A survey of small mammals in the Volta Region of Ghana with comments on zoogeography and conservation

ABSTRACT We examined small mammal (insectivores, bats and rodents) diversity in community and legally protected forest remnants in the Ghana-Togo Highlands of the Volta Region of Ghana, West Africa, a zoologically understudied area compared to neighboring Togo to the East, or Ghana west of the Volta River. We recorded 34 small mammal species: three species of shrews (Soricidae Fischer, 1815), 12 species of rodents, one primate (Galagidae Gray, 1825) and 17 species of bats (Chiroptera Blumenbach, 1779). The rodent Stochomys longicaudatus (Tullberg, 1893) appears to be a first record for Ghana. Two shrew, three rodent and one bat species were first records for the Volta Region. By comparing our small mammal captures and limited microhabitat data from 1999 and 2001 to forest cover change maps for the period 2000-2015 we discuss trends in species community changes due to forest cover loss and other disturbance regimes. Aside from contributing to our understanding of the distribution of several small mammal species, the study demonstrates the progressive loss of forest habitat in the Volta Region.


INTRODUCTION
In the last 100 years wildlife abundance and diversity in West Africa have declined on a large scale (Oates 1999;Wilkie et al. 2011). This decline has greatly accelerated during the last two to three decades (Visconti et al. 2011;Habel et al. 2019) with dire projections (Tilman et al. 2017). Reasons for this decline are the widespread loss of original habitat due to uncontrolled logging, and species-poor monocultures (cocoa, oil-palms, rubber trees etc.), an increase of local subsistence farms, and increasing bushmeat hunting pressure driven by the growing urban population (Brashares et al. 2011;Greengrass 2015). These act in combination with an increasing frequency of uncontrolled fires aggravated by climate change to gradually diminish the remaining forest. In Ghana, the only remains of original forest are found in more remote and inaccessible areas of the forest reserves established between 1919 and 1939 (Oates 1999), in the three high forest National Parks (Bia NP, Kakum NP, and Ankasa Conservation Areas), and in the so-called "sacred groves" -traditionally protected and locally maintained forest patches ranging in size from a few trees to several hundred hectares. Previous studies of small mammal ecology and conservation in Ghana indicated that these traditionally protected sites contained higher small mammal biomass than surrounding areas and that they function as refuges for some small mammal species no longer found anywhere else (Yeboah 1984;Decher 1997b;Decher & Bahian 1999).
In the Volta Region the condition of the forest reserves is especially serious. The forest has been greatly reduced by extensive deforestation for cocoa plantations, rice cultivation and shifting agriculture, accelerated by frequent dry season fires and a relatively dense human population ranging from 26-100 persons per km2 (Bakarr et al. 2001). A review of forest conservation in Ghana attributed the widespread deforestation in the Volta Region reserves to "…the high demand for farmland, associated partly with movement of dispossessed people before the damming of the Volta Lake in 1966." (Hawthorne 2001: 493).
The Ghana-Togo Highlands, including the Fazao-Malfakassa Wildlife Reserve in Togo, have been classified as an area with "exceptionally high" conservation priority for all organisms and "very high" conservation priority for mammals. This assessment was based on species richness, species endemism, rare and endangered species, and critical habitats (Bakarr et al. 2001). Zoogeographically much of the Volta Region belongs to the so-called Dahomey Gap, the interruption of the forest belt in West Africa stretching from Accra, west of the Volta eastward through Benin, formerly called "Dahomey". For a discussion of the Dahomey Gap phenomenon see Booth (1954Booth ( , 1958, Robbins (1978) and Decher et al. (1997). A number of species unique to the Dahomey Gap were recently described, including the Dahomey Gap Wood Mouse, Hylomyscus pamfi Nicolas, Olayemi, Wendelen & Colyn, 2010 and Walter's Duiker, Philantomba walteri Colyn, Hulselmans, Sonet, Oudé, de Winter, Natta, Nagy & Verheyen, 2010(Colyn et al. 2010Nicolas et al. 2010b).
