Review of the Hylomyscus denniae group (Rodentia: Muridae) in eastern
Africa, with comments on the generic allocation of Epimys
endorobae Heller
Michael D. Carleton, Julian C. Kerbis Peterhans, and William T. Stanley
(MDC) Department of Vertebrate Zoology, National Museum of Natural History, Smithsonian
Institution, Washington, D.C. 20560-0108, U.S.A., e-mail: carleton@si.edu;
(JKP) University College, Roosevelt University, Chicago, Illinois 60605, U.S.A.;
Department of Zoology, Division of Mammals, The Field Museum of Natural History,
Chicago, Illinois 60605, U.S.A., e-mail: kerbis@fieldmuseum.org;
(WTS) Department of Zoology, Division of Mammals, The Field Museum of Natural History,
Chicago, Illinois 60605, U.S.A., e-mail: stanley@fieldmuseum.org
Abstract.?The status and distribution of eastern African populations
currently assigned to Hylomyscus denniae are reviewed based on morpho-
logical and morphometric comparisons. Three species are considered valid,
each confined largely to wet montane forest above 2000 meters: H. denniae
(Thomas, 1906) proper from the Ruwenzori Mountains in the northern
Albertine Rift (west-central Uganda and contiguous D. R. Congo); H.
vulcanorum Lo?nnberg & Gyldenstolpe, 1925 from mountains in the central
Albertine Rift (southwestern Uganda, easternmost D. R. Congo, Rwanda,
and Burundi); and H. endorobae (Heller, 1910) from mountains bounding
the Gregory Rift Valley (west-central Kenya). Although endorobae has been
interpreted as a small form of Praomys, additional data are presented that
reinforce its membership within Hylomyscus and that clarify the status of
Hylomyscus and Praomys as distinct genus-group taxa. The 12 species of
Hylomyscus now currently recognized are provisionally arranged in six
species groups (H. aeta, H. alleni, H. anselli, H. baeri, H. denniae, H. parvus)
based on 8 qualitatitive characters. Biogeography of the three species of the
H. denniae group is discussed in the context of broad distributional patterns
and area relationships evident among other terrestrial small mammals also
confined to the Afromontane biotic region in eastern Africa.
Until recently, Hylomyscus denniae, the
Montane Woodland Mouse, was under-
stood to be a highly variable murid
species, broadly if patchily distributed in
forested highlands across south-central
and eastern Africa (e.g., Misonne 1974,
Bishop 1979, Musser & Carleton 1993).
Carleton & Stanley (2005), however,
demonstrated that nominal denniae masks
a complex of species and perhaps two
species groups. Members of the H. anselli
group occur in mountains of south-
central Africa: from the central Angolan
highlands (cf. anselli group); through the
disconnected mountain chains in north-
ern Zambia and western Tanzania (an-
selli); to Mount Rungwe and the Eastern
Arc Mountains of central and eastern
Tanzania (arcimontensis). In contrast,
forms of the H. denniae assemblage are
centered in eastern African mountains
(Fig. 1): the central portion of the Alber-
tine Rift system in easternmost Demo-
cratic Republic of Congo, Burundi,
Rwanda, and southwestern Uganda (vul-
canorum); astride the northern section of
PROCEEDINGS OF THE BIOLOGICAL SOCIETY OF WASHINGTON
119(2):293?325. 2006.
the Albertine Rift in northeastern Dem-
ocratic Republic of Congo and western-
most Uganda (denniae); and mountains
surrounding the Gregory Rift Valley in
western Kenya (endorobae). Members of
the two groups differ in three qualitative
characters: relative length of the incisive
foramina (longer in the denniae group,
extending between the first molars;
shorter in the anselli group, terminating
about the anterior border of the first
molars), patency of the subsquamosal
fenestra (tiny or occluded in the denniae
group; small to medium size in the anselli
group), and occurrence of pectoral teats
(present in the denniae group, total
mammae 5 8; absent in the anselli group,
total 5 6).
The primary purpose of the current
study is to evaluate morphological differ-
entiation among populations of the H.
denniae complex, as a basis for comment-
ing on the status of the three forms named
from wet montane forest in eastern
Fig. 1. Mountainous regions of eastern Africa inhabited by populations of the Hylomyscus denniae
complex, as circumscribed herein. The 12 geographic samples (OTUs) used in multivariate analyses are
numbered and generally plotted; see Materials and Methods for specific localities comprising each and
sample sizes.
294 PROCEEDINGS OF THE BIOLOGICAL SOCIETY OF WASHINGTON
African mountains?denniae Thomas
(1906), endorobae Heller (1910), and
vulcanorum Lo?nnberg & Gyldenstolpe
(1925). Our secondary goal is to review
the generic affinity of the Kenyan form
endorobae, which, although historically
considered a form of Hylomyscus (e.g.,
Hatt 1940, Musser & Carleton 1993), has
been recently transferred to Praomys
(Van der Straeten & Robbins 1997).
Materials and Methods
Specimens reported herein consist prin-
cipally of skins with their associated
skulls and are contained in the following
museum collections: American Museum
of Natural History, New York City
(AMNH); Carnegie Museum of Natural
History, Pittsburgh (CM); Field Museum
of Natural History, Chicago (FMNH);
Museum of Comparative Zoology, Har-
vard University, Cambridge (MCZ); Na-
tional Museum of Natural History,
Smithsonian Institution, Washington
D.C. (USNM, formerly the U.S. National
Museum); Natural History Museum of
Los Angeles County, Los Angeles
(LACM); The Natural History Museum,
London (BMNH, formerly the British
Museum of Natural History); and the
Rijksuniversitair Centrum, Antwerpen,
Belgium (RUCA).
Fourteen cranial and two dental vari-
ables were measured by MDC to
0.01 mm, using hand-held digital calipers
while viewing crania under a stereomicro-
scope. These measurements, and their
abbreviations as used in text and tables,
are (see Carleton & Van der Straeten 1999
for most landmarks): occipitonasal length
(ONL); greatest zygomatic breadth (ZB);
breadth of braincase (BBC), measured
across the parietal flanges just behind the
zygomatic arches; breadth across occipital
condyles (BOC); interorbital breadth
(IOB); length of rostrum (LR); breadth
of rostrum (BR); postpalatal length
(PPL); length of bony palate (LBP);
breadth of bony palate (BBP), measured
across the maxillary bones above the
second molars; length of incisive foramen
(LIF); length of diastema (LD); breadth
of zygomatic plate (BZP); length of
auditory bulla (LAB); coronal length of
maxillary toothrow (CLM); width of the
upper first molar (WM1). Five external
dimensions (to nearest whole mm) and
body mass (to nearest 0.5 gm) were
transcribed from skin tags or field cata-
logs as given by the collector: total length
(TOTL); head and body length (HBL);
tail length (TL); hindfoot length (HFL);
ear (pinna) length (EL); and weight (WT).
Most external data of the Burundi and
Uganda samples were recorded in the
field by JKP and WTS, reducing among-
collector variability for these big, ex-
tremely age-sensitive, dimensions. The
large series from Mount Kenya was
collected early in the 20th century by J.
A. Loring and E. A. Mearns (Hollister
1919), and specimens lack field-obtained
values for head-and-body length and
weight; the former datum was obtained
by subtraction of TL from TOTL. Re-
cording of measurements was limited to
animals judged to be adult, as based on
the possession of fully erupted third
molars and adult pelage. Three crude
age classes of ??adult?? specimens were
further identified based on degree of
upper molar wear, from little (young
adult), to moderate (full adult), to heavy
(old adult), generally following the pat-
terns of coronal change outlined by
Carleton & Martinez (1991). Except for
a priori exclusion of juvenile specimens,
we did not attempt to remove age-related
size effects (post-weaning growth) in
portraying the multivariate results. Sexual
size dimorphism was statistically unap-
preciable in large population samples of
H. arcimontensis and H. denniae (Carle-
ton & Stanley 2005); measurements of
males and females were accordingly
combined in analyses and tables.
Throughout the text, we use the abbrevia-
VOLUME 119, NUMBER 2 295
tions M1-3 or m1-m3 to individually
reference the upper (maxillary) and lower
(dentary) molars, respectively; individual
cusps of the M1-3 are referenced using the
positional t-system developed by Miller
(1912).
The following 12 operational taxonom-
ic units (OTUs) of Hylomyscus, arranged
according to country of origin, were used
to generate the various tabular summaries
and undertake descriptive and multivari-
ate analyses. Many of these OTUs are
allopatric to one another, restricted to
wet montane forest in isolated mountain
systems (Fig. 1). Locality numbers corre-
spond to the geographic placenames and
coordinates given in Appendix 1; full
provenance and museum registration
numbers are provided in the Taxonomic
Summary. We included a sample of
Praomys delectorum (Kenya, Taita Hills,
Mounts Mbololo and Umengo, N 5 29)
to assess morphometric discrimination
between large-bodied taxa of Hylomyscus
(i.e., denniae, OTU 8, and endorobae,
OTU 2) and a small form of Praomys.
Abbreviations are used here and through-
out the text for forest reserves (FR),
national parks (NP), mount or mountains
(Mt, Mts), and Democratic Republic of
Congo (D. R. Congo). For consistency,
we use the names Albertine Rift Valley
(also 5 Western Rift) and Gregory Rift
Valley (also 5 Eastern Rift or Kenyan
Rift) for the western and eastern
branches, respectively, of the Great Rift
Valley system in eastern Africa.
Burundi: OTU 1?Kibira NP (localities
2?5, 7, 10, 11; N 5 38). Kenya: OTU 2?
Mt Kenya (localities 25, 28; N 5 43).
Rwanda: OTU 3?Virunga Volcanoes
NP, Gahinga and Gasiza (localities 29,
30; N 5 17); OTU 4?Virunga Volcanoes
NP, Kinigi (locality 33; N 5 14); OTU
5?Virunga Volcanoes NP, Karisoke and
Visoke (localities 31, 39) (N 5 24); OTU
6?Nyungwe Forest, Routabansougera
and Uwinka (localities 37, 38; N 5 43);
OTU 7? Nyungwe Forest, Kitabi (local-
ity 34; N 5 12). Uganda: OTU 8?
Ruwenzori Mts, Kasese District, Mubuku
Valley, 1890?2670 m (localities 42, 43,
49; N 5 56); OTU 9?Ruwenzori Mts,
Kasese District, highland localities, 2955?
4040 m (localities 40, 41, 45?47; N 5 28);
OTU 10?Kabale and Kisoro Districts
(localities 48, 50?52; N 5 38). D. R.
Congo: OTU 11?Ruwenzori Mts, Bu-
gongo Ridge and Kalonge (localities 12,
14, 15, 20; N 5 30); OTU 12?Kivu
District (localities 16, 18, 19; N 5 23).
Standard descriptive statistics (mean,
range, and standard deviation) were de-
rived for adult specimens (young, full,
and old age classes combined) of the 12
OTUs. Means and ranges of external
variables are provided as guidance to
identification and as a rough index of
body size (Tables 1, 2), but were not
treated in morphometric analyses. Dis-
Table 1.?External and cranial measurements of
the holotypes of Epimys endorobae Heller and
Epimys taitae Heller. Although closely similar in
size, the former is a large example of the genus
Hylomyscus (endorobae, here considered a valid
species) and the latter a small form of Praomys
(taitae, currently a junior synonym of Praomys
delectorum).
Variable endorobae (USNM 162858) taitae (USNM 181797)
TOTL 251.0 243.0
HBL 100.0 105.0
TL 151.0 138.0
HFL 22.5 23.0
EL 19.5 19.0
ONL 28.2 28.5
ZB 13.6 13.1
BBC 12.1 11.5
IOB 4.4 4.3
LR 9.1 9.9
BR 4.9 4.9
PPL 9.6 9.6
LBP 4.6 4.9
BBP 5.3 5.3
LIF 6.0 6.2
LD 7.7 7.9
BZP 2.4 3.0
LAB 4.5 4.6
BOC 6.2 6.0
CLM 4.4 4.2
WM1 1.4 1.3
296 PROCEEDINGS OF THE BIOLOGICAL SOCIETY OF WASHINGTON
criminant function classification and prin-
cipal component scores were computed
using only the 16 craniodental variables,
all of which were first transformed to
natural logarithms. Principal components
were extracted from the variance-covari-
ance matrix, and variable loadings are
expressed as Pearson product-moment
correlation coefficients of the derived
components or canonical variates with
Table 2.?External and cranial sample statistics (mean, 6 1 SD, and observed range) for Praomys
delectorum (Kenya, Taita Hills, N 5 29) and selected OTUs of the Hylomyscus denniae species group: H.
vulcanorum (OTU 1?Burundi, Kibira NP, N 5 36); H. denniae (OTU 8?Uganda, Ruwenzori Mts, N 5
56); and H. endorobae (OTU 2?Kenya, Mt Kenya, N 5 45).
Variable H. vulcanorum H. denniae H. endorobae P. delectorum
TOTL 227.5 6 14.1 236.1 6 14.3 253.0 6 16.8 229.2 6 7.5
206?260 200?266 215?290 218?243
HBL 94.7 6 7.0 101.7 6 5.4 103.8 6 7.3 101.6 6 4.7
81?107 90?115 92?120 95?112
TL 132.8 6 10.6 135.7 6 8.6 149.4 6 10.8 127.5 6 5.4
117?155 122?157 123?172 118?138
HFL 20.5 6 0.9 21.4 6 1.0 22.8 6 1.2 23.0 6 0.7
19?22 19?23 18?26.5 21?24
EL 17.6 6 1.2 19.9 6 1.1 ? 19.4 6 0.7
16?22 18?23 18?21
WT 21.5 6 4.4 31.5 6 9.2 ? ?
