Phytocoenologia, 39 (2), 157–174 Berlin – Stuttgart, June 16, 2009 Old growth mature forest types and their floristic composition along the altitudinal gradient on Silhouette Island (Seychelles) – the telescoping effect on a continental mid-oceanic island by Bruno SENTERRE, Bruxelles, Justin GERLACH, Cambridge, James MOUGAL and Denis MATATIKEN, Seychelles with 4 figures, 2 tables and 3 photos Abstract. The granitic Seychelles are the only mid-oceanic islands of continental origin. Botanists have long focused on taxonomy, and plant communities were described in a qualitative way, based on simple observation. Therefore the altitudinal belts, their floristic characteristics and distribution are still poorly understood and conservation efforts focus mainly on species-centred actions. Here we describe a quantitative study of plant communities and indicator species differentiated along the altitudinal gradient on Silhouette, the most pristine and second highest (740 m) island of this archipelago. Twelve plots were sampled from 80 m to 640 m above sea level. Each plot contains three nested subplots corresponding to three forest strata: 50 x 10 m for all individuals of the tree layer (i.e. trunk diameter > 5 cm), 50 x 4 m for the shrub layer (ligneous > 90 cm height), and 50 x 4 m for estimation of abundance-dominance coefficient in the herbaceous stratum (all herbaceous plants, plus ligneous plants < 90 cm height). The results are summarized in a two-way table. Both indicator species and vegetation types are discussed in relation to previous studies in Seychelles. Three altitudinal belts are distinguished: lowland, submontane and lower montane rain forests. Most of the best indicator species are found in the understory, especially ferns. The submontane belt develops at about 350 m and turns into a typical lower montane belt at ca. 550 m. Such transition zones occur respectively at about 900 and 1500 m in most of the tropical mountains, illustrating here a perfect example of the “telescoping effect”. Although the flora of Seychelles is relatively species poor, the strong characterisation of these altitudinal belts is unusual compared to younger islands in the Pacific, and may be a result of the longer evolution of its flora. eschweizerbartxxx author Keywords: altitudinal zonation, cloud forest, IndVal, Massenerhebung effect, submontane, understory. Introduction Our capability to protect species is ultimately dependent on the conservation of representative and functionally sustainable examples of the plant communities, i.e. vegetation types that support these species. Conserving biological diversity at the level of natural communities is an important complementary approach to single-species conservation efforts, not only because they are home for such species but also because natural communities contain important assemblages of both plant and animal species, both known and still undescribed. Why are some rare species limited to small areas? Which are the factors associated with such special habitats and where should we expect to find new populations of these species? Which other species have an important place in the functionality of such ecosystems? Which species can be used as indicators for recognizing and mapping such plant communities? On the Seychelles islands, botanists have long focused on taxonomy (Baker 1877, Christensen 1912, Friedmann 1994, Robertson 1989), and plant communities have been described in a qualitative way, DOI: 10.1127/0340 – 269X/2009/0039– 0157 based on simple observation (Carlström 1996, Gerlach 1993, Jeffrey 1962, Procter 1984b, VeseyFitzgerald 1940), or focussing on invasive species (Fleischmann et al. 2003, 2005), or not considering all strata (Gerlach 1997). As far as we know, quantitative studies on plant communities considering all strata and all vascular plant groups (i.e. including the poorly known ferns) have not been carried out in Seychelles (see also Gerlach 1993, p.19). The main forest types distinguished by most authors correspond to the three main classes already described by Vesey-Fitzgerald: the “lowland forest” (0–300 m), the “intermediate forest” (300–550 m) and the “mountain moss forest” (from 550 m to the top, at 905 m). However, the location of these intervals and the definition of vegetation types can vary widely between authors. Such vegetation belts have been reported only for the two higher islands of the Seychelles, i.e. Mahé and Silhouette. During more than two centuries of human presence, both islands have suffered extensive human impact due to forestry, agriculture and cinnamon plantations. The lowland and intermediate levels are the most highly degraded, especially on Mahé where nearly no pristine or even old growth humid forest 0340 – 269X/09/0039 – 0157 $ 08.10 © 2009 Gebrüder Borntraeger, D-14129 Berlin · D-70176 Stuttgart 158 B. Senterre et al. still occurs below 550 m. In contrast, Silhouette is undoubtedly the most pristine island for humid forest types. There, human disturbance is mainly restricted to areas below 300 m or only in limited areas above that altitude (see Vesey-Fitzgerald 1940, p.466, for detailed explanations). In the lowland belt, many areas are now covered by an old growth secondary forest, characterized by many introduced species in the upper layer but retaining many native species in the understory. In the present study, we propose an original description of the floristic variation along the altitudinal gradient in the humid forests of Silhouette. The forest types are described using both the most abundant and the most characteristic species. Reference is made to other plant communities and other names given in the literature. Study area The Seychelles archipelago comprises a group of 115 islands located in the western Indian Ocean. Of these, 41 are granitic and constitute most of the socalled inner islands, lying between 4–5° South and 55–56° East. These are the only mid-oceanic granitic islands in the world and are located about 1200 km of the African coasts, 1000 km north of Madagascar and 2800 km south-east of India. Their granitic nature denotes their continental origin as fragments of Gondwanaland that were separated millions of years ago when India drifted north toward Asia. Isolated for about 75 million years, the Seychelles now hosts a unique biota, many species being extremely primitive (Procter 1984a, WWF 2007). The majority of those species are restricted to the larger granitic islands. The main granitic islands have a combined area of about 230 km2, of which Mahé (154 km2), Praslin (37 km2), Silhouette (20 km2) and La Digue (10 km2) are the most important in terms of size. According to Senterre (2009), the climate is an intermediate between tropical wet (Af) and tropical monsoon (Am), following the Köppen’s classification (Kottek et al. 2006). The average annual rainfall is 2200 mm at sea level, but exceeding 4000 mm on the summits. Mean temperature ranges from 24 °C to 30 °C (Walsh 1984). The prevailing winds are NW monsoon (December to March) and SE trade winds (May to October) and the inner islands lie outside the cyclone belt. eschweizerbartxxx author Fig. 1. Location of the 12 plots inventoried for the present study of plant communities on Silhouette Island. Forest types on Silhouette Island 159 considered all ligneous plant individuals with a dbh t 5 cm (i.e. diameter at breast height, measured at 1.3 m above ground level, or following international standards, White & Edwards 2001) and included in a 50 x 10 m area (i.e. 500 m2). For each individual, dbh, total height and structural position (i.e. emergent, canopy or under-canopy tree; see Oldeman 1990) were recorded. Hemi-epiphytes like Ficus spp. or Schefflera procumbens were also counted in this subplot if they reached the canopy. The second subplot included all ligneous individuals belonging to the shrub stratum, i.e. small understory lianas (if any), palms, pandans, tree ferns and mainly true shrubs, all higher than 90 cm and with dbh < 5 cm. This second subplot was sampled along the same 50 m line but only 4 m wide (i.e. 200 m2) and all individuals were counted. The third subplot, sampled in the same area as subplot 2, recorded herbaceous species and included all plants < 90 cm height thus also accounting for the regeneration of species from the upper layers. Taller herbaceous species (e.g. Mapania spp.) and understory epiphytes were also included in this subplot. Individuals were not counted but the abundance coefficient was estimated using van der Maarel’s (1979) categories (Table 1). For each plot, environmental and geographical parameters were recorded: GPS coordinates, altitude, slope inclination and orientation, percentage rock cover on the ground, bryophyte cover on the ground, trunks and branches, thickness of bryophyte cover, and percentage cover of each of the strata (i.e. canopy trees, under-canopy trees, shrubs and herbs) and for palms specifically. An important variety of habitat types exists on these islands, mostly related to altitude (up to 905 m), coastal influence, soil structure, topography and perturbation. The tropical wet climate and small size of these islands do not allow the presence of either mesophilous or dry forests or related formations, except in special habitats such as “glacis”. These glacis are granite outcrops and have been considered as inselbergs (Fleischmann et al. 1996). They are very common on Silhouette and are found from sea level up to the highest point. Silhouette, the third largest island, has the second highest point of the archipelago (Mont Dauban at about 740 m) and is covered by some of the most pristine forests of Seychelles. This island is the least developed of the granitic islands and has a population of fewer than 300 people. Its flora is characterised by several island and archipelago endemics, some of which are already extinct from Mahé (Friedmann 1994). Although the geology of Silhouette differs from that of the other large islands of Seychelles (Piggott 1968), the vegetation is similar to that found on Mahé. Materials and methods The field sampling was carried out between the 11th of February and the 23rd of March 2008. Considering the local logistic and climatic constraints, this period included 23 days in the forest. Plot sampling design followed the basic principles of quantitative plant community studies. The plot disposition was chosen to study the variation of the floristic composition along the altitudinal gradient, all other ecological variants being as constant as possible, especially stand maturity and the proximity of ravine/valley areas. Each plot was within an area which was considered ecologically and structurally homogeneous. Twelve vegetation plots were studied on Silhouette Island (Fig. 1). To include all strata, each plot was subdivided into three subplots. The tree layer subplot eschweizerbartxxx author Results and discussion Within the 12 studied plots (comprising 36 subplots), we recorded 108 taxa in 95 genera and 56 families. The tree layer contained 1068 individuals of 45 taxa. Shrub subplots included 1756 individuals in 41 taxa, Table 1. Abundance-dominance coefficients, based mainly on van der Maarel (1979) and Braun-Blanquet (1932). If a species is observed only out of the plot but still in the same stand, its presence is quoted by 0.5. The categories are defined with correspondence between semi-quantitative and quantitative scales, where “f” is respectively the estimated abundance-dominance coefficient or the relative abundance of individuals. Coefficient value 0.5 Definition Average cover out of the plot 0.5 Braun-Blanquet van der Maarel 1 r r + 1 1 individual 2 2 individuals 2 + 3 f d 5% 3.5 1 1 4 5< f d 10 % 7.5 2 2m 5 10< f d 15 % 12.5 2 2a 6 15< f d 25 % 20 2 2b 7 25< f d 50 % 37.5 3 3 8 50< f d 75 % 62.5 4 4 9 f > 75 % 87.5 5 5 160 B. Senterre et al. and the herb layer included 92 taxa, 29 of which were ferns and fern allies. To identify floristic groups in the dataset, we performed a Detrended Correspondence Analysis (DCA, Legendre & Legendre 1998) based on data synthesized from all strata (Fig. 2). The results indicate that floristic variability is highly correlated with the first ordination axis, which explains already 23.4 % of the total floristic variability. A Canonical Correspondence Analysis (CCA, Legendre & Legendre 1998) carried out on the same dataset, with altitude as an environmental variable, showed that this factor is highly correlated with the first ordination axis, explaining 23.2 % of the floristic variability (p = 0.001). We conclude that altitudinal belts correspond to distinct plant communities and that four groups of plots could be considered for the indicator species analysis. The relatively isolated position of plot 9 on the second DCA axis, could indicate the relatively low level of disturbance and invasion of alien species in this plot within the lowland group. What is the floristic relationship between the different floristic groups, and which species are peculiar to some groups, i.e. which can be regarded as indicator species? To investigate the first question, we used a classification method based on the Bray-Curtis similarity index and UPGMA (Unweighted Pair Group Method with Arithmetic mean, Fig. 3). The results were not totally concordant for the distinct strata, i.e. using the herb layer submontane (“intermediate forests”) and lower montane (“mountain moss forest”) belts were more clearly distinct than when the upper tree layer was used. In fact, the difference between these two belts was marked mainly by the presence of Glionnetia sericea within an otherwise impoverished tree flora, while the understory flora was much more differentiated between these two altitudinal belts (see Table 2). This observation underlines the importance of considering the herb layer for forest type classification. Both results show that the submontane forests have a higher floristic affinity with the lower montane forests than with the lowland. Therefore, we considered four groups in the analysis of indicator species, using IndVal (Dufrêne & Legendre 1997) with a 4-levels typology synthesised from the ordination and classification results. The justification of each group at different levels of the hierarchy was estimated by the number and/or goodness of indicator species and a two-way table is produced emphasizing such indicator species (Table 2). Results of the IndVal analysis, detailed in the Table 2, confirm the differentiation of three altitudinal belts in the rain forests of Silhouette. The two submontane variants share only few species, including typically submontane species and excluding the most important lower montane ones. Nevertheless, the upper variant of this submontane group (plots near Mon Plaisir) included some lower montane species probably because of its somewhat transitional position at 500–540 m altitude. At Mon Plaisir, these lower montane species are mainly present as fallen epiphytes. eschweizerbartxxx author We therefore consider these two submontane variants as belonging to the same plant community. More vegetation plots are required to clarify the variability in the submontane belt. Hereafter, we describe forest types distinguished in the present study and make reference to the literature on Seychelles and to worldwide synthesis of mountain vegetation habitats. The nomenclature used here for the three forest types distinguished is based on review of worldwide studies of vegetation types and their nomenclature (Areces-Mallea et al. 1999, Aubréville 1965, Camirand & Evelyn 2003, Grossman et al. 1998, Mueller-Dombois & Ellenberg 1974) and implemented using more detailed studies of altitudinal belts in the tropics (Ashton 2003, Boughey 1955a, 1955b, 1965, Bruijnzeel & Hamilton 2000, Bussmann 2006, Churchill et al. 1995, Grimshaw 2001, Hedberg 1951, Letouzey 1968, Ohsawa 1993, Schnell 1976, 1977, Trochain 1980, Van Steenis 1935). In addition, one of us recently revised the definitions, homologies and nomenclature of forest types in the tropical mountains worldwide (Senterre 2005b) focussing on the lower belts. Altitudinal belts have often been misinterpreted, especially with respect to the submontane belt (Ashton 2003, White 1978), for reasons related to levels of perturbation (Tchouto et al. 1999, Thomas & Achoundong 1994), heterogeneous transect data, and a lack of consideration of ecological transgressors, i.e. species narrowly associated with more than one habitat, and ecological equalizations (Senterre 2005b). The vegetation belts and their definition, as understood here, are synthesized in our previous work (Senterre 2005b): lowland, submontane, lower montane (may include two mostly physiognomic subdivisions), upper montane, subalpine and alpine. From these, only the lowland, submontane and lower montane belts are reported from Silhouette and the Seychelles islands overall. For each forest type, a conservation rank was estimated using categories proposed in the international vegetation classification system (Grossman et al. 1998). We also accounted for information on the quantitative data from our plots, field observations and GIS (Geographic Information System) data. Our personal observations on the transition from one altitudinal belt to another in Seychelles are based on extensive field explorations on Silhouette, mainly from La Passe to Mon Plaisir, Mont Pot à Eau, Mont Dauban, from Grand Barbe to Rende D’Avance, Gratte Fesse, Mont Cocos Marrons, on the northern slopes of Mont Dauban and on the eastern and northern slopes of Grand Congoman. We also explored the Congo Rouge – Morne Seychellois – Pérard area and Montagne Brûlée area, on Mahé. Throughout these explorations, we focussed on the identified indicator species. The transitions appeared to be always marked at the same altitude and within an altitudinal interval of less than 50 m. Forest types on Silhouette Island 161 Fig. 2. Ordination of the floristic data (DCA, Detrended Correspondance Analysis) compiling all the strata (12 plots, 108 taxa), with van der Maarel coefficient indexes. The first two axes of the ordination explain 30.7 % of the total variance of floristic data (sum of all eigenvalues = 2.565). The first axis alone explains already 23.4 % of the total variance. Samples names are: L#, lowland; S#, submontane; M#, lower montane. eschweizerbartxxx author Fig. 3. Classification using Bray-Curtis similarity index and UPGMA (Unweighted Pair Group Method with Arithmetic mean) (a) for the tree layer and (b) for herb layer (using van der Maarel coefficient indexes). 162 B. Senterre et al. Table 2. Two-way table produced with IndVal (Dufrêne & Legendre 1997) for all strata, i.e. 108 species in 12 plots (made of 36 subplots). Species are ordered following their IndVal value for the floristic groups identified. The most relevant species are emphasized in bold. The values in the table are, for a given species, on the left, the sum of van der Maarel index values for all plots of a plant community and, on the right, the number of plots with presence of the species in the concerned plant community. The “Lowland” plots are the number 1 (180 m a.s.l.), 6 (90 m), 8 (300 m), 9 (80 m); “Submont1” plots are the number 7 (400 m) and 10 (380 m); “Submont2” plots are the number 2 (540 m), 3 (500 m), 4 (580 m); “Montane” plots are the number 5 (640 m), 11 (590 m) and 12 (590 m). The origin of the species (“Orig.”) is abbreviated as follows: end. = endemic to Seychelles; ind. = indigenous; inv. = invasive; inv.p. = potentially invasive; nat. = naturalized but non invasive. The stratum (“Str.”) to which the species belongs to is mentioned: A = canopy tree; Ad = under-canopy tree; ar = shrub; H = herbaceous understory. Species Family Orig. Str. IndVal Lowland Submont1 Submont2 Lower Montane 1. Characteristic species of the Calophyllum inophyllum / Deckenia nobilis / Psilotum nudum lowland rainforests Hypoxidia rhizophylla (Baker) F.Friedmann Nephrosperma vanhoutteanum (H.Wendl. ex vanHoutt.) Balf.f. Paraserianthes falcataria (L.) I.C.Nielsen Cocos nucifera L. Pyrostria bibracteata (Baker) Cavaco Tabebuia pallida (Lindl.) Miers Memecylon elaeagni Bl. Allophyllus pervillei Blume Dracaena reflexa Lam. var. angustifolia Baker Adenanthera pavonina L. Anacardium occidentale L. Deckenia nobilis H.Wendl. ex Seem. Flagellaria indica L. Morinda citrifolia L. Wielandia elegans Baill. Calophyllum inophyllum L. Oplismenus compositus (L.) P.Beauv. Phymatosorus scolopendria (Burm.f) Pic.Serm. Abrus precatorius L. subsp. africanus (Vatke) Verdc. Albizia lebbeck (L.) Benth. Artocarpus heterophyllus Lam. Averrhoa bilimbi L. Cananga odorata (Lam.) Hook.f. & Thomson Casuarina equisetifolia L. Citrus reticulata Blanco Hevea brasiliensis (Kunth) Müll.Arg. Melia dubia Cav. Psilotum nudum (L.) P.Beauv. Pterocarpus indicus Willd. Sandoricum koetjape (Burm.f.) Merr. Tacca leontopetaloides (L.) Kuntze Tarenna sechellensis (Baker) Summerh. Selaginella fissidentoides (Hook. & Grev.) Spring Hypoxidaceae Arecaceae end. end. H Ad 100 86 9./ 4 16./ 4 0./ 0 4./ 2 0./ 0 0./ 0 0./ 0 1./ 1 Mimosaceae Arecaceae Rubiaceae Bignoniaceae Melastomataceae Sapindaceae Dracaenaceae Mimosaceae Anacardiaceae Arecaceae Flagellariaceae Rubiaceae Euphorbiaceae Clusiaceae Poaceae Polypodiaceae Fabaceae Mimosaceae Moraceae Oxalidaceae Annonaceae Casuarinaceae Rutaceae Euphorbiaceae Meliaceae Psilotaceae Fabaceae Meliaceae Taccaceae Rubiaceae Selaginellaceae inv. ind. ind. inv. end. ind. ind. nat. nat. end. ind. nat. ind. ind. ind. ind. ind. nat. nat. nat. nat. nat. nat. inv.p. nat. ind. nat. inv.p. nat. end. ind. A Ad ar A ar ar ar A Ad A H ar ar A H H H A A Ad A A Ad A A H A A H ar H 76 75 75 75 72 70 69 59 50 50 50 50 50 47 39 35 25 25 25 25 25 25 25 25 25 25 25 25 25 25 19 8./ 12./ 12./ 14./ 12./ 7./ 11./ 11./ 4./ 4./ 3./ 5./ 2./ 7./ 7./ 6./ 2./ 1./ 3./ 1./ 1./ 1./ 2./ 9./ 1./ 3./ 1./ 8./ 3./ 3./ 3./ 4./ 0./ 0./ 0./ 1./ 1./ 6./ 6./ 0./ 0./ 0./ 0./ 0./ 1./ 3./ 0./ 0./ 0./ 0./ 0./ 0./ 0./ 0./ 0./ 0./ 0./ 0./ 0./ 0./ 0./ 0./ 0./ 0./ 0./ 0./ 0./ 0./ 2./ 0./ 0./ 0./ 0./ 0./ 0./ 0./ 1./ 2./ 0./ 0./ 0./ 0./ 0./ 0./ 0./ 0./ 0./ 0./ 0./ 0./ 0./ 0./ 2./ 1./ 0./ 0./ 0./ 0./ 0./ 2./ 0./ 0./ 0./ 0./ 0./ 0./ 0./ 0./ 3./ 0./ 0./ 0./ 0./ 0./ 0./ 0./ 0./ 0./ 0./ 0./ 0./ 0./ 0./ 0./ 1 0 0 0 0 0 1 0 0 0 0 0 0 0 0 1 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 eschweizerbartxxx author 4 3 3 3 3 3 4 3 2 2 2 2 2 2 2 2 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 2 0 0 0 1 1 2 2 0 0 0 0 0 1 1 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 1 0 0 0 0 0 0 0 1 1 0 0 0 0 0 0 0 0 0 0 0 0 0 0 1 2. Characteristic species of the Northea hornei / Roscheria melanochaetes / Haplopteris ensiformis cloud forests (submontane + lower montane) Northea hornei (M.M.Hartog) Pierre Dillenia ferruginea (Baill.) Gilg Lindsaea kirkii Hook. Mapania floribundum (Nees ex Steud) Koyama Roscheria melanochaetes (H.Wendl.) H.Wendl. ex Balf.f. Elaphoglossum hornei C.Chr. Protarum sechellarum Engl. Trichomanes cupressoides Desv. Elaphoglossum lancifolium (Desv.) C.V.Morton Haplopteris ensiformis (Sw.) E.H.Crane Canthium carinatum (Baker) Summerh. Platylepis occulta (Thouars) Rchb.f. Sapotaceae Dilleniaceae Dennstaedtiaceae Cyperaceae Arecaceae end. end. end. end. end. A A H H Ad 100 88 88 88 85 0./ 0./ 0./ 0./ 3./ 0 0 0 0 1 7./ 1./ 4./ 5./ 9./ 2 1 1 2 2 11./ 9./ 11./ 12./ 14./ 3 3 3 3 3 15./ 8./ 10./ 7./ 12./ 3 3 3 2 3 Lomariopsidaceae Araceae Hymenophyllaceae Lomariopsidaceae Vittariaceae Rubiaceae Orchidaceae end. end. ind. ind. ind. end. ind. H H H H H ar H 75 75 74 63 50 25 25 0./ 0./ 2./ 0./ 0./ 0./ 0./ 0 0 1 0 0 0 0 0./ 2./ 3./ 0./ 2./ 0./ 0./ 0 1 1 0 1 0 0 9./ 6./ 11./ 7./ 5./ 1./ 3./ 3 2 3 3 2 1 1 9./ 9./ 8./ 5./ 3./ 1./ 2./ 3 3 3 2 1 1 1 2.1 Characteristic species of the Drypetes riseleyi / Lomariopsis pervillei - Trichomanes fulgens submontane forests Pandanus hornei Balf.f. Lomariopsis pervillei Mett. ex Kuhn Cynanchum callialatum Buch.-Ham. ex Wight Drypetes riseleyi Airy Shaw Ixora pudica Baker Scleria sieberi Ness ex Kunth Pandanaceae Lomariopsidaceae Asclepiadaceae Euphorbiaceae Rubiaceae Cyperaceae end. end. ind. end. end. ind. A H H A Ad H 87 79 50 50 50 35 0./ 0./ 0./ 0./ 0./ 0./ 0 0 0 0 0 0 9./ 5./ 1./ 1./ 3./ 3./ 2 2 1 1 1 1 2./ 2./ 0./ 0./ 0./ 0./ 1 1 0 0 0 0 0./ 0./ 0./ 0./ 0./ 2./ 0 0 0 0 0 1 Forest types on Silhouette Island 163 Species Family Orig. Str. IndVal Ludia mauritiana J.F.Gmel. var. sechellensis F.Friedmann Verschaffeltia splendida H.Wendl. Trichomanes fulgens C.Chr. Syzygium jambos (L.) Alston Bolbitis bipinnatifida (Kuhn) Ching Cyathea sechellarum Mett. Haplopteris scolopendrina (Bory) C.Presl Phaius tetragonus (Thouars) Reichb.f. Polystichopsis wardii (Bak. in Hook. & Baker) Tardieu Procris insularis H.Schröter Schefflera procumbens (Hemsl.) F.Friedmann Sphaerostephanos subtruncatus Holttum Merremia peltata (L.) Merr. Psychotria pervillei Baker Angiopteris evecta (G.Forst.) Hoffm. Asplenium protensum Schrad. Syzygium aromaticum (L.) Merr. & L.M.Perry Trema orientalis (L.) Blume Flacourtiaceae Arecaceae Hymenophyllaceae Myrtaceae Lomariopsidaceae Cyatheaceae Vittariaceae Orchidaceae Dryopteridaceae Urticaceae Araliaceae Thelypteridaceae Convolvulaceae Rubiaceae Marattiaceae Aspleniaceae Myrtaceae Ulmaceae end. end. end. inv. ind. end. ind. ind. end. end. end. ind. inv. end. ind. ind. nat. ind. Ad A H Ad H ar H H H H A H A ar H H Ad Ad 33 30 80 40 67 67 67 67 67 67 67 60 53 36 33 33 33 33 Lowland Submont1 Submont2 1./ 4./ 0./ 0./ 0./ 0./ 0./ 0./ 0./ 0./ 0./ 1./ 1./ 1./ 0./ 0./ 0./ 0./ 1./ 4./ 2./ 1./ 0./ 0./ 0./ 0./ 0./ 0./ 0./ 0./ 0./ 0./ 0./ 0./ 0./ 0./ 0./ 0./ 9./ 1./ 7./ 7./ 6./ 3./ 8./ 3./ 3./ 7./ 3./ 2./ 4./ 4./ 2./ 1./ 1 1 0 0 0 0 0 0 0 0 0 1 1 1 0 0 0 0 1 1 1 1 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 3 1 2 2 2 2 2 2 2 2 2 2 1 1 1 1 Lower Montane 0./ 0 1./ 1 0./ 0 0./ 0 0./ 0 0./ 0 0./ 0 0./ 0 0./ 0 0./ 0 0./ 0 0./ 0 0./ 0 1./ 1 0./ 0 0./ 0 0./ 0 0./ 0 2.2 Characteristic species of the Glionnetia sericea / Gastonia crassa / Elaphoglossum macropodium lower montane forests Ctenopteris albobrunnea (Baker) Tardieu Gastonia crassa (Hemsl.) F.Friedmann Glionnetia sericea (Baker) Tirv. Nepenthes pervillei Bl. Timonius sechellensis Summerh. Pandanus sechellarum Balf.f. Hymenophyllum polyanthos (Sw.) Sw. Hymenophyllum hygrometricum (Poir.) Desv. Agrostophyllum occidentale Schltr. Colea seychellarum Seem. Elaphoglossum macropodium (Fée) T.Moore Gynura sechellensis (Baker) Hemsl. Syzygium wrightii (Baker) A.J.Scott Xiphopteris serrulata (Sw.) Kaulf. Paragenipa wrightii (Baker) F.Friedmann Erythroxylum sechellarum O.E.Schulz Rumohra adiantiformis (G.Forst.) Ching Gleichenia linearis (Burm.f.) C.B.Clarke Humata repens (L.f.) J.Small ex Diels Aphloia theiformis (Vahl) Benn. subsp. madagascariensis (Clos) H.Perr. var. seychellensis (Clos) F.Friedmann Curculigo sechellensis Boj. Asplenium caudatum G.Forst. Bulbophyllum intertextum Lindl. Craterispermum microdon Baker Hederorkis seychellensis Bosser Mapania seychellaria D.A.Simpson Microsorum punctatum (L.) Copel. Polystachya sp.1 Psilotum complanatum Sw. Psychotria sp.1 Senterre Psychotria sp.2 Senterre cf. dupontiae Seychellaria thomassetii Hemsl. Grammitidaceae Araliaceae Rubiaceae Nepenthaceae Rubiaceae Pandanaceae Hymenophyllaceae Hymenophyllaceae Orchidaceae Bignoniaceae Lomariopsidaceae Asteraceae Myrtaceae Grammitidaceae Rubiaceae Erythroxylaceae Davalliaceae Gleicheniaceae Davalliaceae Flacourtiaceae end. end. end. end. end. end. ind. ind. end. end. ind. end. end. ind. end. end. ind. inv. ind. end. H Ad A H Ad A H H H Ad H H Ad H ar ar H H H Ad 100 100 100 100 100 90 88 87 67 67 67 67 67 67 66 60 53 51 51 48 0./ 0./ 0./ 0./ 0./ 0./ 0./ 0./ 0./ 0./ 0./ 0./ 0./ 0./ 4./ 0./ 0./ 0./ 0./ 0./ 0 0 0 0 0 0 0 0 0 0 0 0 0 0 1 0 0 0 0 0 0./ 0./ 0./ 0./ 0./ 2./ 0./ 0./ 0./ 0./ 0./ 0./ 0./ 0./ 2./ 1./ 0./ 0./ 0./ 0./ 0 0 0 0 0 1 0 0 0 0 0 0 0 0 1 1 0 0 0 0 0./ 0./ 0./ 0./ 0./ 0./ 3./ 3./ 0./ 0./ 0./ 0./ 0./ 0./ 0./ 0./ 3./ 2./ 3./ 4./ 0 0 0 0 0 0 1 1 0 0 0 0 0 0 0 0 1 1 1 1 9./ 7./ 12./ 10./ 9./ 11./ 13./ 12./ 5./ 2./ 7./ 3./ 5./ 5./ 8./ 5./ 7./ 4./ 6./ 6./ 3 3 3 3 3 3 3 3 2 2 2 2 2 2 3 2 2 2 2 2 Hypoxidaceae Aspleniaceae Orchidaceae Rubiaceae Orchidaceae Cyperaceae Polypodiaceae Orchidaceae Psilotaceae Rubiaceae Rubiaceae Triuridaceae end. ind. ind. end. end. end. ind. H H H A H H H ind. end. end. ind. H ar ar H 45 33 33 33 33 33 33 33 33 33 33 33 2./ 0./ 0./ 0./ 0./ 0./ 0./ 0./ 0./ 0./ 0./ 0./ 1 0 0 0 0 0 0 0 0 0 0 0 3./ 0./ 0./ 0./ 0./ 0./ 0./ 0./ 0./ 0./ 0./ 0./ 2 0 0 0 0 0 0 0 0 0 0 0 0./ 0./ 0./ 0./ 0./ 0./ 0./ 0./ 0./ 0./ 0./ 0./ 0 0 0 0 0 0 0 0 0 0 0 0 7./ 3./ 2./ 1./ 4./ 2./ 2./ 2./ 1./ 4./ 2./ 3./ 2 1 1 1 1 1 1 1 1 1 1 1 Cinnamomum verum J.Presl. Clidemia hirta (L.) D.Don Phoenicophorium borsigianum (K.Koch.) Stuntz Nephrolepis biserrata (Sw.) Schott Planchonella obovata (R.Br.) Pierre Psidium cattleianum Sabine Grisollea thomassetii Hemsl. Lauraceae Melastomataceae Arecaceae Oleandraceae Sapotaceae Myrtaceae Icacinaceae 100 92 92 75 67 67 25 21./ 14./ 24./ 5./ 6./ 15./ 1./ 4 4 4 2 3 4 1 17./ 4./ 13./ 4./ 5./ 0./ 1./ 2 1 2 1 2 0 1 25./ 21./ 13./ 11./ 6./ 11./ 0./ 3 3 3 3 2 2 0 24./ 17./ 7./ 8./ 1./ 7./ 2./ 3 3 2 3 1 2 1 eschweizerbartxxx author 3. Ubiquist species inv. inv. end. ind. ind. inv. end. A ar Ad H A Ad Ad 164 B. Senterre et al. Tropical ombrophilous lowland forest (Lowland rain forest) Synonyms: Lowland forest (Fleischmann et al. 2003, Gerlach 1993, Vesey-Fitzgerald 1940); Natural intermediate forest, pro parte (Gerlach 1993, within the Mid-altitude forest); forêt de basse altitude, facies à Mimusops sechellarum et Vateriopsis seychellarum (Friedmann 1994). Physiognomy: The lowland forests are characterised by a high canopy, at about 25–30 m, and commonly emergent trees up to 30–50 m, giving the canopy an undulated aspect. The main four strata are typically well differentiated: (1) the upper tree layer (including sub-canopy trees, canopy trees and emergent trees); (2) the under canopy trees; (3) the shrub layer; and (4) the herb layer (Photo 1). Epiphytes and lithophytes are rare or not developed, or localized in wet microhabitats (e.g. in the shade of big granite boulders). Floristics: The lowland forests are well differentiated not only by their remarkable abundance of nonindigenous species (Adenanthera pavonina, Hevea brasiliensis, Paraserianthes falcataria, Sandoricum koetjape, Tabebuia pallida) and more or less scattered coconut trees (Cocos nucifera), but also by some na- tive indicator species such as Calophyllum inophyllum, Casuarina equisetifolia, Deckenia nobilis (in the upper tree layer), Nephrosperma vanhoutteanum (in the under-canopy tree layer), Allophyllus pervillei, Memecylon elaeagni, Pyrostria bibracteata, Wielandia elegans (in the shrub layer), Hypoxidia rhizophylla, Psilotum nudum, Tacca leontopetaloides (in the herb layer). The main diagnostic species are Calophyllum inophyllum, Deckenia nobilis, Wielandia elegans, Tacca leontopetaloides, and most of the non-indigenous timber trees. In addition, lowland rain forests are easily recognized from higher belts by the absence of the most characteristic species of both submontane and lower montane belts, e.g. Northea hornei, Roscheria melanochaetes, and Haplopteris spp. Nevertheless, if some of these (sub)montane species are present, they are marginal and most probably associated with wetter microhabitat (ravine/valley, rock shelter, etc.) revealing ecological equalizations (Budowski 1965, Rübel 1935, Senterre 2005b), or are relict populations from past climate changes. Ecology: On all the areas explored, both windward or leeward the lowland belt extended from about 20 m up to about 300–350 m altitude. It is well developed on non-superficial soils, with 0–50(-75) % cover of granite blocks. eschweizerbartxxx author Photo 1. Lowland rain forest at plot 1, with many Tabebuia pallida trees, some Cocos nucifera, and a quite diverse understory with still a lot of native species. Forest types on Silhouette Island 165 Distribution: It has been reported, in the Seychelles inner islands, mainly on Mahé, Praslin, Silhouette, La Digue, Marianne, Curieuse and Félicité. On Silhouette, a more or less pristine patch of this forest type was observed on the hills between Grand Barbe and Mont Cocos Marrons (plot 9, see DCA results). Some other well conserved areas are expected on the northern slopes of Mare aux Cochons, the south-western slope of main Mont Dauban (probably never explored), and near Pointe Civine (Fig. 4). At all other sites the lowland forest was semi-natural, mainly dominated by non native timber tree species. On Silhouette most areas are old growth secondary forests, i.e. closed canopy with late secondary species (Oldeman 1990), still hosting many native species. The lowland rain forests are mixed with other vegetation types like the open glacis formations, the ravine/ valley forests variants, and secondary eco-units (secondary woodlands and pioneer scrub habitats). Variability and related vegetation types: Occasionally one or two non-indigenous species can become dominant providing a floristically and physiognomically distinct appearance. In such places, the conservation value, vulnerability and management of lowland forest may differ with more pristine parts of the forest. Therefore, some authors have decided to consider highly invaded forests as distinct forest types. Gerlach (1993) described three main types within his “mid-altitude forests”: 1-”tree plantations”, characterised by non native timber species (on Silhouette, mainly Sandoricum koetjape and Tabebuia pallida) or rubber trees (Hevea brasiliensis); 2-Psidium littorale forest; and 3-Cinnamomum verum forest (= “Cinnamon coppice”, Vesey-Fitzgerald 1940). Off course, all intermediates could exist from a non-invaded forest to a highly invaded one, but such distinctions are useful in a conservation perspective, for the reason already mentioned. On Silhouette, ancient tree plantations and Psidium littorale forests are more common at low altitude while Cinnamomum verum forests are mainly found in the submontane belt. Nevertheless, apart from tree plantations, these forest types can be observed from sea level to the higher summits. Where perturbations have occurred in the past (tree fall gaps, forest fire, logging), early to late secondary forests are found in relatively localized stands, sharing some species from the old growth mature lowland rain forests, mainly in the understory, mixed with successional species. These successional eco-units are still poorly understood in Seychelles and have caused much confusion in previous classifications. Nevertheless, it seems that two main variants could be distinguished: the “secondary palmetum” and the “Agati woodland”. The “secondary palmetum” of Vesey-Fitzgerald (1940) is mainly characterised by the abundance and size of three palm species (Nephrosperma vanhoutteanum, Phoenicophorium borsigianum and especially Deckenia nobilis) occasionally associated with Pandanus hornei. Because of the common organisa- eschweizerbartxxx author tion of these palm species into a single layer or a few strata, such secondary eco-units appear to be associated with fire events or rare tropical storms. This forest type is quite localized on Mahé and Silhouette but can cover more extensive areas on Praslin. It can also be observed in the submontane belt, i.e. on Montagne Brûlée on Mahé. Several species of this community are shared with quite distinct plant communities, associated with distinct environmental conditions and characterised by a distinct physiognomy, i.e. the “dry land type of Palmetum”, or Deckenia-Memecylon society (Vesey-Fitzgerald 1940). The dry land type of Palmetum occurs on unequal sized blocks divided from each other by “Gleichenia-brake” or bare slabs of rock, where rapid drainage through a coarse, gritty soil caused conditions of physiological dryness. This vegetation type includes some of the glacis vegetation units defined by Procter (1984b) and Gerlach (1993). It corresponds mainly to what Friedmann (1994) called “forêt de basse altitude, facies à Dillenia ferruginea, Deckenia, Phoenicophorium”. The ecological transgression (Senterre 2005b, White 1978, 1979) of species from such dry habitats to secondary eco-units of humid forest, especially fire-disturbed, is a phenomenon often observed in other tropical regions (Budowski 1965, Senterre et al. 2004). Note that the term “palm forest” have been used for different plant communities in Seychelles, often without making any distinctions between the range of variants, i.e. Verschaffeltia ravine/valley forests, Deckenia secondary Palmetum, Deckenia-Memecylon dry land Palmetum, etc. The other secondary variant distinguished is dominated by fast-growing naturalised trees, i.e. Adenanthera pavonina and Paraserianthes falcataria (the “Agati woodland” of Vesey-Fitzgerald). Here the term woodland gives a very good picture of this forest type, characterized by an irregular canopy layer, not entirely closed, with high emergent trees and an understory with much light penetration. This forest type covers extensive areas in the lowland but also in the submontane belts. It seems related to local perturbations such as tree fall gaps (more frequent on steep slopes) or logging. Along the ravines and valleys that run through the humid forests, with or without water on the ground surface, a distinct lowland forest type develops. This ravine/valley forest is mainly characterised by the palm Verschaffletia splendida and the screw palm Pandanus hornei, both of which have stilt roots as a physiognomic adaptation to such habitats (Richards 1952, Schnell 1976, Senterre 2005b, Senterre & Lejoly 2001). This community is also developed in the submontane and lower montane ravine/valley rain forests, though associated with a different suit of understory species (see next section). In addition, the lowland ravine/valley rain forests are often characterised by Barringtonia racemosa (L.) Spreng. (Friedmann 1994: 189). At the coast between 0 and 20 m altitude, the lowland rain forest turns into another plant community: 166 B. Senterre et al. the coastal rain forest. This forest type, shaped by the presence of the sea can be considered as a distinct entity, which is characterised by many species of the inland lowland rain forest (as in other tropical islands: Keppel 2002) coupled with the natural presence of Cocos nucifera, with more abundant Casuarina equisetifolia, and some more characteristic species like Barringtonia asiatica (L.) Kurz and Terminalia catappa L. (Friedmann 1994: 187, 337). Pisonia grandis R.Br. is another coastal forest species but may be more associated with the presence of bird populations (see Friedmann 1994: 110) and could therefore be characteristic of what Gerlach (1993) called the “seabird island coastal forest”. The coastal forests were called the “coastal plateau forest” by Procter (1984b), the “typical coastal forest” by Gerlach (1993), or the “littoral forest” (Fleischmann et al. 2003). These coastal forests should not be confused with the coastal swamp forests characterised mainly by Heritiera littoralis Ait. (see “lowland marsh” of Gerlach 1993). Such coastal swamp forests are localised in the coastal swamps, at the back of the mangrove, and sometimes penetrate inland along the lower course of rivers. Mangrove forest types constitute a totally distinct floristic unit, including several distinct forms (Chapman 1976, Macnae 1971, Pascal et al. 2001). Conservation rank: G1-GH (critically imperilled or historically eliminated: Grossman et al. 1998). Within the four lowland forest plots, we recorded 49 taxa of which 29 are indigenous (59 %). About 45 % of these indigenous taxa are endemic to Seychelles. Although the diversity of indigenous taxa is higher than that of exotic species, native trees comprised only 25 % of the total tree abundance. Undisturbed lowland rain forest on Silhouette is considered extremely rare. Restoration efforts are deemed vital to maintain this forest type, although detailed information on the original forest composition is absent. We estimate the original area of lowland rain forests to about 1309 ha (66 % of Silhouette island, Fig. 4), nowadays reduced to 5–20 % of more or less undisturbed fragmented stands. Tropical ombrophilous submontane forest (Submontane rain forests) Synonyms: Natural intermediate forest, pro parte (Gerlach 1993, within the Mid-altitude forest); High-altitude forest, pro parte but mainly (Gerlach 1993); forêt de basse altitude (Friedmann 1994); Moist eschweizerbartxxx author Photo 2. Submontane rain forest at plot 10 with a lot of Northea hornei in the tree layer, next to many cinnamon trees. We can also see the typical undercanopy tree Roscheria melanochaetes. Forest types on Silhouette Island 167 forest slopes and valleys habitats (Procter 1984b: type 1d); Intermediate forest (Vesey-Fitzgerald 1940: the typical form); slopes and valleys of the intermediate forests, pro parte (Fleischmann et al. 2003). In other tropical regions, homologous vegetation types have received multiple, sometimes very confusing names: lower montane forest (Bruijnzeel & Hamilton 2000, Whitmore 1998), premontane rain forest or wet forest (many authors in central America, e.g. Holdridge 1967), transition or intermediate forests (in most tropical studies), cloud forest (mostly by authors not differentiating submontane and lower montane belts). However, the identity of submontane forest as a distinct altitudinal belt rather than a transition zone is clear, at least in most tropical areas (Senterre 2005b) and in Seychelles as demonstrated here. Physiognomy: The physiognomy (Photo 2) is similar to that of lowland rain forest, except that emergent trees are less common. The four strata are still well differentiated and the understory is marked by the presence of relatively abundant vascular epiphytes, on the trunk base and extending to rock surfaces, i.e. lithophytes. The tree ferns may appear very locally, especially near ravines, but are not abundant and their size is quite small. Bryophyte patches are localized on some tree trunks and never form a continuous cover. They may be more abundant in ravine/valley variants of the submontane forest, but then their thickness is still less than 1 cm. Floristics: The upper tree stratum is dominated by Northea hornei and Dillenia ferruginea, more or less mixed with cinnamon trees. The under-canopy layer is mainly dominated by Roscheria melanochaetes, Phoenicophorium borsigianum and cinnamon trees and, locally, by Pandanus hornei and Verschaffeltia splendida. The main species of the shrub stratum includes cinnamon with some scattered Psychotria pervillei and locally abundant Ixora pudica. Non shrub species of the understory include most of the characteristic submontane species: Lomariopsis pervillei, Trichomanes fulgens, Haplopteris scolopendrina and Grammitis pervillei (Mett. ex Kuhn) Tardieu (the last one only observed out of the plots). Therefore, the submontane rain forests can be distinguished from the lowland rain forests by the abundance of Northea hornei, Roscheria melanochaetes and Haplopteris spp. They can be distinguished from the lower montane belt, on the one hand, by some preferential submontane species such as those cited here for the understory and, on the other hand, by the absence of the characteristic lower montane species. Within the tree strata, we consider only two species as characteristic submontane species, i.e. Craterispermum microdon and Drypetes riseleyi, even though our data did not emphasize that observation, these species being quite rare and also marginally present in lower montane rain forests. On Silhouette, they are more abundant on the southern part of Jardin Marron and Grand Congoman, which are probably the best remnants of submontane rain forests in Seychelles. eschweizerbartxxx author Ecology: The lower limit of the submontane rain forest has mostly been removed, especially on Mahé. We had the opportunity to observe it very clearly on the following sites on Silhouette: on the eastern slopes of Grand Congoman, on the slopes of Rende D’Avance near Gratte Fesse, on the northern slope of Mont Cocos Marrons, and on the northern slope of Mont Dauban. In all sites, the first species appearing, at about 300 m, are Northea hornei, Roscheria melanochaetes, Haplopteris ensiformis, Lindsaea kirkii and Mapania floribundum, thus species common to submontane and lower montane forests. The submontane belt is generally fully developed at 350 m and is very homogeneous up to about 550 m, where the transition to the lower montane belt occurs. Distribution: This vegetation type once occupied extensive areas on Mahé and Silhouette. On Silhouette, the best representatives are found, from the most pristine to the least pristine, at: Grand Congoman, the southern part of Jardin Marron, the eastern slopes (unexplored) and northern slopes of Mont Dauban and the northern slope of Mont Cocos Marrons. On Mahé, we do not know of any locality for a well conserved submontane rain forest. The submontane rain forest has not been recorded for Praslin (summit at 367 m) or La Digue (333 m). Nevertheless, some submontane indicators are cited for the flora of both islands, especially Praslin, and, therefore, further studies are needed in order to examine the exact position and ecology of such submontane indicators, focussing on fern species (not considered by previous authors). Variability and related vegetation types: The variants invaded by alien species, and distinguished for the lowland belt, can occur also in the submontane belt, having very similar floristic and physiognomic patterns. Nevertheless, understory species, including ferns and herbs, will differ from lowland to submontane invaded variants. In addition, soils and ecology will differ. Therefore, even though lowland and submontane invaded forests presents very similar characteristics, they should be differentiated based on their understory flora and/or ecological parameters (mainly altitude) and considered as distinct units for conservation/restoration actions. In addition, another type of invaded forest has been observed in the submontane belt only and is based on Syzygium jambos. This species already invaded quite large areas on Mahé and is now also becoming a problem on Silhouette (especially in the eastern part of the Mont Dauban ridge). The successional stages in the submontane belt present mainly common characteristics with homologous eco-units in the lowland belt. We found, for example at Montagne Brûlée (Mahé), a forest type related to the “secondary palmetum” cited for the lowland variants, but differing clearly by the presence of Roscheria melanochaetes and most of the understory (sub)montane indicators. On Silhouette, the submontane forests have been degraded mainly on 168 B. Senterre et al. the slopes of Jardin Marron, where we now can see a submontane form of the “Agati woodland”. This forest type is also well developed on the northern and southern faces of the Mont Dauban ridge, where the steep slopes promote tree fall gaps and natural succession to take place. Even more common than in the lowland belt, submontane ravine/valley rain forests alternate with the typical submontane rain forests. This submontane ravine/valley variant differs from the lowland one by the absence of Barringtonia racemosa, disappearing at about 300 m, and by the simultaneous appearance of abundant populations of Begonia seychellensis Hemsl., Cyathea sechellarum, Sphaerostephanos subtruncatus, Angiopteris evecta. The tree layer of these ravine/valley forests is often irregularly closed. Pandanus hornei and Verschaffeltia splendida are especially characteristic, accompanied by other casual (sub)montane species like Northea hornei. This forest type has been highly degraded and has a very high conservation value. Indeed, we collected several unknown species in the submontane ravine/valley rain forests of the northern face of Mont Dauban. Some fauna elements also seem limited to this forest type, i.e. between Gratte Fesse and Mont Corgat (Silhouette, Gerlach 2003: 27). This forest type includes the “Pandanus hornei-Verschaffeltia splendida society” of Vesey-Fitzgerald, the “Cyathea seychellarum patches” of Gerlach (1993), and the “moist forest ravines habitat” of Procter (1984b: type 1c). In only a few places, the topography allows the development of submontane swamp forests, see “high altitude marsh habitat” of Gerlach (1993). This habitat is still poorly understood. We only have few data on its flora, based on observations from the Mahé’s Mare aux Cochons (Carlström 1996). On Silhouette, such a forest type probably once occurred on its homonymous locality, i.e. Mare aux Cochons, but this area has been totally remodelled before any detailed botanical observation could have been done. Based on the Carlström qualitative observations, this habitat is not floristically different from the submontane ravine/valley rain forests. The forest ecotone of this swamp habitat is not detailed by the previous authors who were more interested by the general flora. eschweizerbartxxx author Conservation rank: G1(-G2) (critically imperilled). Within the 4 most typical submontane forest plots (excluding plot 4), we recorded 59 taxa including 50 indigenous ones (85 %). About 56 % of these indigenous taxa are endemic to Seychelles. Here the indigenous taxa are still clearly more important than exotics in terms of species, and about one tree out of two (47 %) belongs to a native species. This vegetation type is nearly extinct on Mahé, and imperilled on Silhouette. Its status on Praslin, if it exists, is totally unknown. We estimate the original area covered on Silhouette up to about 365 ha (18 % of the island, Fig. 4) of which 60 % remain undisturbed. Restoration is still possible because the original undisturbed composition still exists on Silhouette, and the pres- ent paper gives a new contribution to our knowledge of it. But here the expansion of invasive species is of great concern and additional plots are needed to understand the variability of this plant community more precisely. Permanent plots would also be useful to follow up the invasion of non-native species into the very well conserved areas mentioned here. Tropical ombrophilous lower montane forest (Lower montane rain forest) Synonyms: Mossy montane forest, not High-altitude forest (Gerlach 1993); forêt hygrophile d’altitude, faciès à Northea hornei et Glionnetia sericea (Friedmann 1994); Moist forest boulder peat habitat (Procter 1984b: type 1a); Mountain moss forest, Randia-Nepenthes society (Vesey-Fitzgerald 1940); Mountain mist forest (Fleischmann et al. 2003: type 5c). In other tropical regions, homologous vegetation types have received a diversity of names: cloud forest, lower montane cloud forest (Bruijnzeel & Hamilton 2000), moist montane forest (Trochain 1957), dwarf cloud forest, mist montane forest (Boughey 1955a). Physiognomy: At first sight, the most striking difference with the lowland and submontane belts is the spectacular physiognomy of these lower montane forests (a characteristic shared worldwide). Here the canopy becomes much lower, i.e. about 8–16 m (depending on the soil structure), and emergent trees are nearly absent, giving to the canopy a very flat aspect. The reduction of the canopy height is clearly associated with the reduction of the trunks, nearly undifferentiated in the extreme examples, where the soil is less developed. The lower canopy height causes a lack of differentiation of the under-canopy tree layer. As with the submontane rain forests, vascular epiphytes are abundant (Photo 3), but in this case very abundant and mixed with an impressive covering of bryophytes on the “trunks”, on the main branches and on the ground. The bryophytes cover can easily reach more than 5 cm and up to 15 cm of thickness. Floristics: Here the invasive species become clearly less dominant and the main tree is now Northea hornei. It is associated in the upper tree layer with Glionnetia sericea, Timonius sechellensis (a common species often abundant in such habitat). Pandanus sechellarum and Roscheria melanochaetes are sub-canopy trees. Dillenia ferruginea, Cinnamomum verum and Psidium littorale may replace most of the Northea and Glionnetia trees in more degraded areas. The shrub layer is mainly composed of Gastonia crassa, Rapanea seychellarum Mez, Syzygium wrightii, Erythroxylum sechellarum, Paragenipa wrightii, Psychotria pervillei and its related taxa or forms. Most of the herbaceous species found in the understory are typically lower montane, i.e. not seen in the submontane belt: Ctenopteris albobrunnea, Nepenthes pervillei, Hymenophyllum polyanthos, H. hygrometricum, Forest types on Silhouette Island 169 Agrostophyllum occidentale, Elaphoglossum macropodium, Gynura sechellensis, Xiphopteris serrulata, Rumohra adiantiformis, Humata repens, Bulbophyllum intertextum, etc. (Table 2). Among the species already present in the submontane belt, the most important ones include Lindsaea kirkii, Mapania floribundum, Protarum sechellarum and Haplopteris ensiformis. For the shrub layer, the best indicator species are Gastonia crassa and Rapanea seychellarum. The main diagnostic tree species is Glionnetia sericea. Within the herbaceous characteristic species cited here, the first ones appearing at the transition level and best diagnostic species are the ferns Elaphoglossum macropodium, Humata repens, Ctenopteris albobrunnea and Xiphopteris serrulata. Another understory species not mentioned in our plots but an excellent indicator of the lower montane belt is Oleandra distenta Kunze. Ecology: In all visited sites, on Mahé and Silhouette, the transition from submontane to lower montane forests can still be observed quite easily, especially if we focus on the understory indicator species. We found generally the first lower montane indicators at about 550 m, and then the typical lower montane belt become well developed before the 600 m (ca. 580 m), except where secondary eco-units or ravine/ valley variants cover too large an area, i.e. the Casse Dent entry to Congo Rouge. On steep or precipitous mountain ridges, several lower montane species can often be observed much below the 550 m transition: in some limited areas of Montagne Brulée, and on the north-eastern side of Mont Dauban, at as low altitude as 450 m. This example of ecological equalization is very well explained by Vesey-Fitzgerald (1940): when “the prevailing wind strikes these spots on the ridge and the strong up-draught created by the steepness of the terrain causes a great expansion and cooling of the air with a consequent excessive development of mist” at a lower altitude than in areas less exposed to winds or without precipitous steep slopes. This phenomenon has probably not been considered by most of the botanists in Seychelles while observing montane species at unusual lower altitude. Distribution: The lower montane rain forest is mainly found on the higher ridges of Mahé and Silhouette, above 550 m. On Mahé, it could be present at Mont Varigault (550 m), Montagne Planneau (680 m), Copolia (580 m), Trois Frères (690–760 m), Congo Rouge (740–830 m), Morne Seychellois (905 m), Morne Blanc (655 m) and Pérard (845 m). On Silhouette, this forest type occurs only on Mont Pot à Eau (620 m) and along the Mont Dauban ridge (740 m), where it eschweizerbartxxx author Photo. 3. Lower montane rain forest at plot 5 with a very important cover of epiphytes, including many fern species. Most trees are Northea hornei, mixed with Glionnetia sericea, Timonius sechellensis and some cinnamon trees. 170 B. Senterre et al. is very well preserved. On Mahé, the best preserved examples of this forest type are found on the slopes of Congo Rouge, Morne Seychellois and Pérard. Variability and related vegetation types: Here, the invaded variants distinguished for the lowland and submontane belts could in theory still be distinguished, for the same reasons. Nevertheless, monodominant forest of non-indigenous species are not so developed above the transition of 550 m, although cinnamon and Syzygium jambos can locally constitute monodominant forest eco-units on Mahé. On Madagascar, the lower montane forest belt can be totally replaced by Psidium forests (Carrière et al. 2008) but such a situation has not yet been attained in Seychelles. As already mentioned, secondary lower montane forests may be those areas with more Dillenia ferruginea trees, another native species replacing the Northea hornei and Glionnetia sericea individuals. Therefore, the floristic composition of the tree layer is no longer very distinct from the submontane forests and this emphasizes again the importance of the understory flora for the precise recognition of these two mountain vegetation belts. In addition, on the higher slopes of both northern and eastern sides of the main summit of Mont Dauban, the forest above 550 m altitude is on somewhat deeper soils and takes the appearance of the Agati woodlands, but detailed data or observations have not been collected. Getting to this site with time for observation is especially difficult because of the incredible density of Clidemia hirta, both in the understory of these Agati woodlands (northern slope) and in the open patches of the eastern slope (also invaded by Gleichenia linearis and Rubus rosifolius J.E. Smith). It is clear that this area of Mont Dauban needs urgent detailed studies. This is also true for the south-western side, which is probably the most pristine and least prospected (probably never explored), and which is the highest well conserved lower montane forest of Silhouette. Apart from these invaded and secondary variants, we also find ravine/valley variants within the lower montane belt. Such lower montane ravine/valley rain forests are floristically very similar to the submontane ravine/valley forests, especially for the herbaceous understory. But several species are, as far as we know, endemic to such habitat type: Pisonia sechellarum F.Friedmann and Psychotria silhouettae F.Friedmann (see the “Pisonia forest” of Gerlach, 1993). This variant has been observed only in one small area of Silhouette, but it is probable that similar plant communities, or other micro-endemics, will be found when other high altitude exposed ravines have been prospected in detail, especially in the north of Mont Dauban (in the higher part of the rivière Machabée valley) and on the northern parts of Morne Seychellois and Pérard (on Mahé). Finally, when climbing the highest peaks of Seychelles (on Mahé), it appears that further transitions could be recognized at higher altitude. At about 780 m altitude, both on the main summit of Congo Rouge eschweizerbartxxx author and on the north-eastern slope of Morne Seychellois, the tree ferns become extraordinary abundant and attain huge sizes, even far from the ravines and valleys. Rare and/or small species from the typical lower montane rain forest become at the same time both more common and larger, i.e. Canthium sechellense Summerh., Syzygium wrightii, Gastonia crassa. This higher altitude variant of the lower montane rain forest corresponds to what has also been described on central African mountains (Letouzey 1968, Monod 1960). Such a tree fern lower montane rain forest is only present on the two mentioned mountains and on Pérard. In addition, the very top of Morne Seychellois, above ca. 880 m, is marked in the upper tree layer by the monodominance of Pandanus sechellarum (usually an uncommon shrub or under-canopy tree in lower montane rain forests) and the absence of most of the typical lower montane understory species, and could therefore also constitute another vegetation belt. These very high altitude variants have never been studied in detail. Conservation rank: G2(-G3) (imperilled to vulnerable). Within the 3 lower montane forest plots, we recorded 58 taxa including 52 indigenous ones (90 %). About 62 % of these indigenous taxa are endemic to Seychelles. Here the indigenous taxa are still clearly more important than exotics in terms of species, and about one tree out of two (46 %) belongs to a native species. This vegetation type is still well represented on Mahé, and imperilled on Silhouette. We estimate the potential area of lower montane rain forests at about 52 ha (2.6 % of Silhouette island, Fig. 4), of which 60–75 % are undisturbed stands. As for submontane forests, restoration will be problematic. The unstudied uppermost altitudinal belts (the tree fern variant of the lower montane belt and the summit Pandanus forest) are only present on Mahé and must be considered as critically imperilled due to their very limited surface areas. Conclusions In the literature on Seychelles, the lower montane and submontane rain forests have been distinguished mainly based on tree species and moss abundance. In the present study, however, we identified a number of indicator species within all forest strata and biological types. We conclude that most of the best indicator species are to be found in the understory, i.e. ferns, and that the lack of detailed consideration of this stratum is responsible for the incomplete description of the altitudinal belts and partial misunderstanding in previous literature on Seychelles (see variation in given altitudinal intervals, mixing of cited indicator species belonging to different ecological groups, etc.). The importance of the understory species in forest types studies has been underlined by other authors in other tropical regions (Hemp 2002, Poulsen et al. 2005, Senterre 2005a). Forest types on Silhouette Island 171 eschweizerbartxxx author Fig. 4. Potential distribution of the main forest types of Silhouette. Light grey = lowland rain forest; dark grey = submontane rain forest; black = lower montane rain forest; white = ravine/valley forests; crosshatch = mangroves, swamp forests, marshes; dotted = glacis. The altitudinal zonation of forest types on Seychelles is characterised by a marked telescoping effect (Bruijnzeel & Hamilton 2000, Senterre 2005b). On most of the larger, or drier, tropical islands, and humid continental mountains, the submontane belt develops above ca. 700–900 m and turns into the lower montane belt above ca. 1500 m. Where the level of humidity is very high (tropical islands like Seychelles, coasts like some humid parts in the Gulf of Guinea) and combined with a small land mass resulting in lower temperatures (only on small high islands), the altitude needed for the air to reach condensation is lower. Such mountains with a marked telescoping effect are rare throughout the tropics: Mont Doudou in Gabon (Sosef et al. 2004, associated with cold sea surface currents), the Monte Mitra in Equatorial Guinea (littoral mountain), Korup and Rumpi mounts in Cameroon, Mayotte (Pascal et al. 2001), Mount Payung near the western tip of Java, Mount Finkol on Kosrae island (Micronesia), Gau island in the Fiji archipelago, Mt. Tinggi (Bawean island). As far as we know, few of these regions have been studied in detail for the complete gradient, from lowland to lower montane, passing through submontane belt (Pendry & Proctor 1997). In many areas, the reason is that the two lowermost belts (lowland and submontane) have already been removed (FAO 2000, MuellerDombois & Fosberg 1998). In Seychelles, we still have well preserved submontane forests, and further explorations may reveal more undisturbed lowland forests on Silhouette (i.e. south-western slopes of Mont Dauban). In addition, islands like Silhouette or Mahé are both relatively small but high, which reduces the land mass elevation effect (“Massenerhebung effect”). Further studies are needed on the factors influencing the telescoping effect, such as comparing the mountains with a telescoping effect on distinct islands: floristic patterns, diversity, phytogeographical and phylogenetical affinities, island size, climate, isolation and geographical context. The better understanding of the submontane belt resulting from the present study also underlines the importance of Silhouette as the last island of Seychelles where enough old growth forests exist to be able to identify suitable indicator species. Indeed, on Mahé, most of the submontane forests are early sec- 172 B. Senterre et al. ondary forests with a very irregular canopy and many more individuals of invasive species in all strata. Actually we have not seen or heard of any area of submontane forest on Mahé that could be compared to areas such as Grand Congoman on Silhouette. Once the indicator species have been identified based on well preserved areas like those of Silhouette, then it is possible to use indicator species from the understory and thus still be able to recognize the altitudinal belts even in secondary woodlands as on Mahé, where all the tree indicator species have disappeared. Nevertheless, Mahé also has its originality. The area of Congo Rouge, Pérard and Morne Seychellois includes considerable large areas of undisturbed lower montane rain forests and are therefore useful for a more precise understanding of this forest type (not so well represented on Silhouette). When comparing the altitudinal zonation in Seychelles to other tropical regions, and the typical telescoping effect observed here, it seems very probable that some further altitudinal belts should be distinguished between the interval of 550 m (beginning of the lower montane belt) and 905 m (top of Morne Seychellois). The very top of the still pristine Morne Seychellois, dominated by Pandanus sechellarum and characterised by a clear reduction of the diversity of vascular plants, could correspond to the upper montane belt, which develops mainly above ca. 2300 m in most tropical areas and is characterised by such a reduction of the vascular epiphytes diversity. Such habitat is unique to the summit of Morne Seychellois and has never been studied in detail (Carlström 1996). Therefore, and despite a reduced native flora, the altitudinal zonation on Seychelles seems to be strongly marked, i.e. presence of diagnostic species and constancy of the transition intervals. This is unusual when compared to younger, species poor islands (i.e. islands east of the Wallace line: Ashton 2003), and may result from the longer evolution of the flora of Seychelles allowing more developed niche differentiation. In addition, glacis from the higher summits of Mahé need to be explored since they could be refuges for some relict species of a subalpine belt. In Africa, we know that montane species can be seen at very low altitude on inselbergs (Parmentier & Maley 2001, Senterre 2005b: 75), due to a complex combination of historical and ecological parameters (glaciations, steep rock faces, dryer habitat but periodically humid, etc.). This study, and especially the more detailed configuration of the altitudinal belts, is important for conservation in Seychelles. A map of these forest types has been prepared (Fig. 4) and will be the base for decisions in terms of conservation and further studies or explorations. We identified priority areas on Silhouette for further studies, especially the three main areas of primary lowland rain forests. Their study and conservation must be considered as a high priority. They may be the very last examples of lowland rain forests in the archipelago and some of them have never been studied nor explored. Studying such areas will help us to better understand how to restore de- eschweizerbartxxx author graded habitats. We also note that most of the high altitude ravine/valley forests have not been prospected and should also be considered as a high priority since such habitat has revealed the presence of microendemic species, e.g. Pisonia sechellarum, and may play an important role as refugia for species from upper altitudinal belts. Following the observations on Silhouette and comparisons on Mahé, we expect to find the remnants of a submontane belt on Praslin (possibly at Fond Azore, still not described), where the highest point is about 367 m. Indeed, some submontane species are listed for this island, i.e. Northea hornei, Roscheria melanochaetes, Protarum sechellarum, Mapania floribundum, Scleria sieberi, Lindsaea kirkii, and even the more montane specialist Gastonia crassa. So, where are these species occurring, and in which habitats in Praslin? We expect most of them to occur in humid microhabitats of boulders of big granite blocks, like at La Réserve (Mahé), but what about the top of Praslin? Finally, to understand how these ecosystems are functioning, their history, their evolution on a short to long term, we need more detailed study of their phytogeographical (Procter 1984a) and phylogenetic patterns (Hardy & Senterre 2007), as well as plant functional types (Senterre 2005a). Indeed, we know that such characteristics may be very different from one vegetation type to another. For example, in Hawai’i, the lowland flora has a typically southeast Asian affinity while the alpine flora (on the top of the Haleakalā, Maui) has the strongest affinity with the Americas, where homologous belts are much more frequent (Medeiros & Loope 1994: p.33). In the light of such considerations, what is the phytogeographical pattern of the flora of Seychelles when vegetation types are considered separately: e.g. glacis vs. forest, lowland vs. montane? Such questions can be investigated when considering the total flora of such vegetation types (based on plant community plots) or just the indicator species (ecological groups, Tanghe 1995). Therefore, in future, the ecology of native species needs to be re-evaluated when ecological equalizations are considered. Acknowledgements. We are grateful to the Island Development Company (IDC) and Seychelles Bureau of Standards (SBS) for supporting the present study. The travel expense and local accommodation were supported by the FNRS (Fonds National pour la Recherche Scientifique), in Belgium, with the support of the Laboratoire de Botanique systématique et de Phytosociologie of the Free University of Brussels, i.e. Professor Jean Lejoly. Finally, the set-up of this project was in collaboration with the Nature Protection Trust of Seychelles (NPTS) and the Botanical Garden of Mahé. We are also grateful to Mr. Charles Morel for giving his support for the collection of plant vouchers and for providing access to the national collections (SEY), and to Dr. Christopher Kaiser for his contribution to the writing of the manuscript. The two referees also provided very useful comments and corrections of the paper. Finally, we thank people having contributed to the field data collection or drying of the specimens: Abdula Jumaye, Danny Malbrook and Frances Seilor. Forest types on Silhouette Island 173 References Areces-Mallea, A.E., Weakley, A.S., Li, X., Sayre, R.G., Parrish, J.D., Tipton, C.V. & Boucher, T. (1999): A Guide to Caribbean Vegetation Types: Preliminary Classification System and Descriptions – N. Panagopoulos (ed.) The Nature Conservancy. Ashton, P.S. (2003): Floristic zonation of tree communities on wet tropical mountains revisited. – Perspect. in Plant Ecol. Evol. and Syst. 6 (1–2): 87–104. Aubréville, A. (1965): Principes d’une systématique des formations végétales tropicales. – Adansonia 5 (2): 153–197. Baker, J.G. (1877): Flora of Mauritius and the Seychelles. – Reeve & Co., London. Boughey, A.S. (1955a): The nomenclature of the vegetation zones on the mountains of the Tropical Africa. – Webbia 11: 413–423. – (1955b): The vegetation of the mountains of Biafra. – Proc. Linn. Soc. (Bot.) 165 (2): 144–150. – (1965): Comparisons between the montane forest floras of America, Africa and Asia. – Webbia 19: 507–517. Braun-Blanquet, J. (1932): Plant sociology: the study of plant communities. – McGraw-Hill Book Company, Inc., New York. Bruijnzeel, L.A. & Hamilton, L.S. (2000): Decision time for cloud forests. – IHP Humid Tropics Programme Series, Vol. 13. UNESCO. Budowski, G. (1965): Distribution of tropical American rain forest species in the light of successional processes. – Turrialba 15 (1): 40–42. Bussmann, R.W. (2006): Vegetation zonation and nomenclature of African mountains – An overview. – Lyonia 11 (1): 41–66. Camirand, R. & Evelyn, O.B. (2003): Ecological Land Classification for Forest Management and Conservation in Jamaica. – Tecsult International, Québec, Canada. Carlström, A. (1996): Areas of Special Conservation Value for the Plants of the Granitic Islands of Seychelles. – Unpublished report, Seychelles Government. Ministry of Foreign Affairs, Planning and Environment. Conservation & National Parks Section, Victoria, Seychelles. Carrière, S.M., Randrianasolo, E. & Hennenfent, J. (2008): Aires protégées et lutte contre les bioinvasions: des objectifs antagonistes? Le cas de Psidium cattleianum Sabine (Myrtaceae) autour du parc national de Ranomafana à Madagascar. – VertigO 8 (1): 1–14. Chapman, V.J. (1976): Mangrove vegetation. – J. Cramer, Vaduz. Christensen, C. (1912): On the ferns of the Seychelles and the Aldabra group. – Trans. Linn. Soc. Lond. 2nd Ser. Zool. 15: 407–422. Churchill, S.P., Balslev, H., Forero, E. & Luteyn, J.L. (1995): Biodiversity and conservation of neotropical montane forests: Proceedings of the neotropical montane forest biodiversity and conservation symposium. – The New York Botanical Garden, USA. Dufrêne, M. & Legendre, P. (1997): Species assemblages and indicator species definition: the need of an asymmetrical and flexible approach. – Ecol. Monogr. 67: 345–366. FAO (2000): FAO Workshop – Data Collection for the Pacific Region. – Forest Resources Assessment Working Paper – 051 Forestry Department, Apia, Samoa. Fleischmann, K., Porembski, S., Biedinger, N. & Barthlott, W. (1996): Inselbergs in the sea: vegetation of granite outcrops on the island of Mahé, Praslin and Silhouette (Seychelles). – Bulletin of the Geobotanical Institute ETH 62: 61–74. eschweizerbartxxx author Fleischmann, K., Héritier, P., Meuwly, C., Küffer, C. & Edwards, P.J. (2003): Virtual gallery of the vegetation and flora of the Seychelles. – Bulletin of the Geobotanical Institute ETH 69: 57–64. Fleischmann, K., Edwards, P.J., Ramseier, D. & Kollmann, J. (2005): Stand structure, species diversity and regeneration of an endemic palm forest on the Seychelles. – Afr. J. Ecol. 43 (4): 291–301. Friedmann, F. (1994): Flore des Seychelles: Dicotylédones. – Orstom, Paris. Gerlach, J. (1993): The conservation of Silhouette island, Seychelles. I. Plants. – Phelsuma 1: 18–29. – (1997): A study of habitat structure and vegetation in Seychelles. – Phelsuma 6: 53. – (2003): The biodiversity of the granitic islands of Seychelles. – Phelsuma 11 (B):1–47. Grimshaw, J.M. (2001): What do we really know about the Afromontane Archipelago? – Syst. Geogr. Plants 71: 949–957. Grossman, D.H., Faber-Langendoen, D., Weakley, A.S., Anderson, M., Bourgeron, P., Crawford, R., Goodin, K., Landaal, S., Metzler, K., Patterson, K.D., Pyne, M., Reid, M. & Sneddon, L. (1998): International classification of ecological communities: terrestrial vegetation of the United States. Volume I. The National Vegetation Classification System: development, status, and applications. – The Nature Conservancy, Arlington, Virginia, USA. Hardy, O. & Senterre, B. (2007): Characterizing the phylogenetic structure of communities by an additive partitioning of phylogenetic diversity. – J. Ecol. 95: 493–506. Hedberg, O. (1951): Vegetation belts of the East African mountains. – Svensk Botanisk Tidskrift 45 (1): 141–202. Hemp, A. (2002): Ecology of the Pteridophytes on the southern slopes of Mt. Kilimanjaro – I. Altitudinal distribution. – Plant Ecol. 159 (2): 211–239. Holdridge, L.R. (1967): Life zone ecology. – Tropical Science Center, San Jose, Costa Rica. Jeffrey, C. (1962): The Botany of the Seychelles (Report by the visiting botanist of the Department of Technical Cooperation Seychelles Botanical Survey, 1961–62). – London. 37 pp. Keppel, G. (2002): Coastal Vegetation of Taunovo Bay, Pacific Harbour, Viti Levu, Fiji – A Proposed Development Site. – S. Pac. J. Nat. Sci. 20 (A): 25–29. Kottek, M., Grieser, J., Beck, C., Rudolf, B. & Rubel, F. (2006): World Map of the Köppen-Geiger climate classification updated. – Meteorol. Z. 15: 259–263. Legendre, P. & Legendre, L. (1998): Numerical Ecology. – Developments in Environmental Modelling, Vol. 20. Elsevier Science B.V., Amsterdan. Letouzey, R. (1968): Etude phytogéographique du Cameroun. – P. Lechevalier, Paris. Macnae, W. (1971): Mangroves on Aldabra. A Discussion on the Results of the Royal Society Expedition to Aldabra 1967– 68. – Philos. Trans. R. Soc. Lond. Ser. B Biol. Sci. 260 (836): 237–247. Medeiros, A.C. & Loope, L.L. (1994): Rare animals and plants of Haleakalā National Park. – Hawai’i Natural History Association. Monod, T. (1960): Notes botaniques sur les îles de São Tomé et Príncipe. – Bulletin de L’IFAN T. XXII, sér.A, n°1: 19–79. Mueller-Dombois, D. & Ellenberg, H. (1974): Vegetation types: a consideration of available methods and their suitability for various purposes. – Technical Report 49, Island Ecosystems IRP, U.S. International Biological Program. 50 pp. Mueller-Dombois, D. & Fosberg, F.R. (1998): Vegetation of the Tropical Pacific Islands. – Ecological Studies, Vol. 132. Springer, New York. 174 B. Senterre et al. Ohsawa, M. (1993): Latitudinal pattern of mountain vegetation zonation in southern and eastern Asia. – J. Veg. Sci. 4 (1): 13–18. Oldeman, R.A.A. (1990): Forests: elements of silvology. – Springer-Verlag, Berlin, New York. Parmentier, I. & Maley, J. (2001): L’arbre et le pigeon, ou le pigeon et l’arbre? – Canopée 19: 12–14. Pascal, O., Labat, J.-N., Pignal, M. & Soumille, O. (2001): Diversité, affinités phytogéographiques et origines présumées de la flore de Mayotte (Archipel des Comores). – Syst. Geogr. Plants 71: 1101–1123. Pendry, C.A. & Proctor, J. (1997): Altitudinal zonation of rain forest on Bukit Belalong, Brunei: Soils, forest structure and floristics. – J. Trop. Ecol. 13: 221–241. Piggott, C.J. (1968): A soil survey of the Seychelles. – Min. Overseas Dev. Tech. Bull. 2. Overseas Surveys, England. Poulsen, A.D., Hafashimana, D., Eilu, G., Liengola, I.B., Ewango, C.E.N. & Hart, T.B. (2005): Composition and species richness of forest plants along the Albertine Rift, Africa. – Biol. Skr. 55: 129–143. Procter, J. (1984a): Floristics of the granitic islands of the Seychelles. – In: D.R. Stoddard (ed.): Biogeography and ecology of the Seychelles islands, Pp. 209–220. – Junk Publishers, The Hague. – (1984b): Vegetation of the granitic islands of the Seychelles. – In: D.R. Stoddard (ed.): Biogeography and ecology of the Seychelles islands, Pp. 193–208. – Junk Publishers, The Hague. Richards, P.W. (1952): The tropical rain forest. – Cambridge University Press, Cambridge, UK. Robertson, S.A. (1989): Flowering plants of Seychelles (An annotated checklist of angiosperms and gymnosperms with line drawings). – Royal Botanic Gardens, Kew. Rübel, E. (1935): The replaceability of ecological factors and the law of the minimum. – Ecology 16 (1): 336–341. Schnell, R. (1976): Flore et Végétation de l’Afrique Tropicale. – Vol. 1. Gauthier-villars, Bordas, Paris. – (1977): Flore et Végétation de l’Afrique Tropicale. – Vol. 2. Gauthier-villars, Bordas, Paris. Senterre, B. (2005a): Recherches méthodologiques pour la typologie de la végétation et la phytogéographie des forêts denses d’Afrique tropicale. – Acta Bot. Gall. 152 (3): 409–419. – (2005b): Recherches méthodologiques pour la typologie de la végétation et la phytogéographie des forêts denses d’Afrique tropicale. – Ph.D. thesis, Université Libre de Bruxelles, Bruxelles. 345 pp. – (2009): Distribution and determinants of forest fires and land degradation on Praslin, Seychelles. – Consultancy report, Plan Conservation Action Group, Victoria, Mahé, Seychelles. 72 pp. Senterre, B. & Lejoly, J. (2001): Los grandes tipos de hábitat forestales de Río Muni. – In: C. Aedo, R. Morales, M.T. Tellería & M. Velayos (eds.): Botánica y Botánicos en Guinea Ecuatorial, Pp. 171–199. – Real Jardín Botánico – Agencia Española de Cooperación Internacional, Madrid. Senterre, B., Lejoly, J. & Sonké, B. (2004): Analyse du gradient de continentalité et identification de communautés végétales en forêts denses d’Afrique centrale par la méthode du mégatransect. – Phytocoenologia 34 (3): 491–516. Sosef, M.S.M., Issembe, Y.A., Bourobou, H.P. & Koopman, W.J.M. (2004): Botanical biodiversity of the Pleistocene forest refuge Monts Doudou (Gabon). – In: B.L. Fisher (ed.): Monts Doudou, Gabon: A floral and faunal inventory with references to elevational variation, 28, Pp. 17–91. – Memoirs of the California Academy of Sciences, California Academy of Sciences, San Francisco. eschweizerbartxxx author Tanghe, M. (1995): Groupes écologiques d’espèces végétales et application à la végétation forestière de la Wallonie. – In: J.F. Dulière, M. Tanghe & F. Malaisse (eds.): Répertoire des groupes écologiques du fichier écologique des essences. – Ministère de la Région Wallonne, Direction générale des Ressources naturelles et de l’Environnement, Belgique. Tchouto, M.G.P., Edwards, I., Cheek, M., Ndam, N. & Ackworth, J. (1999): Mount Cameroon cloud forest. – In: J. Timberlake & S. Kativu (eds.): African Plants: Biodiversity, Taxonomy and Uses, Pp. 263–277. – Royal Botanic Gardens, Kew. Thomas, D.W. & Achoundong, G. (1994): Montane forest of western Africa. – In: J.H. Seyani & A.C. Chikuni (eds.): Proc. XIIIth Plenary Meeting AETFAT. Pp. 1015–1024, Malawi. Trochain, J.-L. (1957): Accord interafricain sur la définition des types de végétation de l’Afrique Tropicale. – Bulletin de l’Institut d’Etudes centrafricaines, nouvelle série 13–14: 55–93. – (1980): Écologie végétale de la zone intertropicale non désertique. – Université Paul Sabatier, Toulouse. van der Maarel, E. (1979): Transformation of cover-abundance values in phytosociology and its effects on community similarity. – Vegetatio 39 (2): 97–114. Van Steenis, C.G.G.J. (1935): On the origin of the Malaysian mountain flora. Part 2. Altitudinal zones, general considerations and renewed statement of the problem. – Bull. Jard. bot. Buitenz. 3 (13): 289–417. Vesey-Fitzgerald, D. (1940): On the vegetation of the Seychelles. – J. Ecol. 28: 465–484. Walsh, R.P.D. (1984): Climate of the Seychelles. – In: D.R. Stoddard (ed.): Biogeography and ecology of the Seychelles islands, Pp. 39–62. – Junk Publishers, The Hague. White, F. (1978): The afromontane region. – In: M.J.A. Werger & A.C. Van Bruggen (eds.): Biogeography and ecology of southern Africa, Pp. 463–513. – Dr. W. Junk b.v. Publishers, The Hague. White, F. (1979): The Guineo-Congolian Region and its relationships to other phytochoria. – Bull. Jard. Bot. Nat. Belg. 49 (1/2): 11–55. White, L. & Edwards, A. (2001): Conservation en forêt pluviale africaine – Méthodes de recherche. – Wildlife Conservation Society, New York, USA. Whitmore, T.C. (1998): An introduction to tropical rain forests. – Oxford University Press, Oxford. WWF (2007): Granitic Seychelles forests (AT0113). -. URL http://www.nationalgeographic.com/wildworld/profiles/ terrestrial/at/at0113.html Address of authors: Bruno Senterre*, 1, 4, Justin Gerlach2, 3, James Mougal4 & Denis Matatiken4 1 Laboratoire de Botanique systématique et de Phytosociologie, Université Libre de Bruxelles (ULB), CP 169, 50 av. F. Roosevelt, 1050 Bruxelles, Belgium. 2 133 Cherry Hinton Road, Cambridge, CB1 7BX, England. 3 Nature Protection Trust of Seychelles, PO Box 207, Victoria, Mahé, Seychelles 4 Present address: Ministry of Environment, Natural Resources & Transport, Division of Nature and Conservation, Terrestrial & Ecological Research Centre, Botanical Gardens, Mont Fleuri, P.O. Box 445, Republic of Seychelles. Corresponding author, e-mail: bsenterre@gmail.com