AMERICAN MUSEUM Norntates PUBLISHED BY THE AMERICAN MUSEUM OF NATURAL HISTORY CENTRAL PARK WEST AT 79TH STREET, NEW YORK, N.Y. 10024 Number 3157, 44 pp., 14 figures, 8 tables February 14, 1996

Systematic Revision of Sub-Saharan African Dormice (Rodentia: Myoxidae: Graphiurus) Part 1: An Introduction to the Generic Revision, and a Revision of Graphiurus Surdus1

MARY ELLEN HOLDEN2

ABSTRACT African dormice, genus Graphiurus (Rodentia: of Graphiurus surdus, represents the first distilla- Myoxidae), occur from the Cape of Good Hope tion of the large amount of data that has been to Senegal. The genus has never been adequately collected. The completed revision ofthe genus will revised, and thus the species diversity within the form one of the few comprehensive taxonomic genus, interspecific relationships, and biogeogra- revisions for any group ofAfrican in which phy have not been documented. This report con- the intent is to present a reliable estimate ofspecies stitutes the first of seven papers whose collective diversity and document patterns of geographical purpose is to provide a morphological revision of distributions. It will provide a framework for the the genus, explore the phylogenetic relationships future tasks oftesting hypotheses of species limits of species or species groups within the genus, and and interspecific relationships within Graphiurus document biogeographic patterns inferred from using molecular and perhaps other anatomical data, species distributions and relationships. Over 2700 and determining the significance of the docu- museum specimens ofAfrican dormice have been mented geographic distributional patterns relative examined and measured from institutions in the to distributions of other African and United States, , and for this study. environmental associations. This introductory paper, that includes a revision This first report contains an introduction to the 1 This report is dedicated to Guy Graham Musser, who ignited my interest in mammalogy, and provided support and inspiration for my research endeavors. 2 Instructor, Department of Biology, Orange County Community College, 115 South Street, Middletown, New York, 10940; Research Associate, Department of Mammalogy, American Museum of Natural History.

Copyright © American Museum of Natural History 1996 ISSN 0003-0082 / Price $5.60 2 AMERICAN MUSEUM NOVITATES NO. 3157 entire revision, including an overview of the in- ences between the two similarly sized species. Al- tended scope and organization, and a review of though the geographic range ofboth extends from historic taxonomic treatments of species of Gra- southern Cameroon to northeast and south-central phiurus. A species account of Graphiurus surdus, Zaire, the two have never been taken at the same and a comparison of that species with G. christyi collecting locality, and may have different micro- and G. lorraineus, follows. Graphiurus surdus was habitat requirements. previously known only from southern Cameroon; Comparisons are also made between G. surdus its range is shown to extend at least to northeast and G. lorraineus, as G. surdus has previously been and central Zaire. Graphiurus schwabi, previously synonymized with G. lorraineus as a subspecies of thought to be a synonym of G. lorraineus, is al- G. murinus. Graphiurus lorraineus is not allied located as a junior synonym of G. surdus. Gra- with the G. murinus complex, and is sympatric phiurus surdus has alternately been recognized as with representatives of that assemblage through- a valid species, included in the problematic species out its range, including G. christyi. Sympatry of complex G. murinus, or obscured along with G. G. lorraineus with G. surdus in southern Came- lorraineus under G. murinus lorraineus. roon and central Zaire is documented. Compari- Comparisons with G. christyi, a member of the sons between the two species show that they are G. murinus complex from the same geographic easily separated by significant size and shape dif- area, show that there are significant shape differ- ferences.

RESUME Les loirs africains du genre Graphiurus (Roden- travail et une recapitulation historique de la po- tia: Myoxidae) ont une distribution geographique sition taxonomique des especes de Graphiurus. Suit allant du Cap au Senegal. Ce genre n'a jamais ete une redefinition specifique de G. surdus, et une revise de facon adequate. Par consequent, ni la comparaison de cette espece avec G. christyi et G. diversite specifique intragenerique, ni les relations lorraineus. interspecifiques non plus que la biogeographie ne Jusqu'alors, G. surdus etait connu uniquement sont bien documentees. Ce rapport est la premier au Sud du Cameroun mais nous avons demontre d'une serie de sept publications decrivant la mor- que son aire de distribution s'etend au moinsjusq- phologie du genre Graphiurus, les relations phy- u'aux centrales et du Nord-Est du Zaire. logenetiques entre especes ou groupes d'especes, Graphiurus surdus a et6 successivement retenu ainsi que les modeles biogeographiques qui de- comme espece valide, inclu dans le complexe spe- coulent des distributions specifiques et des rela- cifique douteux de G. murinus, ou regroupe comme tions phylogenetiques. sous-espece avec G. lorraineus. Graphiurus schwa- Pour realiser cette etude, plus que 2700 speci- bi doit etre mis en synonymie avec G. surdus. mens conserves dans des institutions scientifiques Des comparaisons avec G. christyi, un membre americaines, europeennes, et africaines ont ete ex- du complexe specifique de G. murinus, habitant amines et mesures. le meme geographique et de meme taille, Cette introduction, qui comprend une revision ont montre qu'il y a des difflerences morpholo- de Graphiurus surdus, doit etre consideree comme giques importantes. Bien que la distribution geo- le premier "distillat" de la masse de donnees dont graphique de ces deux especes couvre une region nous disposons. La revision complete du genre allant du Sud du Cameroun au Nord-Est du Zaire, formera une des rares etudes taxonomiques ex- elles n'ont jamais ete recoltees dans la meme lo- haustives d'un groupe de rongeurs africains dans calite mais pourraient exiger des microhabitats dif- le but d'en presenter une evaluation fidele de sa ferents. diversite specifique et de formuler des schemas de Des comparaisons de G. surdus et de G. lorrai- distribution geographique. Elle fournira le cadre neus, s'imposaient aussi parce que le premier a ete approprie pour tester les hypotheses futures con- mis en synonymie et regroupe avec le second sous cernant les limites specifiques et les relations in- le nom de G. murinus lorraineus. Toutefois, G. terspecifiques dans le genre Graphiurus a l'aide de lorraineus ne peut etre inclu dans le complexe G. donnees moleculaires et, peut-etre d'autres don- murinus. Il est sympatrique, sur toute son aire de nees anatomiques, ainsi que pour evaluer la valeur distribution, des representants de ce groupe d'es- des modeles biogeographiques par rapport 'a ceux peces, y compris G. christyi. La sympatrie de G. d'autres Mammiferes africains et les associations lorraineus et de G. surdus au Sud du Cameroun et avec l'environnement. au Zaire central a ete demontree. Ces deux especes Ce premier texte comprend outre une introduc- peuvent etre distinguees aisement par des diffTer- tion 'a la revision globale, un resume du plan de ences de taille et de forme. 1 996 HOLDEN: SUB-SAHARAN AFRICAN DORMICE. PART 1 3

INTRODUCTION Recent quality systematic revisions of in- sons for following the conceptual approach adequately studied groups often result in the taken in this revision are detailed below un- recognition of a greater diversity of species der Materials and Methods. in those groups than was previously acknowl- Workers who adopt the same theoretical edged. For some smaller and relatively well- species concept may obtain different esti- known groups, such as mammals, species di- mates of species diversity for a given group, versity estimates might produce higher num- because there are often several subjective es- bers and be made more accurate by focusing timates of diversity that can be justified by more attention on recent systematic revisions differing operational practices consistent with in addition to the relatively few descriptions the same theoretical concept. Therefore, the of new taxa that are published each year. operational concept followed in a given work These estimates may then provide finer doc- should be clearly stated. In addition, it is es- umentation of biogeographic patterns, and sential that adequate data are presented for aid in assessing conservation priorities. reasonable geographic samples (populations) Objections might be made that estimates of each alleged species, if such samples are of species diversity depend on the particular available. Researchers who adopt different species concepts used by the authors of sys- operational or theoretical species concepts tematic revisions. The same arguments apply may then make use ofdata contained in such to descriptions of new species; one worker revisions and possibly build upon them, even might reason that a "new species" is simply if they reject the overall estimate of species a diagnosable population that should not be diversity, or species diagnoses. recognized as a separate species, whereas oth- New species ofsmall African mammals are ers may feel that the population should be described each year (for example, look at spe- recognized as a species precisely because it is cies accounts of African shrews in Hutterer, diagnosable. Species concepts obviously af- 1993, and African muroid rodents in Musser fect the numbers of species recognized by a and Carleton, 1993), and new species oflarge given author, and if the concept followed is African mammals are even occasionally dis- at either end of the theoretical spectrum (see covered (Cercopithecus solatus from Gabon, Endler, 1989; Frost and Hillis, 1990; O'Hara, for example; see Groves, 1993) or rediscov- 1994; and Templeton, 1989, for reviews), the ered (Phacochoerus africanus, the other spe- differences in numbers of species recognized cies ofwarthog; see Grubb, 1993). However, can be dramatic (see discussions in Cracraft, many groups of small and some larger Afri- 1989; O'Hara, 1994; and Patton and Smith, can mammals require taxonomic revisions 1994). A perusal of mammalian systematic that address questions regarding diversity of literature reveals that most authors explicitly genera and species, as well as biogeographic or implicitly use the reproductive isolation patterns. species concept sensu Mayr (1942, 1969), In the absence of sound revisions describ- providing either genetic proof or evidence ing species limits, ecological studies are often (sympatry or morphological divergence) of compromised, if not rendered useless, if the reproductive isolation. The theoretical (but followed is incomplete or incor- not necessarily operational) species concept rect. For example, the very few ecological followed by most modern mammalian sys- studies that include African dormice are of- tematists is therefore somewhat pervasive. ten only informative at the generic level. The The relative strengths and weaknesses of the assignment of the specimens to problematic isolation and other process-based species species or species complexes without com- concepts versus phylogenetic concepts have ment obscures the true identities of the spe- been extensively discussed (e.g., Baum, 1992; cies reported upon. Voucher specimens are de Queiroz and Donoghue, 1988, 1990a, frequently not collected, and detailed de- 1990b; Nixon and Wheeler, 1990; O'Hara, scriptions of the caught are often 1994). Although I do not support universal lacking. Well-documented revisions ofprob- usage ofa particular species concept, my rea- lematic groups would help alleviate such 4 AMERICAN MUSEUM NOVITATES NO. 3157 weakened efforts by providing accessible ref- I will review the number of species detected erences upon which ecologists may rely. and their diagnostic morphological features These systematic revisions would ideally in the context of characterizing the genus be based on a combination of morphological Graphiurus, attempt to reconstruct possible and molecular data; however, for most small phylogenetic relationships among the species African taxa, potential morpholog- or species groups, explore biogeographic pat- ical samples (in the form of museum speci- terns found within the genus, and summarize mens) abound, whereas molecular samples any concordance between geographic distri- are few, if they exist at all. For this reason butions ofdormice species with ranges ofoth- morphological systematic studies are often er African groups. Finally, the sig- essential before any sense can be made ofthe nificance of the evolutionary interspecific al- often incidental molecular data. Morpholog- liances, and the geographic and ecological ical studies also give critical information, such distribution ofeach species, in understanding as locations of potential hybrid or contact present biogeographic patterns in Africa will zones, that could help make molecular sam- be explored. pling over this vast more efficient. One group of small African mammals that has not yet received rigorous review is the MATERIALS AND METHODS genus Graphiurus (Myoxidae: Rodentia), a INSTITUTIONS: Specimens I examined for group of dormice occurring throughout most this report (Part 1) ofthe Graphiurus revision of the habitats found in Africa south of the are housed in collections of the following in- . Estimates of the number of species stitutions (identified throughout the report by in Graphiurus have been made in several the appropriate acronym): checklists (e.g., Allen, 1939; Ellerman, 1940; Holden, 1993), and an identification manual AMNH American Museum ofNatural His- (Misonne, 1974), and were the foci of both a tory, New York generic revision (Genest-Villard, 1978) and BM(NH) The Natural History Museum, for- regional taxonomic revisions (e.g., merly the British Museum (Natural Robbins History), London and Schlitter, 1981; Schlitter et al., 1985; An- CM Carnegie Museum of Natural His- sell, 1974; Ansell and Dowsett, 1988). How- tory, Pittsburgh ever, no comprehensive systematic revision FMNH Field Museum of Natural History, of the genus is available that documents the Chicago number of species in the group, their un- KM Kaffrarian Museum, King Wil- ambiguous morphological and distributional liam's Town limits, their phylogenetic relationships, and MCZ Museum of Comparative Zoology, their contribution to understanding African Harvard University, Cambridge biogeography. MHNG Museum d'Histoire Naturelle de I intend to provide a new estimate of the Geneve, Geneve MNHN Museum National d'Histoire Na- number ofspecies in Graphiurus that will de- turelle, Paris rive from a comprehensive, substantially MRAC Musee Royal de l'Afrique Centrale, documented systematic revision ofthe genus, Tervuren based on detailed examination of more than RUCA Universite d'Anvers, Rijksuniver- 2700 specimens. A revision of G. surdus is sitair Centrum Antwerpen, Ant- included in this first of an intended series of werpen reports. Each subsequent paper will include SMF Forschungsinstitut und Naturmu- accounts of one or more additional species. seum Senckenberg, Frankfurt am Each species account will contain history of Main taxonomic treatment, synonyms (if any), TM Transvaal Museum, Pretoria morphological description, geographic dis- USNM National Museum of Natural His- tory, Smithsonian Institution, tribution, morphometric comparisons with Washington, D.C. other taxa, and discussion of significant geo- ZMB Zoologisches Museum und Institut graphic variation. Any karyological, molec- fur Spezielle Zoologie, Museum fdr ular, or ecological information (though often Naturkunde der Humboldt-Univ- minimal) will be included. In the final report ersitat zu Berlin, Berlin 1 996 HOLDEN: SUB-SAHARAN AFRICAN DORMICE. PART 1 5

CHOICE OF MEASUREMENTS: Genest-Villard measurements used in statistical analyses. (1978) explained why she found certain mea- Most values for the latter three were taken surements of the skull and dentition to be from labels attached to study skins; I mea- important in dormice systematics. The 24 sured lengths of hind foot (including claws) cranial and dental measurements I employed and ear on a few dry skins and fluid-preserved for this study were chosen because: 1) based specimens. Total length was also recorded on on a preliminary survey I thought them to most labels, and I obtained length of head be potentially informative ofspecies-level dif- and body by subtracting tail length from the ferences (they varied among species); 2) they total length. Weights (WT), in grams, were sample a variety of cranial and dental fea- sometimes recorded by collectors and these tures; 3) I could compare them directly with were included in some summary statistics. data presented in the only previous taxonom- External measurements reflect not only the ic revision (Genest-Villard, 1978), and re- variation due to age, sex, and individual ex- gional studies (e.g., Robbins and Schlitter, pression inherent in any sample, but to mea- 1981; Schlitter et al., 1985), even if I had to suring techniques and competence of the va- take redundant measurements, such as crown riety ofcollectors who are represented in these length of M1-M3 (preferred by me), and samples. In particular, two techniques of crown length of P4-M3 (published in the lit- measurement of hind foot length (inclusive erature), and breadth ofM I (preferred by me) and exclusive of claws) were represented in versus breadth ofM2 (found in the published samples of all species. I used the hind foot literature); 4) some regions of preserved cra- length and the other external measurement nia are commonly damaged, and I sometimes values in my analyses as a general guide to took alternative measurements that estimat- differences and similarities in size and pro- ed a similar dimension, because it was im- portions but excluded them from multivar- possible to predict which measurement would iate analyses. Despite differences in mea- yield a larger sample. (For example, I pre- surement techniques and accuracy, the size ferred condyloincisive length for estimating and proportional differences between the spe- cranial length, but many examples of some cies summarized in the ratio diagrams and species have missing or broken incisors and summary statistics mirror those observed damaged incisor alveoli; greatest length of upon first-hand examination of specimens. skull was also taken so that I would poten- Genest-Villard (1978), who has provided the tially have larger samples with an estimate only published revision of Graphiurus, used of skull length of those species to include in some of these same external measurements morphometric analyses.) In selecting cranio- and presents her interpretation of their dental measurements with the above consid- strengths and weaknesses in diagnosing spe- erations in mind, I was sensitive to the dis- cies. cussion by Voss (1988: 361), in which ideal Cranial and dental measurements were measurement protocols were weighed against taken with dial calipers graduated to tenths practical limitations. of millimeters. The limits of each measure- I did not restrict my suite ofmeasurements ment are illustrated in figure 1. These mea- to include only those that were obtainable surements are abbreviated in text, tables, and from most crania, which could be included figures as follows: in multivariate analyses, for descriptive sta- tistics could still be calculated for those po- GLS greatest length of skull tentially informative measurements. For the CIL condyloincisive length multivariate analyses I simply extracted the ZB zygomatic breadth sample of specimens that had complete mea- IB interorbital breadth surement sets for an informative subset of BBC breadth of braincase the 24 dimensions measured. HBC height of braincase BR breadth of rostrum MEASUREMENTS: Values for external, cra- LHR least height of rostrum nial, and dental measurements are given in LN length of nasals millimeters. Length ofhead and body (LHB), LD length of diastema length oftail (LT), length ofhind foot (LHF), PL palatal length and length of ear (LE) were the only external LBP length of bony palate 6 AMERICAN MUSEUM NOVITATES NO. 3157