There are a number of pre-World War I records from the Volta Region when it was still part of German colonial "Togoland". This includes the historical locality "Bismarckburg", type locality of the famous "lost" Büttner's African Forest Mouse Leimacomys buettneri Matschie, 1893b -one of the declared targets of this survey -just a few miles east of today's Ghana-Togo border. Only two specimens of Leimacomys Matschie, 1893 were preserved at the Zoological Museum Berlin (ZMB) and were subsequently described as a new species (Matschie 1893b), but they remain the only known specimens today. Several studies have established that Leimacomys feeds mostly on insects (Misonne 1966;Dieterlen 1976;Denys 1993), and that its short tail suggests a terrestrial rather than arboreal life style (Dieterlen 1976;D. Kock, pers. comm.). A taxonomic enigma, it has been classified in the subfamilies Murinae Illiger, 1811(e.g. Simpson 1945) or Dendromurinae G. M. Allen , 1939(e.g. Rosevear 1969) until Musser & Carleton (2005) placed Leimacomys in its own subfamily Leimacomyinae Musser & Carleton, 2005 within the family Muridae Illiger, 1811, deeming it sufficiently different from all other muroid rodents. This forest endemic represents an evolutionary lineage that has been long isolated from other rodents (Denys 1993).The only previous "modern" mammal survey in the Volta Region, which included ten sites, was that of the US Smithsonian Institution's African Mammal Project expeditions to Ghana (between 14 July 1967 and19 August 1969;Schmidt et al. 2008). The present study examines small mammal communities in different forest remnants in the Volta Region of Ghana with the following objectives: 1) present a snapshot of the diversity and abundance of small mammal species (shrews, bats, and rodents) in forest remnants along the Togo border in the Volta Region of Ghana based on our 1999 and 2001 capture data; 2) infer subsequent loss of small mammal habitat based on high-resolution 21st century forest cover change maps (Hansen et al. 2013); and 3) provide national and international conservation managers with new data and recommendations to improve, or initiate protection of the remaining forest and wildlife sites in the Volta Region. Despite some researchers arguing that small mammals may receive adequate protection in reserves set aside for the protection of larger or so-called "flagship species" such as forest elephant, forest buffalo or large antelopes, a recent critique argued that, similar to commercially branded goods and services, focusing on conservation flagships can lead to standards of comparison that may decrease the attractiveness and public acceptance of non-flagship species (Douglas & Winkel 2014). Also, with few exceptions in the Volta Region, such as Kalakpa Resource Reserve, Kyabobo National Park, or the Tafi Atome Monkey Sanctuary (Ormsby & Edelman 2010), forest remnants and sacred groves in the Volta Region may not be large enough and the traditional taboos may not be strong enough to sustain and protect larger mammals, but they may still harbor a diverse medium and small mammal fauna. One review of mammal conservation priorities also concluded, that "…evidence indicates that threats to small mammals are as extensive as those faced by large mammals…" and "[i]t appears that the needs of small mammals, in relation to the degree of threat, are not being met by the current conservation agenda" (Entwistle & Stephenson 2000: 137).

Study area
Our study area encompassed large parts of the Volta Region of Ghana. This region is part of the Ghana-Togo Highlands and part of the Akwapim-Togo escarpment (Fig. 1B). In Ghana maximum elevation reaches 955 m at Mount Afadjato (Gnielinski 1986) and 986 m at Mount Agou in Togo (IGN Togo Map 1: 500 000). For the purposes of this paper we define the Ghana-Togo Highlands as extending from Akwamufie in the Eastern Region (east bank of lower Volta, near the Volta Dam) north-eastward, including Kalakpa Resource Reserve, Mount Adaklu and Mount Agou in Togo. From there it extends northward to Atakpamé and along the line Sotouboua-Sokodé, including the Faille d'Alédjo region; westward to Bassar, then south to include the Fazao Malfakassa Wildlife Reserve and the new Kyabobo National Park in Ghana, then along the line Pawa -Nkwanta -Worawora -Kpandu back to Akwamufie (Fig. 1). Most of the Ghanaian part of this region is covered by dry semi-deciduous forest of the "fire zone subtype" (Hall & Swaine 1981: 17). For this study, terrestrial and volant small mammals (shrews, and rodents and bats) were captured in 1999 and 2001 at the following six sites in the Volta Region from south to north: 1) Kalakpa Resource Reserve ( Fig. 2A) Except for site 1 (Kalakpa RR), some microhabitat characterizations based on trap site measurements for sites 2-6 can be found in Table 1.

