Accepted Manuscript

Diversity depends on scale in the forests of the Central Highlands of Vietnam

Ha Thi Thanh Do, John C. Grant, Bon Ngoc Trinh, Heidi C. Zimmer, J. Doland Nichols

PII: S2287-884X(17)30100-0 DOI: 10.1016/j.japb.2017.08.008 Reference: JAPB 252

To appear in: Journal of Asia-Pacific Biodiversity

Received Date: 17 November 2016 Revised Date: 18 August 2017 Accepted Date: 24 August 2017

Please cite this article as: Thanh Do HT, Grant JC, Trinh BN, Zimmer HC, Nichols JD, Diversity depends on scale in the forests of the Central Highlands of Vietnam, Journal of Asia-Pacific Biodiversity (2017), doi: 10.1016/j.japb.2017.08.008.

This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and review of the resulting proof before it is published in its final form. Please note that during the production process errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain. ACCEPTED MANUSCRIPT

1 Diversity depends on scale in the forests of the Central

2 Highlands of Vietnam

3

4 Ha Thi Thanh Do a*,b , John C. Grant a, Bon Ngoc Trinh b, Heidi C. Zimmer a, J. Doland Nichols a

5

6 a Forest Research Centre, Southern Cross University, Lismore NSW Australia 2480. * Corresponding author

7 [email protected]

8 b Silviculture Research Institute, Vietnam Academy of Forest Science, Bac Tu Liem, Ha Noi, Vietnam

9

MANUSCRIPT

ACCEPTED ACCEPTED MANUSCRIPT

10 Abstract: Tropical forests are among the most diverse ecosystems on earth. They are also

11 the most threatened. The montane forests in the Central Highlands Region of Vietnam have

12 outstanding biodiversity and suite of unique species, yet we know little about them. This

13 study focuses on characterising natural forest at three sites: Dam Rong, Ha Nung and Yok

14 Don. We identified six discrete communities and their indicator species. One community,

15 Highland Floodplain forest, had tree species richness of up to 22 species/400 m 2 and 70

16 species/ha. In the lowland forests of Yok Don we identified three distinct communities,

17 despite that area having the lowest mean species richness (5 species/400 m 2). This study

18 illustrates the high species richness of the forests of Vietnam, and provides an important

19 record of the tree species (including rare and threatened species) at each of these sites. Our

20 community determinations can be used in future conservation management planning. 21 Moreover, the presence of three distinct tree commuMANUSCRIPTnities at Yok Don, which had the 22 lowest species richness, highlights that biodiversi ty should be assessed at multiple scales.

23

24 Keywords : Annamite mountains, Dam Rong, dipterocarp forest, Ha Nung, montane

25 rainforest, Yok Don

26

ACCEPTED ACCEPTED MANUSCRIPT

27 Introduction

28

29 The diversity of life on earth maintains the ecosystem services on which humans rely

30 (Chapin III et al 2000). Yet recent species losses implicate human actions as the cause of a

31 sixth mass extinction (Chapin III et al 2000). Resources for biodiversity conservation are

32 limited, and it is for this reason we must prioritise which areas are most critical for

33 protection (Mittermeier et al 1998). ‘Biodiversity hotspots’ provide one such prioritisation

34 strategy. Across the world, twenty-five regions have been named biodiversity hotspots.

35 Many of these high biodiversity areas with high levels of animal and plant endemism, are

36 threatened (Myers et al 2000). Tropical forests encapsulate more than half of the world’s

37 plant species and appear in 15 of the 25 biodiversity hotspots. They are also being depleted 38 faster than any other ecosystem (Myers 1988). MANUSCRIPT 39 Primary tropical forests, because of their high spe cies richness, constitute some of the most

40 complex ecosystems on earth (Gibson et al 2011; Wilson et al 2012) and are known for

41 being difficult to sample effectively and efficiently (Phillips et al 2003). Nevertheless,

42 distinct classification of tropical forests, as with all vegetation types, is fundamental to the

43 management, mapping and study of these systems (Biondi and Zuccarello 2004; De

44 Cáceres et al 2015). Classification of the world’s vegetation communities (and development

45 of classification methodology) has been underway for over a century (Mucina 1997). The 46 conceptualisationACCEPTED of discrete communities has been a large challenge (Looijen & van Andel 47 1999; Wilson & Chiarucci 2000), but there continues to be a focus on producing well- ACCEPTED MANUSCRIPT

48 functioning local vegetation classifications (cf. ill-fitting broad ones) that are fit for purpose

49 (e.g. conservation management) (Mucina 1997).

50 Historically, Thái (1963; 1999) classified Vietnam’s vegetation into five types and 14

51 subtypes. The first order classification was based on geography (highland or lowland),

52 canopy structure (closed or open), and climate. Vegetation communities were then

53 identified by edaphics, the level of disturbance and floristics, and finally identified by the

54 dominant species (or genera or families) (Thái 1963; 1999). The biodiversity of Southeast

55 Asia, which includes four biodiversity hotspots, has been under assault in recent decades

56 (Sodhi et al 2004). The region has the highest relative rate of deforestation of all major

57 tropical regions. Sodhi et al (2004) emphasised that the extent of this disaster may be far

58 greater than is currently understood, because of the paucity of research data. In addition, 59 biodiversity conservation research in the tropics MANUSCRIPTis chronically underfunded (Balmford and 60 Whitten 2003; Gardner et al 2008; Vieilledent et al 2016). The Indo-Burma biodiversity

61 hotspot (Myers et al 2000), is one of the four South-east Asian hotspots, and it includes all

62 of Vietnam. Most of this biodiversity occurs in Vietnam’s mountains, located in the north-

63 west, north-east and central regions, while the majority of flatlands have been cleared for

64 cultivation and urbanisation (Meyfroidt and Lambin 2008).

65 The Central Highlands region (CHR) in central Vietnam encapsulates most of the

66 remaining forests with high biodiversity value in Vietnam (Meyfroidt and Lambin 2008).

67 The CHR spansACCEPTED five provinces and is topographically dominated by the Annamite

68 Mountains (Day Truong Son). The montane rainforests of the northern and southern

69 Annamite Mountains have been highlighted by the WWF as ‘global ecoregions’ (Olson et ACCEPTED MANUSCRIPT

70 al 2001). The WWF designation indicates that these areas contain geographically unique

71 species, communities and conditions, with globally outstanding biodiversity (Olson et al

72 2001). Most famously, Annamite montane rainforests include regionally significant conifer

73 species richness, and the recently discovered large mammals, saola and giant muntjac

74 (Wikramanayake et al n.d.). In an analysis of biodiversity of ecoregions in the Indo-Pacific,

75 based on combined species richness and endemism, Krupnick and Kress (2003) found that

76 within the Indo-Burma biodiversity hotspot, the Southern Annamites montane rainforest

77 had the highest biodiversity. Biodiversity in the CHR is subject to pressures typical

78 throughout Vietnam, including deforestation as a result of immigration and the

79 development of market crops (Meyfroidt et al 2013), and dams, including Yali Falls dam –

80 the largest dam in the lower Mekong Basin (Polimeni et al 2014). Approximately half 81 (2,864,100 ha) of the CHR is forest, while approximMANUSCRIPTately one third (1,952,800 ha) is 82 devoted to agriculture of paddy rice, coffee, sugarcane and other commercial crops

83 (General Statistical Office of Vietnam 2013).

84 There are few studies of Vietnamese forests that have been published in the international

85 literature. Tran et al (2013) detailed the relationship between biodiversity and biomass of

86 major natural forest types in Vietnam. Blanc et al (2000) described communities and

87 succession of forests in Cat Tien National Park, Dong Nai Province. There are also two

88 studies that report on the spatial distributions (typically aggregated) of trees in the forests of 89 northern VietnamACCEPTED (Hai et al 2014; Nguyen et al 2016) and several studies that focus on the 90 influences of humans on threatened tree species (Dao and Hölscher 2015) and on plant

91 composition in general (Hoang et al 2011). These studies were conducted in northern ACCEPTED MANUSCRIPT

92 Vietnam. Despite outstanding biodiversity, very little has been published in the

93 international scientific literature about the tropical forest communities of Vietnam, and less

94 on attempts to characterise vegetation communities. Studies on the globally significant

95 CHR are particularly lacking. We aim to address this research gap.

96 This study describes the forest tree communities, and their indicator species, at three sites in

97 the little-studied, biologically significant Central Highlands region of Vietnam.

98 Biodiversity indices and species accumulation curves are presented for each community. To

99 support descriptions of these communities we present soil, climate and elevation data.

100

101 Materials and methods 102 Study area MANUSCRIPT 103 The Central Highlands Region (CHR) is located in the southeast of the Indochina Peninsula,

104 between longitude 11°11’ N (Lam Dong) to 15°25’ N (Kon Tum), and across longitude

105 107°12’ E ( Đak Nong) to 109°30’E. This region is at the southern end of Annamite Range,

106 Vietnam (Appendix Fig. 1). The majority of the CHR is from 100 to 800 m, although the

107 region also encompasses high mountains, including Ngoc-Linh (2,598 m). The CHR has

108 three main topographic classes: mountains, plateau and plain/delta (Nguy ễn et al 2000).

109 The climate of the CHR is dominated by the Asian monsoon, and is characterised by 110 distinct wet andACCEPTED dry seasons. Range annual rainfall is 1,400 –2,000 mm. The wet season 111 occurs from early May to mid-October, when 85–90% of the annual rainfall occurs. The

112 dry season is from November to April. The CHR covers two main catchments: Ba River ACCEPTED MANUSCRIPT

113 and Mekong Rivers watersheds. The soils of the CHR are derived from sandstone and

114 igneous bedrock and belong to six main groups: Fluvisols, Acrisols, Luvisols, Ferralsols,

115 Leptosols and Gleysols (Berdin et al 1999; Moormann 1961).

116

117 Study design

118 We selected contrasting study sites in natural forests within the forest reserves: (1) Kon Ha

119 Nung special use forest (scientific research) established in 1981 (affected by minor harvest

120 in 1978), (2) Yok Don National Park established in 1986 (undisturbed), and (3) Dam Rong

121 watershed protection forest, established in 1992 (undisturbed). Kon Ha Nung forest reserve

122 has been strictly protected for natural regeneration since 1981, and is representative of

123 intact natural forest (Tran et al. 2013). Environmental conditions in the three study sites are 124 shown in Table 1. The three sites are spread across MANUSCRIPT the CHR (Appendix Fig. 1).

125 Fieldwork was carried out in November 2013, at the end of the wet season. At each site we

126 established two replicate 1 ha-plots (100 x 100 m). These plots were 1 km apart. Each plot

127 was divided into 25 subplots (20 x 20 m; Appendix Fig. 2). At each site, morphology was

128 described using FAO Guidelines For Soil Description (Food And Agriculture Organization

129 Of The United Nations (FAO) 2006). Elevation was recorded using a Garmin GPS 76csX.

130 Slope and aspect (in azimuth) were measured using a Brunton pocket transit. 131 Vegetation wasACCEPTED surveyed within each 20 x 20 m subplot. All trees with stems ≥ 10 cm 132 diameter at breast height (DBH; 1.3 m high) were identified and their positions were ACCEPTED MANUSCRIPT

133 recorded. All woody species were identified using An Illustrated Flora of Vietnam (Pham

134 1999).

