Addressing the Vepris verdoorniana complex (Rutaceae) in West Africa, with two new species

Vepris verdoorniana (Rutaceae) has long been regarded as a widespread and variable species occurring from Guinea to Gabon. Recent evidence has shown that the original material from the type locality in Cameroon consists of two different species, V. verdoorniana in the strict sense, endemic to Lower Guinea (Cameroon, Gabon and C.A.R. in W-C Africa) and V. letouzeyi Onana, which was thought to occur from Guinea to Cameroon. In this paper we show that the West African (Upper Guinea) material previously referred to as V. letouzeyi/V. verdoorniana is separate morphologically from that in Cameroon (Lower Guinea). In fact the West African material referred to comprises two distinct species, here described as V. occidentalis Cheek & Onana (Guinea to Ghana, but probably extending to Benin) a tree of lowland marginal dry evergreen and semi-deciduous forest, and V. fer Cheek, a species restricted to four mainly lower submontane forest locations on iron substrates in Liberia and Sierra Leone (also Upper Guinea). Vepris occidentalis is here assessed as Least Concern due to its wide range and high frequency, whilst V. fer is assessed as Endangered due to the low number of locations and threats from iron ore mining projects. Both new species are illustrated and mapped.


Introduction
Vepris Comm. ex A.Juss. (Rutaceae-Toddalieae), is a genus with 86 accepted species, 22 in Madagascar and 65 in Continental Africa with one species extending to Arabia and another endemic to India (Plants of the World Online, downloaded 25 May 2018). Four new species were recently described from Cameroon (Onana & Chevillotte 2015;Cheek et al. 2018a) and one more is in press (Onana et al. 2019), taking the total in Cameroon to 22 species, the joint highest number for any country. Many of these 22 species are endemic to western Cameroon (SouthWest Region and NorthWest Regions of Cameroon) and several are threatened (Onana & Cheek 2011). Western Cameroon is the most species-diverse area for vascular plants per degree square in tropical Africa and contains several refugia areas (Barthlott et al. 1996;Cheek et al. 2001).
In continental Africa, Vepris are easily recognised. They differ from all other Rutaceae because they have digitately (1 -) 3 (-5)-foliolate (not pinnate) leaves, and unarmed (not spiny) stems. The genus consists of evergreen shrubs and trees, predominantly of tropical lowland evergreen forest, but with some species extending into submontane forests and some into drier forests and woodland. Vepris species are often indicators of good quality, relatively undisturbed evergreen forest since they are not pioneers (Cheek, Onana pers. obs. 1992. Species of Vepris in Africa extend from South Africa, e.g. Vepris natalensis (Sond.) Mziray, to the Guinean woodland in the fringes of the Sahara desert (V. heterophylla (Engl.) Letouzey). Mziray (1992) subsumed the genera Araliopsis Engl., Diphasia Pierre, Diphasiopsis Mendonça, Oricia Pierre, Teclea Delile, and Toddaliopsis Engl. into Vepris, although several species were only formally transferred subsequently (e. g. Cheek et al. 2009;Onana & Chevillotte 2015). Mziray's conclusions were confirmed by the molecular phylogenetic studies of Morton (2017) but sampling was limited, identifications appeared problematic (several species appear simultaneously in different parts of the phylogenetic trees) and more molecular work would be desirable. Morton studied about 14 taxa of Vepris, mainly those from eastern Africa. Nevertheless, characteristics of some of the former genera are useful today in grouping species. The "araliopsoid" species have subglobose, 4-locular fruit with 4 external grooves; the "oricioid" species are apocarpous in fruit; the fruits of "diphasioid" species are laterally compressed in one plane, bilocular and bilobed at the apex; while "tecleoid" species are unilocular in fruit and 1-seeded, lacking external lobes or grooves. There is limited support for these groupings in Morton's study, Due to the essential oils distributed in their leaves, and the alkaloids distributed in their roots, several species of Vepris have traditional medicinal value (Burkill 1997 (Burkill 1997: 651 -653). Research into the characterisation and antimicrobial and anti-malarial applications of alkaloid and limonoid compounds in Vepris is active and ongoing, although sometimes published under generic names no longer in current use, e.g. Wansi et al. (2008). Applications include as synergists for insecticides (Langat 2011). Cheplogoi et al. (2008) and Imbenzi et al. (2014) respectively list 14 and 15 species of Vepris that have been studied for such compounds.