15.5?30.0 17.5?65.0
ONL 25.0 6 0.9 26.7 6 1.1 27.7 6 0.9 28.3 6 0.8
23.1?26.8 24.3?28.7 25.6?29.4 26.9?29.8
ZB 12.6 6 0.4 13.1 6 0.6 13.4 6 0.4 12.9 6 0.3
11.6?13.2 11.4?14.5 12.4?14.5 12.3?13.5
BBC 11.3 6 0.3 11.6 6 0.3 12.1 6 0.3 11.4 6 0.3
10.7?12.0 11.0?12.2 11.5?12.7 10.8?11.7
IOB 4.1 6 0.1 4.1 6 0.1 4.3 6 0.1 4.4 6 0.1
3.9?4.6 3.9?4.4 4.0?4.6 4.2?4.5
LR 7.6 6 0.4 8.2 6 0.4 8.7 6 0.3 9.8 6 0.3
6.9?8.3 7.3?9.2 8.1?9.4 9.3?10.5
BR 4.4 6 0.3 4.5 6 0.3 4.7 6 0.2 4.8 6 0.2
3.9?5.1 4.0?5.1 4.3?5.3 4.5?5.4
PPL 8.9 6 0.6 9.6 6 0.6 9.7 6 0.6 9.4 6 0.4
7.8?10.5 8.1?10.6 8.6?11.0 8.8?10.1
LBP 3.9 6 0.2 4.1 6 0.2 4.5 6 0.2 5.1 6 0.3
3.5?4.3 3.6?4.5 4.1?4.8 4.7?6.1
BBP 4.9 6 0.1 5.1 6 0.2 5.2 6 0.2 5.2 6 0.1
4.6?5.1 4.6?5.5 4.8?5.5 5.0?5.5
LIF 5.5 6 0.3 6.0 6 0.3 6.3 6 0.3 5.7 6 0.3
4.8?6.0 5.3?6.7 5.9?7.0 5.2?6.3
LD 6.8 6 0.3 7.4 6 0.4 7.7 6 0.4 7.7 6 0.4
5.8?7.3 6.5?8.5 6.7?8.2 7.1?8.5
BZP 2.2 6 0.1 2.3 6 0.1 2.4 6 0.2 2.9 6 0.1
1.9?2.4 1.9?2.7 2.0?3.0 2.7?3.2
LAB 4.3 6 0.1 4.3 6 0.1 4.6 6 0.2 4.5 6 0.1
4.0?4.7 3.9?4.6 4.2?4.9 4.2?4.5
BOC 5.8 6 0.2 6.1 6 0.2 6.2 6 0.2 6.0 6 0.2
5.4?6.3 5.6?6.4 5.8?6.5 5.6?6.4
CLM 3.94 6 0.14 4.12 6 0.11 4.34 6 0.13 4.14 6 0.11
3.64?4.20 3.93?4.43 4.12?4.67 3.83?4.39
WM1 1.17 6 0.04 1.24 6 0.04 1.28 6 0.05 1.26 6 0.04
1.07?1.27 1.14?1.32 1.16?1.36 1.17?1.34
VOLUME 119, NUMBER 2 297
the original cranial measurements. All
analytical procedures were implemented
using statistical packages contained in
Systat (Version 10.2.01, 2002). Faunal
similarities among the mountain systems
inhabited by forms of the H. denniae
group were assessed by means of the
Jaccard Index, a matching coefficient for
binary data that scales similarity to range
from 0 to 1 (Hayek 1994). The indices
from the pairwise comparisons were
clustered using UPGMA as implemented
by Systat.
Results
Generic allocation of Epimys endorobae
Heller.?Van der Straeten & Robbins
(1997) referred endorobae Heller (1910)
to Praomys based on their principal
components analyses of holotypes attrib-
utable to Mastomys and related genus-
group taxa (Hylomyscus, Myomys, Myo-
myscus, Praomys). Heller?s (1910) endor-
obae was long associated with Hylomys-
cus, in particular as a subjective synonym
of the species H. denniae (e.g., Allen 1939,
Hatt 1940, Misonne 1974, Musser &
Carleton 1993). In the principal compo-
nent results of Van der Straeten &
Robbins (1997), scores for the holotypes
of endorobae and taitae, the latter epithet
then and now considered a synonym of P.
delectorum, plotted closely in multivariate
space and influenced their decision to
reallocate endorobae to Praomys. Limited
to the holotypes, these results are un-
derstandable: not only do the type speci-
mens of endorobae and taitae resemble
one another in size for most craniodental
dimensions quantified (Table 1), but ho-
mogeneous population samples overlap
substantially in most univariate ranges
(Table 2). Their similarity in size and
possible confusion are also conveyed in
the report of Demeter & Hutterer (1986),
who initially identified Hylomyscus speci-
mens from Mt Meru, Tanzania, as H.
denniae and speculated that taitae from
the Taita Hills, Kenya, is a synonym (the
Meru specimens have been subsequently
redetermined to be examples of Praomys
delectorum?R. Hutterer, pers. comm.).
In order to evaluate these taxonomic
recommendations, we morphometrically
compared a population series of endor-
obae from Mt Kenya (OTU 2) with those
of a large Hylomyscus, H. denniae sensu
stricto from the Ruwenzori Mts, Uganda
(OTU 8), and a small Praomys, P.
delectorum taitae from the Taita Hills,
Kenya (We use this trinomial simply as
a geographic tag in lieu of any evaluation
of infraspecific variation and credible
subspecific divisions within P. delec-
torum).
A multivariate perspective discloses
that the phenetic affinity of endorobae
unambiguously lies with Hylomyscus den-
niae and cannot be confused with a small
Praomys (Fig. 2). Whether principal com-
ponent or canonical variate, the pro-
nounced separation along the first two
factors extracted underscores basic shape
differences between a Hylomyscus skull
and a Praomys skull. The cranium of H.
denniae or endorobae, compared with that
of P. delectorum, is absolutely shorter
(ONL) but comparatively broad across
the zygomatic arches (ZB) and braincase
(BBC, BOC); the interorbital region
(IOB) is more constricted; the rostrum is
truncate and less robust (LR, BR, LD);
the bony palate (LBP) is shorter and the
incisive foramina (LIF) correspondingly
long; and the zygomatic plate (BZP) is
conspicuously narrower (Table 3). Most
of these proportional differences are
readily apparent in mean univariate con-
trasts and side-by-side inspection of skulls
(Tables 1, 2). Molar dimensions (LM1-3,
WM1), on the other hand, contribute
little to specimen dispersion along the
first factor; indeed, the molar rows of
large Hylomyscus such as denniae and
endorobae are fully as robust as small
Praomys (Table 2). In discriminant func-
tion classification, the holotypes of en-
298 PROCEEDINGS OF THE BIOLOGICAL SOCIETY OF WASHINGTON
Fig. 2. Results of multivariate ordinations performed on 16 log-transformed craniodental variables as
measured on 103 intact specimens representing samples of Hylomyscus denniae (OTU 8?Uganda,
Ruwenzori Mts), H. endorobae (OTU 2?Kenya, Mt Kenya), and Praomys delectorum taitae (Kenya, Taita
Hills). Open, inverted triangles correspond to the scores of type specimens of endorobae (USNM 162858)
and taitae (USNM 181797). Top, projection of specimen scores on first two principal components (PC)
extracted. Bottom, projection of specimen scores on first two canonical variates (CV) extracted; average
between-centroid divergences (Mahalanobis D2) among the three samples are indicated along the two-way
arrows. See Table 3 for variable correlations and percent variance explained.
VOLUME 119, NUMBER 2 299
O H. denniae      # H. endorobae      A P. delectorum 
o 
Q_ 
cr> 
> O 
1     - 
-1     - 
- 
I I                I                 I 
% 
o 
o   ? 
o   t.?       # 
o_                               # 
I 
- 
<b           ?       ? 
atf>         ? o          ?9W          Y# 
Rf o      ^ ?*       ^endorobae 
o    o ~.a @*o 
00od9ocP^#^#                     A - A - 
o 
cp                                 te/Yae ?v4 
A 
A 
??                                               A - 
o 
A     A    A 
O                                                          A  A    A       A A o 
0 
A A 
I 
A 
II                              I I 
4     - 
2     - 
0    - 
-1 0 
PC I (56.2%) 
? 
I I 
* 
* ? # 
%         ? # endorobae 
. ?? *? ? 
? / V# 
? _ 
?? ? 
?         ? 
o 
o       o V ^^^ 
A
   taitae / 
AV 
A 
A 
%o       o ^              A >A             A A - 
r,     ?                %         O o                  o   u       > ^_J0]^8______  J%A^     A A 
CL O  O    O     O 
o o o ^(%       o 
o    o o ??     o      ( 
A 
A D   ? 4 Q   o    o    o A" 
(P%o      ? 
oo   0 
A 
o 
o 
I I 
10 
CV1 (80.1%) 
dorobae and taitae were entered as
unknowns, and each was confidently
assigned to their respective species sam-
ple, at a level of mathematical certainty (P
5 1.000) based on the 16 variables we
measured and the 3 groups defined
a priori.
The pattern of variable correlations
lends quantitative validation to funda-
mental shape contrasts that were visually
appreciated by Thomas (1926) in erecting
Hylomyscus as a genus distinct from
Praomys. These include the truncated
rostrum, relatively expanded braincase,
and slender zygomatic plate, which di-
mensions all weigh heavily (Table 3) in
discriminating specimens of denniae-en-
dorobae (Hylomyscus) from those of
taitae (Praomys). The shorter and more
attenuate muzzle of a Hylomyscus versus
Praomys offers a first clue to generic
identification where members of the two
co-occur in the field: the distance between
the tip of the nose and eyeball is notably
less in the former. In addition to obvious
mean differences (BZP?Table 2), Thom-
as? character of the zygomatic plate is
easily grasped in dorsal view, wherein the
junction of the anterior zygomatic arch
and rostrum in Praomys is characterized
by a deep and conspicuous dorsal notch,
but the notch in Hylomyscus is shallow
(Fig. 3). Breadth of the zygomatic-plate,
in combination with its correlate, depth
of the dorsal notch, serves to distinguish
species samples of Hylomyscus and Pra-
omys available to us.
Other features of the holotype and Mt
Kenya series of endorobae similarly con-
form to Hylomyscus denniae and differ
from the Praomys delectorum sample. The
fur in specimens of endorobae is soft and
fine, versus coarser in P. delectorum. The
tail is absolutely and relatively longer in
species of the H. denniae group, including
endorobae (Table 2): tail length relative to
head-and-body length averages about
140% versus about 125% in P. delec-
torum. In addition to relative tail length,
development of terminal penicillation
offers another character to separate east-
ern African examples of Praomys and
Hylomyscus. In Praomys, caudal hairs are
short over the entire tail such that
scalation remains macroscopically evident
even near the tip. In Hylomyscus, in-
cluding specimens of endorobae, caudal
hairs lengthen over the distal one-quarter
to one-fifth of the tail, obscuring the scale
pattern to the naked eye and extending
beyond the terminal vertebra to form
a weak brush or pencil (??the tip rather
more hairy???Thomas, 1926:178). The
hindfoot of endorobae is relatively short
and broad, in metatarsal and phalangeal
proportions like those of H. vulcanorum
and H. denniae and unlike the relatively
longer and narrower pes of P. delectorum.
In particular, digit V in endorobae is long,
its claw reaching nearly to the end of the
Table 3.?Results of principal component and
three-group discriminant function analyses per-
formed on intact specimens (N 5 103) representing
Hylomyscus denniae (OTU 8?Uganda, Ruwenzori
Mts), H. endorobae (OTU 2?Kenya, Mount
Kenya), and Praomys delectorum taitae (Kenya,
Taita Hills); see Fig. 2 and text for discussion.
Variable
Correlations
PCI PCII CV1 CV2
ONL 0.90 0.37 0.50 0.41
ZB 0.34 0.79 20.21 0.35
BBC 0.08 0.70 20.37 0.72
BOC 0.28 0.65 20.20 0.37
IOB 0.70 0.01 0.53 0.32
LR 0.95 20.12 0.86 0.23
BR 0.77 0.34 0.41 0.29
PPL 0.41 0.74 20.15 0.06
LBP 0.87 20.39 0.89 0.25
BBP 0.61 0.14 0.37 0.27
LIF 0.13 0.90 20.40 0.53
LD 0.74 0.57 0.24 0.36
BZP 0.91 20.22 0.82 0.03
LAB 0.63 0.41 0.29 0.66
CLM 0.23 0.40 20.10 0.72
WM1 0.30 0.21 0.07 0.40
Canonical correlations ? ? 0.97 0.88
Eigenvalues 0.30 0.01 14.0 3.48
Percent variance
explained
56.2 21.5 80.1 19.9
300 PROCEEDINGS OF THE BIOLOGICAL SOCIETY OF WASHINGTON
third or ungual phalanx of digits II?IV; in
P. delectorum, the fifth digit is notably
shorter relative to digits II?IV, extending
to the middle of the second phalanx
(Fig. 4). This distinction is easily ob-
served in the field or museum and helps
to differentiate juveniles of Hylomyscus
and Praomys. The first digit (hallux) is
also relatively longer in Hylomyscus,
extending to the middle of the first
Fig. 3. Lateral (left) and dorsal (right) views of the rostrum and anterior zygomatic arch in species of
Praomys and Hylomyscus: top, P. delectorum (FMNH 151303) from Tanzania; bottom, H. denniae (FMNH
144611) from Uganda, Ruwenzori Mts. Abbreviations: fr, frontal; max, maxillary; na, nasal; nc,
nasolacrimal capsule; pre, premaxillary; zn, zygomatic notch; zp, zygomatic plate.
VOLUME 119, NUMBER 2 301
phalanx of digit II, but the contrast with
Praomys is not so visually prominent and
less useful for field identification to genus.
