Molecular Systematics of Mouse Opossums (Didelphidae: Marmosa): Assessing Species Limits using Mitochondrial DNA Sequences, with Comments on Phylogenetic Relationships and Biogeography

2010·2.3 MB·English

by  Eliécer E. Gutiérrez

PUBLISHED BY THE AMERICAN MUSEUM OF NATURAL HISTORY
CENTRAL PARK WEST AT 79TH STREET, NEW YORK, NY 10024
Number 3692, 22 pp., 4 figures, 5 tables June 25, 2010
Molecular Systematics of Mouse Opossums
(Didelphidae: Marmosa): Assessing Species Limits
using Mitochondrial DNA Sequences, with
Comments on Phylogenetic Relationships
and Biogeography
ELIE´CER E. GUTIE´RREZ,1,2 SHARON A. JANSA,3 AND ROBERT S. VOSS4
ABSTRACT
The genus Marmosa contains 15 currently recognized species, of which nine are referred to the
subgenus Marmosa, and six to the subgenus Micoureus. Recent revisionary research based on
morphologicaldata,however,suggeststhatthesubgenusMarmosaismorediversethanthecurrently
accepted taxonomy indicates. Herein we report phylogenetic analyses of sequence data from the
mitochondrial cytochrome-b gene representing 12 of the 14 morphologically defined taxa recently
treatedasvalidspeciesofMarmosa(Marmosa)intheaforementionedrevisionarywork.Thesedata
provideabasisfortestingthemonophylyofmorphologicallydefinedtaxainthesubgenusMarmosa,
and they afford the first opportunity to assess phylogenetic relationships among the majority of
species currently referred to the genus. Ten of 11 species of Marmosa (Marmosa) represented by
multiplesequencesinouranalyseswererecoveredasmonophyletic.Incontrast,oursamplesofM.
mexicanawererecoveredastwodeeplydivergenthaplogroupsthatwerenotconsistentlyassociatedas
sistertaxa.Amongotherresults,ouranalysessupporttherecognitionofM.isthmicaandM.simonsi
asspeciesdistinctfromM.robinsoni,andtherecognitionofM.macrotarsusandM.waterhouseias
speciesdistinctfromM.murina.Thevalidityofthreeotherspecieslongrecognizedasdistinct(M.
rubra,M.tyleriana,andM.xerophila)isalsoclearlysupportedbyourresults.Althoughcytochrome-b
sequencedataarenotconsistentlyinformativeaboutinterspecificrelationshipsinthisstudy,wefound
1Department of Biology, City College of New York, City University of New York, New York, NY 10031
([email protected]).
2TheGraduateSchoolandUniversityCenter,CityUniversityofNewYork,NewYork,NY10016.
3DepartmentofEcology,Evolution,andBehavior;andJ.F.BellMuseumofNaturalHistory.UniversityofMinnesota,
1987UpperBufordCircle,St.Paul,MN55108([email protected]).
4DivisionofVertebrateZoology,AmericanMuseumofNaturalHistory([email protected]).
CopyrightEAmericanMuseumofNaturalHistory2010 ISSN0003-0082
2 AMERICAN MUSEUMNOVITATES NO. 3692
strong support for several clades, including (1) the subgenus Micoureus; (2) a group comprised of
Marmosa macrotarsus, M. murina, M.tyleriana,andM.waterhousei;(3)a groupcomprisedofM.
robinsoniandM.xerophila;and(4)agroupcomprisingallofthespeciesinthesubgenusMarmosa
that occur north and west of the Andes (M. isthmica, M. mexicana, M. robinsoni, M. simonsi, M.
xerophila,andM.zeledoni).OurdiscoveryofthelattercladesuggeststhattheAndesmayhaveplayed
amajorroleintheearlydiversificationofthisspecioseradiationofsmallNeotropicalmarsupials.
INTRODUCTION pical subgenus of Marmosa. Based on his
examinationofapproximately2500specimens
Species of the didelphid marsupial genus (including most of therelevant type material),
Marmosa inhabit tropical and subtropical he resurrected five species that had previously
vegetation from Mexico to northern beentreatedasjuniorsynonymsorsubspecies:
Argentina, including such diverse habitats as M. simonsi and M. isthmica (formerly synon-
xerophytic thorn scrub, savannas, lowland ymized with M. robinsoni); M. zeledoni (for-
rain forests, and humid-montane (‘‘cloud’’) merlysynonymizedwithM.mexicana);andM.
