EB, Avian Amazona Auk04

7/29/2019 EB, Avian Amazona Auk04

1/15

B Neotropicalbirds have been of central importance in devel-opment of models aimed at explaining the highspecies diversity of the Neotropics (e.g. Cracra1985, Haffer 1985). An obvious example is theForest Refuge hypothesis initially outlined byHaffer (1969, 1974), which suggests that isola-tion of remnant patches of rainforest duringdry glacial periods fostered diversification

of tropical flora and fauna. Other importantvicariant models of speciation that have beensupported by studies of the diversification ofNeotropical avifauna include the Andean up-li (Chapman 1917, Cracra and Prum 1988),formation of river systems in the Amazon

basin (Wallace 1853, Simpson and Haffer 1978,Capparella 1988), formation of the Panama land

bridge (Cracra 1985, Cracra and Prum 1988),and marine incursions (Nores 1999).

It is within this rich ornithological tradi-

tion that we present results of a phylogeneticanalysis of the Amazona ochrocephala complexof parrots, a group of biogeographic interest

because of its broad Neotropical distribution.

A.We present a phylogenetic analysis of relationships among members of theAmazonaochrocephala species complex of parrots, a broadly distributed group in Middle andSouth America that has been a taxonomic headache. Mitochondrial DNA sequence data areused to infer phylogenetic relationships among most of the named subspecies in the complex.Sequence-based phylogenies show that Middle American subspecies included in the analysisare reciprocally monophyletic, but subspecies described for South America do not reflect pat-terns of genetic variation. Samples from the lower Amazon cluster with samples collected inwestern Amazonianot with samples from Colombia and Venezuela, as was predicted bysubspecies classification. All subspecies of the complex are more closely related to one anotherthan to otherAmazona species, and division of the complex into three species (A. ochrocephala,

A. auropalliata, andA. oratrix) is not supported by our data. Divergence-date estimates suggest

that these parrots arrived in Middle America aer the Panama land-bridge formed, and thenexpanded and diversified rapidly. As in Middle America, diversification of the group in SouthAmerica occurred during the Pleistocene, possibly driven by changes in distribution of foresthabitat. Received 20 January 2003, accepted 3 December 2003.

R.Presentamos un anlisis de las relaciones filogenticas entre miembros delcomplejo de loros Amazona ochrocephala, un grupo ampliamente distribuido en Mesoamricay Suramrica, y que ha sido un dolor de cabeza taxonmico. Utilizamos secuencias de ADNmitocondrial para reconstruir la relaciones filogenticas entre la mayora de las subespeciesnombradas del complejo. Las filogenias basadas en estas secuencias muestran que las sub-especies mesoamericanas incluidas en el anlisis son recprocamente monofilticas, pero lassubespecies descritas para Suramrica no reflejan patrones de variacin gentica. Muestras de

la baja Amazona se agrupan con muestras de la Amazona occidental, en vez de agruparse conlas muestras de Colombia y Venezuela, como se esperaba con base en la clasificacin actualde subespecies. Todas las subespecies del complejo estn estrechamente relacionadas entres, separadas por distancias menores que las distancias entre miembros del complejo y otrasespecies deAmazona, y la divisin del complejo en tres especies (A. ochrocephala,A. auropalliata,yA. oratrix) no es apoyada por nuestros datos. Las fechas de divergencia estimadas con los da-tos moleculares sugieren que estos loros llegaron a Mesoamrica despus de la formacin delistmo de Panam y luego expandieron su distribucin y se diversificaron rpidamente. Comoen Mesoamrica, la diversificacin del grupo en Suramrica occuri durante el Pleistoceno,posiblemente como resultado de cambios en la distribucin de hbitats forestales.

Smithsonian Tropical Research Institute, Apartado 2072, Balboa, Repblica de Panam

J R. E1 E B

PHYLOGENY AND BIOGEOGRAPHY OF THEAMAZONAOCHROCEPHALA (AVES: PSITTACIDAE) COMPLEX

The Auk 121(2):318332, 2004

1Present address: Department of Biological Sciencesand Museum of Natural Science, 202 Life Sciences,Louisiana State University, Baton Rouge, Louisiana70803, USA. E-mail: [email protected]

318


2/15

Amazona ochrocephala PhylogenyApril 2004] 319

A historical perspective on relationships amongthese parrots serves as a baseline from which tostudy the interaction of history and ecology thathas led to their contemporary diversity and dis-tribution. Furthermore, a molecular systematic

analysis of the group is of taxonomic interest,because classification of the complex using mor-phological characters has been a taxonomicheadache (Howell and Webb 1995). Finally,identification of conservation units, particularlyof the Mesoamerican subspecies, is important

because their populations have suffered pre-cipitous declines due to habitat loss and the pettrade (Collar et al. 1994).

Study species.The A. ochrocephala complexincludes eleven named subspecies that are

distributed from Mexico to the Amazon basin(Fig. 1). Characters used to identify the varioussubspecies include plumage (in particular, ex-tent and position of yellow on head and thighs,and coloration at the bend of the wing), billand foot pigmentation, and body size (Monroeand Howell 1966, Forshaw 1989, Juniper andParr 1998). However, those characters can varysignificantly, even among individuals from thesame locality (Howell and Webb 1995, Lousadaand Howell 1996, Juniper and Parr 1998, J. R.

Eberhard pers. obs.), in part because of age-related variation (Howell and Webb 1995,

Lousada and Howell 1996). Some taxonomistshave considered the entire complex to con-stitute a single species, A. ochrocephala (e.g.Monroe and Howell 1966, Forshaw 1989), butothers (e.g. Sibley and Monroe 1990, American

Ornithologists Union 1998, Juniper and Parr1998) divide the complex into three species: theYellow-crowned Amazon (A. ochrocephala [A. o.ochrocephala,A. o. xantholaema,A. o. naereri, and

A. o. panamensis]); the Yellow-naped Amazon(A. auropalliata [A. a. auropalliata, A. a. parvipes,and A. a. caribaea]); and the Yellow-headedAmazon (A. oratrix [A. o. oratrix, A. o. tresmariae,

A. o. belizensis, andA. o. hondurensis]). Two otherraces of oratrix are mentioned in the literature:magna, from the Caribbean slope of Mexico,

is not considered valid (Juniper and Parr 1998);and guatemalensis has not been formally de-scribed and is included in belizensis by Juniperand Parr (1998).

