latter was not available for this study) (Hall and this phylogeny makes it possible, for the first time, to Moreau, 1970; Keith et al., 1992). The ML phylogeny supports Clancey’s (1985) recognition of nyassae and begin reconstructing the biogeographic history, evolu- hoeschi (see also Mendelsohn, 1984) as being distinct tion of migration, and evolution of clutch size from a from similis. historical perspective in this diverse passerine genus. The two ‘‘similis’’ specimens included are not close relatives. Reexamination of the voucher specimen for ACKNOWLEDGMENTS similis 1 (housed at the MacGregor Museum, Kimber- ley, South Africa) has shown this specimen to be I am very grateful to S. Rohwer and S. V. Edwards for ongoing distinctly different from any described race of similis in support, advice, and discussion. Special thanks to S. V. Edwards for coloration, wing shape, and several other morphologi- allowing this work to be conducted in his lab. I also thank R. cal characters of which the most striking is a beak Liversidge for ongoing discussion of pipit systematics and behavior. J. about 30% shorter than mean beak lengths published Felsenstein, J. Huelsenbeck, and D. Swofford provided advice on for similis (Long-billed pipit; Keith et al., 1990). The analyzing large data sets using ML; D. Swofford kindly granted permission to use PAUP*. S. V. Edwards, R. Huey, R. Liversidge, S. specimen is also different from any other described Pearson, S. Rohwer, C. E. Smith, and two anonymous reviewers made species of Anthus. Formal description of this new pipit valuable comments on the manuscript. I thank the Burke Museum species is in progress (Voelker and Liversidge, unpub. staff, B. Schmidt, and J. Gerwin for field assistance and collecting manuscript). efforts. The University of Washington Burke Museum (UWBM), Los Given the great morphological similarity and num- Angeles County Museum (LACM), Field Museum of Natural History ber of species in this genus, it is hardly surprising that (FMNH), American Museum of Natural History (AMNH), Zoologisk no previous hypotheses of the phylogenetic relation- Museum, Copenhagen (COP), British Antarctic Survey (BAS), Acad- emy of Natural Sciences, Philadelphia (ANS), Macgregor Museum, ships among all species of Anthus exist. Hall (1961) Kimberley, South Africa (MM), National Museums of Kenya (NMK), elaborated on the values and limitations of five sets of and K. Hustler provided tissue or blood samples critical to this study. morphological characters in classifying pipits: color I thank A. Berruti, T. Cassidy, R. Nuttall, and the Leitch, Pienaar, and pattern, size, hind claw, tail pattern, and wing and Whipp families for assistance and hospitality in South Africa, V. formula. That these characters are more limiting than Molkov, V. Khrokov, and N. Borisova for the same in Kazakhstan and useful seems apparent from the general lack of relation- Russia, and P. Capllonch, Dr. R. Barquez, the Larega family, and E. Krauczuk for the same in Argentina. C. Miranda (Div. Fauna, ships proposed. For instance, juvenile plumage and Tucuma´n), Drs. M. Costa and O. Nassiff (Recursos Naturales y Medio molt patterns may be useful to some degree (Hall, 1961; Ambiente, Corrientes), the Direccio´n de Recurso y Vida Silvestre Arctander et al., 1996), but plumages and patterns are (Misiones), and Lic. D. Ramadori and Lic. V. Lichtschein (Direccio´n undocumented for many species (e.g., Keith et al., 1992; de Fauna y Flora Silvestres, Buenos Aires) provided collecting and Maclean, 1993) so few relationships can be postulated export permits in Argentina; I also thank the permitting agencies in based on these characters. Another potentially useful Russia, Kazakhstan, and South Africa for their consideration and character, patches of color, is of limited value as the assistance. Throughout this work I have been supported by an Eddy Graduate Fellowship in Ornithology and the Leaders Five Fellowship majority of species lack colors other than browns. The for Molecular Systematics, both from the Burke Museum. While few relationships that could be implied using morpho- collecting specimens for this research I was supported by grants from logical characters do not necessarily agree with the ML the University of Washington, the American Ornithologists’ Union tree. While possession of pink underparts would link (Alexander Wetmore Award and Blake Trust), the Frank M. Chap- cervinus and roseatus as close relatives, in agreement man Fund (American Museum of Natural History), and Sigma Xi with the ML tree, possession of yellow axillaries (rosea- Grants-in-Aid. Labwork was supported by an NSF Dissertation tus, lineiventris, and crenatus) or green wing edges Improvement Grant (DEB-9700995), the University of Washington Department of Zoology (Chair’s Fund for Molecular Research), grants (roseatus and hodgsoni) would imply close relation- from the University of Washington Royalties Research Fund and the ships which conflict with the ML tree. Overall, the National Science Foundation (DEB-9419738) to S. V. Edwards, and problems of morphological similarity and lack of life contributions by F. Ayer to establish the Edwards lab. Anthus SYSTEMATICS 93

REFERENCES in pipits (Anthus: Motacillidae) in New Zealand and some offshore islands. Ibis 139: 366–373. Alstro¨m, P., and Mild, K. (1987). Some notes on the of the Friesen, V. L., and Anderson, D. J. (1997). Phylogeny and evolution of Water Pipit complex. Proc. 4th Int. Identification Meeting (Eilat). the Sulidae (Aves: Pelecaniformes): A test of alternative modes of speciation. Mol. Phylogenet. Evol. 7: 252–260. Alstro¨m, P., and Mild, K. (1993). The taxonomic status of Anthus berthelotii. Bull. B. O. C. 113: 88–92. Glutz von Blotzheim, U. N., and Bauer, K. M., Eds. (1985). ‘‘Hand buch der Vo¨gel Mitteleuropas,’’ AULA-Verlag, Weisbaden, Ger- Arctander, P., Folmer, O., and Fjeldså, J. (1996). The phylogenetic many. relationships of Berthelot’s Pipit Anthus berthelotii illustrated by DNA sequence data, with remarks on the genetic distance between Griffiths, C. S. (1997). Correlation of functional domains and rates of Rock and Water Pipits Anthus spinoletta. Ibis 138: 263–272. nucleotide substitution in cytochrome b. Mol. Phylogenet. Evol. 7: 352–365. Avise, J. C., and Walker, D. (1998). Pleistocene phylogeographic effects on avian populations and the speciation process. Proc. R. Hackett, S. J. (1996). Molecular phylogenetics and biogeography of Soc. London B 265: 457–463. Tanagers in the genus Ramphocelus (Aves). Mol. Phylogenet. Evol. 5: 368–382. Bleiweiss, R., Kirsch, J. A. W., and Lapointe, F.-J. (1994). DNA–DNA hybridization-based phylogeny for ‘‘higher’’ nonpasserines: Reevalu- Hall, B. P. (1961). The taxonomy and identification of pipits (genus ating a key portion of the avian family tree. Mol. Phylogenet. Evol. Anthus). Bull. Br. Mus. (Nat. Hist.) 7: 243–289. 