135 Soil samples were taken from a subset of five of the 20 x 20 m subplots (Appendix Fig. 2).

136 First, we prepared and described a soil profile down to bedrock/1.5 (FAO 2006). A set of

137 soil samples was taken from the soil profile at depth intervals 0–20 cm, 20–40 cm, and 40–

138 60 cm. Each sample for analysis was a subsample from a bulked and mixed set of four

139 samples for each depth. Soil sample preparation and analysis followed standard methods

140 (Rayment and Higginson 1992). The analysis included total organic carbon (Walkley-

141 Black), total nitrogen (Kjeldahl), available phosphorus (Oniani, 0.1M H 2SO 4; Oniani et al

142 1973), soil-exchangeable cations (ammonium acetate extract - Rayment and Higginson

143 1992). 144 MANUSCRIPT 145 Data analysis

146 The data used for analysis included all trees with DBH ≥ 10 cm, in three sites, each of

147 which had two one-hectare plots, and each plot encapsulating 25 subplots. Four additional

148 400 m 2 subplots were surveyed adjacent to the main plot in Yok Don, because of the

149 noticeably different forest that occurred there (to better represent the range of forest types at

150 Yok Don). In total there were 154 subplots. ANOVA was undertaken to compare key soil 151 variables amongACCEPTED communities; Fligner tests were used to check the homogeneity of 152 variance. Differences among sites were assessed using adjusted P-values calculated in

153 Tukey's HSD tests. Multiple comparisons increase the risk of Type 1 error; for this reason ACCEPTED MANUSCRIPT

154 only P-values < 0.001 are considered significant. All analyses were conducted in R version

155 3.0.3 (2014-03–06) (R Core Team).

156

157 Identifying communities and indicator species

158 The raw matrix used for analysis contained the counts of occurrence of each species

159 according to subplot. To identify subplots with similar species composition, and hence

160 identify community types, we used Ward's hierarchical cluster analysis (Ward’s minimum

161 variance method) with the Bray-Curtis indices as a measure of dissimilarities, in the Vegan

162 package for R (Oksanen et al 2013). This is a hierarchical agglomerative method of

163 vegetation classification, with advantages such as the ability to produce groupings at

164 multiple levels. However, a drawback of this method is that to define new groupings (i.e., if 165 new data are added) the whole classification needsMANUSCRIPT to be re-built, and that it has a low 166 robustness to sampling variation (De Cáceres et al 2015).

167 Species indicator values (INDVAL) were calculated using the package Indicspecies (De

168 Cáceres and Legen dre 2009; Dufrêne and Legendre 1997). Indicator values are calculated

169 for each species based on its abundance within the community, also known as “specificity”,

170 and presence/absence or “fidelity” to communities.A subset of potential indicator species

171 were separated from the total species list as species where P ≤ 0.05. Several indicator 172 species were chosenACCEPTED to characterise each community. 173 Indicator species can provide information on the presence and status of a range of other

174 species. Some indicator species are stenotopic, that is, where the species is indicator of only ACCEPTED MANUSCRIPT

175 one community; other indicator species are eurytopic, where they are indicators of several

176 communities. A good indicator species (stenotopic species) can be used to reduce costs in

177 inventory and monitoring, not only for environmental change, but also biodiversity

178 (Williams and Gaston 1994).

179

180 Calculating site and community biodiversity

181 Three biodiversity indices, Shannon, Simpson, and Evenness, were calculated for each site

182 and each community using the BiodiversityR package (Kindt and Coe 2005). While species

183 richness, or number of species, is the simplest measure of biodiversity, Shannon and

184 Simpson diversity indices account for relative abundances in addition to richness. Simpson

185 biodiversity is weighted towards dominant/common species, and is not sensitive to the MANUSCRIPT 186 occurrence of rare species with few individuals, whereas Shannon diversity is weighted 187 equally towards rare and common species (Magurran 1988). Finally Evenness indicates

188 how well individuals are spread among species; low values indicate a small number of

189 species dominate, and high numbers indicate a spread among many species (Magurran

190 1988). Abundance here is simply number of stems per subplot.

191 Species accumulation curves were generated in Vegan (Oksanen et al 2013). The

192 Chapman-Richards model was used to fit the curves (Flather 1996) for six communities 193 (Appendix TableACCEPTED 1). The models have upper asymptote as α parameter and other 194 parameters b or c (or both), and the inflection point, relative to the upper asymptote.

195 ACCEPTED MANUSCRIPT

196 Results

197

198 Tree communities and their indicator species and biodiversity indices

199 Across the three sites, in a total of six plots and 154 subplots, we recorded 3,732

200 individuals from 157 species. A complete list of species is presented in Appendix Table 2.

201 Cluster analysis identified six communities (Fig. 1). The first subdivision separates forests

202 into two main groups by elevation, a lowland group (Yok Don) and a highland group (Dam

203 Rong and Ha Nung). The next subdivision resulted in six clusters in total, each cluster

204 representing a community.

205 There were three communities in Yok Don, two at Dam Rong and one at Ha Nung. Yok

206 Don, at the lowest elevation (192 m), was dominated by deciduous (dipterocarp) forest 207 (Appendix Fig. 4). In contrast, Dam Rong (elevation MANUSCRIPT 924 m) and Ha Nung (elevation 686 m) 208 were dominated by evergreen forest (Appendix Fig. 4). It is notable that Ha Nung, rather

209 than lower-elevation Yok Don, had the lowest annual rainfall at 1,533 mm. Dam Rong had

210 the highest at annual rainfall 1,865 mm, and a higher number of rain days (Table 1).

211 The Highland Floodplain (HLF) community had the greatest number of species (85

212 species), although it also had the highest total area sampled of 2.0 ha, with modelling

213 suggesting that this figure is approaching the true number of species for this community ( α

214 = 86.89). The lowest species richness was found in Lowland Deciduous – type 2 (LLD2)

215 community inACCEPTED Yok Don, at only 12 species identified in 1.0 ha. This figure is also near the

216 estimated true number of species (14.2 species). ACCEPTED MANUSCRIPT

217 Across the three sites we found 15 rare and threatened species, of which 12 species are on

218 the the IUCN Red List (Table 2), ranging from Near Threatened (NT) to Endangered (EN)

219 or with Data Deficient (DD) status, and an additional three species on the Vietnam Red List

220 ( Vietnamese Academy of Science and Technology 2007) which were not listed or listed as

221 least concern, by IUCN. The communities with the highest number of rare and threatened

222 species were HLF and LLM, each with six species.

223 Ninety-nine indicator species were selected from the total group of 157 species. Of these,

224 69 species were associated with one cluster (community), 22 species were associated with

225 two clusters, seven species were associated with three clusters, one species was associated

226 with four clusters and five species showed no association.

227

228 Lowland tree communities (Yok Don) MANUSCRIPT

229 Lowland dry deciduous (type 1):

230 Twelve subplots were characterised by Lowland Dry Deciduous community (type 1;

231 LLD1). LLD1 only occurred at Yok Don, and was the least common community in this

232 study.

233 The most common species in this community were, in order of dominance, Dillenia hookeri ,

234 Shorea obtusa , Xylia xylocarpa , Dipterocarpus obtusifolius and Dipterocarpus 235 tuberculatus . TwoACCEPTED individuals of Sindora siamensis , listed as Endangered in Vietnam, were 236 also found. ACCEPTED MANUSCRIPT

237 Dipterocarpus tuberculatus and Dillenia hookeri were identified as indicator species.

238 Dillenia hookeri was only present in LLD1 subplots (indicator value = 90.5 ***), and D.

239 tuberculatus was present in 82% of subplots characterised as LLD1 (indicator value = 57.7

240 ***).

241 Subplot-scale species richness for LLD1 ranged from 4 to 11 (Table 3), and a total of 21

242 species were recorded in this community (Fig. 3). This is the second lowest total number of

243 species for a community, although LLD1 also occupied the smallest number of subplots of

244 any community. However, 21 species was far from the asymptote/ total number of species

245 (63.03 species). LLD1 had the second highest stem abundance of all communities in the

246 study.

247

248 Lowland dry deciduous (type 2): MANUSCRIPT

249 Lowland Dry Deciduous community (type 2; LLD2) was the most common community at

250 Yok Don, occurring in 25 subplots.

251 The most common species in this community were Mitragyna speciosa , Shorea obtusa ,

252 Dipterocarpus tuberculatus , Terminalia chebula and Terminalia alata . The plots in this

253 community included no rare or threatened species. Terminalia alata (indicator value =

254 80.0***) and Terminalia chebula (indicator value = 66.3***) were both common species, 255 and were identifiedACCEPTED as the best indicator species for LLD2. 256 Mean species richness at the subplot scale for LLD2 was 3.52, and ranged from 2 to 5. The

257 species accumulation curve showed a total of 12 species, and this was near asymptote at ACCEPTED MANUSCRIPT

258 14.16 species. LLD2 had the lowest species richness of all communities in this study, and

259 the lowest stem abundance. LLD2 had the second highest evenness of all communities in

260 the study.

261

262 Lowland Mixed forest:

263 The Lowland Mixed forest community (LLM) occurred in 17 subplots and was

264 characterised by indicator species Shorea siamensis (indicator value = 70.2***), which

265 occurred almost exclusively in this community (93% of occurrence in LLM subplots),

266 although it was not present at all LLM forest sites (only 53%).

267 The most common species in LLM were Hopea pierrei , Shorea siamensis , Shorea obtusa , 268 Dipterocarpus obtusifolius and Dipterocarpus tuberculatusMANUSCRIPT. We also found six rare and 269 threatened species. These were the IUCN Red-listed Endangered species Hopea pierrei ,

270 Hopea recopei , Anisoptera costata, Shorea roxburghii and the Vietnam Red-listed Sindora

271 siamensis (Endangered) and Pterocarpus macrocarpus (Vulnerable).

272 The LLM community had the highest subplot species richness of all the communities at

273 Yok Don at 6.1 species (range 3 to 10), and also the highest Shannon and Simpson diversity

274 (1.45 and 0.69). The total number of species recorded was 32, across 17 subplots, and

275 modelling suggested true species richness was 39.62 (Appendix Table 1, Fig. 3).

276 ACCEPTED

277 Highland tree communities (Ha Nung and Dam Rong)

278 Highland Floodplain: ACCEPTED MANUSCRIPT

279 The Highland Floodplain (HLF) community characterised all subplots at Ha Nung (50

280 subplots). The most common species were Aglaia perviridis , Syzygium sp ., Machilus

281 odoratissima , Pavieasia annamensis , Michelia floribunda and Polyalthia cerasoides . We

282 also identified six rare and threatened species: the IUCN Red-listed Knema pierrei ,

283 Mangifera minutifolia, Xylopia pierrei (all Vulnerable), Nageia fleuryi (Near-Threatened)

284 and Michelia floribunda – which is listed as Data Deficient but was common at this site.