Field techniqueS
Survey techniques followed those published by Voss & Emmons (1996), standard methods for measuring and monitoring mammal diversity in Wilson et al. (1996), and complied with guidelines approved by the American Society of Mammalogists (Animal Care and Use Committee 1998) and the Institutional Animal Care and Use Committee (IACUC) of the University of Vermont. Trapping effort varied somewhat for the six sites visited (Tables 2; 3). Only three of the sites could be visited in both years. Kyabobo Range National Park (KRNP) was visited in both years, but the 1999 site in the park (Odome) was replaced by another site (Laboum Creek) in 2001 because it yielded no captures due to recent burning. We used up to 142 traps at each site: 100 standard (LFA) and one large (XLF15) Sherman live traps, seven large and two medium Tomahawk live traps, 20 Museum Special snap traps, and 12 Victor rat traps. Where possible, we also employed lines of pitfall traps using local plastic buckets connected by plastic drift fences in each habitat. Bats were captured using standard 2.6 to 12 metre long mist nets (50 denier/2 ply, 38 mm mesh; Avinet Inc.) at ground level on forest paths, at forest edges and across small streams mostly during evening hours and sometimes all night long. Species were identified, sexed and representative vouchers were kept for all species and of all individuals with uncertain identification. Sex is either denoted as male (♂), female (♀) or unknown (?) in the following list. Preliminary field identification of terrestrial small mammals and bats was based on Rosevear (1965Rosevear ( , 1969, Meester & Setzer (1971), Hutterer & Happold (1983 and Happold (1987) and subsequently updated also using more recent sources cited in the species accounts below. Bat and some rodent voucher specimens were identified and deposited at the Senckenberg Museum Information on some of the larger arboreal and diurnal species (squirrels, anomalures, primates, small carnivores) was obtained by interviewing local people and by slow observa-tion walks (Emmons 1980) using binoculars (Zeiss Classic 10 × 40) and a small digital video camera (Sony DCR-PC1). In 2001 we also recorded microhabitat data on standardized habitat data sheets.

Molecular identiFication
To facilitate identification, we amplified and sequenced a fragment of the cytochrome b (Cytb) gene for 10 individuals from two taxa. DNA was extracted from a small (5-10 mg) piece of liver or kidney that had been stored in 95% ethanol. Ethanol was removed via soaking in distilled water (Kilpatrick 2002) prior to extraction using a Gentra Puregene Mouse Tail Kit (QIAGEN -Germantown, Maryland) according to manufacturer's instructions.
A 401 bp fragment of the Cytb mitochondrial gene was amplified with PCR using the primers CytbA and CytbE (Sullivan et al. 1997) using PuRe Taq Ready-To-Go PCR Beads (GE Healthcare Life Sciences). Amplification occurred over 35 cycles with denaturation at 94° C for 1 min, annealing at 50° C for 1 min and extension at 72° C for 1 min. Sanger sequencing was performed using both forward and reverse primers at the DNA Analysis Facility, Vermont Integrative Genomics Resource, University of Vermont. Resulting sequences were compared to previously published results using BLASTn for nucleotides (Altschul et al. 1990). We have deposited our sequences that were either new or from different localities into GenBank under accession numbers MT311307-MT311314).

