Study area Satchari Forest covers an area of about 40 km2and is comprised of Satchari Reserved Forest (SRF), Satchari National Park (SNP) and surrounding areas(SA) characterised by tea (Camellia sinensis) estates and fallow lands. The forest was declared as a reserve on 1914, and on 2005 part of that reserve (243 ha) was declared as a national park (Mukul, 2007). The area is unique in biodiversity, and is one of the last habitats of critically endangered Hoolock Gibbons (Hoolockhoolock) in the country (Muzaffar et al., 2007; Arefin et al., 2011). Acidic soils (sandy loam to silt clay) dominate the area, with the relief is characterised by a gently undulating to hilly topography. The altitude is generally low with hilltops reaching 104 m a.s.l. and increasing towards India with elevation reaching 144 m a.s.l.beyond the border. The area enjoys a sub-tropical monsoon climate with three distinct seasons, viz., summer (March to mid June), monsoon(mid June to mid October) and the winter (mid October to Febru-ary). The mean annual precipitation however varies considerably year to year, and is about 4160 mm. Most rainfall occurs during the monsoon, whereas July receives the highest amount of rainfall (about 1250 mm) with relative humidity ranges between 74% and90%. The average minimum and maximum temperatures are 12°C and 32°C respectively. The prominent vegetation type in the area is mixed tropical evergreen forest (Champion, 1936). Relatively undisturbed forest is almost entirely restricted in the national park (i.e. SNP). Some areas outside the reserved forest have converted extensively into agricultural fields, fruit orchards, human settlements, sungrass (Saccharum spontaneum) fields and plantations of fast growing timber species. Most of the plantations, however, are dominated byexotic species such as, Acaciaspp., Eucalyptus camaldulensis, Tecton-a grandis, Albizia falcataria, with one of the pioneer Oil palm (Elaeisguinensis) plantations of the country established on mid-seventies in the area (Choudhury et al., 2004). Sampling design A systematic sampling approach was followed to avoid any bias in data collection. Plots were established in the SRF, the SNP, and the SA. A species-area curve was followed to determine the sample size and sample adequacy. All tree species were recorded using acircular plot with a 10 m radius (314.16 m2). Within each circular plot, three nested square (2 m2 m) plots were laid randomly to sample the herbs, grasses, shrubs, and climbers. Altogether,139 circular plots (N= 139) were established in the area, of which88 were located in the SRF, 14 in the SNP, and 37 in the SA. Since the SA has a near-to-homogeneous vegetation structure dominated by tea estates and fallow grass lands, the distance between plots in the SA was 800 m, whilst in the SRF and in the NP, plots were laid 400 m apart. Disturbances and protection regime Disturbance data, both at the plot and landscape level, were collected following the methodological framework proposed by Buhk et al. (2007). For each plot, all identifiable disturbance events were recorded based on the following temporal and spatial descriptors: frequency (1/year, 2/year, >2/year), duration (year), distribution (homogeneous or heterogeneous), form (puncti, linear, laminar), and size (small, large). However, in the analyses, the following features were incorporated: (i)number of disturbances; (ii) number of different disturbance selectivities; (iii) number of different disturbances distribution; (iv) number of different disturbance sizes; and (v) number of different disturbance frequencies. The protection regime was represented by the spatial location of the plots in the Satchari Forest, i.e., the SNP, the SRF, and the SA, with a decreasing order of protection.The SNP, due to its high conservation value and its designation as a protected area, has been enjoying the highest protection focus (Mukul et al., 2012), whereas the SRF is under the jurisdiction of the Forest Department and subject to substantial resource collection by locals, both legally and illegally. The SA, on the other hand,has no conservation restriction imposed on it. Statistical analysis Statistical analyses were performed using the ‘R’ statistical package (version 2.10.0;R Development Core Team, 2009). We used the ‘‘vegan’’ version 1.17-0 (Oksanen et al., 2010), and ‘‘gbm’’ version 1.6-3 packages as provided in the appendix of Elithet al. (2008). Boosted Regression Trees (BRTs) were constructed to assess the relationship between explanatory variables and the number and the percentage of exotic species. The environmental and edaphic variables incorporated in the analysis were: (i) number of native species; (ii) bulk density; (iii) soil moisture; (iv) pH,(v) organic carbon; (vi) P; (vii) N; (viii) Ca, (ix) Mg; (x) K, (xi) elevation; (xii) aspect west; (xiii) aspect north; and (xiv) slope. All environmental attributes (14), protection regime (1), and disturbance variables (5) were included as predictors. Particular focus was given to: (i) the relationship between exotic and native species; (ii) the effect of the environment (topographic and edaphic)on exotic species richness; and (iii) the effect of forest protection status and disturbances on exotic species. BRTs were calculated by function gbm.step (gbm 1.6–3) as provided in Elith et al.(2008). As implemented, cross-validation was used to determine the optimal model settings, while all data were used to fit the final model. The measure of the relative influence of predictor variables was based on the number of times a variable was selected for split-ting, weighted by the improvement of model results (see Elithet al., 2008for more detail). Final model settings were a bag fraction of 0.75, a Gaussian error distribution, a learning rate of 0.008, and a tree complexity of 2. The optimal number of regression tress (‘‘trees’’ in this context was not a biotic but a statistical tree, i.e., a decision or regression tree) in the BRT model was found to be 1100 and 1250 for the analysis of the number and the percentage of exotic species, respectively.