285 Data Deficient species have inadequate data for threat assessment and because of this they

286 are often treated as threatened (IUCN 2001). We also found the Vietnam Red-listed species

287 Aglaia spectabilis .

288 The set of indicator species for this community was complex with 16 species. The species

289 were, in order of indicator value: Michelia floribunda (indicator value = 88.3***) and 290 Polyalthia cerasoides (indicator value = 82.5***),MANUSCRIPT both common species, followed by 291 Nephelium lappaceum , Machilus odoratissima , Syzygium sp ., Cinnamomum bejolghota ,

292 Baccaurea harmandii , Syzygium zeylanicum , Grewia bulot , Symplocos laurina , Symplocos

293 laurina var. acuminata , Wendlandia paniculata , Symplocos lancifolia , Cinamomum sp .,

294 Canarium album , and Prunus arborea (indicator value = 51.0***).

295 This community also had the highest mean subplot species richness, at 16 species (range 8

296 to 22), compared to all other communities in the study. Similarly, it had the highest

297 Simpson and Shannon diversities. The total number of species recorded in this community

298 was 85 – by thisACCEPTED measure HLF was clearly the richest of species of all communities. This

299 figure was near the estimate of asymptote of 86.89. The species accumulation curve shows

300 HLF tracking higher than all other communities, even at lower subplot counts. ACCEPTED MANUSCRIPT

301

302 Highland upslope and lowslope forests:

303 There were two communities at Dam Rong: Highland Upslope (HLUS) in 26 subplots, and

304 Highland Lowslope (HLLS) in 24 subplots (Table 3).

305 The most common species in HLUS and HLLS were Psychotria poilanei , Machilus

306 parviflora and Castanopsis hystrix . The two communities differ in that HLUS had the

307 additional common species Lithocarpus stenopus and Podocarpus neriifolius , while HLLS

308 was characterised by Michelia mediocris and Gomphia striata .

309 The communities also shared the rare and threatened species Knema pierrei (Vulnerable).

310 HLLS had the additional Vietnam Red listed species Lithocarpus truncatus , while HLUS 311 had the IUCN Red-listed Hopea pierrei . MANUSCRIPT 312 The indicator species for these two highland communities (together) were, in order of

313 indicator value, Castanopsis hystrix (indicator value = 100***) and Machilus parviflora

314 (indicator value = 94.9***), both common species, followed by Beilschmiedia

315 roxburghiana , Gomphia striata , Canarium subulatum , Michelia mediocris , Archidendron

316 robinsonii , Pyrenaria jonquieriana , Eriobotrya angustissima and Magnolia candollei

317 (indicator value = 56.6 **). Castanopsis hystrix is a perfect indicator species for these

318 communities, as it was present in all sites. In addition, Machilus parviflora occurred only in 319 this community.ACCEPTED ACCEPTED MANUSCRIPT

320 HLUS had a good indicator species in Podocarpus neriifolius (indicator value = 96.3),

321 while the species with the highest indicator value for HLLS was Cinnamomum iners

322 (indicator value = 67.0).

323 HLLS mean subplot species richness was 15.7 species, lower than HLF at Ha Nung.

324 However, HLLS species ranged from 10 to 23, the highest minimum and maximum species

325 richness in this study – higher than HLF. The subplot species richness of HLUS was similar

326 to HLLS at 15.3 species (range 8 to 22). HLLS was also characterised by the highest

327 abundance. Despite occurring in similar numbers of subplots, species accumulation curves

328 showed that HLLS (24 subplots, 58 species) was more species rich than HLUS (26 subplots,

329 47 species). Modelled curve asymptotes similarly showed HLLS with more species than

330 HLUS (59.03 compared to 49.17 species). 331 MANUSCRIPT 332 Relating communities to soil factors

333 The topsoils of all communities were strongly acidic (Table 4). All nutrients (except

334 nitrogen) were at low to very low levels across all sites. Nitrogen levels were medium to

335 high at the Ha Nung and Dam Rong plots, and equivalent to four to five times the levels

336 recorded at Yok Don. Phosphorus levels were low to very low at all sites, but again they

337 were two to three times higher at Ha Nung and Dam Rong compared with Yok Don. Base 338 saturation percentagesACCEPTED were low at the Ha Nung and Dam Rong sites and moderate at Yok 339 Don. ACCEPTED MANUSCRIPT

340 Key plant nutrients, nitrogen, phosphorous and potassium varied significantly among

341 communities (Appendix Table 3, Appendix Fig. 3). In general, nitrogen, phosphorous and

342 potassium were lower at the low elevation communities (LLD1, LLD2 and LLM; Yok Don)

343 compared to the high elevation communities (HLLS, HLUS, HLF). Exchangeable calcium

344 levels were recorded as low in the topsoils across all the communities and

345 calcium/magnesium ratios were also low.

346 Within Yok Don there were significant differences in soil among communities, in particular

347 the soils at LLD2 had significantly more silt/significantly less sand than the soils at LLD1

348 and LLM (which were not significantly different from one another; Table 4, Appendix

349 Table 3, Appendix Fig. 3, Appendix Fig. 5). LLD2 also had higher exchangeable

350 magnesium than LLM (not significant at P < 0.001 level). LLM and LLD1 could not be 351 easily differentiated in terms of soil; the biggestMANUSCRIPT difference was that LLD1 had lower pH, 352 but this difference was not significant. The two communities within Dam Rong had no

353 significant differences.

354 All sites (and communities) had soils that were acidic, leached and generally low in

355 nutrients. These issues would be ameliorated to some extent by the depth of the soils at Ha

356 Nung and Dam Rong, which would provide a larger total biodiversity resource. This is in

357 contrast to the soils Yok Don, which was much shallower.

358 ACCEPTED 359 Discussion

360

361 Community and site species richness ACCEPTED MANUSCRIPT

362 This study describes six distinct communities at three forest sites. The communities defined

363 in this study extend the vegetation community classification of Vietnam (Thái 1999), by

364 providing a detailed and objective analysis of communities at the fine scale(< 1000 m2),

365 using hierarchical agglomerative modelling. Based on Thái (1999) the communities at Ha

366 Nung and Dam Rong, and LLM in Yok Don, were simply classified as closed evergreen

367 moist tropical forest, while LLD1 and LLD2 were classified as open broadleaf dry tropical

368 forest.

369 Species richness varied considerably among communities, and according to size of sample

370 area (i.e., subplot versus plot). The highest mean subplot species richness, 16 species/400

371 m2 (trees > 10 cm DBH), was within the HLF community at Ha Nung. The mean species

372 richness of forests at Dam Rong was slightly lower (15.7 and 15.3 species/400 m 2). These 373 results demarcate the highland forests of the CentrMANUSCRIPTal Vietnam as being characterised by 2 374 species richness comparable tropical forests of Pasoh, Malaysia (18.8 species/400 m ), and

375 higher than tropical forests at Barro Colorado Island, Panama and Mudumalai, India (12.5

376 species/400 m 2 and 5.2 species/400 m 2, respectively; Condit et al 1996). Outside tropical

377 regions, studies have shown mean tree species richness of 1 to 7.8 species/400 m 2 (in New

378 Zealand; Bellingham et al 1999). Moreover, if we consider maximum (cf. mean) species

379 richness, HLLS forest species richness at Dam Rong (24 species/400 m 2) is higher again.

380 At the plot scale (1 ha), Ha Nung had a maximum species richness of 70 species/ha, and 381 Dam Rong, 55ACCEPTED species/ha. These values are much lower than the Pasoh (206 species/ha) 382 and Barro Colorado Island (91 species/ha), but still higher than Mudumalai (22 species/ha)

383 (Condit et al 1996). The highest recorded species richness at this scale is tropical rainforest ACCEPTED MANUSCRIPT

384 in Ecuador at 942 species/ha (Wilson et al 2012). Compared to tropical forests from across

385 the world, the species richness of highland forests of Vietnam (HLF, HLLS, HLUS) are

386 ranked highly at smaller scales (subplot or 400 m 2), but lower at larger scales (plot or1 ha).

387 Species accumulation curves have historically been used to assist in determining adequate

388 sampling regimes, characterising community structure and estimating species richness (He

389 et al 1997). The key to this process is whether the species accumulation curve approaches

390 or reaches asymptote (with no or few species being added with additional sampling effort

391 indicating adequate sampling). In our study, the communities nearer to asymptote,

392 indicating adequate sampling, were LLD2 (25 subplots, total number of species = 12, α =

393 14.16), HLLS (24 subplots, total number of species = 58, α = 59.03), HLUS (26 subplots,

394 total number of species = 48, α = 49.17) and HLF (50 subplots, total number of species = 395 85, α = 86.89). Alternatively LLM (17 subplots, MANUSCRIPTtotal of species = 32, α = 39.62) and LLD1 396 (12 subplots, total number of species = 21, α =63.03) were further from reaching asymptote,

397 and require larger survey area (more than 50 and 600 subplots for LLM and LLD1,

398 respectively) to adequately capture species richness and community characters (Fig. 3,

399 Appendix Table 1).

400

401 Community richness 402 In contrast to ACCEPTEDthe highland sites, Yok Don had very low mean subplot species richness (4.8 403 species/400 m 2). The occurrence of low diversity forests in the tropics is not uncommon,

404 Huston (1994) describes their association with extreme soil conditions, using ACCEPTED MANUSCRIPT

405 Dipterocarpaceae in Malaysia on very wet and very dry soils as an example. Indeed, the

406 soils at Yok Don were sandy, and low in nutrients, but rainfall was relatively high. On the

407 other hand, Yok Don had three distinct communities, where the same analysis of the same

408 size survey area resulted in only one community defined in species-rich Ha Nung. This

409 result indicates that the spatial co-occurrence of species at Yok Don was more

410 heterogenous at smaller scales, compared to the other sites. Community diversity at Yok

411 Don is likely to be driven by soil heterogeneity; previous studies have correlated

412 dipterocarp species distributions with soil characters (Sukri et al 2012; Webb and Peart

413 2000; see below: Environmental drivers of tree community diversity). Despite low overall

414 species richness, because of its high community diversity, Yok Don should be also be

415 considered of high conservation value. 416 MANUSCRIPT 417 Communities and indicator species: lowland

418 The indicator species for LLD1 (Yok Don) were Dipterocarpus tuberculatus

419 (Dipterocarpaceae) and Dillenia hookeri (Dilleniaceae). These species were common in the

420 study plots. Dipterocarpus tuberculatus is a large tree growing to 15-20 m in height, with

421 typical DBH of 30 cm (but up to 60 cm). It is a light-demanding tree that grows well on

422 many soil types, and is distributed predominately in the Western Highlands and in the 423 Southeast regionsACCEPTED of Vietnam (Nguyen 2013). Dipterocarpus tuberculatus , with several 424 other Dipterocarpus species, is characteristic of the dry deciduous dipterocarp forest that is

425 widespread in Southeast Asia (Stott 1990). Dipterocarpus tuberculatus is the most common

426 dipterocarp at Yok Don (Nguyen and Baker 2016). Dillenia hookeri occurs in southern ACCEPTED MANUSCRIPT

427 Vietnam, Cambodia, Thailand and Laos. It is a small tree, although it can occur as a shrub

428 in dry areas. It is primarily found at higher elevations, particularly in humid or wet areas in

429 the Central Highlands of Vietnam (Tr ần 2002).

430 LLD2 (Yok Don) was characterised by the indicator species Terminalia alata and

431 Terminalia chebula (Combretaceae). These species were common in LLD2, but uncommon

432 in LLD1, making these communities easy to distinguish. Terminalia alata is a deciduous

433 tree that grows to 35 m, and prefers moderately dry sites. It has a distribution that extends

434 across Asia. Terminalia chebula is also deciduous, and grows to 20 m, preferring slightly

435 moister sites and extending to lower elevations than Terminalia alata (Gardner 2000). Both

436 species are typically found in semi-open forests. Jackson (1994) s reported that Terminalia

437 alata seedlings are shade intolerant. Moreover, in tropical southern India, Terminalia alata 438 is a savanna tree, but has a role in forest extensi MANUSCRIPTon, as it provides microclimatic conditions 439 for forest tree seedling establishment by reducing grass cover (Puyravaud et al 1994).

440 Eventually forest tree seedlings establish, and can prevent Terminalia alata seedlings from

441 recruiting (Puyravaud et al 1994). In contrast, Terminalia chebula is described as a slow-

442 growing, late successional species (Khurana and Singh 2004). Terminalia chebula has large

443 seeds that can increase dormancy in response to water stress, and has relatively drought

444 tolerant seedlings (Khurana and Singh 2004).

445 The indicator species for LLM forest community was Shorea siamensis , also from

446 Dipterocarpaceae.ACCEPTED It was uncommon in LLD1 and LLD2, but common in LLM2. Shorea

447 siamensis is typical of dry deciduous dipterocarp forests in Vietnam, Indonesia, Malaysia,

448 Myanmar and Thailand (IUCN 2001). It is a large deciduous tree growing to 30 m in height, ACCEPTED MANUSCRIPT

449 and 80 cm DBH. Shorea siamensis is drought tolerant, growing in hot dry conditions on

450 poor sandy soil, to elevations of 1,000 m (Nguyen 2013), where it occurs at low density.

451 Nguyen and Baker (2016) recorded low numbers of Shorea siamensis seedlings at Yok Don,

452 highlighting a possible regeneration bottleneck. Ghazoul et al (1998) noted that viable seed

453 production can be pollen limited, because Shorea siamensis is self-incompatible. In terms

454 of rare and threatened species, the presence of four Endangered Dipterocarpaceae species,

455 Anisoptera costata , Hopea pierrei , Hopea recopei and Shorea roxburghii in LLM at Yok

456 Don is significant. These species are on the IUCN Red List because of population

457 reductions. Hopea recopei and Hopea pierrei are characterised by the additional threat of

458 occurring only in a small area, with Hopea pierrei thought to have a population of less than

459 250 mature individuals (IUCN Endangered [Criterion D]). This is interesting as Hopea

460 pierrei was the most common threatened species in the LLM community at Yok Don. MANUSCRIPT 461

462 Communities and indicator species: highland

463 The HLF forest community (Ha Nung) was characterized by the indicator species Michelia

464 floribunda (Magnoliaceae) and Polyalthia cerasoides (Annonaceae). Michelia floribunda is

465 an evergreen tree that grows to 20 m, typically found in less-disturbed forests above 1500

466 m (Gardner et al 2000) and is listed as a Data Deficient species by the IUCN (IUCN 2001). 467 Michelia florbundaACCEPTED is a dominant species in some evergreen forests in mountainous China 468 (Gong et al 2013), and Thailand (Viranant et al 2009). Michelia species are considered to

469 be shade tolerant (Tang et al 2013). The other indicator species, Polyalthia cerasoides, is a ACCEPTED MANUSCRIPT

470 medium-sized tree, growing to 10-20 m and 20-50 cm in diameter, and is found in Vietnam,

471 China, Laos, Cambodia and India (Tr ần 2002).

472 HLF had the highest number of rare and threatened species. Both Knema pierrei

473 (Myristicaceae), Mangifera minutifolia (Anacardiaceae) exist only in Vietnam and are

474 threatened by small population and area of occupancy (IUCN Criterion D2). Knema pierrei

475 is medium-sized tree, growing to 15-20 m height. It is shade tolerant, grows in moist soil,

476 and is associated with lowland tropical moist forest, but has a widely scattered distribution

477 in Vietnam (Tr ần 2002). Xylopia pierrei (Annonaceae) is found in Vietnam and Cambodia,

478 and is listed as Vulnerable because of population declines (IUCN Criterion A1a). Nageia

479 fleuryi (Podocarpaceae) is near threatened because of population declines – its timber is

480 highly valued. Generation length for Nageia fleuryi is thought to be 30 years (Thomas 481 2013). Nageia fleuryi occurs in Vietnam, Laos and MANUSCRIPT China. 482 The highland forest groups (HLUS and HLLS) had indicator species Castanopsis hystrix

483 (syn. Castanopsis purpurella subsp. purpurella ) (Fagaceae) and Machilus parviflora

484 (Lauraceae). Castanopsis hystrix is listed as vulnerable in Vietnam’s Red Data Book of rare

485 and endangered species, and studies of montane forest in northwest Vietnam place it as

486 only occurring in core/undisturbed areas of reserves (Dao and Hölscher 2015). Machilus

487 parviflora (prev. Persea minutiflora ) is a medium-sized tree (12-14 m) occurring in

488 Vietnam, India and Laos. Machilus parviflora regenerates well in shade, and at high

489 humidity on fertileACCEPTED soil (Tr ần 2002). This species is distributed in primary forest in north-

490 east, north-central and the Central Highlands of Vietnam (Tr ần 2002). All these

491 communities had recordings Knema pierrei which is classified as vulnerable species. ACCEPTED MANUSCRIPT

492 The factors distinguishing HLLS and HLUS are that HLUS had a strong indicator species,

493 Podocarpus neriifolius (indicator value = 96.3), and included records of the rare and

494 threatened species Hopea pierrei . Alternatively HLLS was characterised by an indicator

495 species with an indicator value of only 67 ( Cinnamomum iners ) and had no unique rare or

496 threatened species records.

497

498 Environmental drivers of tree community diversity

499 Our study sites were positioned across the Central Highlands, varying in elevation by

500 almost 800 m. A key factor driving the differences among sites is elevation, which affects

501 climate and has an impact on soil. The role of climate in determining the distributions of

502 ecosystems, at broader spatial scales, is well known (Schimper et al 1903); desert, grassland, MANUSCRIPT 503 forest distributions across the world are predicted by temperature and precipitation 504 (Holdridge, 1947). Indeed, the distributions of evergreen and deciduous forests at higher

505 and lower elevation, respectively (as seen in this study) are controlled by seasonal water

506 availability (as in Vázquez and Givnish 1998). Dam Rong is higher (924 m), cooler, wetter

507 (1,865 mm/ year rainfall), and has clay loam soils, and Yok Don is lower (192 m) and

508 hotter, and although it receives more rainfall (1,789 mm/year), its shallow sandy loam soils

509 have lower capacity to capture and store water for extended periods. Ha Nung, at the north 510 of the CHR, ACCEPTED is at 686 m, is on clay-dominated soils and is characterised by the lowest 511 rainfall of all sites at 1,533 mm/yr. The relationship between elevation and diversity is

512 complex. For example, Van der Ent et al (2016) showed plant diversity decreasing with

513 elevation in Borneo, while Teejuntuk et al (2003) showed diversity increasing with ACCEPTED MANUSCRIPT

514 elevation in the montane forests of Thailand. In this study the highest diversity was at Ha

515 Nung/HLF, at the ‘middle’ elevation.

516 High plant diversity has also been related to complex and changing topography,

517 microclimate and soil physical and chemical characteristics. This occurs where there is a

518 great variety in soil parent material and geomorphological processes are strongly active,

519 (such as in high rainfall, hilly upland country) (Sollins 1998). This is broadly true for our

520 study, the upland sites did have higher diversity. But the highest species diversity did not

521 occur with the highest soil variability (Ha Nung and Yok Don, respectively). High diversity

522 has also been related to low fertility (Nadeau and Sullivan 2015; Toledo et al 2012)

523 although Ha Nung/HLF (was not significantly lower than the other sites in the nutrients

524 tested. There are myriad climate-soil-vegetation feedbacks, and the complexity of these 525 relationships means there is no single driver MANUSCRIPT of. For example, while Yok Don is 526 characterised by the high rainfall, its shallow soils combine with low levels of nutrients to

527 create a more stressful edaphic environment than at the other sites.

528 The relationships between species composition and soil in dipterocarp forests, such as those

529 in Yok Don, have been researched extensively. Webb and Peart (2000) in Borneo found

530 that mature tree species composition was strongly correlated with topography and soil

531 characteristics. Key soil characteristics related to dipterocarp distribution have included

532 texture, carbon content, pH, depth, drainage and nutrient status (Davies et al 2005;

533 Palmiotto et alACCEPTED 2004; Slik et al 2009). Annual rainfall, rainfall seasonality and droughts are

534 an important influence on these forests (Slik et al 2009). We found clear differences in soil

535 texture between distinct forest tree communities growing alongside each other at Yok Don. ACCEPTED MANUSCRIPT

536 LLD2 (dominated by Mitragyna speciosa , Shorea obtusa and Dipterocarpus tuberculatus )

537 had significantly higher silt, and less sand, than nearby LLD1 and LLM. LLD1 and LLM

538 were more difficult to differentiate in terms of soil, although there were clear distinctions in

539 species composition, LLM dominated by Hopea perreri and Shorea siamensis and LLD1

540 dominated by Dillenia hookeri and Shorea obtusa .

541 The results from Yok Don support previous studies, which indicated that differences in

542 dipterocarp community composition correlate with soil texture (which is acting as a

543 surrogate for soil nutrients) (Davies et al 2005; Palmiotto et al 2004). Key soil factors

544 driving differences in forest tree communities at Dam Rong and Ha Nung are less clear;

545 however, there was some evidence that potassium differed among communities at Dam

546 Rong. The forest communities described in this study, particularly those at Yok Don, are 547 growing in remarkably low nutrient environments. MANUSCRIPT The nutrient dynamics at play between 548 the low nutrient, acidic soils, and the diverse and complex plant communities are likely to

549 be finely balanced. The species present in all of these forests appear to have adapted to

550 challenges, such as deficiency and toxicity, caused by very low pH.

551 Beyond local-plant soil relationships, there is scope to develop this Central Highlands

552 forest survey elevation transect (currently three points: 192 m, 686 m and 924 m) and test if

553 forests of Vietnam follow trends identified by other transect-based research, such as in

554 Mexico (Vazquez and Givnish, 1998) and Malaysia (Kitayama 1992; Palmiotto et al

555 2004). Moreover,ACCEPTED these little-studied forests provide new opportunities to test theoretical

556 relationships, such as between tropical plant diversity and soil fertility, known from other

557 tropical evergreen forests (Gentry 1988; Huston, 1980), and dipterocarp forests (Davies et ACCEPTED MANUSCRIPT

558 al 2005; Palmiotto et al 2004). Finally, the low soil fertility that characterises these

559 communities indicates that they are likely to contain a larger proportion of total ecosystem

560 nutrients in the standing biomass (compared with systems with higher soil nutrient pools).