Parrots of the ochrocephala complex are gener-ally found below 750 m (Forshaw 1989, Juniperand Parr 1998), inhabiting deciduous woodland,gallery forest, savannah woodland, dry forest,secondary growth along major rivers, and sea-sonally flooded forests (Forshaw 1989, Juniperand Parr 1998). In Middle America, most of

the subspecies appear to be allopatric, thoughthe biological barriers (if any) that separate theranges are not obvious (Juniper and Parr 1998).A zone of contact among yellow-headed, yel-low-naped, and yellow-crowned forms may oc-cur along the Atlantic slope of Central Americafrom Belize to Nicaragua (Lousada and Howell1996). Unfortunately, free-flying birds are veryrare in that region, and we were unable tosecure representative samples for inclusionhere. Given currently available information on

distribution of ochrocephala subspecies in SouthAmerica, no range discontinuity is known

between A. o. ochrocephala (eastern Colombia,Venezuela, Trinidad, Guianas, and northernBrazil) and A. o. naereri (southern Colombia,eastern Ecuador and Peru, western Brazil, andnorthern Bolivia) (Juniper and Parr 1998). Therange of A. o. panamensis is mostly separatedfrom other yellow-crowned forms by the Andes,though it may be continuous with A. o. ochro-cephala in northwestern Venezuela (Juniper and

Parr 1998).Four congeneric species were included as

outgroups in our phylogenetic analyses:A.aes-tiva,A. amazonica,A.farinosa, andA. autumnalis.

F. 1. Distribution of theAmazona ochrocephala com-plex (after Juniper and Parr 1998); taxa sampled for the

present study are indicated in bold type. Distributionof A. aestiva is outlined with the dashed line. Samplelocations are indicated by points, and are numberedto correspond with the sample listing in Table 1.


3/15

E B320 [Auk, Vol. 121

Two of those species,A. aestiva andA. amazonica,are consistently listed next to the A. ochrocephalacomplex in linear taxonomies (Forshaw 1989,Sibley and Monroe 1990, Juniper and Parr 1998).The other two were included because samples

were obtained opportunistically. An additionaloutgroup species, A. barbadensis, was includedin analysis of cytochrome oxidase I (COI) se-quences. That species was of particular interest

becauselike members of the ochrocephala com-plex,A. aestiva, andA. amazonicait is character-ized by yellow plumage on parts of the head.

M

Samples.Where possible, we used vouchered tis-

sue samples from museum frozen-tissue collections.However, because the ochrocephala complex is poorlyrepresented in those collections, many of the sampleswere obtained from field workers, captive breedingfacilities, and pet owners (Table 1). Use of materialfrom captive birds was contingent on availability of in-formation on the sampled individuals geographic ori-gins. Source material included frozen tissues (muscle,liver), blood, feathers (both emerging and full-grown),and small pieces of museum skins (toe and body skin).Three of the outgroup samples (representingA. aestivaandA. barbadensis) were from captive birds of unknown

origin. Collection locations for the ochrocephala samplesincluded here are shown in Figure 1.

Laboratory procedures.Total cellular DNA extrac-tions from frozen tissue, blood, and feather sampleswere done by incubating samples overnight in CTAB

buffer (Murray and Thompson 1980) and proteinase K,followed by a standard phenol-chloroform extractionand dialysis. Extractions from museum skin sampleswere done using the Qiamp kit (Qiagen, Valencia,California), following the protocol outlined by Mundyet al. (1997). Three mitochondrial DNA (mtDNA)fragmentsthe complete ATP synthase 6 and 8 genes

(ATPase6,8), a 622-bp portion of COI, and the completeNADH dehydrogenase 2 (ND2) genewere amplifiedvia the polymerase chain reaction (PCR) for most sam-ples; only COI was analyzed for museum skin samples.In addition, a 694-bp fragment of the cytochrome-b (cytb) gene was sequenced for a subset of the samples (seeTable 1) selected to include one member of each of theclades identified using the other three coding regions.

To amplify and sequence the ATPase6, ATPase8,and ND2 genes, we used primers originally designed

by G. Seutin for studies of Neotropical passerine birds.Primer sequences are given (5 to 3) followed by the

base position of the primers 3 base relative to thedomestic chickens (Gallus gallus) mtDNA sequence(Desjardins and Morais 1990); the H or L indicateswhether the primer is located on the heavy or light

strand, respectively. The primers CO2GQL (GGACA-ATGCTCAGAAATCTGCGG, L8929) and CO3HMH(CATGGGCTGGGGTCRACTATGTG, H9947) wereused to amplify a 1,074-bp fragment that included thefull ATPase6 and ATPase8 genes; along with those,the internal A6PWL (CCTGAACCTGACCATGAAC,

L9245) was used for sequencing the fragment. Forone sample (a molted feather stored at room tem-perature for about eight years), the ATPase region wasamplified in two overlapping pieces, using CO2GQLwith A6VALH (AGAATTAGGGCTCATTTGTGRC,H9436), and A6PWL with CO3HMH. The ND2 genewas amplified with primer pairs METB (CGAAA-ATGATGGTTTAACCCCTTCC, L5233) and TRPC(CGGACTTTAGCAGAAACTAAGAG, H6343), andMETB with ND2LSH(GGAGGTAGAAGAATAGGCY-TAG, H6102). The COI fragment was amplified usingprimers COIa and COIf (Palumbi 1996), and the cyt-b

fragment was amplified and sequenced using primersCB1 and CB3 (Palumbi 1996). Except for those involv-ing museum skins, PCRs were done using AmpliTaq(Perkin-Elmer, Wellesley, Massachuses) and five cy-cles with an annealing temperature of 50C followed

by 30 cycles at 56C.A set of additional COI primers was designed

to amplify and sequence a series of five overlap-ping fragments ranging in size from 106 to 196 bp.Sequences (5 to 3) of those primers, named ac-cording to the position of the primers 5 end, areas follows: L7506 (TAGGGTTYATCGTATGGGCC),

H7523 (ACTGTGAATATGTGGTGGGC), L7628(GACTCGCCACACTACACGG), H7642 (CTCATTT-GATGGTCCCTCCG), L7773 (GTCTCACAGGRATC-GTCC), H7813 (GTATGTGTCGTGTAGGGCA),L7804 (AATAGGTGCCGTCTTTGCC), and H7879(GAATAGGGGGAATCAGTGGG).

That primer set was used in conjunction with prim-ers COIa and COIf to amplify and sequence DNA ex-tracted from museum skin samples. Polymerase chainreaction amplifications of museum skin extracts weredone using AmpliTaq Gold (Perkin-Elmer) in 25-l re-actions and 40 cycles with an annealing temperature of

60C. Those reactions were set up in a UV hood to avoidcontamination.

Amplification products were visualized in agarosegels, and then cleaned and purified using GELase(Epicentre Technologies, Madison, Wisconsin) fol-lowing the manufacturers protocol. PCR fragmentswere then sequenced using either Dyedeoxy or dRho-damine (Applied Biosystems, Foster City, California;Perkin-Elmer) cycle sequencing reactions and an ABI377 automated sequencer. Amplification primers wereused for sequencing both the heavy and light strandsof PCR fragments, and an additional internal primer,

A6PWL, was used to sequence the ATPase region.Three samplestwo A. o. ochrocephala samplesfrom Venezuela and the A. barbadensis sampleweresequenced by M. Rusello (Columbia University, New


4/15


5/15


6/15


York), using museum skin samples from the AmericanMuseum of Natural History. Those sequences wereobtained using the primer set listed above.