forests from sea level to about 3000 meters tobagi and M. waterhousei (formerly synony-
(Creighton and Gardner, 2008). As currently mized with M. murina).5 Although Rossi’s
understood (Voss and Jansa, 2009), the genus unpublished results (summarized, in part, by
contains 15species,of which nine are referred Rossi et al., 2010) are compellingly supported
to the paraphyletic subgenus Marmosa Gray, by morphometric analyses and by qualitative
1821, and six to the monophyletic subgenus characters of the integument, skull, and denti-
Micoureus Lesson, 1842. By virtue of its wide tion,hisproposed taxonomy(table 1)remains
ecogeographic range, the genus is of excep- tobetestedwithmoleculardata.
tional biogeographic interest, but effective Herein we report phylogenetic analyses of
analysisofdistributionalpatternsisprevented DNA sequences from the mitochondrial cyto-
byahostoftaxonomicproblems,nottheleast chrome-b gene representing most of the
of which concerns species delimitation. species recognized by Rossi (2005) in the
Tate (1933) recognized 10 species referable subgenus Marmosa as well as several species
to the subgenus Marmosa (sensu Voss and of the subgenus Micoureus. These data pro-
Jansa, 2009), which he organized into ‘‘sec- vide a basis for testing the monophyly of
tions’’ based on subjectively inferred relation- Rossi’s morphologically defined species, and
ships (table 1). Subsequently, Hershkovitz they afford an opportunity to infer phyloge-
(1951) synonymized all of the taxa in Tate’s netic relationships among the majority of
Mitis Section (for which the oldest available species currently referred to the genus.
name is robinsoni; Cabrera, 1958), and new Although our results include novel insights
specieswerelaterdescribedbyPine(1972)and concerning biogeography andsubgeneric clas-
Handley and Gordon (1979). As a result, sification, we defer formal treatment of these
recent taxonomic synopses (Gardner, 2005; topics to future reports that will incorporate
Creighton and Gardner, 2008; Voss and additional sequence data from other genes.
Jansa, 2009) have recognized nine species:
M. andersoni, M. lepida, M. mexicana, M.
MATERIALS AND METHODS
murina, M. quichua, M. robinsoni, M. rubra,
M. tyleriana, and M. xerophila. Despite such SOURCEOF MATERIAL: Except as noted, all
consensus, several of these species have voucher specimens and associated tissues are
improbably wide geographic distributions preserved in the following collections (listed
(e.g., M. mexicana, M. murina, and M. alphabetically by institutional abbreviation):
robinsoni), and previously published analyses AMNH, American Museum of Natural
of mitochondrial gene sequences suggest that
at least some include genetically divergent 5Rossi (2005) additionally suggested that macrotarsus
forms (Steiner and Catzeflis, 2003, 2004; Wagner,1842,istheoldestavailablenameforthespecies
Patton and Costa, 2003). formerly known as quichua Thomas, 1899. Contra
Creighton and Gardner (2008), macrotarsus Wagner,
In a recent revisionary study, Rossi (2005)
1842, is not preoccupied by macrotarsos Schreber, 1777
recognized 14 valid species in the nominoty- (aprimate).
2010 GUTIE´RREZET AL.:MOLECULAR SYSTEMATICS OFMOUSE OPOSSUMS 3
TABLE1 University of New Mexico (Albuquerque);
Speciesof Marmosa(subgenus Marmosa) MVZ, Museum of Vertebrate Zoology,
Recognized as ValidbyAuthorsa University of California (Berkeley); ROM,
Royal Ontario Museum (Toronto); T-, tissue
Tate(1933)b Gardner(2005)c Rossi(2005) collection oftheLaboratoirede Paleontologie
MurinaSection M.andersonid M.mexicana at the Institut des Sciences de l’Evolution de
M.murina M.lepida M.zeledonig Montpellier (ISEM; Montpellier); TTU,
M.rubra M.mexicana M.isthmicah Museum of Texas Tech University (Lub-
M.tyleriana M.murina M.robinsoni bock); UFMG, Universidade Federal de
M.quichua M.quichua M.simonsih Minas Gerais (Belo Horizonte); UMSNH,
MitisSection M.robinsonie M.xerophila Universidad Michoacana de San Nicolas de
M.mitis M.rubra M.rubra
Hidalgo (Morelia); USNM, United States
M.chapmani M.tyleriana M.andersoni
National Museum of Natural History
M.simonsi M.xerophilaf M.tyleriana
(Washington); V-, voucher collection of
M.ruatanica M.lepida
MexicanaSection M.murina Francois M. Catzeflis (currently at ISEM,
M.mexicana M.macrotarsusi,j these specimens will eventually be deposited
LepidaSection M.waterhouseii either at the Muse´um National d’Histoire
M.lepida M.tobagii Naturelle, Paris, or at MHNG; F.M.
aOnlytaxareferabletothenominotypicalsubgenus(as Catzeflis, in litt.).