561 For this reason, these forest tree communities highly susceptible to damage through the

562 unrestricted removal of that biomass (e.g. harvesting or fire). The nutrient dynamics of

563 these systems need to be better understood in order to be able to manage them into the

564 future.

565

566 Conclusion

567 The forests of the Central Highlands of Vietnam are biologically significant, threatened and

568 poorly described in the international scientific literature. From three sites, spanning 800 m 569 elevation, we describe six forest communities, MANUSCRIPT their indicator species and soil 570 characteristics. Patterns in diversity varied with scale and methods: species richness ranged

571 from a mean of five to 23 species/400 m 2 subplot, whereas sites (2 ha) had one to three

572 communities. Remarkably, the site with the lowest subplot species richness was also the

573 site with the most communities, and poorest soils. Many questions remain about drivers of

574 diversity in these forests - and this study provides a strong baseline.

575 576 Conflicts of interestACCEPTED 577 The authors declare that there is no conflicts of interest.

578 ACCEPTED MANUSCRIPT

579 Acknowledgments

580 For field work and data collection, we thank: Tran Hoang Hoa, Ngo Van Cam For

581 assistance with analysis: Jerry Vanclay and Michael Whelan. For comments on earlier

582 drafts of this manuscript: Tran Van Con. Funding: Australian Development Scholarship.

583

584 References

585 Balmford A, Whitten T. 2003. Who should pay for tropical conservation, and how could 586 the costs be met? Oryx 37: 238–250.

587 Bellingham PJ, Stewart GH, Allen RB 1999. Tree species richness and turnover throughout 588 New Zealand forests. Journal of Vegetation Science 10: 825–832.

589 Berdin FR, Tran MT, Truong DT, Tran VH Deckers J, Langohr R. 1999. Soil resources of 590 Gialai Province: Correlation of the Vietnamese Soil Classification System with the World 591 Reference Base for Soil Resources . K.U. Leuven University, Belgium, Internal Project 592 Report—NIAPP-KULeuven, Land Evaluation forMANUSCRIPT Land Use Planning and Development of 593 Sustainable Agriculture in South Vietnam.

594 Biondi E, Zuccarello V. 2004. Modelling environmental responses of plant associations: a 595 review of aome critical concepts in vegetation study. Critical Reviews in Plant Sciences 23: 596 149–156.

597 Blanc L, Maury-Lechon G, Pascal JP. 2000. Structure, floristic composition and natural 598 regeneration in the forests of Cat Tien National Park, Vietnam: An analysis of the 599 successional trends. Journal of Biogeography 27: 141–157.

600 Chapin FS, Zavaleta ES, Eviner VT, Naylor RI, Vitousek PM, Reynolds Hl et al. 2000. 601 Consequences of changing biodiversity. Nature 405: 234–42.

602 Condit R, Hubbell SP, Lafrankie JV, Sukumar R, Manokaran N, Foster RB, Ashton PS. 603 1996. Species-areaACCEPTED and species-individual relationships for tropical trees: a comparison of 604 three 50-ha plots. Journal of Ecology 84: 549–562. ACCEPTED MANUSCRIPT

605 Dao THH, Hölscher D. 2015. Red-listed tree species abundance in montane forest areas 606 with differing levels of statutory protection in north-western Vietnam. Tropical 607 Conservation Science 8: 479–490.

608 Davies SJ, Tan S, Lafrankie JV, Potts MD. 2005. Soil-related floristic variation in a 609 hyperdiverse dipterocarp forest. Ecological Studies: Pollination Ecology and Forest 610 Canopy : 22–34.

611 De Cáceres M, Legendre P. 2009. Associations between species and groups of sites: indices 612 and statistical inference. Ecology 90: 3566–3574.

613 De Cáceres M, Chytrý M, Agrillo E, Attorre F, Botta-Dukát Z, Capelo J, et al. 2015. A 614 comparative framework for broad-scale plot-based vegetation classification. Applied 615 Vegetation Science 18: 543–560.

616 Dufrêne M, Legendre P. 1997. Species assemblages and indicator species the nees for a 617 flexible asymmetrical approach. Ecological Monographs 67: 345–366.

618 Flather CH. 1996. Fitting species-accumulation functions and assessing regional land use 619 impacts on avian diversity. Journal of Biogeography 23: 155–168.

620 Food And Agriculture Organization Of The United Nations. 2006. Guidelines for Soil 621 Description (Fourth). Rome: FAO. MANUSCRIPT 622 Gardner S, Sidisunthorn P, Anusarnsunthorn V. 2000. A field guide to forest trees of 623 northern Thailand . Kobfai Publishing Project, Bangkok.

624 Gardner TA, Barlow J, Araujo IS, Ávila-Pires TC, Bonaldo AB, Costa JE, et al. 2008. The 625 cost-effectiveness of biodiversity surveys in tropical forests. Ecology Letters 11: 139–150.

626 General Statistical Office of Vietnam (GSO). 2013. Statistical Handbook of Vietnam . 627 (General Statistic Office of Vietnam, Ed.). Statistical Publishing House, Ha Noi.

628 Gentry AH. 1988. Tree species richness of upper Amazonian forests. Proceedings of the 629 National Academy of Sciences of the United States of America 85: 156–159.

630 Ghazoul J, Liston KA, Boyle TJB. 1998. Disturbance-induced density-dependent seed set 631 in Shorea siamensis (Dipterocarpaceae), a tropical forest tree. Journal of Ecology 86: 462– 632 473. ACCEPTED

633 Gibson L, Lee TM, Brook BW, Gardner TA, Barlow J et al. 2011. Primary forests are 634 irreplaceable for sustaining tropical biodiversity. Nature 478: 378–381. ACCEPTED MANUSCRIPT

635 Gong H, Ye W, Hu X, Yang X. 2013. Tree species diversity and related mechanism in an 636 evergreen broad-leaved forest in Ailao Mountains, Yunnan, China. African Journal of 637 Agricultural Research 8: 134–144.

638 Hai NH, Wiegand K, Getzin S. 2014. Spatial distributions of tropical tree species in 639 northern Vietnam under environmentally variable site conditions . Journal of Forestry 640 Research 25: 257–268.

641 He F, Legendre P, Lafrankie JV. 1997. Distribution patterns of tree species in a Malaysian 642 tropical rain forest. Journal of Vegetation Science 8: 105–114.

643 Hoang VS, Baas P, Keßler PJA, Slik JWF, Ter Steege H, Raes N. 2011. Human and 644 environmental influences on plant diversity and composition in Ben En National Park, 645 Vietnam. Journal of Tropical Forest Science 23: 328–337.

646 Holdridge LR. 1947. Determination of World Plant Formations From Simple Climatic Data. 647 Science 105: 367–368.

648 Huston M. 1980. Soil nutrients and tree species richness in Costa Rican forests. Journal of 649 Biogeography 7: 147–157.

650 Huston MA. 1994 . Biological Diversity: The coexistence of species on changing 651 landscapes . Cambridge University Press, Cambridge. MANUSCRIPT 652 IUCN 2001. 2001 IUCN Red List Categories and Criteria: Version 3.1. IUCN Species 653 Survival Commission. IUCN, Gland, Switzerland and Cambridge, UK. .

654 Jackson JK. 1994. Manual of afforestation in Nepal (2nd ed.). Forest Resarch and Survey 655 Centre, Kathamandu.

656 Khurana E, Singh JS. 2004. Germination and seedling growth of five tree species from 657 tropical dry Ffrest in relation to water stress: Impact of seed size. Journal of Tropical 658 Ecology 20: 385–396.

659 Kindt R, Coe R. 2005. Tree diversity analysis: A manual and software for common 660 statistical methods for ecological and biodiversity studies . World Agroforestry Centre 661 (ICRAF), Nairobi.

662 Kitayama K. 1992.ACCEPTED An altitudinal transect study of the vegetation on Mount Kinabalu, 663 Borneo. Vegetatio 102: 149–171.

664 Krupnick GA, Kress WJ. 2003. Hotspots and ecoregions: A test of conservation priorities 665 using taxonomic data. Biodiversity and Conservation 12: 2237–2253. ACCEPTED MANUSCRIPT

666 Looijen RC, Van Andel J. 1999. Ecological communities: conceptual problems and 667 definitions. Perspectives in Plant Ecology, Evolution and Systematics 2: 210–222.

668 Magurran AE. 1988. Ecological Diversity and Its Measurement . Springer Netherlands.

669 Meyfroidt P, Lambin EF. 2008. Forest transition in Vietnam and its environmental impacts. 670 Global Change Biology 14: 1319–1336.

671 Meyfroidt P, Vu TP, Hoang VA. 2013. Trajectories of deforestation, coffee expansion and 672 displacement of shifting cultivation in the Central Highlands of Vietnam. Global 673 Environmental Change 23: 1187–1198.

674 Mittermeier RA, Myers N, Thomsen JB, Da Fonseca GAB, Olivieri S. 1998. Biodiversity 675 Hotspots and Major Tropical Wilderness Areas: Approaches to Setting Conservation 676 Priorities. Conservation Biology 12: 516–520.

677 Moormann FR. 1961. The soils of the Republic of Vietnam. Sai Gon, Republic of Vietnam: 678 Ministry of Agriculture.

679 Mucina L. 1997. Classification of vegetation: Past, present and future. Vegetation 680 Description and Data Analysis: A Practical Approach. Journal of Vegetation Science 8: 681 751–760. 682 Myers N. 1988. Threatened biotas: “Hot spots” inMANUSCRIPT tropical forests. Environmentalist 8: 187– 683 208.

684 Myers N, RUSSELL A, Mittermeier RA, Fonseca GAB, Kent J. 2000. Biodiversity hotspots 685 for conservation priorities. Nature 403: 853–858.

686 Nadeau MB, Sullivan TP. 2015. Relationships between plant biodiversity and soil fertility 687 in a mature tropical forest, Costa Rica . International Journal of Forestry Research 2015: 1- 688 13 .

689 Nguyen HH, Uria-Diez J, Wiegand K. 2016. Spatial distribution and association patterns in 690 a tropical evergreen broad-leaved forest of north-central Vietnam. Journal of Vegetation 691 Science 27: 318–327.

692 Nguyen HN. 2013. Atlas of Vietnam’s forest tree species . Agricultural Publishing House, 693 Ha Noi. ACCEPTED

694 Nguy ễn KV, Nguy ễn TH, Phan KL, Nguy ễn TH. 2000 . Bioclimatic Diagrams of Vietnam - 695 Các bi ểu đồ sinh khí h ậu Vi ệt Nam . Ha Noi National University, Ha Noi. ACCEPTED MANUSCRIPT

696 Nguyen TT, Baker P. 2016. Structure and composition of deciduous dipterocarp forests in 697 Central Vietnam: patterns of species’ dominance and regeneration failure . Plant Ecology 698 and Diversity 9: 589-601.

