MATERIALS AND METHODS

Seed sources and study sites: The study was done using six seed sources of A. excelsa. The seeds originated from Bukit Lagong Forest Reserve (FRIM, Selangor), Pengkalan Arang (Terengganu), Narathiwat (Thailand), Manong Forest Reserve (Perak), Semengoh (Sarawak) and Pasir Mas (Kelantan). Table 1 shows all of seed origin and the location of the sources. The seeds were sown and germinated immediately in seed tray and later the seedlings were transplanted into polythene bags. They were about 3 months-old before being transplanted to the field in Merchang (Terengganu) and Batu Arang (Selangor) in year 2002.

Table 1:	Details of the six seed sources of A. excelsa

Evaluation of genetic variation: The selection of 30 trees from each seed source in both provenance trials was done randomly and leaves which were still attached to the twigs were preferred, in order to avoid wilting. The leaf samples were kept in plastic bags and placed in a cool container containing crushed ices. The samples were then ground immediately, if possible or preserved fresh in refrigerator at -20 °C to avoid the protein denaturation. An amount of 0.6 g of leaf was homogenised with liquid nitrogen using mortar and pestle. Then 1200 μL of leaf extraction buffer was added into the powdered form of leaf material with a ratio of 1:2. This is to create a slurry homogenate. The sample was filled into Eppendorf tube and labelled and later was kept in crushed ices before centrifugation at -10 °C, 1000 rpm for 3 min. The clear filterate was stored in a freezer at -70 °C.

Three types of gel and electrode buffers were used to detect the various enzymes. These include Histidine (H), Lithium (L) and Morpholine Citrate (MC). The mould frame was made of perspex, with internal dimensions of 18.8x15.0x0.6 cm and a glass plate of equal area. Both the frame and the glass plate were cleaned and dried properly by placing them in the oven at 40 °C to prevent breakage when pouring the hot gel solution. A 10.5% of potato hydrolysed starch (SIGMA-ALDRICH Inc.) was prepared by adding 31.5 g of starch into 100 mL of cold buffer in a dry Buchner flask. The remainder 200 mL of boiled buffer was then added into the flask and the buffer was swirled thoroughly to create an even suspension. The buffer was once again cooked together with the starch in the microwave oven until it boiled or big bubbles started to form. When the starch was fully cooked, the flask that contained the gel mixture was connected to the suction pump for a degassing process. The starch gel solution was then poured into the perspex mould and was left overnight at room temperature for hardening.

Table 2:	Enzymes assayed, their abbreviations and Enzymes Commission (EC) designations
Source: Weeden (1989)

Each resulting homogenate was absorbed onto a filter paper wick (i.e., 0.6x0.4 cm Whatman No. 3). This saturated wick and a tracking marker were inserted into the sliced gel that was tightened up by space bars before the electrophoretic run. The prepared gel (with wicks) was placed between the 2 electrode tanks containing 500 mL of buffer. The gel and the buffers were connected by two sets of bridge wicks (J-cloth or sponge) that were placed with one end overlapping about 1.0 to 3.0 cm on the gel and the other end dipping into the buffer. The set-up was covered with a thin polythene sheet to prevent evaporation of the buffer. A bag or plastic container containing ice was placed on the polythene sheet to avoid the drying up of gel during electrophoretic run due to high voltage or current. The electrophoretic run was carried out in a 4 °C refrigerator with an appropriate electric voltage depending on the buffer used. The wicks were taken out from the gel after 30 min of running.

After the electrophoretic run, the gel was removed from the mould in the refrigerator and cut into a smaller size by discarding only the sides. A mark was made to determine the origin of the sample arrangement by slashing the side. Later, the gel was sliced into three or four sheets with a thickness of approximately 2 mm per slice. The slices were stained for 19 different enzyme systems. The different enzyme systems used are as listed in Table 2. The stained gels were incubated in an oven at 37 °C for 30 to 60 min or until coloured bands developed. Then the gel was fixed using a fixing solution. Observation on the pattern of the bands was made and a clear zymogram was photographed whenever necessary. The banding of phenotypes was recorded in the prepared data sheets and followed with the interpretation.

