BIOTROPIA Vol. 27 No. 1, 2020: 69 - 79 DOI: 10.11598/btb.2020.27.1.1122 69 ASSOCIATION OF TREE COMMUNITIES WITH SOIL PROPERTIES IN A SEMI DECIDUOUS FOREST OF PERLIS, PENINSULAR MALAYSIA RADHIAH ZAKARIA1*, MOHD NIZAM MOHD SAID2,3 AND FARIDAH HANUM IBRAHIM4 1Faculty of Public Health, Universitas Muhammadiyah Aceh, Banda Aceh 23123, Indonesia 2School of Environmental and Natural Resource Sciences, Faculty of Science and Technology, Universiti Kebangsaan Malaysia, 43600 Bangi, Selangor, Malaysia 3Institute of Climate Change, Universiti Kebangsaan Malaysia, 43600 Bangi, Selangor, Malaysia 4Faculty of Forestry, Universiti Putra Malaysia, 43400 Serdang, Selangor, Malaysia Received 20 August 2018 / Accepted 16 November 2018 ABSTRACT Plant community distribution is associated with environmental factors, particularly, the soil properties of habitats. This study was conducted to determine the effect of soil properties on the association of tree communities within three distinct habitats in a semi-deciduous forest in Perlis State Park (PSP), Perlis. Eighteen plots of 40 × 60 m (0.24 ha each) with sampling areas of 1.92 ha (8 plots) in Setul Formation, 0.96 ha (4 plots) in Granite and 1.44 ha (6 plots) in Kubang Pasu Formation (totalling 4.32 ha) were established at the PSP. All trees with 5.0 cm and above diameter at breast height (dbh) were enumerated, while the top soil samples were collected from each plot for soil analyses. A total of 412 tree species, 207 genera, and 68 families were recorded; 270 tree species from 152 genera and 57 families in the Setul forest; 204 tree species of 130 genera and 50 families in the Granite forest; and 109 tree species from 76 genera and 31 families in the Kubang Pasu forest. Euphorbiaceae was the most represented family at Setul, Granite and Kubang Pasu with 36, 19 and 12 species, respectively. Soil properties significantly varied among the study sites. Setul had loam, Kubang Pasu had clay- loam, and Granite had the sandy-loam texture. The soils were acidic and had low to high concentrations of available nutrients. Ordinations using canonical correspondence analysis indicated that the soil factors play an important role in the distribution and diversity of plants in these forest habitats. Keywords: canonical correspondence analysis, Perlis State Park, semi-deciduous forest, vegetation– environment relationship INTRODUCTION In tropical rain forest, the tree species composition varies with the type of habitat (Richards 1952) and their distributions are often associated with environmental factors (Newbery & Proctor 1984; Fujii et al. 2018). Studies conducted in different tropical regions had emphasized the influence of environmental factors, especially soil properties, on plant species distributions (Wentworth 1981; Oliveira- Filho et al. 2001; Nizam et al. 2013). Relationships between soil variables and plant species distribution had been discovered in various habitats, such as in granite and limestone areas (Nizam et al. 2013), grasslands (Cachovanová et al. 2012), savannas (Barruch 2005) and the tropical rain forests (Silk et al. 2010; Sukri et al. 2012). Peninsular Malaysia is predominantly of the rainforests type and comprise only a small restricted part of a semi-deciduous forest. The semi-deciduous forest in Peninsular Malaysia is located near Kra Isthmus, a transitional zone where Malesian and Indochinese floristic regions intersect (van Steenis 1979; Whitmore 1984; Middleton 2003). The State of Perlis, which is situated in the northernmost of Peninsular Malaysia, is the only part in this country with a semi-deciduous forest (Faridah-Hanum 2006). *Corresponding author, e-mail: radhiah@unmuha.ac.id https://link.springer.com/article/10.1007/s12224-012-9131-3#auth-1 BIOTROPIA Vol. 27 No. 1, 2020 70 The Perlis State Park (PSP) exhibits a semi- deciduous forest with most areas consisting of limestone hill forests and a small portion of granite-based parent material. This study aimed to determine the association between the tree communities and their soil properties under different forest habitats with different rock formations. This study is timely because the increasing human pressures on limestone