699 Oksanen J, Blanchet FG, Kindt R, Legendre P, Minchin PR, O’Hara RB, et al. 2013. vegan: 700 Community Ecology Package.

701 Olson DM, Dinerstein E, Wikramanayake ED, Burgess ND, Powell GVN, Underwood EC 702 et al. 2001. Terrestrial Ecoregions of the World: A New Map of Life on Earth. BioScience 703 51: 933 - 938.

704 Oniani OG, Chater M, Mattingly GEG. 1973. Some effects of fertilizers and farmyard 705 manure on the organic phoshorus in soils. Journal of Soil Sience 24: 1–9.

706 Palmiotto PA, Davies SJ, Palmiotto PA, Vogt KA, Davies SJ, Vogt KA, et al. 2004. Soil- 707 related habitat specialization in dipterocarp rain forest tree species in Borneo. Journal of 708 Ecology 92: 609–623.

709 Pham HH. 1999. Cây C ỏ Vi ệtnam (An Illustrated Flora of Vietnam ). The Youth Publishing 710 House, Ho Chi Minh City.

711 Phillips OL, Núñez Vargas P, Monteagudo AL, Cruz AP, Zans MEC, Sánchez WG, et al. 712 2003. Habitat association among Amazonian treeMANUSCRIPT species: A landscape-scale approach. 713 Journal of Ecology 91: 757–775.

714 Polimeni JM, Iorgulescu RI, Chandrasekara R. 2014. Trans-border public health 715 vulnerability and hydroelectric projects: The case of Yali Falls Dam. Ecological Economics 716 98: 81–89.

717 Puyravaud J-P, Pascal J-P, Dufour C. 1994. Ecotone Structure as an Indicator of Changing 718 Forest-Savanna Boundaries (Linganamakki Region, Southern India). Journal of 719 Biogeography 21: 581–593.

720 Rayment GE, Higginson FR. 1992. Australian laboratory handbook of soil and water 721 chemical methods . Inkata Press, Melbourne.

722 Schimper AFW, Balfour IB, Fisher WR, Groom P. 1903. Plant-geography upon a 723 physiological ACCEPTEDbasis . Clarendon Press, Oxford. 724 Slik JWF, Raes N, Aiba S-I, Brearley FQ, Cannon CH, Meijaard E, et al. 2009. 725 Environmental correlates for tropical tree diversity and distribution patterns in Borneo. 726 Diversity and Distributions 15: 523–532. ACCEPTED MANUSCRIPT

727 Sodhi NS, Koh LP, Brook BW, Ng PKL. 2004. Southeast Asian biodiversity: An 728 impending disaster. Trends in Ecology and Evolution 19: 654–660.

729 Sollins P. 1998. Factors influencing species composition in tropical lowland rain forest: 730 does soil matter? Ecology 79: 23–30.

731 Stott P. 1990. Stability and stress in the savanna forests of mainland South-East Asia. 732 Journal of Biogeography 17: 373–383.

733 Sukri RS, Wahab RA, Salim K A, Burslem DFRP. 2012. Habitat associations and 734 community structure of dipterocarps in response to environment and soil conditions in 735 Brunei Darussalam, Northwest Borneo. Biotropica 44: 595–605.

736 Tang CQ, Chiou C-R, Lin C-T, Lin J-R, Hsieh C-F, Tang J-W, et al. 2013. Plant diversity 737 patterns in subtropical evergreen broad-leaved forests of Yunnan and Taiwan. Ecological 738 Research 28: 81–92.

739 Teejuntuk S, Sahunalu P, Sakurai K, Sungpalee W. 2003. Forest Structure and Tree Species 740 Diversity along an Altitudinal Gradient in Doi Inthanon National Park, Northern Thailand. 741 Tropics 12: 85–102.

742 Thái VT. 1963. Phát sinh qu ần th ể và phân lo ại th ảm th ực v ật nhi ệt đới ở Vi ệt Nam. Nhà 743 xu ất b ản Nông nghi ệp, Hà N ội. MANUSCRIPT 744 Thái VT. 1999. Nh ững h ệ sinh thái r ừng nhi ệt đới Vi ệt Nam. Nhà xu ất b ản Khoa h ọc và K ỹ 745 Thu ật, H ồ Chí Minh.

746 Thomas P. 2013. Nageia fleuryi. The IUCN Red List of Threatened Species 2013 : 747 e.T39606A2930266. http://dx.doi.org/10.2305/IUCN.UK.2013- 748 1.RLTS.T39606A2930266.en

749 Toledo M, Peña-Claros M, Bongers F, Alarcón A, Balcázar J, Chuviña J, et al. 2012. 750 Distribution patterns of tropical woody species in response to climatic and edaphic 751 gradients. Journal of Ecology 100: 253–263.

752 Tr ần H. 2002. Tài nguyên cây g ỗ Vi ệt Nam . Nhà xu ất b ản Nông nghi ệp, H ồ Chí Minh.

753 Tran VC, Nguyen TT, Do HTT, Cao CK,Tran HQ, Vu TI, et al. 2013. Relationship 754 between abovegroundACCEPTED biomass and measures of structure and species diversity in tropical 755 forests of Vietnam. Forest Ecology and Management 310: 213–218. ACCEPTED MANUSCRIPT

756 Van Der Ent A, Erskine P, Mulligan D, Repin R, Karim, R. 2016. Vegetation on ultramafic 757 edaphic “islands” in Kinabalu Park (Sabah, Malaysia) in relation to soil chemistry and 758 elevation. Plant and Soil 403: 77–101.

759 Vázquez JAG, Givnish TJ. 1998. Altitudinal gradients in tropical forest composition, 760 structure, and diversity in the Sierra de Manantlan. Journal of Ecology 86: 999–1020.

761 Vieilledent G, Gardi O, Grinand C, Burren C, Andriamanjato M, Camara C, et al. 2016. 762 Bioclimatic envelope models predict a decrease in tropical forest carbon stocks with 763 climate change in Madagascar. Journal of Ecology 104: 703–715.

764 Vietnamese Academy Of Science And Technology . 2007. Vietnam Red Data Book part II. 765 Plants . Nature Science and Technology Publishing House, Ha Noi.

766 Viranant V, Kaewaumput T, Charoensuk S, Buajung P. 2009. Natural watershed recovery 767 estimation after 40 years of forest rehabilitation in Khun Khong Watershed Research 768 Station, Chiang Mai, Thailand. Thai Journal of Forestry 28 :58-72.

769 Webb CO, Peart DR. 2000. Habitat associations of trees and seedlings in a Bornean rain 770 forest. Journal of Ecology 88: 464–478.

771 Wikramanayak ED, Rundel PW, Boonratana R. (n.d.). Ecoregions: Southeastern Asia: 772 Vietnam into Laos and Cambodia . World Wildlife MANUSCRIPT Fund. 773 Williams PH, Gaston KJ. 1994. Measuring more of biodiversity: Can higher-taxon richness 774 predict wholesale species richness? Biological Conservation 67: 211–217.

775 Wilson, JB & Chiarucci A. 2000. Do plant communities exist? Evidence from scaling-up 776 local species-area relations to the regional level. Journal of Vegetation Science 11(5): 777 773–775.

778 Wilson JB, Peet RK, Dengler J, Pärtel M. 2012. Plant species richness: The world records. 779 Journal of Vegetation Science 23: 796–802.

780

ACCEPTED ACCEPTED MANUSCRIPT

781 Figure captions

782 Figure 1. Ward’s hierarchical clustering diagram of communities – based on Bray-Curtis 783 distance.

787

788 Figure 2. The significant indicator species for each community. The values behind each 789 species code are the indicator values, which range between 0 and 100 (zero means no 790 association with the cluster, and 100 means maximum association with the cluster); 791 Following the indicator value is the significance value (p): 0 ‘***’ 0.001 ‘**’ 0.01 ‘*’. 792

795

796 Figure 3. Species accumulation curves for plant communities, subplot area = 400 m 2. 797 Different lines refer to different communities, grey lines are estimated asymptote, black 798 lines are measured species richness.

ACCEPTED ACCEPTED MANUSCRIPT APPENDICES

Appendix Figure 1. Vietnam showing study sites (black circles), major cities (grey circles). Central Highlands region is coloured in grey. MANUSCRIPT

ACCEPTED ACCEPTED MANUSCRIPT

Appendix Figure 2. Plot and subplot sampling design, filled cells indicate where soil sample were taken.

MANUSCRIPT

ACCEPTED ACCEPTED MANUSCRIPT

MANUSCRIPT

Appendix Figure 3. Key soil factors according to communities

ACCEPTED ACCEPTED MANUSCRIPT

Appendix Figure 4 . Forest communities: Highland Lowslope - HLLS (A), Highland Upslope - HLUS (B), Highland Floodplain - HLF (C), Lowland DeMANUSCRIPTciduous – type 1 - LLD1 (D), Lowland Deciduous – type 2 - LLD2 (E), Lowland Mixed - LLM (F).

ACCEPTED ACCEPTED MANUSCRIPT

MANUSCRIPT

Appendix Figure 5. Soil profiles of communities: Highland Lowslope - HLLS (A), Highland

Upslope - HLUS (B), Highland Floodplain - HLF (C), Lowland Deciduous – type 1 - LLD1 (D),

Lowland Mixed -LLM (E), Lowland Deciduous – type 2 - LLD2 (F).

ACCEPTED ACCEPTED MANUSCRIPT

Appendix TABLE 1. Species area curve models for six communities

Communities a b c S r

HLF 86.89 -0.052 0.469 0.938 0.998

HLLS 59.03 -0.124 0.559 0.576 0.999

HLUS 49.17 -0.107 0.451 0.246 0.999

LLD1 63.03 -0.009 0.447 0.039 0.999

LLD2 14.16 -0.037 0.333 0.048 0.999

LLM 39.62 -0.078 0.672 0.133 0.999

Model function: y=a*(1-exp(b*x))^c where Y is number of species and x is effort (plots). a, b and c are constants with a being the asymptote. S = standard error and r = correlation coefficient MANUSCRIPT

ACCEPTED ACCEPTED MANUSCRIPT

Appendix Table 2. Species list.