The most recent revision of the African Toddalieae was by Verdoorn (1926), who recognised 49 taxa in the genera now included in Vepris. Her revision underpinned subsequent Flora accounts. In the account for Conspectus Florae Angolensis, Exell & Mendonça (1951) evidenced that she had mistakenly attributed the name Teclea grandifolia Engl. (now Vepris grandifolia (Engl.) Mziray) for which the type is Angolan, to another, undescribed species diagnosed by pedicellate flowers and axillary inflorescences, for which they published the new name T. verdoorniana Exell & Mendonça. Verdoorn (1926) had cited three specimens for her "Teclea grandifolia" from Cameroon (Staudt 472, 493 and Mildbraed 9234) and one from Sierra Leone (Barter s.n.). Exell & Mendonça (1951) selected one of these, Staudt 472 (K) as type. In the second edition of Flora of West Tropical Africa,  recognised three species of Teclea in West Africa, separating T. verdoorniana from T. afzelii Engl. and T. sudanica A.Chev. (now V. heterophylla) by the possession of inflorescences both axillary and terminal, hairy, to 4 cm long, (vs all axillary, glabrous, to 2.5 cm long), petioles often winged and leaflets sessile (vs non-winged, leaflets petiolulate). Letouzey followed this delimitation in Flore du Cameroun (1963) and pointed out that the Mildbraed specimen is from C. A.R. Onana & Chevillotte (2015) showed that since Verdoorn (1926), authors had unwittingly included more than one species in Vepris verdoorniana. Onana & Chevillotte (2015) linked Staudt 493, which has winged, not terete petioles, with additional specimens from Southwest Region Cameroon and described them as V. letouzeyi Onana, showing that it is a diphasoid, not a tecleoid species. They gave V. letouzeyi the range formerly attributed to V. verdoorniana, i.e. Guinea to Gabon, that is occurring in both Upper Guinea (Guinea to Ghana) and Lower Guinea (Nigeria to Congo Republic).
These two phytogeographic units are separated by the Dahomey Gap (see e.g. Demenou et al. 2017). Vepris verdoorniana, differentiated among other characters by terete petioles and based on the type Staudt 472, was now shown to be restricted to the Cameroon coastal plain (Onana & Chevillotte 2015), Bioko, Gabon and C. A.R. However, comparison of the West African (that is, Upper Guinea) material attributed to, or attributable to, Vepris verdoorniana by the definition of  shows that it is morphologically separate from both V. verdoorniana and V. letouzeyi of Lower Guinea. The Upper Guinean (Guinea to Ghana, possibly Benin) winged-petiole material considered to be V. letouzeyi in Onana & Chevillotte (2015) is a different taxon, named below as V. occidentalis. Further, a second undescribed taxon is present in Upper Guinea, which had also been included in Teclea verdoorniana by . The species represented by Barker 1181 (K!) has opposite leaves, nearly glabrous inflorescences and very large, pitted fruits. It occurs in the Gola Forest of Sierra Leone near the border with Liberia and in Liberia at the southern end of Mt Nimba and in the Putu Hills. This species is here termed V. fer Cheek.
Here we present a key and table of characters (Table 1) separating the taxa of the Vepris verdoorniana complex based largely on morphology. We also present a key and table of characters (Table 2) to all the tecloid (unilocular) West African (Upper Guinean) species and present a synopsis of the species with descriptions, illustrations and conservation assessment of the two new species to science.
