The research was carried out at Popa Mountain Park (PMP), where high diversity of plants (including medicinal plants) were found. The area was classified by Myanmar’s forest department as forest reserve in 1902. Popa Reserve was proposed as protected area by the Nature Conservation National Park Project conducted between 1981 and 1984. The area was subsequently declared a Protected Area in 1989. The park covers an area of about 10,000 ha, of which 88.7% is covered by forest [20] . Elevation ranges from about 300 to 1500 m above sea level. Mean maximum and minimum monthly temperatures respectively are 31.30˚C and 8.52˚C and the mean annual rainfall is about 1170 mm (Popa Forest Department office, 2013, unpublished data). The PMP includes a diverse range of vegetation types such as dry mixed deciduous forest, and dry dipterocarp forest (scrub indaing forest), dry forest (Than-dahat forest) and dry hill or evergreen forest [20] . All forests in PMP are second or third growth forests as results of timber harvesting and clearing for agriculture in the early twentieth century [21] .

A volcanic plug at the western foot of Mount Popa or Taung-kalat in local name, is a prominent landmark and widely known as a religious site for local people. Popa Mountain Park is also an important watershed for the surrounding area (especially for Kyaupadaung Township). Therefore, Popa Mountain Park was designated as a protected area for conservation of the forest, protection of the watershed of the Kyet-mauk-taung dam located at the southern edge of the park, conservation of medicinal plants for sustainable use, preservation of existing religious sites and to ensure sustainability of water sources, including natural springs [22] . To provide better protection for the forest within the area, and to regulate the local use of vegetation outside the boundary, an area of 10,360 ha was established as a buffer zone surrounding the Park in 2010. There are 45 villages located in the surrounding areas of the park. The major agricultural practices on the eastern side are cultivation of bananas, fruits and other seasonal crops, while the main activities on the west side are cultivation of rain-fed rice paddies, palm-sugar production, small-scale fisheries, and seasonal crops [23] (Figure 1).

Data for this study were collected from dry mixed deciduous forest, dry dipterocarp forest, dry forest and dry hill or evergreen forest. Twenty five sample plots, each with an area of 400 m² (20 m × 20 m) were laid out in each forest type. The height (Ht) and diameter at breast height (DBH) of all trees (DBH ≥ 5 cm, Ht ≥ 1.3 m) were measured in each plot. “A Checklist of the Trees, Shrubs, Herbs and Climbers of Myanmar” [24] was used to identify plant species by botanical and local names with assistance of taxonomic experts.

The data were analyzed for species composition, richness, diversity, species Importance Value Index (IVI). The species richness were estimated using the Jackknife estimator [25] . To provide quantitative estimates of plant diversity, Simpson’s index [26] and Shannon’s index [27] were used. Evenness indices, which are a structural composition index reflecting the dominance of species were also calculated. The Sorenson’s quantitative index which accounts for the relative abundance of shared species was used to assess the degree of floristic similarity within and between the forest stands [28] . For assessing the ecological importance or significance of a species, the IVI were used [29] . To assess taxonomic and structural composition between the forest types, the IVI value was calculated at the level of families. Family important values was computed as the average of the relative basal area, density and frequency.

Species richness is the basic component of diversity of any community [30] . The Jackknife estimator for species richness of trees (DBH ≥ 5 cm) showed that dry mixed deciduous forest (DMDF) occupied the greatest number of species (74 species) followed by dry dipterocarp forest (DDF) (53 species), dry forest (DF) (45 species) and dry hill/evergreen forest (DHEF) (40 species) (Table 1). Our findings of tree species are in the range of that found in mature tropical forest, 56 - 282 species ha⁻¹ (DBH ≥ 10 cm) [31] . Our findings are also in line with other study [32] in five tropical dry forests on Oceanic islands. Their study found 242 species ha⁻¹ (DBH ≥ 2.5 cm) in New Caledonia, 112 species ha⁻¹ in Fiji, 39 species ha⁻¹ in the Mariannas, 24 species ha⁻¹ in the Marquesas and 96 species ha⁻¹ in Hawaii. But our findings are lower than that found in dry tropical forests of India, where 190 species ha⁻¹ (DBH ≥ 5 cm) were recorded [33] and 105 species ha⁻¹ (DBH ≥ 10 cm) were found in the tropical seasonal forest in Khao Yao National Park in Thailand [34] . The difference may be due to the forest locations and tree size (i.e. DBH) recorded during forest inventory. We enumerated trees with only having above 5 cm DBH. The difference in terrain, gradient and slope direction causes differences in the soils, water and microclimate which causes differences in species adaptability [19] .