3: 248–255. Hall, B. P., and Moreau, R. E. (1970). ‘‘An Atlas of Speciation in Bourne, W. R. P. (1995). The origin and affinities of Berthelot’s Pipit African Passerine Birds,’’ British Museum (Nat. Hist.), London. Anthus bertheloti. Bull. B. O. C. 115: 22–24. Hasegawa, M., Kishino, H., and Yano, T. (1985). Dating of the Christidis, L., and Boles, W. E. (1994). ‘‘The Taxonomy and Species of human–ape splitting by a molecular clock of mitochondrial DNA. J. Birds of Australia and its Territories,’’ Royal Australasian Ornitholo- Mol. Evol. 22: 160–174. gists Union, Victoria. Helm-Bychowski, K., and Cracraft, J. (1993). Recovering phylo- Cicero, C., and Johnson, N. K. (1995). Speciation in sapsuckers genetic signal from DNA sequences: Relationships within the (Sphyrapicus): III. Mitochondrial-DNA sequence divergence at the corvine assemblage (Class Aves) as inferred from complete se- cytochrome-B locus. Auk 112: 547–563. quences of the mitochondrial DNA cytochrome b gene. Mol. Biol. Evol. 10: 1196–1214. Clancey, P. A. (1985). Species limits in the Long-billed Pipits of the southern Afrotropics. Ostrich 56: 157–169. Hellmayr, C. E. (1935). Catalogue of birds of the Americas, part 7. Clancey, P.A. (1990). A review of the indigenous pipits (Genus Anthus Field Mus. Nat. Hist. Zool. Ser. 13(7): 1–541. Bechstein: Motacillidae) of the Afrotropics. Durban Mus. Novit. 15: Huelsenbeck, J. P. (1995). Performance of phylogenetic methods in 42–72. simulation. Syst. Biol. 44: 17–48. Cooper, M. R. (1985). A review of the genus Macronyx and its Irwin, D. M., Kocher, T. D., and Wilson, A. C. (1991). Evolution of the relationship to the Yellow-bellied Pipit. Honeyguide 31: 81–92. cytochrome b gene of mammals. J. Mol. Evol. 32: 128–144. Cracraft, J. (1985). Historical biogeography and patterns of differen- Jermiin, L. S., Graur, D., and Crozier, R. H. (1995). Evidence from tiation within the South American avifauna: Areas of endemism. In analyses of intergenic regions for strand-specific directional muta- ‘‘Neotropical Ornithology’’ (P.A. Buckley, M. S. Foster, E. S. Morton, tion pressure in metazoan mitochondrial DNA. Mol. Biol. Evol. 12: R. S. Ridgely, and F. G. Buckley, Eds.), pp. 49–84, Ornithological 558–563. Monographs 36, American Ornithologists’ Union. Keith, S., Urban, E. K., and Fry, C. H. (1992). ‘‘The Birds of Africa,’’ Cramp, S., Ed. (1988). ‘‘The Birds of the Western Palearctic,’’ Vol. 5 Vol. 4, Academic Press, London. Oxford Univ. Press, London. Kishino, H., and Hasegawa, M. (1989). Evaluation of the maximum Dement’ev, G. P., and Gladkov, N. A., Eds. (1970). ‘‘The Birds of the likelihood estimate of the evolutionary tree topologies from DNA Soviet Union,’’ Vol. 5, Israel Program for Scientific Translations, sequence data, and the branching order in Hominoidea. J. Mol. Jerusalem. Evol. 29: 170–179. Desjardins, P., and Morais, R. (1990). Sequence and gene organiza- Klicka, J. T., and Zink, R. M. (1997). The importance of recent ice ages tion of the chicken mitochondrial genome: A novel gene order in in speciation: A failed paradigm. Science 277: 1666–1669. higher vertebrates. J. Mol. Biol. 212: 599–634. Knox, A. (1988). Taxonomy of the Rock/Water Pipit superspecies Dowling, T. E., Moritz, C., and Palmer, J. D. (1990). Nucleic acids II: Anthus petrosus, spinoletta, and rubescens. Br. Birds 81: 206–211. Restriction site analysis. In ‘‘Molecular Systematics’’ (D. M. Hillis Kocher, T. D., Thomas, W. K., Meyer, A., Edwards, S. V., Pa¨a¨bo, S., and C. Moritz, Eds.), pp. 250–317, Sinauer, Sunderland, MA. Villablanca, F. X., and Wilson, A. C. (1989). Dynamics of mitochon- Edwards, S. V. (1997). Relevance of microevolutionary processes to drial DNA evolution in : Amplification and sequencing with higher level molecular systematics. In ‘‘Avian Molecular Evolution conserved primers. Proc. Natl. Acad. Sci. USA 86: 6196–6200. and Systematics’’ (D. P. Mindell, Ed.), pp. 251–278, Academic Press, Krajewski, C., and King, D. G. (1996). Molecular divergence and London. phylogeny: Rates and patterns of cytochrome b evolution in cranes. Edwards, S. V., and Wilson, A. C. (1990). Phylogenetically informa- Mol. Biol. Evol. 13: 21–30. tive length polymorphism and sequence variability in mitochon- Lara, M. C., Patton, J. L., and da Silva, M. N. F. (1996). The drial DNA of Australian songbirds (Pomatostomus). Genetics 126: simultaneous diversification of South American Echimyid rodents 695–711. (Hystricognathi) based on complete cytochrome b sequences. Mol. Edwards, S. V.,Arctander, P., and Wilson, A. C. (1991). Mitochondrial Phylogenet. Evol. 5: 403–413. resolution of a deep branch in the genealogical tree for perching Liversidge, R. (1996). A new species of pipit in southern Africa. Bull. birds. Proc. R. Soc. London (Biol.) 243: 99–107. B. O. C. 116: 211–215. Felsenstein, J. (1985). Confidence limits on phylogenies: An approach Maclean, G. L. (1993). ‘‘Roberts’ Birds of Southern Africa,’’ Trustees of using the bootstrap. Evolution 39: 783–791. the John Voelcker Bird Book Fund, Capetown. Foggo, M. N., Hitchmough, R. A., and Daugherty, C. H. (1997). Mariaux, J., and Braun, M. J. (1996). A molecular phylogenetic Systematics and conservation implications of geographic variation survey of the Nightjars and allies (Caprimulgiformes) with special 94 GARY VOELKER

emphasis on the Potoos (Nyctibiidae). Mol. Phylogenet. Evol. 6: Sueoka, N. (1992). Directional mutation pressure, selective con- 228–244. straints, and genetic equilibria. J. Mol. Evol. 34: 94–114. Mendelsohn, J. (1984). The Mountain Pipit in the Drakensberg. Sullivan, J., Holsinger, K. E., and Simon, C. (1995). Among-site rate Bokmakierie 36: 40–44. variation and phylogenetic analysis of 12S rRNA in Sigmodontine Moore, W. E. (1995). Inferring phylogenies from mtDNA variation: rodents. Mol. Biol. Evol. 12: 988–1001. Mitochondrial-gene trees versus nuclear-gene trees. Evolution 49: Sullivan, J., and Swofford, D. L. (1997). Are guinea pigs rodents? The 718–726. importance of adequate models in molecular phylogenetics. J. Naylor, G. J. P., and Brown, W. M. (1997). Structural biology and Mamm. Evol. 2: 77–86. phylogenetic estimation. Nature 388: 527–528. Swofford, D. L., Olsen, G. J., Waddell, P. J., and Hillis, D. M. (1996). Nazarenko, A. A. (1978). On species validity of Anthus rubescens Phylogenetic analysis. In ‘‘Molecular Systematics’’ (D. M. Hillis and Tunstall (Aves, Motacillidae). Zool. Zh. 57: 1743–1744. C. Moritz, Eds.), pp. 407–514, Sinauer, Sunderland, MA. Nunn, G., and Cracraft, J. (1996). Phylogenetic relationships among Takezaki, N., Rzhetsky, A., and Nei, M. (1995). Phylogenetic test of the major lineages of the Birds-of-Paradise (Paradisaeidae) using the molecular clock and linearized trees. Mol. Biol. Evol. 12: mitochondrial DNA gene sequences. Mol. Phylogenet. Evol. 5: 823–833. 