data analySiS
Trap success was calculated by dividing the number of captured small mammals of each site by the no. of trapnights (= nights trapped × no. of traps) and multiplying by 100 (Table 2). Specimens brought to us by local people were excluded from trap success. For bats the distribution of captures, netting effort and number of bats per net night and net unit for the six localities visited in 1999 and 2001 (Table 3). Basic statistics (mean, standard deviation) for microhabitat data were generated using SPSS, Version 24 (IBM 2016). Conservation status of each species was taken from the current version of the IUCN Red List (IUCN 2020). Species accumulation curves, diversity and similarity indices were generated using the program EstimateS (Colwell 2013). Although it is perhaps somewhat unorthodox to use the species richness predictor EstimateS for such a large area of the Volta Region, we wanted to facilitate a rough species richness comparison with a previous study on the Accra Plains, also focussing on forest remnants (Decher 1997a, b;Decher & Bahian 1999), and studies in other areas in Ghana (Decher & Fahr 2005). Our sites across the Volta Region were selected to represent the whole partially forested area. All were supposed to be surveyed twice in two different seasons. In spite of some logistical challenges -some sites had to be replaced or added in year 2 -we argue that it is one large study area focussing on at least partially protected forest remnants in the Volta Region.

RESULTS
During a total of 26 days of fieldwork covering the 1999 and 2001 expeditions, we captured 308 individual small mammals from 33 species (Tables 2; 3).

reMark
This species was previously known from several localities in the Volta Region (Grubb et al. 1998). In their Kyabobo expedition report Hurst et al. (1995) mention several individuals of this large shrew (as C. flavescens) from KRNP, but we did not encounter it there. Our specimen from Liati Wote (ZFMK 2003(ZFMK .1090) was included in a recent phylogeographic study of C. olivieri (Jacquet et al. 2015) and grouped with their Clade IIB with Dahomey Gap specimens from Togo and Benin.
conServation StatuS. -Crocidura olivieri is listed as "Least Concern" on the IUCN Red List.  No. of bats per net unit 3 9.2 9.2 11.5 11.5 7.9 -9.2 6.9 6.9-11.5 14.8 6.9 -- reMark Initially identified as Crocidura cf. douceti Heim de Balsac, 1958, this tiny shrew has been re-assigned to Crocidura eburnea, which has only recently been elevated from a synonym of C. obscurior (Hutterer 2005) to species level (Jacquet et al. 2014). This is the first record of C. cf. eburnea from the Volta Region and since the Volta River is considered the eastern boundary of the C. obscurior complex (Jacquet et al. 2014), this possible range extension awaits further study.       conServation StatuS. -Anomalurus beecrofti is listed as "Least Concern" on the IUCN Red List, however, progressive forest fragmentation with the loss of mature trees, specifically oil palms (Elaeis guineensis), will greatly compromise the habitat of this treedependent glider.   [1988,1989] and Grubb et al. [1998]) but it seems to occur more commonly in or near forest remnants than the latter species, suggesting a greater dependence on forest. In our survey up to eight individuals would hit the nets at the same time indicating that they fly and forage in groups.