To obtain an independent molecule-based estimateof the relationship between A. aestiva and the ochro-cephala complex (see below), a nuclear intron fragment

from the glyceraldehyde-3-phosphate dehydrogenase(Gapdh) gene was sequenced for a subset of samples.The nuclear sequence for A. aestiva was obtained us-ing a DNA extract taken by P. Wainright from a bloodsample (A.aest3; aempts to obtain nuclear sequencesusing extracts of A.aest1 and A.aest2 were unsuccess-ful). Primers GapdL890 and GapdH950 (Friesen etal. 1997) were used for amplification and sequencingof the nuclear fragment. Polymerase chain reactionswere done using AmpliTaq or TaqGold (Perkin-Elmer),

beginning with 5 min at 94C, and then five cycles withan annealing temperature of 50C followed by 30 cycles

at 56C. In some cases, the initial PCR product had to bere-amplified (30 cycles at 56C) prior to sequencing.

All sequences have been deposited in GenBank, un-der accession numbers AY194295AY194327 (ATPase6),AY194328AY194360 (ATPase8), AY194367AY194403(COI), AY194434AY194466 (ND2), AY194404AY194413(cyt b), and AY194425AY194433 (Gapdh).

Sequence analysis.Sequences generated by theautomated sequencer were aligned and proofreadusing SEQUENCHER (version 3.1.1; GeneCodes, AnnArbor, Michigan). The ATPase6,8, COI, and ND2 se-quences were then concatenated for most subsequent

phylogenetic analyses, which were done using PAUP*(version 4.0b8; Swofford 1999). Sequences were com-

bined because the mitochondrial gene regions arefully linked and thus represent a single phylogeneticmarker (a partition-homogeneity test showed that thegene regions were not significantly heterogeneous [P >0.50]). The PAUP* and SEQUENCER 5.0 programs (seeAcknowledgments) were used to calculate descriptivestatistics about nucleotide variation. Analyses thatincluded the museum skin specimens were based onCOI sequences, because we aempted amplificationsof that gene region only from the DNA extracted from

skin samples. The cyt-b sequences, which were ob-tained for a subset of the samples, were used only forestimates of divergence dates (see below).

Phylogenies were reconstructed using neigh-bor-joining (NJ), maximum-parsimony (MP), andmaximum-likelihood (ML) algorithms in PAUP*, anda Bayesian approach as implemented in MRBAYES(Huelsenbeck and Ronquist 2001). Outgroup root-ing was used to root trees. The MP and NJ analyseswere done with all characters weighted equally; MPsearches were also done with weights of 1 and 18assigned to transitions (Ti) and transversions (Tv),

respectively. For analysis of COI sequences, the Ti:Tv weighting was 1:17. Those weightings reflect theTi:Tv ratios determined empirically from the data.Parsimony trees were found using heuristic searches

and random branch addition. Neighbor-joining treeswere obtained using Tamura-Nei distances (Tamuraand Nei 1993).

Substitution model parameters for ML analy-ses in PAUP* were found using MODELTEST 3.1(Posada and Crandall 1998), which uses hierarchical

likelihood-ratio tests to compare the fit of differentnested models of DNA substitution to the data ma-trix. For the ATPase+COI+ND2 data set, MODELTESTsupported the Tamura-Nei model with I = 0.7666 andequal rates at all variable sites. To reduce computingtime, the ML analyses of ATPase+COI+ND2 in PAUP*were done using a reduced data set of 18 taxa (seeTable 1) that included representatives of all majorclades identified by MP and NJ analyses. For the COIdata set, the best fit found by MODELTEST was anHKY model with a Ti:Tv ratio of 30.9904, the propor-tion of invariable sites set to 0.6426, and a gamma

shape parameter of 0.4975. Those parameters werespecified in PAUP* for heuristic ML tree searches and

bootstrapping analyses.For the Bayesian Markov chain Monte Carlo

(MCMC) searches, a general time-reversible modelwas specified, with site-specific variation partitioned

by codon position. Four chains were run for 500,000generations and sampled every 1,000 generations. Inthe ATPase+COI+ND2 analysis, because stationaritywas reached by 15,000 generations, the first 20,000generations were discarded, and the remaining treeswere used to obtain a majority-rule consensus. For

analysis of the COI data set, trees from the first 35,000generations were discarded prior to generating theconsensus tree.

Nodal support was assessed by bootstrap analysisin the MP, NJ, and ML analyses (1,000, 1,000, and 125replicates, respectively), and by posterior probabili-ties in the Bayesian analyses. Posterior probabilitiesindicate percentage of the time that a given cladeoccurs among trees sampled in the Bayesian analyses(Huelsenbeck and Ronquist 2001).

Divergence times among clades in the ochrocephalacomplex were estimated using two different molecular

clock calibrations (no specific calibration exists for thePsiaciformes). A 2% sequence divergence per millionyears (my) calibration (based on restriction-site varia-tion across the mitochondrial genome; see Shields andWilson 1987, Tarr and Fleischer 1993) was used withthe ATPase+COI+ND2 data set. Another set of diver-gence time estimates was calculated using the parrotcyt-b data and a fossil-based molecular clock calibra-tion for cyt b in cranes (Krajewski and King 1996). Thatsecond calibration uses cyt-b maximum-likelihooddistances (as calculated using the DNADIST pro-gram in PHYLIP; Felsenstein 1995), which in cranes

diverge by 0.7%1.7% my1

. For both of those datasets, the assumption of clock-like sequence changewas first tested by using a likelihood ratio test (LRT;Felsenstein 1988) to compare likelihood scores of ML


7/15

E B324 [Auk, Vol. 121

trees found by heuristic searches in PAUP* with a mo-lecular clock enforced versus not enforced. The LRTsfor both the ATPase+COI+ND2 and the cyt-b data setsshowed no statistical difference between trees foundwith or without a clock enforced (P = 0.2286 and P =0.2032, respectively).

To determine whether our data support division ofthe ochrocephala complex into three species, alternativetree topologies were compared using the nonparamet-ric Shimodaira-Hasegawa (S-H) test (Shimodaira andHasegawa 1999) in PAUP*, using RELL bootstrapping.As pointed out by Goldman et al. (2000), that test isapplicable when one of the trees being compared isone selected with reference to the same data beingused in the test. Using both the ATPase+COI+ND2and the COI data sets, we compared the Bayesian treewith a test tree composed of three clades: an oratrixclade (oratrix + belizensis + tresmariae), an auropalliata

clade, and an ochrocephala clade (panamensis + ochro-cephala + naereri + xantholaema). Within each of thoseclades, relationships among subspecies were notresolved; the A.ae1 and beliz1 samples were omit-ted from the analysis (see below). The S-H test wasalso used to compare the Gapdh tree obtained in theparsimony analysis with a tree in which the A. aestiva sample is forced within the ochrocephala clade. Forthat test, we used likelihood parameters suggested

by MODELTEST (Posada and Crandall 1998) for theGapdh data (F81, with equal rates for all sites and noinvariable sites).

