Introduction

Anomaluridae are small to large rodents restricted to equatorial Africa between latitudes 15°N and 15°S. All except one species, Zenkerella insignis, are able to perform gliding flight. All anomalurids live in rain forest, montane forest, or gallery forest. High trees provide takeoff and landing sites for glides and provide hollows for the daytime rest. As anomalures are nocturnal and seldom maintained in captivity, very little is known about their biology Durrell, 1952; Rahm, 1969; Schlitter 1989; Julliot et al., 1998.

Seven species of Anomaluridae are currently recognized Dieterlen, 1993, 2005 and placed in three or four genera: Anomalurus which occasionally contains Anomalurops, Zenkerella, and Idiurus.

Most current authors agree that Idiurus consists of two species, I. macrotis and I. zenkeri Misonne, 1971; Rahm, 1988; Nowak, 1991; Dieterlen, 1993, 2005. They are the smallest members of the family, with head and body lengths ranging from about 95 mm in I. macrotis to about 75 mm in I. zenkeri. These animals have a gliding membrane supported by a cartilaginous rod extending from the elbow and a long, sparsely haired tail that considerably exceeds the head and body length.

A preliminary analysis of the distributions of both species Schunke and Hutterer, 2000 revealed that large continuous distributions across the Central African forest belt, as postulated by some authors e.g. Kingdon, 1997, were not supported by specimen-based distribution maps, which in contrast suggest patchy distributions for both species. Geographical isolation may cause genetic drift and result in speciation of populations of these arboreal rodents. To test this hypothesis, we have analyzed the morphological variation among geographical populations of Idiurus species across their entire range from Sierra Leone in the west to Tanzania in the east.

Specimens and Methods

Specimens in 14 museum collections form the basis of our report. In total, we examined 116 Idiurus macrotis and 83 I. zenkeri appendix 1. Of these, craniometric data taken from 63 skulls of I. macrotis and 51 skulls of I. zenkeri were used for the calculations. Only skulls with all molars completely emerged were used in order to exclude juveniles. From each skull, up to 57 measurements were taken with an electronic caliper to the nearest 0.01 mm fig. 1, table 1.

Figure 1

Definition of linear measurements taken from skulls, as exemplified by a skull of Idiurus macrotis ZMB 22885. Greatest length and width of single teeth not marked in figure were also taken. For abbreviations, see table 1. The scale for the dentition represents 1 mm and for the skull, 10 mm.

Figure 2

Distribution of I. macrotis based on specimens examined and literature records. Closed symbols, specimens examined; open symbols, data taken from literature Sanderson, 1940; Kuhn, 1965; Rahm, 1966; Adam, et al. 1970; Aellen et al., 1970; Julliot et al., 1998. Scale represents 1,000 km.

Figure 3

Distribution of I. zenkeri based on specimens examined and literature records. Closed symbols, specimens examined; open symbols, data taken from literature Rahm, 1966; Jones, 1971; Delany, 1975. Scale represents 1,000 km.

Table 1

0: character not used in final PCA; 1 and 2: character used in analyses; 2: character used in figs. 4 to Figure 5Figure 67. Abbreviations as in fig. 1.

Figure 4

Geographic variation of I. macrotis according to the first and second principal components as obtained from logarithmized absolute skull measurements. For characters, see table 1. Open symbols, males; closed symbols, females; symbols with a central dot, sex unknown; diamonds, West Africa; upright triangles, northwestern Cameroon; triangles pointing upside down, southern Cameroon to Gabon; circles, Democratic Republic of Congo; star, Tanzania.

Figure 5

Variation of I. zenkeri according to the first and second principal components as obtained from the logarithmized absolute skull measurements. For characters, see table 1. Open symbols, males; closed symbols, females; symbols with a central dot, sex unknown; upright triangles, northwestern Cameroon; triangles pointing upside down, southern Cameroon and Equatorial Guinea; square, Central African Republic; circles, Democratic Republic of Congo.

Figure 6

Size variation of I. macrotis and I. zenkeri from the first principal component as obtained from logarithmized absolute skull measurements. For characters, see table 1.

Figure 7

Shape variation of I. macrotis and I. zenkeri from the second principal component as obtained from residuals calculated for palatilar length. For characters, see table 1.

Variation in color was assessed in a sample of 55 skins of I. macrotis and 41 skins of I. zenkeri; color slides showing the skins along with a color reference were used for comparison.

The specimens originated from the entire ranges of the two species figs. 2–3. We defined six geographical clusters and used them as operational units:

The two species were analyzed separately. Measurements were converted to logarithms before entering the calculations. The normal distribution of the data was tested with a Kolmogorov-Smirnov test independently for both species. For this test, specimens from the northeastern part of the Democratic Republic of Congo were used because the largest numbers of specimens were available from this area. The same subsets were also tested for sexual dimorphism with t-tests with the exclusion of specimen of unknown sex.