IB F -> I I- z -J

-A cn 0

II IiII -

I I I I, q 'I BIP~~~~~~~~~

ZB BBC

cv) i- -

-I

Fig. 1. Diagrammatic views of cranium and toothrows of an adult Graphiurus christyi illustrating limits of cranial and dental measurements. 1 996 HOLDEN: SUB-SAHARAN AFRICAN DORMICE. PART 1 7

LIF length of incisive foramina clusal patterns present in African dormice BIF breadth of incisive foramina may reveal whether the patterns exhibit in- BBP breadth across bony palate at Ml teresting interspecific variation for phyloge- CLP4-M3 crown length of maxillary toothrow netic studies, and may aid in diagnoses of CLM1-M3 crown length ofmaxillary molar row extant and extinct species. BP4 breadth of fourth upper permanent AGE CRITERIA AND SAMPLES: Specimens premolar to one offour classes based BM1 breadth of first upper molar were assigned age BM2 breadth of second upper molar primarily on stage oftooth eruption and wear, BIT breadth of upper incisor tips but degree of fusion of cranial sutures as well PPL postpalatal length as pelage characteristics were also evaluated. LB length of auditory bulla The range from young adults through old BBO breadth of basioccipital adults provided the data for morphometric analyses and computation of summary sta- MANDIBLE AND DENTITION: The Graphiu- tistics and ratio diagrams. Combining this rus dentary was illustrated by Wahlert et al. broad range of adult stages was necessary to (1993: 15); that conformation typifies all the obtain adequate samples for analysis; series species. I neither measured the mandibles of ofAfrican dormice are, with a few exceptions, Graphiurus nor used their traits to help dis- characteristically small and from scattered tinguish among taxa. I surveyed the dentaries localities. in all my samples and, except for differences The age classes recognized are: in size, could not detect qualitative traits that Juvenile: Deciduous upper fourth premolar increased my ability to distinguish among taxa (DP4) present, or occlusal surface offourth over what was present in the crania. premolar (P4) not yet even with chewing My study of dentition is limited to mor- surface of molar row. phometric comparisons among taxa. I did not Young Adult: Occlusal surface of P4 level survey occlusal patterns to determine wheth- with top of molar row, but exhibits little er the variation seen enabled discrimination or no wear, molars unworn or only slightly among species, or provided characters useful so. in resolving phylogenetic relationships. The Adult: All teeth moderately worn (lingual and premolars and molars of all species of Gra- labial facets worn, transverse ridges worn phiurus are simple and basined, their occlusal but not obliterated). surfaces vaguely sculptured by low ridges and Old Adult: Premolars and molars very worn shallow valleys in patterns illustrated by and sometimes sharply beveled or eroded Wahlert et al. (1 993: 1 8) for G. lorraineus and on their labial surfaces, many or all trans- in figure 5 for G. surdus. These indistinct pat- verse ridges obliterated or distorted, occlu- terns are typical of those found in most spe- sal surfaces often worn flat to the roots. cies of Graphiurus, with the notable excep- tions of G. crassicaudatus and G. hueti. They SEXUAL DIMORPHISM AND SAMPLES: Fe- exhibit high variability and I found them ex- males and males ofthe species of Graphiurus tremely difficult to examine critically without included in this first report are approximately electron microscopy. Determining the sig- equal in body size and very similar in mor- nificance of variation in occlusal patterns phological characters not associated with sex- within and between species would be an ar- ual traits. No adequate sample (at least 10 duous study by itself. Results of such an in- females and 10 males, young adults to old quiry could be used to test hypotheses for- adults with complete crania, from the same mulated in my revision, which is based on locality) exists from which to derive uni- other characters. variate statistics of each sex in the samples The occlusal patterns of most species of of either G. surdus or G. christyi. There is Graphiurus have never been documented, and such a series available for G. lorraineus, but if adequate specimens are available, a scan- it is from Ndola, Zambia, which is outside ning electron micrograph of the occlusal sur- the geographic region that is encompassed by face ofone example ofeach species diagnosed this report on G. surdus. Nevertheless, this will be included. Such a coarse survey of oc- sample illustrates the nature of intergender 8 AMERICAN MUSEUM NOVITATES NO. 3157

TABLE 1 Summary Statistics for Cranial and Dental Measurements (mm) of Female and Male Graphiurus lor- raineus from Ndola, Zambia (Mean, ± 1 SD, observed range [in parentheses], and number of specimens in sample are listed for each measurement. An ontogenetic range from young to old adults is represented in the total sample of each gender. No statistically significant differences [at the a = 0.05 level] exist between sample means of males and females for any of the measurements listed below.) Femalesa Malesb

CIL 21.1 ± 0.44 (20.2-21.8) 14 21.2 ± 0.33 (20.5-21.9) 14 ZB 13.8 ± 0.41 (13.0-14.4) 14 14.1 ± 0.30 (13.5-14.5) 14 IB 3.9 ± 0.15 (3.6-4.2) 14 4.0 ± 0.13 (3.8-4.2) 14 BBC 11.8 ± 0.34 (10.9-12.2) 14 11.9 ± 0.27 (11.4-12.3) 14 HBC 7.1 ± 0.25 (6.6-7.5) 14 7.3 ± 0.21 (7.0-7.6) 14 BIT 1.4 ± 0.12 (1.3-1.6) 14 1.5 ± 0.12 (1.2-1.7) 14 LHR 4.3 ± 0.18 (4.0-4.5) 14 4.4 ± 0.12 (4.3-4.7) 13 LN 8.1 ± 0.24(7.8-8.6)14 8.2 ± 0.19 (8.0-8.5) 13 LD 4.7 ± 0.16 (4.5-5.0) 14 4.7 ± 0.12 (4.5-4.9) 14 PL 7.6 ± 0.23 (7.2-7.9) 13 7.7 ± 0.13 (7.4-7.9)14 PPL 10.3 ± 0.35 (9.6-10.9) 13 10.3 ± 0.26 (9.8-10.6) 14 LIF 2.8 ± 0.21 (2.4-3.0) 14 2.7 ± 0.20 (2.4-3.2) 14 BIF 1.7 ± 0.08 (1.5-1.8) 14 1.7 ± 0.11 (1.5-1.8) 14 BBP 2.7 + 0.18 (2.4-3.1) 14 2.7 ± 0.16 (2.3-2.9) 14 CLP4-M3 3.1 ± 0.12(2.9-3.3)14 3.1 ± 0.10 (3.0-3.3) 14 CLM1-3 2.5 ± 0.11 (2.3-2.7) 14 2.5 ± 0.08 (2.4-2.7) 14 BM1 1.1 ± 0.05 (1.0-1.1) 14 1.1 ± 0.05 (1.0-1.1) 14 BM2 1.2 ± 0.05 (1.1-1.2) 14 1.1 ± 0.07 (1.0-1.2) 14 BP4 0.9 ± 0.06 (0.8-1.0) 14 0.9 ± 0.07 (0.8-1.0) 14 LB 7.3 ± 0.18 (7.0-7.6)14 7.3 ± 0.16 (6.9-7.4) 14 LBP 3.9 ± 0.18 (3.6-4.1) 13 4.0 ± 0.16 (3.8-4.3) 14 BBO 1.9 ± 0.17 (1.5-2.3) 14 2.1 + 0.11 (1.9-2.3) 14 BR 4.9 ± 0.30 (4.4-5.5) 14 4.9 ± 0.26 (4.5-5.4) 14 a BM(NH) 20.11.3.106-20.11.3.108, 20.11.3.110-20.11.3.114; KM 3765; TM 2570, 2576, 2587, 2589, 2593. bBM(NH) 20.11.3.96, 20.11.3.101-20.11.3.105; KM 3766; TM 2571, 2573, 2577, 2580, 2582, 2583, 2585. sexual variation also encountered within general size (growth) and growth-invariant samples of G. surdus and G. christyi, and it divergence (Voss et al., 1990; Voss and Mar- is the one for which meaningful summary cus, 1992). All analyses were based on mea- univariate statistics could be calculated sep- surements transformed to their natural log- arately for females and males. Comparisons arithms. The factor coefficients ofgeneral size between the sexes of G. lorraineus are pre- (which includes allometry) were estimated as sented in table 1. There are no statistically the elements of the first eigenvector of the significant differences (at the a = 0.05 level) pooled-within covariance matrix. Specimen between sample means of females and males scores on general size are age-correlated with- for any of the cranial and dental measure- in species, specimens with unworn teeth gen- ments listed. All other univariate and mul- erally having smaller size scores than con- tivariate statistical analyses were performed specifics with well-worn teeth. The normal- on samples that included whatever combi- ized coefficients of size-invariant species di- nation ofgenders were present in a collection vergence were calculated from mean from a particular locality (or pooled locali- differences estimated in variable-by-variable ties): either all females, all males, or a mixture analyses of covariance (ANCOVAs) with of both. general size as the covariate; the requisite as- MULTIVARLATE ANALYSES: Morphometric sumption of equal slopes was evaluated by comparisons among species samples of Gra- eye. In view of the motley composition of phiurus were analyzed by factors representing species samples (which contain, in some cases, 1 996 HOLDEN: SUB-SAHARAN AFRICAN DORMICE. PART 1 9 conspecifics from widely scattered popula- inferred to have a genetic basis, be fixed with- tions), and the general lack ofinformation to in the species (present in all individuals), and evaluate sampling assumptions, tests for sta- absent from closely related species (Baum, tistical significance of species differences are 1992; Cracraft, 1983, 1987, 1989; Davis and best avoided. Instead, I emphasize the rela- Manos, 1991). Groups of African dormice tive magnitude ofsize-adjusted difference co- that are obviously "species" (in some cases efficients and the congruence of multivariate sympatric) can sometimes only be diagnosed results with visually obvious contrasts in ar- by a combination ofcharacters, because some guing the biological significance of morpho- characters are either not present or hard to metric comparisons. distinguish in some individuals (particularly SPECIES CONCEPT: The theoretical concept juveniles), and in those cases other distin- followed in this report is the cohesion species guishing characters (that also may not be ex- concept (Templeton, 1989). Beurton (1995) hibited in 100 percent ofthe specimens) need recently proposed that gene circulation, which to be considered. Even if the requirement of he usefully differentiated from gene flow, is 100 percent diagnosability were met, some the fundamental cohesion mechanism. As he of the distinguishing characters are not al- pointed out, the concept of "gene flow" used ways absent from "close relatives." The in- by Mayr (1942, 1963, and other references terspecific relationships of African dormice to Mayr listed by Beurton, 1995: 195), Tem- have yet to be explored, so that all (except pleton (1989), and others implicitly includes for a few obviously morphologically diver- gene circulation. However, Beurton does not gent species) hypothesized species of Gra- demonstrate that gene flow, in the strict sense phiurus are at present considered close rela- defined by him, is not an important cohesion tives. mechanism in some cases. Rather than view- The alternative phylogenetic species con- ing Templeton's broad cohesion species con- cept emphasizes the criterion of monophyly, cept and the relatively limited "Heraclitean" with the intention of circumventing prob- species concept ofBeurton, that espouses gene lems with the diagnosability standard (Baum, circulation as the fundamental cohesion 1992; de Quieroz and Donoghue, 1988, 1990; mechanism, as alternative concepts, it seems Donoghue, 1985; Mishler and Brandon, 1987; better to add gene circulation (or separate it Ridley, 1989; Rosen, 1979). Applying the cri- from the broader category of gene flow) to terion of monophyly to species, however, is Templeton's classification ofcohesion mech- problematic (Wheeler and Nixon, 1990; Nix- anisms (Templeton, 1989: 13). on and Wheeler, 1990), and summarized by A species concept ". .. that is consistent Baum (1992): with recovered phylogenetic history" (Frost Thus, species monophyly is a difficult criterion and Hillis, 1990) is intellectually appealing, to apply using currently available cladistic and it is recognized that ".... the potential methodologies, and even ifmonophyly could be for gene exchange is only loosely coupled to accurately assessed this would still leave the historical relatedness-the central consider- problem of establishing exclusivity of descent. ation of systematics" (Baum, 1992). How- It remains to be seen whether protocols can be ever, monophyly of species may be the ex- developed for studying species monophyly and ception rather than the rule (Patton and exclusivity ... and whether these will be suffi- Smith, 1994), and in many cases the appli- ciently practical to gain acceptance in the sys- cation ofphylogenetic concepts is fraught with tematic community. at least as many difficulties as process-based As no practical protocol for studying spe- species concepts (Baum, 1992; O'Hara, 1994; cies monophyly and exclusivity has been gen- Patton and Smith, 1994). erated, for this study it seems reasonable to One phylogenetic species concept requires focus on subjective estimates of genetic or that a group oforganisms possess at least one ecological cohesion, based on shared distinc- "diagnostic character," either primitive or tive, diagnosable morphologies (of some- derived (Cracraft, 1983, 1987, 1989; Wheeler times unknown polarity, and sometimes only and Nixon, 1990; Nixon and Wheeler, 1990), diagnosable by a combination of characters) and furthermore that the character must be over geography, and ecological data. It is pos- AMERICAN MUSEUM NOVITATES NO. 3157 sible that species diversity may be underes- ciously provided technical advice and exten- timated if morphological divergence has not sive discussions. accompanied speciation, but data to discern Peter Goldberg is responsible for figure 3 such situations are unavailable at present. and most photographic prints of crania (figs. Additionally, molecular data might provide 6-10); I am most grateful for his superior evidence for recognizing a greater (or lesser) efforts. The holotype of G. surdus was pho- diversity of species of Graphiurus than hy- tographed by me. I thank Peling Melville for pothesized by this revision, but at present no her expertise in producing the SEM for figure molecular samples exist for most species of 5. Figures 1, 2, and 11-14 were rendered by African dormice, and no species has been ad- Patricia Wynne; I appreciate the aesthetic equately sampled across its geographic range. quality that her excellent drawings lend to the manuscript. ACKNOWLEDGMENTS This study serves as partial fulfillment of the Doctor of Philosophy requirements for I thank the curators, associates, students, the University of California, Berkeley. I am and technicians of the institutions listed in indebted to my major professor, James L. Materials and Methods for facilitating my Patton, for his support and guidance through- visits, and for providing extensive loan ma- out my graduate education, and I also thank terial, without which the results ofthis study the other members of my dissertation com- would have been greatly compromised: Paula mittee, Harry Greene and Dale McCullough. Jenkins [BM(NH)]; Susan MacLaren, Duane Schlitter (CM); Bruce Patterson, Julian Ker- HISTORICAL TAXONOMIC bis Peterhans, Bill Stanley (FMNH); Lloyd TREATMENTS OF TAXA Wingate (KM); Terri McFadden, Maria Rutzmoser (MCZ); Louis de Roguin Members of Graphiurus occur throughout (MHNG); Jacques Cuisin, Michel Tranier sub-Saharan Africa, from the Cape of Good (MNHN); Michel Louette, Wim Van Neer Hope north to Senegal in the northwest and (MRAC); Luc de Bruyn, Herwig Leirs, Walter Somalia in the northeast. Significant varia- Verheyen (RUCA); Dieter Kock, Gerhard tion in morphology and ecology would be Storch (SMF); Deborah Bellars, Gary Bron- expected in small-bodied rodents found over ner (TM); Michael Carleton, Linda Gordon, such a vast continental region. Such variation Helen Kafka, Dave Schmidt (USNM); Ren- is typical of Graphiurus as is evident by the ate Angermann (ZMB). four generic and 76 specific and subspecific I could not have completed this study with- scientific names proposed by taxonomists to out the support of the curators and chair of represent the impressive morphological di- the Department of Mammalogy, American versity detected within the genus. Museum ofNatural History, and I would es- Taxonomic information about subsaharan pecially like to warmly thank my husband dormice has been summarized in eight pri- and colleague, Guy G. Musser, as well as Karl mary reports published between the late 1800s Koopman and Ross MacPhee. I am expressly and the present. Highlights of each are listed grateful to Rob Voss for his help with the below. These references cover the spectrum morphometric analyses. Eric Brothers, Pat of species and are not regionally focused. Brunauer, Wolfgang Fuchs, Helmut Som- Other taxonomic reports and checklists of mers, and Muriel Williams assisted with many species present taxonomic information for aspects ofthe project, and their help is greatly particular geographic regions: appreciated. (Shortridge, 1934; Roberts, 1951; Ellerman I genuinely thank Les Marcus for his pa- et al., 1953; de Graaff, 1981; Meester et al., tience and help with morphometric analyses, 1986; Skinner and Smithers, 1990); Bots- as well as technical rescues from a variety of wana (Smithers, 1971); Zimbabwe (Smithers computer hardware and software problems. and Wilson, 1979); Zambia (Ansell, 1974, Norman Newell generously allowed me to 1978); Malawi (Ansell and Dowsett, 1988); use his computer to perform multivariate Mozambique (Smithers and Tello, 1976); analyses. Jane Mason and John Wahlert gra- Tanzania (Swynnerton and Hayman, 1951); 1 996 HOLDEN: SUB-SAHARAN AFRICAN DORMICE. PART 1 I1I