R

A total of 2,515 bp of coding sequence(ATPase6, 684 bp; ATPase8, 168 bp; COI, 622 bp;and ND2, 1,041 bp) was obtained for each parrotindividual, except museum skin specimens, forwhich only the COI fragment was sequenced.We also sequenced a 694-bp fragment of cyt bfor 10 of the parrots (see Table 1). Overlap be-

tween sequences generated using heavy- andlight-strand primers averaged approximately72% for the ATPase coding region, 86% for theCOI fragment, 58% for the ND2 gene, and 90%for the cyt-b fragment; no nucleotide differenceswere found between overlapping complemen-tary sequences.

No indels or stop codons were observed,as expected for protein-coding mitochondrialregions. In the ATPase+COI+ND2 data set, 378(15.0%) base positions are variable and 205

(8.2%) are parsimony-informative. Sequencevariability differed across codons (Table 2).Third-position changes are the most common(49.5% of variable sites), whereas 38.9% of

changes occur at first position, and 11.6% at thesecond codon position. Transitions outnumbertransversions by 18.6 to 1, averaging over allpairwise comparisons. In the COI data set,which includes the additional skin samples, 95

(15.3%) characters are variable, and 49 (7.9%)are parsimony informative. In that region, thevast majority (92.6%) of changes occur at thirdposition, and the remainder occur at first posi-tion. All base changes in COI are synonymous.In the cyt-b fragment, most changes (83.5%) oc-cur at third position, with 15.2% at first position,and the remaining 1.3% at second position.

According to the phylogenies generated byall of the tree-reconstruction algorithms usingthe ATPase+COI+ND2 sequences, the ochro-

cephala complex forms a well-supported clade(bootstrap values of 100% in MP, NJ, and MLanalyses, and 100% posterior probability valuein the Bayesian tree). Topology of the Bayesiantree (Fig. 2) is nearly identical to that of the treesfound by PAUP* using parsimony, distance,and maximum-likelihood algorithms. The MPand NJ trees differ only slightly in the arrange-ment of the South American clade that includessamples A.aest1, ochro1, ochro2, and nater3.The Middle American subspecies (oratrix, tres-

mariae, belizensis, auropalliata, and panamensis)consistently form reciprocally monophyleticclades that are strongly supported by bootstrapanalysis, with bootstrap values 99% in the MP

T 2. Nucleotide variability at different codonpositions in the ATPase6, ATPase8, COI, and ND2genes of Amazona. Both ingroup and outgroupspecies were included in the calculations.

Codon Number Percentage

Region position Total bp variable variableATPase6 all 684 100 14.6

1 228 25 11.02 228 7 3.13 228 68 29.9

ATPase8 all 168 29 17.31 56 7 12.52 56 5 8.93 56 17 30.4

COI all 622 95 15.31 208 7 7.42 207 0 0.0

3 207 88 92.6ND2 all 1041 167 16.0

1 347 32 9.22 347 25 7.23 347 110 31.7


8/15


and NJ trees, and 95% in the ML tree, andwith posterior probability values of 100% inthe Bayesian analysis. Down-weighting transi-tions in the MP and NJ analyses of the completedata set resulted in trees with almost no reso-lution within the ochrocephala complex clade.

Transversion weighting of the COI data set alsoresulted in trees with decreased resolution, butno results were in conflict with tree topologiesfound using the equally weighted data. A singlebelizensis sample (beliz1) consistently clusteredwith the oratrix samples, but because the beli-zensis samples were from wild-caught captive

birds for which exact location data were notavailable, and because morphological differ-ences used to distinguish oratrix and belizensisare fairly subtle, we hesitate to overinterpret

that result, assuming instead that the bird wasmisidentified.The samples representing named subspe-

cies from South America do not form clades

(see Fig. 2). Individuals representing naereri(three localities in Bolivia), xantholaema (Ilha deMaraj, at the mouth of the Amazon River), andochrocephala (Xingu River, Brazil) intermingle toform a well-supported clade that does not in-

clude the remaining ochrocephala sample, whichis from Colombia. The Colombian ochrocephalasample falls at the base of the complex and isseparated from other South American samples

by almost 2% (uncorrected) sequence diver-gence. The separation of the Colombian ochro-cephala sample from the other South Americansamples was further investigated by sequencingthe COI fragment for three additional samplestaken from museum skins: one from a secondlocality in Colombia, and two others from dif-

ferent localities in Venezuela (see Table 1 andFig. 1). The COI sequence was also obtainedfor an A. barbadensis sample. That was done inresponse to placement of A. aestiva within theochrocephala complex (see below). LikeA. aestiva,

A. barbadensis has yellow plumage on the headthat is similar in hue and extent to that found inmembers of the ochrocephala complex.

Phylogenetic analysis of the COI sequences,which includes the additional samples fromColombia and Venezuela, again demonstrates

a clear phylogenetic separation of ochrocephalasamples from northern South America versusthose from central South America (Fig. 3). TheColombian and Venezuelan samples form anmtDNA clade that is sister to the remainingSouth American individuals from Brazil andBolivia, with moderate (79%) support in theBayesian analysis (Fig. 3). Bootstrap values forthe same node were 88% in an NJ tree, 67% inan MP tree, and 64% in an ML tree (trees notshown). Overall, the COI tree (Fig. 3) is not as

well resolved as one based on the full mitochon-drial data set (Fig. 2), presumably because theCOI data set is much smaller. Together, the twotrees indicate that parrots from the Amazon,northern South America, and Mesoamerica forma polytomy uniting three lineages of equivalentevolutionary distinctiveness. The COI data alsoshow that A. barbadensis is a distinct species,and not particularly closely related to the ochro-cephala complex.

The unexpected placement of A. aestiva

within the ochrocephala complex was confirmedusing a second extraction from a differentfeather of the same bird, and using a secondsample (A.aest2) obtained from the NMNH (the

F. 2. Phylogram of theAmazona ochrocephala com-plex obtained in a Bayesian analysis of the mitochon-drial ATPase6,8, COI, and ND2 sequences (2,514 bp)obtained in the present study. The tree represents a50% majority-rule consensus of 479 trees generated

by MRBAYES during an MCMC search (see text), andnumbers at nodes are posterior probabilities. The tree

was rooted using A. amazonica, A. autumnalis, and A.farinosa as outgroups. Current subspecies groupings(e.g. Forshaw 1989, Juniper and Parr 1998) are indi-cated to the right of sample names.