and granite habitats compromise their integrity, diversity and functions. Limestone forests are being replaced by plantations, agricultural crops and forages to increase their productivity. MATERIALS AND METHODS Study Site and Tree Sampling A semi-deciduous forest in PSP, Perlis, Peninsular Malaysia, was selected as the study site (latitude 634'N to 643'N, longitude 10010'E to 10013'E) (Fig. 1). Tree samplings were conducted in 18 plots of 40 × 60 m each, covering a survey area of 4.32 ha. Due to different topography of the selected forest habitats, several plots were selectively established to avoid rocks and huge boulders. As such, the number of plots that were established also varied between forest habitats of different geological formations; eight plots (1.92 ha) in the Setul, six plots (1.44 ha) in the Kubang Pasu, and four plots (0.96 ha) in the Granite. All trees with 5 cm and above diameter at breast height (dbh) were enumerated and identified using the published keys (Whitmore 1972; Whitmore 1973; Ng 1978; Ng 1989). Since the number of plots in each forest habitat was unequal as mentioned earlier, the rarefaction analysis was conducted using EcoSim software program (Gotelli & Entsminger 2003) to produce the rarefaction curves. Rarefaction allows comparison of the observed richness and diversity between sites even though sampling efforts are not equal, or samples differ in the total number of individuals (Lee & Chao 1994). The rarefaction analysis of the study site confirmed that although the sampling efforts between habitats were unequal, nevertheless the species richness was of the same trend with the actual observation in the survey plots (Zakaria et al. 2015). Perlis State Park Figure 1 Location of the Perlis State Park in the state of Perlis, Peninsular Malaysia Association of tree communities with soil properties – Zakaria et al. 71 Soil Sampling and Analysis Using the soil corer, the top soil samples at 0-15 cm depth were collected from 18 established plots of various habitats within the PSP. Five replicate samples were obtained from each plot and air dried at room temperature. The replicates were then homogenized together to represent one composite sample from each plot. Root fragments and unwanted materials were removed and the soils were lightly ground and passed through a 2 mm mesh sieve before analysis. The soil samples were analyzed for physical properties, i.e., particle size distribution and organic matter (OM) content as well as chemical properties, i.e., exchangeable acid cations (Al3+ and H+), exchangeable base cations (K+, Mg2+, Na+ and Ca2+) and available nutrients (phosphorus (P), potassium (K) and magnesium (Mg). OM content was measured using loss on ignition techniques, while the soil pH was recorded based on soil–water ratio of 1:2.5 (McLean 1967). By titration, the exchangeable acid cations were analyzed in 1 M KCl while the exchangeable base cations used the 1 M ammonium acetate extract (Black 1967; Shamshuddin 1981). To determine the cation concentrations, the extracts were run using the Perkin-Elmer Atomic Absorption Spectrophotometer (AAS) Model 3300. Available nutrients (P, K and Mg) in the soil samples were extracted using 1 M ammonium acetate–acetic acid. The concentration of P in the extract was recorded using Ultraviolet (UV) spectrophotometer, whilst K and Mg concentrations were measured using the AAS. Data Analysis The enumerated trees in the plots were investigated for overall floristic composition. The soil parameters among the study plots were tested for significant difference using t-test. Association between tree communities and soil variables was analyzed using the multivariate techniques of canonical correspondence analysis (CCA) (ter Braak 1987; ter Braak & Prentice 1988; ter Braak 1992), which is available in CANOCO for Windows 4.56 (ter Braak 2009). Tree species with four occurrences, or less, were not included in the analysis because they weaken the uniformity when assigned to the groups (Legendre & Gallagher 2001; Barruch 2005). The species with four occurrences or less were disregarded to limit the number of species to less than 200 and consequently increase the