Angola (Hill and Carter, 1941); Zaire (Hatt, nus Graphiurus. It has long been obvious to 1940); Kenya (Hollister, 1919); Uganda (De- us that the Cape (Graphiurus ocu- lany, 1975); Ethiopia (Yalden et al., 1976), laris) as not properly congeneric with the oth- and (Rosevear, 1969; Robbins er African species and we may take this op- and Schlitter, 1981; Schlitter et al., 1985). A portunity to reclassify the group." Thomas survey of the treatments of dormice taxa and Hinton proposed the genus Gliriscus for found in those publications is not included rock dormice, with G. platyops as the type here. That information will be evaluated in species, and provided key characteristics by appropriate accounts of species in forthcom- which it, Graphiurus (containing only the ing sections of my revision of Graphiurus. Cape dormouse), and Claviglis (all the forest I include this historical perspective on spe- dormice) could be distinguished from one an- cies diversity within Graphiurus here for two other. reasons. It summarizes various taxonomic Finally, Allen (1936) proposed Aethoglis, treatments and philosophies of researchers with "genotype Graphiurus nagtglasii Jen- who have worked with the group and pro- tink" for hueti, an older name for the same vided differing hypotheses of species num- species represented by nagtglasii. The large bers and definitions, and it will be useful for body size of hueti, along with a variety of my subsequent reports. The different treat- pelage, cranial, and dental traits, were used ments also provide the past taxonomic con- by Allen to diagnose the new genus. text in which surdus, the taxon I discuss in In his "Checklist of African Mammals," this first paper, has been viewed by other Allen (1939) recognized all four genera. Lists workers. and revisions published after Allen's com- Four generic names, all published between pilation, however, admitted only the genus 1832 and 1936, have been proposed for Graphiurus, but taxonomic arrangements be- grouping the morphological variation - low this level were characterized by disagree- served in sub-Saharan African dormice. The ment. Some easily distinguished species, such validity of the species clusters designated by as G. crassicaudatus, G. hueti, and G. ocu- these generic titles will be addressed in the laris, retained their integrity in most check- final section ofmy systematic revision. Until lists and revisions, and rarely were the sub- then I will use Graphiurus for all the species jects of taxonomic confusion (crassicaudatus of sub-Saharan dormice treated in my revi- as treated by Misonne, 1974, is an exception). sion. Graphiurus is the oldest valid name for Others, such as G. parvus, G. surdus, and G. the genus and was proposed by Smuts (1832) lorraineus have been alternately lumped into for capensis, which is a synonym ofthe South one widely distributed, morphologically vari- African ocularis, one of the most distinctive able species, usually identified as G. murinus, species in the genus. The three other generic or separated into geographically restricted names used for sub-Saharan species are in- species without substantive documentation troduced below. of their species limits. These disparate treat- More than 50 years after Graphiurus was ments of the species are summarized below. proposed, Jentink (1888) described a new 1. Reuvens (1890a): "Die Myoxidae Oder species, the small-bodied crassicaudatus from Schlaefer," Reuvens's privately published Liberia. He thought its tail to be so unusual "Inaugural-Dissertation" that was the first that he placed the species in a new genus, comprehensive treatise on the osteology and Claviglis ("characterized by a club-shaped, systematics of dormice. Extracts of the Dis- not distichous tail," Jentink, 1888: 42). sertation were published the same year in Jentink's genus was recognized by Thomas Notes from the Leyden Museum (Reuvens, and Hinton in 1925 (p. 232) when they re- 1890b). The sub-Saharan species recognized ported on material collected in "South-West by Reuvens were placed in three subgenera Africa" (Namibia). They described a new of the genus Myoxus. species and subspecies ofdormouse and not- ed that ". . . in determining these specimens FAMILY MYOXIDAE we have made an examination of all the Af- Genus Myoxus rican Dormice commonly referred to the ge- Subgenus AMERICAN MUSEUM NOVITATES NO. 3157

"E. nagtglasii" Saharan species he had formerly listed under "E. kelleni" Eliomys and made minor adjustments in syn- "E. crassicaudatus" onomies. "E. murinus" "E. orobinus" Subgenus Graphiurus FAMILY MYOXIDAE "G. capensis" Genus Graphiurus "G. hueti" G. hueti (synonym: natglasi) Subgenus Myoxus G. crassicaudatus M. elegans G. kelleni G. angolensis 2. Trouessart (1897): "Catalogus mam- G. smithi malium tam viventium quam fossilium" and G. johnstoni an appendix of"addenda et corrigenda." This, G. parvus the first real catalog or checklist that included G. nanus sub-Saharan dormice taxa, was a prominent G. murinus reference in its time and the list to which G. platyops subsequent authors-Allen (1939), for ex- G. ocularis (synonym: capensis) ample-compared their contributions. No- G. orobinus tice that during Trouessart's time, many sub- 4. Allen (1939): "A Checklist of African Saharan species were allocated to the genus Mammals," an influential reference of its Eliomys, a group we now realize contains only time. Explaining the reason behind the North African, Middle Eastern, and Euro- checklist's creation, Allen noted "that the pean forms, and one that is phylogenetically mammalian fauna of Africa is fairly well as- not closely related to sub-Saharan Graphiu- certained, and that the likelihood of many rus (Wahlert et al., 1993). more really new genera or species being found FAMILY MYOXIDAE is small. Nevertheless much revisionary work is yet to be done as adequate collections ac- Genus Eliomys (synonym: Claviglis) cumulate in the larger museums. As a first E. natglasii (sic) necessity in indicating what has been done E. kelleni E. smithii and where further study is urgent, a complete E. parvus checklist of the names applied to African E. nanus mammals is here presented, intended to in- E. crassicaudatus clude all those currently regarded as valid as E. murinus (synonyms: coupei, lalandianus, er- well as all synonyms through July 1938, so ythrobronchus, cineraceus, microtis) far as possible" (p. 3). Allen recognized all E. orobinus four genera, one of which he had described, Genus Graphiurus and listed 37 species. G. capensis (synonyms: ocularis, cattoiri, typ- icus, elegans) FAMILY MUSCARDINIDAE G. hueti Genus Aethoglis In an appendix of "addenda et corrigenda" A. hueti (valid subspecies: hueti, argenteus, Trouessart (1899) made these changes and monardi, nagtglasii additions: Genus Claviglis C. alticola Genus Graphiurus C. ansorgei G. hueti (synonym: natglasii [sic]) C. cuanzensis G. johnstoni C. brockmani (valid subspecies: brockmani, in- G. angolensis ternus) C. butleri 3. Trouessart (1904): "Catalogus mam- C. christyi malium tam viventium quam fossilium," the C. coupeii "quinquennale supplementum." In this last C. crassicaudatus (valid subspecies: crassicau- and most complete revision of his catalog, datus, dorotheae) Trouessart transferred to Graphiurus the sub- C. foxi 1 996 HOLDEN: SUB-SAHARAN AFRICAN DORMICE. PART 1

C. griselda groups ofrelated forms, but apart from perhaps C. haedulus four easily distinguished forms the remainder C. schwabi do not appear separable into definite groups. C. kelleni Variation in size is considerable in some forms, C. littoralis so that overlapping invalidates any arrangement C. lorraineus based on size, while although extremes ofcolour C. murinus (synonyms: lalandianus, erythro- in the subgenus are wide (from pale grey to bronchus, cineraceus, cinerascens, microtis; brown), division into groups based on colour valid subspecies: griseus [synonym: john- breaks down when it is seen that in a very large stoni], isolatus, pretoriae, raptor, saturatus, series from northern nearly all the col- selindensis, tzaneenensis) our shades found in forms from elsewhere in C. nanus the range of the genus are represented." C. olga C. orobinus Ellerman could easily distinguish four spe- C. parvulus cies and noted that "certainly monardi and C. parvus (valid subspecies: parvus, dollmani) crassicaudatus and, I think, probably woos- C. personatus nami and surdus are sufficiently distinct to C. smithii be regarded as types of specific groups. The C. soleatus (valid subspecies: soleatus, collaris) remainder will have to be referred to a single C. spurrelli group, in which it appears that there are far C. streeteri too many outstanding 'distinct species' at the C. surdus present day" (p. 607). C. tasmani C. vandami C. vulcanicus FAMILY MUSCARDINIDAE C. woosnami Genus Graphiurus C. zuluensis Subgenus Graphiurus Genus Gliriscus G. ocularis (synonyms: capensis, typicus, ele- G. angolensis (valid subspecies: angolensis, jor- gans, cattoiri) dani) Subgenus Gliriscus G. platyops G. platyops G. rupicola (valid subspecies: rupicola, monto- G. eastwoodae sus) G. rupicola (valid subspecies: rupicola, mon- Genus Graphiurus tosus) G. eastwoodae Subgenus Claviglis (synonym: Aethoglis) G. ocularis (synonyms: typicus, capensis, ele- hueti group gans) G. hueti (valid subspecies: hueti, argenteus, nagtglasi [sic]) 5. Ellerman (1940): "The Families and crassicaudatus group Genera of Living Rodents. Volume I. Ro- G. crassicaudatus (valid subspecies: cras- dents Other than Muridae." Following Al- sicaudatus, dorotheae) len's list by only a year, Ellerman's work rec- surdus group ognized only one African dormouse genus, G. surdus Graphiurus, but did group the 40 species list- woosnami group ed into three subgenera. He noted (p. 607) G. woosnami that Graphiurus, monardi group G. monardi

... which is in much need of revision, is ex- murinus group tremely difficult to arrange in any natural order. G. olga Apart from the subgenera, Graphiurus, for ocu- G. orobinus laris, and Gliriscus for platyops group, there re- G. butleri mains a large assemblage of more normal Af- G. brockmani (valid subspecies: brock- rican Dormice. Both Mr. Hayman and myself mani, internus) have tried to arrange these into groups, but with- G. foxi out much success. Mr. Hayman reports: "Sub- G. parvus (valid subspecies: parvus, doll- genus Claviglis, this contains the remainder and mani) by far the largest number of African dormice. G. soleatus (valid subspecies: soleatus, col- Attempts have been made to divide them into laris) AMERICAN MUSEUM NOVITATES NO. 3157

G. personatus toriae, raptor, saturatus, schneideri, selin- G. murinus (synonyms: coupei, erythro- densis, smithi [sic], soleatus, spurelli [sic], bronchus, cineraceus, lalandianus; val- streeteri, sudanensis, surdus, tasmani, id subspecies: murinus, tzaneenensis, tzaneenensis, vandami, vulcanicus, woos- isolatus, griseus [synonym: johnstoni], nami, zuluensis; "perhaps not readily saturatus, raptor) separable from murinus:" crassicauda- G. microtis tus, dorothae [sic] G. smithi (sic) (synonym: subrufus) G. ansorgei 7. Genest-Villard (1978): "Revision sys- G. lorraineus G. spurrelli tematique du genre Graphiurus (Rongeurs, G. haedulus Gliridae)." Genest-Villard did not recognize G. christyi subgenera, retained only six species, but not- G. angolensis (valid subspecies: angolensis, ed that certain subspecies might turn out to jordani) be valid species. Although preceded by G. griselda (valid subspecies: griselda, pre- checklists and faunal accounts dealing with toriae) taxonomy of sub-Saharan dormice, her con- G. kelleni tribution was the only attempt to understand G. nanus the array ofmorphological and ecological di- G. johnstoni versity represented by samples and present Representatives of nine forms were not seen hypotheses of species limits within the con- by Ellerman and "not allocated to group:" text of a systematic revision of the entire ge- nus. Her results would eventually be tested G. alticola and some conclusions not accepted (for ex- G. littoralis ample, Ansell, 1978, 1989; Robbins and G. streeteri Schlitter, 1981; Schlitter et al., 1985), and G. tasmani Holden (1993: 763) commented that "the re- G. vandami vision by Genest-Villard (1978), based most- G. zuluensis ly on size grades, underestimated species di- G. schwabi G. parvulus versity, particularly in the murinus group." G. vulcanicus FAMILY GLIRIDAE 6. Misonne (1974): "Order Rodentia" in Genus Graphiurus "The Mammals of Africa, an Identification G. parvus ("Formes mises en synonymie:" Manual." One genus, two subgenera, and five brockmani [valid as a subspecies], doll- species were recognized by Misonne, and not mani, foxi, internus, nanus, olga, persona- all synonyms were listed. tus, tasmani; "Formes douteuses:" kelleni, Claviglis ansorgei cuanzensis) G. murinus ("Liste par ordre alphabetique des FAMILY GLIRIDAE formes decrites referables a G. murinus:" Genus Graphiurus alticola, angolensis, ansorgei, butleri, chris- Subgenus Graphiurus tyi, cineraceus, collaris, coupei, erythro- G. ocularis bronchus, griselda [valid as a subspecies], subgenus Claviglis griseus, haedulus, isolatus, johnstoni, jor- G. monardi dani, lalandianus, littoralis, lorraineus [val- G. hueti ("includes" argenteus, nagtglasii) id as a subspecies], marrensis, microtis [val- G. platyops ("includes" angolensis, australis, id as a subspecies], murinus [valid as a sub- eastwoodae, jordani, kaokensis [sic], species], orobinus [valid as a subspecies], montosus, parvulus, rupicola) parvulus, pretoriae, raptor [valid as a sub- G. murinus ("includes" alticola, ansorgei, species], saturatus, schwabi, selindensis, brockmani, butleri, christyi, collaris, smithii, soleatus, spurelli [sic] [valid as a cuanzensis, coupei, dasilvai, dollmani, subspecies], surdus, tzaneenensis, vandami, etoshae, foxi, griselda, griseus, haedulus, vulcanicus, woosnami, zuluensis) internus, isolatus, johnstoni, kelleni, lit- G. crassicaudatus ("Forme mise en synonymie:" toralis, lorraineus, marrensis, microtis, dorotheae) nanus, olga, orobinus, personatus, pre- G. platyops ("Formes mises en synonymie:" 1 996 HOLDEN: SUB-SAHARAN AFRICAN DORMICE. PART 1 15