9/15

E B326 [Auk, Vol. 121

COI sequences of A.aest1 and A.aest2 differedby a single base substitution). We are confidentthat those results are not due to species mis-identification, because A. aestiva can be unam-

biguously distinguished fromA. ochrocephala by

the presence of blue plumage on the forehead(species identity of A.aest1 was confirmed withphotographs, A.aest2 was identified by NMNHpersonnel, and A.aest3 was identified by staff atthe University of Georgia Veterinary School).The relationship between A. aestiva and theochrocephala group was explored further usingthe Gapdh nuclear sequences. A total of 404 bpwere sequenced for an A. aestiva sample, fourmembers of the ochrocephala complex (repre-senting belizensis, panamensis, ochrocephala, and

auropalliata), as well as four other Amazonaspecies (A. farinosa, A. amazonica, A. viridigena-lis, and A. autumnalis). The Gapdh sequenceswere easily aligned, with only a single one-nucleotide indel shared by the A. autumnalisandA. viridigenalis samples. Of the 404 bp in the

Gapdh data set, 28 are variable and only two areparsimony-informative; 2 bp changes separate

Amazona aestiva and the ochrocephala complex.Parsimony and distance-based analyses of thenuclear data produced trees with consistent to-

pologies in which A. aestiva falls outside of theochrocephala clade (parsimony tree shown in Fig.4), but an S-H test shows that those topologiesare not significantly different from one in which

A. aestiva is forced within the ochrocephala clade(P = 0.15).

Short internode distances uniting MiddleAmerican subspecies indicate a rapid geo-graphic expansion across Mesoamerica andrecent diversification. The genetic distances(uncorrected p distance, calculated using the

ATPase+COI+ND2 data set) between individualsof different named Mesoamerican subspecies inthe ochrocephala complex are small, rangingfrom 0.007 (belizensis vs. auropalliata) to 0.016

F. 3. Phylogram of the Amazona ochrocephala com-plex obtained in a Bayesian analysis of the COI data-set (622 bp). The tree represents a 50% majority-ruleconsensus of 464 trees generated by MRBAYES duringan MCMC search (see text), and numbers at nodes areposterior probabilities. The tree was rooted using A.

amazonica,A. autumnalis,A. barbadensis, andA. farinosa asoutgroups. Current subspecies groupings (e.g. Forshaw1989, Juniper and Parr 1998) and geographic origin ofsamples are indicated to the right of sample names.

F. 4. Phylogeny showing the relationship betweenfour members of theAmazonaochrocephala complex andseveral Amazona species based on nuclear Gapdh se-quences (404 bp). The tree shown is one of the two mostparsimonious trees found in an exhaustive maximumparsimony search by PAUP*, and is identical to thestrict consensus of the two most parsimonious trees.

The phylogeny was rooted using A. autumnalis as theoutgroup; the same topology is obtained using mid-point rooting. Numbers at the nodes indicate bootstrapvalues obtained in 1,000 bootstrap replicates.


10/15


(panamensis vs. tresmariae). Genetic distancesbetween Middle American and South Americanochrocephala subspecies average 0.014, whereasdistances between members of the ochrocephalagroup and outgroup species range from 0.059

(A. amazonica vs. naereri) to 0.076 (A. farinosavs. naereri). Assuming that Amazona mtDNAsrate of sequence divergence is approximately2% Ma1, as in a variety of other birds, such asgeese (Shields and Wilson 1987) and honey-creepers (Tarr and Fleischer 1993), the abovedistances indicate that lineages sampled withinthe ochrocephala complex shared a common an-cestor 1.2 million years ago (mya).

Maximum-likelihood distances calculatedusing DNADIST and the cyt-b data range from

0.003 (oratrix vs. auropalliata) to 0.039 (tresmariaevs. xantholaema) within the ochrocephala com-plex, whereas distances to A. amazonica rangefrom 0.088 (A. amazonica vs. belizensis) to 0.100(A. amazonica vs. tresmariae). According to thecrane calibration of Krajewski and King (1996),the cyt-b ML distances among taxa in the ochro-cephala complex suggest that diversification ofthe group occurred during the past 0.25.6 Ma.That interval is broad, but consistent with cal-culations based on the ATPase+COI+ND2 data;

taken together, estimates indicate that the ochro-cephala complex diversified recently, probablywithin the past 2 Ma.

The ATPase+COI+ND2 and COI data setsstrongly support reclassification of the SouthAmerican ochrocephala subspecies. SouthAmerican ochrocephala parrots separate alonggeographic rather than currently described sub-species lines. The mean genetic distance betweenthe northern and central South American birdsthat we have examined is 0.02 (uncorrectedp dis-

tance), indicating a split 1.0 mya under the 2% di-vergence per my calibration (the mean cyt-b MLdistance is 0.026, yielding an estimate of 1.53.7mya, with the cyt b 0.7%1.7% Ma1 calibration).Results of the comparison of alternative treetopologies do not support division of the ochro-cephala complex into three species. According tothe S-H test, the three-species tree is rejected atthe P < 0.001 level when compared to the log-likelihood of the Bayesian tree (Fig. 2).

D

Molecular systematics of the ochrocephalacomplex.The molecular data presented here

provide an informative counterpoint to themorphological characters that have previously

been used to classify members of the ochroceph-ala parrot complex. Consistent with taxonomicarrangements that group all of the subspecies

under a single species name (e.g. Forshaw1989), sequence data indicate that membersof the complex are very closely related, andmuch more closely related to one another thanto other surveyed Amazona species. Molecularphylogenetic analysis supports the monophylyof named subspecies from Middle America(oratrix, tresmariae, belizensis, auropalliata, and

panamensis), but not of the South Americanones (ochrocephala, naereri, xantholaema). Wenote that our analysis did not include samples

from the Caribbean slope of Honduras andNicaragua, where there may be a contact zonebetween several different subspecies (Lousadaand Howell 1996).

Subdivision of the ochrocephala complex intothree species is not supported by phyloge-netic analysis of parrot mtDNA genes. In thecomparison of alternative tree topologies, theATPase+COI+ND2 and COI data sets rejectedthe three-species topology that reflects divisionof the complex into A. ochrocephala, A. auropal-

liata, and A. oratrix. Although plumage char-acters support subdivision of the ochrocephalacomplex into three species, plumage paernsare quite variable and appear to be very labile.For example, an examination of museum skinsat the American Museum of Natural History by

J.R.E. found an ochrocephala specimen (AMNH133032) with a yellow feather on its nape, anda naereri skin (AMNH 255153) with a yellowfeather on its throat; in both of those subspe-cies, yellow feathers are typically confined to

the forehead and crown (Forshaw 1989, Juniperand Parr 1998). Similarly, Monroe and Howell(1966) note a well-documented instance of anindividual captive parrot that had a yellow-crowned plumage paern for 10 years, andaerward developed a yellow nape in additionto the crown.

Under the phylogenetic species concept(Cracra 1983), the mtDNA sequence datawould support elevating the Middle Americansubspecies included here to species status, and

regrouping the South American taxa into twospecies,A. ochrocephala (ochrocephala from north-ern South America) andA. naereri (ochrocephalafrom Amazonia, plus the currently recognized


11/15

E B328 [Auk, Vol. 121

naereri and xantholaema), representing themtDNA clades from northern and central SouthAmerica. Alternatively, the data support rec-ognition of a single monophyletic species, A.ochrocephala (with South American subspecies

revised as described here), which conforms tothe opinions of Monroe and Howell (1966) andForshaw (1989).

