Variables were ranked by importance using IncNodePurity scores (Fig. 12b), highlighting VPD as a key factor affecting bamboo presence and coverage. The examination of VPD values revealed consistently low levels in bamboo-dense areas, suggesting that bamboo thrives in high-humidity conditions favourable to its growth and survival. Recognizing the climatic drivers of bamboo distribution enables researchers and policymakers to monitor it as an early indicator of ecological stress or potential shifts in bamboo cover. This insight serves as a strategic tool for climate-resilient planning, helping forest managers to prioritize conservation efforts and identify suitable sites for bamboo-based livelihoods, particularly in climate-vulnerable regions.
Bamboo is difficult to identify through satellite images because its spectral characteristics are similar to those of other vegetation classes when a single date image is considered. So, a combination of different seasons’ imagery was used to distinguish bamboo from other vegetation. The combination of March and November satellite imagery proved to be more effective for mapping bamboo than using either single month imagery. This finding aligns with previous research by Li et al. who also reported higher classification accuracy when using multi-temporal imagery rather than single-month data. Other studies have also indicated that combination of multiple imagery sources enhances classification accuracy. Li et al. also highlighted April-May and December-February as optimal periods for distinguishing bamboo forests from other forest types, emphasizing that winter is the best time for bamboo mapping. Similarly, Zhang et al. found the highest mapping accuracy during the winter season (September to November), followed by the March-May period. Cloud contamination is also an inevitable issue for bamboo classification based on optical features. For the study area, cloud-free imagery was more readily available in March and November, making these months practical for analysis. Ecologically, these months also represent key phenological stages for bamboo. Due to the rapid growth of bamboo from shoots to fully developed plant in spring, March marks the onset of the growing season, and the selective logging of mature bamboo during winter, leads to changes in canopy and structure. The study area is predominantly covered by evergreen and semi-evergreen forests and is influenced by widespread shifting cultivation, creating complex vegetation patterns. Notably, the clearing of shifting cultivation patches typically begins during the winter months making this period crucial for distinguishing bamboo. In this context, March and November imagery provides an optimal spectral window to distinguish bamboo from surrounding vegetation and active shifting cultivation patch, enhancing classification accuracy.
This study found that the RF classifier outperformed other models in terms of performance accuracy metrics for mapping bamboo distribution. Previous research similarly highlights the effectiveness of RF in mapping bamboo spatial distribution. The overall classification accuracy achieved was 87.54%, aligning closely with findings by Xiang et al. who reported an accuracy of 87.96% using Sentinel-2 data with the RF classifier. SVM and ANN can be prone to overfitting, especially when dealing with high-dimensional data whereas RF due to its ensemble nature and ability to handle complex relationships in the data is beneficial. Sentinel-2 data proves beneficial for mapping the bamboo due to its high spatial resolution. Furthermore, the incorporation of textural variables along with different spectral variables also enhanced the classification accuracy. Bamboo’s unique structure with upright cylindrical stems and irregular upper leaf arrangement creates a diffused reflective surface, resulting in distinct spectral and textural patterns that enhance its separability from other vegetation. Among the GLCM features, the mean was found to be more effective in mapping bamboo. This finding aligns with Ghosh and Joshi who similarly identified the GLCM mean as the key contributor to improved classification accuracy.
The SWIR bands emerged as important variables for generating the LULC map and were effective in distinguishing bamboo from other land cover classes. Similar findings were reported by Chen et al., Feng et al. and Xiang et al. SWIR reflectance is primarily influenced by leaf water content and helps differentiate vegetation from soil. Since water in leaves absorbs energy in the SWIR region, this band is valuable for vegetation analysis. For the bamboo class NDVIre2 for the month of November was the top predictor for effectively mapping. The red edge-based spectral indices were found to be significant for identifying bamboo forests. Red-edge bands are more sensitive to the chlorophyll content and are helpful in identifying the canopy changes of bamboo forests. NDVIre1 and NDRE2 for the month of November were among the top five important variables for bamboo classification. Higher NDVIre2 values are generally associated with healthier and more abundant vegetation, while lower values may indicate stress, damage, or reduced vegetation cover. Overall, the red-edge bands, the red-edge index and SWIR bands are of high importance for bamboo forest information extraction.