Tree diversity inclines provide important information about rarity and commonness of species in a community [19] . Shannon’s diversity index (H’) places more weight on the rare species while Simpson’s diversity index (1-D) gives more weight to those species which occur more frequently [28] [35] . In Popa Mountain Park, DMDF occupied the highest value of Shannon-Wiener index and Simpson diversity index, 3.61 and 0.96, respectively. Both diversity indexes indicated that the species diversity of DMDF is the highest among all forests in PMP. The larger the value of H’, the greater the species diversity and vice versa [36] . The Shannon-Weiner diversity Index (H’) value stood at 3.61 in DMDF, 2.96 in DDF, 1.45 in DF and 2.41 in DHEF, indicating that among forest types, DMDF was the most complex in species diversity whereas DF is the simplest community in term of species composition. This implies that elevation and aspect favors diversity and may be partly responsible for the diversity index obtained at Popa Mountain Park.

Our study findings are in line with the findings of Bhat [37] in the six fresh water swamp forests of Karnataka in India, with H’ of 2.53, 3.69, 2.46, 4.04, 3.25 and 4.90. Similarly, in the northern forest-savana ecotone of Ghana [38] obtained H’ values of 3.02, 0.04 and 0.39 at near-forest ecotone, near-savana ecotone and mid-eco- tone. However, when compared with the findings of Kumar [39] in three sites of tropical dry deciduous forest in western India, where H’ values of 0.67 - 0.79 were recorded, the H’ value of dry deciduous forest (2.96) in PMP is significantly higher. This indicates that diversity and richness in terms of species, and their distribution, is largely dictated by climate and ecological conditions, a view supported by Bello [12] .

^*DMDF = dry mixed deciduous forest, DDF = dry dipterocarp forest, DF = dry forest, DHEF = dry hill/evergreen forest.

Species evenness (E) is a measure of equitability of spread. Values obtained were 0.83 in DMDF, 0.74 in DDF, 0.38 in DF and 0.65 in DHEF. The species in DMDF were more abundant, and the percentage of evenness j (%) was close to 1.0. Therefore, Shannon’s evenness (j) shows that DMDF have highest species diversity and DF have the lowest species diversity in PMP. The slope direction influences tree species diversity at different altitudes [40] .

To assess the species composition and stand structure, the important value index (IVI) was used. No single species clearly dominated in dry mixed deciduous forest (Table 2). In dry mixed deciduous forest, the most abundant species were Shorea obtusa (8.58% important value index, IVI), Croton roxburghianus (7.34% IVI), Pittosporum napaulensis (4.83% IVI). The highest IVI value belonged to the species Shorea obtusa in dry dipterocarp forest whereas Tectona hamiltoniana in dry forest and Vitex canescens in Dry hill or evergreen forest.

Among all forest types, the most frequently occurring species is Techtona hamiltoniana (in dry forest), with 610 individuals recorded with a relative frequency of 17.12%. Due almost solely to its high relative frequency, Techtona hamiltoniana also has the highest species importance value of 54.25. The next most frequently occurrance species and most important species are Vitex canescens (in DHEF) with important values of 29.13 representing by 263 individual trees with relative frequencies of 12.72%. Result of our study suggest that Techtona hamiltoniana is an ecologically important species in Popa Mountain Park. In DMDF, 52.59% of the relative abundance included 10 common species (13% of total in DMDF) while 55.45% of RA (Relative Abandance) included five common species (9% of total species) in DDF. In DF only one species made up 70.03% of RA whereas two species made up 52.61% of RA in DHEF. These findings indicate that the number of species per unit area were high in all investigated forests.