The t3 (labial anterocone) of the upper
first molar in endorobae, as well as that of
H. denniae and vulcanorum, is small and
weakly demarcated from the t2 (lingual
anterocone). Consequently, the anterior
chevron of Hylomyscus generally appears
narrow with suppressed cuspidation. In
the Praomys jacksoni and P. delectorum
complexes (but not P. tullbergi?see
Lecompte et al. 2002a), the t3 of the M1
constitutes a small but discrete cusp, even
in adults with moderately worn molars,
and the anterior chevron is correspond-
ingly broader. Contrary to Thomas?
(1926) claim???M1 without antero-exter-
nal cusplet???a t3 is not wholly absent in
Hylomyscus and its development varies
among species. In most, its homologous
position is usually indicated by a short
and shallow indentation on the antero-
lateral face of the M1, as observed in
unworn to barely worn molars of juve-
niles and very young adults. After occlu-
sal wear begins, anatomical demarcation
of a t3 is lost, such that it appears as little
more than a posterolateral spur from the
medial t2. Only in H. parvus is the t3 (and
t9) of M1 so miniscule that it may be said
to be absent; the anterior chevron of this
species assumes a very narrow and highly
asymmetrical shape. Specimens of H. aeta
and H. baeri possess a relatively large t3,
only slightly smaller than the t1, and the
first chevron of M1 is basically symmet-
rical. Still, the t3 in these species of
Hylomyscus is not as physically well
demarcated from the t2 as in the Praomys
jacksoni and P. delectorum complexes and
their anterior chevron remains narrower.
A skeletal trait, occurrence of the
entepicondylar foramen, also associates
endorobae as a large Hylomyscus instead
of a small Praomys. Among muroid
rodents, an entepicondylar foramen may
occur on the distal humerus, above the
medial epicondyle, and transmits the
median nerve and sometimes the brachial
artery (Carleton 1980:Fig. 13, Landry
1958). LeCompte et al. (2002a) documen-
ted that this humeral foramen occurs in
two species of Hylomyscus (H. aeta, H.
stella), and we have found it to be
uniformly present in seven other species,
including endorobae (Table 4). Examples
of P. jacksoni, type species of the genus,
and P. delectorum lack an entepicondylar
foramen, but the foramen?s occurrence
varies within the genus as currently
arranged?present in species of the tull-
bergi complex and absent in those of the
jacksoni and delectorum complexes (Le-
Compte et al. 2002a, this study). The
latter variability bears on the larger issue
of the generic boundaries of Praomys
proper (see LeCompte et al. 2002a, 2002b
for discussion). For our purposes, it is
significant that specimens of endorobae
agree with those of H. denniae and other
Hylomyscus in possessing an entepicon-
Fig. 4. Dorsal view of metatarsal and phalan-
geal proportions in the left hind foot of Praomys
and Hylomyscus: left, P. delectorum (FMNH
151303; Tanzania); right, H. denniae (FMNH
144587; Uganda, Ruwenzori Mts). Roman numerals
I through V identify pedal digits one through five.
Note the relatively longer digits I and V in H.
denniae as compared to P. delectorum.
302 PROCEEDINGS OF THE BIOLOGICAL SOCIETY OF WASHINGTON
dylar foramen (Fig. 5), a structure that is
lacking in a small Praomys such as P.
delectorum.
Morphometric differentiation among
samples of the H. denniae complex.?
Three well defined clusters are evident in
plots of specimen scores onto the first two
canonical variates extracted from discrim-
inant function analysis of our 12 OTUs
(Fig. 6). The largest grouping includes
mice that originated from mountains in
southwestern Uganda (OTU 10), Burundi
(OTU 1), Rwanda (OTUs 3?7), and
nearby D. R. Congo (OTU 12); a second
consists of those from west-central
Uganda (OTUs 8, 9) and contiguous D.
R. Congo (OTU 11); and the third is
composed of the single sample from
Kenya (OTU 2). Each of these three
major clusters of eastern African Hylo-
myscus correspond to already named
populations?vulcanorum, denniae, and
endorobae, respectively?that inhabit iso-
lated mountain systems. Multivariate
overlap of specimen scores among in-
dividual OTUs within each of the denniae
and vulcanorum clusters is substantial,
and a posteriori misclassification of speci-
mens between them is high (typically 40?
70%); few individuals, however, were
misallocated among samples of the three
taxa (2?7%). Craniodental size plays
a predominant role in the dispersion of
samples along the first canonical variate
(most variable correlations moderate to
large and all positive?Table 5), in par-
ticular accounting for the separation of
the smaller vulcanorum from the larger
denniae and endorobae. The closer phe-
netic association of the latter two relative
to vulcanorum is similarly reflected in the
last clustering division using Mahalanobis
distances between group centroids
(Fig. 6). Segregation of endorobae from
denniae and vulcanorum along CV2 is
emphasized by certain weakly correlated
Table 4.?Occurrence of the entepicondylar fora-
men in species of Hylomyscus and Praomys (see
Appendix 2 for localities and specimen numbers).
Species (No. specimens) Present (+)/Absent (2)
H. aeta (17) +
H. alleni (2) +
H. anselli (14) +
H. arcimontensis (18) +
H. denniae (20) +
H. endorobae (5) +
H. parvus (3) +
H. stella (20) +
H. vulcanorum (22) +
P. degraaffi (21) 2
P. delectorum (27) 2
P. jacksoni (20) 2
P. misonnei (20) +
P. rostratus (2) +
P. tullbergi (13) +
Fig. 5. Frontolateral view of the left humerus in
examples of Hylomyscus denniae (left?FMNH
144626; Uganda, Ruwenzori Mts) and Praomys
delectorum (right?FMNH 147314; Tanzania, West
Usambara Mts). Note the presence of an entepi-
condylar foramen above the medial epicondyle in H.
denniae, a structure found in all Hylomyscus species
examined but lacking in small Praomys (see Table 4).
VOLUME 119, NUMBER 2 303
Fig. 6. Results of discriminant function analysis performed on 16 log-transformed craniodental variables
as measured on 304 intact specimens representing 12 OTUs of eastern African Hylomyscus: denniae (8, 9,
11), endorobae (2), and vulcanorum (1?7, 10, 12). Top, projection of specimen scores onto the first two
canonical variates extracted; irregular polygons enclose maximal dispersion of specimen scores around an
OTU centroid (numbered symbols). Bottom, UPGMA phenogram based on Mahalanobis distances between
the 12 OTUs. See Table 5 for variable correlations and percent variance explained.
304 PROCEEDINGS OF THE BIOLOGICAL SOCIETY OF WASHINGTON
CM 
CM 
> 
o 
2     - 
0    - 
-2     - 
-5 
# H.vulcanorum A /-/. denniae A H. endorobae 
- 
i i         i         i i           l           l           I i         i         i 
\            11X 
? \ 
\y^\^K2/^ 
^\^            - 
2A                  /      " 
- 
i         i         i         i 1           1           1           1 I            I            I 
-2-1012 
CV 1 (72. 4%) 
H. endorobae 
H. denniae 
H. vulcanorum 
Kenya (2) 
Uganda (8) 
Uganda (9) 
D.R.Congo (11) 
Uganda (10) 
Rwanda (5) 
DR. Congo (12) 
Rwanda (7) 
Rwanda (3) 
Rwanda (6) 
Rwanda (4) 
Burundi (1) 
50 40 30 20 10 
Mahalanobis D 
variables that reflect the relatively wider
braincase (BBC, IOB), longer bony palate
(LBP), and larger bullae (LAB) in speci-
mens of endorobae (Tables 2, 5). Al-
though clearly circumscribed in multivar-
iate space, as depicted in our 12-group
discriminant function analysis, the gener-
alized distance distinguishing endorobae
from the three samples of denniae (MD2
5 27.5) is a small value in our experience
with other African species comparisons,
which typically match or exceed that
recorded here for denniae versus vulca-
norum (MD2 5 36).
Only modest separation of denniae and
endorobae is also evident in our principal
component analyses, both that obtained
in comparisons of denniae-endorobae-tai-
tae (Fig. 2) and that derived only from
samples representing denniae-endorobae
(Fig. 7). Although PCA scores for Hylo-
myscus from the Ruwenzoris (OTUs 8, 9,
11 5 denniae) and Mt Kenya (OTU 2 5
endorobae) marginally overlap, the major
axes of their elliptical spreads are phenet-
ically discrete, their Y-intercepts being
significantly different (F 5 125.7, P ,
0.001) while their slopes are comparable
(F 5 0.2, P 5 0.643). This covariation
pattern among the craniodental variables
of denniae and endorobae is a familiar one
that resembles multivariate contrasts re-
covered for other closely related, conge-
neric species of Muroidea (Voss & Mar-
cus 1992, Voss et al. 1990). Although
juveniles were omitted from our analyses,
post-weaning growth still accounts for
much of the elongate scatter observed
within the species samples and for the
oblique orientation of the group constel-
lations on the first and second principal
components. Thus, in one-way analyses
of variance, age-class as a categorical
effect contributes significantly to the
spread of specimen scores along PC I (F
5 78.0, P , 0.001) but not PC II (F 5
1.3, P 5 0.275); whereas, species (or
OTU) as an effect contributes substan-
tially to dispersion on PC II (F 5 131.5, P
, 0.001) but much less so on PC I (F 5
5.9, P 5 0.017). Predictably, all variables
load positively on the first component
and most correlations are moderate to
large (r 5 0.50?0.90; Table 6). Those
variables that load heavily on PC II,
more or less orthogonal to the major axes
of the taxon samples, include the same
ones highlighted for CV 2 in the discrim-
inant function analysis (BBC, IOB, LBP,
LAB), although component correlations
are accentuated. Two additional large
correlations with PC II emphasize the
more robust molars (LM1-3, WM1)
observed in examples of endorobae as
compared with those of denniae (Tables 2,
6).
Taxonomic Summary
Based on the morphometric and mor-
phological evidence presented above, we
continue to view endorobae of central
Kenya as a constituent form of Hylomys-
cus, one related to H. denniae in the
Table 5.?Results of discriminant function anal-
ysis performed on intact specimens (N 5 304)
representing 12 OTUs of eastern African Hylomys-
cus; see Fig. 6.
Variable
Correlations
CV1 CV2
ONL 0.77 20.10
ZB 0.52 0.02
BBC 0.73 20.29
BOC 0.75 0.08
IOB 0.25 20.28
LR 0.74 20.16
BR 0.21 20.24
PPL 0.53 0.11
LBP 0.67 20.29
BBP 0.73 0.14
LIF 0.62 20.23
LD 0.68 20.06
BZP 0.59 20.01
LAB 0.32 20.37
CLM 0.81 20.18
WM1 0.84 20.10
Canonical Correlations 0.93 0.70
Eigenvalues 5.63 0.93
Percent variance 72.4 12.1
VOLUME 119, NUMBER 2 305
Fig. 7. Results of principal component analysis (Varimax rotation) performed on intact specimens (N 5
137) representing samples from the Ruwenzori Mountains (OTUs 8, 9, 11) and Mount Kenya (OTU 2). Top,
projection of specimen scores onto the first two principal components extracted; regression lines of PC II on
PC I indicated for specimens representing each taxon. Bottom, regression lines and centroids as calculated
a posteriori for each OTU; ellipses correspond to 95% confidence limits of a group?s centroid; individual
specimen scores not plotted. See Table 6 for variable correlations and percent variance explained.
306 PROCEEDINGS OF THE BIOLOGICAL SOCIETY OF WASHINGTON
o Ruwenzori Mtns (denniae) 
Mt Kenya 
(endorobae) 
CO 
^ 0 
O 
0_ 
o 
o    o ^        ^^^-%--z2L^        o 
o 
o o 
o q.n; 
-2     - 
-3 -2 
Kenya (2) 
1     - 
CO 
^ 0 
O 
Q_ 
Uganda (9) 
D.R. Congo (11) 
Uganda (8) 
-2 
-3 _L _L 
-1 0 
PC I (52.0%) 
Ruwenzori Mts, Uganda. Trenchant fea-
tures that support this generic assignment
include the very long tail, the relatively
long fifth digit of the hindfoot, truncate
rostrum with slim zygomatic plate, in-
distinct t3 of M1, and presence of an
entepicondylar foramen. Thomas long
ago anticipated this conclusion, for in
his description of Mus denniae, he re-
marked (1906:145) that ??A mouse from
Mount Kenya ? would also seem to be
assignable to Mus Denniae.?? The Kenyan
populations were soon after formally
proposed as a new species, Epimys
endorobae Heller (1910), and the taxon
was subsequently considered either as
a full synonym of Hylomyscus d. denniae
(Hollister 1919, Allen 1939) or as a defin-
able subspecies of H. denniae (Hatt 1940).
These past allocations more accurately
convey the relationship of endorobae,
although we dispute the taxonomic rank
accorded by those authors.
We consider each of the three well
circumscribed phena revealed within east-
ern African ??denniae?? to represent valid
species, for which the names denniae
Thomas (1906), vulcanorum Lo?nnberg &
Gyldenstolpe (1925), and endorobae Hell-
er (1910) are available and geographically
applicable. The three differ predominant-
ly in size (Table 2; Figs. 8, 9)?H. vulca-
norum the smallest to H. endorobae the
largest?and more subtly in shape?H.
endorobae with a broader skull, longer
palate, inflated bullae, and larger molars
relative to H. denniae and H. vulcanorum.
One could plausibly attribute such differ-
ences among isolated populations to geo-
graphic variation and interpret their
status as subspecies of a highly variable
H. denniae, as did Hatt (1940) and Bishop
(1979 [who did not specifically mention
endorobae, but attributed Kenyan exam-
ples to H. d. denniae]). However, mor-
phometric differentiation between popu-
lations that represent each epithet is
sharply drawn and intimates morphoge-
netic discontinuity (species) instead of
population intergradation (subspecies).