australis, eastwoodae, montosus, rupicola SPECIES LIMITS OF [valid as a subspecies], schoutedeni) GRAPHIURUS SURDUS G. hueti ("Formes mises en synonymie:" argen- teus [valid as a subspecies], monardi [valid as a subspecies], nagtglasi [sic]) "A small-eared species allied to murinus," G. ocularis ("Formes mises en synonymie d'apres was Dollman's (1912: 314) introduction to Ellerman:" capensis, typicus, elegans, cat- the new species he named and described as toiri) Graphiurus surdus. He drew attention to the 8. Holden (1993): "Family Myoxidae" a ears, "... remarkably small, about 5 mm chapter in "Mammal Species of the World, shorter than those of the South-African spe- a taxonomic and geographic reference." I did cies," and was impressed by the cranium, so not recognize subgenera and listed 14 species, ". . . markedly different from that ofany oth- commenting that a "systematic revision of er known form, the occipital part ofthe brain- Graphiurus is in progress and the species rec- case broad and almost rectangular in shape, ognized below reflect information in the lit- giving the skull a very unusual appearance." erature as well as examination of museum Dollman was convinced that "The small ears, specimens; results are provisional and as- large hind feet, and unusual shape of the oc- signment of synonyms incomplete" (p. 763). cipital region immediately distinguish this The list reflected my unpublished research dormouse from all the other allied forms." based on museum specimens as well as the Graphiurus surdus maintained its identity conclusions in certain reports defining limits in relevant checklists for at least six decades. of species in particular African regions (for Allen (1939), for example, in his checklist of example, Ansell and Dowsett, 1988; Robbins African mammals, listed surdus as a species, and Schlitter, 1981; Schlitter et al., 1985). and Ellerman (1940), in his monograph on families and genera of living rodents, char- FAMILY MYOXIDAE acterized G. surdus as being easily distin- from other of Genus Graphiurus guishable species Graphiurus. G. christyi Morphology of G. surdus seemed so distinc- G. crassicaudatus (synonym: dorotheae) tive to Ellerman that he made the species a G. hueti (synonyms: argenteus, nagtglasii) member ofa separate group (see the historical G. kelleni (synonyms: ansorgei, johnstoni, nan- outline above). us) The species that seemed so distinctive to G. lorraineus (synonyms: collaris, haedulus, Dollman and Ellerman lost its species status spurrelli) in the 1970s. In his chapter on rodents for G. microtis an African mammal identification manual, G. monardi (synonym: schoutedeni) Misonne (1974) relegated surdus, without G. murinus (synonyms: alticola, butleri, ciner- aceus, cinerascens, dasilvail, erythrobron- comment, to G. murinus. A similar action chus, etoschae, griselda, isolatus, lalandi- was taken by Genest-Villard (1978) in the anus, littoralis, marrensis, pretoriae, raptor, only taxonomic revision of Graphiurus prior saturatus, schneideri, selindensis, soleatus, to my present report; she referred surdus to streeteri, sudanensis, tzaneenensis, vanda- G. murinus lorraineus. mi, vulcanicus, woosnami, zuluensis; "the A few years later, Robbins and Schlitter synonyms ... almost certainly contain (1981) reported on their collection of Cam- names which are actually synonyms of G. eroon dormice and reinstated surdus as a spe- microtis and other species," p. 764) cies representing a distinctive element of the G. ocularis (synonyms: capensis, elegans, typi- Cameroon fauna. They were able to provide cus) G. olga key morphological characters that would dis- G. parvus (synonyms: brockmani, dollmani,foxi, tinguish specimens ofG. surdus from samples internus, personatus) ofthe other four kinds of dormice they iden- G. platyops (synonyms: angolensis, eastwoodae, tified from Cameroon: G. hueti, G. crassi- jordani, parvulus) caudatus, G. lorraineus, and G. christyi. Based G. rupicola (synonyms: australis, kaokoensis, on my study of those same Cameroon spec- montosus) imens, as well the type series ofsurdus, I too G. surdus (synonym: schwabi) could not accept the arrangements defended AMERICAN MUSEUM NOVITATES NO. 3157 by either Misonne (1974) or Genest-Villard "deaf, silent, mute, faint" (Brown, 1956). (1978), but could agree with Ellerman (1940) Presumably Dollman (1912) chose this name and Robbins and Schlitter (1981), and rec- to dramatize the pinnae of the specimen he ognized surdus as a species in the myoxid had selected as the holotype. He thought its chapter ofthe taxonomic and geographic ref- ears were very small (table 2), and were unique erence to mammal species of the world to and diagnostic of the species. (Holden, 1993). DISTRIBUTION: The actual geographic range Redescribing G. surdus is the subject ofthis of G. surdus is unknown. Specimens have first section of my systematic revision of the been collected from eight localities in south- genus. Previous workers had access to only ern Cameroon, one in Equatorial (Rio a few examples of the species collected in Muni), one in northeastern Gabon, and two southern Cameroon and Rio Muni. I have in Zaire, one in the northeast and the other been able to locate 22 specimens in institu- in the south-central part of the country (see tional collections. Some of these, the type fig. 2 and gazetteer below). All sites are below series and those reported on by Robbins and 1000 m in the central African portion of the Schlitter (1981), for example, had already "Guineo-Congolian regional centre of en- been identified as G. surdus. Others were hid- demism," the floristic region described by den in collections under misidentifications or White (1983), in the memoir accompanying simply identified as "Graphiurus sp.," and a the Unesco/AETFAT/UNSO vegetation map few have been recently collected and gener- ofAfrica. Within that regional center are sub- ously loaned to me for study by their holding divisions described by White and delimited institutions. on the map. Although 22 specimens constitute a small Localities 1-7 in Cameroon, locality 10 in sample, it is large enough to provide mor- Gabon, and locality 12 in Zaire fall in "Wet- phological data by which I am able to affirm ter Guineo-Congolian rain forest," locality Dollman's view that G. surdus is one of the 11 in Zaire is in "Drier Guineo-Congolian more morphologically distinctive species of rain forest," and localities 8 in Cameroon and Graphiurus. The evidence is presented below 9 in Rio Muni are congruent with a mapping in the context of description, diagnosis, and unit reflecting a forest mosaic of"wetter" and comparisons with morphologically similar "drier" Guineo-Congolian rain forest. Lo- species. Because I have studied more speci- cality 11 is the only locality from the region mens of G. surdus, obtained from localities that is not at least partially characterized as covering a larger geographic region than were being wetter Guineo-Congolian rain forest on available to previous researchers, I am able the vegetation map. Each general delimited to sketch a more revealing depiction of its region on the map is described in terms of geographic distribution. the dominant vegetation types found therein. Locality 11 is "Inkongo, Sankuru river" (In- kongo is situated on the left bank ofthe San- Graphiurus surdus Dollman, 1912 kuru river), and it is plausible that the spec- Graphiurus surdus Dollman, 1912: 314. imen was taken from a wetter forest than is Graphiurus schwabi G. M. Allen, 1912: 441. the dominant forest type of that region. The Masako specimen was taken from the Ma- HOLOTYPE AND TYPE LocALITY: BM(NH) sako Forest Reserve (locality 12), which is 0.2.5.55, an adult female collected by G. L. situated in a loop of the Tshopo river (Hut- Bates along the Benito River, "French Con- terer and Dudu, 1990); this locality is dis- go" (actually Rio Muni in Equatorial Guinea) cussed further under Natural History below. in July, 1899. The specimen consists of a It is possible that G. surdus will be found to stuffed skin (originally preserved in fluid), be associated with wetter Guineo-Congolian cranium, and mandible. The skin and skull rain forest. (fig. 4) are intact and undamaged. Measure- The general geographic distribution of G. ments of the holotype are listed in table 2. surdus may coincide with what Carleton and ETYMOLOGY: The Latin surdus means Robbins (1985: 987) called the "evergreen 1996 HOLDEN: SUB-SAHARAN AFRICAN DORMICE. PART 1 17

TABLE 2 rainforests ofCentral Africa," or the "Central Comparisons of Measurements (mm) Among Ho- African Forest" (p. 984), which is ".... ap- lotypes Representing Three Species of Graphiurus proximately equivalent to the Congo Forest from Block of Booth [1958] or the Lower Guinea G. surdus Forest of Moreau [1966, 1969]." Although G. lorrai- from widely scattered localities, the examples schwabi surdus neus G. christyi of G. surdus have come only from the Zaire MCZ BMNH BMNH BMNH 8607a 0.2.5.55b 7.7.8.115c 14.2.1.1.8d basin and the southern Cameroon-northern Gabon region-the "Central African For- Age juv adult adult adult est"-and not from the West African portion Sex ? female female male of the Guineo-Congolian rain forest that ex- LHB 74 87e 95 105 LT 64 67e 62 95 tends west of the Cross and Niger rivers. LHF 18.5 20f 16 18 Only two specimens ofG. surdus have been LE 7 ge 12 12 taken in Zaire; one at Inkongo in 1912, and GLS 26.2 27.5 the other at Masako in 1986, despite exten- CIL 22.5 25.9 sive collecting expeditions to various parts of ZB 13.1 14.1 14.3 15.2 that country. It is conceivable that future small IB 4.3 4.5 5.0 4.8 mammal surveys may extend the distribu- BBC 12.5 12.8 12.3 12.3 tional limits of G. surdus further to the north HBC 8.3 7.7 and west. BR 5.0 5.6 5.3 5.9 GAZETTEER AND SPECIMENS: 22 specimens LHR 4.4 4.8 5.0 5.5 LN 10.3 10.7 9.4 10.2 of Graphiurus surdus from 12 localities com- LD 6.2 6.4 5.1 5.7 prise all the known examples of the species PL 9.0 9.4 8.0 8.3 preserved in institutional collections. The LBP 5.1 5.2 4.5 4.8 numbers preceding these localities match the LIF 2.5 2.8 2.7 2.4 numbered dots on the distribution map of G. BIF 1.6 1.7 1.8 2.0 surdus (fig. 2). BBP 3.6 3.3 3.1 3.7 CLP4-M3 3.2 3.3 3.3 CAMEROON CLM1-3 2.5 2.7 2.7 BP4 0.9 0.9 1. Ayina (Aina) River (1.48N, 13.10E): BM1 0.9 1.1 1.1 1.1 MRAC 1093. BM2 1.0 1.1 1.1 1.2 2. Bitye (3.1ON, 12.20E): BM(NH) BIT 1.7 1.9 1.6 1.9 14.7.23.15. PPL 10.7 11.7 3. Edea (3.47N, 10.13E): ZMB 71341. LB 7.0 7.2 7.1 7.2 4. Eseka (3.38N, 10.47E): CM 42202, BBO 2.5 2.7 2.3 42205, 42208, 42209, 42210; AMNH 236482. a Cameroon; skin and skull; fourth premolar erupted 5. Kribi (2.56N, 9.54E): MCZ 8607 (ho- halfway between alveolus and occlusal surfaces of mo- lotype of Graphiurus schwabi). lars. 6. Efulan (Efulen) (2.40N, 10.45E): AMNH b Equatorial Guinea; skin and skull, originally pre- 90067. served in fluid; maxillary teeth slightly to moderately 7. Metet (3.23N, 11.43E): MCZ 14669, woM. c Zaire; skin and skull; maxillary teeth slightly to mod- 17605, 17606. erately worn. 8. Sangmelima(2.56N, 11.58E): CM 10331. d Zaire; skin and skull; maxillary teeth slightly to mod- erately worn, occlusal surface obscured by glue. e Recorded by Dollman (1912: 314), who indicated the dimensions were "measured from spirit-specimen." reported by Dollman (1912: 314), although we may not I could not obtain accurate measurements for length of have measured the ear the same way. head and body and length oftail. I measured the dry ear f Obtained by me from the dry hind foot (includes and obtained 8 mm for its length; presumably shrinkage claws). Dollman (1912: 314) gave 19 mm, obtained pre- accounts for the difference between my value and that sumably when the specimen was wet. 18 AMERICAN MUSEUM NOVITATES NO. 3157

'0G GO 10' 200

100 0 0 100 Fig. 2. Geographic distribution of Graphiurus surdus as indicated by specimens I identified. Each numbered locality corresponds to the numbered place name in the gazetteer of collection sites and specimens for G. surdus.

EQUATORIAL GUINEA ZAIRE 9. Rio Muni, Benito River: BM(NH) 11. Inkongo, Sankuru River (4.55S,23.1 SE): 0.2.5.53, 0.2.5.54, 0.2.5.55 (holotype of BM(NH) 13.4.7.2. Graphiurus surdus), 0.2.5.56. 12. Masako(0.36N, 25.13E),500 m: RUCA D1655. GABON SYMPATRY: The geographic ranges of four 10. Belinga (1.09N, 13.12E): MNHN 1983- species of Graphiurus overlap the known dis- 186. tribution of G. surdus, but records of sym- 1 996 HOLDEN: SUB-SAHARAN AFRICAN DORMICE. PART 1 19

200

100

00

100

200 300 400 500 Fig. 2. Extended. patry, based on specimens I identified, exist specimen of G. surdus was obtained (see the only between G. surdus and G. lorraineus map of Cameroon localities in Robbins and (Cameroon, Equatorial Guinea, and Zaire), Schlitter, 1981). G. crassicaudatus (Cameroon), and G. hueti EMENDED DiAGNosIs: When compared with (Cameroon); table 3. Most records of the G. christyi and G. lorraineus, the only species fourth species, G. christyi, are from north- of Graphiurus with which it is likely to be eastern and northern Zaire (see the map in confused, G. surdus exhibits 1) small ears rel- Schlitter et al., 1985, for example), but a se- ative to size of head and body, 2) silky fur, ries was collected at Lolodorf in southern 3) inconspicuous eye mask, 4) relatively Cameroon (see gazetteer ofG. christyi below), straight conformation of the zygomatic arch a place not far from Efulan from which a in lateral view, 5) long palatal region relative 20 AMERICAN MUSEUM NOVITATES NO. 3157

TABLE 3 Records of Graphiurus lorraineus, G. crassicaudatus, and G. hueti (referenced by museum acronym and catalog number) taken together with G. surdus at localities in Cameroon, Equatorial Guinea, and Zaire (Numbers in parentheses refer to localities in gazetteer and symbols on map in fig. 2. Specimens of G. surdus from each place are listed in the gazetteer.) Locality G. lorraineus G. crassicaudatus G. hueti CAMEROON (2) Bitye BM(NH) 13.2.8.6, BM(NH) 23.1.22.52 13.9.12.8, 23.1.22.53 MRAC 10939 (4) Eseka - AMNH 236483; CM 42204 (5) Kribi MCZ 8678 - - (6) Efulan - - BM(NH) 1.11.21.13, 96.3.20.12, USNM 125434 (7) Metet CM 4645, 4646, 4674 - MCZ 17607 FMNH 48926 (8) Sangmelima CM 42720, 9529, - CM 9525, 9527, 10332-10335, 9530, 10336 10337, 10338 EQUATORIAL GUINEA (9) Benito River BM(NH) 58.336 ZAIRE (11) Inkongo BM(NH) 13.4.7.3, MRAC 7220, 12190 to skull length, 6) relatively short and narrow and two inguinal), and both sexes have short, incisive foramina, 7) relatively narrow P4, wide arboreal feet with six plantar pads. and 8) low rostrum relative to its breadth. The skull of G. surdus is exemplified by the DESCRIPrION: A compact body and rela- holotype in figure 4 and specimens from tively short tail characterize the body form Cameroon and Zaire in figures 6-10. Its gen- of G. surdus (fig. 3; table 4), a configuration eral conformation, relationships of cranial common to most species of subsaharan Af- bones to one another, and position and num- rican dormice. The grayish brown, silky dor- ber of cranial foramina are very similar to sal pelage is composed ofoverhairs, short and those in the skull drawings ofGraphiurus pre- fine guard hairs that barely extend beyond sented by Wahlert et al. (1993: 11-14), in the overfur, but no intermixed layer of un- which the bony elements and foramina are derhairs. The short and silky ventral coat is also labeled. The rostrum is wide and mod- dark gray washed with whitish buff: the hairs erately long, the interorbit shaped like an are dark gray for most of their lengths and hourglass in dorsal perspective, the braincase tipped with either white or pale buff. The rounded, and the zygomatic arches flair intermixing of long brown and unpigmented slightly outward. The palatal region is rela- hairs, and brown hairs with unpigmented tips, tively long and the auditory bullae are not along the caudal vertebrae provide G. surdus enlarged. Each zygomatic arch meets the ros- with its characteristic frosted brown tail. The trum in a smooth, concave wide angle, and small pinnae are dark gray, their inner and not sharply at a right angle. The dorsal max- outer surfaces clothed with short, fine hairs. illary root ofeach zygomatic arch lies anterior Facial mask and eye rings are indistinct. Dor- to the ventral maxillary root. Structural de- sal surfaces of metacarpal and metatarsal tails will be elaborated in the following sec- regions are brown, tops of digits are covered tion where I compare the cranium of G. sur- with silvery hairs. As in all other species of dus with that of other species. Graphiurus, G. surdus females have four pairs The dentary of G. surdus is shaped very of mammae (one pectoral, one postaxillary, much like the labeled drawing of a Graphiu- 1 996 HOLDEN: SUB-SAHARAN AFRICAN DORMICE. PART 1 21