Data analysis: The soluble enzymatic proteins from the leaves were assayed for each enzyme. The locus alleles were labeled as fast (F), medium (M) and slow (S) according to the decreasing anodal mobility. Standard measures of allozyme diversity within species and seed sources were calculated as described in Hamrick and Godt (1989). These measures included the percentage polymorphic loci (P), the mean number of alleles per locus (A), the effective alleles per locus (Ae), observed (H_o) and expected (H_e) heterozygosity, corrected for sample size (Nei, 1978). Other analyses done on the isozyme data also included Shannon`s information index (I) (Shannon and Weaver, 1949) and population genetic structure was measured using F-Statistics and, fixation index (Wright, 1978), gene flow estimation (Slatkin and Barton, 1989) and gene diversity among provenances (Nei, 1973). F<sub>IS</sub> compares the observed and expected levels of heterozygosity within provenances. A positive F<sub>IS</sub> indicates a lower than expected level of heterozygosity, possibly due to nonrandom mating (Wright, 1969). F<sub>IT</sub> measures the heterozygosity of an individual relative to the entire population. It is an indication of inbreeding due to nonrandom mating and random genetic drift within the subpopulations. G<sub>ST</sub> is a measure of the level of variation that resides within, rather than between, populations (Nei, 1973). Finally, the construction of a dendrogram was made based on Nei`s genetic distance using UPGMA (Nei, 1978). The programme is actually an adoption programme of NEIGHBOR and PHYLIP version 3.5c. The data were analysed using POPGENE version 1.21. The banding patterns were first encoded using Microsoft Excel for easier editing before being transformed into a POPGENE data file. The formulas for calculating those analyses are as follows:

The formula was used for allelic frequency are:

(1)

The observed heterozygosity (H_o) is simply the proportion of all genotypes that was heterozygote and may be expressed on a per-locus basis or averaged over all loci. Meanwhile, the expected heterozygosity was estimated using formula:

Where, X_ik is the frequency of the k^th allele at a locus in the i^th population.

The gene diversity in this populations is defined as H_e = (1 - _kΣ X_ik²). The average of H_e over loci is H_S. The gene diversity in the total population is:

Where, X_k = _iΣ X_ik/n , n being the number of populations. Then H_T-H_S, the diversity due to interpopulational gene differences, is denoted as D_ST and the coefficient of gene differentiation is D_ST/H_T denoted as G_ST.

The genetic identity (GI) between the two population or operational taxanomix unit (OTUs) was estimated according to Nei`s coefficient using this formula:

(4)

Where, X_i and Y_i = Frequencies of the i^th allele in the two OTUs or populations.

Morphological traits: A total of 40 trees were selected randomly from each seed source in both provenance trials to determine the variability of morphological characteristics when the trees were about two years old. Morphological traits such as leaf shape, leaf margin, leaf base, leaf angle and internode were also determined. The leaf angle was determined using the triangulation method. The length between nodes (inter-node) of two leaves (a) was measured. Then the length from the base of the node to the centre length of the leaf (b) was measured. The distance from the centre of the leaf to the base of the next leaf was then measured as (c). The value of the distances (c) was obtained using the following formula:

(5)

The general analysis for morphological traits was done according to one-way ANOVA. In addition, similarity coefficients based on these morphological characteristics were estimated using the formula given below;

(6)

RESULTS

Isozyme analysis

Intra-population variation: Table 3 shows the summary of genetic diversity from 35 loci scored including observed (H_o) and expected (H_e) heterozygosities, Shannon`s information index (I), effective number of alleles (A_e), proportion of polymorphic loci and the average number of alleles per locus per seed source.

The values of the observed heterozygosity varied from 0.0000 to 0.3333 with mean values ranging from 0.0229 (Pengkalan Arang) to 0.0451 (Bukit Lagong). Meanwhile, the values of expected heterozygosity ranged from 0.0000 to 0.5161. The mean values of H_e were found to vary from 0.0575 (Pengkalan Arang) to 0.0983 (Manong). Moreover, Shannon`s information index showed the diversity within the seed source to vary from 0.0000 to 0.8897 with mean values ranging from 0.1098 in the Pengkalan Arang seed source to 0.1852 in the Manong seed source. The effective number of alleles was found to range between 1.0000 and 2.0666 with mean values found to be within 1.0808 to 1.1468.