The disturbance map indicated that bamboo is predominantly distributed within high-disturbance areas. Approximately 78.9% of the district’s bamboo growing areas experienced disturbance. Shifting cultivation, practiced by local tribes, was identified as a key disturbance factor that appears to facilitate bamboo growth in these areas. Field observations also revealed that shifting cultivators tend to prefer areas with high bamboo density for their cultivation practices over typical forest patches. This preference is attributed to several factors including the ease of cutting bamboo and the high ash content produced after burning bamboo forests, which serves as an effective fertilizer for the crops. In this study, NBR trends were analyzsed for studying the disturbances. Schneibel et al. and Cohen et al. identifed NBR as the most sensitive index to disturbances in moist forests. In Tripura state of India, Bhat et al. also observed that NBR was very effective in studying the disturbance and regrowth in shifting cultivation landscape.
Disturbances play a pivotal role in influencing bamboo distribution and regeneration by creating conditions favourable for its proliferation. In northeastern Thailand, human disturbances such as logging promote bamboo dominance, which suppresses tree regeneration. Similarly, in subtropical China, bamboo expansion is widespread in areas affected by logging or fire, significantly altering the dominance from trees to bamboos, showcasing its resilience and adaptability. M. baccifera is particularly adapted to thrive in disturbed environments, especially in areas affected by shifting cultivation. Troup observed vast expanses of this species in the Chittagong and Arakan hills, where shifting cultivation and frequent fires had destroyed tree growth. These bamboo brakes are considered anthropogenic climax communities resulting from prolonged disturbances. Its rhizomes exhibit exceptional vitality, resilience, and prolificity surviving even after the burning of culms, leaves, and branches. Unlike most species whose rhizomes are destroyed by repeated fires, M. baccifera thrives due to its robust underground rhizome system, regenerating rapidly into pure bamboo forests that act as fire climax communities and secondary vegetation. This ability allows M. baccifera to dominate abandoned fields and expand through practices like shifting cultivation, underscoring its role as a pioneer species in ecological succession and forest regeneration.
To interpret the overall disturbance and recovery of vegetation in the study area, an RGB composite (Fig. 13) was derived using the LandTrendr-based parameters: rate of disturbance, rate of growth, and duration of growth. Purplish areas indicate strong cultivation disturbance with low regrowth rates and long recovery periods, suggesting abandoned land unsuitable for further cultivation. Bluish areas show prolonged regeneration with limited recovery rates. Greenish-yellowish areas indicate high disturbance with fast recovery, likely due to short cultivation cycles. These areas also include sites where shifting cultivation has been replaced by monoculture plantations. Over time, fallow shifting cultivation patches are predominantly colonized by bamboo, indicating favourable conditions for the establishment of gap-dependent, light-demanding pioneer species. The fallow patches provide an environment for bamboo plant and establish itself more effectively than other plant species. Overall, the composite provides insights into land-use changes and vegetation dynamics within the study area.
The spatial dynamics of bamboo regrowth across disturbance gradients reveal the remarkable resilience and recovery potential of bamboo-dominated landscapes. A stacked bar chart (Fig. 14) was generated to visualize how regrowth levels vary across different disturbance categories. In undisturbed areas, approximately 60% of the landscape exhibited no detectable regrowth, indicative of stable, mature vegetation with minimal structural change. In zones with low disturbance, a relatively even distribution among no, low, and moderate growth was observed, suggesting stable bamboo stands or mixed forests with limited anthropogenic impact. In moderately disturbed areas, around 45% of the landscape showed moderate growth, accompanied by a notable increase in high growth. This suggests that partial canopy openings in these regions may facilitate bamboo sprouting and lateral spread. In areas subjected to high disturbance, nearly 50% exhibited high growth, reflecting bamboo’s role as a pioneer species capable of rapid clonal propagation and recolonization. These zones likely represent sites undergoing natural succession or those influenced by post-disturbance management efforts. However, about 20% of the highly disturbed areas remained in the no or low growth category, potentially due to factors such as overharvesting of culms or rhizomes, soil erosion or land-use conversions like permanent plantations.