Fig. 3. Stuffed skin of an adult Graphiurus surdus, BM(NH) 14.7.23.15, from Cameroon. Measure- ments (mm): LHB = 95; LT = 65; LHF = 18; LE = 13. The tail tip appears to be slightly damaged, and the accuracy of length of tail in this specimen is questionable. 22 AMERICAN MUSEUM NOVITATES NO. 3157

Fig. 4. Cranial views (x 3.5) ofthe holotype ofGraphiurus surdus: BM(NH) 0.2.5.55, an adult female from the Benito River, Rio Muni, Equatorial Guinea. Measurements are listed in table 2. rus dentary illustrated in Wahlert et al. (1993: KARYOTYPIC AND MOLECULAR DATA: No 15). I did not study the mandible in any com- sample of G. surdus has been analyzed for parative detail for the reasons listed above chromosomal or biochemical information. under Materials and Methods. Molecular data may only be retrieved from An example of the dentition of G. surdus museum skins, or carcasses preserved in al- in occlusal perspective is provided in figure cohol; no specially preserved tissue samples 5. See the section on Mandible and Dentition have been obtained. under Materials and Methods above for ad- GEOGRAPHIC VARIATION: The geographi- ditional discussion. cally biased sampling, combined with the 1 996 HOLDEN: SUB-SAHARAN AFRICAN DORMICE. PART 1 23

Fig. 5. Scanning electron micrographs of the left lower (on the left; clp4-m3 = 3.2 mm) and upper right (on the right; CLP4-M3 = 3.3 mm) toothrows (fourth premolar to third molar) of a young adult Graphiurus surdus (AMNH 90067) from Efulan, Cameroon. small sample size from each locality, pre- torial Guinea, and Gabon (table 4). The only vents any useful analysis of geographic vari- exception is the incisor tips of the ation. The known geographic range of the from Masako, which are wider than any of species is represented mostly by collecting the other specimens measured. localities in southern Cameroon, one in near- DISCUSSION OF SYNONYM: Graphiurus by Rio Muni and one in Gabon, and two schwabi, based on one specimen from Kribi, scattered far to the east in Zaire. The col- Cameroon, was described by G. M. Allen in lecting localities therefore do not provide ad- April, 1912, just one month after Dollman equate geographic samples to discern any sig- named G. surdus. Allen (1912: 441) charac- nificant patterns ofvariation. The total num- terized G. schwabi as "A small dark gray spe- ber of specimens is only 22, and each geo- cies, with the orbital rings scarcely distin- graphic sample is too small to provide guishable from the general dark slaty color information regarding variation within and of the head. Feet except the distal portion of between populations. Out ofthe 12 localities, the toes entirely dusky. Tail white bordered, nine are represented by only one specimen, ears small." and the other three by 4-6 animals. Without Allen apparently revised his opinion about large samples containing a range in age so schwabi, for in 1939 he placed the name in series ofcomparable stages can be compared, the synonymy of G. haedulus (a taxon I in- and without samples that cover most of the clude at present in G. lorraineus; Holden, geographic range, no rigorous inquiry into 1993). Subsequent workers did not always how morphology or other traits may vary agree, and during the next 50 years, schwabi geographically is possible. was treated again as a species, relegated to Nearly all values obtained from the two the synonymy of G. surdus, omitted from a specimens from Zaire fall within the ob- checklist, referred to G. murinus, combined served ranges of respective measurements in with haedulus again to be placed in the syn- the pooled sample from Cameroon, Equa- onymy of G. lorraineus, and finally returned 24 AMERICAN MUSEUM NOVITATES NO. 3157

TABLE 4 TABLE 4-(Continued) Comparisons of Measurements (mm) and Weight (g) Between Geographic Samples of Graphiurus Eq. Guinea surdus (Benito River), Cameroon (Mean ± 1 SD, range [in parentheses], and num- (Kribi, Metet, ber of specimens are listed for the large sample Bitye, Eseka, Zaire containing more than one specimen. Only indi- Sangmelima), viduals in the age range young to old adults were Gabon (Belinga)a Inkongob Masakoc measured.) CLP4-M3 3.2 ± 0.16 3.2 3.1 Eq. Guinea (2.9-3.5) 16 (Benito River), CLM1-3 2.6 ± 0.13 2.6 2.5 Cameroon (2.4-2.8) 16 (Kribi, Metet, BP4 0.8 ± 0.07 0.9 0.8 Bitye, Eseka, Zaire (0.7-0.9) 14 Sangmelima), BM1 1.0 ± 0.08 1.0 1.0 Gabon (Belinga)a Inkongob Masakoc (0.9-1.1) 14 BM2 1.1 ± 0.06 1.0 1.0 LHB 99.0 ± 7.93 95 98 (1.0-1.2) 14 (87-110) 8 BIT 1.9 ± 0.16 - 2.6 LT 72.3 ± 6.19 70 70 (1.7-2.2) 15 (65-82) 6 PPL 11.8±0.62 - 12.1 LHF 20.8 ± 1.29 18 19 (11.0-12.9) 11 (18-22) 12 LB 7.3 ± 0.20 - 7.7 LE 12.3 ± 1.62 11 14 (6.9-7.7) 14 (9-14) 11 BBO 2.7 ± 0.13 - 2.5 WT 24.8 ± 5.31 - 26 (2.4-2.8) 13 (18-34) 6 GLS 27.6 ± 0.98 - 28.6 a AMNH 90067,236482; BMNH 0.2.5.55 (holotype), (26.5-29.4) 10 0.2.5.53-0.2.5.56, 14.7.23.15; CM 10331, 42202, 42205, CIL 24.9 ± 1.03 - 25.4 42208-42210; MCZ 17605, 17606; MNHN 1983-186. (23.5-26.3) 11 bBMNH 13.4.7.2. ZB 14.6 ± 0.61 - 15.2 c RUCA D1655. (13.4-15.7) 11 IB 4.5 ± 0.19 4.5 4.6 to G. surdus, as outlined below. Ellerman (4.1-4.8) 14 (1940) was the only one who recognized BBC 12.6 ± 0.41 - 12.1 schwabi as a species, but because he had not (11.7-12.8) 11 examined the holotype, listed the taxon un- HBC 8.1 ± 0.27 - 8.1 der the heading "not seen and not allocated (7.7-8.5) 10 to group." Perret and Aellen (1956) also did BR 5.7 ± 0.33 - 5 9 not examine the (5.2-6.4) 16 holotype of G. schwabi, but LHR 4.8 ± 0.39 4.8 4.9 they proposed, based on Allen's (1912) orig- (4.0-5.5) 13 inal description, that schwabi was probably LN 10.4 ± 0.53 10.6 11.6 a synonym of G. surdus, not G. haedulus. (9.8-11.3) 13 Their knowledge of G. surdus was limited to LD 6.5 ± 0.36 6.6 6.5 Dollman's (1912) description; they had also (5.9-7.0) 16 not studied that holotype. In his checklist, PL 9.3 ± 0.43 - 9.7 Misonne (1974) did not include schwabi. Ge- (8.6-10.3) 16 nest-Villard (1978: 409) placed both schwabi LBP 5.2 ± 0.29 - 5.6 and surdus as synonyms of G. murinus lor- (4.8-5.8) 16 LIF 2.8 ± 0.20 3.0 2.9 raineus. (2.5-3.2) 16 Robbins and Schlitter (1981) maintained BIF 1.8 ± 0.12 - 1.8 that schwabi and haedulus were synonyms of (1.6-2.0) 16 G. lorraineus. However, they provided no BBP 3.4 ± 0.16 3.4 reason for their allocation of schwabi to G. (3.1-3.6) 15 lorraineus. Their only discussion of schwabi is in the account of G. surdus where they 1 996 HOLDEN: SUB-SAHARAN AFRICAN DORMICE. PART 1 25

TABLE 5 Comparisons of Selected Measurements (mm) Among Three Species of Graphiurus from Equatorial Africa (Mean ± 1 SD, range [in parentheses], and number of specimens are listed for each measurement. The range in age from young to old adults is contained within each sample.) G. surdusb G. crassicaudatusa Eq. Guinea, G. huetic Cameroon Cameroon, Gabon Cameroon LHB 92.6 ± 4.0 99.0 ± 7.93 145.9 ± 5.54 (83-98) 11 (87-110) 8 (134-152)11 LT 59.4 ± 4.42 72.3 ± 6.19 131.1 ± 11.31 (55-70) 9 (65-82) 6 (114-149) 8 LHF 17.7 ± 0.89 20.8 ± 1.29 31.5 ± 0.82 (16-19) 12 (18-22) 12 (30-32) 11 LE 13.0 ± 1.12 12.3 ± 1.62 22.3 ± 3.69 (11-14)9 (9-14) 11 (20-31)8 GLS 26.6 ± 0.79 27.6 ± 0.98 38.1 ± 0.98 (25.2-27.8) 8 (26.5-29.4) 10 (36.3-39.5) 15 ZB 16.1 ± 0.35 14.6 ± 0.61 (15.7-16.6) 6 (13.4-15.7) 11 IB 4.9 ± 0.17 4.5 ± 0.19 6.2 ± 0.23 (4.7-5.2) 14 (4.1-4.8) 14 (5.7-6.5) 20 LN 8.9 ± 0.53 10.4 ± 0.53 (8.1-9.9) 12 (9.8-11.3) 13 CLP4-M3 3.8 ± 0.21 3.2 ± 0.16 5.2 ± 0.15 (3.4-4.2) 14 (2.9-3.5) 16 (5.0-5.6) 17 CLM1-3 3.0 ± 0.12 2.6 ± 0.13 4.2 ± 0.17 (2.8-3.2) 14 (2.4-2.8) 16 (3.8-4.6) 18 BP4 1.1 ± 0.07 0.8 ± 0.07 (0.9-1.1) 12 (0.7-0.9) 14 BM1 1.3 ± 0.05 1.0 ± 0.08 1.8 ± 0.05 (1.2-1.3) 12 (0.9-1.1) 14 (1.7-1.8) 16 BM2 1.3 ± 0.06 1.1 ± 0.06 (1.2-1.4) 13 (1.0-1.2) 14 BIT 1.5 ±0.11 1.9±0.16 (1.3-1.7) 13 (1.7-2.2) 15 LB 6.7 ± 0.12 7.3 ± 0.20 - (6.5-6.9) 11 (6.9-7.7) 14 aAMNH 89582, 89583, 236483; BM(NH) 23.1.22.52; CM 2973, 42203, 42204, 42206, 42207, 42211, 42715, 42718, 59453; ZMB 71312. b Data derived from specimens listed in table 4. CBM(NH) 1.11.21.13, 36.10.28.26, 36.10.28.27, 96.3.20.12; CM 2888, 3676, 6123, 6124, 9525, 10332-10335, 10337, 42200, 42201, 16134; MCZ 17607, 17920; USNM 125434. outlined the treatment of schwabi by Allen ear length has any relevance here is mysti- (1939), and Perret and Aellen (1956), and fying. Absolute length does not distinguish noted that the original descriptions of both samples of G. surdus from G. lorraineus, as G. surdus and G. schwabi included reference is apparent with their data (p. 274) and mine to their small ears ("9 and 7 mm, respec- (table 6). They finally asserted that "Exami- tively"), and continued, "However, external nation of specimens from Rio Muni and measurements of G. surdus were taken from southern Cameroon indicates that G. surdus fluid preserved specimens, whereas those of is a species distinct from G. haedulus and G. G. schwabi were from a dried skin originally schwabi; the latter two taxa are synonyms of preserved in formalin. The ears of our spec- G. lorraineus." imens measured 12 to 14 mm" (p. 286). Why After examining all existing holotypes, as 26 AMERICAN MUSEUM NOVITATES NO. 3157

Fig. 6. Comparisons among adult crania of Graphiurus from a dorsal perspective. Left: G. crassi- caudatus, BM(NH) 12.1.12.1 (holotype ofG. c. dorotheae), Nigeria. Middle: G. surdus, BM(NH) 14.7.23.15, Cameroon. Right: G. hueti, CM 42201, Cameroon. x 2.5. Note that the anterolateral margin of the dorsal maxillary process of each zygomatic arch joins the rostrum at a right angle in G. crassicaudatus and G. hueti, but at an obtuse angle in G. surdus. That species is obviously smaller than G. hueti and contrasts with G. crassicaudatus, which is similar in size, by its longer nasals and a narrower interorbit (see table 5) that is not defined by supraorbital shelves (also see discussion in text). well as most museum specimens of Graphiu- have shrunken unduly"), the pinnae of G. rus, I do not hesitate to arrange the name schwabi are actually very small, even for a schwabi as a junior synonym of G. surdus. young dormouse. The holotype ofG. schwabi is ajuvenile, with permanent premolarsjust beginning to erupt. The small size of this young specimen may Comparisons with Other Species of account for its erroneous association with Graphiurus haedulus and placement in the synonymy of Four other species of Graphiurus occur in G. lorraineus. Almost all specimens ofG. sur- the general forested region where G. surdus dus that I examined have the characteristic is found. Two of these have very different relatively horizontal and straight and robust morphologies and have rarely been confused zygomatic arch that distinguishes this species with G. surdus. Set apart from all the other from G. lorraineus; the holotype ofG. schwa- West African forest species of Graphiurus by bi shares this zygomatic conformation. Gra- its large body and skull dimensions, G. hueti phiurus surdus also has very short and narrow has always been recognized as a distinctive incisive foramina, another trait that is con- species easily separated not only by size but spicuous in the holotype of G. schwabi. And by morphology (table 5; figs. 6, 7; also see the even allowing for some shrinkage (in his de- key to Cameroon forest species in Robbins scription ofschwabi, Allen (1912: 442) noted and Schlitter, 1981); hueti has never been as- that "The ears seem unusually small, but may sociated with surdus in synonymy or group- 1996 HOLDEN: SUB-SAHARAN AFRICAN DORMICE. PART 1 27