Moreover, the degree and pattern of
craniodental differentiation among these
montane Hylomyscus conform to that
which has emerged from other multivar-
iate studies of morphologically similar,
closely related (congeneric) species of
African murids?e.g., those of Dasymys
(Carleton & Martinez 1991, Verheyen et
al. 2003), Hybomys (Van der Straeten &
Verheyen 1982, Carleton & Robbins
1985), Lemniscomys (Van der Straeten
1980, Carleton & Van der Straeten 1997),
Lophuromys (Verheyen et al. 2002), Ma-
lacomys (Van der Straeten & Verheyen
1979), and Praomys (Van der Straeten &
Kerbis Peterhans 1999). In several of the
foregoing studies, karyological and-or
genetic data are reported that supplement
the conventional morphometric results
and thereby augment the case for separate
species. Such independent data sets
should be applied to evaluate our recom-
Table 6.?Results of principal component analy-
sis (Varimax rotation) performed on intact speci-
mens (n 5 137) representing samples from the
Ruwenzori Mountains (OTUs 8, 9, 11) and Mount
Kenya (OTU 2); see Fig. 7.
Variable
Correlations
PC I PC II
ONL 0.90 0.29
ZB 0.83 0.13
BBC 0.49 0.39
BOC 0.56 0.25
IOB 0.42 0.40
LR 0.84 0.35
BR 0.85 0.24
PPL 0.87 20.05
LBP 0.26 0.86
BBP 0.38 0.36
LIF 0.84 0.17
LD 0.91 0.17
BZP 0.52 0.19
LAB 0.51 0.64
CLM 0.07 0.85
WM1 0.01 0.65
Eigenvalues 0.021 0.006
Percent variance
explained
52.0 14.8
VOLUME 119, NUMBER 2 307
mendation for the specific distinction of
endorobae and vulcanorum from H. den-
niae. In one such preliminary analysis,
using two mitochondrial gene fragments
(Huhndorf 2001), substantial sequence
divergence (4.8?5.8%) was demonstrated
between a sample from the Ruwenzoris
(5 H. denniae) and three others from
mountains in the central Albertine Rift
(Mgahinga Gorilla NP, Bwindi Impene-
trable NP, and Kibira NP 5 H. vulca-
norum). Denser geographic representation
and wider taxonomic sampling within and
among forms of Hylomyscus are needed.
Fig. 8. Dorsal and ventral views (ca. 2.53) of adult crania of: left, H. denniae (FMNH 144495; ONL 5
28.0 mm), a female from Uganda, Ruwenzori Mts, near the confluence of the Mubuku and Bujuku rivers;
middle, Hylomyscus endorobae (USNM 163374; ONL 5 27.1 mm), a male from Kenya, Mt Kenya, western
side; and right, H. vulcanorum (FMNH 137661; ONL 5 25.5 mm), a female from Burundi, Kibira NP,
2.3 km N and 1.9 km W Teza.
308 PROCEEDINGS OF THE BIOLOGICAL SOCIETY OF WASHINGTON
The kinship hierarchy among the three
species of the H. denniae group is
ambiguous based on our data. Hylomys-
cus denniae and H. endorobae are phene-
tically most alike, as revealed in ordina-
tions of craniodental variables. On the
other hand, the mountain systems inhab-
ited by H. denniae and H. vulcanorum are
geographically much closer to one anoth-
er, forming the northern and central
portions of the Albertine Rift system,
respectively, and faunal association be-
tween these is much stronger than that
with the mountains in west-central Kenya
where H. endorobae occurs (see Discus-
sion). The geographic context, along with
weak similarities in craniodental shape,
instead offers circumstantial evidence for
closer kinship of H. denniae and H.
vulcanorum. Although the three species
appear to be derived from a recent
common ancestor, we view their relation-
ship as an unresolved polytomy at this
stage of understanding. A phylogeo-
graphic approach using gene-sequence
information should decidedly resolve this
uncertainty.
Synonymies presented below for the
three species of the H. denniae group trace
earliest identification and first subsequent
usage of other name combinations. We
have personally examined the holotypes
of denniae Thomas (1906) and endorobae
Heller (1910); moreover, in the case of
denniae, the large series recently collected
from the valley of the Mubuku River by
FMNH personnel essentially represents
topotypic material. The holotype of
vulcanorum Lo?nnberg & Gyldenstolpe
(1925) was not studied, but the ample
material included from other volcanoes in
the Virunga mountains supplies confi-
dence in the application of this name to
the smallest of the three species. Speci-
mens examined include all individuals
personally seen and identified by us; see
Appendix 1 for geographic coordinates of
principal localities.
Hylomyscus denniae (Thomas)
Mus denniae Thomas, 1906:144 (type
locality?Uganda, Ruwenzori East,
Mubuku Valley, 7000 ft; holotype?
BMNH 6.7.1.137).
Hylomyscus denniae, Thomas, 1926:178.
Hylomyscus denniae denniae, Allen,
1939:386.
Rattus (Hylomyscus) denniae denniae,
Ellerman, 1941:209.
Praomys (Hylomyscus) denniae, Misonne,
1974:28.
Praomys (Hylomyscus) denniae denniae,
Delany, 1975:85.
Distribution.?Restricted to middle to
upper slopes of the Ruwenzori Mts, in
both west-central Uganda and contiguous
Democratic Republic of Congo (Fig. 10);
known elevational range, 1890?4040 m.
Remarks.?As we understand the spe-
cies, Hylomyscus denniae is endemic to the
Fig. 9. Lateral views (ca. 2.53) of adult crania
of the same specimens portrayed in Fig. 8: top, H.
denniae; middle, Hylomyscus endorobae; and bot-
tom, H. vulcanorum.
VOLUME 119, NUMBER 2 309
Ruwenzori Mountains, but specimens
from montane forest on Mt Elgon also
have been referenced as H. denniae
(Clausnitzer & Kityo 2001). Clarification
of their species identification is the subject
of on-going research (Carleton et al.,
unpubl.). Although the occurrence of
some denniae-like form on Mt Elgon
seems geographically plausible, the few
specimens from the Kenyan slopes that
we have examined to date prove to be
examples of the H. anselli complex (Mt
Elgon NP?USNM 589874, 589876,
589879?589881).
Elevational transects along both west-
ern (D. R. Congo) and eastern (Uganda)
slopes of the Ruwenzoris have documen-
ted H. denniae from wet montane forest
through alpine habitats (Kerbis Peterhans
et al. 1998; Misonne 1963). In Uganda,
the species was uncommon (4.3?6.6%)
relative to other rodents obtained in the
forest zone (1920?2100 m) but was more
abundant in habitats above 2100 m, in-
cluding Podocarpus forest, bamboo (Ar-
undinaria alpina), tree heath and moss,
and alpine vegetation characterized by
Senecio and Lobelia (Kerbis Peterhans et
Fig. 10. Known distributions of Hylomyscus denniae, H. endorobae, and H. vulcanorum in eastern Africa,
as based on specimens studied by us (see Taxonomic Summary).
310 PROCEEDINGS OF THE BIOLOGICAL SOCIETY OF WASHINGTON
al. 1998). Within Podocarpus forest
(2700 m), H. denniae supplanted Praomys
jacksoni (montis) as the dominant member
of the rodent community (32?39% of
captures). A similar elevational range (ca.
2000?4500 m), pattern of habitat occupa-
tion, and relative abundance of H.
denniae versus P. jacksoni are apparent
along the western Ruwenzori slope in D.
R. Congo (Heller, 1925 field notes,
FMNH archives; Misonne 1963). Mis-
onne (1963) recorded a 1000-m gap
between the highest occurrence of H.
stella, 1000 m in evergreen rainforest,
and the lowest capture of H. denniae,
2000 m in montane forest, suggesting
elevational fidelity to different vegetation
zones.
Specimens examined.?DEMOCRAT-
IC REPUBLIC OF THE CONGO: Mt
Ruwenzori, Bugongo Ridge, 8500 and
9000 ft (AMNH 82448, 82457, 82458,
82467?82469, 82509, 82511; FMNH
26083?26085, 26088, 26089, 96029?
96031); Mt Ruwenzori, summit of Bu-
gongo Ridge, 11,000 ft (FMNH 26086);
Mt Ruwenzori, Butego Valley, Ibale,
7500 ft (FMNH 26103, 26105?26107,
26109); Mt Ruwenzori, Butagu Valley,
Kalonge Village, 7000 ft (FMNH 26026);
Mt Ruwenzori, Katahuleko Creek west of
Kalonge, 7000 ft (AMNH 82462; FMNH
26096?26099); Mt Ruwenzori, Old Ka-
longe, 7500 ft (FMNH 26091, 26092); Mt
Ruwenzori, Ugongo (5 Old Kalonge),
7500 ft (FMNH 26095).
UGANDA: Mt Ruwenzoni, Mabuku
Valley, east slope, 8500 ft (USNM
172919); Kasese District, Ruwenzori
Mts, confluence of Mubuku and Bujuku
rivers, Nyabitaba Hut, 8750 ft (FMNH
144475, 144479?144482, 144485?144489,
144491?144495, 144502, 144504, 144505,
144507, 144511, 144513?144515, 144518,
144519, 144522?144525, 144527, 144530?
144532, 144536, 144538, 144540?144542,
144550, 144552, 144557, 144558, 144560?
144562, 144568, 144571, 144577, 144579,
144580); Kasese District, Ruwenzori Mts,
periphery of Lake Mahoma, 9700 ft
(FMNH 144583, 144584); Kasese Dis-
trict, Ruwenzori Mts, left bank Bujuku
River, John Mate Camp, 11,050 ft
(FMNH 144586, 144587, 144589,
144591?144596, 144598?144605, 144608,
144611); Kasese District, Ruwenzori Mts,
right bank Bujuku River, Bigo Hut,
13,000 ft (FMNH 144612); Kasese Dis-
trict, Ruwenzori Mts, above Lake Bu-
juku, Bujuku Hut, 13,050 ft (FMNH
144613, 144614, 144626?144633, 144642?
144645); Kasese District, Ruwenzori Mts,
Lower Kitandara Lake, 13,250 ft
(FMNH 144646); Kasese District, Ru-
wenzori Mts, Mubuku Valley, right bank
Mubuku River near confluence of Mu-
buku and Kyoha Rivers, 3 km W Ibanda,
6200 ft (FMNH 144448?144451, 144453?
144457); Kasese District, Ruwenzori Mts,
Mubuku Valley, confluence of Mubuku
and Mahoma rivers, 6900 ft (FMNH
144459?144470).
Hylomyscus endorobae (Heller)
Mus Denniae, Thomas, 1906:145 (in
part?referral of Mt Kenya specimen).
Epimys endorobae Heller, 1910:3 (type
locality?Kenya, western edge of Mau
Forest, 25 mi N Eldoma [5 Eldama]
Ravine, 8600 ft; holotype?USNM
162888).
Rattus denniae, Hollister, 1919:79 (allo-
cated as junior synonym).
Hylomyscus denniae denniae, Allen,
1939:386 (considered full synonym of
nominate race, following Hollister?s
[1919] action).
Hylomyscus denniae endorobae, Hatt,
1940:536 (reinstated as a valid sub-
species).
Rattus (Hylomyscus) denniae denniae, El-
lerman, 1941:209 (generic reallocation).
Praomys (Hylomyscus) denniae, Misonne,
1974:28 (listed in synonymy without
indication of rank).
Distribution.?Highlands of west-cen-
tral Kenya (Fig. 10), including Mt Kenya,
VOLUME 119, NUMBER 2 311
the Aberdare Mts, and Mau Escarpment;
extent of geographic range uncertain;
known elevational range, 2135?3260 m.
Remarks.?Heller (1910) considered his
new species endorobae to be a close ally of
H. denniae, ??distinguished by much dar-
ker coloration, smaller ears, and larger
molars.?? We appreciate his first and third
differentiating features but cannot quan-
titatively substantiate the smaller pinnae
of H. endorobae. The direction of the
mean difference is supportive (?x < 20 mm
in H. denniae versus ?x < 18 mm in H.
endorobae), but sample size of H. endor-
obae with reliable ear measurements was
too small (N 5 5) to tabulate and
statistically contrast.
Detailed ecological information is lack-
ing for H. endorobae, although most
examples can be generally associated with
montane forest. At the type locality
(25 mi N Eldama Ravine, 2620 m) in
the Mau Escarpment, Heller (1910) re-
ported that his new species was found
only in montane forest, commonly in
association with Praomys jacksoni. In his
field notes, Heller (27 Nov 1909) further
characterized the habitat as ??vegetation
in forest, juniper, yew, and bamboo with
gray olive and Hypericum on margin.??
The extensive series from Mt Kenya was
obtained at five collecting camps sited
within the forest zone, approximately
from 2135 to 3260 m. Most specimens
were collected in a ??belt composed mainly
of bamboo [Arundinaria] and yellow-
wood [Afrocarpus],?? 2590?3260 m, but
some were obtained from a lower forest
belt composed of cedars (Juniperus) and
various hardwood trees, 2135?2590 m
(Mearns, in Roosevelt 1910:489). Again,
Praomys jacksoni was often taken in
sympatry with H. endorobae on Mt
Kenya, especially at the 3050-m and
3260-m camps (Loring, in Roosevelt
1910:484). No H. endorobae was
recorded around the highest and only
collecting station established in afroalpine
habitat proper (4175 m???a zone of
giant heath, mixed with giant groundsels
and shrubs??; Mearns, in Roosevelt,
1910:489).
In addition to H. endorobae, at least
three other species of Hylomyscus occur
in Kenya, H. aeta, H. stella (kaimosae),
and H. cf. anselli. The lone sample of H.
aeta so far recorded is from the Kaka-
mega FR (CM 98254) in extreme west-
central Kenya. Examples of H. stella
are known to us from Kaimosi (USNM
183182?183184) and Kakamega FR (CM
98251?98253, 98256, 98257, 102459?