conServation StatuS. -Micropteropus pusillus was classified as "Least Concern" on the IUCN Red List. It remains a common fruit bat species in the Ghana-Togo Highlands. However, this species may depend on forest remnants during its foraging flights (Fahr 1996). (Thomas, 1908) Rousettus smithi Thomas, 1908: De Vree et al. 1969, 1970Robbins, 1980;Grubb et al. 1998). Myonycteris a. smithii appears to be a relatively common forest and forest edge species throughout the Ghana-Togo Highlands and does not appear too affected by the widespread forest fragmentation. Most specimens in Ghana, Togo and Côte d'Ivoire have been recorded from the forest savanna mosaic and dry forests, but only a few from evergreen rain forest (Fahr 1996). We captured no males and only one female  reMark Previous records of this species were from Odomi Jongo in the Volta Region (USNM) and from Bismarckburg, Misahohé, Aledjo, Atakpamé, Evou, Ezimé, Fazao, Odjolo, Plateau Akpossi in Togo (De Vree et al. 1969, 1970De Vree & Van der Straeten 1971). This is a small fruit bat similar in appearance to Micropteropus pusillus from which it can be distinguished by its slender snout and the nine undivided and thinner palatal ridges (see drawings in Bergmans [1997] and Happold [1987]). With just two specimens captured at the beginning of dry season in 1999, N. veldkampi was much less common than M. pusillus (23 specimens; Table 2). The seasonal occurrence could be explained by the migratory behavior of this species, which was shown to follow the progression of the rains northward to savanna areas (Fahr 1996, Thomas 1983. Similarly, at Mount Nimba, Wolton et al. (1982) did not obtain this species at all between early July and early September, whereas it was common there at other times of the year (Monadjem et al. 2016). During the African Small Mammal Project, between January and June 1968 (Robbins 1980), no N. veldkampi were encountered, yet the same Project captured 20 M. pusillus in Togo and Benin. During a study on the Accra Plains between November 1991 and June 1992 no N. veldkampi were encountered, but 45 M. pusillus were captured (Decher 1997a).

Myonycteris angolensis smithii
conServation StatuS. -Nanonyteris veldkampi is classified as "Least Concern" on the IUCN Red List. Its migratory behavior and dependence on forest remnants still need to be investigated in more detail in the Ghana-Togo Highlands.

Hypsignathus monstrosus H. Allen, 1861
Hypsignathus monstrosus H. Allen, 1861: 157.  (1985) apparently list these specimens as being from "Akenim", Togo. In Côte d'Ivoire, Fahr (1996) recorded six out of seven specimens of N. arge within the rainforest zone and one in forest-savanna mosaic. Our specimen of N. arge appears to be the first one reported from the Volta Region.
conServation StatuS. -Although classified as a species of "Least Concern" on the IUCN Red List, as a mostly forest-dependent species, N. arge is of some conservation concern in the Ghana-Togo Highlands.
conServation StatuS. -Listed as "Least Concern" on the IUCN Red List, this species was the most common rhinolophid bat in our survey and is probably of lesser conservation concern in the Ghana-Togo Highlands.

Rhinolophus landeri Martin, 1838
Rhinolophus landeri Martin, 1838: 101.  reMark The single specimen of R. landeri could be distinguished from R. alcyone by its smaller size (weight 7.8 g; forearm 43.2 mm). A British Museum specimen of R. landeri is labeled as originating from "Wraura" (BMNH 55.378, leg. A. H. Booth), which may be the same as Worawora (near Apesokubi). A review of the bats of Côte d'Ivoire, showed that R. landeri occurs in all savanna formations to the northern edge of the Sudan savanna and that rainforest is actually being avoided (Fahr 1996;Fahr & Kalko 2011), contrary to Rosevear's (1965) assessment of R. landeri as a rainforest species, which at his time included R. guineensis Eisentraut, 1960 as a subspecies. This is also supported by captures from Togo with just one specimen from the eastern edge of the Ghana-Togo Highlands at Atakpamé (De Vree et al. 1969) and 22 specimens caught in Northern Togo at Namoundjoga (De Vree et al. 1970). On the Accra Plains nine R. landeri were caught in forest remnants and in more open savanna (Decher 1997a).
conServation StatuS. -R. landeri is listed as "Least Concern" by IUCN Red List. However, its uncommon occurrence in the Ghana-Togo Highlands and the fact that we found just one specimen in a forested valley make it of some conservation concern for the Volta Region.
Family hippoSideridae Lydekker, 1891 Hipposideros cf. ruber (Noack, 1893) (Fig. 11 bat uses a perch-hunter foraging strategy and it requires large hollow trees for its day roosts. On the Accra Plains this species was found only in a traditionally protected sacred grove that was a remnant of high forest (Decher 1997a). In Côte d'Ivoire 70% of localites were in the area of moist forests, 10% in forestsavanna mosaic and 20% in savanna formations (Fahr 1996).
conServation StatuS. -Listed as "Least Concern" on the IUCN Red List, its dependence on large and hollow trees and preference for gallery forest (Decher & Fahr 2005) make this a species of high conservation concern in the Ghana-Togo Highlands.