445–459. Tarr, C. L., and Fleischer, R. C. (1993). Mitochondrial DNA variation Perna, K., and Kocher, T. (1995). Unequal base frequencies and the and evolutionary relationships in the ’Amakihi complex. Auk 110: estimation of substitution rates. Mol. Biol. Evol. 12: 359–361. 825–831. Randi, E. (1996). A mitochondrial cytochrome B phylogeny of the Voelker, G., and Edwards, S. V. Can weighting improve bushy trees? Alectoris partridges. Mol. Phylogenet. Evol. 6: 214–227. Models of cytochrome b evolution and the molecular systematics of Ridgely, R. S., and Tudor, G. (1989). ‘‘The Birds of South America,’’ pipits and wagtails (Aves: Motacillidae). Syst. Biol., 47: 589–603. Vol. 1, Univ. of Texas Press, Austin. Wakeley, J. (1993). Substitution-rate variation among sites and the Ridgway, R. (1904). The birds of North and Middle America, part 3. estimation of transition bias. Mol. Biol. Evol. 11: 436–442. Bull. U.S. Natl. Mus. 50(3): 1–801. Wakeley, J. (1996). The excess of transitions among nucleotide Saitou, N., and Nei, M. (1987). The neighbor-joining method: A new substitutions: New methods of estimating transition bias under- method for reconstructing phylogenetic trees. Mol. Biol. Evol. 4: score its significance. Trends Ecol. Evol. 11: 158–163. 406–425. Walsh, P. S., Metzger, D. A., and Higuchi, R. (1991). Chelex 100 as a Scho¨niger, M., and von Haeseler, A. (1995). Performance of the medium for simple extraction of DNA for PCR-based typing from maximum likelihood, neighbor joining, and maximum parsimony forensic material. BioTechniques 10: 506–513. methods when sequence sites are not independent. Syst. Biol. 44: Wood, T. C., and Krajewski, C. (1996). Mitochondrial DNA sequence 533–547. variation among subspecies of Sarus Crane (Grus antigone). Auk Sheldon, F. H., and Gill, F. B. (1996). A reconsideration of songbird 113: 655–663. phylogeny, with emphasis on the evolution of titmice and their Yang, Z. (1996a). Among-site rate variation and its impact on sylvioid relatives. Syst. Biol. 45: 473–495. phylogenetic analyses. Trends Ecol. Evol. 11: 367–372. Shields, G. F., and Wilson, A. C. (1987). Calibration of mitochondrial Yang, Z. (1996b). Phylogenetic analysis using parsimony and likeli- DNA evolution in geese. J. Mol. Evol. 24: 212–217. hood methods. J. Mol. Evol. 42: 294–307. Sibley, C. G., and Ahlquist, J. E. (1981). The relationships of the Yang, Z., and Goldman, N. (1997). Are big trees indeed easy? Trends wagtails and pipits (Motacillidae) as indicated by DNA–DNA Ecol. Evol. 12: 357. hybridization. L’Oiseau R.F.O. 51: 189–199. Zink, R. M., and McKitrick, M. C. (1995). The debate over species Sibley, C. G., and Ahlquist, J. E. (1990). ‘‘Phylogeny and Classification concepts and its implications for ornithology. Auk 112: 701–719. of Birds: A Study in Molecular Evolution,’’ Yale Univ. Press, New Zink, R. M., and Slowinski, J. B. (1995). Evidence from molecular Haven, CT. systematics for decreased avian diversification in the Pleistocene Sibley, C. G., and Monroe, B. L., Jr. (1990). ‘‘Distribution and Epoch. Proc. Natl. Acad. Sci. USA 92: 5832–5835. Taxonomy of Birds of the World,’’ Yale Univ. Press, New Haven, CT. Zink, R. M., Rohwer, S., Andreev, A. V., and Dittmann, D. (1995). Strimmer, K., and von Haeseler, A. (1996). Accuracy of neighbor Trans-Beringia comparisons of mitochondrial DNA differentiation joining for n-taxon trees. Syst. Biol. 45: 516–523. in birds. Condor 97: 639–649.