recognizedbyVossandJansa,2009)arelisted.Taxaare TAXON SAMPLING: Our taxonomic sample
listedinthesameorderasinthecitedworks. (table 2) includes 71 individuals representing
bNotethatspecieswereorganizedby‘‘sections’’within 12 of the 14 species of Marmosa (Marmosa)
Tate’s(1933)system. recognized by Rossi (2005) together with four
cAlso the taxonomy followed by Creighton and of the six currently recognized species of the
Gardner (2008) and Voss and Jansa (2009). Names are
subgenus Micoureus. We were unable to
usedinthesamesenseasbyTate(1933)exceptasnoted
obtain samples of Marmosa (M.) andersoni,
otherwise.
dDescribedbyPine(1972). M. (M.) tobagi, M. (Mi.) alstoni, or M. (Mi.)
eSenior synonym of mitis. Includes chapmani, simonsi, phaea for this study. Among other didelphid
andruatanica(afterHershkovitz,1951). genera,TlacuatzinandMonodelphishavebeen
fDescribedbyHandleyandGordon(1979). identified as phylogenetically closest to
gFormerlyincludedinM.mexicana. Marmosa (e.g., by Voss and Jansa, 2009, and
hFormerly included in M. robinsoni (sensu Gardner, references cited therein); therefore, we used
2005).
sequences from two individuals of Tlacuatzin
iFormerlyincludedinM.murina.
canescens and one of Monodelphis brevicauda-
jIncludesquichua.
ta as outgroups to root our trees.
Withineach recognizedspeciesofMarmosa
(Marmosa), we chose individuals to represent
as many nominal taxa (subspecies or subjec-
History (New York); BMNH, Natural
History Museum (London); CM, Carnegie tive synonyms) and regions of vertebrate
Museum of Natural History (Pittsburg); endemism (Mu¨ller, 1973; Cracraft, 1985) as
EBRG, Museo de la Estacio´n Biolo´gica de available tissue resources would allow (fig. 1).
Rancho Grande (Maracay); FMNH, Field Forthemajorityofoursamples(60outof71),
MuseumofNaturalHistory(Chicago);INPA, we extracted high-molecular-weight DNA
Instituto Nacional de Pesquizas da Amazoˆnia from field-preserved tissues. We extracted
(Manaus); ISEM, Institut des Sciences de relatively poor-quality DNA from museum
l’Evolution de Montpellier (Montpellier); skins of five individuals (two of M. tyleriana,
LSUMZ, Louisiana State University, Mu- andoneeachofM.rubra,M.zeledoni,andM.
seum of Natural Science (Baton Rouge); xerophila), and we obtained six additional
MHNG, Muse´um d’Histoire Naturelle de sequences from GenBank: three of M. murina
Gene`ve (Geneva); MNK, Museo de Historia (AJ486984, AJ486990, AJ486995), two of M.
Natural Noel Kempff Mercado (Santa Cruz); demerarae (AJ487005, AJ487006), and one of
MSB, Museum of Southwestern Biology, M. mexicana (AJ606454). After removing
4 AMERICAN MUSEUMNOVITATES NO. 3692
TABLE2
SequencedSpecimensof IngroupandOutgroup Taxa
Taxon Tissue/DNA#a Voucherb Localityc bpd
Ingroup
M.(Marmosa)isthmica TK135686 TTU102969 Ecuador:Esmeraldas(17) 1145
M.(Marmosa)isthmica FMG2716 USNM575395e Panama:BocasdelToro(37) 1140
M.(Marmosa)isthmica FMG2736 USNM575397e Panama:BocasdelToro(37) 1146
M.(Marmosa)isthmica TK22555 TTU39118e Panama:Darie´n(39) 1146
M.(Marmosa)lepida F38809 ROM107034e Guyana:Potaro-Siparuni(30) 1146
M.(Marmosa)lepida JLP7844 MVZ155245e Peru:Amazonas(42) 1146
M.(Marmosa)lepida DWF717 AMNH273186e Peru:Loreto(46) 1146