Because of the low levels of nuclear sequencedivergence among members of the ochrocephalacomplex, reciprocal monophyly of the mtDNAlineages that form the Middle Americansubspecies included here is not suffi cient to con-clusively identify them as evolutionarily signifi-cant units (ESUs), following criteria proposed

by Moritz (1994). However, the morphological

differences (e.g. yellow nape and lack of red atthe bend of the wing in auropalliata, or extensiveyellow on the head in oratrix and tresmariae)probably reflect meaningful divergences at nu-clear loci, even if such variation alone is not di-agnostic for some of the subspecies. Regardlessof the taxonomic nomenclature adopted forthe ochrocephala complex, the combination ofmtDNA data and plumage variation amongthe Mesoamerican members indicates that eachsubspecies should be considered distinct units

for conservation purposes.Analyses of the mitochondrial and nuclearsequence data produce conflicting results withrespect to the position of A. aestiva relative tothe ochrocephala complex. The Gapdh phylogeny(Fig. 4), in which A. aestiva falls outside of theochrocephala clade, is in agreement with the mor-phological characters that can be used to distin-guish the two. However, low levels of variationin the Gapdh sequence data make it impossibleto conclusively reject the hypothesis supported

by mtDNA analysis, which places A. aestivawithin the ochrocephala complex. Agreement be-tween the Gapdh analysis and the morphologi-cal distinctiveness of A. aestiva suggests to usthat the mtDNA data do not accurately reflect

A. aestivas phylogenetic history, but additionalnuclear sequence data are necessary to resolvethat issue. Our mtDNA data placing A. aestivawithin the ochrocephala clade are consistent withthose of Rusello and Amato (2004), who used

both mitochondrial and nuclear sequences in

a phylogenetic analysis of Amazona. However,they also found A. barbadensis to fall withinthe ochrocephala complex, whereas we did not.When Rusello and Amatos (2004) nuclear data

were analyzed separately, they recovered ayellow-headed clade, which presumably in-cluded A. aestiva, though a tree is not shown.The distributions of A. ochrocephala and A. aes-tiva are largely separate, although there is some

overlap in their ranges (see Fig. 1). That regionof overlap could permit hybridization betweenA. ochrocephala and A. aestiva, possibly leadingto differential mtDNA introgression, whichcould explain the mitochondrial sequencedata. Such introgression has been reported instudies of a range of animal taxa (Ferris et al.1983, Powell 1983, Tegelstrom 1987, Dowlinget al. 1989, Lehman et al. 1991, Boyce et al.1994, Quesada et al. 1995, Rohwer et al. 2001).Although no hybridization between those spe-

cies has been noted in the wild,Amazona speciesare known to hybridize in captivity (Nichols1980). Another possible explanation is that mi-tochondrial primers preferentially amplified anancestral mitochondrial pseudogene frozenin the nuclear genome of theA. aestiva samples.However, that does not seem likely, given thesize of the pseudogeneor the repeated natureof translocationrequired to explain the coin-cident paern observed in the four differentmitochondrial gene regions.

Biogeography of the ochrocephala complex.Recent work by Rusello and Amato (2004) in-dicates that the ochrocephala complex arose froma South American ancestor, and our phyloge-netic hypothesis (Fig. 2) is consistent with theiranalysis. Two of the three principal ochrocephalamtDNA clades are confined to South America,and the level of intraclade mtDNA divergenceobserved in the Amazonian lineages (Brazilianand Bolivian parrots) is greater than the ge-netic distances observed between individuals

representing the relatively large and taxonomi-cally diverse sample of Mesoamerican parrots.Failure to reject a molecular clock promptedour application of available avian mtDNA clockcalibrations, which indicated that the threeprincipal ochrocephala mtDNA clades formedcontemporaneously in the Pliocene or earlyPleistocene.

The short genetic distances among MiddleAmerican subspecies, and the close relationship

between the Middle and South American lin-

eages, suggests that Middle America was colo-nized by ochrocephala parrots well aer the riseof the isthmus of Panama 3.5 mya (Coates 1997).Both the short internodes and the short terminal


12/15


branch lengths of Middle American subspeciesindicate that the area was colonized relativelyquickly and recently. A similar paern of rapidexpansion across the Mesoamerican landscape,followed by in situ phylogenetic diversifica-

tion, has been demonstrated for freshwaterfish (Bermingham and Martin 1998, Perdiceset al. 2002, G. Reeves and E. Bermingham un-publ. data) and howler monkeys (Corts-Ortizet al. 2003). The branching order of the tree inFigure 2 is consistent with a south-to-north step-ping-stone paern of colonization for MiddleAmerica, a dispersal paern that has been rec-ognized in a number of organisms (see Savage1982).

The allopatry of Middle American subspe-

cies, and their diversification, may be aribut-able to habitat preferences. These parrots arelowland birds, generally found in relativelydry or deciduous forests, forest edges and gal-lery forest, and savannahs; in South America,they appear to avoid continuous moist forest,perhaps being replaced byA. amazonica in thosehabitats (Juniper and Parr 1998). Expansionof ochrocephala parrots across Middle Americamay have occurred during glacial periods ofthe Pleistocene, when dry forest and savan-

nah vegetation were probably more continu-ous over much of the region (Colinvaux 1997).Subsequent warmer and weer periods wouldhave permied an extension of wet forests(Colinvaux 1997), possibly leading to fragmen-tation of the drier habitat preferred by parrotsof the ochrocephala complex. In addition, G.Reeves and E. Bermingham (unpubl. data) haveproduced a model suggesting that phylogenetic

breaks between lineages can be maintained inthe absence of discrete barriers to gene flow,

owing to inertia resulting from behavioral inter-actions (repulsion) or demographic interactionsresulting from differences in population sizes ofresident and immigrant populations.

The strong phylogeographic structure inMesoamerican ochrocephala parrots stands incontrast to the apparent lack of geographicstructure in parrots collected across a regionextending from the mouth of the Amazonto Bolivia and Peru, a distance of >2,000 km.Contrary to current subspecies descriptions,

our results clearly show that parrots describedas A. o. ochrocephala and A. o. xantholaema, fromRio Xingu and Ilha de Maraj in the the lowerAmazon River, are closely allied with parrots

recognized as A. o. naereri from westernBrazil, Bolivia, and Peru. Recent population-genetic analysis of mahogany (Swietenia macro-

phylla) distributed along the southern arc of theAmazon also failed to demonstrate strong geo-

graphic subdivision in the region, in contrast toMesoamerican Swietenia macrophylla, which wasstrongly structured into four regional popula-tion groups (Novick et al. 2003). The compellingdata for geographically structured populationsof Middle American freshwater fish, howlermonkeys, and mahogany may indicate a geo-graphic history of population expansion andsubdivision for many groups that is consider-ably more dynamic than that of Amazona par-rots. Differences in regional histories such as the

one documented here for South American ochro-cephala parrots versus their Middle Americancounterparts might also provide a partialexplanation for the reduction in beta diversity(species turnover) that characterizes westernAmazonian rainforest tree communities in com-parison to those in Panama (Condit et al. 2002).