Macronycteris gigas (Wagner, 1845) De Vree et al. 1969;Robbins 1980). Our record confirms previous observations (Grubb et al. 1998) that in Ghana this species often occurs in forests, although it is also found in Guinea woodland and even in the coastal savanna like the Accra Plains (Decher 1997a). In Côte d'Ivoire, Fahr (1996) made six of his eight captures in the rain forest zone. Six ectoparasitic Ascodipteron variisetosum Maa, 1965 [Diptera: Streblidae] with (2 × 3) were found on the ventral side of the upper arm on the specimen from Agumatsa.
conServation StatuS. -Listed as "Least Concern" on the IUCN Red List, this species is of some conservation concern in the Ghana-Togo Highlands because it appears to be patchily distributed and may be limited not so much by the availability of large tracts of forest as by the presence of large hollow trees and caves for its roosts (Grubb et al. 1998 Robbins et al. 1985). The species was also caught at three locations on the Accra Plains (Decher 1997a and USNM, as S. dinganii) and seems to be most commonly associated with Guinea savanna, forest savanna mosaic, and high forest edge. Records from Côte d'Ivoire are all from the northern tree savanna (Fahr 1996, as S. dinganii). This species also seems to adapt to rooftops and thatched huts for its roosts.
conServation StatuS. -The conservation classification of S. livingstonii on the IUCN Red List is "Least Concern". Thomas, 1904 (Fig. 13) Myotis bocagei cupreolus Thomas, 1904 (Hurst et al. 1995) and the present study attempted to rediscover the enigmatic rodent species Leimacomys buettneri Matschie, 1893 on the Ghana side, but so far without success. Leimacomys was first collected by R. Büttner or his local assistant in 1890 at Bismarckburg, which was then the German colonial district headquarter located near the modern town of Yegué, about 20 km east of KRNP in Togo located at 08°11'06"N, 00°38'57"E, 707 m above sea level and only 5 km from the Ghanaian border (Krell 1994).

Myotis bocagii cupreolus
Local Wildlife Division field staff at Kyabobo National Park who were shown pictures of Leimacomys claimed they knew the species, but we did not succeed in obtaining additional specimens. Similarly, at Shiare, which is even closer to the original Togo locality, people seemed to know this rodent from our description by the local name "Yefuli" or "Yiefuni". Attempts to obtain Leimacomys in the hills near Shiare have so far been unsuccessful. Future searches will have to include further probing of the indigenous knowledge of this species, various trapping methods, and local participation to eventually rediscover it on the Ghana or Togo side of the border.

volta reGion terrreStrial SMall MaMMal richneSS and diverSity
Twenty-four nights of trapping in the Volta Region, amounting to 2399 trap nights (two field seasons) resulted in a total of 247 individuals from 16 species captured (Table 1). The species extrapolation curve calculated in EstimateS 9 (Colwell et al. 2012;Colwell 2013) and the cumulative number of species over the individuals captured show no asymptotic levelling off (Fig. 14A). The maximum value of the classic (non bias-corrected) Chao 1 richness estimator, defined as the "absolute number of species in an assemblage," is a function of the ratio of species represented by a single individual (F1 = singletons) and species represented by only two individuals (F2 = doubletons) (Magurran 2004): SChao1 = Sobs + (F12 / 2*F2). In this case: SChao1 = 17 + (82 / 2*3) = 27.7 species. This shows that our two-season sampling in the Volta Region did not get close to the maximum number of terrestrial small mammal species to be expected in the Volta Region. With 39.3%, capture success was highest during the one night of trapping at Tagbo Falls, but with just two species caught, diversity (Simpson's 1 -D = 0.12) was lowest. The locally managed Bedibem Forest at Apesokubi yielded more terrestrial small mammal species (11) than any of the other sites with several species only encountered there.