M.(Marmosa)macrotarsus LHE1516 USNM584462e Bolivia:SantaCruz(3) 797
M.(Marmosa)macrotarsus LHE1548 MNK[uncataloged] Bolivia:SantaCruz(3) 1146
M.(Marmosa)macrotarsus JRM202 MVZ191187f Brazil:Amazonas(5) 1146
M.(Marmosa)macrotarsus MNFS746 INPA2912f Brazil:Amazonas(6) 1087
M.(Marmosa)macrotarsus JRM450 INPA2911f Brazil:Amazonas(9) 1146
M.(Marmosa)macrotarsus RSV2303 AMNH272816e Peru:Loreto(46) 1146
M.(Marmosa)macrotarsus RSV2413 AMNH272870e Peru:Loreto(46) 860
M.(Marmosa)mexicanaA MHNG1812007 MHNG1812007 Belize:Corozal(1) 800i
M.(Marmosa)mexicanaA FN32277 ROM99608e Guatemala:ElPete´n(26) 1146
M.(Marmosa)mexicanaA FN34135 ROM99776e Guatemala:ElProgreso(27) 1146
M.(Marmosa)mexicanaA FN30771 ROM96968e Mexico:Campeche(31) 1146
M.(Marmosa)mexicanaA FN30134 ROM96318e Mexico:Campeche(32) 1145
M.(Marmosa)mexicanaA FN29881 ROM96090e Mexico:Campeche(34) 1144
M.(Marmosa)mexicanaA FN29586 ROM95795e Mexico:Campeche(33) 1146
M.(Marmosa)mexicanaB JOM7269 USNM569858e Guatemala:AltaVerapaz(24) 1087
M.(Marmosa)mexicanaB FN31448 ROM98459e Guatemala:BajaVerapaz(25) 1146
M.(Marmosa)mexicanaB WB8515 USNM570071e Guatemala:Zacapa(28) 1146
M.(Marmosa)murina LPC436 MVZ197421f Brazil:MatoGrosso(11) 1146
M.(Marmosa)murina JLP16986 UFMG2599f Brazil:MatoGrossodoSul 1146
(10)
M.(Marmosa)murina LHE503 USNM549291e Brazil:Para´ (12) 1146
M.(Marmosa)murina LHE582 USNM549292e Brazil:Para´ (12) 1146
M.(Marmosa)murina LPC715 MVZ197433f Brazil:Tocantins(14) 1092
M.(Marmosa)murina T2704 MHNG1885048 FrenchGuiana:Cayenne(21) 820i
M.(Marmosa)murina T2084 V-909g FrenchGuiana:Cayenne(22) 820i
M.(Marmosa)murina T2471 V-1206g FrenchGuiana:Cayenne(23) 820i
M.(Marmosa)murina F50629 ROM113649e Guyana:Demerara-Mahaica 1146
(29)
M.(Marmosa)murina F41351 ROM114321h Surinam:Brokopondo(47) 770
M.(Marmosa)murina TK17359 CM68346e Surinam:Para(49) 1146
M.(Marmosa)murina TK17387 CM68353e Surinam:Para(49) 1146
M.(Marmosa)robinsoni NK101529 MSB94363e Panama:LosSantos(40) 1146
M.(Marmosa)robinsoni NK101606 MSB94366e Panama:LosSantos(40) 1146
M.(Marmosa)robinsoni NK101633 MSB94368e Panama:Veraguas(41) 1146
M.(Marmosa)robinsoni NK101634 MSB94369e Panama:Veraguas(41) 1146
M.(Marmosa)robinsoni RPA262 EBRG25389e Venezuela:Falco´n(52) 1146
M.(Marmosa)rubra F54196 ROM118744e Ecuador:Orellana(20) 1146
M.(Marmosa)rubra — FMNH84253e Peru:Cusco(43) 402
M.(Marmosa)simonsi NK37836 MSB87086e Ecuador:ElOro(16) 1146
M.(Marmosa)simonsi NK37837 MSB87087e Ecuador:ElOro(16) 1146
M.(Marmosa)simonsi TK134911 TTU103308e Ecuador:Guayas(18) 1146
M.(Marmosa)tyleriana — AMNH130510e Venezuela:Bol´ıvar(50) 398
M.(Marmosa)tyleriana — AMNH130511e Venezuela:Bol´ıvar(50) 399
M.(Marmosa)waterhousei F40140 ROM105889e Ecuador:Orellana(19) 1146
M.(Marmosa)waterhousei F37580 ROM105257e Ecuador:Orellana(20) 727
2010 GUTIE´RREZET AL.:MOLECULAR SYSTEMATICS OFMOUSE OPOSSUMS 5
TABLE 2
(Continued)
Taxon Tissue/DNA#a Voucherb Localityc bpd
M.(Marmosa)waterhousei JLP7480 MVZ154754e Peru:Amazonas(42) 726
M.(Marmosa)waterhousei TK73294 TTU98717e Peru:Loreto(44) 1146
M.(Marmosa)waterhousei TK73276 TTU100922e Peru:Loreto(44) 1050
M.(Marmosa)waterhousei JMC88 LSU28017e Peru:Loreto(45) 1146
M.(Marmosa)xerophila — USNM443814e Colombia:LaGuajira(15) 402
M.(Marmosa)xerophila RPA315 AMNH276582e Venezuela:Falco´n(51) 1146
M.(Marmosa)xerophila RPA324 AMNH276586 Venezuela:Falco´n(51) 1146
M.(Marmosa)zeledoni — AMNH269997e Panama:Chiriqu´ı(38) 402
M.(Micoureus)constantiae NK15501 MSB59883e Bolivia:SantaCruz(2) 1146
M.(Micoureus)constantiae NK23272 AMNH275466e Bolivia:SantaCruz(4) 1146
M.(Micoureus)demerarae T2006 V-972 FrenchGuiana:Cayenne(22) 820i
M.(Micoureus)demerarae T2083 V-884e FrenchGuiana:Cayenne(22) 820i
M.(Micoureus)demerarae RSV2029 AMNH272667e Peru:Loreto(46) 1146
M.(Micoureus)demerarae RSV2085 MUSM13294e Peru:Loreto(46) 1146
M.(Micoureus)paraguayana MAM46 MVZ182064e Brazil:Sa˜oPaulo(13) 1146
M.(Micoureus)paraguayana MAM47 MVZ182065e Brazil:Sa˜oPaulo(13) 1146
M.(Micoureus)regina JLP15435 MVZ190323e Brazil:Amazonas(7) 1146
M.(Micoureus)regina MNFS1232 MVZ190332e Brazil:Amazonas(8) 402
Outgroups
Monodelphisbrevicaudata TK17069 CM68359e Surinam:Nickerie(48) 1146
Tlacuatzincanescens TK11826 TTU37700e Mexico:Jalisco(35) 1146
Tlacuatzincanescens TK45085 UMSNH2993e Mexico:Michoaca´n(36) 1146
aAlphanumericidentifiersusedbyinstitutionaltissuecollections(andtolabelterminalsinaccompanyingtrees;figs.2,
3).Sequencesamplifiedfrommorphologicalspecimenslacktissue/DNAnumbers.
bSeeMaterialsandMethodsfornamesofmuseumcollectionsidentifiedbyabbreviationsinthistable.
cCountry and next-largest administrative unit (state, department, province, etc). Numbers in parentheses refer to
gazetteerentries(appendix),whichprovideadditionalgeographicinformation.
dNumberofbasepairssequenced.Allsequenceswereobtainedbyusexceptasindicatedotherwise.
eExaminedbytheauthors.
fExaminedbyRossi(2005).
gExaminedbySteinerandCatzeflis(2003).
hExaminedbyLimetal.(2005).
iFromGenBank.
identical haplotypes, our phylogenetic analy- published sequences were obtained and found
ses were based on a matrix that included several discrepancies. For example, our se-
cytochrome-b sequences from 66 individuals. quence of AMNH 272816 should be identical
Although many sequences identified as to GenBank accession number AJ487003,
Marmosa murina are available in GenBank, butthesedifferinanA/Cmutationatposition
mostofthemarefromlocalitiesontheGuiana 783 (as numbered from the start codon).
Shield, where there is very little genetic Additionally, our sequence of AMNH
variation and apparently no phylogeographic 272870wasgeneratedfromthesamespecimen
structure (Steiner and Catzeflis, 2003, 2004); as GenBank accession AJ487002, but the two
therefore, we included only three (all from differatfoursites(59A/G,246C/T,813G/A,
French Guiana) in our study. Two other 819 T/C). Our sequences, which were generat-
GenBank sequences reported as M. murina edfromatleasttwostrands,areunambiguous
in previous studies are from Peru and corre- at these sites. Any number of reasons,
spond to M. macrotarsus (sensu Rossi, 2005). including error incorporated by Taq polymer-
We resequenced the tissues from which these ase, could explain such differences. However,
6 AMERICAN MUSEUMNOVITATES NO. 3692
Fig.1. ProvenanceofsequencedspecimensofMarmosa(localitiesofsequencedoutgroupspecimensare
notshown). Numbersrefer to entriesin the Gazetteer (appendix).
because we do not have the original chro- Initial PCR amplifications using genomic
matograms from the GenBank reports, we DNA as a template were performed in 20 mL
used the sequences generated in our lab from reactions using GoTaq DNA polymerase
these specimens. (Promega Corp.) and recommended concen-
LABORATORY METHODS: Genomic DNA trations of primers, unincorporated nucleo-
was extracted from all samples using DNeasy tides,buffer,andMgCl .Thesereactionswere
2
extraction kits (Qiagen, Inc.). Whenever pos- performed in a four-stage touchdown proto-
sible, we amplified the entire cytochrome-b col. The first stage consisted of 5 cycles of
gene using primers CYTB-F1 and CYTB-R1 denaturation at 95uC for 20 seconds, anneal-
(table 3) located in the flanking tRNAs. To ing at 59uC for 20 seconds, and extension at
generate fragments of a suitable size for 72uC for 30 seconds. The second and third
sequencing, we used this PCR product in stages were identical to the first except for
two separate reamplification reactions, one lowered annealing temperatures of 57uC and
using primer CYTB-F1 paired with CYTB- 55uC,respectively.Thefinalstageconsistedof
730R and one using either CYTB-540F or 25 cycles with an annealing temperature of
CYTB-650F paired with CYTB-R1. In cases 52uC. Subsequent reamplification reactions
where we extracted poor-quality DNA from using this product as a template consisted of
skin samples (two samples of Marmosa a single stage of 25 cycles of denaturation at
tyleriana and one each of M. xerophila, M. 95uC for 30 seconds, annealing at 55uC for
zeledoni, and M. rubra), we generated a short 30 seconds, and extension at 72uC for 1 min-
(,400 bp) PCR product using CYTB-F1 ute. All reactions were preceded by an initial
paired with CYTB-420R and sequenced it denaturation at 95uC for 2 minutes and
directly using amplification primers. followed by a 7 minute extension at 72uC.
2010 GUTIE´RREZET AL.:MOLECULAR SYSTEMATICS OFMOUSE OPOSSUMS 7
TABLE 3
NameandDNASequenceof Primers Used forDNAAmplification andSequencing
Primername Primersequence
CYTB-F1-Didelphidae 59ATAACCTATGGCATGAAAAACCATTGTTG
CYTB-R1-Didelphidae 59CCTTCATTGCTGGCTTACAAGGC
CYTB-420R-Didelphidae 59GCTCCTCAGAAGGATATTTGTCCTCA
CYTB-730R-Marmosa 59TCWCCTAATARRTCWGGTGARAATATTGC
CYTB-540F-Marmosa 59GAGGAGGMTTYTCHGTTGATAAAGC
CYTB-650F-Marmosa 59CTATTCCTTCACGAAACAGGCTC
CYTB-217R-Marmosa 59TCTGTAGCCCAYATYTGYCGWGAYG
CYTB-70F-Marmosa 59CCMTCAAATATTTCAGCCTGATG
Gene fragments were sequenced in both criterion (AIC) as implemented in ModelTest
directions using amplification primers and v. 3.7 (Posada and Crandall, 1998) and
ABI BigDye version 3.1 terminator chemistry PAUP* (Swofford, 2002). For ML analyses,
(Applied Biosystems, Inc.). Reactions were we performed 20 independent searches in
run on either an ABI 3130xl or ABI 3730xl GARLI 0.96 beta (Zwickl, 2006) using the
capillary sequencer. Sequences were edited default settings. Maximum-likelihood boot-
and compiled using Sequencher version 4.8 strapanalyseswereperformedinGARLI0.96
(GeneCodesCorporation,2007).Allsequenc- betausing100pseudoreplicateddatamatrices,
es, along with their specimen voucher num- with10 searchesperformed on each. Bayesian
bers, have been deposited in GenBank with analyses were performed using the Markov
accession numbers HM106338–HM106402. Chain Monte Carlo (MCMC) sampling ap-
ANALYTICAL METHODS: We performed proach in MrBayes v. 3.1.2 (Huelsenbeck and
multiple sequence alignment in Clustal X Ronquist,2001;Altekaretal.,2004;Ronquist
version 2.0 (Larkin et al., 2007) and adjusted et al., 2005) through the Computational
the resulting alignment with reference to Biology Service Unit from Cornell University
translated amino-acid sequences. We used (http://cbsuapps.tc.cornell.edu/mrbayes.aspx).
maximum parsimony (MP), maximum likeli- The search started with a random tree, and
hood (ML), and Bayesian inference (BI) to consisted of one cold chain and three heated
analyze the resulting data matrix; missing chains(temperature50.2)anddefaultpriors.
bases were coded as unknown for all phylo- The Markov chains were run for 1 3 106
genetic analyses. To assess nodal support, generations, and trees were sampled every
we used nonparametric bootstrapping 1000generations.Defaultvalueswerekeptfor
(Felsenstein, 1985) for the MP and ML the ‘‘relburnin’’ and ‘‘burninfrac’’ options in
analyses and nodal posterior probability MrBayes (i.e., relburnin 5 yes; burninfrac 5
estimates for the BI analysis (Ronquist and 0.25); therefore, the first 250,000 generations
Huelsenbeck, 2003). Parsimony analyses were (250 trees) were discarded as burn-in, and
performedinPAUP*4.0b10(Swofford,2002) posterior probability estimates of all model
using equal weighting and the heuristic search parameterswerebasedontheremaining(750)
option with 1000 replicate searches, 10 ran- trees.