accuracy of the results of CCA ordination diagram. Thus, 161 tree species were selected for the CCA. Detrended Correspondence Analysis (DCA) was first conducted on the species data to confirm their unimodality and subsequently justify the appropriateness of using CCA (Lepš & Šmilauer 2003). Soil variables selected for the CCA were soil pH, available P, available magnesium (Mg2+), calcium (Ca2+), natrium (Na+), potassium (K+) content, ammonium-N (NH4-N) and nitrate-N (NO3-N). The abundance values were log-transformed prior to matrix processing because their distributions were skewed toward extremely large values (ter Braak 1995). The correlation significance between matrices was tested using Monte-Carlo permutation test based on 499 random trials at a 0.05 significant level (Lepš & Šmilauer 2003). The ordination diagrams were subsequently plotted by CANODRAW 4.14 to illustrate the association patterns of tree communities in relation to soil properties. RESULTS AND DISCUSSION Tree Species Composition A total of 4,300 trees were recorded within the sampling area of 4.32 ha at Perlis State Park. The floristic composition included 412 tree species, 207 genera and 68 families. The Setul habitat showed high species richness of 270 species from 1,722 trees; belonging to 152 genera and 57 tree families and the plots in Granite habitat showed 204 tree species, from 1,245 trees; belonging to 130 genera and 50 families and the plots in the Kubang Pasu habitat contained 109 tree species from 1,333 trees, belonging to 76 genera and 31 families. In all three habitats, Euphorbiaceae was the most represented family with 36 species in Setul, 19 species in Granite and 12 in Kubang Pasu. On the contrary, 11 families were represented by only one species with a single individual in Setul, Granite and Kubang Pasu habitats. These families include Actinidiaceae (Saurauia pentapetala), Ancistrocladaceae (Ancistroclados BIOTROPIA Vol. 27 No. 1, 2020 72 tectorius), Anisophylleaceae (Anisophyllea apetala), Convolvulaceae (Erycibe albida), Ctenolophonaceae (Ctenolophon parvifolius), Hypericaceae (Cratoxylum formosum), Magnoliaceae (Magnolia elegans), Oleaceae (Chionanthus macrocarpus), Opiliaceae (Milientha suavis), Oxalidaceae (Sarcotheca griffithii) and Rhizophoraceae (Gynotroches axillaris). Hence, these species were not considered in the analysis of species association with soil variables. Soil Properties The three habitats have different soil properties as follows: the soil in Setul was loam whereas those in Granite and Kubang Pasu were sandy and silty-loam, respectively. Granite habitat had the lowest silt and clay contents but had the highest sand content at more than 69% among all study sites (Table 1). The high sand content in Granite is related to sandy parent material; the quartz sand derived from granitic hill was washed out and deposited a long time ago as reflected by the high sandy materials and extremely low clay and silt contents (Osumi 1979; Zaidey et al. 2010). In terms of OM content, the Granite soil had the significantly lowest OM content (3.12 ± 0.08) followed by the Setul soil (6.97 ± 1.96) and Kubang Pasu soil (10.72 ± 2.01) (Table 1). The lowest OM content in Granite is related to the lower percentage of silt and clay in the soil; because the absence of silt and clay generally increases decomposition rates and decreases OM for a particular set of environment interactions (Paul & Clark 1996). Overall, it is apparent that the amount of OM in the study sites is generally low. This low OM content is due to the high decomposition rate of OM in tropical rainforest soils (Longman & Jenik 1987). The low soil pH in Granite (4.39), Setul (5.64) and Kubang Pasu (5.98) indicated that the soils were strongly to moderately acidic. The soil pH in Granite was significantly lower than those of Setul and Kubang Pasu (Table 1). Granite soil has the highest acidity probably because it originated from the acidic parent rocks (Juo 1981) and the decomposition of OM also adds to the acidification of surface soil (Zaidey et al. 2010). A similar study recorded that most soils in Peninsular Malaysia tropical rainforests are acidic with pH values between 