102461; USNM 589877, 589878), nearby
localities again in far west-central Kenya.
Examples of H. cf. anselli are found on
the eastern slopes of Mt Elgon (see H.
denniae account), the Guas Ngishu Pla-
teau (USNM 163349?reported as endor-
obae by Hollister 1919), and from 4 mi E
Molo (USNM 437428, 437468). In places,
the species may prove to be sympatric
with H. endorobae; e.g., the species we
identify from Molo is H. endorobae
(FMNH 17024) and that from 4 mi E
Molo is H. cf. anselli. Additional collect-
ing, especially along designed elevational
transects, is required to illuminate the
number, distribution, and ecology of
Hylomyscus species in Kenya.
Specimens examined.?KENYA: Aber-
dare Mts, 10,000 and 10,500 ft (USNM
163348, 183312); Aberdare Mts NP,
Fishing Lodge, 9000 ft (USNM 589869);
Aberdare Mts, Bamboo Camp, 8650 ft
(AMNH 114025); Nasin Gishu District,
7500 ft (AMNH 34789); western edge of
Mau Forest, 25 mi N Eldama Ravine,
8600 ft (USNM 162888, holotype); west
side of Mt Kenya, various elevations from
7000 to 10,700 ft (AMNH 80984; CM
2843, 2856; FMNH 43454?43457; USNM
163347, 163357?163362, 163364, 163365,
163367?163377, 163379?163384, 163386,
1163387, 163389, 163390, 163395?
163401); Mt Kenya, east slope, 10,300 ft
(AMNH 82452, 82453); Mt Kenya, Naro
Moru, 9000 and 10,000 ft (USNM
589870?589873, 589875); Kikuyu, Mu-
312 PROCEEDINGS OF THE BIOLOGICAL SOCIETY OF WASHINGTON
gaga North, ca. 7200 ft (FMNH 86322,
86335); Kijabe (FMNH 16987, 16989,
17013); Rift Valley Province, Nakuru
District, Molo (FMNH 17024); Nyeri
District, South Laikipia Forest, 6 mi W
Bellevue, 10,000 ft (AMNH 187595).
Hylomyscus vulcanorum (Lo?nnberg &
Gyldenstolpe, 1925), new rank
Rattus (Praomys) denniae vulcanorum
Lo?nnberg & Gyldenstolpe, 1925:4 (type
locality?Belgian Congo [5 D. R.
Congo], Birunga [5 Virunga] Volca-
noes, Mount Karisimbi, 3800 m).
Hylomyscus denniae vulcanorum, Allen,
1939:386.
Rattus (Hylomyscus) denniae vulcanorum,
Ellerman, 1941:209.
Praomys (Hylomyscus) denniae, Misonne,
1974:28 (listed in synonymy without
indication of rank).
Praomys (Hylomyscus) denniae vulca-
norum, Bishop, 1979:528.
Distribution.?Mountains bordering
the central Albertine Rift valley, from
southwestern Uganda and east-central
Democratic Republic of Congo, through
Rwanda, to southern Burundi (Fig. 10);
known elevational range, 1670?3100 m.
Remarks.?As herein documented, H.
vulcanorum occurs as far south as 4uS
latitude in southern Burundi, in mountain
chains to the west and east of Lake Kivu
in the central Albertine Rift. Along the
southern portion of the Albertine Rift,
mountains to the east of Lake Tanga-
nyika, in western Tanzania (Mahale Mts,
ca. 06u309S, and Mbizi Mts, ca. 07u509S;
specimens in FMNH), harbor forms of
the H. anselli group (Carleton & Stanley
2005, this study); mountains at compara-
ble latitudes to the west of Lake Tanga-
nyika, in southeastern D. R. Congo
(Itombwe, Marungu, Mt Kabobo), have
been little surveyed for small mammals
and their Hylomyscus remain unknown.
To the north, H. vulcanorum ranges into
SW Uganda at least as far as Bwindi
Impenetrable NP, about 1uS latitude.
Bishop (1979) also referred specimens
from the Cherengani Hills, Kenya, and
Mt Elgon, Uganda, to his Praomys
denniae vulcanorum. We have not yet
examined these specimens but question
their identification as H. vulcanorum
proper. As noted in the H. denniae
account, those high-elevation occurrences
of Hylomyscus on Mt Elgon seen by us
are a species referable to the H. anselli
complex. Perhaps Bishop?s series also
represent this species, which in aspects
of general size recalls H. vulcanorum sensu
stricto.
At Bwindi Impenetrable NP in south-
western Uganda, H. vulcanorum occurs at
2350 m (Ruhizha), but at lower sites
(FMNH collections) it is replaced by its
congeners H. stella (e.g., at Omubiyanja
Swamp, 1850 m; Nteko Parish, 1600 m;
Byumba, 1540 m; Buhoma, 1500 m) or
H. aeta (also at Nteko Parish, 1600 m),
an elevational replacement analogous to
distributions of Hylomyscus on the slopes
of the Ruwenzori Mts. In montane forests
of Burundi and Uganda, H. vulcanorum
was commonly obtained in traps placed
above ground, averaging about 20% in
arboreal captures (Kerbis, unpubl.).
Specimens examined.?BURUNDI: Bu-
banza Province, Kibira NP, Ntamba
Zone, Kubutare Colline, Ruhondo,
2000, 2020, and 2030 m (FMNH
148455, 148469?148471, 148474); Bu-
banza Province, Kibira NP, Rugazi Com-
mune, Kabanga Colline, Nyabikona Riv-
er, 1950 m (FMNH 148458); Bubanza
Province, Bururi FR, Nyagatarugwa Riv-
er Valley, 1800 m (FMNH 148462);
Cibitoke Province, Bukinanyana Com-
mune, Kibira NP, Ndoro zone, Kwo-
gofe Colline, 2200 m (FMNH 148463,
148472); Cibitoke Province, Bukinanyana
Commune, Kibira NP, Ndora zone,
Giserama Colline, Ruvyirame River,
2.1 km N Kirumura, 1670 m (FMNH
149084?149086, 149088); Cibitoke Prov-
ince, Bukinanyana Commune, Kibira
VOLUME 119, NUMBER 2 313
NP, Ndora zone, Giserama Colline,
Gatare Stream, 1.9 km N Kirumura,
1950 m (FMNH 149090, 149091, 149094,
149095); Kayanza Province, Matongo
Commune, Kibira NP, Abris Kivuso,
2430 m (FMNH 148464); Kayanza Prov-
ince, Muruta Commune, Kibira NP,
Gitenge River, 2177 m (FMNH 148465?
148468); Muryamuya Province, Kibira
NP, 2.9 km N and 1.4 km W Teza, 2480
and 2520 m (FMNH 137657, 137658);
Muryamuya Province, Kibira NP, 2.3 km
N and 1.9 km W Teza, 2330?2390 m
(FMNH 137659?137662, 137699); Mur-
yamuya Province, Kibira NP, 3.5 km S
and 2.6 km W Teza, 1970?2050 m
(FMNH 137663?137671); Muryamuya
Province, Bukeye Commune, Kibira
NP, Nyamugari Colline, Teza, 2080 m;
03 14 S, 29 34 E (FMNH 148473); Bururi
Province, Bururi FR, Mumushwizi Val-
ley, 1880 m (FMNH 155981, 155982,
155984?155990); Bururi Province, Bururi
FR, Ruhinga Hill, 2170 m (FMNH
155991, 155992).
DEMOCRATIC REPUBLIC OF
THE CONGO: Kivu District, Kalehe,
Lemera (USNM 375361, 375365, 375367,
375368); Kivu District, Bukavu, Mugaba
(USNM 375348); Tshibati (RUCA 812,
1023, 1122, 1163, 1223, 1124, 1218,
1230, 1478, 1507, 2055, 2210, 2215,
2279, 2293, 2298, 2327, 2328); Kivu
District, Mount Mikeno, northwest slope,
7900 ft (AMNH 82512).
RWANDA: Gahinga Volcano (RUCA
4894?4896, 4935, 4936, 4992, 5004, 5055?
5057, 5082, 5084); Gasiza Volcano,
2360 m (5178, 5179, 5188, 5195, 5226);
Karisoke, 3100 m (RUCA 3659, 3662,
3682, 3706, 3720); Kayove (RUCA 5279);
Kinigi, 2250 m (RUCA 2407, 2410, 2417,
2419, 2488, 2653, 2667, 2669, 2687, 2740,
2763, 2764, 2825, 2836, 2837, 2855, 3010,
3881, 3882); Kitabi, 2200 m (RUCA 853,
863, 864, 887?889, 892?896, 903, 1006,
1015); Mutura, 2200 m (RUCA 670);
Pindura (RUCA 150); Routabansougera,
1750 m (RUCA 214?216, 220, 221, 225,
228, 266, 4145, 4152, 4153, 4173, 4230,
4282, 4331); Uwinka, 2450 m (RUCA
161, 163, 164, 168, 173, 178, 191, 199,
201, 206, 207, 2061, 2129, 2130, 2136,
2137, 2139, 2147, 2166, 2173, 2190, 2198,
2252, 2260, 2319; USNM 340837, 340839,
340841, 340842, 340848, 340850, 340859);
Visoke Volcano, 2700 m (RUCA 2849,
2850, 2949?2957, 3027?3034, 3049, 3050,
3742, 3903).
UGANDA: Virunga Volcanoes, pass
between Mt Sabinio and Mt Mgahinga,
8000?8500 ft (FMNH 26124, 26125,
26128, 26521?26525); Kigezi, Kabale Dis-
trict, Bwindi Impenetrable NP, Ruhizha,
2350 m (FMNH 157903?157907, 160502;
LACM 35608); Kabale District, Bwindi
Impenetrable NP, Mubwindi Swamp,
2070 m (LACM 53328, 53329); Kabale
District, Bwindi Impenetrable NP, Itama
area, 1615 m (LACM 35610); Kigezi,
Kisoro District, Mgahinga Gorilla NP,
saddle between Mts Muhavura and Mga-
hinga, 2980 m (FMNH 157511?157513);
Kigezi, Kisoro District, Mgahinga Goril-
la NP, edge of encroached area along
pipeline, 2680 m (FMNH 157515,
157516, 157518, 157521, 157524, 157527,
157528, 157530); Kigezi, Rubanda Dis-
trict, Echuya FR, 2380 m (FMNH
161049, 161050).
Discussion
Generic and specific taxonomy.?The
taxonomic status of Hylomyscus as genus
or subgenus has vascillated since its
description. Thomas (1926) expressly
created the taxon as a genus (H. aeta,
type species) to segregate certain small
African murines that had been associated
with Praomys, and considered the two
genera to be closely related, species of
Hylomyscus exhibiting more arboreal
adaptations. Allen (1939) and Hatt
(1940) acknowledged Thomas? generic
distinction, but Ellerman (1941) reduced
Hylomyscus, along with Mastomys and
Praomys, to subgenera of a broadly de-
314 PROCEEDINGS OF THE BIOLOGICAL SOCIETY OF WASHINGTON
fined Rattus. Misonne (1969) effectively
argued the in situ African origin of
Praomys and kin, including Hylomyscus,
and their distant phylogenetic relation-
ship to Rattus sensu stricto, although he
maintained them in separate generic
groupings within his Rattus Division.
Recent molecular studies of informative
taxonomic breadth decidedly sustain the
exclusion of Praomys, Hylomyscus, and
Mastomys from Rattus proper (Watts &
Baverstock 1995, LeCompte et al. 2002b,
Jansa & Weksler 2004). Aside from
Rattus, Hylomyscus has been considered
a subgenus of Praomys by some systema-
tists (Davis 1965, Misonne 1969, 1974,
Delany 1975). This taxonomic stature too
has been rejected by recent authorities
(Rosevear 1969, Robbins et al. 1980,
Musser & Carleton 1993), and phyloge-
netic evaluations of morphological or
molecular data portray Hylomyscus both
as monophyletic and as cladistically re-
moved from Praomys and related taxa
(Fadda et al. 2001, LeCompte et al.
2002a, 2002b). Within the radiation of
African murines, this clade of six gen-
era?Heimyscus, Hylomyscus, Mastomys,
Myomyscus, Praomys, Stenocephalemys?
has been identified as the Praomys group
(LeCompte et al. 2002b) or Stenocephal-
emys Division (Musser & Carleton 2005).
Consensual recognition of Hylomyscus
as genus has thus returned full circle. In
retrospect, this rank most effectively
captures the conservative nature and
diagnostic value of the traits originally
emphasized by Thomas (1926:178): fur
soft and fine; tail relatively long with fine
caudal scales and a weak terminal pencil;
hind feet small but relatively broad, digit
V relatively long; skull with a diminutive
rostrum, expanded brain-case, and nar-
row zygomatic plate; and t3 of the M1
weakly or not defined. Our comparisons
of craniodental morphometry and mor-
phology, in concert with the cladistic
results of LeCompte et al. (2002a,
2002b), serve only to support his insights
on generic boundaries among these en-
demic African murines. In essence, the
disagreement over endorobae as small
Praomys (Van der Straeten & Robbins
1997) or large Hylomyscus (Musser &
Carleton 1993, 2005; this study) repre-
sents just another context that under-
scores the basic shape differences that
characterize the crania of these two
genus-group taxa, again as first appreci-
ated by Thomas (1926). Principal compo-
nent ordination of covariation among
continuous variables has proven to be
a sensitive technique for discerning subtle
cranial size and shape differences between
closely related (congeneric) species popu-
lations of muroid rodents (e.g., Voss &
Marcus 1992). In our view, application of
principal component analysis solely to
holotypes in order to delimit muroid
generic boundaries (as per Van der
Straeten and Robbins 1997), taxonomic
entities historically founded on qualitative
traits, overextends the power and purpose
of the method.
At the time Thomas (1926) erected
Hylomyscus, he allocated six species as
generic members: H. aeta (type species),
H. alleni, H. carillus, H. denniae, H.
schoutedeni, and H. stella. Five of these
are still considered valid, and subsequent
descriptions and revisions have raised the
number of Hylomyscus species to 12,
a small but distinctive murine radiation
characteristic of moist lowland and mon-
tane forests in Subsaharan Africa (see
Musser & Carleton 2005 for taxonomic
updates; also Carleton & Stanley 2005,
this study). With increase in the number
of species recognized, certain themes or
patterns of morphological affinity among
them are becoming apparent. Six species
groups are provisionally arranged here
(Table 7), four of them (aeta, alleni, baeri,
and parvus) distributed within sectors of
lowland rainforest in West and Central
Africa, and two (anselli and denniae)
confined to wet montane forest in south-
central and eastern Africa. Within spe-
VOLUME 119, NUMBER 2 315
ciose muroid genera, species groups or
complexes have traditionally emerged,
following the early phase of discovery
and description, as phenetic foci or
tendencies of character variation and lack
a persuasive phylogenetic perspective to
sustain their monophyly (other than the
overall morphological similarity exhibited
by the constituent species). So it is with
the six groups identified here: many of
their character ??differences?? consist of
crude verbalizations of subtle size and
shape variation of continuous variables
(Table 7), and few of these differences can
be objectively systematized as cladistically
informative transformation series.
Other data sets must be mustered to
test the naturalness of these six groups
and to place their phylogenetic diversifi-
cation within the context of broad geo-
Table 7.?Provisional species groupings1 discernable within Hylomyscus and distribution of certain
character states (see Discussion).
Character
Species Groups
H. aeta H. alleni H. anselli H. baeri H. denniae H. parvus
Pectoral mammae2
(total)
absent (6) present (8) Absent (6) present? (8) present (8) absent? (6)
Upper incisors opisthodont orthodont, weakly
proodont
opisthodont opisthodont opisthodont proodont
T3 on M13 large medium medium large medium Minute
T9 on M1 distinct indistinct indistinct distinct indistinct Absent
Interorbital
condition4
shelf with
bead
amphoral weak shelf shelf weak shelf amphoral
Rostral length medium short medium medium long short
Incisive foramen5 medium short medium medium long short
Subsquamosal
foramen6
medium medium medium medium tiny, absent medium
1 Hylomyscus aeta group 5 H. aeta (Thomas, 1911), H. grandis Eisentraut, 1969. Hylomyscus alleni group
5 H. alleni (Waterhouse, 1838), H. carillus (Thomas, 1904), H. stella, (Thomas 1911). Hylomyscus anselli
group 5 H. anselli Bishop, 1979, H. arcimontensis Carleton & Stanley, 2005. Hylomyscus baeri group 5 H.
baeri Heim de Balsac & Aellen, 1965. Hylomyscus denniae group 5 H. denniae (Thomas, 1906), H. endorobae
(Heller, 1910), H. vulcanorum Lo?nnberg & Gyldenstolpe, 1925. Hylomyscus parvus group 5 H. parvus
Brosset, Dubois, & Heim de Balsac, 1965.
2 In general, counts are based on many females with enlarged mammae (near-term pregnancy or nursing)
surveyed per species. The count for H. baeri is based on a single young adult (USNM 450446); typical
condition uncertain. Only two suitable females of H. parvus were located (FMNH 167829, 167830), one of
which possessed a pair of small pectoral mammae in addition to prominently swollen post-axial and two
inguinal pairs; typical condition needs verification.
3 Size and distinctiveness relative to t1: large 5 area of t3 subequal to t1, physically discrete only in
juveniles-young adults, anterior chevron approximately symmetrical; medium 5 t3 smaller than t1,
physically discrete only in juveniles and indistinct in adults, anterior chevron moderately asymmetrical;
minute 5 t3 minute relative to t1, indistinct or seemingly absent even in juveniles, anterior chevron highly
asymmetrical.
4 Amphoral 5 no supraorbital shelf present, rear edges of orbit more or less smoothly contoured, shape
hourglass-like; weak shelf 5 rear edges of orbit squared, forming weak supraorbital shelf in older animals,
shape hourglass-like; shelf 5 rear edges of orbit squared, distinct supraorbital shelf evident even in young
animals, shape weakly cuneate; shelf with bead 5 supraorbital shelf present at rear of orbit, lateral shelf
edges dorsally reflected (beaded), shape moderately cuneate.
5 Posteriormost extent of incisive foramina: short 5 terminates short of or even with anterior root of the
M1s; medium 5 extends to middle of anterior root or even with t2 of M1s; long 5 extends to the level of the
t2-t1 of M1s.
6 Area and hamular process: medium 5 foramen present, one-third to one-half the area of the contiguous
postglenoid foramen, hamular process thin and long; tiny 5 foramen distinctly less than one-third the area
of the postglenoid foramen or occluded, hamular process thick and short or indistinct if foramen occluded.
316 PROCEEDINGS OF THE BIOLOGICAL SOCIETY OF WASHINGTON
graphic patterns. Several questions invite
immediate attention in this regard. For
example, although H. parvus is morpho-
logically distinctive within the genus
(Brosset et al. 1965), certain characteris-
tics of the alleni group approach the
conditions observed in H. parvus and
suggest that it should belong in the same
species group. Certainly, this is the
conclusion indicated by the cladistic
hierarchy (H. stella (H. alleni-H. parvus))
disclosed by the cytochrome-b analyses of
Lecompte et al. (2002b). However, their
representation within Hylomyscus is lim-
ited, geographically and taxonomically,
and merger of the two groups should be
argued more substantially. Hylomyscus
baeri is a species endemic to far West
Africa and seemingly is not closely related
to the broadly distributed West African
congenor, H. alleni. By virtue of its
relatively large t3 on M1 and distinctive
post-orbital shelf (albeit not beaded), is
H. baeri the West African vicar to H. aeta
of the Congo Basin? Are the H. anselli
and H. denniae species groups, which
occupy isolates of wet montane forest in
south-central and eastern Africa, respec-
tively, derived from a common ancestor,
as implied by their former inclusion under
the broadly ranging, polytypic construct
of H. denniae (e.g., sensu Bishop 1979)?
Or did each originate from different
species groups that inhabit lowland rain-
forest via independent dispersions into
mountainous settings? The distribution of
our few character states speaks inconclu-
sively to these questions (Table 7). In
view of the singular development of an
interorbital bead and relatively large t3 on
M1, does the H. aeta group form a basal
kinship group to all other Hylomyscus?
For the traits Thomas (1926) emphasized
in his diagnosis of Hylomyscus, in partic-
ular the weak expression of a t3, he
curiously selected an exceptional species
to serve as the standard of his new genus
(Table 7). The monophyly of the genus,
as its specific contents are currently
understood, remains a legitimate issue
that deserves stronger resolution based on
expanded species sampling within Hylo-
myscus and among related genera. Such
phylogenetic and biogeographic issues
will simultaneously profit from continued
alpha taxonomic revision and documen-
tation of distributions, which are espe-
cially required for three of the groups
(aeta, alleni, anselli). This much-needed
attention will undoubtedly further in-
crease the species diversity within Hylo-
myscus.
Biogeographic remarks.?The three
species of the Hylomyscus denniae group
are restricted to moist montane environ-
ments in eastern Africa, the majority of
localities falling within an elevational belt
covering 1800?3400 m. The species are
common in middle to upper elevation
montane forests, including bamboo
zones, but also occur higher into erica-
ceous vegetation and tropical alpine
habitats (Misonne 1963, Kerbis Peterhans
et al. 1998, this study). On the basis of
such an ecological occurrence and natu-
rally discontinuous mountain setting, the
H. denniae group qualifies as an element
of the Afromontane biotic region, here
used in the broad sense to connote both
wet montane forests and afroalpine habi-
tats above continuous tree canopy (i.e.,
following White 1978, 1981, although the
afroalpine is considered a separate region
by some?e.g., Hedberg 1986, Clausnitzer
2001). By its nature, the Afromontane
region of Subsahara Africa is highly
disjunct, consisting of several mountain
ranges and highlands that are diverse in
geological age and orogenic formation;
these nonetheless share a characteristic
and largely stenotopic high-altitude flora
and fauna. Ornithological biogeographers
have denoted seven physiographic groups
within the Afromontane region (Moreau
1966, Dowsett 1986), and species of the
H. denniae group occupy portions of two
of these, the East Congo (H. denniae, H.
vulcanorum) and Kenya (H. endorobae)
VOLUME 119, NUMBER 2 317
Table 8.?List of terrestrial small mammals (orders Soricomorpha and Rodentia) known from
afromontane habitats in eastern Africa. (E) 5 probable endemic; (?) 5 occurrence uncertain, not used
for statistical summaries.
Albertine rift Gregory rift
Ruwenzoris Kivu Mts Mt Elgon Aberdares-Mt Kenya
Soricomorpha
Crocidura elgonius X X
Crocidura fumosa X (E)
Crocidura kivuana X (E)
Crocidura lanosa X (E)
Crocidura maurisca X (E)
Crocidura montis X X X
Crocidura nigrofusca X X X
Crocidura niobe X X
Crocidura stenocephala X (E)
Crocidura tarella X (E)
Myosorex babaulti X (E)
Myosorex blarina X (E)
Paracrocidura maxima X X
Ruwenzorisorex suncoides X X
Scutisorex somereni X X
Surdisorex norae X (E)
Surdisorex polulus X (E)
Sylvisorex granti X X X X
Sylvisorex lunaris X X
Sylvisorex vulcanorum (?) X
Rodentia
Funisciurus carruthersi X X
Heliosciurus ruwenzorii X X
Tachyoryctes audax X (E)
T. rex X (E)
T. ruandae X (E)
T. ruddi X X
Cricetomys ansorgei X X
C. kivuensis X (E)
Delanymys brooksi X (E)
Dendromus insignis X X X X
D. kahuziensis X (E)
D. nyasae X X
Lophuromys aquilus X X X
L. medicaudatus X (E)
L. rahmi X (E)
L. woosnami X X
L. zena X (E)
Colomys goslingi X X X
Dasymys incomtus X X X X
D. montanus X (E)
D. rwandae X (E)
Grammomys dryas X X
G. gigas X (E)
G. dolichurus X X
G. ibeanus X
Hybomys lunaris X (E)
Hylomyscus cf. anselli X X
H. denniae X (E)
H. endorobae X (E)
318 PROCEEDINGS OF THE BIOLOGICAL SOCIETY OF WASHINGTON
montane groups. We judge the southern
geographic extent of H. denniae to be well
delimited, but additional collecting in the
mountains west of Lake Albert may
extend its northern limits. Similarly, the
ranges of H. vulcanorum and H. endor-
obae may reach slightly farther south than
presently documented within their princi-
pal montane groups (see remarks under
Taxonomic Summary). Notwithstanding
such local distributional refinements, we
doubt that members of the H. denniae
group will be found to occur in the
Ethiopian Plateau or Cameroon High-
lands, farther to the north and west,
respectively. Nor are members of the H.
denniae species group so far known from
Afromontane zones to the south and
southwest, in the Tanganyika-Nyasa
ranges or Angola Plateau where examples
of the H. anselli group are found (Carle-
ton & Stanley 2005). As discussed above,
whether the two species groups of Hylo-
myscus indigenous to montane settings
share a recent common ancestor relative
to those in the Guinean-Congo lowlands
merits investigation.
Within the East Congo and Kenya
Afromontane groups, the geographic
distribution of each Hylomyscus species
is more narrowly congruent with local
mountain systems: H. denniae in the
Ruwenzori Mts of the northern Albertine
(Western) Rift; H. vulcanorum in moun-
tains rimming Lake Kivu of the central
Albertine Rift; and H. endorobae in the
Mau Escarpment, Aberdare Mts, and Mt
Kenya along the Gregory (Eastern) Rift
Valley (Fig. 10). Based on shrew and
rodent distributions, the specific richness
and amount of endemism of the Ruwen-
zori-Kivu sectors of the Albertine Rift are
jointly impressive (Table 8), and the
historical role of the Ruwenzoris as
a lesser area of endemism is clearly
appreciable, as previously remarked by
Kerbis Peterhans et al. (1998). The total
number of species found in the Kenya
Afromontane group is notably less, but
the level of endemism is still high,
especially for shrew and rodent species
that inhabit the Aberdare Mts-Mt Kenya
subregion (Table 8). Few species occur
across all four mountain systems, and
some that do?such as Sylvisorex granti,
Praomys jacksoni, and Otomys tropical-
is?are believed to represent species com-
posites in need of careful taxonomic
Albertine rift Gregory rift
Ruwenzoris Kivu Mts Mt Elgon Aberdares-Mt Kenya
H. vulcanorum X (E)
Mus bufo X X
M. triton X X X X
Praomys degraaffi X (E)
P. jacksoni X X X X
Rhabdomys dilectus X X
Thamnomys kempi X (E)
T. major X (E)
T. venustus X X
Otomys barbouri X (E)
O. dartmouthi X (E)
O. denti X X
O. jacksoni X (E)
O. orestes X (E)
O. tropicalis X X X X
# species 27 41 18 23
# endemics 5 17 2 9
% endemicity 18.5 41.5 11.1 39.1
Table 8.?Continued.
VOLUME 119, NUMBER 2 319
revision (see Hutterer 2005, Musser &
Carleton 2005, for comments). Others,
such as Dendromus insignis and Mus
triton, lack a current revisionary standard
employing morphometric and-or molecu-
lar analyses. A pronounced faunal dis-
continuity is predictably recorded be-
tween mountains of the Albertine Rift
(Ruwenzoris-Kivu) and those of the
Gregory Rift (Mt Elgon-Aberdares-Mt
Kenya) (Fig. 11). Such a division among
small terrestrial mammals follows from
comparable biogeographic patterns ex-
hibited by other animals, notably birds
(Dowsett 1986, Stuart et al. 1993); from
geographic proximity and topographic
connectivity of the mountain ranges;
and from the probable former continuity
of montane forests among those ranges,
especially during cooler and moister
stadia of the Pleistocene (White 1981,
Lovett 1993, DeBusk 1998).