Fig. 7. Ventral views (x 2.5) of the same crania of Graphiurus shown in figure 6. Left: G. crassicau- datus. Middle: G. surdus. Right: G. hueti. x 2.5. Compared with G. crassicaudatus, G. surdus has a narrower cranium, larger auditory bullae, wider incisors, shorter toothrows and narrower maxillary teeth (see table 5, and discussion in text). ings in any published checklist or faunal ac- the infrequently collected G. crassicaudatus count. as a distinct species. The other, G. crassicaudatus, although I will present my taxonomic revisions of similar in body and skull size to G. surdus G. crassicaudatus and G. hueti in future re- (table 5), is at once separable by its shorter ports; further comparison, other than the nasal bones, wide interorbit defined by prom- morphometric contrasts mentioned above, inent supraorbital shelves, smaller auditory indicated in table 5, shown in figures 6 and bullae, narrower incisors, and much larger 7, and outlined by Robbins and Schlitter teeth (table 5; figs. 6, 7; Robbins and Schlit- (1981), are not needed here to clarify the mor- ter, 1981). Also, in both G. crassicaudatus phological identity of G. surdus. and G. hueti, the anterolateral margin of the Samples of two other species do require dorsal maxillary root ofeach zygomatic arch careful comparison with material represent- contacts the rostrum at a right angle, not the ing G. surdus. Individuals of G. christyi and wide angle forming a concave outline be- G. lorraineus are superficially similar in mor- tween arch and end of rostrum that is char- phology to specimens of G. surdus, found in acteristic of G. surdus and most other species the same forest region, and these two species of Graphiurus. Except for Misonne (1974), have been misidentified as G. surdus in the who included crassicaudatus ("perhaps not published literature and collections of mu- readily separable from murinus"), along with seums; comparisons among the three are surdus, in G. murinus, everyone who has pub- critical to the definition of G. surdus. lished checklists and relevant faunal reports Dollman (1914: 80) described G. christyi, concerning subsaharan dormice has accepted stating that it was "related to Graphiurus 28 AMERICAN MUSEUM NOVITATES NO. 3157 spurrelli Dollm., similar in colour and size once relegated to the synonymy of G. muri- but differing in the general form ofthe skull." nus, each is morphologically diagnosable, The name was accepted to indicate a species particularly in those regions where their geo- of Graphiurus in the checklists provided by graphic ranges overlap. Furthermore, they Allen (1939) and Ellerman (1940), but was have been caught at the same localities [re- included in G. murinus by Misonne (1974) cords documented below in the gazetteers, and in G. murinus lorraineus by Genest-Vil- and also noted by Hatt (1940)] in Cameroon lard (1978); those two authors also immersed (Lolodorf) and Zaire (Niangara, Medje, Ya- surdus within their vague definitions of G. losemba). Based on my observations, Gra- murinus. Robbins and Schlitter (1981) ex- phiurus lorraineus is not morphologically tracted christyi as a diagnosable species with- similar to any member of the G. murinus in the context of documenting the identifi- complex. Graphiurus christyi, however, is al- cation of samples from Cameroon, and lied with the G. murinus group, but its in- Schlitter et al. (1985) reasserted its integrity terspecific relationships within that complex and outlined its geographic distribution in are unclear, and it is possible that detailed southern Cameroon and northern Zaire. study may place it as a junior synonym of I agreed with the observations made by another species now contained within the G. Schlitter and his colleagues (Holden, 1993). murinus complex. Morphological traits of all the specimens I Systematic revisions of G. christyi and G. have seen from southern Cameroon and lorraineus will be the subjects of future sec- northern Zaire are like those characterizing tions of my overall revision of Graphiurus, the holotype of christyi and other specimens and maps showing their known geographic Dollman described, and separate them from ranges as documented by specimens will be any ofthe other species ofdormouse recorded included in each ofthose respective accounts. from the same region. What I intend to present here are the con- Graphiurus lorraineus was also named by trasts between those two species and G. sur- Dollman (1910: 285) who noted that the dus. To accomplish this, I brought together "species is easily distinguished from the other 35 adults of G. christyi and 112 of G. lorrai- members of the genus by the brilliant buff neus. The examples that I studied and the coloration ofthe upper parts ofthe body, the localities at which they were collected are list- comparatively small ears, and short and rath- ed in the following gazetteers. These are only er stoutly built skull." Through the years, the a small portion of the known museum spec- treatment ofthis species in the literature par- imens of the two species that I examined. I alleled that ofchristyi: recognized as a species used these smaller samples obtained from by Allen (1939) and Ellerman (1940); merged southern Cameroon and northeastern and with G. murinus, along with christyi and sur- central Zaire, to compare with the sample of dus, by Misonne (1974) and Genest-Villard G. surdus, whose known geographic distri- (1978); reestablished as morphologically dis- bution is in the same general region. tinctive by Robbins and Schlitter (1981), who documented samples ofit and G. christyi from Graphiurus christyi southern Cameroon (Lolodorf); reported upon again by Schlitter et al. (1985), who noted CAMEROON that G. lorraineus "is widely distributed and 1. Lolodorf (3.14N, 10.48E): CM 2912, 2953, frequently common in forested and second 2972, 4606. growth situations in West and Equatorial Af- ZAIRE rica"; and recognized as a species by me (Holden, 1993). Based on specimens studied, 2. Avakubi (L.18N, 27.35E): AMNH 49890. my 3. Niangara (3.45N, 27.54E): AMNH 49882. estimate ofthe geographic distribution of 4. Mambaka (0.51N, 27.33E): BM(NH) G. lorraineus is much broader than indicated 14.2.11.8, 14.2.11.9. in the literature; its geographic distribution 5. Medje (2.25N, 27.18E): AMNH 49891,49894, will be fully documented in the forthcoming 49895, 49897, 49900-49902, 49905, 49912- systematic revision of that species. 49917, 49919, 49920, 51297; BM(NH) Although both christyi and lorraineus were 19.5.8.66; USNM 259140. 1996 HOLDEN: SUB-SAHARAN AFRICAN DORMICE. PART 1 29

6. Yalosemba (2.35N, 21.47E): CM 86748- GRAPHIURUS SuRDus COMPARED WITH G. 86754, 86756. CHRISTYi: Graphiurus christyi, out of all the dormice species found in the central African forest region, is the most similar to G. surdus Graphiurus lorraineus in body size, fur coloration, and cranial mor- CAMEROON phology. Adults of both species have about the same body size (judged by length of head 1. 7 km NW Ambam (Ambam = 2.23N, 11.16E), Meyo/Ambam: MHNG 912/65. and body), but G. surdus has, on the average, 2. Assobam (3.15N, 14.02E): BM(NH) 9.10.2.21. absolutely shorter tail and ears, and longer 3. Bafia (4.40N, 1l.05E): CM 9528. hind feet (table 6). Proportional differences, 4. Bitye (3.10N, 12.20E): BM(NH) 13.2.8.6, diagrammed in figure 13, reflect these con- 13.9.12.8, 23.1.22.53; MRAC 10939. trasts in absolute values listed in table 6: Rel- 5. Ebolowa (2.56N, 1l. llE): CM 16135. ative to length of head and body, G. surdus 6. Kribi (2.56N, 9.54E): MCZ 8678. has significantly shorter tail and ears but lon- 7. Lolodorf (3.14N, 10.48E): CM 2879, 2915, ger hind feet, compared to the sample of G. 2969, 2971, 2999, 3685, 3914, 3915, 4560, christyi. In this set ofcomparisons, short ears, 4561, 4867, 4868, 5638, 5641, 5679, 5686, as Dollman (1912) noted, are diagnostic of 5699, 5704, 5733, 5737, 6120-6122, 13252, 14927, 16136, 16142-16144; MCZ 8676, G. surdus, and the difference is evident when 17921, 17922. examples of each species are compared side- 8. Metet (3.23N, 1 1.43E): CM 4645,4646,4673, by-side. The differences in lengths oftail and 4674; FMNH 48926. hind feet are not always apparent by exam- 9. Sangmelima (2.56N, 11.58E): CM 42720, ination of specimens, but become evident 9529, 9530, 10336. when means of samples are compared. 10. Yaounde (3.51N, 11.31E): CM 42721. The two species differ in texture and color 11. Yokadouma (3.25N, 15.08E): CM 42719. ofpelage. The dorsal coat of G. surdus is very soft and silky to the touch, whereas that of EQUATORIAL GUINEA G. christyi is velvety. This contrast may be 12. Rio Muni, Benito River: BM(NH) 0.2.5.57. produced by differences in hair morphology, by a denser distribution of hairs in the coat GABON of G. surdus, by the slightly greater length of the dorsal fur and guard hairs in G. surdus, 13. Mitzic (0.48N, 11.30E): BM(NH) 58.336. or some combination thereof. The rump hairs measure between 5 and 7 mm (dorsal guard ZAIRE hairs 9 to 11 mm) in specimens of G. surdus, 14. Bafwabaka (2.1ON, 27.39E): AMNH 49888, and rump hairs range from 4 to 6 mm (dorsal 49889. guard hairs 7 to 11 mm) in examples of G. 15. Barumbu (I.15N, 23.29E): MRAC 9209. christyi. 16. Faradje (3.45N, 29.43E): AMNH 49886. Dark brown upperparts character- 17. Inkongo (4.55S, 23.15E): BM(NH) 13.4.7.3, grayish 7220; MRAC 12190. ize specimens of G. surdus; the range from 18. Kananga (5.53S, 22.26E): AMNH 86311, brownish gray to dark brownish buff or even 86316, 86318; BM(NH) 26.7.6.192; MRAC reddish brown is typical of G. christyi. The 7144, 7491; TM 9767, 9768. overall color of grayish brown of G. surdus 19. , (Stanley Pool) (4.15S, is flat, but the silky hairs ofspecimens in good 15.25E): SMF 6299. condition have a sheen, so that the dorsal 20. Equateur Province, Lukolela (1.1OS, 17.1 1E): coat of these specimens gleams when held AMNH 86893, 86894, 86896. near a light source, not unlike the fur ofsome 21. Medje (2.25N, 27.18E): AMNH 49904,49906. amphibious mammals. I do not believe that 22. Niangara (3.45N, 27.54E): AMNH 49878; USNM 259142; MRAC 12993. this is due to so many of the skins having 23. Nzoro (Vankerckhovenville) (3.2 IN, 29.32E): been prepared from fluid specimens, as some AMNH 49885. examples of G. christyi were also prepared 24. Poko (3.08N, 26.52E): BM(NH) 19.5.8.67. from specimens preserved in alcohol, and do 25. Yalosemba (2.35N, 21.47E): CM 86755, not have this characteristic sheen, and some 86758-86761. specimens of G. surdus that exhibit this trait 30 AMERICAN MUSEUM NOVITATES NO. 3157

TABLE 6 Comparisons ofMeasurements (mm) and Weight (g) Among Three Species of Graphiurus from Equatorial Africa (Mean ± 1 SD, range [in parentheses], and number of specimens are listed for each measurement. The range in age from young to old adults is contained within each sample.) G. surdusa Cameroon, G. christyib G. lorraineusc Gabon, Zaire, Cameroon, Cameroon, Gabon, Eq. Guinea Zaire Zaire, Eq. Guinea LHB 98.5 ± 7.11 96.9 ± 5.52 80.5 ± 5.28 (87-110) 10 (86-170) 30 (70-93) 57 LT 71.8 ± 5.34 80.2 ± 5.39 66.8 ± 6.13 (65-82) 8 (73-95) 27 (45-77) 50 LHF 20.4 ± 1.45 18.1 ± 1.22 16.5 ± 1.55 (18-22) 14 (16-20) 31 (12.7-19.1) 57 LE 12.5 ± 1.66 14.2 ± 1.09 12.4 ± 1.49 (9-15) 13 (12-17) 27 (9-15) 27 WT 25.0 ± 4.86 29.0 ± 3.22 16.8 ± 4.15 (18-34) 7 (25-33) 6 (12-24) 5 GLS 27.7 ± 0.98 28.0 ± 1.00 24.5 ± 0.74 (26.5-29.4) 11 (26.7-29.7) 25 (22.7-26.2) 47 CIL 25.0 ± 0.99 24.1 ± 1.03 21.3 ± 0.71 (23.5-26.3) 12 (22.5-26.0) 26 (19.6-22.8) 52 ZB 14.6 ± 0.61 15.1 ± 0.76 13.8 ± 0.48 (23.4-15.7) 12 (13.3-16.7) 23 (13.0-14.9) 41 IB 4.5 ± 0.18 4.7 ± 0.17 4.2 ± 0.18 (4.1-4.8) 16 (4.3-5.2) 34 (3.7-4.7) 82 BBC 12.5 ± 0.41 12.6 ± 0.42 11.7 ± 0.38 (11.7-11.9) 12 (12.1-13.7) 27 (10.8-12.4) 51 HBC 8.1 ± 0.25 8.1 ± 0.28 7.4 ± 0.33 (7.7-8.5) 11 (7.5-8.7) 27 (6.4-8.2) 43 BR 5.7 ± 0.33 5.8 ± 0.26 5.1 ± 0.28 (5.2-6.4) 17 (5.4-6.2) 21 (4.5-5.7) 66 LHR 4.8 ± 0.36 5.3 ± 0.39 4.5 ± 0.25 (4.0-5.5) 15 (4.4-6.1) 31 (3.9-5.0) 75 LN 10.5 ± 0.57 10.5 ± 0.54 9.1 ± 0.44 (9.8-11.6) 15 (8.7-11.3) 32 (7.5-10.0) 96 LD 6.5 ± 0.34 6.1 ± 0.39 5.1 ± 0.29 (5.9-7.0) 18 (5.2-6.9) 34 (4.2-6.0) 81 PL 9.3 ± 0.43 8.6 ± 0.39 7.8 ± 0.32 (8.6-10.3) 17 (7.7-9.1) 29 (6.9-8.4) 70 LBP 5.2 ± 0.29 4.7 ± 0.25 4.2 ± 0.26 (4.8-5.8) 17 (4.2-5.0) 30 (3.6-4.8) 66 LIF 2.8 ± 0.19 3.0 ± 0.25 2.7 ± 0.22 (2.5-3.2) 18 (2.4-3.3) 32 (2.0-3.2) 77 BIF 1.8 ± 0.11 2.2 ± 0.13 1.7 ± 0.13 (1.6-2.0) 17 (2.0-2.4) 32 (1.2-2.0) 76 BBP 3.4 ± 0.15 3.5 ± 0.15 3.1 ± 0.16 (3.1-3.6) 16 (3.3-3.9) 32 (2.7-3.5) 77 CLP4-M3 3.2 ± 0.15 3.2 ± 0.09 3.1 ± 0.15 (2.9-3.5) 18 (2.9-3.5) 33 (2.8-3.4) 79 CLM1-3 2.6 ± 0.14 2.5 ± 0.09 2.5 ± 0.13 (2.4-2.8) 18 (2.4-2.8) 33 (2.1-2.7) 80 BP4 0.8 ± 0.07 0.9 ± 0.05 0.8 ± 0.07 (0.7-0.9) 16 (0.8-0.9) 20 (0.7-1.1) 70 BM1 1.0 ± 0.07 1.0 ± 0.03 1.0 ± 0.06 (0.9-1.1) 16 (1.0-1.1) 24 (0.8-1.1) 75 1996 HOLDEN: SUB-SAHARAN AFRICAN DORMICE. PART 1 31

TABLE 6-(Continued) G. surdusa Cameroon, G. christyib G. lorraineusc Gabon, Zaire, Cameroon, Cameroon, Gabon, Eq. Guinea Zaire Zaire, Eq. Guinea BM2 1.1 ± 0.07 1.1 ± 0.05 1.1 ± 0.07 (1.0-1.2) 16 (1.0-1.2) 26 (0.9-1.2) 76 BIT 1.9 ± 0.15 1.9 ± 0.13 1.6 ± 0.12 (1.7-2.2) 16 (1.7-2.2) 33 (1.3-1.9) 74 PPL 11.8 ± 0.60 12.2 ± 0.65 10.5 ± 0.55 (11.0-12.9) 12 (11.3-13.3) 21 (9.5-11.7) 41 LB 7.3 ± 0.22 7.4 ± 0.37 7.0 ± 0.31 (6.9-7.7) 15 (6.6-7.9) 31 (6.4-7.8) 76 BBO 2.7 ± 0.13 2.5 ± 0.24 2.1 ± 0.16 (2.4-2.8) 14 (2.1-3.0) 23 (1.8-2.5) 63 a Data derived from specimens listed in table 4. b AMNH 49884, 49890, 49891, 49894, 49895, 49897, 49900-49902, 49905, 49912-49917, 49919, 49920, 51297; BM(NH) 14.2.11.8, 14.2.11.9, 19.5.8.66; CM 2953, 2912, 2972, 4606, 86748-86754; USNM 259140. c BM(NH) 9.10.2.21, 13.2.8.6, 13.9.12.8, 23.1.22.53; CM 2869, 2879, 2915, 2971, 2999, 3685, 3914, 3915, 4560, 4561, 4645, 4646, 4673, 4674, 4867, 4868, 5704, 5638, 5641, 5679, 5686, 5699, 5733, 5737, 6120-6122, 9528, 10336, 13252, 14927, 16135, 16142-16144, 16186, 46720; FMNH 48926; MCZ 8676, 8678, 17921.

were never immersed in spirit. The brownish Specimens of G. christyi in general have a buff of the dorsal coat of G. christyi is rich more clearly delineated eye mask, and a dis- and lustrous, but lacks the shininess seen in tinct eye mask is commonly exhibited by examples of G. surdus. members of the G. murinus complex, based Portions ofthe long hairs that make up the on my examination of hundreds of speci- dark gray and bushy tail of G. surdus are mens. This difference in prominence of the unpigmented at their tips, and some hairs are eye mask is another trait used by Robbins unpigmented from base to tip; this results in and Schlitter (1981) to help distinguish the the tail appearing frosted. The pattern is char- two species (however, they described it as acteristic ofalmost all examples examined of absent in G. surdus, but present in G. chris- G. surdus; it is conspicuous in some (partic- tyi), and one that I could verify. ularly younger) samples, and while less prom- Skulls of the two species appear alike in inent in other specimens, it is almost always size and shape (figs. 8-10), but the similarity evident upon careful inspection. This frosting is superficial. Significant absolute and pro- pattern on the tail was not detected in some portional differences exist. Robbins and skins preserved in alcohol, and because those Schlitter (198 1: 286) remarked that G. surdus specimens are discolored (by being either "has a long and narrow skull with a short overly darkened or foxed), the pattern, if maxillary toothrow and short depth of ros- present, is obscured. Robbins and Schlitter trum." The accuracy of this statement de- (1981) considered this trait to be diagnostic pends on the species being compared to G. of the species. Hairs in the tail of G. christyi surdus. Absolute length and width of skull are pigmented to their tips so the tails are the and length oftoothrow are not useful for sep- same color as the dorsum and not frosted. arating specimens of G. surdus from those of This pattern holds for most of the adult G. G. christyi. Means of greatest length of skull christyi I studied; a few individuals had slight and condyloincisive length (estimating skull frosting on parts of the tail, but never over length), zygomatic breadth and breadth of the entire tail. braincase (indices of skull width), and length Examples of G. surdus have a very faint oftoothrow (that includes premolars and mo- eye mask extending from the nose or base of lars or only molars) in my two samples are the whiskers, circumscribing the eyes, and not significantly different (table 6). Further- ending anterior to the base ofthe ear pinnae. more, none of these measurements influence L

.-

O~N

Ii' cU- ) 0 1D;I w < 't. '.t 400-1. 004 co It Af4vt 0

-Y V.- *00

Oob te o3 2j4 ,*>.;/2>X coXr2Pa^

32 0 0

C-

L.