4.5 and 5.5 (Othman & Shamshuddin 1982). Table 1 Summary of soil variables in Setul, Granite and Kubang Pasu habitats at Perlis State Park, Perlis Soil parameters Granite (mean ± s.e.) Setul (mean ± s.e.) Kubang Pasu (mean ± s.e.) p value Organic matter (OM) (%) 3.12 ± 0.08a 6.97 ± 1.96b 10.72 ± 2.01b p < 0.05 Sand (%) 69.75 ± 2.12a 41.45 ± 4.68b 36.60 ± 8.52b p < 0.01 Silt (%) 14.15 ± 0.77a 35.61 ± 3.89b 44.20 ± 2.17b p < 0.01 Clay (%) 16.05 ± 1.42 21.42 ± 2.80 19.20 ± 2.23 NS Soil pH 4.39 ± 0.12a 5.64 ± 0.35b 5.98 ± 0.30b p < 0.01 Exchangeable cations (meq/100g) Ca2+ 0.94 ± 0.27a 5.5 ± 2.05b 7.87 ± 2.18b p < 0.05 Mg2+ 0.41 ± 0.02a 0.75 ± 0.16b 0.95 ± 0.44a p < 0.05 Na+ 0.20 ± 0.04 0.13 ± 0.02 0.13 ± 0.02 NS K+ 0.32 ± 0.03a 0.50 ± 0.06ab 0.57 ± 0.12b p < 0.01 Al3+ 0.60 ± 0.16 0.19 ± 0.14 0 NS H+ 0.50 ± 0.03 0.35 ± 0.04 0.34 ± 0.02 NS CEC 2.92 ± 0.2a 7.58 ± 2.09b 9.87 ± 2.54b p < 0.05 Available nutrients (ug/g) Phosphorus (P) 9.04 ± 0.31a 7.18 ± 0.69b 7.27 ± 0.37b p < 0.05 Nitrate-N (NO3-N) 27.00 ± 1.5 34.39 ± 6.45 30.96 ± 4.94 NS Ammonium-N (NH4-N) 3.59 ± 0.46a 18.64 ± 3.57b 30.81 ± 3.57b p < 0.01 Magnesium (Mg) 28.56 ± 0.51 105.35 ± 48.56 139.49 ± 66.78 NS Potassium (K) 175.02 ± 8.43a 227.27 ± 32.4b 253.80 ± 26.35ab p < 0.05 Note: Mean values in a row with the same letter were not significantly different. Association of tree communities with soil properties – Zakaria et al. 73 The cation exchange capacity (CEC in Granite was the lowest (2.92 ± 0.20 meq/100 g,) was significantly different from those in Setul (7.58 ± 2.09 meq/100 g) and Kubang Pasu (9.87 ± 2.54 meq/100 g). CEC varies with the clay–humus content of the soil. Clay soils usually have large CEC, whereas sandy soils have low CEC (Cruickshank 1972). The sites showed various concentrations of available macronutrients, namely magnesium (Mg), potassium (K) and phosphorus (P), and soluble nutrients, i.e., ammonium-N and nitrate-N (Table 1). The phosphorus concentration in Granite plots (9.04 ± 0.31 μg/g) was significantly higher than those in Setul (7.18 ± 0.69 μg/g) and Kubang Pasu (7.27 ± 0.37 μg/g). The available magnesium (Mg) in Granite (28.56 ± 0.51 μg/g), Setul (105.35 ± 48.56 μg/g) and Kubang Pasu (139.49 ± 66.78 μg/g) did not differ significantly. The available inorganic soil nitrogen is in the form of nitrate-N (NO3-N) and ammonium- N (NH4-N). The concentrations of these two soluble nutrients showed a consistent trend; the Granite habitat had lower concentration than the Setul and Kubang Pasu habitats. Ammonium-N was significantly low at Granite (3.59 ± 0.46 μg/g), higher at Setul (18.64 ± 3.57 μg/g) and the highest at Kubang Pasu (30.81 ± 3.57 μg/g). Nevertheless, the concentration of nitrate-N was not significantly different among the habitats. The lowest availability of inorganic soil nitrogen in Granite habitat was in line with the OM in the habitat. Tan (2009) mentioned that soil inorganic N content increases linearly as organic matter increases in the soil. The inorganic N was characterized by a large NO3-N, which exceeds the NH4-N at all the habitats; this might be attributed to the drier conditions in the study sites (Yamashita et al. 2003). Relationships between Tree Communities and Soil Variables DCA confirmed that the data on tree communities were unimodal with the length of gradient of the first axis at 5.758 (Table 2), which was greater than 4 standard deviation (SD) before being analyzed by CCA (Lepš & Šmilauer 2003). The eigenvalue produced by the DCA axis 1 (0.758) was high, indicating an environmental gradient where most species vary essentially in their abundance (ter Braak 1995). Plots in Granite (symbol x) were clustered together, while plots in Setul (symbol ○) and Kubang Pasu (symbol +) exhibited the gradient that occurs from granite to limestone (Fig. 2). These DCA ordination plots indicated the approximate locations of the sample plots. The plots that clumped together represented the plots with relatively similar floristic attributes, while the separated plots indicate dissimilar floristic composition. The unclearly separated plots in Kubang Pasu and Setul proved that Kubang Pasu Formation is overlain by Setul Formation (Basir & Zaiton 