volta reGion bat richneSS and diverSity
In 18 nights of bat netting in the Volta Region we captured 63 individuals from 17 species (Table 2). For microbats the Agumatsa Valley (Wli) topped the list with six species caught, of which Rhinolophus landeri, Macronycteris gigas, Scotophilus livingstonii and Myotis bocagii were caught only at Agumatsa. The only nycterid, Nycteris arge, was caught at Apesokubi and Doryrhina cyclops was only caught at Shiare. The most widely distributed bats were Micropteropus pusillus (four localities) and Myonycteris angolensis (four localities). Figure 3B shows the species extrapolation curve calculated in EstimateS 9 (Colwell et al. 2012;Colwell 2013), the cumulative number of species over the individuals captured, and the maximum value of the classic (non bias-corrected) Chao 1 richness estimator as defined above (Magurran 2004): SChao1 = Sobs + (F12 / 2*F2). In the case of bats (see Table 2): SChao1 = 17 + (82 / 2*2) = 33 species. This indicates that our two-season sampling of bats in the Volta Region was not an exhaustive sampling of the bat species to be expected in the region. coMMentS on the zooGeoGraphy oF the Ghana-toGo hiGhlandS Our results show that the Ghana-Togo Highlands are a crossroads of Upper Guinea and Lower Guinea elements and highlight the presence of species and intraspecfic clades that are restricted to the Dahomey Gap Region. Several of our forest taxa could be categorized into at least one of the following groups: 1) endemic to the Dahomey Gap region; 2) allied to Upper Guinea forests; 3) allied to Lower Guinea and Central African forests; 4) found throughout West Af-rican forests -defined here as closed-canopy forest from the Senegal-Guinea border to the Sanaga River in Cameroon -; and 5) tolerant of savanna and exhibiting weaker zoogeographic patterns.
One of our captured species (Hylomyscus pamfi) is endemic to the Togo Highlands and Dahomey Gap region, while three exhibit intraspecific patterns wherein populations from the region form a distinct clade. Crocidura olivieri from the region is distinct at an intraspecific level as Clade II-B in the Dahomey Gap and the Sudanian savanna zone of Benin, Togo, Burkina Faso and Niger (Jacquet et al. 2015). Nicolas et al. (2011) recovered Praomys misonnei from the Dahomey Gap region as a unique clade (II). Malacomys edwardsi from the area is unique at an intraspecific level as lineage F2, separate from south-central Ghanaian lineage F1 (Bohoussou et al. 2015).
Five forest taxa exhibit affinities to Upper Guinea. The aforementioned Malacomys edwardsi is an otherwise Upper Guinea species. Crocidura eburnea is part of the Upper Guinean C. obscurior complex (Jacquet et al. 2014). Epomops buettikoferi is an Upper Guinean fruit bat with a few populations reaching Lower Guinea. Megaloglossus azagnyi is a newly described Upper Guinean nectar-feeding fruit bat with as yet unknown ability to penetrate the Dahomey Gap (Nesi et al. 2013). Finally, according to Nicolas et al. (2010b;2020), the sister species to Hylomyscus pamfi is H. simus, which is an Upper Guinea endemic.
Four forest taxa are primarily allied with Lower Guinea. Stochomys longicaudatus is the Lower Guinea "replacement" of Dephomys defua. Epomops franqueti is an essentially Lower Guinean/Central Africa fruit bat that has been able to penetrate the Upper Guinea forest block just to western Côte d'Ivoire.
Praomys missonei is a species of Lower Guinea and Central African forests. Finally, although Crocidura olivieri is known from both sides of the Dahomey Gap, Jacquet et al. (2015) showed that populations from the Dahomey Gap region are more closely related to Lower Guinea populations than those in Upper Guinea.
