Study Area: The study was conducted in forest ecosystems of Bangladesh, where knowledge about diversity patterns is scarce. We focused on the Satchari forest, which was subdivided into two parts of different protection status: Satchari National Park (NP) and Satchari Reserve Forest (RF). The whole study area has been under nominal protection since the 1974 Wild Life Preservation Act. However, human disturbances were almost entirely excluded from Satchari National Park only after its establishment in 2005, while the Reserve Forest is still subjected to different illegal usage (especially firewood collection. Tea plantations, fallow land and settlements cover almost all of the adjoining areas of the forest. The study area is located in north-eastern Bangladesh and stretches between approximately N24°5” and 24°9” and E91°24” and 92°29”. It covers an area of about 20 km2. The natural vegetation of the study area consists of a mixed tropical evergreen forest. A well-developed tree stratum with evergreen top canopy trees characterizes the less disturbed sites of the area. The study area is characterized by sandy loam to silty clay acidic soils and gently undulating to hilly topography (10 to 105 m a.s.l.). The climate is sub-tropical and monsoonic. The rainy season usually extends from June to September; however, the rainfall is irregularly distributed. Mean annual rainfall is 4,162 mm, peaking in the monsoon season. The relative humidity ranges from 74% during December to 90% during July and August. Research on Satchari forests is sparse and concentrates on management and conservation of rural livelihoods and wildlife.
Sampling Design: The study is based on an assessment of tree species characteristics and composition. A representative, unbiased and systematic sampling procedure was applied for the Satchari National Park (15 plots) and Satchari Reserve Forest (96 plots). An equidistant sampling grid (distance of 400 m) was implemented in those parts of the protected area that are actually forested. For all tree individuals in the plots, we measured diameter at breast height (DBH) at 1.3 m height from the ground, which is recognized as a standard and most widely used reference diameter for expressing tree or stand characteristics associated to wood biomass. In our investigation, a “tree” is regarded as a woody plant with a minimum DBH of 10 cm. DBH was recorded within each circular plot of 10 m radius. Basal area of each tree was calculated (basal area = pi × (DBH/2)2) and summed for each plot. For some species that could not be identified directly in the field, pressed samples were taken to the National Herbarium at National Botanical Garden, Dhaka University and Jahangirnagar University, for identification and confirmation. The nomenclature of families and genera follows the Encyclopedia of Flora and Fauna of Bangladesh. The four botanical collections, which could not be identified to species or genus level, due to the lack of diagnostic features, were not included in our floristic analyses.
Data Analysis: Alpha (average plot-level species richness) and beta diversity (relation of gamma and alpha diversity) as well as gamma diversity (overall species richness) [9] were calculated for the National Park and the Reserve Forest. Species composition between the two differently protected areas was compared using function ANOSIM in R-package vegan version 1.17-0. For this and the four distance-decay analysis plots from the Reserve Forest were removed, as they did not contain trees. Floristic similarities in tree species composition were calculated for all possible pairs of plots using the Bray-Curtis Index that allows for consideration of quantitative information for single species. Hence, species frequency (number of tree individuals of a specific species) within the plots is incorporated in the calculated values for this index. To compare spatial species turnover between the two differently protected areas, the distance-decay relation was assessed for two equal sized sub samples from Satchari National Park and Satchari Reserve Forest. We choose both a similar plot size and area under study, i.e., National Park and Reserve Forest as the distance-decay analysis are suspected to be sensitive to spatial grain and extent. The distance-decay relation was calculated by a linear regression (leased squared with R2 as the goodness-of-fit measure) on a pair wise comparison of geographical distance and similarity in species composition. Distance-decay relations are frequently calculated using log-transformed similarity values, log transformed distance values or both. Most studies dealing with plants use an untransformed relationship. Untransformed relations have the advantage of clear interpretation and comparison but bear the problem that: (a) the fitted line has an intercept with the x-axis implying negative similarity values after a certain distance; and (b) that the fitted line has an intercept with the y-axis implying negative distances for specific similarity values.
In their seminal paper Nekola and White (1999) tried all combinations (log- and untransformed) and selected the combination with best values for model residuals. Here we follow this procedure choosing our model based on: (a) the goodness of fit indicator r-squared; and (b) the visual model diagnostics. A significance test for a linear relation between distance matrices has to account for the problem of pseudoreplication. One possible solution is the use of permutation tests. Here, a Mantel-test with 10,000 permutations was performed (conservative two-tailed p-value: null hypothesis: r = 0) that is offered in the R-function mantel in package ecodist version 1.1.4 [48]. To test a potential influence of elevation, the Euclidean distance in elevation between the plots was related to similarity in species composition. In case of significance (again mantel test), the residuals of that relation were related to geographic distance. Thus a significant distance decay relation is tested after accounting for an influence of elevation. The statistical software used was R 2.10.0.. Stand characteristics such as tree density, number of tree species, number of tree individuals and basal area (summed basal area of each tree in a plot) were compared between the areas of different protection status using t-Test. Two plots were excluded from the calculations of basal area (t-Tests) as they contained trees with extensive buttress roots, which rendered the assessment of comparable DBH impossible.