2002). Therefore, these habitats shared similar needs for mineral nutrient elements and other edaphic variables. Table 2 Summary of the Detrended Correspondence Analysis (DCA) of the vegetation data in 18 plots at the Perlis State Park, Perlis Axes 1 2 3 4 Total inertia Eigenvalues 0.758 0.464 0.393 0.174 6.527 Length of gradient 5.758 3.879 3.012 2.509 Cumulative percentage variance of species data 11.6 18.7 24.7 27.4 Sum of all eigenvalues 6.527 BIOTROPIA Vol. 27 No. 1, 2020 74 -1 6 -1 4 Figure 2 DCA ordination diagram of plots based on the abundance of tree species at Perlis State Park, Perlis Notes: x = plots in Granite; ○ = plots in Setul; + = plots in Kubang Pasu. Based on the CCA analysis, the species– environment correlations were high on the first and second axes with values of 0.988 and 0.974, respectively (Table 3). The eigenvalue was 0.561 for the first axis and 0.444 for the second axis. Additionally, the cumulative variation explained by the first three axes of the species- environment relationship was 55.2%. The Monte-Carlo permutation test also indicated a significant difference on the eigenvalues among the ordina-tion axes (p = 0.002). The CCA ordination plot showed the approximate locations of sample plots and the locations, lengths and directions of soil chemical variables (Fig. 3). Plots in Granite (1-4) were clustered together and associated with available phosphorus. Meanwhile, Setul and Kubang Pasu plots did not cluster into the well-defined group. Their plots were mostly assembled on the bottom right part of the diagram and were strongly correlated with several soil variables, i.e., magnesium (Mg), potassium (K), calcium (Ca), pH, available nitrate-N (NO3) and available ammonium-N (NH4). The CCA biplot of species and soil variables showed the species occurrence in relation to soil variables (Fig. 4). Most species were clumped on the centre part of the diagram and correlated with several soil variables such as available phosphorus (P) and available ammonium-N (NH4). Some species such as Antidesma cuspidatum (47) were clearly associated with calcium, whilst Ficus annulata (102) and Duabanga grandiflora (88) were strongly associated with ammonium-N (Fig. 4; Table 4). Furthermore, Hopea ferrea (34) had strong association with sodium. Aglaia affinis (93) had strong association with magnesium, whilst Bombax valetonii (23) was strongly associated with potassium and magnesium. However, Alstonia macrophylla (19), Terminalia subspathulata (30), Senna timorensis (87), Ficus aurata (103) and Colona merguensis (155) showed a weak association with soil factors when they are farther away from all the soil gradients (Fig. 4; Table 4). Table 3 Summary of the canonical correspondence analysis (CCA) of the vegetation and soil chemical properties in the 18 plots at Perlis State Park, Perlis Axes 1 2 3 4 Total inertia Eigenvalues 0.561 0.444 0.356 0.301 4.238 Species-environment correlations 0.988 0.974 0.961 0.987 Cumulative percentage variance of species data 13.2 23.7 32.1 39.2 Cumulative percentage variance of species-environment relation 22.7 40.8 55.2 67.5 Sum of all eigenvalues 4.238 Sum of all canonical eigenvalues 2.464 Association of tree communities with soil properties – Zakaria et al. 75 The variations in species distribution within the study plots were strongly correlated to edaphic factors and vegetation distribution pattern. Calcium and pH were some factors that strongly influenced the vegetation pattern in the limestone study area (Setul and Kubang Pasu), whereas sodium and phosphorus strongly affected the vegetation pattern in Granite area. Soils which developed on granite are low in calcium and magnesium (Burnham 1974), whereas soils that developed on limestone parent material such as in Setul and Kubang Pasu are mainly high in calcium and pH (Gauld & Robertson 1985). -1.5 1.5 -2 .0 3 .0 pHP Ca Mg Na K NO3 NH4 1 2 3 4 5 6 7 89 10 11 12 13 14 15 16 17 18 Figure 3 CCA ordination plot showing the approximate locations of sample plots and location, length and directions of soil variables Notes: x = plots in Granite (1-4); ○ = plots in Setul (5-12); + = plots in Kubang Pasu (13-18). -0.8 1.0 -1 .0 1 .0 1 