Several forest species are found across West Africa, but their distributions are interrupted by the arid Dahomey Gap. These include Graphiurus nagtglasii, Rhinolophus alcyone, Doryrhina cyclops, and Myotis bocagii. None have been the subject of detailed phylogeographic study, although, in spite of limited geographic sampling, the results of Patterson et al. (2019) might suggest that a strong geographic pattern in M. bocagii is unlikely. For D. cyclops mitochondrial results for Afrotropical Hipposideridae clearly separate Upper Guinean (Liberia and Senegal) from central African specimens (Patterson et al. 2020) In contrast to forest species, savanna-tolerant species appear to occur more or less across arid West Africa including the Dahomey Gap. These include Crocidura foxi, Cricetomys gambianus, Gerbilliscus kempi, Lemniscomys striatus, Mus musculoides, and Epomophorus gambianus. Clearly the region's forest-dwelling terrestrial mammals are more easily isolated by zoogeographic barriers such as the Volta River or stretches of unsuitable habitat when compared to most savanna or bat species.
uncertain Future oF SMall MaMMalS in the volta reGion oF Ghana Our maps of the Volta Region of Ghana and the Togo Highlands (Fig. 1) illustrate the extent of the forest cover and the location of our study sites (Fig. 1A) and the gen-  eral topography, overlaid with red pixel clusters (Fig. 1B) illustrating the forest cover loss for the 18-year period from 2000 to 2018, thus mostly in the period since our two data collection events (1999 & 2001). This forest cover loss is based on online earth observation satellite data (Hansen et al. 2013). Another source reports that Ghana's dense evergreen rain forest and moist deciduous forest "shows a small decline in area from about 16 400 sq km in 1975 to 15 500 sq km in 2000, a reduction of 5 percent. This decline accelerated rapidly between 2000 and 2013, as forests were reduced by an additional 20 percent, to 12 400 sq km in 2013" (CILSS 2016).
Most of the mapped forest cover loss is in addition to the signs of forest loss observed during our study periods, as exemplified by our Figure 3C at Wli Waterfall, bare stretches of fragmented forests on hillsides east of Apesokubi (Kabo River Forest Reserve) and encounters with burned areas and active bushfires during our dry season expedition in 1999. Both the legally protected and savanna-dominated study sites, Kalakpa Resource Reserve and Kyabobo National Park are highly susceptible to the spread of bush fires in the dry season. For the years following our study, Dowsett-Lemaire & Dowsett (2011a: 25) report that "Most of the savanna sections burn every dry season, more or less severely. The wide galleries of forest crossing Kalakpa from north to south need protection through a policy of early burning". Dowsett-Lemaire & Dowsett (2011b: 15) also documented "…a lack of coordination between farmers on the one hand, and those members of the communities who wish to develop ecotourism and protection" both at Wli and Tagbo Falls. Further alarming recent data from the World Resources Institute (WRI) document that "Ghana and Côte d'Ivoire experienced the highest percent rise in primary forest loss between 2017 and 2018 of any tropical country (60% and 26%, respectively)" with illegal mining and expansion of cocoa farms causing some of the loss (Weisse & Goldman 2019). While forest loss is not as rapid in the Volta Region as in southwestern and central Ghana based on Global Forest Watch data ( Fig. 1) even in the current view (https:// www.globalforestwatch.org/map), there is concern for the future of the mammal fauna of the region, an urgent need to strengthen the conservation function of protected areas like Kalakpa Resource Reserve and Kyabobo NP, but also a need to strengthen local efforts to manage and preserve community forests and sacred groves.
Our Volta Region survey has shown how much of a Dahomey Gap and Lower Guinean faunal element is already present in the Volta Region part of the Togo Highlands, with species like Hylomyscus pamfi, Praomys misonnei and Stochomys longicaudatus. Some details were only unraveled in recent years through molecular phylogeographic studies on various rodent and bat genera (Nicolas et al. 2010(Nicolas et al. a, b, 2020Olayemi et al. 2012;Nesi et al. 2013). Clearly the Volta River is an important geographic barrier between Upper Guinea and the Dahomey Gap, particularly for rodents (Nicolas et al. 2011). Appendix 1 summarizes these findings.