-....A N-:, 4qpp I k t%N-:

A14 *1*

cisE0

^ a ) .-- rCr.

Ln - 1. 00

CU

'T-lw Cd c

(o C) 41 /

33 .s X 7S.

~A- 00~ o

.-4 - U

0

0

A U

0. I. 61I IiX, , Y .£ o -o

0=

3oZ3 1996 HOLDEN: SUB-SAHARAN AFRICAN DORMICE. PART 1 35

TABLE 7 TABLE 8 Factor Coefficients for General Size and Species Factor Coefficients for General Size and Species Differences (adjusted for general size) Estimated Differences (adjusted for general size) Estimated from Log-Transformed Values of 14 Cranial and from Log-Transformed Values of 14 Cranial and 1 Dental Measurements of G. christyi and G. 3 Dental Measurements of G. lorraineus and G. surdus surdus (Sample of G. christyi is from Cameroon and (Samples of G. surdus are from Cameroon, Zaire; that of G. surdus comes from Cameroon, Equatorial Guinea [Rio Muni], and Zaire; those Gabon, Equatorial Guinea, and Zaire. Specimen of G. lorraineus were obtained in Cameroon, scores are plotted in fig. 11.) Equatorial Guinea [Rio Muni], Gabon, and Zaire. Specimen scores are plotted in fig. 12.) General Species size differences General Species (factor 1) (factor 2) size differencesa (factor 1) (factor 2) GLS .259 -.035 IB .058 .115 GLS .180 .154 BBC .115 .005 IB .052 .129 BIT .436 -.136 BBC .157 .024 LHR .334 .281 BIT .381 .232 LN .280 -.057 LHR .225 -.063 LD .372 -.280 LN .313 .113 PL .240 -.322 LD .206 .423 LIF .342 .122 PL .191 .308 BIF .290 .679 LIF .258 -.173 BBP .173 .100 BIF .292 -.166 CLP4-M3 .021 .041 BBP .110 .122 CLM1-3 .037 -.008 CLMI-3 .234 -.087 LB .217 .053 BM1 .288 -.188 LBP .224 -.461 BP4 .406 -.520 LB .153 -.005 LBP .107 .470 BR .258 .087 the spread of scores on factors representing general size versus size-adjusted differences a Signs ofcoefficients changed following convention in between the two species (table 7; fig. 11). Voss and Marcus (1992: 1923). Robbins and Schlitter (1981) were correct about depth of rostrum; G. surdus is char- acterized by a significantly shallower ros- indicate that relative to greatest length ofskull, trum, not only in absolute mean value (table the palatal region (indexed by the signifi- 6), but also relative to breadth of rostrum cantly greater palatal length, and lengths of when compared with G. christyi (fig. 13). Least diastema and bony palate) is significantly height of rostrum is also one of the size-ad- longer than in G. christyi; postpalatal length, justed differences influencing morphometric however, is relatively shorter in G. surdus, separation of the two species in multivariate an expected configuration given that overall space (table 7; fig. 11). skull length is nearly the same in the two Skulls of G. surdus are characteristically species. elongate, but in the palatal region, not in Breadth of incisive foramina also helps overall length. Palatal length, length of bony separate the two species (table 7; fig. 11). It palate, and length ofdiastema strongly influ- is absolutely narrower in G. surdus than in ence separation of the samples of G. surdus G. christyi (table 6), and significantly narrow- and G. christyi in the principal components er relative to breadth of rostrum (fig. 13). analysis (table 7; fig. 11). Absolute values of Other morphometric distinctions between means of these measurements are also dis- the two samples exist in absolute mean values similar between the two species (table 6), as (table 6) but are not significant. However, are proportional relationships illustrated in there are additional significant proportional the ratio diagram (fig. 13). There the profiles distinctions. For example, compared with G. 36 AMERICAN MUSEUM NOVITATES NO. 3157

* G. christyi A G. surdus -0.5 II II

LU 0 * -0.6 z 0 * LU 0 o00 LlJ -0.7 0 0 @0*. 0 .

LU. 0 C-) -0.8

D A A -0.9 k LU A N A A L7i A A -1.0 A A

-1.1 II II 5.3 5.4 5.5 5.6 5.7 5.8 5.8 6.0 GENERAL SIZE Fig. 11. Discrimination between samples of Graphiurus surdus from Cameroon, Gabon, Equatorial Guinea, and Zaire (N = 1 1) and G. christyi from Cameroon and Zaire (N = 21) as indicated by individual specimen scores plotted on factors representing general size and size-adjusted species differences. The separation along the size-adjusted axis is predominantly influenced by differences in breadth of incisive foramina and length of bony palate, and to a lesser extent by palatal length, least height of rostrum, and length of the diastema (table 7). christyi, the sample of G. surdus has shorter clearly exhibited by the holotype of G. surdus incisive foramina relative to length of dia- (fig. 4). stema, narrower bony palate relative to its G. SURDUS COMPARED WITH G. LORRAINEUS: length, narrower fourth upper premolar rel- The primary reasons for contrasting samples ative to the first upper molar, and wider ba- of G. surdus and G. lorraineus is simply that sioccipital relative to bullar length (reflecting surdus, along with lorraineus, was either re- the absolute smaller bullae of G. surdus; fig. ferred to another species, G. murinus, (Mi- 13). sonne, 1974), or allocated as a synonym of One qualitative cranial character found in G. murinus lorraineus (Genest-Villard, 1978), most specimens examined is unique to G. despite evidence for sympatry and apparent surdus. In lateral view, the side of each zy- dissimilar morphologies of surdus and lor- gomatic arch (primarily the jugal and squa- raineus. mosal process) has relatively straight dorsal Within the geographic region where both and ventral surfaces; the dorsal surface is G. surdus and G. lorraineus occur, examples strongly arcuate and ventral margin shallow- of G. surdus are conspicuously larger, a con- ly concave in other species of Graphiurus, trast reflected in mean differences of most such as G. christyi and G. lorraineus. The measurements (table 6). Among the external extremes in variation of this character, and measurements, mean ear length is about the comparisons with G. christyi and G. lorrai- same in the two species, but relative to length neus can be seen in figure 10. The G. surdus of head and body, ear length is significantly specimen from Masako, Zaire (top left) shorter in G. surdus than in G. lorraineus (fig. strongly exhibits this trait, whereas it is much 14). The tail is absolutely longer in G. surdus, less evident in the specimen from Bitye, but significantly shorter relative to length of Cameroon (bottom left). This character is also head and body compared with G. lorraineus l1996 HOLDEN: SUB-SAHARAN AFRICAN DORMICE. PART 1 37

m G. lorraineus . G. surdus 3.70

A LUJ A U 3.50 z AA LUJ A A LU LL- 3.30 A LLJ A A

(.) D 3.10 F C3 *in*.*

LU N U * uNo 2.90 U

2.70 L I I 4.C)O 4.50 5.00 5.50 GENERAL SIZE Fig. 12. Discrimination between samples of Graphiurus surdus from Cameroon, Equatorial Guinea, and Zaire (N = 1) and G. lorraineus from Cameroon, Equatorial Guinea, Gabon, and Zaire (N = 20) as reflected by specimen scores plotted on factors representing general size and size-adjusted species differences. Spread along the size-adjusted axis is most strongly influenced by differences between samples ofthe two species in breadth ofthe fourth upper premolar, length ofbony palate, length ofthe diastema, and to some extent by differences in palatal length and breadth of incisor tips (table 8).

(fig. 14). Relatively shorter tail and ear also ample, certain dimensions reflecting cranial distinguished G. surdus from G. christyi. lengths in G. surdus (condyloincisive and pal- Length ofhind foot, although absolutely lon- atal lengths, lengths of nasals, diastema, and ger in G. surdus, does not differ significantly bony palate) are significantly longer relative relative to length of head and body when to particular breadth dimensions (zygomatic compared with G. lorraineus (fig. 14). and interorbital breadths, breadths of brain- Contrasts between G. surdus and G. Ior- case, rostrum, and incisive foramina), com- raineus in texture and color of fur parallel pared with G. lorraineus. Thus, when com- those observable between G. surdus and G. pared with G. lorraineus (but not with G. christyi. Compared with the silky, grayish christyi, see discussion above), G. surdus does brown dorsal coat of G. surdus, that of G. not have a long, narrow skull as described by lorraineus is velvety and coppery brown; G. Robbins and Schlitter (1981: 286). In G. sur- surdus has inconspicuous eye rings, G. Ior- dus, the rostrum is lower (least height of ros- raineus has conspicuous dark rings; the tail trum) relative to rostral breadth (also noted of G. surdus is frosted, that of G. lorraineus by Robbins and Schlitter, 1981), and the in- is usually monocolored (although it should cisors are significantly broader relative to ros- be noted that some juvenile specimens of G. trum and other breadth dimensions com- lorraineus have some unpigmented hairs pared with G. lorraineus. The auditory bullae mixed in their otherwise brown tails). of G. surdus are smaller relative to condy- In addition to the greater mean values of loincisive length than those of G. lorraineus; most cranial measurements of G. surdus this is reflected in the broad basioccipital of compared with G. lorraineus (table 6), there G. surdus, which is wider relative to most of are significant proportional differences illus- the other breadth dimensions, as well as length trated in the ratio diagram (fig. 14). For ex- of the bullae (fig. 14). AMERICAN MUSEUM NOVITATES NO. 3157

.08 -.1. -.08 -.06 -.04 -.02 0 .02 .04 .06 -F-----f II I~~~~~~~~~~~ G. christyi - LHB k LT G. surdus

lHF & I LE CIL ZB IB . i ._*-- HBC P- .4 tr.-, BBC 0-- r I

BR 9-

*.. 0 LHR ---

lN 0 0-1-1 LD 0 1\ PL "I LBP lIF

_ BIF .-p- -- _-

BBP 0-1

OZ14 CLM 1 -3 41-

-- BP4 0 op-- 0 i

BM 1 0 0 - BIT 0 4 or_, PPL 0::.'N

LB I I 9 4---Z- BBO

-.1 -.08 -.06 -.04 -.02 0 .02 .04 .06 .08

Fig. 13. Ratio diagram indicating some proportional relationships between samples of Graphiurus. Dimensions are compared between samples of Graphiurus christyi (the standard) from Cameroon and Zaire and G. surdus from Cameroon, Zaire, Gabon, and Equatorial Guinea (see discussion in text). For each measurement, the absolute value ofthe mean, and plus and minus two standard errors of the mean, were converted to logarithms. For each dimension, the logarithm ofthe mean ofthe standard (G. christyi) was subtracted from the logarithm of the mean of the series of G. surdus, and the logarithms of plus and minus two standard errors of the mean of the standard were subtracted from the logarithms of plus 1 996 HOLDEN: SUB-SAHARAN AFRICAN DORMICE. PART 1 39

Portions of the palatal region of G. surdus resembles the shape characteristic ofG. chris- are relatively long. Compared with G. Ior- tyi, and differs from the horizontal configu- raineus, the diastema is significantly longer ration with its nearly straight top and bottom relative to condyloincisive length in G. surdus that is diagnostic of most specimens of G. (influenced in part by the relatively shorter surdus (fig. 10, top left) (see discussion of toothrow of G. surdus), and the incisive fo- variation ofthis character under comparison ramina are significantly shorter relative to with G. christyi). diastemal length (fig. 14). Similar propor- tional differences in the diastemal region were NATURAL HISTORY found between G. surdus and G. christyi (fig. Very little information is available regard- 13). The palatal length of G. surdus is only ing the ecology of G. surdus. The only data I slightly longer than that of G. lorraineus rel- have located comes from collectors. A spec- ative to condyloincisive length (fig. 14), al- imen collected in Bitye, Cameroon, by G. L. though this small difference does aid in dis- Bates was "smoked out of a hollow tree" crimination between the species in multivar- (written on skin tag, BM(NH) 14.7.23.15). iate space (fig. 12, table 8). This suggests that like G. crassicaudatus, G. Mean values of the various dental mea- hueti, and G. lorraineus, and possibly some surements do not differ significantly between other species of Graphiurus, G. surdus may the two species; basically, the upper premo- often nest in hollow trees. lars and molars of G. surdus are the same However, G. surdus may also have the ca- absolute size as those of G. lorraineus, but pacity to build nests in other suitable cavities set in a larger skull (table 6). So although these or locations. Robbins and Schlitter (1981: dimensions do not differ in absolute terms, 286), for example, wrote that "the specimens they are relatively dissimilar. Compared with [of G. surdus] from Eseka were taken in the G. lorraineus, length of molar row, for ex- same trapline as G. crassicaudatus." They de- ample, is significantly shorter relative to scribed the trapline in their discussion ofthat length of bony palate in G. surdus, and the species (p. 284): fourth upper premolar is significantly nar- rower relative to the first upper molar (fig. Specimens [of G. crassicaudatus] from the vi- 14). cinity of Eseka were trapped in Sherman non- Separation along the size-adjusted axis in folding aluminum live-traps set on vines and multivariate space of samples ofthe two spe- horizontal branches in secondary high forest. cies is most strongly influenced by differences No hollow trees were seen in the immediate area. All traps were baited with the nut of the in breadth of the fourth upper premolar, and oil palm (Elaeis). Traps were in position for two lengths of diastema and bony palate (table 8; or more weeks before dormice entered. fig. 12). Finally, the qualitative trait that often sep- Notes written on the skin tag of the spec- arates G. surdus from G. christyi also dis- imen from Inkongo, Zaire (BM(NH) criminates between the former species and 13.4.7.2), recorded only that the animal was G. lorraineus. The configuration of the zy- "killed in forest. This or a similar species [is] gomatic arch in G. lorraineus in lateral view often found in thatched roofs of houses. Na-

and minus two standard errors of the mean of G. surdus. Measurements larger than the standard are represented by positive values, those smaller by negative values. In each sample, the solid or dashed lines connect the means of measurements, the horizontal bars represent plus or minus two standard errors of the mean. A sample with the same proportions as the standard will be represented by mean values on a line parallel to that ofthe standard regardless ofabsolute size. Also, ifvalues for the samples being compared with the standard are similar in absolute size, they will be close together on the diagram. Ifproportions between any ofthe measured dimensions are similar, the positions oftheir points relative to each other on the horizontal scale will be similar. The diagram was constructed from values for mean, standard deviation, and sample size of variates listed in table 6. 40 AMERICAN MUSEUM NOVITATES NO. 3157

-.04 -.02 0 .02 .04 .06 .08 .10 I I I~ ~~I"I~ "-~ _ LHB G. Iorraineus

------LT --- - - G. surdus

LHF 0 0 -----

--*- LE OL- CIL ZB __- IB HBC

BBC .4 0-4-. BR LHR LN LD Pl LBP N lIF BIF BBP

CLM 1 -3 *----4

BP4 0 0 -- PC- --- - BM1 9-- BIT

PPL a i LB BBO I -.04 -.02 0 .02 .04 .06 .08 .10 Fig. 14. Ratio diagram indicating some proportional relationships between species of Graphiurus. Dimensions are compared between samples ofG. lorraineus (the standard) and G. surdus from Cameroon, Gabon, Zaire, and Equatorial Guinea (see discussion in text). Basic properties of the ratio diagram are explained in the legend to figure 13. This visual proportional representation was derived from the values for mean, standard deviation, and sample size of the variates listed in table 6. 1996 HOLDEN: SUB-SAHARAN AFRICAN DORMICE. PART 1 41 tive name Kankyesa." It is doubtful that G. Zaire), in a loop of the river Tshopo. Of the surdus was the species commonly found in 2105 ha of the reserve, one-third is covered thatched roofs of houses at Inkongo. Speci- by primary forest with Gilbertiodendron dew- mens of G. lorraineus are often recorded as evrei as the dominant tree, and two-thirds by having been taken from either thatched roofs secondary forest grown upon primary forest or simply from houses, and this species was logged between 1948 and 1955." also collected in greater numbers at Inkongo Much remains to be learned about the ecol- (table 3). ogy of G. surdus. Its diet, reproductive bi- The animal collected at Masako (from the ology, behavior, kinds of nest sites, and nest Masako Forest Reserve) was captured in a construction are unknown. Specimens have Victor snaptrap baited with fruit of oil palm been trapped in forests, but we are ignorant (Elaeis guinensis). The trap was set on the of the animal's microhabitat. How G. surdus ground in secondary forest (W. Verheyen, and the other three (possibly four) species of personal commun.). The Reserve, according dormice with which it is sympatric coexist to Hutterer and Dudu (1990: 305) ".. .is sit- and partition resources is an intriguing ques- uated 14 km northeast of (Haut- tion that needs to be explored.

