2004 
most lowland specimens have more-or-less 
unicolored (all dark) tails, tails of highland 
specimens are usually indistinctly bicolored 
(dark above, pale below) and _ particolored 
(paler distally than basally). Despite such el- 
evation-correlated phenotypic variation, cy- 
tochrome-b sequences analyzed by Mustran- 
gi and Patton (1997) and by Patton et al. 
(2000) indicate only minor divergence (about 
3%) among lowland and highland haplotypes 
of noctivagus-like Marmosops, including 
specimens from the Amazonian lowlands of 
western Brazil (e.g., INPA 2931), the An- 
dean slopes of southeastern Peru (e.g., MVZ 
171408), and northern Bolivia (e.g., AMNH 
268936). The nominal taxon that Tate (1931) 
described as albiventris, currently synony- 
mized with M. impavidus, is morphologically 
indistinguishable from M. noctivagus and has 
similar mtDNA sequences (Mustrangi and 
Patton, 1997).! 
Traditional taxonomic distinctions among 
forms that we consider conspecific with Mar- 
mosops noctivagus are almost entirely based 
on coat color. Tate (1933: 153-161), for ex- 
ample, described the dorsal fur as “‘varying 
from natal brown ... to Mars brown’’ in 
typical noctivagus, by contrast with “‘bone 
brown ... or warm sepia’ in keaysi and 
**pecan brown” in dorothea. These chromat- 
ic differences are real, but similarities among 
these nominal taxa in other morphological 
characters, the existence of intermediate pel- 
age hues in larger series of noctivagus-like 
eastern-slope specimens obtained in recent 
years, and the modest range of mtDNA var- 
iation among geographically widespread 
samples suggest that such color contrasts are 
not taxonomically significant. 
Paradoxically, several holotypes in this 
complex exhibit atypical craniodental mea- 
surements, a factor that may have contributed 
to the traditional view that these specimens 
represent distinct taxa. For example, the ho- 
' We examined the holotype of Marmosa caucae al- 
biventris (USNM 194378) and the specimens that Mus- 
trangi and Patton (1997) identified as Marmosops albi- 
ventris (MVZ 171408, 173930—173935). All of this ma- 
terial fits our working diagnosis of M. noctivagus (table 
4) and exhibits no close similarity with M. impavidus. 
In particular, the ventral fur is mostly self-white, the su- 
praorbital margins are distinctly beaded in mature adults, 
and all of the adult male skins exhibit well-developed 
gular glands. 
VOSS ET AL.: BOLIVIAN MARMOSOPS 29 
lotype of keaysi is an exceptionally large 
male with a very broad interorbit that is not 
matched by any paratype or topotype that we 
examined. By contrast, the holotype of do- 
rothea has an unusually narrow interorbit, 
and the holotype of yungasensis has a con- 
spicuously short molar toothrow. Unfortu- 
nately, there are no large series from any lo- 
cality to serve as a relevant standard by 
which such morphometric variation might be 
assessed. Just among the 11 specimens of 
‘“‘yungasensis’’ from Pitiguaya, however, in- 
terorbital breadth varies from 5.6 to 6.4 mm, 
and length of the molar series varies from 6.3 
to 7.1 mm. Therefore, none of the exception- 
al values for these dimensions in table 10 
would appear inconsistent with the hypoth- 
esis that the taxa represent one geographi- 
cally variable species. 
Most of the specimens that we refer to 
Marmosops noctivagus were identified by 
Anderson (1997) as M. dorothea or M. noc- 
tivagus keaysi, but some were identified as 
M. impavidus as explained in the preceding 
account, and three others (UMMZ 155829, 
156006, 156007) were identified as Gracili- 
nanus agilis unduaviensis. 
BOLIVIAN SPECIMENS EXAMINED: Beni, 1 km E 
La Embocada (UMMZ 156008, 156013), Rurren- 
abaque (AMNH 247651); Cochabamba, Incacha- 
ca (AMNH 38718), Cordillera de Mosetenes 
(CBF 7527, 7560, 7573, 7577), Tablas Monte 
(AMNH_ 275451-—275458; MSB 70276-70280, 
87080); La Paz, 20 km NNE Caranavi (UMMZ 
126680), Chijchijpa (AMNH 268936, 268937, 
275459; MSB 57004, 68333, 68334, 87092, 
87110), El Vertigo (CBF 4003), 13 km SW Ixia- 
mas (USNM 579250), Kahua (CBF 3880), La Re- 
serva (MSB 68335), Pitiguaya (AMNH 72550, 
72558-72562, 72564, 72566; UMMZ 155829, 
156006, 156007), Rio Solacama (BMNH 
1.2.1.36, 1.6.7.79), Serrania Bella Vista (AMNH 
275460), Ticunhuaya (AMNH 72567); Pando, 
Palmira (AMNH 262402, 262403), Santa Rosa 
(AMNH 262404). 
Marmosops ocellatus 
Judging from the large numbers of speci- 
mens in museum collections, Marmosops 
ocellatus is the commonest (or most fre- 
quently trapped) small marsupial throughout 
most of Santa Cruz department, from the 
base of the Andes to the Brazilian frontier 
(fig. 12). Currently considered to be a junior 