8 9 12 16 19 23 30 31 34 37 40 47 56 58 62 66 75 83 87 88 90 91 92 93 101 102 103 105 109 110 115 132 134 138 143 144 147 155 157 161 pHP Ca Mg Na K NO3 NH4 Figure 4 Canonical correspondence analyses of the biplot for tree species and soil variables at the Perlis State Park Note: Numbers denote the species in the plots as listed in Table 4. BIOTROPIA Vol. 27 No. 1, 2020 76 Table 4 List of species with code numbers referring to diagram in Fig. 4 Species code Species Species code Species 1 Alangium kurzii 61 Sapium baccatum 2 Boea oppositifolia 62 Sauroupus suberosus 3 Buchanania arborescens 63 Trigonostemon villosa 4 Dracontomelon dao 64 Lithocarpus cantleyanus 5 Gluta elegans 65 Casearia capitellata 6 Gluta velutina 66 Homalium longifolium 7 Parishia insignis 67 Hydnocarpus castanea 8 Pentaspodon curtisii 68 Hydnocarpus curtisii 9 Semecarpus cochinchinensis 69 Hydnocarpus filipes 10 Semecarpus curtisii 70 Osmelia maingayi 11 Swintonia floribunda 71 Garcinia eugeniifolia 12 Alphonsea curtisii 72 Garcinia hambroniana 13 Goniothalamus tenuifolius 73 Garcinia nigroleniata 14 Mitrephora maingayi 74 Garcinia parvifolia 15 Orophea cuneiformis 75 Kayea kunstleri 16 Polyalthia glauca 76 Mesua ferrea 17 Xylopia ferruginea var. oxyantha 77 Stemonurus malaccensis 18 Xylopia magna 78 Cryptocarya ferrea 19 Alstonia macrophylla 79 Cryptocarya rugulosa 20 Kibatalia maingayi 80 Litsea machilifolia 21 Ilex cymosa 81 Barringtonia scortechinii 22 Radermachera glandulosa 82 Abdulmajidia rimata 23 Bombax valetonii 83 Leea aequata 24 Durio lowianus 84 Callerya artropurpurea 25 Canarium littorale f. rufum 85 Cynometra malaccensis 26 Dacryodes rubiginosa 86 Saraca cauliflora 27 Dacryodes rugosa 87 Senna timoriensis 28 Lophopetalum javanicum 88 Duabanga grandiflora 29 Parastemon urophyllus 89 Lagerstroemia floribunda 30 Terminalia subspathulata 90 Lagerstroemia ovalifolia 31 Tetrameles nudiflora 91 Memecylon caeraleum 32 Dipterocarpus costatus 92 Memecylon minutiflorum 33 Dipterocarpus fagineus 93 Aglaia affinis 34 Hopea ferrea 94 Aglaia argentea 35 Hopea latifolia 95 Aglaia simplicifolia 36 Parashorea stellata 96 Aglaia spectabilis 37 Shorea macroptera 97 Aphanamixis polystachya 38 Shorea siamense 98 Chisocheton ceramicus 39 Vatica cinerea 99 Chisocheton patens 40 Diospyros andamanica 100 Chukrasia tabularis 41 Diospyros buxifolia 101 Artocarpus dadah 42 Diospyros scortechinii 102 Ficus annulata 43 Diospyros sumatrana 103 Ficus aurata 44 Diospyros venosa 104 Ficus fistulosa 45 Elaeocarpus rugosus 105 Ficus microcarpa 46 Erythroxylum cuneatum 106 Ficus oligodon 47 Antidesma cuspidatum 107 Ficus sundaica 48 Aporosa aurea 108 Ficus variegata 49 Baccaurea griffithii 109 Streblus elongatus 50 Chondrostylis kunstleri 110 Streblus ilicifolius 51 Cleistanthus hirsutulus 111 Streblus macrophyllus 52 Croton argyratus 112 Streblus toxoides 53 Croton cascarilloides 113 Horsfieldia polyspherula 54 Croton laevifolius 114 Honsfieldia sucosa 55 Dimorphocalyx muricatus var. minor 115 Horsfieldia tomentosa 56 Drypetes longifolia 116 Knema laurina 57 Koilodepas longifolium 117 Knema patentinervia 58 Macaranga andamanica 118 Ardisia crassa 59 Macaranga lowii 119 Ardisia pachysandra 60 Mallotus peltatus 120 Maesa ramentacea Association of tree communities with soil properties – Zakaria et al. 77 Tabel 4 Continued Species code Species Species code Species 121 Syzygium cerasiforme 141 Pometia pinnata 122 Syzygium glaucum 142 Xerospermum noronhianum 123 Ochna integerrima 143 Chrysophyllum roxburghii 124 Galearia fulva 144 Palaquium hexandrum 125 Galearia maingayi 145 Palaquium microphyllum 126 Xanthophyllum affine 146 Eurycoma apiculata 127 Xanthophyllum griffithii 147 Pterocymbium javanicum 128 Prunus grisea 148 Pterospermum javanicum 129 Aidia densiflora 149 Pterospermum pectiniforme 130 Diplospora malaccensis 150 Pterygota alata 131 Ixora pendula 151 Sterculia cordata 132 Morinda elliptica 152 Sterculia gilva 133 Psydrax sp. 7 153 Sterculia parviflora 134 Psydrax sp. 8 154 Tetramerista glabra 135 Dimocarpus longan ssp. longan 155 Colona merguensis 136 Guioa bijuga 156 Grewia viminea 137 Harpullia cupanioides 157 Pentace strychnoidea 138 Nephelium costatum 158 Schoutenia