Confidence in the above area relation-
ships, foci of endemism, and inferred
vicariance zones suffers from several
glaring weaknesses in our systematic
framework: (1) inadequacy of biological
inventory to convincingly vouch specific
distributions and of long-term site study
to assess their ecological fidelity to
afromontane environments; (2) need for
improved taxonomic understanding of
many genera and species; and (3) the
general lack of credible interspecific
phylogenies to implicate historical geo-
graphic events in the genesis of kinship
patterns. All three facets require con-
certed attention to enhance comprehen-
sion of the biogeography of shrews and
rodents found in Afromontane land-
scapes.
Acknowledgments
For loans of specimens and-or access to
collections, we thank Guy G. Musser
(AMNH); Suzanne McLaren (CM); John
D. Phelps III (FMNH); Jim Dines
(LACM); Judy Chupasko and Mark
Omura (MCZ); and Walter N. Verheyen
(RUCA). Rebecca Banasiak (FMNH)
rendered the anatomical illustrations
(Figs. 3?5), David F. Schmidt (USNM)
undertook the skull photography (Figs. 8,
9), and Dan Cole (USNM) offered
assistance and advice on map prepara-
Fig. 11. Area relationships based on the geographic occurrences of small terrestrial mammals restricted
to Afromontane habitats in eastern Africa (see Table 8); phenogram of faunal similarity produced from
UPGMA clustering of Jaccard?s Index (coefficient of cophenetic correlation 5 0.98, P , 0.001).
320 PROCEEDINGS OF THE BIOLOGICAL SOCIETY OF WASHINGTON
Albertine Rift 
Kivu-Area Mts 
Ruwenzori Mts 
Gregory Rift 
i i 
Mt Elgon 
Aberdares- 
Mt Kenya 
0.1 0.2 0.3 0.4 0.5 0.6 0.7 
Faunal Similarity 
0.8 0.9 1.0 
tion. We appreciate their individual skills,
knowledge, and dedication in helping us
to communicate our results. JCKP is
especially indebted to the following for
their efforts in the field: Burundi?P. K.
Austin (Chicago), A. J. Fisher (Perth,
Australia), A. Nibizi (Burundi) and J.
Udelhoven (USA); Uganda?R. Mbabazi
and E. Tibenda (Ruhiza, Uganda), P. K.
Austin. For permission to work in Kibira
National Park, (Burundi), we gratefully
acknowledge the Institut National pour
l?Environment et la Conservacion de la
Nature (Mr. A. Nyokindi & Dr. L.
Ntahuga, Directeurs General, and Mr.
Chef du Parc). Logistical support in
Burundi was provided by the U.S. Peace
Corps/USAID Biological Diversity Pro-
ject (Mr. P. Trenchard, former director).
Mr. Leif Davenport was responsible for
all logisitical details and outstanding field
support. Financial support from the Ellen
Thorne Smith Fund, Marshall Field III
Fund and Barbara Brown Fund of The
Field Museum partially covered field
expenses in Burundi. Ugandan specimens
from Bwindi Impenetrable NP and Mga-
hinga Gorilla NP were collected under the
auspices of a John D. and Catherine T.
MacArthur Foundation award (JCKP,
PI) that sponsored the African Tropical
Biodiversity Program. S. Jennings (Di-
rector, Institute of Tropical Forest Con-
servation) and M. Grey provided critical
assistance and logistical support during
the BINP/Mgahinga surveys. In Uganda,
we thank R. Kityo (Curator, Makerere
Museum of Zoology) for far-reaching
support. The efforts of D. Pomeroy and
P. Kasoma (Makerere University) are
also acknowledged as is permission from
J. Okua (Chief Game Warden) and K.
Musana and Dr. E. Edroma, Uganda
Wildlife Authority. G. G. Musser and P.
J. Taylor graciously undertook review of
the manuscript and provided helpful
remarks. Any errors of commision or
omission are, of course, very much our
own.
Literature Cited
Aggundey, I. R., & D. A. Schlitter. 1984. Annotated
checklist of the mammals of Kenya. I.
Chiroptera.?Annals of Carnegie Museum
53:119?161.
Allen, G. M. 1939. A checklist of African mam-
mals.?Bulletin of the Museum of Compara-
tive Zoology, Harvard 83:1?763.
Bishop, I. R. 1979. Notes on Praomys (Hylomyscus)
in eastern Africa.?Mammalia 43:521?530.
Brosset, A., G. Dubost, & H. Heim de Balsac. 1965.
Mammife`res inedits recoltes au Gabon.?
Biologia Gabonica 1:148?174.
Carleton, M. D. 1980. Phylogenetic relationships in
neotomine-peromyscine rodents (Muroidea)
and a reappraisal of the dichotomy within
New World Cricetinae.?Miscellaneous Pub-
lications Museum of Zoology, University of
Michigan 157:1?146.
???, & C. Martinez. 1991. Morphometric
differentiation among West African popula-
tions of the rodent genus Dasymys (Muroi-
dea: Murinae), and its taxonomic implica-
tion.?Proceedings of The Biological Society
of Washington 104:419?435.
???, & C. B. Robbins. 1985. On the status and
affinities of Hybomys planifrons (Miller, 1900)
(Rodentia: Muridae).?Proceedings of The
Biological Society of Washington 98:956?
1003.
???, & W. T. Stanley. 2005. Review of the
Hylomyscus denniae complex (Rodentia:
Muridae) in Tanzania, with description of
a new species.?Proceedings of the Biological
Society of Washington 118:619?646.
???, & E. Van der Straeten. 1997. Morphological
differentiation among Subsaharan and North
African populations of the Lemniscomys
barbarus complex (Rodentia: Muridae).?
Proceedings of The Biological Society of
Washington 110:640?680.
Clausnitzer, V. 2001. Rodents of the afro-alpine
zone of Mt. Elgon. Pp. 427?443 in C. Denys,
L. Granjon, & A. Poulet, eds., African small
mammals. IRD E?ditions, Collection collo-
ques et se?minaires, Paris, 570 pp.
???, & R. Kityo. 2001. Altitudinal distribution
of rodents (Muridae and Gliridae) on Mt.
Elgon, Uganda.?Tropical Zoology 14:95?
118.
Davis, D. H. S. 1965. Classification problems of
African Muridae.?Zoologia Africana 1:121?
145.
???, & X. Misonne. 1964. Gazetteer of collecting
localities of African rodents.?Koninklijk
Museum voor Midden-Afrika, Tervuren.
Zoologische Documentatie 7:1?100.
VOLUME 119, NUMBER 2 321
DeBusk, Jr., G. H. 1998. A 37,500-year pollen
record from Lake Malawi and implications
for the biogeography of afromontane for-
ests.?Journal of Biogeography 25:479?500.
Delany, M. J. 1975. The Rodents of Uganda. British
Museum (Natural History), London, viii +
165 pp.
Demeter, A., & R. Hutterer. 1986. Small mammals
from Mt. Meru and its environs (Northern
Tanzania).?Cimbebasia 8:199?207.
Dowsett, R. J. 1986. Origins of the high-altitude
avifaunas of tropical Africa. Pp. 557?585 in
F. Vuilleumier & M. Monasterio, eds., High
altitude tropical biogeography. Oxford Uni-
versity Press, New York?Oxford, x + 649 pp.
Ellerman, J. R. 1941. The families and genera of liv-
ing rodents, vol. 2. Family Muridae. London:
British Museum (Natural History), xii + 690 pp.
Fadda, C., M. Corti, & E. Verheyen. 2001.
Molecular phylogeny of Myomys/Stenoce-
phalemys complex and its relationships with
related African genera.?Biochemical Sys-
tematics and Ecology 29:585?596.
Hatt, R. T. 1940. Lagomorpha and Rodentia other
than Sciuridae, Anomaluridae, and Idiuridae,
collected by the American Museum Congo
Expedition.?Bulletin of the American Mu-
seum of Natural History 76:457?604.
Hayek, L. C. 1994. Analysis of amphibian bio-
diversity data. Pp. 207?269 in W. R. Heyer,
M. Donnelly, R. W. McDiarmid, L. C.
Hayek, & M. S. Foster, eds., Measuring and
monitoring biological diversity?standard
methods for amphibians. Smithsonian Insti-
tution Press, Washington, D.C., xix + 364 pp.
Hedberg, O. 1986. Origins of the afroalpine flora.
Pp. 443?468 in F. Vuilleumier & M. Mon-
asterio, eds., High altitude tropical biogeog-
raphy. Oxford University Press, New York-
Oxford, x + 649 pp.
Heller, E. 1910. Descriptions of seven new species of
East African mammals.?Smithsonian Mis-
cellaneous Collections 56:1?5.
Hollister, N. 1919. East African mammals in the
United States National Museum, part II.
Rodentia, Lagomorpha, and Tubuliden-
tata.?Bulletin of the United States National
Museum 99:x + 184 pp.
Huhndorf, M. H. 2001. Molecular phylogenetics
and phylogeography of three species of small
mammals from the Albertine Rift, East
Africa. M.Sc. thesis, Illinois State University,
Bloomington, Illinois.
Hutterer, R. E. 2005. Order Soricomorpha.
Pp. 220?311 in D. E. Wilson, & D. M.
Reeder, eds., Mammal species of the World,
a taxonomic and geographic reference, 3rd
edition. Johns Hopkins University Press,
Baltimore, 1:xxxv + 1?743.
Hutterer, R., E. Van der Straeten, & W. N.
Verheyen. 1987. A checklist of the shrews of
Rwanda and biogeographic considerations on
African Soricidae.?Bonner Zoologische Bei-
tra?ge 38:155?172.
Jansa, S. A., & M. Weksler. 2004. Phylogeny of
muroid rodents: relationships within and
among major lineages as determined by IRBP
gene sequences.?Molecular Phylogenetics
and Evolution 31(1):256?276.
Kerbis Peterhans, J. C., R. M. Kityo, W. T. Stanley,
& P. K. Austin. 1998. Small mammals along
an elevational gradient in Rwenzori Moun-
tains National Park, Uganda. Pp. 149?171 in
H. Osmaston, J. Turkahirwa, C. Basalirwa, &
J. Nyakaana, eds., The Rwenzori Mountains
National Park, Uganda. Uganda: Makerere
University.
Landry, S. O. 1958. The function of the entepicon-
dylar foramen in mammals.?American Mid-
land Naturalist 60:100?112.
Lecompte, E?. L. Granjon, & C. Denys. 2002a. The
phylogeny of the Praomys complex (Roden-
tia: Muridae) and its phylogeographic im-
plications.?Journal of Zoological Systemat-
ics and Evolutionary Research 40:8?25.
???, ???, J. Kerbis Peterhans, & C. Denys.
2002b. Cytochrome b-based phylogeny of the
Praomys group (Rodentia, Murinae): a new
African radiation??Comptes Rendus Biolo-
gies 325:827?840.
Lo?nnberg, E., & N. Gyldenstolpe. 1925. Vertebrata.
2. Preliminary diagnoses of seven new mam-
mals.?Arkiv fo?r Zoologi, 17B 5:1?6.
Lovett, J. C. 1993. Climatic history and forest
distribution in eastern Africa. Pp. 23?32 in J.
C. Lovett & S. K. Wasser, eds., Biogeography
and ecology of the rain forests of eastern
Africa. Cambridge University Press, Cam-
bridge, ix + 341 pp.
Miller, G. S. 1912. Catalogue of the mammals of
Western Europe (Europe exclusive of Russia)
in the collection of the British Museum.
London: British Museum (Natural History),
xv + 1019 pp.
Misonne, X. 1963. Les Rongeurs du Ruwenzori et
des region voisines.?Exploration du Parc
National Albert (Duxieme Serie) 14:1?164.
???. 1969. African and Indo-Australian Mur-
idae: Evolutionary trends.?Annales Muse?e
Royal de l?Afrique Centrale, Tervuren, Belgi-
que, Serie IN-8, Sciences Zoologiques 172:
1?219.
???. 1974. Rodentia, main text. Part 6 in J.
Meester & H. W. Setzer, eds., The mammals
of Africa, an identification manual. Smithso-
nian Institution Press, Washington, D.C., 39
pp (not continuously paginated).
322 PROCEEDINGS OF THE BIOLOGICAL SOCIETY OF WASHINGTON
Moreau, R. E. 1966. The bird faunas of Africa and
its islands. Academic Press, New York, 424
pp.
Musser, G. G., & M. D. Carleton. 1993. Family
Muridae. Pp. 501?755 in D. E. Wilson & D.
M. Reeder, eds., Mammal species of the
World, a taxonomic and geographic refer-
ence, 2nd edition. Smithsonian Institution
Press, Washington D.C., 1206 pp.
???, & ???. 2005. Superfamily Muroidea.
Pp. 894?1531 in D. E. Wilson & D. M.
Reeder, eds., Mammal species of the World,
a taxonomic and geographic reference, 3rd
edition. Johns Hopkins University Press,
Baltimore, vol. 2:xviii + 745?2142.
Robbins, L. W., J. R. Choate, & R. L. Robbins.
1980. Nongeographic and interspecific varia-
tion in four species of Hylomyscus (Rodentia:
Muridae) in southern Cameroon.?Annals of
Carnegie Museum 49:31?48.
Roosevelt, T. 1910. African game trails. An account
of the African wanderings of an American
hunter-naturalist. [Appendix C. Loring?s
Notes, pp. 483?488.] [Appendix D. Biological
Survey of Mount Kenia, pp. 489?490.]
Charles Scribner?s Sons, New York, xv +
529 pp.