REFERENCES Allen, G. M. terials for the practice of logotechnics. 1912. New African rodents. Bull. Mus. Comp. Baltimore: published by author. Zool. Harvard Coll. 54(14): 439-447. Carleton, M. D., and C. Brian Robbins 1936. A new genus and a new subspecies of 1985. On the status and affinities ofHybomys African dormouse. J. Mammal. 17: 292- planifrons (Miller, 1900) (Rodentia: 293. Muridae). Proc. Biol. Soc. Washington 1939. A checklist of African mammals. Bull. 98(4): 956-1003. Mus. Comp. Zool. Harvard Coll. 83: 1- Cracraft, J. 763. 1983. Species concepts and speciation analy- Ansell, W. F. H. sis. Curr. Ornithol. 1: 159-187. 1974. Some mammals from Zambia and ad- 1987. Species concepts and the ontology of jacent countries. Puku. Occas. Pap. Nat. evolution. Biol. Philos. 2: 63-80. Parks and Wildlife Serv., Zambia. Suppl. 1989. Speciation and its ontology: The em- 1: 21-49. pirical consequences of alternative spe- 1978. The mammals of Zambia. Chilanga: cies concepts for understanding patterns National Parks and Wildlife Service. and processes of differentiation. In D. 1989. Mammals from Malawi: Part II. Nyala Otte and J. A. Endler (eds.), Speciation 13(1-2): 41-65. and its consequences, pp. 28-59. Sun- Ansell, W. F. H., and R. J. Dowsett derland, MA: Sinauer. 1988. Mammals of Malawi: An annotated Davis, J. L., and P. S. Manos check list and atlas. St. Ives: Trendrine 1991. Isozyme variation and species delimi- Press. tation in the Puccinellia nuttalliana Baum, D. complex (Poaceae): An application of 1992. Phylogenetic species concepts. TREE the phylogenetic species concept. Syst. 7(1): 1-2. Bot. 16: 431-445. Beurton, P. J. de Graaff, G. 1995. How is a species kept together? Biol. 1981. The rodents of Southern Africa. Notes Philos. 10: 181-196. on their identification, distribution, Booth, A. M. ecology and taxonomy. Durban, Pre- 1958. The Niger, the Volta, and the Dahomey toria: Butterworths. Gap as geographic barriers. Evolution de Queiroz, K., and M. J. Donoghue 12: 48-62. 1988. Phylogenetic systematics and the spe- Brown, R. W. cies problem. Cladistics 4: 317-338. 1956. Composition of scientific words. A 1990a. Phylogenetic systematics of Nelson's manual ofmethods and a lexicon ofma- version of cladistics? Ibid. 6: 61-75. 42 AMERICAN MUSEUM NOVITATES NO. 3157

1990b. Phylogenetic systematics and species re- Washington, D.C.: Smithsonian Inst. visited. Ibid. 6: 83-90. Press. Delany, M. J. Hatt, R. T. 1975. The rodents ofUganda. London: British 1940. Lagomorpha and Rodentia other than Museum (Natural History). Sciuridae, Anomaluridae and Idiuridae, Dollman, G. collected by the American Museum 1910. Two new species ofAfrican dormice be- Congo Expedition. Bull. Am. Mus. Nat. longing to the genus Graphiurus. Ann. Hist. 76(9): 457-604. Mag. Nat. Hist., ser. 8, 5: 285-288. Hill, J. E., and T. D. Carter 1912. Seven new African dormice. Ibid. 9: 1941. The mammals of Angola, Africa. Bull. 312-320. Am. Mus. Nat. Hist. 78(1): 211 pp. 1914. Notes on mammals collected by Dr. Hollister, N. Christy in the Congo and by Dr. Bayer 1919. East African mammals in the United in Uganda and British . Rev. States National Museum. Part II. Ro- Zool. Africaine 4(1): 77-80. dentia, Lagomorpha, and Tubuliden- Donoghue, M. J. tata. Bull. U.S. Natl. Mus. 99(2): 184 1985. A critique ofthe biological species con- PP. cept and recommendations for a phy- Holden, M. E. logenetic alternative. The Bryologist 88: 1993. Family Myoxidae. In D. E. Wilson and 172-181. D. M. Reeder (eds.), Mammal species Ellerman, J. R. ofthe world, a taxonomic and geograph- 1940. The families and genera of living ro- ic reference, 2nd ed., pp. 763-770. dents. Vol. 1. Rodents other than Mur- Washington, D.C.: Smithsonian Inst. idae. London: British Museum (Natural Press. History), 689 pp. Hutterer, R. Ellerman, J. R., T. C. S. Morrison-Scott, and R. 1993. Order Insectivora. In D. E. Wilson and W. Hayman D. M. Reeder (eds.), Mammal species 1953. Southern African Mammals 1758 to ofthe world, a taxonomic and geograph- 1951: a reclassification. London: British ic reference, 2nd ed., pp. 69-130. Wash- Museum (Natural History). ington, D.C.: Smithsonian Inst. Press. Endler, J. A. Hutterer, R., and A. Dudu 1989. Conceptual and other problems in spe- 1990. Redescription of Crocidura caliginea, a ciation. In D. Otte and J. A. Endler rare shrew from northeastern Zaire (eds.), Speciation and its consequences, (Mammalia: Soricidae). J. African Zool. pp. 625-648. Sunderland, MA: Sinauer. 104: 305-311. Frost, D. R., and D. M. Hillis 1990. Species in concept and practice: her- Jentink, F. A. petological applications. Herpetologica 1888. Zoological researches in Liberia. A list 46(1): 87-104. ofmammals, collected by J. Biittikofer, Frost, D. R., A. G. Kluge, and D. M. Hillis C. F. Sala and F. X. Stampfli, with bi- 1992. Species in contemporary herpetology: ological observations. Notes Leyden comments on phylogenetic inference and Museum 10: 1-58. taxonomy. Herpetol. Rev. 23(2): 46-54. Mayr, E. Genest-Villard, H. 1942. Systematics and the origin of species. 1978. Revision systematique du genre Gra- New York: Columbia Univ. Press. phiurus (Rongeurs, Gliridae). Mam- 1963. Animal species and evolution. Boston, malia 42(4): 391-426. MA: Belknap Press. Groves, C. P. 1969. Principles of systematic zoology. New 1993. Order Primates. In D. E. Wilson and D. York: McGraw Hill. M. Reeder (eds.), Mammal species of Meester, J. A. J., I. L. Rautenbach, N. J. Dippe- the world, a taxonomic and geographic naar, and C. M. Baker reference, 2nd ed., pp. 243-277. Wash- 1986. Classification of Southern African ington, D.C.: Smithsonian Inst. Press. mammals. Transvaal Mus. Monogr. No. Grubb, P. 5. Pretoria: Transvaal Museum. 1993. Order Artiodactyla. In D. E. Wilson and Mishler, B. D., and R. N. Brandon D. M. Reeder (eds.), Mammal species 1987. Individuality, pluralism, and the phy- ofthe world, a taxonomic and geograph- logenetic species concept. Biol. Philos. ic reference, 2nd ed., pp. 377-414. 2: 397-414. 1996 HOLDEN: SUB-SAHARAN AFRICAN DORMICE. PART 1 43

Misonne, X. Rosen, D. E. 1974. Order Rodentia. In J. Meester and H. 1979. Fishes from the uplands and intermon- W. Setzer (eds.), The mammals of Af- tane basins of Guatemala: Revisionary rica: an identification manual, Part 6, studies and comparative geography. pp. 1-39. Washington, D.C.: Smithson- Bull. Am. Mus. Nat. Hist. 162: 267- ian Inst. Press. 376. Moreau, R. E. Rosevear, D. R. 1966. The bird faunas ofAfrica and its islands. 1969. The rodents of West Africa. London: New York: Academic Press. British Museum (Natural History). 1969. Climatic change and the distribution of Schlitter, D. A., L. W. Robbins, and S. L. Williams forest vegetation in West Africa. J. Zool. 1985. Taxonomic status of dormice (genus 158: 39-61. Graphiurus) from west and central Af- Musser, G. G., and M. D. Carleton rica. Ann. Carnegie Mus. 54(1): 1-9. 1993. Family Muridae. In D. E. Wilson and Shortridge, G. C. D. M. Reeder (eds.), Mammal species 1934. The mammals of South West Africa. A ofthe world, a taxonomic and geograph- biological account of the forms occur- ic reference, 2nd ed., pp. 501-755. ring in that region. Vol. 1. London: Wil- Washington, D.C.: Smithsonian Inst. liam Heinemann. Press. Skinner, J. D., and R. H. N. Smithers Nixon, K. C., and Q. D. Wheeler 1990. The mammals of the Southern African 1990. An amplification of the phylogenetic , 2nd ed. Pretoria: Univ. Pre- species concept. Cladistics 6: 211-223. toria. O'Hara, R. J. Smithers, R. H. N. and the species 1971. The mammals ofBotswana. Mus. Mem. 1994. Evolutionary history No. 4. Salisbury: National Museums of problem. Am. Zool. 34: 12-22. Rhodesia. Patton, J. L., and M. F. Smith Smithers, R. H. N., and J. L. P. L. Tello 1994. Paraphyly, polyphyly, and the nature of 1976. Checklist and atlas of the mammals of species boundaries in pocket gophers Mozambique. Mus. Mem. No. 8. Sal- (genus Thomomys). Syst. Biol. 43(1): 11- isbury: National Museums and Monu- 26. ments of Rhodesia. Perret, J. L., and V. Aellen Smithers, R. H. N., and V. J. Wilson 1956. Mammiferes du Cameroon de la collec- 1979. Checklist and atlas of the mammals of tion J. L. Perret. Rev. Suisse Zool. Zimbabwe Rhodesia. Mus. Mem. No. 63(26): 395-447. 9. Salisbury: National Museums and Reuvens, C. L. Monuments of Rhodesia. 1890a. Die Myosidae oder schlaefer. Ein bei- Smuts, J. trag zur osteologie und systematik der 1832. Enumeration em Mammalium Capen- Nagethiere. "Inaugural-Dissertation zyr sivm, 4th ed. Leidae. Erlangung der Doctorwuerde der Hihen Swynnerton, G. H., and R. W. Hayman Philosophischen Fakultaet der KGL. 1951. A checklist of the land mammals of the Friedrich-Alexander-Universitaet zu Tanganyika Territory and the Zanzibar erlangen." Leiden: P. W. M. Trap. Protectorate. J. East African Nat. Hist. 1890b. Note IV. Einiges uber die Myoxidae oder Soc. 20(6): 274-392. schlafer. Notes from the Leyden Mu- Templeton, A. R. seum 13: 65-76. 1989. The meaning of species and speciation: Ridley, M. a genetic perspective. In D. Otte and J. 1989. The cladistic solution to the species A. Endler (eds.), Speciation and its con- problem. Biol. Philos. 4: 1-16. sequences, pp. 3-27. Sunderland, MA: Robbins, L. W., and D. A. Schlitter Sinauer. 1981. Systematic status ofdormice (Rodentia: Thomas, O., and M. A. C. Hinton Gliridae) from southern Cameroon, Af- 1925. On mammals collected in 1923 by Cap- rica. Ann. Carnegie Mus. 50(9): 271- tain G. C. Shortridge during the Percy 288. Sladen and Kaffrarian Museum Expe- Roberts, A. dition to South-West Africa. Proc. Zool. 1951. The mammals of South Africa. Johan- Soc. London, 1925: 221-246. nesburg: Trustees of "The mammals of Trouessart, E. L. South Africa" book fund. 1897. Catalogus mammalium tam viventium 44 AMERICAN MUSEUM NOVITATES NO. 3157

quam fossilium. Nova editio (prima Voss, R. S., L. F. Marcus, and P. Escalante completa). Fasc. 3. Rodentia II. (My- 1990. Morphological evolution in muroid ro- omorpha, Hystrichomorpha, Lagomor- dents I. Conservative patterns of cra- pha.), pp. 453-664. Berlin: R. Friedlan- niometric covariance and their onto- der. genetic basis in the Neotropical genus 1899. Catalogus mammalium tam viventium Zygodontomys. Evolution 44(6): 1568- quam fossilium. Nova editio (prima 1587. completa). Fasc. 6. Appendix (addenda Wahlert, J. H., S. L. Sawitzke, and M. E. Holden et corrigenda). Index alphabeticus, pp. 1993. Cranial anatomy and relationships of 1265-1469. Berlin: R. Friedlander. dormice (Rodentia, Myoxidae). Am. 1904. Catalogus mammalium tam viventium Mus. Novitates 3061: 32 pp. quam fossilium. Quinquennale supple- Wheeler, Q. D., and K. C. Nixon mentum, anno 1904, pts. 1-2, pp. 1- 1990. Another way of looking at the species 546. Berlin: R. Friedlander. problem. Cladistics 6: 77-81. Voss, R. S. White, F. 1988. Systematics and ecology of ichthy- 1983. The vegetation of Africa. A descriptive omyine rodents (Muroidea): patterns of memoir to accompany the Unesco/ morphological evolution in a small AETFAT/UNSO vegetation map ofAf- adaptive radiation. Bull. Am. Mus. Nat. rica (published in 1981). Paris: Unesco. Hist. 188(2): 259-493. Yalden, D. W., M. J. Largen, and D. Kock Voss, R. S. and L. F. Marcus 1976. Catalogue ofthe mammals of Ethiopia. 1992. Morphological evolution in muroid ro- 2. Insectivora and Rodentia. Monit. dents II. Craniometric factor divergence Zool. Italaliano 8(1): 118 pp. in seven neotropical genera, with ex- perimental results from Zygodontomys. Evolution 46(6): 1918-1934.

Recent issues of the Novitates may be purchased from the Museum. Lists of back issues of the Novitates, Bulletin, and Anthropological Papers published during the last five years are available free of charge. Address orders to: American Museum of Natural History Library, Department D, Central Park West at 79th St., New York, N.Y. 10024. TEL: (212) 769-5545. FAX: (212) 769- 5009. E-MAIL: [email protected]

This paper meets the requirements of ANSI/NISO Z39.48-1992 (Permanence of Paper).