At the family level, it was found that the taxonomic composition of the forests in PMP are different. DMDF was dominated by Dipterocarpaceae, Euphorbiaceae and Combretaceae. Dipterocarpaceae also dominant family in DDF, followed by Verbenaceae, Combreaceae. While, Verbenaceae (mainly Techtona hamiltoniana), Combretaceae and Rhamnaceae were the most common forest tree families in DF. In DHEF, Verbenaceae is the dominant family of trees mostly represented by the Vitec canescens tree and Myrsinaceae and Bixaceae (Table 3).

Dipterocarpaceae family was the family with the highest ecological importance given the IVI value in DMDF. The Dipterocarpaceae with only 3 species ranking 4^th in the species-rich family in DMDF, is the most important family based on the IVI values. This highlights that the most species-rich families are not necessarily the most important families based on the IVI values. For example, in DDF, the Dipterocarpaceae with only 3 species ranking 3^rd in the species-rich family was the most important family based on the results on IVI. This is due to the high number of individuals and high frequencies. In DF, the highest important value was found in Verbenaceae family, which has 4 species, ranking as the 2^nd species-rich family. In DHEF, Verbenaceae family, although with only one species, was the highest ecologically important family according to the IVI value. This is because each species was represented with many individuals.

Verbenaceae, Caesalpiniaceae and Moraceae were observed as species rich families for DMDF, representing 5 species (8% of the total species) in each family (Table 4). For DDF, the species rich families were Verbenaceae, which possessed 5 species with 9% of the total species, Anacardiaceae and Rubiaceae which possessed 4 species with 8% of the total species. The species rich families for DF were found to be Fabaceae and Mimosaceae which were represented by 5 species with 11% of the total species. In the case of DHEF, the species rich families were Fabaceae and Mimosaceae which processed 3 species with 8% of the total species.

An approximate indication of the homogeneity of a stand and of high diversity of tree species can be expressed by frequencies [35] . Species were assigned the frequency classes I, II, III, IV and V using their absolute frequencies 0% - 20%, 20% - 40%, 40% - 60%, 60% - 80% and 80% - 100%, respectively. If high values are found in frequency classes I/II and low values in frequency classes IV/V, it would indicate a high degree of floristic heterogeneity. If high values were found in frequency classes IV/V and low values in I/II, it indicates constant or similar tree species composition in the area. Frequencies depend on the size of the subplots. The larger the subplots, the higher the number of species that will be found in the higher frequency class. Comparison of frequency diagrams is therefore possible when based on the areas with the same subplot sizes. As shown in Figure 2,

SD = stand density, BA = basal area, Vol = volume, RD = relative density, RF = relative frequency, RBA = relative basal area.

NS = number of species, SD = stand density, BA = basal area, Vol = volume, RD = relative density, RF = relative frequency, RBA = relative basal area.

frequency values in I/II classes are higher than in IV/V classes indicating that all forests in PMP have a high degree of floristic heterogeneity and high diversity of species.

The similarities in species composition between the forests are presented in Table 4. Similarity indices range from 0.0 - 1.0, corresponding to a 0% - 100% similarity between any two plant communities. Coefficients of similarity of species composition (Sorensen’s index) between DMDF and DDF showed higher values. Their respective vegetation communities share 86% of the species between the DMDF and DDF. When the forest stands were paired and compared against each other, only DMDF and DDF showed a highly similar species composition. It was found that 46 species were common between the DMDF and DDF, 27 species between DMDF and DF, 30 species between DMDF and DHEF, 23 species between DDF and TDF, 20 species between DDF and DHEF and 11 species between DF and DHEF. Likewise, high similarity values of the family were found in DMDF and DDF.