For the main analysis, the remaining data were submitted to a principal component analysis PCA. Two PCAs were conducted. The first was based on the ln-transformed absolute measurements and thus included variation in size. In the second analysis, linear regression lines of individual ln-transformed measurements on ln-transformed palatilar length PL, fig. 1 were calculated, and residuals of the individual measurements on this line were used instead of original variables Thorpe and Leamy, 1983 to compare only shape differences. Palatilar length was chosen as reference because of its relatively high values and thus relatively low measuring error and for its availability from most specimens.

In addition, a PCA was performed for standard body measurements collectors' field data, including total length, tail length, ear length, and hind-foot length appendix 2. Total length was taken as combined tail length and head and body length, where necessary. Residuals were calculated for the hind-foot length.

After grouping the populations according to the results of the PCA see later, a discriminant analysis was conducted to specify differences between groups.

As Idiurus skulls are very fragile, it was rarely possible to take all measurements from a single skull. Therefore, the PCA was conducted first with all characters and then repeated after a stepwise exclusion of measurements that were unavailable for several specimens. While the number of specimens increased, the number of characters decreased. By comparing the results, it was possible to test their robustness.

Examples of typical results with intermediate numbers of characters and specimens are shown in scatter plots for the first and second principal component, or in box plots. Calculations were processed with the program SPSS version 10.0.

Additional statistical details are given in appendix 3, tables 8–Table 9Table 10Table 1112.

Table 2

Percentage of “grouped” cases correctly classified: 94.23%.

Table 3

Percentage of “grouped” cases correctly classified: 94.87%.

Institutional Abbreviations

Results

Discriminant Analysis

Most individuals of the clusters obtained through the PCA analysis could be identified with only few characters. More than 90% of specimens were classified correctly. In populations of I. macrotis, a combination of four characters height of rostrum, length of bulla, length of mandible, and distance between articular and angular process yielded 94.2% correct classifications tables 2, 11. However, the results for specimens from northwestern Cameroon must be taken with caution, as only five skulls were available from this area.

The discriminant analysis gave similar results for I. zenkeri populations, with 94.9% correctly classified individuals tables 3, 12. The characters used were height of rostrum, width of bulla, and upper and lower rostrum length. Unfortunately, only specimens from southern Cameroon plus Equatorial Guinea and the Democratic Republic of Congo yielded a sufficient number of measurements and could be used in this analysis.

I. macrotis and I. zenkeri may be separated from each other by a single character, such as height of rostrum, basal length, or palatilar length. Each of these characters correctly classified 100% of the specimens in a discriminant analysis.

Tables 4 and 5 provide means and ranges of five selected skull measurements of the different geographical populations of each species, and figures 8–Figure 910 show examples of crania of both species. Skulls of all holotype specimens are figured, with the exception of kivuensis Lönnberg, 1917, the skull of which is lost.

Table 4

Mean Values and Ranges of Standard Skull Measurements in I. macrotis

Table 5

Mean Values and Ranges of Standard Skull Measurements in I. zenkeri

Figure 8

Dorsal and ventral views of crania of Idiurus macrotis. From left to right: Oda, Ghana BMNH 46.579, holotype of cansdalei Hayman, 1946; Ossidinge, northwestern Cameroon ZMB 12992; Efulen, southern Cameroon NMNH 83625, holotype of macrotis Miller, 1898.

Figure 9

Dorsal and ventral views of crania of Idiurus macrotis. From left to right: Medje, Democratic Republic of Congo AMNH 50542, holotype of langi Allen, 1922; Panga, DRC AMNH 50605, holotype of panga Allen, 1922; northwestern Lake Victoria, Tanzania NMW 31089.

Figure 10

Dorsal and ventral views of crania of Idiurus zenkeri. From left to right: Eshobi, northwestern Cameroon BMNH 48.885, holotype of haymani Verheyen, 1963; Yaoundé, southern Cameroon ZMB 7993, holotype of zenkeri Matschie, 1894; La Maboké, Central African Republic MNHN 1966-1946; Kashewe, Democratic Republic of Congo MRAC 30986.

Body Size and Skin Characters

A PCA for total length, tail length, ear length, and hind-foot length field measurements yielded no separation between I. macrotis and I. zenkeri apart from size differences result not shown. The statement of Kingdon 1997: 179 that the tail of I. macrotis “is proportionately shorter” than that of I. zenkeri was not supported by the data used in our analysis fig. 11. The tails of both species differ in absolute length tables 6, 7, but the relative length is about the same.

Figure 11

Bivariate plot of tail length versus total length in Idiurus macrotis N = 34 and I. zenkeri N = 22. All geographic samples combined.