dom-addition replicates, and tree bisection- To estimategeneticdivergence, we calculat-
reconnection (TBR) branch swapping. ed average uncorrected (p) distance within
Maximum-parsimonybootstrapanalyseswere each species and average pairwise p distances
performed in PAUP* using 1000 pseudodo- among species. In addition, we report K2P-
replicated data matrices, each with 5 random- corrected distances for interspecific compari-
addition sequences and TBR branch-swap- sons. These model-corrected statistics are the
ping. To determine the appropriate model of traditionalmetricforgeneticdivergenceinthe
evolution for ML and BI analyses, we didelphid literature (e.g., Patton et al., 2000;
considered both hierarchical likelihood-ratio Patton and Costa, 2003), so we computed
tests (hLRT) and the Akaike information them to allow comparisons with values re-
8 AMERICAN MUSEUMNOVITATES NO. 3692
ported in previous studies. All distances were M. tobagi), and we had only a single
calculated using MEGA version 4 (Tamura et representative sample of M. zeledoni. For 10
al., 2007). of these 11 cases, morphologically defined
species were recovered as monophyletic
groups, usually with moderate to very strong
RESULTS
support in both the MP and the model-based
There are five pairs of identical haplotypes analyses (figs. 2, 3). The only noteworthy
among our 71 sequences: two specimens of exception concerns M. mexicana, samples of
Marmosa simonsi (NK37836 and NK37837) whichformtwodeeplydivergenthaplogroups
from Ecuador, two specimens of M. robinsoni (hereafterreferredtoas‘‘M.mexicanaA’’and
(NK101606 and NK101633) from Panama, ‘‘M. mexicana B’’) that were not consistently
two specimens of M. murina (TK17359 and recovered as sister taxa. Although the model-
TK17387)fromSurinam,twospecimensofM. based analyses recovered these two hap-
murina (T2471 and T2704) from French logroups as a clade, the MP analysis placed
Guiana, and two specimens of M. paraguaya- M.zeledoniasthesistertaxontoM.mexicana
na (MAM46 and MAM47) from Brazil. In AandM.isthmicaassistertoM.mexicanaB;
each of these cases, we excluded the sequence as might be expected, both of these alterna-
corresponding to the second-listed specimen tives are weakly supported.
from all subsequent phylogenetic analyses, Mean uncorrected sequence divergence
resulting in a final data matrix comprising 42 within species (provisionally including M.
complete cytochrome-b sequences (each with mexicana A and M. mexicana B, see below;
1146 bp) and 24 partial sequences (ranging in table 5) ranges from 0.2 to 4.2%. However,
length from 398 to 1145 bp; table 2). As sequence divergence across the basal split
expected of mitochondrial sequences, average within some species is considerably higher
base composition across this dataset is rela- than these average within-group values. In
tively poor in guanine (30.7% A, 22.9% C, particular, Panamanian sequences of M.
12.5% G, 33.9% T), but there is no significant robinsoni differ from the single available
departure from base-compositional stationar- Venezuelan sequence by 6.2%, Bolivian se-
ity among taxa (x2 5 121.83, df 5 186, p 5 quences of M. macrotarsus differ from
0.99; see Saccone et al., 1989). All sequences Brazilian and Peruvian sequences by 6.5%,
translate to open reading frame. and Peruvian sequences of M. demerarae
Our dataset contains 508 variable charac- differ from French Guianan sequences by
ters, 465 of which are parsimony informative. 5.7%. By contrast, average interspecific diver-
Maximum-parsimony analysis recovered 96 gence values within three consistently recov-
minimum-length trees, the strict consensus of ered sister-species pairs (M. constantiae + M.
which is shown in figure 2. For the model- regina, M. demerarae + M. paraguayana, and
based analyses (ML and BI), the hierarchical M.robinsoni +M.xerophila)rangefrom9.5%
likelihood-ratio test (hLRT) selected the most to 18.6%.
complex model (GTR+I+C), whereas the PHYLOGENETICRELATIONSHIPS: Whereascy-
simpler HKY+I+C model was preferred using tochrome-b sequences are clearly useful for
the AIC. To test for possible effects of model testing the monophyly of morphologically
selection on our phylogenetic analyses, we defined species and for assessing intraspecific
performed ML analyses specifying each of genetic divergence, they are less consistently
these models and obtained identical topolo- informative about phylogenetic relationships
gies; therefore, we report only the results among species. Approximately half of the
obtained under the more complex GTR+I+C interspecific nodes resolved in our trees were
model (table 4; fig. 3). recoveredwithstrongsupportinbothMPand
SPECIES LIMITS: We were able to test the model-based analyses. Among these well-sup-
monophyly of just 11 of the 14 morphologi- ported nodes are the sister-species pairs
callydefinedspeciesinthesubgenusMarmosa Marmosa robinsoni + M. xerophila, M. con-
recognized by Rossi (2005) because we lacked stantiae + M. regina, and M. demerarae + M.
samples of two taxa (Marmosa andersoni and paraguayana. At deeper levels, all of our
2010 GUTIE´RREZET AL.:MOLECULAR SYSTEMATICS OFMOUSE OPOSSUMS 9
Fig. 2. Strict consensus of 96 equally most-parsimonious trees (L 5 2198; CI 5 0.36; RI 5 0.80).
Bootstrap support values are indicated above branches subtending species and conspecific haplogroups
discussed in the text. For each terminal, country of origin, next-largest political unit (state, department,
province, etc.), and an alphanumeric specimen identifier (from table 2) are provided. Numbers in
parenthesesrefer to localities mappedin figure 1and listedin the Gazetteer (appendix).
10 AMERICAN MUSEUMNOVITATES NO. 3692
phylogeneticanalysessupportedthemonophy- TABLE4
lyofthegenusMarmosasensuVossandJansa Parameter Estimates fromthe Best-FitModel of
(2009). Also, all analyses recovered a well- Nucleotide SubstitutionforCytochrome b
supported group comprising M. robinsoni, M.
xerophila, M. isthmica, M. mexicana, M. -lnL 10807.817
Basefrequencies
zeledoni, and M. simonsi (hereafter the ‘‘mex-
icana-robinsoni clade’’); within this group, M. pA 0.349
pC 0.251
simonsiwasconsistentlyrecoveredasthesister
pG 0.059
taxon to the remaining species with moderate
pT 0.341
to strong support. Marmosa murina and three
RatematrixR
other species (M. tyleriana, M. waterhousei,
and M. macrotarsus) formed another consis- rA-C 1.320
rA-G 13.495
tently well-supported clade, and the subgenus
rA-T 1.026
Micoureus (represented by M. constantiae, M.
rC-G 1.500
regina, M. demerarae, and M. paraguayana)
rC-T 13.189
was also recovered as monophyletic in all of rG-T 1.000
ouranalyses. Proportionofinvariantsites 0.515
By contrast, our MP and model-based ShapeparameterfortheCdistribution(a) 1.139
analyses were notably inconsistent in their
placement of Marmosa lepida and M. rubra.
Whereas model-based analyses recovered M. Patternsofinterspecificrelationshipswithin
lepida as sister to the murina cluster + the robustly supported murina cluster are
Micoureus (with weak ML bootstrap but similarly equivocal. Whereas the ML analysis
strong Bayesian support), the parsimony recoveredthesister-speciespairM.tyleriana+
analysis recovered M. lepida as the sister M. waterhousei, with M. murina and M.
taxon to Micoureus (with negligible bootstrap macrotarsus as sequentially more distantly
support). In the model-based analyses, M. related sister taxa (fig. 3), the MP analysis
rubra was recovered as the sister taxon to the placedM.murinaandM.macrotarsusassister
mexicana-robinsoni clade (again with weak taxaandleftthepositionsofM.tylerianaand
ML bootstrap but impressive Bayesian sup- M. waterhousei unresolved (fig. 2); neither of
port), whereas M. rubra was recovered as the thesealternativesreceivedcompellingsupport.
sister taxon to all other analyzed congeners in A clade comprising the four species of
the parsimony tree (with ,50% bootstrap Micoureus was recovered as the sister taxon
support). to the murina clade in both of our model-
The remaining interspecific nodes either based analyses, but always with low support.
agree or differ between the MP and model-
based analyses, but all have uniformly weak DISCUSSION
supportvalues.Withinthemexicana-robinsoni
clade,forexample,theMLanalysisrecovered Despite ongoing debate about species con-
the two haplogroups of M. mexicana (A and cepts in the systematic literature (reviewed by
B) as a clade, with M. zeledoni and M. de Queiroz, 1998; Mayden, 1999; Coyne and
isthmica as sequentially less closely related Orr, 2004; Baker and Bradley, 2006), most
sister taxa (fig. 3), whereas the MP analysis researchers agree that genetically independent
placed M. zeledoni as sister to M. mexicana A lineages are fundamentally important units
and M. isthmica as sister to M. mexicana B of evolutionary diversification. We therefore
(fig. 2); neither alternative received strong adoptalineage-basedconceptofspecies(after
Bayesianorbootstrapsupport.Althoughboth de Queiroz, 1998), for which we use mtDNA
ML and MP analyses recovered the more haplotype monophyly (as recovered by this
inclusive clade comprised of M. mexicana, M. study) and morphological diagnosability (as
isthmica, and M. zeledoni, Bayesian and documentedbyRossi,2005;Rossietal.,2010)
bootstrap support for this relationship is as operational criteria for species recognition.
negligible. Whereas mtDNA sequences provide crucial