Ten loci were found to be monomorphic for all seed sources whereas the other 25 loci were polymorphic. The percentages of proportion polymorphic loci were found to vary between 31.43% (Pengkalan Arang and Pasir Mas) and 42.86% (Bukit Lagong and Manong). Meanwhile, the mean number of alleles per locus ranged from 1.48 to 1.69.

Genetic structure: F-statistics revealed varying fixation indices among the loci (Table 4). The estimates of F_IS revealed one locus (Pgi-3) with excess of heterozygotes, as revealed by the negative mean fixation index value. The 34 remaining loci exhibited heterozygote deficiencies. The F_IS values calculated ranged from -0.0265 at Pgi-3 to 1.0000 at Pgm-2 with the average of 0.5643. The F_IT values for all loci ranged from -0.0043 at Pgi-3 to 1.0000 at Pgm-2 with the mean value being 0.8038.

Genetic differentiation among seed sources as measured by F_ST showed that 21.32% of the total genetic variation was due to differences among seed sources; 78.68% of the isozyme variation resided within seed sources. The F_ST values ranged from 0.0000 to 0.9681 among the polymorphic loci. The estimated gene flow (Nm) based on the F_ST values was 2.9124.

The results obtained for genetic diversity in the seed source, measured by Nei`s (1973) index, through the partition of total genetic diversity (H_T) in the components of diversity within populations (H_S) and between populations (D_ST) showed relatively low values for total diversity (Table 5). The average G_ST (coefficient of differentiation) value indicated that approximately 29.84% of the observed variability was found among seed sources, more than that estimated by F_ST.

Genetic distance: Nei`s genetic distance (Nei, 1978) and the UPGMA dendrogram revealed low levels of genetic distance among seed sources. Most of the seed sources produced high similarities to each other with mean identity values ranging between 0.7727 and 0.9999 (Table 6). The most related seed sources were Pengkalan Arang and Pasir Mas with a genetic identity of 0.9999 between them. There were two clusters formed where the first cluster was made up of Pengkalan Arang, Pasir Mas, Narathiwat and Semengoh seed sources while the second cluster consisted of Bukit Lagong and Manong seed sources (Fig. 1). This clusteration is comparatively similar to the ones based on morphological similarities with a few notable exceptions. Four seed sources (Pengkalan Arang, Pasir Mas, Narathiwat and Semengoh) were assigned to CLUSTER I on the basis of isozyme and morphological data.

Table 3:	Summary of genetic diversity for six seed sources of A. excelsa using isozymes
Ho: Heterozygosity observed, He: Heterozygosity expected, I: Shannon`s information index, Ne: Effective number of alleles

Table 4:	F-statistics and gene flow for all loci
FIS, FIT and FST: Inbreeding coefficients, Nm: Gene flow

Table 5:	Nei`s (1973) statistics of genetic diversity for 18 polymorphic loci in six seed sources of A. excelsa
1: Mean values do not include monomorphic loci

Morphological variability and similarity: Most seed sources have the same pattern of qualitative characteristics for all leaf parameters except for Narathiwat and Semengoh seed sources which had one extra pattern in leaf margin (Table 7). Quantitative characteristics of leaf angle and internode showed significant differences among seed sources for both positions but no significant difference was detected on other parameters (Table 8).

Fig. 1:	Dendrogram of six seed sources of A. excelsa using Un-Weight Pair Group Cluster Analysis of Identity Coefficients (Nei`s, 1972)

Fig. 2:	Dendrogram of six seed sources of A. excelsa based on morphological characteristics (qualitative)

The mean morphological similarities for the qualitative characteristics ranged from 0.7930 between Pasir Mas and Bukit Lagong seed sources to 0.9952 between Semengoh and Bukit Lagong seed sources (Table 9). The dendrogram for six seed sources of A. excelsa based on these similarities is given in Fig. 2. Narathiwat-Pasir Mas seed sources and Pengkalan Arang-Semengoh seed sources formed one cluster while Bukit Lagong and Manong seed sources formed the other cluster.