accrescens 139 Nephelium lappaceum 159 Teijsmanniodendron coriaceum 140 Paranephelium macrophyllum 160 Vitex pinnata 161 Vitex siamica CONCLUSION The differences in floristic patterns between limestone (Setul and Kubang Pasu) and granite habitats at PSP suggested that environmental gradients influenced floristic composition. The soil properties such as organic matter, moisture, clay, silt, calcium and pH were among the factors that strongly associated with the limestone habitats, while sand and available phosphorus were strongly correlated with granite. Differences in the availability of elemental mineral nutrient between the soils of granite and limestone may have also contributed to the contrasting vegetation. In relation to soil properties, the different habitats displayed varied spatial distributions among the tree species. Identifying the key underlying gradients, abiotic conditions and major soil influences on the vegetation pattern is essential to gain information about the ecology of particular species and to formulate plans to protect and conserve this fragile habitat. ACKNOWLEDGEMENTS The authors would like to thank the Perlis Forestry Department, Malaysia for the cooperation and support throughout the study. The gratitude also goes to Mrs Latifah, Mr Ibrahim Edham, Ms Siti Eryani, the late Mr Badrol and Mr Ahmed Zainuddin Ibrahim for their assistance rendered during the fieldworks and laboratory procedures. The authors also acknowledge funding from the Research University Grant GUP-2013-056. REFERENCES Barruch Z. 2005. Vegetation-environment relationships and classification of the seasonal savannas in Venezuela. Flora 200:49-64. Basir J, Zaiton H. 2002. Geomorphology and geology of the Setul formation in Perlis. In: Latiff A, Kasim O, Yusoff AR, Faridah-Hanum I, editors. Kepelbagaian Biologi dan Pengurusan Taman Negeri Perlis: Persekitaran Fizikal, Biologi dan Sosial Wang Mu, Edn 1. Kangar (MY): Perlis Forestry Department. p. 67-82. Black CA. 1967. Soil-plant relationship. New York (US): John Wiley and Sons Inc. Burnham CP. 1974. The role of the soil forming factors in controlling altitudinal zonation on granite in Malaysia, In: Flenley JR, editor. Transaction of the Third Aberdeen-Hull Symposium on Malesia Ecology. Hull (UK): Department of Geography, University of Hull. Cachovanová L, Hájek M, Fajmonová Z, Marrs R. 2012. Species richness, community specialization and soil-vegetation relationships of managed grasslands https://link.springer.com/article/10.1007/s12224-012-9131-3#auth-1 https://link.springer.com/article/10.1007/s12224-012-9131-3#auth-2 BIOTROPIA Vol. 27 No. 1, 2020 78 in a geologically heterogeneous landscape. Folia Geobot 47:349-71. Cruickshank JG.1972. Soil geography, Edn 1. New York (US): John Wiley & Sons Inc. Faridah-Hanum I. 2006. Biodiverse Perlis in all its glory. IMPAK 1. Fujii K, Shibata M, Kitajima K, Ichie T, Kitayama K, Turner BL. 2018. Plant–soil interactions maintain biodiversity and functions of tropical forest ecosystems. Ecol Res 33:149-60. Gauld JH, Robertson JS. 1985. Soils and their related plant communities on the Dalradian limestone of some sites in Central Perthshire, Scotland. J Ecol 73:91-112. Gotelli NJ, Entsminger GL. 2003. EcoSim: Null models software for ecology, version 7. Burlington (US): Acquired Intelligence Inc. and Kesey-Bear. Juo ASR. 1981. Chemical characteristics. In: Greenland DJ, editor. Characterization of Soils in Relation to Their Classification and Management for Crop Production: Example from Some Areas of the Humid Tropics. Oxford (UK): Oxford University Press. Lee SM, Chao A. 1994. Estimating population size for closed capture-recapture models via sample coverage. Biometrics 50:88-97. Legendre P, Gallagher ED. 2001. Ecologically meaningful transformations for ordination of species data. Oecologia 129:271-80. Lepš J, Šmilauer P. 2003. Multivariate analysis of ecological data using CANOCO. Cambridge (UK): Cambridge University Press. Longman KA, Jenik J. 1987. Tropical forest and its environment, Edn 2. London (UK): Longman and Scientific & Technical. McLean ED. 1967. Methods of soil analysis, Part 2. In: Black CA, editor. Soil-plant Relationship. New