A decline in shifting cultivation areas was observed during field visit, which is potentially attributed to government plantation initiatives that have converted shifting cultivation plots into intensified, permanent plantations, thereby altering the landscape. The NITI Aayog in its report ”Shifting Cultivation: Towards a Transformational Approach”, also has mentioned the prioritization of government policies for conversion of shifting cultivation land to settled agriculture practice. Field observations revealed that traditional jhum cultivators are increasingly converting their fallow lands into monoculture plantations, particularly of areca nut, and also other agroforestry systems. Incentives for commercial crops have encouraged a shift from conventional shifting cultivation toward long-term cultivation of high-value cash crops such as pineapple, rubber, and areca nut, as well as intercropping systems of areca nut with rice and maize in their jhumlands. Similar transitions were reported by Hazarika et al. and Bhat et al. highlighting a broader trend of land use change. However, this transition has also triggered a contrasting trend, where farmers clear new patches of forests for shifting cultivation, mainly for rice cultivation while converting older fallow lands into monoculture plantations. This dual land-use strategy raises concerns about a resurgence in overall shifting cultivation activity and its long-term implications for bamboo dynamics and forest ecosystems. These will have negative consequences for the socioeconomic well-being of communities that rely heavily on bamboo resources for their livelihoods.
Bamboo thrives under warm and humid climate and its growth is greatly influenced by the prevailing climatic conditions. To evaluate the impact of climatic variables on bamboo distribution, the RF relative importance was employed. Previous studies have demonstrated RF’s effectiveness in variable ranking, supporting its utility in this analysis. VPD emerged as the most influential factor affecting bamboo distribution in the study area. Several studies have emphasized the critical roles of precipitation and temperature in influencing the growth of bamboo forests, especially for moso bamboo. However, in the study area, the bamboo species is primarily characterized by the prevalence of M. baccifera and B. tulda. Notably, previous research has not accounted for VPD, which could provide important additional insights. VPD influences the movement of water vapour from plants to the atmosphere and is an important driver of plant functioning. The mean VPD value was found to be 0.80 kPa in the study area. Similarly, in Lei bamboo dominated forests in China, the mean VPD for the majority of years was observed to be 0.85 kPa. Low VPD levels, typically found in humid, forested fallow lands, promote moderate evapotranspiration, creating favourable conditions for bamboo growth without inducing water stress. Such optimal VPD conditions support the development of resilient shoots and rhizomes, which are vital for bamboo proliferation and expansion. As a result, VPD serves as a reliable indicator to identify areas where bamboo is likely to regenerate and establish dominance in post-shifting cultivation landscapes.
Precipitation emerged as the second most influential variable after VPD. In China, the extensive proliferation of moso bamboo has been attributed to precipitation. The critical phase for bamboo shoot germination necessitates ample precipitation. Moreover, autumn precipitation significantly influences the subsequent spring’s bamboo shoot yield. Once these shoots appear, their rapid growth is driven by vigorous meristematic tissues between nodes, demanding substantial rainfall. This period marks peak water requirements for bamboo. The mean annual precipitation in the study area was found to be 1596.05 mm. Zhang et al. also estimated that over a span of five years, the mean precipitation was found to be 1523.2 mm in Lei bamboo dominated forests of China. Temperature following precipitation was identified as a key factor influencing the potential distribution of bamboo forests. It significantly influences bamboo shoot germination and emergence, with the optimal average daily temperature for bamboo growth ranging between 15 °C and 25°C. Precipitation and temperature were identified as important climatic factors limiting bamboo distribution by Shi et al. and Zhang et al.. It is essential to identify the most influential environmental factors affecting bamboo growth. Such insights are essential for monitoring bamboo distribution and ensuring its sustainable management and climate adaptation strategies.