The phylogenetic break between ochroceph-ala parrots from northern and central SouthAmerica has not been suggested by previoustaxonomic work but is well supported by the

molecular data (see Figs. 2 and 3). The mostobvious landscape feature that coincides withdivergence between the two South Americanlineages is the Amazon River, which runs be-tween the geographic areas represented by thesamples in the two South American clades. Thisis consistent with the riverine barrier hypothesis(Wallace 1853, Capparella 1988), which arguesthat large river courses impede gene flow be-tween populations on opposite banks, leadingto speciation. Nevertheless, our support for the

riverine barrier hypothesis is weak, and furthersampling on both the north and south banksalong the Amazon River would permit a muchstronger test of the hypothesis. Alternatively,the genetic break could reflect past habitat dis-continuities, such as changes in forest cover re-sulting from climatological cycles (Haffer 1969)or isolation of Guiana Shield populations due tosea-level changes (Nores 1999).

Alternatively, the genetic break could reflectpast habitat discontinuities (e.g. changes in

forest cover resulting from Pleistocene glacialcycles; Haffer 1969), with the Amazon beinga secondary barrier, halting the expansion oflineages from their centers of origin. Observed


13/15

E B330 [Auk, Vol. 121

genetic distances between members of thetwo South American clades, which imply thatdivergences occurred during the Pleistocene,would be consistent with that hypothesis. Thenorthern samples could also reflect isolation of

an ochrocephala lineage on the Guiana shield ofnorthwestern South America, which may havebeen isolated during 100 m sea-level risesthat occurred during the late Tertiary and thePleistocene (Nores 1999).

Of the three named South American subspe-cies, xantholaema is the most different morpho-logically (more extensive yellow on the head),has vocalizations that differ from mainlandochrocephala (C. Yamashita and P. Martuschellipers. comm.), and is somewhat isolated on Ilha

de Maraj. However, according to our sequencedata, xantholaema clusters closely with a naererisample from Bolivia. That lack of divergenceis also shown by phylogenetic analysis of se-quence data from the rapidly evolving controlregion, using samples from the present study(J. R. Eberhard and E. Bermingham unpubl.data) and additional samples from wild-caughtxantholaema (C. Y. Miyaki pers. comm.). Theremay be suffi cient gene flow between islandand mainland populations to prevent genetic

divergence of xantholaema from its relatives.Alternatively, xantholaema may have divergedtoo recently for mtDNA to be a useful marker,

but suffi ciently long ago to permit divergence inplumage and vocalization paerns.

A

M. Rusello kindly provided three of the COI se-quences. We thank G. Amato, M. Gonzlez, J.Groth,

J. Hunt, and I. Lovee for technical suggestions andassistance, and M. Leone for assistance in process-

ing our import permits. G. Seutin designed manyof the primers used in the analysis of mitochondrialregions. We thank R. M. Zink and two anonymousreviewers for comments on an earlier version of thismanuscript. Thanks also to the American Museum ofNatural History for permission to examine specimensin their collection, and to C. Miyaki for permissionto cite unpublished data. This study would not have

been possible without the samples contributed by thefollowing individuals and institutions: Academy ofNatural Sciences, Philadelphia; American Museumof Natural History; Collection of Genetic Resources

at the Louisiana State University Museum of NaturalScience; E. Enkerlin, J. J. Gonzlez, and ClaudiaMacas (Tecnolgico de Monterrey, Mexico); H. deEspinoza; Fundacin ARA; O. Habet and S. Matola

(Belize Zoo); Loro Parque; S. Rodden; U.S. NationalMuseum of Natural History; P. Wainright; M. J. West-Eberhard; and T. Wright. Tissue collection of A. [o.]ochrocephala samples on the Rio Xingu by G. R. Graveswas supported by the Academia Brasileira de Ciencias,through a grant from Electronorte administered by P.

E. Vanzolini, and by the Smithsonians InternationalEnvironmental Sciences Program NeotropicalLowland Research Program. Funding was provided byawards from the American Museum of Natural HistoryFrank M. Chapman Memorial Fund and the AmericanOrnithologists Union to J.R.E., a Smithsonian TropicalResearch Institute postdoctoral fellowship to J.R.E.,and the Smithsonian Institutions Molecular EvolutionProgram. For information on the SEQUENCHER 5.0program, see hp://nmg.si.edu/Sequencer.html

L C

A O U. 1998. Check-list of North American Birds, 7th ed. AmericanOrnithologists Union, Washington, D.C.

B, E., A. P. M. 1998.Comparative mtDNA phylogeography ofNeotropical freshwater fishes: Testing sharedhistory to infer the evolutionary landscape oflower Central America. Molecular Ecology 7:499517.

B, T. M., M. E. Z, C. F. A.1994. Mitochondrial DNA in the bark weevils:

Phylogeny and evolution in the Pissodes strobispecies group (Coleoptera: Curculionidae).Molecular Biology and Evolution 11:183194.

C, A. P. 1988. Genetic variation inNeotropical birds: Implications for the spe-ciation process. Pages 16581664 in Acta XIXCongressus Internationalis Ornithologici (H.Ouellet, Ed.). National Museum of NaturalSciences, Oawa, Ontario.

C, F. M. 1917. The distribution of bird-lifein Colombia: A contribution to a biological sur-vey of South America. Bulletin of the American

Museum of Natural History, no. 36.C, A. G. 1997. The forging of Central America.

Pages 137 in Central America: A Natural andCultural History (A. G. Coates, Ed.). YaleUniversity Press, New Haven, Connecticut.

C, P. 1997. The history of forests on theisthmus from the ice age to the present. Pages123136 in Central America: A Natural andCultural History (A. G. Coates, Ed.). YaleUniversity Press, New Haven, Connecticut.

C, N. J., M. J. C, A. J. S.1994. Birds to Watch 2: The World List of

Threatened Birds. BirdLife International,Washington, D.C.

C, R., N. P, E. G. L, J. C, J.T, R. B. F, P. N, S. A, R.


14/15


V, G. V, H. C. M-L, E.L, S. P. H. 2002. Beta-diversity intropical forest trees. Science 295:666669.

C-O, L., E. B, C. R, E.R-L, I. S, M. R-G. 2003. Molecular systematics and

biogeography of the Neotropical monkeygenus, Alouaa. Molecular Phylogenetics andEvolution 26:6481.

C, J. 1983. Species concepts and speciationanalysis. Current Ornithology 1:159187.

C, J. 1985. Historical biogeography andpaerns of differentiation within the SouthAmerican avifauna: Areas of endemism.Ornithological Monographs 36:4984.