Table 6:	Nei`s (1972) Coefficients of genetic identity (above diagonal) and genetic distance (below diagonal) among six seed sources of Azadirachta excelsa
C: Seed source code

Table 7:	Leaf characteristics of six A. excelsa seed sources

Table 8:	Analysis of variance for qualitative and quantitative traits of morphological variants
1Basal position, 2middle position, *Significant at p < 0.05, ns: Not significant

DISCUSSION

Intra- and inter-provenance variation: Genetic variation within populations is the basis for evolutionary change to occur. The amount of such variation is dependent on the species level. According to Hopper and Coates (1990), this is necessary to occur in order for plants to evolve and adapt under different conditions and numerous environments that they encounter during a single or many generations. There are five evolutionary processes that affect the level of genetic variation such as random genetic drift, selection, migration, mating and mutation. All these evolutionary processes affect the levels and distribution of genetic variation and the present state of the genetic resource is a result of their joint effects (Wickneswari, 1999).

The genetic variation among provenances is quantified by measuring the mean heterozygosity, the percentage of polymorphic loci and the number of alleles per locus. The range of expected heterozygosity calculated in the present study was from 0.0575 (Pengkalan Arang) to 0.0983 (Manong) and found to be within the range recorded by Norwati et al. (1997) on the same species but in different populations. These values were also generally consistent with those reported on tropical species by Hamrick and Loveless (1986).

The range of the proportion of polymorphic loci obtained was from 31.43% (Pengkalan Arang and Pasir Mas) to 42.86% (Bukit Lagong and Manong) and found to be within the range recorded on tropical species by Hamrick and Loveless (1986) in general. This range was also found to be similar to the ones reported by Norwati et al. (1997) on the same species.

The range of the average number of allele per locus for the six seed sources of A. excelsa obtained in the present study was found to be within the range recorded on tropical species. This range was also about similar than that recorded by Norwati et al. (1997) but lower than those recorded on other popular tropical tree species for plantation such as H. brasiliensis (Paiva et al., 1994a, b), P. macrocarpus (Liengsiri et al., 1995) and T. grandis (Kertadikara and Prat, 1995).

The mean average of expected heterozygosities was 0.077 and this value was found to be lower than those reported on tropical species by Hamrick and Loveless (1986) in general. This value is comparable to those reported by Mohamad et al. (1997) on the same species and Wickneswari and Norwati (1993) on A. auriculiformis. The value was again found to be lower than those reported on H. brasiliensis (Paiva et al., 1994a, b), P. macrocarpus (Liengsiri et al., 1995) and T. grandis (Kertadikara and Prat, 1995).

The mean value of the proportion of polymorphic loci of six A. excelsa seed sources was 36.67. This value was found to lower than those reported by Hamrick and Loveless (1986) on tropical species. The value was comparatively similar to those reported by Norwati et al. (1997) on the same species as well as A. auriculiformis (Wickneswari and Norwati, 1993), A. mangium (Moran et al. 1989b) and S. macrophylla (Kanzaki et al., 1996).

Meanwhile, the value of the average mean number of alleles per locus for six A. excelsa seed sources was found to be within those reported by Hamrick and Loveless (1986) for tropical species in general. The mean value obtained was again comparable to those reported by Norwati et al. (1997) on the same species as well as A. auriculiformis (Wickneswari and Norwati, 1993) and A. mangium (Moran et al., 1989b).

Table 9:	Morphological identity (above diagonal) and morphological distance (below diagonal) based on the morphological characteristics
C: Seed source code

Species which are widespread, long-lived woody and primarily outcrossed by wind or insect pollination have been reported to produce high levels of genetic diversity (Brown, 1978; Loveless and Hamrick, 1984; Hamrick and Godt, 1989; Hamrick et al., 1992). A. excelsa is also expected to produce higher values of heterozygosities based on mode of pollen and seed dispersal mechanisms. A. excelsa can be classified as one of the entomophilus species pollinated by insects such as bees and moth while the fruit is transported by bats and birds. In contrast, lower level of heterozygosity was reported on all seed sources, might be assumed that the seed sources consist of small population size. This could be due to the species was confined to scattered distribution along farm or plantation boundaries, roadsides or sparsely grown which could lead to the reduction in population size. According to Kijkar and Boontawee (1995), this species is thought to be an introduced species in Peninsular Malaysia and not indigenous and may established from a very restricted genetic base. Species with small population densities often increased the possibility of genetic drift, which will reduce the genetic variability as a result of bottlenecking and inbreeding (Moran and Hopper, 1987). This phenomena is well discussed by other researchers such as Hamrick (1983), Chamberlain et al. (1996a, b) and Liengsiri et al. (1995).