Rosevear, D. R. 1969. The Rodents of West Africa.
British Museum (Natural History), London,
xii + 604 pp.
Stuart, S. N., F. P. Jensen, S. Br?gger-Jensen, & R.
I. Miller. 1993. The zoogeography of the
montane forest avifauna of eastern Tanzania.
Pp. 203?228 in J. C. Lovett & S. K. Wasser,
eds., Biogeography and ecology of the rain
forests of eastern Africa Cambridge Univer-
sity Press, Cambridge, ix + 341 pp.
SYSTAT for Windows. 2002. Systat Software Inc.,
Version 10.2.01.
Thomas, O. 1906. Descriptions of new mammals
from Mount Ruwenzori.?Annals and Mag-
azine of Natural History, Series 7, 18:136?147.
???. 1926. The generic position of certain
African Muridae, hitherto referred to Aeth-
omys and Praomys.?Annals and Magazine
of Natural History, Series 9, 17:174?179.
Van der Straeten, E. 1980. Etude biometrique de
Lemniscomys linulus (Afrique Occidentale)
(Mammalia, Muridae).?Revue de Zoologie
Africaine 94:185?201.
???, & J. C. Kerbis Peterhans. 1999. Praomys
degraaffi, a new species of Muridae (Mam-
malia) from central Africa.?South African
Journal of Zoology 34(2):80?90.
???, & C. B. Robbins. 1997. Further studies on
Mastomys (Rodentia: Muridae) types and
generic distinctions among African Muri-
dae.?Mittheilungen Aus Dem Zoologischen
Museum zu Berlin 73(1):153?163.
???, & W. N. Verheyen. 1979. Notes taxonomi-
ques sur les Malacomys de l?Ouest Africain
avec redescription du patron chromosomique
de Malacomys edwardsi (Mammalia, Muri-
dae).?Revue de Zoologie africaine 93:10?35.
???, & ???. 1982. Differences biometriques
entre Hybomys univittatus (Peters) et Hybo-
mys trivirgatus (Temminck) de l?Afrique de
l?ouest.?Bonner Zoologische Beitrage 33:
205?213.
Verheyen, W., J. L. J. Hulselmans, T. Dierckx, & E.
Verheyen. 2002. The Lophuromys flavopunc-
tatus Thomas 1888 species complex: a cranio-
metric study, with the description and genetic
characterization of two new species (Roden-
tia-Muridae-Africa).?Bulletin de L?Institut
Royal des Sciences Naturelles de Belgique,
Biologie 72:141?182.
???, ???, ???, M. Colyn, H. Leirs, & E.
Verheyen. 2003. A craniometric and genetic
approach to the systematics of the genus
Dasymys Peters, 1875, selection of a neotype
and description of three new taxa (Rodentia,
Muridae, Africa).?Bulletin de L?Institut
Royal des Sciences Naturelles de Belgique,
Biologie 73:27?71.
Voss, R. S., & L. F. Marcus. 1992. Morphological
evolution in muroid rodents. II. Craniometric
factor divergence in seven Neotropical gen-
era, with experimental results from Zygodont-
omys.?Evolution 46:1918?1934.
???, ???, & P. Escalante, P. 1990. Morpho-
logical evolution in muroid rodents I. Con-
servative patterns of craniometric covariance
and their ontogenetic basis in the Neotropical
genus Zygodontomys.?Evolution 44:1568?
1587.
Watts, C. H. S., & P. R. Baverstock. 1995.
Evolution in some African Murinae (Roden-
tia) assessed by microcomplement fixation of
albumin.?Journal of African Zoology 109:
423?433.
White, F. 1978. The afromontane region.
Pp. 463?513 in M. J. A. Werger, ed., Bio-
geography and ecology of Southern Africa.
Junk, The Hague, xv + 1439 pp.
???. 1981. The history of the Afromontane
archipelago and the scientific need for its
conservation.?African Journal of Ecology
19:33?54.
Associate Editor: Gary R. Graves
Appendix 1
The 52 principal localities (in bold face) used in
preparation of distribution maps (Figs. 1, 10). Most
geographic coordinates were obtained from collec-
VOLUME 119, NUMBER 2 323
tors? skin tags and-or field catalogs (C), some from
USBGN gazetteers (G), faunal publications (P), or
topographic maps (M). Publications consulted in-
clude Aggundey & Schlitter (1984), Davis &
Misonne (1964), Delany (1975), Hutterer et al.
(1987), and Verheyen et al. (2002). See Specimens
examined for full locality data.
BURUNDI
1. Abris Kivuso, Kayanza Prov., Kibira NP,
2430 m: 03u009S, 29u32.59E C.
2. Gitenge River, Kayanza Prov., Kibira NP,
2177 m: 02u579S, 29u309E C.
3. Kirumura, 1.9 km N, Cibitoke Prov., Kibira
NP, Gatare Stream, 1950 m: 02u519S,
29u249E M.
4. Kirumura, 2.1 km N, Cibotoke Prov., Kibira
NP, Ruvyirame River, 1670 m: 02u519S,
29u239E M.
5. Kwogofe Hill, Cibitoke Prov., Kibira NP,
2170?2250 m: 02u509S, 29u259E C.
6. Mumushwizi Valley, Bururi Prov., Bururi FR,
1880 m: 03u559110S, 29u359010E M.
7. Nyabikona River, Bubanza Prov., Kibira NP,
1950?2140 m: 03u109S, 29u329E C.
8. NyagatarugwaRiverValley, Bubanza Prov., Bu-
ruri FR, 1795 m: 03u559360S, 29u359140E C.
9. Ruhinga Hill, Bururi Prov., Bururi FR, 2170?
2250 m: 03u559130S, 29u359180E M.
10. Ruhondo, Bubanza Prov., Kibira NP,
2040 m: 02u569S, 29u299E C.
11. Teza and vicinity, Muryamuya Prov., Kibira
NP, 2120 m: 03u139S, 29u349E C.
DEMOCRATIC REPUBLIC OF THE CONGO
12. Bugongo Ridge, 2745?3350 m: ca. 00u209N,
29u489E P.
13. Ibale, 2285 m: 00u209N, 29u469E M.
14. Kalonge Village, Butagu Valley, 2100 m:
00u209N, 29u489E C.
15. Katahuleko Creek, west of Kalonge: 00u209N,
29u489E C.
16. Lemera, 2100 m: 02u089S, 28u499E P.
17. Mt Mikeno, Kivu District, 2410 m: 01u299S,
29u239E P.
18. Mugaba, 2200 m: 02u189S, 28u369E P.
19. Tshibati: 02u149S, 28u489E P.
20. Ugongo (5 Old Kalonge), 2285 m: 00u209N,
29u499E M.
KENYA
21. Aberdare Mts, 2635?3200 m: 00u199S,
36u429E P.
22. Bellevue, 6 mi W, Nyeri Dist., 3048 m:
00u149S, 36u439E G.
23. Eldama Ravine, 2620 m: 00u039N, 35u439E P.
24. Kijabe, 2652 m: 00u569S, 36u349E P.
25. Mt Kenya, 2135?3325 m: 00u109S, 37u209E P.
26. Molo, 2438 m: 00u159S, 35u449E P.
27. Muguga, 2195 m: 01u119S, 36u399E P.
28. Naro Moru, 2740 m: 00u109S, 37u019E G.
RWANDA
29. Gahinga Volcano, up to 2700 m: 01u249S,
29u409E P.
30. Gasiza Volcano, 2360 m: 01u259S, 29u409E P.
31. Karisoke, 3100 m: 01u289S, 29u299E P.
32. Kayove, 2000 m: 01u489S, 29u219E P.
33. Kinigi, 2250 m: 01u269S, 29u369E P.
34. Kitabi, 2200 m: 02u349S, 29u269E P.
35. Mutura, 2200 m: 01u349S, 29u229E P.
36. Pindura: 02u289540S, 29u139560E G.
37. Routabansougera, 1750 m: 02u269S, 29u119E P.
38. Uwinka, 2450 m: 02u299S, 29u129E P.
39. Visoke Volcano, 2700 m: 01u279S, 29u309E P.
UGANDA
40. Bigo Hut, Kasese Dist., Ruwenzori Mts,
3420 m: 00u239N, 29u559E C.
41. Bujuku Hut, Lake Bujuku, Kasese Dist., Ru-
wenzori Mts, 3980 m: 00u239N, 29u539E C.
42. Ibanda, 3 km W, Kasese Dist., confluence of
Mubuku and Kyoha Rivers, Ruwenzori Mts,
1890 m: 00u219470N, 30u009400E C.
43. Confluence of Mubuku and Mahoma Rivers,
Kasese Dist., Ruwenzori Mts, 2105 m:
00u229N, 30u009E C.
44. Echuya FR, Rubanda Dist., 2380 m:
01u149400S, 29u489380W C.
45. John Mate Camp, Kasese Dist., Ruwenzori
Mts, 3370 m: 00u239N, 29u569E C.
46. Lake Mahoma, Kasese Dist., Ruwenzori Mts,
2960 m: 00u219N, 29u589E C.
47. Lower Kitandara Lake, Kasese Dist., Ruwen-
zori Mts, 4040 m: 00u229N, 29u539E C.
48. Mgahinga Gorilla NP, between Mts Mga-
hinga and Muhavura, Kisoro Dist., 2680 m:
01u229S, 29u399E C.
49. Nyabitaba Hut, Kasese Dist., Ruwenzori
Mts, 2670 m: 00u229N, 29u599E C.
50. Pass between Mts Sabinio and Mgahinga,
2440?2590 m: 01u229S, 29u379E M.
51. Ruhizha, Kabale Dist., Bwindi Impenetrable
NP, 2350 m: 01u029S, 29u459E C.
52. Saddle between Mts Mgahinga and Muhavura,
Kisoro Dist., Mgahinga Gorilla NP, 2980 m:
01u239S, 29u399E M.
324 PROCEEDINGS OF THE BIOLOGICAL SOCIETY OF WASHINGTON
Appendix 2
Specimens with post-cranial skeletons used to
confirm the occurrence of the entepicondylar
foramen on the humerus (See Table 4, Fig. 5).
Hylomyscus aeta (N 5 17).?Burundi, Kibira
NP (FMNH 149064); Ruvyirame River (FMNH
149066?149071, 149078, 149079); Gatare Stream
(FMNH 149082, 149098). Gabon, Mount Doudou
(FMNH 167682). Uganda, Bwindi-Impenetrable
NP, Nteko Parish (FMNH 160492?160496).
H. alleni (N 5 2).?Ivory Coast, Haute Dodo
Forest (USNM 583911, 583915).
H. anselli (N 5 14).?Tanzania, Rukwa Region,
Mbizi Mountains FR (FMNH 171342, 171344?
171356).
H. arcimontensis (N 5 18).?Tanzania, West
Usambaras, Ambangulu Tea Estate (FMNH
150126?150132, 150135?150138, 150154?150156,
150159?150162).
H. denniae (N 5 20).?Uganda, Ruwenzori
Mountains NP, Mubuku-Kyoha Rivers (FMNH
144448?144451, 144453?144457); Mubuku-Ma-
homa Rivers (FMNH 144459?144469).
H. endorobae (N 5 5).?Kenya, Aberdare NP
(USNM 589869); Mount Kenya (USNM 589870?
589873).
H. parvus (N 5 3).?Gabon, Maritime Province,
Ogoue (USNM uncataloged).
H. stella (N 5 20).?Democratic Republic of
Congo, Epula Zone, Ituri Forest (USNM 548741,
548742, 548745, 548751?548754, 548756, 548757,
548759, 548760, 548765, 548766, 548769, 548770,
548773, 548775, 548779). Kenya, Kakamega Forest
Reserve (USNM 589877, 589878).
H. vulcanorum (N 5 22).?Burundi, Bururi
Forest, Mumushwizi Valley (FMNH 155981,
155982, 155984?155988, 155990; Ruhinga Hill
(FMNH 155991, 155992). Uganda, Bwindi-Impen-
etrable NP, Ruhizha (FMNH 157903?157905,
160502); Mgahinga-Gorilla NP, saddle between
Mts Muhavura and Mgahinga (FMNH 157511,
157512); Mgahinga-Gorilla NP, slope of Mt Muha-
vura (FMNH 157515, 157516, 157518, 157528);
Echuya FR (FMNH 161049, 161050).
P. degraaffi (N 5 21).?Burundi, Kibira NP,
near Teza (FMNH 138037, 138039?138052, 138059,
138060, 138073, 138074, 138076, 138081).
Praomys delectorum (N 5 27).?Tanzania,
South Pare Mountains, Chome FR (FMNH
151311, 151313?151315, 151317?151319, 151321?
151323, 151325?151327, 151329); Udzungwa Mts
(FMNH 155624?155636).
P. jacksoni (N 5 20).?Burundi, Kibira NP,
Kwogofe Colline (FMNH 148795, 148796, 148799?
148802, 148806?148808, 148810, 148811, 148813,
148816, 148821). Uganda, Bwindi-Impenetrable NP,
Omubiyanja Swamp (FMNH 160870?160874);
Bwindi-Impenetrable NP, Buhoma (FMNH
160875).
P. misonnei (N 5 20).?Democratic Republic of
Congo, Ituri Forest, Epulu (FMNH 149570, 149574,
149576, 149594, 149595?149602). Uganda, Budongo
Forest (FMNH 165316?165319, 165327?165330).
P. rostratus (N 5 2).?Ivory Coast, Siahe
(USNM 583925); Zagne (USNM 583926).
P. tullbergi (N 5 13).?Liberia, Tars Town
(USNM 481923). Sierra Leone, Outamba-Kilimi
National Park (USNM 547393, 547397, 547399,
547400, 547409, 547413, 547417, 547418, 547426,
547434, 547435, 547437).
VOLUME 119, NUMBER 2 325