It was found that dry evergreen forest has the highest mean diameter at breast height (DBH) 24.03 cm and dry dipterocarp forest has the lowest mean DBH, 11 cm (Table 5). Mean tree density (DBH ≥ 5 cm) of the various forest stands was DMDF, 1061 trees ha⁻¹, DDF, 1293 trees ha⁻¹, DF, 871 trees ha⁻¹ and DHEF, 804 trees ha⁻¹. The standing volumes were 214.69 m³∙ha⁻¹ in DMDF, 126.81 m³∙ha⁻¹ in DDF, 220.33 m³∙ha⁻¹ in DF and 610.72 m³∙ha⁻¹ in DHEF. The mean volume of DHEF was highly significantly different (p > 0.01) among the forests in PMP. However, the stand density was the lowest in DHEF while the standing volume strongly suggestion that the trees in DHEF are more mature than other forests. In term of species, Shorea obtusa made up the highest standing volume (29.48 m³∙ha⁻¹) in DMDF and (34.82 m³∙ha⁻¹) in DDF while Tectona hamiltoniana was the highest (160.97 m³∙ha⁻¹) in DF and Vites canescen (284.52 m³∙ha⁻¹) in DHEF.

Diameter frequency distributions provide a useful substitute for development trends of the stands [35] and help to evaluate potential forest sustainability [41] . For all forest types in PMP, it was found that increasing tree size classes results in drastically decreased species richness (SR), density (D) and diversity (H’) (Table 6). Shannon-Wiener diversity Index (H’) showed that the smaller DBH size classes have the higher diversity. The study found that lower size classes; 5 - 10 cm and 10 - 15 cm, contributed more than 50% of total tree density in the investigated forests. As well, the lowest size class, DBH 5 - 10 cm, possessed the highest species richness in all of the forests. In the 5 - 10 cm DBH size class, we found 55 out of 74 species in DMDF, 47 out of 53 species in DDF, 32 out of 45 species in DF and 20 out of 40 species in DH. The higher numbers of species were found in the lower size class in all forest types.

The findings in this study indicate that, where tree density is generally higher in small DBH classes compared to large DBH classes, this is a secondary forest characteristic. In all the forest stands, the greater numbers of trees were observed in the lowest diameter class (5 - 10 cm). This indicates that the density of smaller trees in a stand is sufficient to replace the current population of larger tree. The diameter distribution of the trees followed the inverse J-shape pattern (Figure 3(a)). This pattern indicated that stands are developing and regeneration is occurring in the forest indicating a high potential for species substitution when mature trees in the dominant species die.

The diameter classes 15.1 - 20 cm, 10.1 - 15 cm, 15.1 - 20 cm and 25.1 - 30 cm occupied the largest basal

^*DMDF = dry mixed deciduous forest, DDF= dry dipterocarp forest, DF= dry forest, DHEF = dry hill/evergreen forest, DBH = diameter at breast height, Ht = height, BA = basal area, Vol = volume.

area per ha in DMDF, DDF, DF and DHEF, respectively (Figure 3(b)). No stems in diameter classes 55.1 cm were found in DDF and no 70.1 cm diameter stems were found in DMDF and DF. Large diameter size classes of tree were rarely found in most of the forests in PMP. This might be due to over-cutting of trees (for firewood) from the surrounding area. In DHEF, large tree with stem diameters up to the 95.1 - 100 cm DBH class. It is therefore probable that less human disturbance in DHEF occurred above 1000 m, being less accessible.

^*SR = species richness, D = density, H’ = Shannon-Wiener diversity index.

Figure 4 shows species-area-curves. It can be seen DMDF has a greater richness of trees of stem diameter ≥ 5 cm than the other forest types while DDF has intermediate richness levels. However, DF and DHEF has similar species richness but lower than DMDF and DDF. In this study, the species-area curves were drawn based on trees with height ≥ 1.3 m and DBH ≥ 5 cm, and the total survey areas were one hectare in each forest. Lamprecht [35] suggested that a stock with DBH > 10 cm is generally adequate to draw species-area curves. He also suggested that the species-area curve is the best criterion for determining a minimal plot size needed to survey a community adequately. The pattern of the DMDF curve increases significantly up to the point of 0.40 ha, after which the increase is much more gradual with the increment of the number of species remaining below 10%. Similarly, in DDF new species were found in areas up 0.48 ha after which the increment remained below 10% from 0.48 ha up. Likewise, the pattern of DF curve and DHEF curve went up gradually and then became constant from 0.64 ha and 0.48, respectively. The minimum representative area would be reached if the number of species increases by less than 10% when the expansion of sampling area is 10% [42] . According to the species-area-curve, one ha sample size represented the minimum area for each forest type since there were only minimal numbers of new species discovered.