Table 6

Mean Values and Ranges of Standard Body Measurements of Idiurus macrotis

(total length partially calculated from head and body plus tail length)

Table 7

Mean Values and Ranges of Standard Body Measurements of Idiurus zenkeri

(total length partially calculated from head and body plus tail length)

When logarithmized absolute measurements of I. macrotis as well as residuals for hind-foot length were used for the PCA, then specimens from all areas strongly overlapped. I. zenkeri was not represented by sufficient geographic samples, so no conclusions were obtained for this species.

No weight data were available for any of the 80 specimens of I. zenkeri examined. Kingdon 1997 gives a range of 14–17.5 g for this species and a range of 25–35 g for I. macrotis. Weight data for 14 museum specimens of I. macrotis table 6 range from 23 g to 40 g.

The coloration of the studied skins appeared very similar in both species, in contrast to other anomalurid species Schunke and Hutterer, 2005. In both species, the dorsal fur is grayish brown to dark brown, and the ventral fur is slightly paler. The skin of the limbs and the tail is lighter colored than the dorsal fur. I. macrotis specimens from Ghana seem to be lighter and more yellowish, but individuals from Ivory Coast are darker and more brownish, as are most specimens from Cameroon and the Congo. In I. zenkeri no differences could be found between populations, except for the holotype of I. zenkeri, which has an entirely blackish skin and dark brown fur.

Discussion

The taxonomic history of the genus Idiurus is brief. Matschie 1894 described the genus and species Idiurus zenkeri based on a single specimen from southern Cameroon fig. 5. Subsequently, Miller 1898 named I. macrotis from the same region based on differences in size and coloration “Much larger than Idiurus zenkeri Matschie; tail and ears relatively longer; color apparently darker; skull larger; bony palate narrower; second lower molar distinctly larger than first”. Lönnberg 1917 named Idiurus zenkeri kivuensis from Congo “These specimens agree much more in their general characteristics with Idiurus zenkeri Matschie than with I. macrotis Miller. Especially the cranial dimensions differ from those of I. macrotis”. The taxon was subsequently given species status by Hayman 1946 or considered a synonym of I. zenkeri Verheyen, 1963, while Dieterlen 1993 listed it under I. macrotis, a conclusion that we share. A second, much smaller specimen reported by Lönnberg was considered a juvenile but may well be a specimen of I. zenkeri. Unfortunately the skull of the holotype of kivuensis is lost Verheyen, 1963, and the skull of the smaller specimen is still inside the skin and not accessible. However, the general size of the holotype and the single skull measurement provided by Lönnberg 1917 led us to the conclusion that kivuensis is rather a synonym of macrotis than of zenkeri. Allen 1922 described I. langi and I. panga as new species from the northeastern Democratic Republic of Congo. I. langi was diagnosed as “[s]ize of and proportions nearly as in Idiurus macrotis Miller, but very different in coloration”, and I. panga as “[s]imilar to Idiurus macrotis Miller, but much smaller and considerably paler throughout, including the basal fur”. Hayman 1946, in a review of the genus, accepted I. macrotis and I. zenkeri as species, but also considered I. kivuensis a species “In my view the conclusion must be drawn that there are only three species of Idiurus, namely zenkeri, macrotis and a species intermediate in size between these two, for which the name kivuensis is available”. He also named a new subspecies I. kivuensis cansdalei “Differing from I. k. kivuensis in the pale general color above and below, and from I. k. panga in the general color of the hair tips above being between avellaneous and wood brown of Ridgway instead of light drab”. Finally, Verheyen 1963 named Idiurus zenkeri haymani from northwestern Cameroon: “Cette nouvelle sous-espèce se distingue de la forme typique par la longueur totale du corps, la longeur de l'oreille, la plus grande longeur du crâne et par la largeur bizygomatique nettement plus petites. … Il n'y pas de différence notable avec Id. zenkeri zenkeri quant aux caractères craniologiques et odontologiques ou de coloration du pelage”. Although most current authors, like Dieterlen 1993, accept only two species, the seven available taxon names indicate that some underlying genetic variability may exist.

Our analysis principally supports the current concept of the two species I. macrotis and I. zenkeri. Both species are mainly separated by size but occur sympatrically in at least two different areas west and east of the Congo Basin figs. 2–3 without intermediate forms.

Still, the results of the principal component analysis indicate the existence of morphologically distinguishable populations in I. macrotis, as we predicted from the patchy distribution of the species fig. 2. Remarkable is the result that northwestern Cameroon and southern Cameroon to Gabon seem to have clearly different populations each fig. 4. The former population matches the West African population in size but in shape is closer to specimens from southern Cameroon to Gabon. Animals from the Democratic Republic of Congo differ mainly in size from the West African and northwestern Cameroon populations but differ in shape from the southern Cameroon to Gabon population. Differences are largest between specimens from northwestern Cameroon and from the Democratic Republic of Congo, on one hand, and specimens from southern Cameroon to Gabon and West Africa, on the other hand. This pattern does not suggest a clinal variation. A single individual from northwestern Tanzania NMW 31089, fig. 9 has a very long and wide skull but is matched by a few other specimens. In the PCA, the specimen is closest to animals from southern Cameroon to Gabon in size and, particularly, in shape.

