Marmosops, MATSCHIE, 1916

Díaz-Nieto, Juan F., Jansa, Sharon A. & Voss, Robert S., 2016, DNA sequencing reveals unexpected Recent diversity and an ancient dichotomy in the American marsupial genus Marmosops (Didelphidae: Thylamyini), Zoological Journal of the Linnean Society 176 (4), pp. 914-940 : 931-932

SUBGENUS MARMOSOPS MATSCHIE, 1916

Type species

Marmosops incanus (Lund, 1840) .

Included taxa

We refer 25 nominal taxa to the subgenus Marmosops : albiventris Tate, 1931; bahiensis Tate, 1931; caucae Thomas, 1900 ; celicae Anthony, 1922; collega Thomas, 1920; creightoni, Voss et al., 2004 ; dorothea Thomas, 1911; keaysi Allen, 1900; leucastrus Thomas, 1927; lugendus Thomas, 1927; madescens Osgood, 1913; neblina Gardner, 1990 ; neglectus Osgood, 1915; noctivagus Tschudi, 1845 ; ocellatus Tate, 1931 ; oroensis Anthony, 1922; paulensis Tate, 1931 ; perfuscus Thomas, 1924; politus Cabrera, 1913; purui Miller, 1913; scapulatus Burmeister, 1856; sobrinus Thomas, 1913; stollei Miranda-Ribeiro, 1936; ucayaliensis Tate, 1931; and yungasensis Tate, 1931.

The nominal species impavidus Tschudi, 1845 , is not definitely identifiable as a member of the genus Marmosops despite current usage of this name for at least two distinct species in the nominotypical subgenus. Briefly (a fuller account will be provided elsewhere), no type specimen is known to exist, the name cannot be applied with confidence based on characters described by Tschudi, its application as a senior synonym for caucae is inconsistent with Tschudi’s description, and its application to specimens with ocellatus - like CYTB sequences from western Brazil and northeastern Peru is inconsistent with Tschudi’s type locality. In our opinion, impavidus is best regarded as a nomen dubium and should be retired from taxonomic service.

Remarks

The sequences that we identify as incanus are from south-eastern Brazilian material that exhibits the diagnostic morphological traits of M. incanus as described and illustrated by Mustrangi & Patton (1997). As in their study, we found substantial geographically structured CYTB sequence divergence amongst incanus -like specimens, which GMYC analysis suggests represent four putative species (differing by 4.2– 10.8%, on average, in uncorrected pairwise comparisons). Although we use alphabetical designations for these haplogroups, the type locality of incanus (Lagoa Santa, in Minas Gerais) falls within the geographical range of ‘ incanus C’ (north of the Rio Paraíba do Sul), to which this epithet would presumably belong if it were to be applied in a stricter sense than it is at present. Additionally, at least on the basis of geography, the name scapulatus (also based on a specimen from Minas Gerais) seem to apply to incanus C, and the name bahiensis (based on a type from Bahia) is geographically closest to incanus D. Apparently , there are no available names for ‘ incanus A’ (from the southern coastal plain of Rio de Janeiro) or ‘ incanus B’ (from São Paulo). However, pending morphological analysis of this complex, and in the absence of any other evidence for nuclear gene divergence, it seems premature to suggest that M. incanus includes more than a single valid species.

Similarly, we apply the name paulensis to specimens from south-eastern Brazil that share the diagnostic morphological traits of M. paulensis as described by Mustrangi & Patton (1997). The three putative species identified by LGMYC in this complex occur allopatrically, each on a different mountain range (‘ paulensis A’ in the Serra da Mantiqueira, ‘ paulensis B’ in the Serra dos Orgaõs, ‘ paulensis C’ in the Serra do Mar) and differ by 3.0–6.5%, on average, in uncorrected pairwise CYTB sequence comparisons. The holotype (FMNH 26576) is from the Serra dos Orgaõs, so the epithet would properly apply to ‘ paulensis B’ if it were to be applied in a stricter sense. No names are currently available for ‘ paulensis A’ or ‘ paulensis C’, but in the absence of diagnostic morphological characters (which we have yet to discover) or other evidence for nuclear gene divergence, formal taxonomic recognition is inappropriate.

Clade D includes two putative species represented by three sequences from morphologically distinctive specimens recently collected in the Cordillera del Condor of north-eastern Peru and from the immediately adjacent Cordillera Oriental of southern Ecuador. No name is available for either putative species, which we distinguish alphabetically. Sequenced specimens of ‘Condor A’ and ‘Condor B’ were collected over 1700 m a.s.l. on opposite sides of the Río Zamora, and their CYTB sequences differ by about 7.5% (uncorrected) – an impressive difference for material obtained only 98 km apart on a straight line – but we lack sequence data from the headwaters of the Zamora, where intermediates might be expected to occur.

The caucae complex (also part of clade D) includes one of the most geographically widespread series of specimens recovered by our analyses. Our sequenced material spans an enormous range, from about 11°N (in northern Colombia) to 7°S (in northern Peru), and from about 66°W (in southern Venezuela) to 80°W (in western Ecuador); most sequenced specimens were collected in montane habitats, but several are from lowland Amazonian sites. As might be expected, there is substantial sequence variation within this complex (4.1%, uncorrected), but only two haplogroups (‘ caucae A’ and ‘ caucae B’) were recovered as weakly divergent (2%, uncorrected; Table S4) putative species. The name caucae is based on a Colombian specimen that morphologi- cally resembles all of the referred material that we examined from that country; other names that can be associated with this complex based on morphology and geography include celicae, madescens, neblina , oroensis, perfuscus, and sobrinus. Sequence data that we obtained from type material (holotypes and/or paratypes) of celicae (AMNH 47182), neblina (USNM 560732, 560735), and oroensis (AMNH 47180) are likewise consistent with the hypothesis that these names are junior synonyms of caucae . The type locality of madescens is geographically closest on a straight line distance to our material of ‘ caucae B’ (e.g. locality 115, Fig. S3 ) but the type was collected at a much higher elevation. [The type locality of madescens is in the highlands at approximately 2700 m a.s.l., whereas geographically adjacent material of ‘ caucae B’ was collected in the lowlands (below 815 m).] Consequently, we are unable to confidently associate a name with ‘ caucae B’ pending a proper revision of this complex.

Sister to the caucae complex is a putative species consisting of five sequences from three localities in southeastern Peru; one of these sequences is from a paratype of ucayaliensis (AMNH 76532), which appears to be the only available name that is definitely assignable to this haplogroup, although purui (from western Brazil; Fig. S3 ) might be a senior synonym. The average pairwise uncorrected difference between CYTB sequences of caucae (including haplogroups A and B) and ucayaliensis is just 3.9%, and we are not aware of any consistent morphological differences amongst relevant voucher specimens. In effect, there is, as yet, no compelling evidence that these are distinct species.

Clade E consists of two putative species that lack available names (‘Gálvez’ and ‘Juruá’) and a third that corresponds to M. ocellatus as diagnosed by Voss et al. (2004); all are lowland haplogroups that occur in eastern Bolivia, eastern Peru, and/or western Brazil. Voss et al. (2004) described morphological differences between Bolivian material of ocellatus and ‘Juruá’ (the latter represented by a sequenced specimen that they called Marmosops impavidus ), which might be valid species despite modest sequence divergence (4.2%, uncorrected). ‘Juruá’ and ‘Gálvez’ differ by an average pairwise CYTB distance of about 7.4% and also appear to be phenotypically distinct.

Clade F includes creightoni , a morphologically distinctive species ( Voss et al., 2004), of which we sequenced the holotype (CBF 6552) and another specimen, both from the eastern Andean versant of northern Bolivia. Sister to creightoni , and differing from it by average uncorrected distances of 7.7–9.0%, is a strongly supported group of four putative species that collectively correspond to Marmosops noctivagus of current usage (sensu Gardner & Creighton, 2008). We use alphabetical designations to distinguish these haplogroups, which differ inter se by only 2.8–4.9%, because the application of available names is uncertain.

Tschudi’s (1844) original description of noctivagus was based on specimens collected in east-central Peru (Junín department), which, unfortunately, is in between the known geographical ranges of ‘ noctivagus B’ and ‘ noctivagus C’. As we do not have sequence data from any surviving syntypes, and because morphological differences between voucher specimens of these haplogroups are not apparent, Tschudi’s epithet could apply to either of them. The application of keaysi, the next-oldest name based on noctivagus -like material, is also uncertain because Allen’s (1900) type locality in southern Peru (Puno department) is in between the known geographical ranges of ‘ noctivagus C’ and ‘ noctivagus D’. Geography can be used to associate albiventris (based on a type from Cusco department, Peru) and two names based on material from Amazonian Brazil (collega and stollei) with ‘ noctivagus C’, and the same logic suggests that politus (from eastern Ecuador) applies to ‘ noctivagus A’. Three names currently treated as synonyms of noctivagus that are based on types from northern Peru (leucastrus, lugendus, and neglectus) might apply to ‘ noctivagus B’, but until this complex is revised taxonomically, these are mere conjectures. Our sequences of ‘ noctivagus D’ include one from a paratype of yungasensis, but the name dorothea (based on a type from La Paz department, Bolivia) is a senior synonym ( Voss et al., 2004).

Figures

Figure 3. Collection localities for sequenced specimens of Atlantic Forest species included in subgenus II of Marmosops. Progressively darker shading indicates the following elevations: pale grey ≥ 500 m, medium grey ≥ 1000 m, and dark grey ≥ 2000 m.