Approximately 2000 new flowering plant species are described each year (Willis 2017), adding to the estimated 369,000 already known to science (Nic Lughadha et al. 2016) although this total is disputed . Evidence based conservation assessments exist for about 21 -26% of these species, and 30 -44% of these assessments rate the species concerned as threatened (Bachman et al. 2018). Widespread species tend to have already been discovered, so that many newly discovered species are those that are range-restricted and so more likely to be threatened, e.g. Vepris bali Cheek (Cheek et al. 2018a). This makes it imperative to discover and publish such species so that they can assessed, and, if merited, conservation actions taken to reduce their extinction risks, such as through designating and implementing Important Plant Areas (Darbyshire et al. 2017).

Materials and Methods
This study is based on herbarium specimens. All specimens seen by the authors are indicated "!". Herbarium citations follow Index Herbariorum (Thiers continuously updated), nomenclature follows Turland et al. (2018) and binomial authorities follow IPNI (continuously updated). Material of the suspected new species was compared morphologically with material of all other African Vepris, principally at K, but also using material and images from BM, BR, FHO, G, GC, HNG, P and YA. Specimens at WAG were unavailable during the period of study due to building work, and but low resolution images of specimens were viewed on the Naturalis website (https:// bioportal.naturalis.nl/). Specimens referred to in Aké Assi (2001) are held privately in Ivory Coast and are not accessible at present. The main online herbarium used during the study apart from that of WAG was that of P (https://science.mnhn.fr/all/search). Points were georeferenced using locality information from herbarium specimens. The map was made using ArcMap version 10.5, with additional layers showing protected areas (IUCN & UNEP-WCMC 2017) and lakes (Esri 1992). The conservation assessment was made using the categories and criteria of IUCN (2012). Herbarium material was examined with a Leica Wild M8 dissecting binocular microscope fitted with an eyepiece graticule measuring in units of 0.025 mm at maximum magnification. The drawing was made with the same equipment using a Leica 308700 camera lucida attachment.

Results
We present tables and keys separating the two new taxa from those with which they have previously been included. The species treatments that follow are presented in the order of the taxa in Table 1 above, beginning with Vepris verdoorniana under which name all four taxa treated in this paper were previously known (hence the "Vepris verdoorniana complex", followed by V. letouzeyi, recently segregated from the former, followed by the two new species.    (2015), was amongst the original material of Vepris verdoorniana (see above under that species and in introduction). Here we also assign sheet 1 of Staudt 493 to this species (see under V. verdoorniana above).

RECOGNITION.
Differing from Vepris verdoorniana in the canaliculate (not terete) petioles, the ovoid-ellipsoid, conspicuously and deeply pitted fruits (not narrowly ellipsoid, pits absent or inconspicuous), leaves opposite (not usually alternate). DISTRIBUTION  In fact the plot voucher specimens were all collected for Environmental Impact Assessments (EIs) for mining projects at these sites.
At Mount Bele, Adam (1975) reported that this species (as Teclea verdoorniana) is locally common in the understorey at the base of the mountain and nearby. This is consistent with the large number (17)  It is estimated that the project offers 2.37 billion tons of iron ore. Although the project is temporarily closed, there is little doubt that it will be re-opened once iron ore prices improve (https://www.thornycroft.co.za/ Gru/2018-4-15/putu-iron-ore-mining-in-liberia.html). Currently the hills are covered in intact lowland evergreen forest with only small areas cleared during the exploration phase of the project (Digital Globe imagery dated 1\2\2016 viewed in Google Earth 6  Of the four known locations, only one, the Gola Forest Reserve (Sierra Leone), is protected.
Using the georeferenced specimens cited in this paper, the area of occupancy (AOO, 44 km 2 ) and extent of occurrence (EOO 27,691 km 2 ) have been calculated using GeoCat (Bachman et al. 2011).
Given the continued loss of plants of this species at the Mont Tokadeh site, and the impending losses at Mount Bele and Putu Hills locations, we here assess Vepris fer as Endangered EN B1+B2a,b(iii). ETYMOLOGY. Named (French noun in apposition) for the iron substrate on which this species usually occurs. LOCAL NAMES. Samakin (Ganta: Fox 124, K). LOCAL USES. None known. NOTES. The description above is based on both the specimens cited and on the description of Adam (1975) which is derived solely from Adam 27597.
Mount Bele is the southern tip of the extensive Nimba Mountain range for which Adam completed the Flora (Adam 1975). Notably his specimen is the only record of the genus recorded in his extensive work on the main range of Nimba.
Initially we connected Barter s.n. (Sugar Loaf Mt, Sierra Leone) with this taxon since it has a canaliculate petiole and similar aspect.
However, it differs from our taxon in that it has alternate only leaves, much more densely hairy inflorescences, a long and well-defined petiolule to the median leaf, and many more secondary nerves per leaflet. The identity of this specimen remains uncertain.
Vepris fer is unique in West Africa (Senegal-Ghana) for the combination of large (>1 cm long) fruit which is 1-seeded, with conspicuous large-pits, and the regularly opposite leaves. Vepris bali Cheek of the Cameroon Highlands (Cheek et al. 2018a) is the only other Vepris species in W-C Africa with regularly opposite leaves. That species however has bilocular ovaries and 8 stamens, and long petiolules (not unilocular, 4 stamens, sessile leaflets).
The inflorescence of the type specimen shows scarring on the main axes suggesting damage early in development. This may account for the discrepancy in dimensions of the male inflorescences between Barker 1181 (c. 5 cm long, long internodes) and that of the type (3.5 cm long, more congested, possibly stunted).
The disjunction between Gola and Nimba locations is about 220 km, and between Nimba and the Putu Hills c. 180 km. It is likely that this is not due to undersampling in the surviving habitat between these locations but due to the preference of the species for submontane forest. This disjunct pattern is seen in other submontane forest species in the Guinea Highlands e.g. Trichilia djalonis A.Chev., and the more recently discovered Isoglossa dispersa I. Darbysh. (Darbyshire et al. 2011 . Extensive surveys in two other submontane forested iron outcrops (Simandou in Guinea, and Tonkolili in Sierra Leone, respectively van der Burgt et al. 2012;Clark et al. 2011), both also iron ore projects, has failed to discover this species, suggesting that its range is no more extensive than that recorded in this paper.
Although fertile specimens are not known from the Putu Hills location, they agree closely in gland dot size, colour and density, in the unusual opposite leaves with those from the other three locations.
The inclusion by  of Barker 1181 (Vepris fer) with its opposite leaves, from the high rainfall southernmost outliers of Mt Nimba in Liberia, as Teclea (Vepris) verdoorniana together with the alternate-leaved material from lower rainfall habitats in West Africa treated in this paper as V. occidentalis, was followed by other authors, such as Adam (1975) and Hawthorne & Jongkind (2006). In fact, these species differ from each other in numerous major characters (see Table 1).
While several species of Vepris sometimes have subopposite leaves, V. fer seems to be consistently fully opposite in its phyllotaxy, with the exception of a section of one stem bearing infructescences where the placement of leaf scars indicate that the phyllotaxy is alternate. This may be a case of stage-dependent variation in phyllotaxy as occurs in the threatened V. felicis Breteler which bears opposite leaves at the fertile stem apex of flowering stems but wherein nodes distant from the stem apex are alternate (Cheek pers.  Teclea grandifolia sensu Aubréville (1936: 92, fig. 163) non Engl. Teclea verdoorniana sensu Keay (1958: 689); Hall & Swaine (1981: 303).
Fruit orange or red when ripe, ellipsoid, 0.6 -0.8 (-0.9) × 0.4 -0.6 (-0.7) cm, apex and base rounded, surface drying wrinkled, oil cavities elliptic, 0.5 mm long, usually slightly raised, convex; 1-seeded. Fig. 2. RECOGNITION. Vepris occidentalis differs from all species within its range, and from V. verdoorniana, in the strongly laterally winged petioles (not terete), and by the oil glands on the lower surface of the leaflets being concolorous, not visible to the naked eye, (not black, conspicuous to the naked eye). It further differs from the last species in the ellipsoid fruits (not narrowly ovoid). DISTRIBUTION. Guinea, Ivory Coast, Ghana, probably HABITAT. Drier evergreen and semi-deciduous forest types, often on shallow soils over rock at the boundary with savannah (woodland); 100 -500 m alt.
Vepris occidentalis is listed (as Teclea verdoorniana) as one of 13 characteristic species of Southern marginal type forest (Hall & Swaine 1981: 83 -87). Associated species include Lecaniodiscus cupanioides Planch., Hildegardia barteri (Mast.) Kosterm., Dialium guineense Willd., Nesogordonia papaverifera (A.Chev.) N.Hallé and Drypetes floribunda Hutch. This forest type, as its name suggests, is marginal to the south of the main block of forest in SW Ghana, and interfaces with "Savannah" (woodland). Surviving forest is principally on thin soil over rock. Rainfall is 1000 -1250 mm p.a. Hall & Swaine (1981: 302, map 580), also map this species, at lower frequency, in Dry Semi-Deciduous Forest on the northern boundary of the main Ghana forest block, in a line proceeding south eastwards from the boundary with Ivory Coast, towards the Akosombo dam of Lake Volta. Rainfall is 1250 -1500 mm p.a. Hall & Swaine equate this vegetation type unequivocally with the "forêt dense semi-dècidue" of Aubréville (1938). Aubréville surmised that before the advent of agriculture it had stretched far to the north, covering what is now "Guinean savanna" (woodland). Guillaumet & Adjanohoun 1971) name this vegetation "Forêt à Aubrevillea kerstingii et Khaya grandifoliola ", and Chatelain et al. (2011) map the occurrence of V. occidentalis in Ivory Coast. However, we have not been able to verify all their 22 points for this species. The westernmost outliers of V. occidentalis occur in the northernmost, drier part of Liberia and in east-central Guinea. It is likely that the easternmost extent of this species occurs in Benin, since Akoègninou & Yédomonhan in Akoègninou et al. (2006) list Vepris verdoorniana in that country citing Houngnon 3659 from Misséssinto. However, we have not been able to view this specimen to confirm the identification, so do not include it in Map 2, although we consider it feasible, even likely. Akoègninou & Yédomonhan give the distribution of this taxon as in their phytogeographic zones 3 (Ouémé) and 4 (Kouffo), in the southern part of Benin, just inland of the coastal belt. This is semideciduous forest receiving rainfall of 900 -1200 mm (Akoègninou et al. 2006). No Vepris species is recorded from Togo (Brunel et al. 1984), but it is quite possible that V. occidentalis occurs there but has been overlooked since there are records immediately to the west in Ghana, and it appears to be present to the east in Benin (see above). Benin appears to be the eastern limit of the species since, although Nigeria is well studied for trees (e.g. Keay 1989), there are no records attributable to the species there.
Tarenna agnata Cheek has a partly overlapping distribution to that of Vepris occidentalis, also occurs in Dry Semi Deciduous Forest in Ghana, extending into Ivory Coast, but is seems much rarer, being known from fewer points and is considered possibly extinct (Cheek et al. 2015). CONSERVATION STATUS. Vepris occidentalis has a large range (EOO = 6,191,100 km 2 , AOO = 120 km 2 ), and we have been able to verify records from 16 locations. Chatelain et al. (2011) map 22 points for Ivory Coast alone, but we have not been able to verify all of these records. However, if confirmed, the number of locations of this species may be much higher, potentially exceeding 30, on which basis the AOO was calculated.
Relatively little of the original 2,360 km 2 of the Ghanaian Southern Marginal Type of forest, (the habitat in which this species is most frequent in Ghana), is today covered by forest (Hall & Swaine 1981: 85), and of this area only 93 km 2 (4%) is within Map 2. Global distribution of Vepris occidentalis in West Africa.