York (US): John Wiley and Sons Inc. Middleton DJ. 2003. Progress on the flora of Thailand. Telopea 10(1):33-42. Newbery DM, Proctor J. 1984. Ecological studies in four contrasting lowland rain forests in Gunung Mulu National Park, Sarawak. IV. Associations between tree distributions and soil factors. J Ecol 72:475- 93. Ng FSP. 1978. Tree flora of Malaya, Vol. 3. Kuala Lumpur (MY): Longman. Ng FSP. 1989. Tree flora of Malaya, Vol. 4. Kuala Lumpur (MY): Longman. Nizam MS, Jeffri AR, Latiff A. 2013. Structure of tree communities and their association with soil properties in two fan-palm dominated forests of east coast Peninsular Malaysia. Trop Ecol 54(2):213-26. Oliveira-Filho AT, Curi N, Vilela EA, Calvalho DA. 2001. Variation in tree community composition and structure with changes in soil properties within a fragment of semi-deciduous forest in south-eastern Brazil. Edin J Bot 58(1):139-58. Osumi Y. 1979. Site classification for the effective land utilization based on soil in Mata Ayer Forest Reserve. Ishigaki (JP): Tropical Agriculture Research Institute, Ministry of Agriculture, Forestry and Fisheries. Othman Y, Shamshuddin J. 1982. Sains tanah [Soil science]. Kuala Lumpur (MY): Dewan Bahasa dan Pustaka. Paul EA, Clark FE. 1996. Soil microbiology and biochemistry, Edn 2. California (US): Academic Press. Richards PW. 1952. The tropical rain forest: An ecological study. Cambridge (UK): Cambridge University Press. Shamshuddin J. 1981. Asas sains tanah [The principle of soil science]. Kuala Lumpur (MY): Dewan Bahasa dan Pustaka. Silk JWF, Alba SI, Brearley FQ, Cannon CH, Forshed O, Kitayama K, … van Valkenburg JLCH. 2010. Environmental correlates of tree biomass, basal area, wood specific gravity and stem density gradients in Borneo’s tropical forests. Glob Eco Biog 19:50-60. Sukri RS, Wahab RA, Salim KA, Burslem DFRP. 2012. Habitat association and community structure of dipterocarps in response to environment and soil conditions in Brunei Darussalam, northwest Borneo. Biotropica 44(5):595-605. Tan KH. 2009. Environmental soil science, Edn 3. Florida (US): CRC Press. ter Braak CJF. 1987. The analysis of vegetation- environment relationships by canonical correspondence analysis. Vegatatio 69:69-77. ter Braak CJF, Prentice IC. 1988. A theory of gradient analysis. Advan in Ecol Res 18:271-317. ter Braak CJF. 1992. CANOCO A FORTRAN program for canonical community ordination. Ithaca (NY, US): Microcomputer Power. ter Braak CJF. 1995. Ordination, Edn 1. In: Jongman RH, ter Braak CJF, van Tongeren OFR, editors. Data Analysis in Community and Landscape Ecology. Wageningen (NL): Pudoc Wageningen. p. 91-173. ter Braak CJF. 2009. Program CANOCO version 4.56 biometric - Quantitative methods in the life and earth sciences, Plant Research International. Wageningen (NL): Wageningen University and Research Centre. https://link.springer.com/journal/12224 https://link.springer.com/journal/12224 Association of tree communities with soil properties – Zakaria et al. 79 van Steenis CGGJ. 1979. Plant-geography of east Malesia. Bot J Linn Soc 79(2):97-178. Wentworth TR. 1981. Vegetation on limestone and granite in the Mule Mountains, Arizona. Ecology 62(2):469-82. Whitmore TC. 1972. Tree flora of Malaya, Vol. 1. Kuala Lumpur (MY): Longman. Whitmore TC. 1973. Tree flora of Malaya, Vol. 2. Kuala Lumpur (MY): Longman. Whitmore TC. 1984. Tropical rain forests of the Far East, Edn 2. Oxford (UK): Clarendon Press. Yamashita T, Kasuya N, Wan Rashidah K, Suhaimi WC, Quah ES, Okuda T. 2003. Soil and belowground characteristics of Pasoh Forest Reserve. In: Okuda T, Manokaran N, Matsumoto Y, Niiyama K, Thomas SC, Ashton PS, editors. Pasoh Ecology of a Lowland Rain Forest in Southeast Asia. Kuala Lumpur (MY): Forest Research Institute Malaysia. p. 89-109. Zaidey AK, Arifin A, Zahari I, Hazandy AH, Zaki MH, Affendy H, … Nik Muhamad M. 2010. Characteristic soil properties of lowland and hill dipterocarp forests at Peninsular Malaysia. Int J Soil Sci 5(3):112-30. Zakaria R, Nizam MS, Faridah Hanum I. 2015. Floristic, diversity and structure of tree communities in three distinct forest habitats of Perlis State Park, Perlis, Peninsular Malaysia. J Sci 5(11):1016-23.