Although the information gathered was inadequate to relate concisely on the actual level of genetic variation especially on the number of mother trees but the information gathered from Pengkalan Arang seed source could be used to explain the lower level of genetic variation. Lower values of heterozygosities would suggest that the seeds were collected from a few mother trees. For instance, ground truthing at Pengkalan Arang revealed that seed sources might originated from only two to three mother trees separated about eight to 10 m away. Also, there is a possibility that the seeds may be from the same maternal tree and thus leading to genetic heterogeneity reduction and limited gene exchange.

In addition, seed sources from Bukit Lagong and Semengoh were obtained from research plot which is an artificial stand where the genetic base of the planted population is usually be rather narrow. This postulation is supported by Dusan (1992) who found a lower value of mean heterozygosity of 0.275 from artificial stand of Picea abies when compared to the value (0.322) obtained from virgin forest. This is because the artificial stand normally receives uniform treatment levels, which explain for unnecessary increased for heterozygosity.

The percentage of proportion polymorphic loci among provenances was 36.67%, which is lower than those reported for other tropical trees. The low value of polymorphism suggests that the seed sources in the present study had lesser opportunities to evolve. Normally, polymorphism is required as part of the adaptive strategies of populations in a heterogenous environment of the forest (Feret and Bergman, 1976). Consequently, a species has to evolve to compete for optimum growth against other species for growth resources such as space, light, nutrient and water. A concrete discussion cannot be done further since information on the actual details of their history, origin and the ecological background of source materials are not available.

Genetic distance: Isozyme analysis in A. excelsa revealed 35 loci where 25 of them being polymorphic. One (Pgi-3) out of 25 polymorphic loci showed significant excess of heterozygosity (Table 4). Meanwhile, there was a general excess of homozygosity within seed sources, as revealed by the positive mean fixation index value. According to Gregorius and Namkoong (1983), a lack of heterozygosity can be observed in a population because of restriction of gene flow within the whole provenance and of increased of relatedness between neighbour individuals. Similarly, the results noted in an isozyme analysis of population structure of Q. rubra (Schwarzmann and Gerhold, 1991), in which observed heterozygosities did not differ significantly from those expected under random mating and in a population study of F. sylvatica (Cuguen et al., 1988; Comps et al., 1990) in which F_IS and F_IT values were generally positive indicating heterozygote deficit relative to panmixia. Houston and Houston (1993) stated that the differences between two species in population structure might reflect differences in their reproductive strategies which result in differing mating patterns and existing stand structures. The positive mean of F_IT and F_IS values observed for this species implies an overall deficit of heterozygotes, perhaps caused by assortative mating between near neighbours related descent.

Generally, F_ST values with average 0.213 obtained in this study indicated moderate differentiation (Wright, 1978) between the two provenances which approximately 78.7% of the observed variation residing within seed source. This value was comparative to the one reported on G. sepium (0.172) by Chamberlain et al. (1996a). However, this value was higher than those reported on P. macrocarpus (0.121) by Liengsiri et al. (1995) and other tropical trees (0.119) by Hamrick et al. (1992). The value was also higher than those reported on F. sylvatica (0.060) by Cuguen et al. (1988), Populus tremuloides (0.030) by Jelinski and Cheliak (1992) as well as on European beech (0.054) by Comps et al. (1990). This level of population subdivision is high when compared with that of temperate tree species and this supports the notion that there is greater population differentiation in the tropics than in the temperate zone (Hamrick et al., 1992). Possible reasons for higher levels of population differentiation in the tropical species have been reviewed by Bawa (1983). They include lower population densities, more widely scattered populations that reduce gene flow and increase genetic drift and greater spatial variation in the natural selection pressure. However, higher levels of population differentiation in present study might also be due to bottleneck effects where the sources of A. excelsa are mostly not from natural population (Nei, 1975).

Nei`s (1978) statistics of genetic diversity was also calculated to analyse the distribution of allelic diversity. The average G_ST value indicated that approximately 29.84% of the observed variability was found among seed sources, more than the one estimated by F_ST and was higher than the average value found in other tropical tree species by Loveless (1992), i.e., G_ST = 10.9%. This value was also found to be higher than those reported on F. sylvatica (3.6%) by Comps et al. (1990), Pinus kesiya (3.9%), a wind-pollinated tropical pine by Boyle et al. (1991), Q. macrocarpa (7.6%) by Schnabel and Hamrick (1990) and Rugelia nudicaulis (1.5%) by Godt and Hamrick (1995). This result once again showed that most of the variation was found within seed sources.