I. macrotis can thus be divided in five geographic populations: West Africa, northwestern Cameroon, southern Cameroon to Gabon, northeastern Democratic Republic of Congo, and Tanzania, by skull size and shape figs. 4, 8–9. The cranial differences are not matched, however, by differences in body size table 6 or coloration.

Results for I. zenkeri show a higher overlap between populations from Cameroon and the Democratic Republic of Congo. Specimens from northwestern Cameroon are less well separated than in I. macrotis fig. 5. Still, the populations can be separated with the PCA, and individuals are likely to be classified correctly with a discriminant analysis table 3. However, a clinal variation can not be excluded in this species, due to the intermediate position of animals from the Central African Republic. Further specimens, however, are necessary to test our results.

A subdivision as in I. macrotis, though more provisional, is possible in I. zenkeri. Animals from southern Cameroon plus Equatorial Guinea and the northeastern Democratic Republic of Congo can be recognized, and less so animals from northwestern Cameroon figs. 5, 10. The position of animals from the Central African Republic remains unsettled until further specimens become available. As in I. macrotis, body measurements and coloration provide no clue to geographical variation.

The case of the enigmatic holotype skin of I. zenkeri fig. 12 requires some explanation. In the series of 80 fairly uniform skins that we examined, the holotype is the only case where the fleshy parts of the skin have an equally blackish color. The skull fig. 10, however, groups nicely with other specimens from southern Cameroon and Equatorial Guinea fig. 5. We assume that the blackish coloration of the type skin in the Berlin Museum is an artifact. A similar case, also related to the Berlin Museum, is a mouse described by Dieterlen 1986 as Togomys melanoderma from older African collections and diagnosed as having “all unhaired parts of the skin deep black”. A subsequent inspection Dieterlen, 1989 of the animal revealed its identity with Asian Rattus exulans but with the skin stained black by some unknown chemical process.

Figure 12

Study skin of the holotype of Idiurus zenkeri ZMB 7993 showing the dark staining of the fleshy parts of the skin and the fur.

Another interesting aspect of our study is the shape difference between populations from southern Cameroon to Gabon and the Democratic Republic of Congo for both I. macrotis and I. zenkeri. Specimens were separated by the second principal component according to their geographic origin rather than according to species fig. 7. The factor had particularly high loadings for bulla size and rostrum length. Animals from Cameroon tended to have a longer rostrum, larger incisors, and smaller bullae, independent of species status. This finding may reflect unknown parallel adaptations in both species, for which we lack any supportive biological data.

Conclusions

Both species of Idiurus show a very uniform coloration throughout their range. Although some skins of I. macrotis from Ghana were slightly paler, no differences in skull size or shape between specimens from Ghana and other West African populations could be detected.

Our morphometric study has revealed some differences in both species at the population level, but we feel that a translation of these differences into named subspecies is not warranted at this moment. Size and shape differences do not follow clear patterns. For I. macrotis, size would produce clusters with specimens from West Africa, northwestern Cameroon, and, perhaps, southern Cameroon to Gabon and Tanzania. Individuals from Congo show intermediate values and overlap with the two main groups. If shape differences are taken, then the populations from southern Cameroon to Gabon are similar to West Africa and northwestern Cameroon, but different from the Congo sample. Only combined data distinguish between allopatric populations.

Populations of I. zenkeri from southern Cameroon and Equatorial Guinea and the Democratic Republic of Congo overlap, and the intermediate position of some specimens from Central African Republic suggest a clinal variation. Individuals from northwestern Cameroon are slightly smaller than those of other populations, but there are no differences in shape.

In summary, it is possible to distinguish between geographic populations of the two species, but neither size nor shape provides clear patterns. If one were to accept size differences as sufficient for a taxonomic division in one case, shape differences would have to be ignored, and vice versa.

Many of the specimens that we studied were collected almost a century ago. The specimens of both species obtained in 1910 by the American Museum Congo Expedition Allen, 1922 still belong to the best series available for systematic studies. The unique specimen from Tanzania fig. 9 was collected by Rudolf Grauer in 1905 in one of the forest islands in the northwest of Lake Victoria K. Bauer, personal commun.. It is unknown whether this forest and its Idiurus population still exist.

With such limited information available, we cannot decide whether the clusters in both species, for which we found evidence, represent populations or perhaps cryptic species. Future chromosome and genetic studies may shed further light on this widely unknown group of forest gliders. For the moment, we suggest the following taxonomic arrangement: