Reconstituting a rainforest patch in southern Benin for the protection of threatened plants

In a twenty-year effort at Drabo, southern Benin, small remnant forests, young fallow and agricultural fields were linked and rehabilitated to develop a 14 ha forest reserve. Forest regrowth was encouraged by managing the natural growth of the local fallow vegetation and by bringing in seeds and other propagules from forest islands of Benin. The succession to shade-tolerant woody forest species of Guineo-Congolian origin at the expense of extra-regional herbs, the co-existence of species with slightly different requirements, and the fate of exotic trees in this natural forest are described. A quantitative assessment of a homogeneous lot indicated 397 trees per ha, with stem diameters >10 cm, 43.7% of them below 20 cm, and a rich undergrowth of 72600 smaller plants per ha, proof of active rejuvenation. Only 4.2% of all plants resulted from the 104


Introduction
During the quaternary Ice Ages, West Africa's rain forests receded to two main refuges situated in today's Côte d'Ivoire/Liberia and southern Cameroon (Figure 1).As climate warmed, forests advanced, though with temporary recessions, but never quite closed the so-called Dahomey Gap in Benin-Togo and south-eastern Ghana, which separates the western Upper Guinean from the eastern Lower Guinean/Congolese forest block (Poorter et al. 2004, Giresse 2008).In present-day Benin, Guineo-Congolese closed semi-deciduous humid forests are limited to tiny patches, most of them so-called sacred forests, which are under threat.The fauna and flora of these forests are composed of species originating from one or the other or both forest blocks, as well as species from different savannah zones (Booth 1958, Robbins 1978, UICN 1996, Sinsin and Kampmann 2010, Neuenschwander et al. 2011).
In a highly populated region with sometimes over 250 people per km 2 (INSAE 2013) and embedded in an agricultural landscape and human-induced so-called derived savannah (Paradis andHoungnon 1977, Mama et al. 2014), these forests are islands of high biodiversity.Covering 0.02% of the national territory, they harbour 20% of all plant species and 64% of the highly threatened plants, but are mostly outside established nature reserves (Adomou et al. 2011).Their protection has been described as the highest priority for nature conservation in Benin (Neuenschwander et al. 2011).Many of these forests are severely degraded in terms of structure and species diversity, which calls for rehabilitation measures.Apart from the detailed study of Adomou (2005), only shortterm mapping of the flora and rapid surveys through interviews have been conducted (Adjanohoun et al. 1989, Sokpon and Agbo 1999, Nagel et al. 2004, Juhé-Beaulaton 2008, Kokou et al. 2008, Hèdégbètan 2011, Agbani 2012, CERF Bénin 2013).
Here we describe a 20-year effort to link up tiny forest fragments and to rehabilitate them through selective management of former farm and fallow land by encouraging forest regrowth and introducing plants from other forest islands of southern Benin.The management of this reconstituted, biodiversity-rich forest with its most important inhabitants, the critically endangered, endemic red-bellied monkey, Cercopithecus erythrogaster (Cercopithecidae) and the interactions with the villagers have already been described and compared with management in other forest sites (Neuenschwander et al. 2015).(2000), UICN (1996), Chatelain et al. (2004), Hawthorne and Jongkind (2006), and Adomou (2011).

Overall situation
Ecological zones in West Africa (Figure 1) were delimited using Arbonnier (2000) for the north, UICN (1996), Chatelain et al. (2004), Hawthorne and Jongkind (2006) for the south, and Adomou et al. (2006Adomou et al. ( , 2011) ) for Benin.The main ice age refuges in western Côte d'Ivoire, today Taï National Park, and western Cameroon, today Korup National Park, are marked.In view of the huge human impact in the region, these zones circumscribe the ecological limits for the corresponding vegetation, rather than showing actual forests.
Within the Guinea-Congolese forest zone, the limits for the vegetation and the outer limits of still existing semi-deciduous and evergreen rainforests are therefore given separately.The site of a detailed study of plant cover (Chatelain et al. 2004) in the 20 × 20 km Sikensi block is indicated.
The Dahomey Gap as shown in Figure 1 has a strong gradient of rainfall, with highest rainfall towards the Nigerian border and lowest rainfall in the Accra Plains in Ghana.In Benin, this gradient is reflected in the distinction of phyto-geographical districts within each ecological zone (Figure 2, Adomou 2011).The Guineo-Congolese zone has two widely separated rainy seasons; further north, there is only one rainy season.

Study site
Drabo Gbo (6°30'N; 2°18'E; ca 60 m asl) (Figure 2) is a village of about 500 inhabitants at the southern edge of the Allada plateau, 30 km north of Cotonou, 12 km from  are 20 °C, rarely down to 15 °C during "Harmattan" periods, when a dust-laden wind from the Saharan desert brings cooler night-time temperatures.The study site has a deep lateritic soil.One small lot, called "Emile" in Figure 3, is sandy.The water table throughout is at 25 m depth below the soil surface.
The "Monkey Sanctuary of Drabo Gbo" (Figure 3, centre of map: 6°30.28'N;2°17.77'E) was founded in 1995 when the first author bought 2.5 ha of teak forest and agricultural land from the elders of Drabo Gbo.The study site was assembled from 25 single lots of different sizes, most of them agricultural fields or recently fallowed land, bought between 1995 and 2005 (for year of purchase of each lot see legend of Figure 3).A small house in the forest and a house for the community called "Maison de Jeunesse" (MdJ), situated on an adjacent triangle of land, were constructed in 1997/98 and 2005, respectively.For some lots, full registration with title deeds has not yet been possible.In January 2014, most of this land was donated by the first author to the International Institute of Tropical Agriculture (IITA) as a site for biodiversity studies.Today there are 14 ha of land that has been managed for forest regrowth from agricultural fallow or teak plantation over 10 to 20 years, as well as tiny sacred forests, like Orojamè (0.8 ha) and Dodja with its old growth forest (0.7 ha).They are adjacent to the village of Drabo Gbo and the smaller villages of Drabo Fanto and Dodja.

History of the land
Speed, quality and abundance of regrowth depend much on the previous occupation of the land (Stanturf et al. 2012, Chazdon 2014).The first author's residence on this site since twenty years offered numerous opportunities for informal interviews with the elders, most of them illiterate farmers, in order to tap into the oral history, which in most families was very much alive.

Management
A flimsy fence of concrete poles, which support five lines of barbed wire, surrounds most of the forest lots; however, effective protection is provided by the "vodun adogba" , i.e., by traditional cults to which the first author has been initiated.Interactions with the surrounding population were described in detail in Neuenschwander et al. (2015), among others that no gathering of fire wood and medicinal plants is allowed.
Management consists of freeing trees from an excessive burden of climbers and maintaining small paths.Exotic timber species, such as Cassia siamea, Acacia auriculiformis and Tectona grandis (teak), are cut on demand by villagers, but under the supervision of the owner.Among the indigenous commercially important species, oil palm (Elaeis guineensis) is by far the most important.On demand, its fruits are collected where easily accessible.During successive fallow cycles, this indigenous species had been protected and enriched to a degree where in many places areas of up to 0.5 ha were so densely stocked that all other vegetation was suppressed.Cutting these oil palms produced sunlit habitats, which triggered the emergence of pioneer trees and offered open space for planting young trees.These were sometimes shaded by cut palm leaves until full rooting was achieved.

Introducing trees, shrubs and lianas from other forests
During the twenty-year period of establishing the forest, numerous collecting trips were made to other patches of forest in southern Benin.Plant material was collected mainly in the following locations (Figure 2): Forest pockets (<1 ha each) near Drabo Gbo and nearby villages; a community forest in Lanzron (50 ha) east of Zinvié in the Ouémé Valley, which is inundated from August to November; the "Forêt de la Panthère" (1.4 ha) of Zinvié (for management of the Lanzron and Zinvié forests see Neuenschwander et al. 2015); the forest on the national agricultural research station of Niaouli (170 ha), where collections were made along public paths to the small, local stream and beyond; the "Forêt de Kè" at Dangbo (2 ha) bordering swampland; the forest of Ahozon near Pahou (150 ha natural forest), which is the last remaining coastal forest between Lagos and Accra, now under the supervision of the national forest services; the forest of Pobè (about 115 ha) on land of the national oil palm research institute, with a small river; the forest of Ewè (150 ha), a rich but unprotected forest contested by two rival communities and under threat of disappearing completely; the natural forest of the "Noyau Central" of the Lama forest (2000 ha), which is the largest dry type rainforest of Benin, locally inundated and supervised by the managers of the surrounding teak plantations and the official forest station; the community forest of Tobè near Koko, West of Bantè (350 ha), in the Guinea Savannah, which is controlled by the local community and supported by an NGO; the forest of Pénéssoulou still further north; plus exchanges of plants with the Botanic Garden of the University of Abomey-Calavi.With the exceptions of Tobè and Pénéssoulou, the flora of these sites belongs to different phyto-districts of the Guineo-Congolese semi-deciduous forest zone (Adomou 2011).
Most collecting expeditions were conducted in the main rainy season to benefit from conditions favourable for transplanting.To facilitate establishment at Drabo, plants, seeds, or cuttings were first held in a 50 m 2 nursery in open forest near the house, where they could be provided with additional shade if needed, water, and mulch.Most plants remained there for up to a year, awaiting the next significant rains.Recalcitrant seeds were scarified or treated with hot water.Some plants were planted out directly into particular microhabitats, mostly into 5 ha near the habitation of the owner.Watering and mulching during the first year made it possible that also plants from slightly moister forests could survive in Drabo.

Quantitative assessment of regrowth
On "Cooun" (Figure 3), which has the highest proportion of introduced plants among all rehabilitated forests of this study, a uniform regrowth area of about 4 ha was assessed quantitatively in May 2016 to describe first the general habitus of this 20 year old Albizia-Antiaris-Blighia forest with a high proportion of oil palms.Second, the importance of introduced plants was to be estimated.The total number of trees >10 cm diameter on 1000 m 2 (a circle measured out with a string of 17.8 m length) and all species of trees, lianas and herbs on 5 times 10 m 2 (in the centre and on the main circle in all four directions, measured with a string of 1.78 m) were assessed.The procedure was repeated three times from randomly chosen points on the narrow main path.Data about trees from 3000 m 2 and all plants from 150 m 2 were extrapolated to numbers of plants per ha.Tree numbers were grouped in 10 cm-diameter intervals, separately for the three dominant Albizia sp., two common Blighia sp., Elaeis guineensis, all other indigenous trees, and all introduced plants.In the small circles, all plants were deter-mined to species and counted.Trees with diameters >10 cm were separately noted to check for congruence in the data, since the surface of the 15 small circles is 5% of the surface of the three big circles.

Inventory of the flora
To register all plant species, parts of the forest were inspected since 1998 daily for "Cooun" and surroundings, weekly for "Grande Forêt" and surroundings, and monthly for Drabo Fanto and Dodja by PN and/or a local guide.Unknown plants were marked with a yellow plastic band for later identification by AA.First identifications in 1997 and 1998 had been made by the late Prof. Paul Houngnon of the National Herbarium of Benin.Plants were identified by means of Akoègninou et al. (2006) and Hawthorne and Jongkind (2006) and by comparison with specimens in the National Herbarium.Additionally, dead or withering plants were noted.If this condition concerned all or most specimens this species was registered as 'decreasing'.
A list of all species registered in these forests is given as supplementary file.Species from aquatic, semi-aquatic, coastal-sand, or rocky habitats as well as horticultural species are excluded because they are not part of the natural flora of Drabo and, where present, survived only under special conditions, i.e., in pots or concrete ponds.The supplementary file lists: -Taxa (Pteridophytes as 1Pteri, Gymnosperms as 2Gym, Monocotyledons as 3Mono, Dicotyledons as 4Dicot, and non-identified as 5nonid.-Plants are described as parasites/epiphytes, herbs, climbers, shrubs or trees based on the description in Akoègninou et al. (2006) and-where different growth forms exist-with the habitus they show in Drabo.Species that are either new or not included in this flora are marked.Family names follow taxonomic revisions (APG I 1998, for least differences with Akoègninou et al. 2006).-Chorology according to Adomou et al. (2010Adomou et al. ( , 2011) ) and Akoègninou et al. (2006): The origin of the species is indicated as follows (Figure 1): GC Guineo-Congo forest species that are distributed across the Upper and Lower Guinean and into the Congolese zone east to Sudan, Uganda, Kenya, GO Upper Guinea forest species from west of the Dahomey Gap with an eastern limit in Benin or nearby Nigeria, GE Lower Guinea forest species from east of the Dahomey Gap with a western limit in Benin, SG Guineo-Sudanian transition zone species, SZ Sudanian savannah species, and S Sahel savannah species.In the evaluation of threatened species, NT-species were excluded.-Suspected reasons for difficulty in establishment were given as: -recalcitrant seeds, which were sometimes treated with hot water or by scarifying, -drought, i.e., temporarily too little water, -savannah species, of which we suspect that they do not support transfer to two rainy seasons and are therefore not capable of reaching the coast, -medicinal use, e.g., roots harvested for increasing male potency.-The number of samples, which include one to maximum 10 plantlets or seeds per species and date, was indicated.-Plants originally found on the 14 ha and in their vicinity of a few hundred meters were marked with x.Some of them, if they were rare, were also reproduced and transplanted to other sites.
Percentages were compared by a two-way Chi-square test with correction for continuity.

History in the 20 th century
Even before the village of Drabo Gbo had been founded at the end of the 19 th century, the area had been under cultivation probably for hundreds of years out of the village of Gbodjo situated at the lagoon near Calavi, with fallow cycles of 15 years or more.Some forests remained untouched and served as sacred forests, like Dodja, the old-est forest in this study.Its Cola gigantea, Celtis milbraedii, Blighia sapida and Antiaris toxicaria are probably several hundred years old and reach 40 m.The forest was not touched for perhaps 100 years; but gaps with trees of only 10 m height that are heavily encumbered by lianas indicate that big trees had been cut before the present inhabitants of Dodja can remember.Another sacred forest, Orojamè, was created around a big Cola tree about 80 years ago and to date is the central site for the "Oro" cult of the entire area.This site of 0.4 ha was bought in 1999 and increased to about 0.8 ha.
In the former teak forest and another fallow lot, two big Cola gigantea trees were included in the newly acquired land.Their preservation was due to a courageous action in the late 1970s by the then "délégué" , who refused political pressure to cut them down because they were reputed to be sites of ghosts.They became islands of biodiversity, from where plants, insects, but also mammals spread to the rest of the developing forest.
About 8 ha, among them "Cooun" where the quantitative study was made, were under forest cover about 50-70 years ago, before the forest was cleared and replaced by maize, teak, Dialium guineense, which were stunted to harvest the sweet seeds, or cow "pasture", effectively bush regrowth since the local cows eat only broad leaves, no grass.Since no heavy equipment was ever used, seeds and roots of the previous forest vegetation remained alive.When fallow or crop land was purchased it was mostly devoid of trees.Once cultivation ended, shrubs, some up to man's height, developed into big trees within a few years.
In the past, fire in Fanto was ignited along the edge to catch grass-cutters, a large rodent.Moreover, at Drabo Gbo, six big Albizia and Rhodognaphalon trees were set on fire by igniting a car tyre at their base.In the same vindictive action in 2013 by villagers, who were enraged because of a local murder, all 65 trees on the triangle of the "Maison de Jeunesse" (MdJ), were cut.This action was ordered by the "Bokonon" (seer), who claimed a bad spirit was located there.The garden of the MdJ has meanwhile been transformed into an ethnobotanical garden.At present, the protection of the forest is good, i.e., no trees are debarked or felled, few lianas cut, no fire is laid, and no hunting is observed, except for some digging-out of ground squirrels or Gambian rats.

Status today
The forests of Drabo are embedded in a landscape of farmland with numerous isolated small houses.The section in Figure 3 covers 2.1 km 2 and shows 80 houses in Drabo Gbo, 16 in Drabo Fanto, 16 on AgoXwè, plus 350 isolated houses on 500 m 2 lots.The surrounding vegetation consists of fallow, which during the period of the study was gradually reduced from a few to zero years.Fields are mostly planted with maize, cassava, or pineapple.Up to now, most land has been sold to people from town by the original owners, who eventually install themselves in this area by building modest onestory houses and grow various fruit trees and bananas.Where construction has not yet started the land is often being abused by the original owners without any consideration to conserve soil fertility.It is therefore often heavily infested with Imperata cylindrica.Gradually, big Cola and other trees, which were typical of a rich landscape, have been felled and the forest of the present study has meanwhile become an island in an ecologically impoverished surrounding.
Numerous woodlots have been planted with Cassia siamea, Tectona grandis, and mostly Acacia auriculiformis.The total area of uniform and biodiversity-poor woodlots in Figure 3 is bigger than the 14 ha of forest of the present study.Such lots provide wood for construction and firewood and take the pressure off the reconstituted, biodiverse forest.
Today, trees in Dodja are around 30 m high, with a few up to 40 m.In Orojamè the canopy is at around 20-25 m.In the oldest, selectively managed fallow land, trees reach 20-25 m, but in Drabo Fanto and on "Emile" in Drabo Gbo on sandy soil, the canopy is at only 10-15 m.
In 2016 a quantitative assessment of the vegetation was made in a mixed Albizia forest with rich undergrowth on "Cooun ", after 15 years of fallow combined with cow pasture (Figure 4).Among all trees, Albizia spp.(half A. zygia, ¼ A. adianthifolia, ¼ A. glaberrima) accounted for 64.7% (N = 119), Elaeis guineensis for 13.4%, all other indigenous trees for 19.3%, and the introduced species for only 1.7%.There were 397 trees with diameters >10 cm per ha, resulting in 25.2 m 2 per tree.Young trees with diameters of 10 to 20 cm accounted for 43.7% of all trees.In the small circles, where all plants were identified, a total of 1093 plants were counted, i.e. 72600 plants per ha.Four trees had diameters of >10 cm, which averages out at 267 tree per ha, slightly below the count from the big lots.This rich undergrowth was variously dominated by Mallotus oppositifolius (in 5 of the 15 samples), Chassalia kolly (in 3 lots) or Reissantia indica, Macrosphyra longistyla, and Brachiaria deflexa (each dominant in one lot); the rest (4 lots) having a highly mixed undergrowth.Though this part of the forest was particularly well enriched with species from nearby natural forest patches the importance of these species remained still modest, namely 4.2% of all small plants in the small lots.
All above named species are part of the 328 naturally occurring species classified as abundant (category 5), yet they account for only 11.6%.For 64.6% of all local species, less than 20 specimens are known, and for 25.0%only 1-4 specimens could be found despite regular and intensive search.Among the 253 introduced species, only one reached a total population of more than 20 specimens.

Origin of species and successional changes
Identification, particularly of juvenile plants without flowers and fruits, proved to be challenging.Up to 1998 a total of 168 plants of local origin had been identified.By 2004, when the bulk of the land had been acquired, the number increased to 317.Up to 2010 only 7 more species were discovered, and 4 more up to 2016, bringing the total of local species to 328.Another 253 species were introduced from forests from southern Benin (Fig. 2).The supplementary file lists 635 plant species, among which four could not yet be identified.Identification or description as a new species will become possible as soon as flowers will be produced.Two have been tentatively attributed to the genus Monanthotaxis (Annonaceae).
Among the 631 remaining species, 50 or 7.9% were lost during the last 20 years, 49 of them introduced species.For five of the 10 extra-regional (At, PAL, Pt) and 10 of the savanna (SG, SZ, S) species, loss is attributed to seeds that would not sprout in the forest environment.By contrast, only one out of seven forest (GC) species that had been collected only as seeds failed to produce seedlings.
Since locally occurring rare species were also transplanted to other parts of the forest, the total number of introduction events in the course of the last 20 years was a high 1041, with a maximum of 10 introductions per species with up to 10 individuals from one collection date.During the last 20 years, changes in composition were observed.Declines in number of plants and, sometimes, plant vigour were unevenly distributed among growth forms and origins (Table 1).Declines were rare (7.0%) among plants from the GC zone, particularly trees and shrubs.They increased gradually the further away the plants originated from, up to 64.6% among plants of pan-tropical origin.Particularly pan-tropical herbs became rare with 80.0% of all 65 species of this category declining and being confined to the few remaining open spaces and alongside paths.
Change to a more mixed composition of large trees with a larger proportion of evergreen species is reflected in the frequent breaking of large branches of Albizia trees during storms, while Celtis and Blighia spp.resist better and become gradually more important.
The decline in numbers and vigour of particular species in Table 1 indicates the prospective changes in species distribution.Here some examples concerning trees and plants of commercial interest: A total of 224 tree species are growing on these 14 ha.Among them, 69 are big trees of 10 m height or more.In the Drabo forest under total protection, the number of Cola gigantea, which had been largely destroyed in the area, increased steadily.The timber trees Afzelia africana, Celtis milbraedii, Diospyros mespiliformis, Milicia excelsa and Triplochiton scleroxylon, which had all disappeared from Drabo, grew well in the study forest.Erythrophleum suaveolens raised from seeds became common and reached 25 m, while Rhodognaphalon brevicuspe planted as sticks reached 22 m.Some formerly rare plants like Oxyanthus racemosus and Pavetta corymbosa, of which only one plant each could be found in 1995-1996, became common.Similarly, some tree species like Albizia ferruginea, Borassus aethiopum, Ceiba pentandra, Celtis milbraedii, Parkia biglobosa, Psydrax parviflora, Vepris verdoorniana, and Zanthoxylum leprieuri existed in only one lot, but grew well when planted elsewhere.Several trees, which were the last ones in the area and were subsequently felled, provided seeds and sticks that were reared in the nursery and then planted out: Cynometra megalophylla, Elaeophorbia drupifera, Kigelia africana, and Pentaclethra macrophylla.
Some species from the Guineo-Sudanian transition zone or even further north reach the coast and were found in Drabo, like Adansonia digitata, Crossopteryx febrifuga, Lawsonia inermis, Parkia biglobosa, and Tamarindus indica, of which only a few big trees survived in the area.Lophira lanceolata from the isolated population near Pahou survived only on sandy soil.
In the course of reforestation of the last 20 years, yearly herbs decreased or disappeared.They resurged, however, immediately wherever an opening provided sunlight habitat.Thus, Trema orientalis trees could not be found anymore, until they suddenly reappeared from dormant seed on MdJ in 2013 in the newly opened up area, hundreds of meters away from the original site.Even in places where Chromolaena odorata was not cut, the encroaching forest within 10 years shaded it out completely.Tithonia diversifolia, another feared invader, appeared only once in 2002 and disappeared the same year, when shaded out.
Commercialized fruits are produced by the following well-established indigenous trees: Chrysophyllum albidum, Dialium guineense, Irvingia gabonensis, and Spondias mombin.Of particular interest is the introduced Synsepalum dulcificum, an endangered plant that has potential for commercialization because of its sugar-free sweet fruits.Tropical crops of closed forests like Coffea canephora, Hevea brasiliensis, Theobroma cacao, and Vanilla planifolia needed irrigation to start growth.By contrast, the exotic fruit trees Anacardium occidentale, Azedirachta indica, Mangifera indica, and Psidium guajava did not last in the closed forest and died within 10-15 years.Similarly, field crops like Ananas sativus, Manihot esculenta, and Musa triploid spp.did not produce any commercial product anymore, but persisted in the closed forest for many years even when shaded.
Among the introduced timber species, Acacia auriculiformis and Cassia siamea grew well in closed forest, whereas Leucaena leucocephala was overgrown by climbers and despite abundant seed-set did not reproduce vigorously.Among the original 1 ha of Tectona grandis many trees were cut; but the remainder grew well in a mixed forest.The one clump of Bambusa vulgaris remains problematic because it is displacing all other plants in the vicinity.
Many medicinal plants, which grow also in the Drabo forest, like Acanthus montanus, Mondia whitei, Acridocarpus smeathmanii and A. alternifolius, as well as Heliotropium indicum are being commercialized and sold on the "juju" markets, without being cultivated or even locally protected.

Endangered plants
Table 2 lists a total of 73 species from the Red List of Benin (EW, CR, EN or VU only), which could possibly survive in the reconstituted forest, among which 52 species are actually growing in Drabo, i.e., 71.2%.Four more species were planted, but could not survive.Ten species were found in Drabo before introductions from other sites started.
Most threatened (CR) are Acanthus montanus which was purchased on the market in Calavi, where material from one location near Porto Novo is sold as medicine; Barteria nigritana from Ahozon; Caloncoba echinata from a moist site in Niaouli; Entandrophragma angolense, which originally had only a few trees in Niaouli; and Dennettia tripetala and Nesogordonia kabingaensis from Ewè, where they are not protected.Caesalpinia bonduc is extinct in the wild (EW), but widely maintained (and also stolen!) in the village for its medicinal roots.A few plants of Garcinia cola (EW) also grow in the study forests.Of special interest are Distemonanthus benthamianus (EN), which exists in Benin only in one small, though protected site in Pobè, and Mansonia altissima (CR) from Ewè, where it is not protected.Both species grow as isolated populations at the eastern edge of their continental distribution, and only few trees exist in Benin.
Thanks to 20 years of careful protection and management, many threatened plants could be established and the species richness of the forest now rivals the one of the forests from where these endangered species come from (Table 2).

Discussion
Biodiversity in general, and of tropical forests in particular, is to be preserved for ethical reasons, for agriculture under the headings of plant protection-biological control (Neuenschwander et al. 2003, van Driesche et al. 2010) and varietal selection (Dansi et al. 2013), and for the preservation of water regimes and mitigation of climate change (Corlett 2014) or other eco-services (Martin 1991, Kuyah et al. 2016, Rowland et al. 2016).In an ideal world, nature protection and agriculture are considered complementary and nature reserves a prudent investment (McNeely and Scherr 2001).Though knowledge and understanding of the interactions with agriculture are increasing rapidly (IAAST 2008, Kuyah et al. 2016, Rowland et al. 2016) destruction of forests continues unabated.For Benin, the legal framework for nature conservation is in place (Republic of Benin 2012), but implementation is often insufficient.The protection and restauration particularly of southern forests has therefore been described as first priority in nature protection (Neuenschwander et al. 2011).
Because of their small sizes, Benin forests would not satisfy the criteria for a key biodiversity area (Secretariat of the Convention on Biological Diversity 2002, Brooks et al. 2006, IUCN 2016, RBG Kew 2016); but they show a particularly interesting history.Generally, forests in West Africa have been advancing since the Ice Ages, 10 000 years BP, though this advance has been marked by recessions and, in the area of Table 2. Distribution of threatened plant species in southern Benin.Only living woody plants (trees, shrubs, lianas) and species on solid ground (i.e., without plants limited to swamp forests, sandy beaches or rocky outcroppings) are considered.Only VU to EW. Presence = x, origin as in Figure 1.
x Zinvié GC Asclepiadaceae Mondia whytei  the Dahomey Gap, slowed by sea currents and trade winds.Since 2000 years, advances are regular and little influenced by grazing, slash-and-burn agriculture with long fallow-periods, originally with endemic crop like rice and yams, firewood collection for iron-smelting, and home use by low human populations (Giresse 2008).Because close forests develop at annual rainfalls of >1400 mm and existing forests decline if rainfall drops below 1200 mm the area of the Dahomey Gap, including southern Benin, did not reach total forest cover (Giresse 2008).Under the rapid population increase of 10 to 20 times in Benin observed in the 20 th century, tree savannah surrounding the rainforest pockets was further degraded (Paradis andHoungnon 1977, Mama et al. 2014) and gave rise to agricultural land under ever shorter fallows.Rainforest vegetation was confined to the often cited, but little studied sacred forests (lately: Agbani 2012, CERF Bénin 2013), some small communal and government forests, as well as riparian forests further north (Natta and van der Maesen 2003).These forest islands harbour plant species with origins in the biodiversity centres of the Upper or Lower Guinea forest blocks.Today, large contiguous forest areas in West Africa are confined to western Côte d'Ivoire and eastern Liberia for the Upper Guinea block and extreme eastern Nigeria to southern Cameroon for the Lower Guinea block.In most areas designated as evergreen or semi-deciduous humid forest (Figure 1), deforestation has been so pervasive that actual conditions are not that far removed from what is found in the moister parts of the Dahomey Gap.In southern Côte d'Ivoire in the Sikensi block of 20 × 20 km, well within the evergreen forest zone (Figure 1), for instance, forest cover changed from 182 km 2 in 1958 to 149 km 2 in 1990, to 17 km 2 in 2000.Forests of less than 4 ha represent 64% of the forest area (Chatelain et al. 2004).Similarly high deforestation rates with maxima in the 1980s were reported from Nigeria (14.3%) and Côte d'Ivoire (15.6%) (Giresse 2008), demonstrating the difficulty of assigning phyto-ecological zones.
Because of rapid deforestation eight out of 26 commercial timber species of Côte d'Ivoire and Ghana are listed as threatened (Poorter et al. 2004).Some of these species exist in Benin, where their threat status is even higher (Table 2, Adomou et al. 2011; for a more stringent list see Adomou 2005).Table 2 demonstrates the richness of threatened species that grow in Drabo, rivalling the one of well-established larger forest reserves.It must, however, be noted that Benin forests with locally open water, like Pobè, Dangbo, and Niaouli, harbour several additional threatened plants, namely those requiring much higher humidity than found in Drabo; such species were not collected nor evaluated.Several of the listed species, like Distemonanthus benthamianus and Mansonia altissima, which are moreover at the edge of their continental distributions, occur in only one locality each, which-in the case of Mansonia-is unprotected, yet they grow well in Drabo.Despite similar conditions, none of the listed forests has all threatened species.We conclude that, in this highly fragmented landscape, rare species were either destroyed by man or suffered from stochastic loss in isolated habitats where important seed dispersers, like some birds, bats or rats, are no longer present (Fahrig 2003, Klein et al. 2014).In a situation, where creating corridors (Damschen et al. 2006) is no longer possible, introducing rare species into convenient habitats under protection counteracts this loss.Our systematic transfer within the same ecological zone, which did not inflict undue harm in the collection sites, thus saved plants that might otherwise disappear from Benin.In Benin, many of these species are at the edge of their ecological and/or geographical distribution and may therefore be genetically different from those in the centre.Such genetic differences could well give them an advantage to face stress from climate change.Special care to ascertain their survival is therefore justified.
The fact that plants from zones with yearly rainfalls 200 mm above the one in Drabo, as for instance Baissea zygodioides, Dennettia tripetala, Dracaena phryniodes, Gardenia nitida, Landolphia togolana, Pararistolochia goldieana, Tapura fischeri, and Turraea heterophylla from evergreen forest pockets (Holmgren et al. 2004) survive in Drabo, if watered modestly during the first dry season, is testimony to their great plasticity and adaptability.
At the other extreme, these forests harbour also tree species that must be considered remnants of the savannah into which the advancing forests intruded.These include Lophira lanceolata, with its isolated population near Pahou (Paradis et al. 1978) as well as Adansonia digitata, Crossopteryx febrifuga, Parkia biglobosa, and Khaya senegalensis, all characteristic trees of the Sudanian savannah, some with occurrence even in the Sahel savannah.They thereby demonstrate their great adaptability to different climates.
The present long-term observations in Drabo describe the succession towards the climax Guineo-Congolese semi-deciduous forest through natural fallow, enriched with plants from other botanical districts of this zone (Adomou 2011).By planting in clearings similar to natural tree fall and without irrigation, traditional forestry practices used for rehabilitation were employed (Sabogal 2007, Stanturf et al. 2012, Chazdon 2014).By planting Celtis spp., the natural succession from leguminous pioneer trees like Albizia spp. to Celtidaceae, as observed in Dodja, is being accelerated.In contrast to forests on shallow soil as in IITA-Ibadan, where in some places 40 years were not enough to develop a forest (Neuenschwander et al. 2015), Drabo forests develop on deep soil and, by adding species, reach the semblance of a secondary forest within 20 years.
As the tree canopies close herbs-most hail from outside the region-gradually decline.This concerns particularly grasses, so that only two forest species, Olyra latifolia and Oplismenus hirtellus, remain.The disappearance of herbs is, however, immediately reversed as soon as free soil appears because of natural tree fall or clearance.Because the Drabo forests are of irregular shape with long edges, many light-demanding species can still survive along the fringes.
Except for bamboo, no plant in Drabo has the potential to become invasive and transform the environment.Even Chromolaena odorata and Imperata cylindrica, two feared weeds, as well as potential invaders like guava or Leucaena leucocephala under the conditions of southern Benin just follow the environmental conditions, but do not transform them (Lincoln et al. 2016).They are easily shaded out and cannot stand up to the pressure from the local vegetation except when situated at the very border of the forest.Moreover, in situations where C. odorata becomes an obnoxious climax stage, its biological control is now promising (Prasad et al. 1996, Timbilla 1996, Day et al. 2013).
One cannot step into the same forest twice (paraphrasing Heraclitus); the present text therefore provides only a snapshot.The fact that only a few species were newly detected in the last years indicates that the cumulative list of species has reached a plateau and we can expect only few more discoveries.Since 21 IUCN-listed spp. with good potential for establishment are still lacking (Table 2) some more threatened plants might, however, be introduced in future.Conversely, among the introduced plants, some species are represented only by a few, some even by only one specimen with uncertain survival.Most cultivated plant still present in the forest will disappear altogether, while plants used in traditional medicine will hold steady if sufficiently protected.Due to successional changes, alpha-biodiversity, expressed as the total number of species, will probably decline as seen in other systems (Barlow et al. 2007).It is doubtful that a stable climax will ever be achieved because natural tree fall and edge effects will continue to offer footholds for transient species.

Conclusions
With 585 plant species or about 20% of the Flora of Benin, the Drabo forest have become a sanctuary not only for monkeys but also for rare plants, which themselves again offer the basis for the establishment of rare butterflies and other specialists.The vicinity to big towns and the relatively easy access should allow some eco-touristic development.Since the ownership has been transferred to IITA sustainability should be guaranteed (Neuenschwander et al. 2015).
The present study shows that with relatively modest means, but much patience and perseverance, it is possible to restore, even create de novo, a rainforest.The techniques are available since long time (Dobson et al. 1997, Mansourian et al. 2005).What is needed now is action to create a network of protected forests with exchange of species and local rehabilitation to round up the area of forests and to fill holes created by earlier logging.Most importantly, the local populations have to be involved and see the advantage of such forests or at least not oppose their creation.The 14 ha reserve is not large, but it represents two dozen sacred forests.Contrary to those, it is open to the public.Though it is not "natural" (Willis and Birks 2006) it effectively protects biodiversity in a human-impacted, so-called "gardenized" , landscape.
For the future, the major question remains whether in this densely populated area it will be possible to maintain this sanctuary which has become the best known and fully sampled forest of the entire region, while all other forests are all less well known and probably also less well protected.To assure its sustainability the forest will have to be used for scientific studies and bring benefits to the local population.The present study should thereby serve as a basis.

Figure 2 .
Figure 2. Map of Benin with phyto-ecological zones with GC, SG and SZ as in Figure 1, phyto-geographical districts according to Adomou (2011) in small print.
Many species occur in different zones; only their main habitat is indicated.At indicates Afrotropical species with distributions beyond West and Central Africa into Madagascar, PAL Paleotropical and Pt Pan-tropical species, i.e. all species that have penetrated or invaded West Africa from other floral regions, for example Cocos palm.West African species that have similarly spread across the world are indicated by their original zone, for example oil palm.-The next column indicates the origin of transplanted species (Figure 2, including Ouéga and the IITA campus a few km south of Drabo, Agongbè north of Drabo, Hévié north of Pahou, Avrankou on the Iguidi River near the Nigerian border east of Dangbo, and Tanougou waterfalls outside the Penjari Park in the Sudan savannah) and their location in the study forests (Figure 3), the years of collecting and transplanting, and-separately-in which form, namely seeds, small plants, sticks, they were collected.-Abundance in 2016 is ranked as follows: 1 = 1-4 plants established; 2 = 5-10 plants; 3 = up to 20 plants; 4 = common species; 5 = abundant species.-The maximum height of trees of each species was estimated in April-June 2016, taking into account all study forests.For most species, biggest trees were found in Drabo; in addition, especially big specimens from Dodja (Do), Fanto (Fa) or Orojamè (Or) were indicated.-The population tendency was roughly estimated as s = stable, i = increasing, d = decreasing.Where necessary, Cut or Lost (L = lost by July 2016) were indicated.-Red List status was given according to Neuenschwander et al. (2011) based on IUCN criteria as NT = near threatened, VU = vulnerable, EN = endangered, CR = critically endangered, EW = extinct in the wild in Benin.All other species were considered as Least Concern because they are not threatened or have not been assessed.

Table 1 .
Number of species alive and percentage declining (d.) in different ecological zones.Drabo July 2016.