Vietnamese Family No Species name code

1 Aceraceae Acer laurinum Hassk. Thích lá qu ế Acelau

2 Buchanania latifolia Roxb Mèn ven Buclat

3 Anacardiaceae Lannea coromandelica (Houtt.) Merr Cóc chu ột Lancor

4 Mangifera minutifolia Evrand Xoài r ừng Manmin

5 Polyalthia cerasoides (Roxb.) Bedd Đuôi trâu Polcer

6 Annonaceae Xylopia pierrei Hance Gi ền tr ắng Xylpie

7 Rauwenhoffia siamensis Scheff. Dù d ẻ Rausia

8 Apocynaceae Astonia scholaris (L.) R Br. Sữa Astsch 9 Araliaceae Schefflera heptaphylla MANUSCRIPT (L.) Frodin Chân chim tám lá Schhep Stereospermum cylindricum Pierre ex Bignoniaceae 10 Dop Quao vàng Stecyl

Canarium album (Lour.) Raeush. ex 11 DC. Trám tr ắng Canalb

Canarium littorela var. rufum (A. W. 12 Burseraceae Benn.) Leenh Trám nâu Canlitruf

13 Canarium subulatum Guillaumin Cà na Cansub

14 Garuga pierrei Guillaum Cóc đá Garpoi

15 Dialium cochinchinense Pierre Xoay Diacoc

ACCEPTEDGleditsia australis Hemsl. ex Forbes & 16 Hemsl Bồ kết Gleaus Caesalpiniaceae Peltophorum dasyrrhachis var. tonkinense (Pierre) K.Larsen & 17 S.S.Larsen Lim x ẹt Peldas

18 Sindora siamensis Teysm. ex Miq. Gụ mật Sinsia ACCEPTED MANUSCRIPT

19 Calophyllum polyanthum Wall. Cồng nhi ều hoa Calpol

20 Garcinia fusca Pierre Bứa l ửa Garfus Clusiaceae 21 Garcinia multiflora Champ. ex Benth. Dọc Garmul

22 Garcinia poilanei Gagnep Bứa poilane Garpie

Anogeissus acuminata (Roxb. ex DC.) 23 Guill. & Perr Chò nhai Anoacu

24 Terminalia alata Heyne ex Roth Chiêu liêu kh ế Terala Combretaceae 25 Terminalia chebula Retz Chiêu liêu ổi Terche

26 Terminalia corticosa Pierre ex Laness Kha t ử Tercor

27 Dillenia hookeri Pierre. Sổ hooker Dilhoo Dilleniaceae 28 Dillenia pentagyna Roxb. Tai t ượng Dilpen

29 Anisoptera costata Korth Vên vên Anicos

Dipterocarpus obtusifolius Teysm. Ex 30 Miq. Dầu trà beng Dipobt MANUSCRIPT 31 Dipterocarpus tuberculatus Roxb. Dầu đồng Diptub Ki ền ki ền phú 32 Dipterocarpaceae Hopea pierrei Hance qu ốc Hoppie

33 Hopea recopei Pierre Chò chai Hoprec

34 Shorea obtusa Wall Cà chít Shoobt

35 Shorea roxburghii G. Don Sến mủ Shorox

36 Shorea siamensis Miq. Cẩm liên Shosia

37 Diospyros apiculata Hiern. Nh ọ nồi Dioapi Ebenaceae 38 ACCEPTEDDiospyros pyrrhocarpa Miq. Th ị lửa Diopyr

39 Elaeocarpus floribundus Blume Côm trâu Ealflo

40 Elaeocarpaceae Elaeocarpus griffithii (Wight) A Gray Côm t ầng Ealgri

41 Elaeocarpus kontumensis Gagnep Côm Kon Tum Elagri ACCEPTED MANUSCRIPT

42 Elaeocarpus sp. Côm Elakon

43 Antidesma ghaesembilla Gaertn Chòi mòi Antgha

44 Aporosa dioica (Roxb.) Muell.-Arg Th ầu t ấu khác g ốc Apodio

45 Baccaurea harmandii Gagnep Dâu da qu ả đỏ Bachar

46 Baccaurea ramiflora Lour Dâu da đất Bacram

47 Bridelia cambodiana Gagnep. Th ẩu m ật lá to Bricam Euphorbiaceae 48 Cleidion spiciflorum (Burm. f.) Merr Mỏ chim Colave

49 Croton argyrata Blume Bạc lá Croarg

50 Endospermum chinense Benth. Vạng tr ứng Elasp.

51 Mallotus apelta (Lour.) Muell.-Arg Ba bét tr ắng Malape

52 Mallotus metcalfianus Croiz Ba bét đỏ Malmet

53 Juglandaceae Engelhardtia roxburghiana Wall. Ch ẹo tía Endchi

54 Ormosia balansae Drake Ràng ràng mít Ormbal

Placolobium vietnamenseMANUSCRIPT N. D. Khoi & Ràng ràng vi ệt Fabaceae 55 Yakovl nam Plavie

56 Pterocarpus macrocarpus Kurz Giáng h ươ ng Ptemac

57 Castanopsis chinensis (Spreng.) Hance Dẻ gai trung qu ốc Caschi

Castanopsis fissa (Champ. ex Benth.) 58 Rehder & E.H.Wilson Dẻ đấu n ứt Casfis

59 Castanopsis hystrix A. DC Kha th ụ nhi ếm Cashys

60 Castanopsis indica (Roxb.) A. DC. Cà ổi ấn độ Casind

Fagaceae Castanopsis pseudoserrata Hick. & 61 ACCEPTEDCam. Kha th ụ nguyên Caspse

Lithocarpus ducampii (Hickel & A. 62 Camus) A.Camus Dẻ đỏ Litduc

Lithocarpus stenopus (Hickel & 63 A.Camus) A. Camus Dẻ cọng m ảnh Litste

64 Dẻ qu ả vát Littru Lithocarpus truncatus (King ex ACCEPTED MANUSCRIPT Hook.f.) Rehd.

65 Flacourtaceae Flacourtia indica (Gurm.f.) Merr. Mùng quân Flaind

66 Hamamelidaceae Rhodoleia championii Hook.f. Hồng quang Rhocha

Cratoxylum cochinchinense (Lour.) 67 Blume Đỏ ng ọn Cracoc

Hypericaceae Cratoxylum formosum subsp. 68 pruniflorum (Kurz) Gogelein Thành ng ạnh nam craforpru

69 Cratoxylum pruniflorum (kurz) Kurz Thành ng ạnh Crapru

70 Incacinaceae Platea latifolia Blume Xươ ng tr ăn Plalat

71 Irvingiaceae Irvingia malayana Oliv. Ex Benn. Kơ nia Irvmal

72 Ixonanthaceae Ixonanthes reticulata Jack. Hà n ụ Ixoret

73 Beilschmiedia roxburghiana Nees Ch ắp ch ại Beirox

Cinnamomum bejolghota (Buch.-Ham. 74 ex Nees) Sweet Re g ừng Cingla

Cinnamomum glaucescens (Nees) 75 Drury MANUSCRIPT Re h ươ ng Cinine 76 Cinnamomum iners Rienw. ex Blume Qu ế rừng Claexc Lauraceae 77 Cinnamomun sp Re Cinsp.

78 Lindera annamensis Liou Liên đàn trung b ộ Linann

79 Litsea verticillata Hance Bời l ời vòng Litver

80 Machilus odoratissima Ness Bời l ời đỏ Macodo

81 Machilus parviflora Meissn. Kháo hoa th ưa Macpar

82 Barringtonia musiformis Kurz. Chi ếc cau Barmus Lecythidaceae 83 ACCEPTEDCareya arborea Roxb Vừng xoan Carabo

84 Fagraea fragrans Roxb. Trai nam b ộ Fagfra

85 Loganiaceae Strychnos nux-blanda A. W. Hill Mã ti ền qu ả cam Strnux

86 Strychnos spireana Dop Mã ti ền Spire Strspi ACCEPTED MANUSCRIPT

87 Lagerstroemia calyculata Kurz Bằng l ăng Larcal Lythraceae 88 Lagerstroemia crispa Pierre ex Laness. Bằng l ăng ổi Lagcri

Dạ hợp Nha 89 Magnolia candollei (Blume) Noot Trang Magcan

Michelia citrata (Noot. & Chalermglin) Q. N. Vu and N. H. Xia, comb. 90 Magnoliaceae nov. Gi ổi xanh qu ả to Miccit

91 Michelia floribunda Fin.& Gagnep. Gi ổi nhi ều hoa Micflo

92 Michelia mediocris Dandy Gi ổi xanh Micmed

93 Malvaceae Grewia bulot Gagnep. Bù l ốt Grebul

94 Aglaia perviridis Hiern Gội r ất xanh Aglper

95 Aglaia spectabilis (Miq.) Jain & Bennet Gội n ếp Aglspe Meliaceae 96 Aglaia tomentosa Teysm. & Binn Ngâu lông Agltom

97 Toona surenii (Blume) Merr. Xoan m ộc Toosur

Archidendron clypearia ( Jack) I. 98 Nielsen MANUSCRIPT Mán đỉa Arccly 99 Archidendron eberhardtii I. Nielsen Đái bò Arcebe Mimosaceae Archidendron robinsonii (Gagnep) I. 100 Nielsen Dái heo Arcrob

101 Xylia xylocarpa (Roxb.) Taub Căm xe Xylxyl

102 Artocarpus gomezianus Wall Chay nhung Artgom

Artocarpus rigidus subsp. asperulus 103 (Gagnep.) F.M.Jarrett Mít nài Artrigasp Moraceae 104 ACCEPTEDFicus racemosa L Sung Ficrac 105 Streblus macrophyllus Blume Du ối lá to Strmac

Sang máu h ạnh 106 Horsfieldia amygdalina (Wall.) Warb nhân Horamy Myristicaceae 107 Knema globularia (Lamk.) Warb Máu chó lá nh ỏ Kneglo

108 Knema pierrei Warb. Máu chó lá to Knepie ACCEPTED MANUSCRIPT

Decaspermum parviflorum (Lam.) 109 A.J.Scott Th ập t ử hoa nh ỏ Decpar

110 Syzygium cumini (L.) Skeels Trâm v ối Syzcum

111 Syzygium hancei Merr. & Perry Trâm hoa nh ỏ Syzhan

112 Syzygium sp Trâm Syzsp. Myrtaceae Syzygium syzygioides (Miq.) Merr. & 113 Perry Trâm ki ền ki ền Syzsyz

Syzygium wightianum Wall. Ex Wight 114 et Arn. Trâm tr ắng Syzwig

115 Syzygium zeylanicum (L.) DC. Trâm đỏ Syzzey

116 Ochnaceae Gomphia striata (Tiegh.) C. F. Wei Lão mai Gomstr

117 Dacrycapus imbricatus (Bl.) D. Laub. Thông nàng Dacimb

118 Podocarpaceae Nageia fleuryi (Hickel) de Laud. Kim giao Nagfle

119 Podocarpus neriifolius D. Don Thông tre Podner 120 Proteaceae Helicia cochinchinensis MANUSCRIPT Lour Mạ sưa nam b ộ Helcoc 121 Rhizophoraceae Carallia brachiata (Lour.) Merr. Trúc ti ết Carabo