The overall result corresponded to long-lived woody plants which acquired relatively high genetic diversity within species but most of the genetic diversity was found within populations and little existed among populations. The distribution of allozyme variation among populations is the product of interactions among several evolutionary factors. Of primary importance are selection, effective population size and the ability of the species to disperse pollen and seeds. On the other hand, species with more pollen and seed movement should have less differentiation among population than species with restricted gene flow (Hamrick, 1989). However, low level of diversity among seed sources and the high level of diversity within seed sources of A. excelsa in this study could be due to the certain relatedness among the seed sources or land races of this species whereby the seed sources might come from similar source in terms of ancestory genetic constituent. The reason is simply that A. excelsa is not usually found in the jungle or natural forests. It occurs mostly on farmland or marginal lands through shifting cultivation (Kijkar and Boontawee, 1995) and therefore likely to be naturalised in this area.

Cluster analyses: The dendrograms produced by the UPGMA clustering technique are presented in Fig. 1 and 2 for morphological and genetic values, respectively. The morphological and genetic identity values were found to range between 0.7930 and 0.9952 and 0.7727 and 0.9999, respectively. Such high ranges revealed that they were very much related both in terms of morphological and genetical aspects.

Two groups of clustering were formed using both techniques. Both dendrograms showed seed source of Pengkalan Arang, Narathiwat, Semengoh and Pasir Mas formed one cluster whereas Bukit Lagong and Manong seed sources formed the other group. Such relatedness among the seed sources are as expected based on their geographic distances among them. Narathiwat, Pasir Mas and Pengkalan Arang seed sources were located at a closer latitudes and longitudes. Similar clusteration based on latitudes, longitudes and altitudes have been reported for Casuarina cunninghamiana (Moore and Moran, 1989) and A. crassicarpa (Nor Aini et al., 2006). Normally, as geographic distance between population is small or decreases, then the genetic similarity will increase.

The pattern of such clusteration is also found to be associated with historical factors. The Bukit Lagong seed source was once thought to be originating from Perak while the source from Thailand was believed to have some relationships with Borneon sources as a result from the scientific expedition made earlier. This is because, (Roland Kueh, unpublished) reported that a group of scientists from Thailand made an expedition for seed collection of A. excelsa to Borneo Island (i.e., Sabah and Sarawak). Perhaps, through such activity could possibly caused such relatedness between Sarawak and Thailand.

The general pattern exhibited by the dendrogram based on morphological similarity values is similar to that based upon genetic identity values with a few notable exceptions. The uncertainty regarding the phenetic-phylogenetic correspondence is due primarily to variation caused by the confounding selective forces as a result of changes in environment. The majority of allozyme loci are neutral with respect to natural selection and their distribution is largely dependent on the stochastic events of mutation, migration, drift and founder effect compared to morphological characteristics. If so, UPGMA clustering of genetic identity values could provide a less biased view of the evolutionary relationships of the species than the clustering of morphological similarity values.

CONCLUSION

Genetic parameters were found to produce low levels of genetic variability for all A. excelsa seed sources. The Manong seed source was found to possess higher level of diversity followed by Bukit Lagong, Semengoh, Narathiwat, Pasir Mas and Pengkalan Arang seed sources. Moreover, low level of polymorphism might indicate that this species has lesser opportunities to evolve in different environment pressures.

The overall result corresponded that most of the genetic diversity found within seed sources with little existing among seed sources where it is generalised the genetic diversity of woody plants. The founder effect could help to explain such phenomenon for this species.

High values of morphological and genetic in cluster analyses suggests that the close relatedness of the seed sources. This argument is made based on the historical background and geographical distance among those seed sources. Seed sources having similar latitudes and longitudes were found to give smaller distance. Moreover, possible explanation on the relatedness between groups was also due to the source of the land races.

ACKNOWLEDGMENTS

We wish to express our deepest gratitude and thanks to Associate Professor Dr. Kamis Awang for his invaluable suggestions and comments. Our special thanks also extended to Mr. Salim Ahmad and Mr. Zakaria Taha for their assistance and co-operation during sample and information collection of the project. We would also like to thanks to Forestry Department of Peninsular Malaysia for providing us the research sites.