C, J., R. O. P. 1988. Paerns andprocesses of diversification: Speciation andhistorical congruence in some Neotropical

birds. Evolution 42:603620.D, P., R. M. 1990. Sequence and

gene organization of the chicken mitochon-drial genome. Journal of Molecular Biology212:599634.

D, T. E., G. R. S, W. M. B.1989. Reproductive isolation and introgression

between Notropis cornutus and Notropis chryso-cephalus (family Cyprinidae): Comparisons ofmorphology, allozymes, and mitochondrialDNA. Evolution 43:620634.

F, J. 1988. Phylogenies from molecular

sequences: Inference and reliability. AnnualReview of Genetics 22:521565.

F, J. 1995. PHYLIP (PylogeneticInference Package) 3.57c. Published by theauthor, University of Washington, Seale.Available for download at evolution.gs.washington.edu/phylip.html.

F, S. D., R. D. S, C.-M. H, J. T.N, U. R, A. C. W. 1983.Flow of mitochondrial DNA across a spe-cies boundary. Proceedings of the NationalAcademy of Sciences USA 80:22902294.

F, J. M. 1989. Parrots of the World, 3rd ed.Landsdowne Editions, Melbourne, Australia.F, V. L., B. C. C, H. E. W,

T. P. B. 1997. Intron variation in MarbledMurrelets detected using analyses of single-stranded conformational polymorphisms.Molecular Ecology 6:10471058.

G, N., J. P. A, A. G. R.2000. Likelihood-based tests of topologies inphylogenetics. Systematic Biology 49:652670.

H, J. 1969. Speciation in Amazonian forestbirds. Science 165:131137.

H, J. 1974. Avian Speciation in TropicalSouth America. Publications of the NuallOrnithological Club, no. 14.

H, J. 1985. Avian zoogeography of the

Neotropical lowlands. Ornithological Mono-graphs 36:113146.

H, S. N. G., S. W. 1995. The Birds ofMexico and Northern Central America. OxfordUniversity Press, New York.

H, J. P., F. R. R. 2001.

MRBAYES: Bayesian inference of phylogeny.Bioinformatics 17:754755.

J, T., M. P. 1998. Parrots: A Guideto Parrots of the World. Yale University Press,New Haven, Connecticut.

K, C., D. G. K. 1996. Moleculardivergence and phylogeny: Rates and paernsof cytochrome b evolution in cranes. MolecularBiology and Evolution 13:2130.

L, N., A. E, K. H, D. M,R. O. P, P. J. P. G, R. K. W.1991. Introgression of coyote mitochondrial

DNA into sympatric North American graywolf populations. Evolution 45:104119.

L, S. A., S. N. G. H. 1996.Distribution, variation, and conservation ofYellow-headed Parrots in northern CentralAmerica. Cotinga 5:4653.

M, B. L., J., T. R. H. 1966.Geographic variation in Middle Americanparrots of the Amazona ochrocephala complex.Occasional Papers of the Museum of Zoology,no. 34. Louisiana State University, BatonRouge.

M, C. 1994. Defining evolutionary signifi-cant units for conservation. Trends in Ecologyand Evolution 9:373375.

M, N. I., P. U, D. S. W. 1997.Skin from feet of museum specimens as a non-destructive source of DNA for avian genotyp-ing. Auk 114:126129.

M, M. G., W. F. T. 1980. Rapidisolation of high molecular weight plant DNA.Nucleic Acids Research 8:43214325.

N, T. D. 1980. An unusual amazon hybrid.ICBP Technical Publication 1:169170.

N, M. 1999. An alternative hypothesis for theorigin of Amazonian bird diversity. Journal ofBiogeography 26:475485.

N, R. R., C. W. D, M. R. L, C. N,A. C, E. B. 2003. Geneticstructure of Mesoamerican populations of

big-leaf mahogany (Swietenia macrophylla) in-ferred from microsatellite analysis. MolecularEcology 12:28852893.

P, S. R. 1996. Nucleic acids II: The poly-merase chain reaction. Pages 205247 inMolecular Systematics, 2nd ed. (D. M. Hillis,

C. Moritz, and B. K. Mable, Eds.). SinauerAssociates, Sunderland, Massachuses.P, A., E. B, A. M,

I. D. 2002. Evolutionary history of the


15/15

E B332 [Auk, Vol. 121

genus Rhamdia (Teleostei: Pimelodidae) inCentral America. Molecular Phylogenetics andEvolution 25:172189.

P, D., K. A. C. 1998.MODELTEST: Testing the model of DNA sub-stitution. Bioinformatics 14:817818.

P, J. R. 1983. Interspecific cytoplasmic geneflow in the absense of nuclear gene flow:Evidence from Drosophila. Proceedings ofthe National Academy of Sciences USA 80:492495.

Q, H., R. W, D. O. F. S. 1995.Differential introgression of mitochondrial DNAacross species boundaries within the marinemussel genusMytilus. Proceedings of the RoyalSociety of London, Series B 262:5156.

R, S., E. B, C. W. 2001.Plumage and mitochondrial DNA haplo-

type variation across a moving hybrid zone.Evolution 55:405422.

R, M. A., G. A. 2004. A molecu-lar phylogeny of Amazona : Implications forNeotropical parrot biogeography, taxonomyand conservation. Molecular Phylogeneticsand Evolution 30:421437.

S, J. M. 1982. The enigma of the CentralAmerican herpetofauna: Dispersals or vicari-ance? Annals of the Missouri Botanical Garden69:464547.

S, G. F., A. C. W. 1987. Calibration

of mitochondrial DNA evolution in geese.Journal of Molecular Evolution 24:212217.

S, C. G., B. L. M, J. 1990.Distribution and Taxonomy of Birds of theWorld. Yale University Press, New Haven,Connecticut.

S, H., M. H. 1999. Multiplecomparisons of log-likelihoods with applica-

tions to phylogenetic inference. MolecularBiology and Evolution 16:11141116.

S, B. B., J. H. 1978. Speciation pat-terns in the Amazonian forest biota. AnnualReview of Ecology and Systematics 9:497518.

S, D. L. 1999. PAUP* PhylogeneticAnalysis Using Parsimony (*and OtherMethods), version 4. Sinauer Associates,Sunderland, Massachuses.

T, K., M. N. 1993. Estimation of thenumber of nucleotide substitutions in the con-trol region of mitochondrial DNA in humans

and chimpanzees. Molecular Biology andEvolution 10:512526.

T, C. L., R. C. F. 1993. MitochondrialDNA variation and evolutionary relationshipsin the Amakihi complex. Auk 110:825831.

T, H. 1987. Transfer of mitochondrialDNA from the northern red-backed vole(Clethrionomys rutilus) to the bank vole (C.

glareolus). Journal of Molecular Evolution 24:218227.

W, A. R. 1853. A Narrative of Travels on theAmazon and Rio Negro. Reeves, London.

Associate Editor: R. Zink

Documents

EB, Avian Amazona Auk04