122 Eriobotrya angustissima Hook.f. Sơn tra lá h ẹp Eriang Rosaceae 123 Prunus arborea (Blume) Kalkm. Xoan đào Pruarb

124 Mitragyna speciosa (Korth.) Havil. Giam đẹp Mitspe

125 Morinda tomentosa Heyn Nhàu nhu ộm Mortom

126 Rubiaceae Psychotria poilanei Pitard Lấu tuy ến Psypoi

127 Wendlandia glabrata DC. Gạc nai Wengla

128 ACCEPTEDWendlandia paniculata (Roxb.) A. DC. Ho ắc quang Wenpan

129 Acronychia pedunculata (L.) Miq. Bưởi bung Acrped

130 Rutaceae Clausena excavata Burm. F Nhâm hôi Clespi

131 Euodia meliaefolia (Hance) Benth Thôi chanh Euomel ACCEPTED MANUSCRIPT

132 Zanthoxylum avicennae (Lam.) DC. Mu ồng tru ống Zanavi

133 Nephelium lappaceum L. Chôm chôm Neplap

134 Sapindaceae Pavieasia annamensis Pierre Tr ường Pavann

135 Sapindus saponaria L. Bồ hòn Sapsap

136 Sapotaceae Madhuca alpinia (Chev.) Chev Sến núi cao Madalp

137 Ailanthus triphysa (Dennst.) Alston Thanh th ất Ailtri Simaroubaceae Eurycoma longifolia Jack subsp. 138 longifolia Bá bệnh Eurlon

Lòng mang đài 139 Pterospermum pierrei Hance tua Ptepie

140 Sterculiaceae Reevesia macrocarpa Li Trú qu ả to Reemac

Scaphium macropodium (Miq.) 141 Beumee. Ươ i Scamac

142 Symplocos lancifolia Sieb . & Zucc Dung lá thon Symlan 143 Symplocos laurina (Retz)MANUSCRIPT Wall. Dung lá to Symlau Symplocaceae Symplocos laurina var. 144 acuminata .(Miq.) Brand. Dung tr ắng Symlauacu

145 Symplocos poilanei Guillaum. Dung poilane Sympoi

Pyrenaria jonquieriana Pierre ex Th ạch châu trung 146 Laness. bộ Pyrjon Theaceae 147 Schima superba (DC.) Korth Chò xót Schsup

148 Colona evecta (Pierre) Gagn. Bồ an ch ở Colpoi Tiliaceae Cọ mai nháp lá 149 ACCEPTEDColona poilanei Gagnep nh ỏ Cracoc 150 Ulmaceae Gironniera subequalis Pl. Ngát Girsub

151 Unknown1 Unkunk Unknown 152 Unknown2 Unkunk

153 Verbenaceae Callicarpa arborea Roxb. Tu hú g ỗ Calarb ACCEPTED MANUSCRIPT

Premna corymbosa (Burm. f.) Rottb. & 154 Willd. Cách núi Precor

155 Vitex quinnata (Lour.) Williams. Đẻn 5 lá Vitqui

156 Vitex trifolia L Đẻn 3 lá Vittri

157 Vitex pinnata L Bình linh lông Vitpin

MANUSCRIPT

ACCEPTED ACCEPTED MANUSCRIPT

Appendix Table 3. ANOVA of soil variables among communities. Adjusted P-

values are from Tukeys HSD test.

Avail. P Total Nitrogen Exch. K

HLLS-HLF 0.795 0.999 0.457

HLUS-HLF 0.318 0.217 0.000***

LLD1-HLF 0.106 0.086 1.000

LLD2-HLF 0.018* 0.027* 0.998

LLM-HLF 0.011* 0.040* 1.000

HLUS-HLLS 0.945 0.477 0.031*

LLD1-HLLS 0.617 0.078MANUSCRIPT 0.745 LLD2-HLLS 0.342 0.030* 0.414

LLM-HLLS 0.210 0.041 0.729

LLD1-HLUS 0.979 0.003 0.005***

LLD2-HLUS 0.917 0.001*** 0.001***

LLM-HLUS 0.766 0.001*** 0.003

LLD2-LLD1ACCEPTED 1.000 1.000 1.000

LLM-LLD1 0.995 1.000 1.000

LLM-LLD2 0.998 1.000 0.999

Significance value (p): 0 ‘***’ 0.001 ‘**’ 0.01 ‘*’ ACCEPTED MANUSCRIPT 1 TABLE 1. Description of climate, soil types and topography in study sites (adapted from 2 Berdin et al. 1999; Nguy ễn et al. 2000; Ty et al. 2012)

Kon Ha Nung Yok Don Dam Rong

Location and elevation Kon Ha Nung EaSoup floodplain R’Mai Mountain; plateau belong to Lang 192 m (s.d.14) Biang Plateau, Chu 686 m (s.d. 16) Yang Sin Mountains

924 m (s.d.4 3)

Slope of site % 0-5 (aver. 2) 0-17 (aver. 3.7) 8-43 (aver. 23)

Mean daily min T of 19.5 21.1 16.7 coldest month ° C

Mean annual T ° C 23.5 23.7 18.2 Mean daily max T of 32.3 MANUSCRIPT 33.9 24.6 warmest month

Precipitation year (mm) 1532.5 1789 1865

Mean humidity (%) 82 81 84

Mean daily sunshine hours NA 6.8 6.4

Mean annual rain days 133 138 165

Weather station An Khe. 422 m, Buon Me Thuot. Da Lat. 1513 m, ~ 70 km south. 490 m, ~50 km ~30 km south-east. ACCEPTED south-east. ACCEPTED MANUSCRIPT

Topography, geography Highland, Lowland, Foot Hill and low Gently slopes, interfluves mountain on undulating to dominantly ancient metamorphic and steeply dissected alluvium areas, low granitic rock areas upland on hills of dominantly Elevation range basaltic and metamorphic or from 600-1613 granitic rock. granitic rock areas, Mean elevation and small alluvial around 800, to plains. <1000 Elevation range from 150-200 m

Soil classification (based on Ferralsols Luvisols/Lixisols; Acrisols (Ferralic) WRB’98) (Acric, Vertic) Planosols (Eutric, Acrisols Dystric, Skeletic); (Chromic), Luvisols Leptosols Acrisols (Alumic, (Dystric) on Hyperdistric, and granitic rock MANUSCRIPT Chomic) occur only. dominantly on metamorphic rock.

3

ACCEPTED ACCEPTED MANUSCRIPT

1 TABLE 2. Rare and threatened species according to community. IUCN is IUCN Red Listing 2 as at 2004, VN is Vietnamese Red List.

Species IU VN HL HL HL LL LL LL CN F LS US D1 D2 M

Anisoptera costata Korth EN EN 2

Hopea pierrei Hance EN EN 1 21

Hopea recopei Pierre EN 3

Shorea roxburghii G. Don EN 4

Knema pierrei Warb. VU 3 2 1

Mangifera minutifolia Evrand VU 2

Xylopia pierrei Hance VU VU 3

Nageia fleuryi (Hickel) de Laud. NT 2

Michelia floribunda Fin.& Gagnep. DD 59 Sindora siamensis Miq. MANUSCRIPT LR EN 2 1 Aglaia spectabilis (Miq.) Jain & Bennet LR VU 9

Lithocarpus truncatus (King ex Hook.f.) Rehd. VU 2

Castanopsis hystrix A. DC VU, 156 248 A1c,d

Magnolia candollei (Blume) Noot (syn DD 11 9 Magnolia betongensis ).

Pterocarpus macrocarpus Kurz VU 4

Total number listed species 6 4 4 1 0 6 3 ACCEPTED ACCEPTED MANUSCRIPT

2 1 TABLE 3. Biodiversity indices by community in 400m plots.

Richness Number of subplots Shannon Simpson E-even. Abun. sampled Aver. Min. Max.

Total 154 2.06 0.80 11.92 2 23 0.80 24.23

Dam Rong 50 2.40 0.87 15.52 8 23 0.74 33.62

HLLS 26 2.39 0.87 15.70 10 23 0.71 39.37

HLUS 24 2.41 0.86 15.30 8 22 0.77 26.87

Ha Nung 50 2.63 0.92 16.02MANUSCRIPT 8 22 0.89 22.72 HLF 50 2.63 0.92 16.02 8 22 0.89 22.72

Yok Don 54 1.20 0.62 4.80 2 11 0.77 16.94

LLD1 12 1.12 0.55 5.58 4 11 0.59 32.42

LLD2 25 1.07 0.60 3.52 2 5 0.87 9.04

LLM 17 1.45 0.69 6.12 3 10 0.74 17.65

2 Aver.: Average;ACCEPTED Min.: Minimum; Max.: Maximum; E-even.: Evenness; Abun.: Abundance ACCEPTED MANUSCRIPT

1 TABLE 4. Topsoil (0-20cm) characteristics and summary plot data.

HLF HLLS HLUS LLD1 LLD2 LLM Soil

pproperties Mean s.d Rating 1 Mean s.d Rating Mean s.d Rating Mean s.d Rating Mean s.d Rating Mean s.d Rating

V.strongly V.strongly V.strongly V.strongly V.strongly V.strongly

pH(CaCl 2) 3.75 0.07 Acidic 3.92 0.05 Acidic 3.80 0.25 Acidic 3.92 0.10 Acidic 4.08 0.11 Acid 4.08 0.05 Acid

TOC (%) 1.99 0.22 Mod. 2.40 0.49 High 3.63 3.10 High 0.57 0.08 Low 0.43 0.18 Low 0.49 0.24 Low

TN (%) 0.21 0.03 Medium 0.22 0.03 Medium 0.32 0.24 High 0.05 0.00 Low 0.06 0.02 Low 0.05 0.03 Low MANUSCRIPT Avail P(ppm) 1.69 0.96 Low 1.30 0.38 V.low 0.94 0.02 V.low 0.61 0.07 V.low 0.53 0.17 V.low 0.37 0.11 V.low

Exch Ca 2 0.71 0.17 V.low 0.95 0.08 V.low 1.26 0.60 V.low 1.16 0.43 V.low 1.47 0.51 V.low 0.90 0.16 V.low

Exch K2 0.07 0.02 V.low 0.18 0.06 V.low 0.42 0.33 Mod. 0.07 0.00 V.low 0.05 0.01 V.low 0.08 0.02 V.low

Exch Mg 2 0.32 0.21 Low 0.34 0.08 Low 0.46 0.13 Low 0.51 0.25 Low 0.86 0.51 Low 0.27 0.08 Low ACCEPTED ACCEPTED MANUSCRIPT

Exch Na 2 0.17 0.02 Low 0.18 0.02 Low 0.42 0.43 Mod. 0.16 0.01 Low 0.16 0.01 Low 0.16 0.01 Low

CEC 2 5.93 1.63 V.low 5.94 0.62 V.low 8.08 3.70 Low 3.76 0.38 V.low 4.14 1.04 V.low 3.28 0.94 V.low

Ca/Mg 2.2 Low 2.8 Low 2.7 Low 2.3 Low 1.7 Low 3.3 Low

BSP% 21 Low 28 Low 32 Low 51 Mod. 61 Mod. 43 Mod.

Texture Clay Clay loam Clay loam Loamy sand Loam Loamy sand

2 1Ratings were based on Hazelton & Murphy (2007) MANUSCRIPT 3 2mequiv./100 g

4 3 Base Saturation Percentage = 100 * sum of exchangeable cations / CEC

ACCEPTED ACCEPTED MANUSCRIPT

MANUSCRIPT

ACCEPTED ACCEPTED MANUSCRIPT

MANUSCRIPT

ACCEPTED ACCEPTED MANUSCRIPT

MANUSCRIPT

ACCEPTED