Introduction
Forests play a vital role in watersheds, significantly impacting the water cycle within these ecosystems. They act as regulators of water management due to the presence of tree vegetation, which directly influences the overall health of the watershed. However, forests, as a valuable natural resource, are often subjected to various human uses that can lead to detrimental effects, such as a reduction in forest area and deforestation (Kusumandari et al. 2024). One significant impact of converting forest land to other land use, which provides fewer ecosystem services, is an increase in soil erosion and the expansion of critically degraded land (Endayani et al. 2019).
In recent years, the condition of watersheds and their ecosystems has emerged as a national issue, largely due to hydrometeorological disasters that indicate a decline in watershed quality. Permatasari et al. (2017) state that watershed damage is characterised by the extent of critical land, which impairs the optimal functioning of the watershed. Critical land is defined as land that has sustained damage resulting in a reduction or loss of its ecological functions (Budiastuti et al. 2020).
Based on the national watershed classification map, there are 42,210 watersheds in Indonesia, with 37,721 watersheds being actively maintained and 4,489 watersheds undergoing restoration. In terms of area, the restored watersheds are larger than the maintained watersheds. One of the 18 priority watersheds in Indonesia is the Kapuas watershed, which contains several sub-watersheds, including the Sepauk sub-watershed.
The Sepauk sub-watershed constitutes the upstream area of the Kapuas watershed, covering an area of 122,025 hectares and is categorised as having a critical level of 53.54%. It features various types of land cover (Kapuas Watershed Management Center 2022) and plays a vital role as a recharge area for the Kapuas watershed. This watershed is recognised as one of the 108 priority recovery watersheds in Indonesia (Indonesian Ministry of Environment and Forestry 2020).
Within the Sepauk sub-watershed, one of the areas with an important ecological role as part of the upstream recharge system of the Kapuas watershed is the Peninsung Catchment, which covers approximately 8,111.36 hectares. Most of the land within this catchment is categorised as moderately critical land, covering approximately 5,771.57 ha (71.15%). Due to its strategic hydrological function and increasing degradation pressure, this catchment represents an important area for watershed rehabilitation and sustainable land management efforts. The Peninsung Catchment functions as an integrated ecosystem, composed of interacting physical and biotic components. These interactions are reflected in the traditional land-use practices of the Dayak community, particularly the tembawang system, which represents indigenous ecological knowledge and long-term landscape management practices (Crevello 2004, Murhaini and Achmadi 2021).
The term "tembawang" refers to a hereditary land-use system characterised by forest gardens. These forest gardens represent a form of complex agroforestry, where tree-dominated systems mimic the structure and function of natural forests (de Foresta et al. 2000, Kaya et al. 2002, Melvani et al. 2020, Melvani et al. 2022) or represent a reconstructed natural forest (Wiersum 2004, Astiani and Ripin 2016). The presence of trees in these areas plays a crucial role in the environment, namely absorbing carbon dioxide and conserving tree species (Marjokorpi and Ruokolainen 2003), reducing erosion rates (Rafdinal et al. 2021a) and also playing a role in the social economy aspects for the local community (Salafsky 1994).
The forest garden can be found in many regions of Indonesia, including in West Kalimantan, where the 'tembawang' system (Marjokorpi and Ruokolainen 2003) integrates tengkawang trees (Shorea spp.) with both fruit and timber trees (de Foresta et al. 2000). This land-use type is an ancient tradition of land management. Previous studies indicate that forest gardens in Kalimantan are closely linked to their historical interaction with swidden (Silvianingsih et al. 2020).
The Sintang Regent Regulation No. 122 of 2021 defines tembawang, or other similar designations, as areas that were once the sites of traditional “betang house” (traditional Dayak houses) or former field huts, which have since evolved into areas dominated by fruit trees. These areas are managed and respected, based on the local community's wisdom and have been maintained by a group of people for generations (Anonymous 2021). Furthermore, Pradityo et al. (2016) state that tembawangs are former settlement areas that continue to be utilised and managed by local communities for specific reasons following its abandonment. These areas have become overgrown with various vegetation, forming forest gardens that are only partially utilised and regulated by customary law.
The role of tembawang in an ecosystem, particularly in watershed ecosystems, is significant due to its vegetation, which structurally and compositionally resembles natural forests. This resemblance contributes to its ecological function, especially in soil and water conservation. Additionally, tembawang has economic importance, as it provides resources for daily life, including food, medicinal plants and building materials. Furthermore, tembawang helps reduce erosion rates through litter dynamics and has the potential for carbon storage (Rafdinal et al. 2021a).
Considering the increasing extent of critical land in the Sepauk Sub-watershed, the ecological characteristics of tembawang suggest its potential contribution to watershed rehabilitation and sustainable land management. Marjokorpi and Otsamo (2006) emphasised that rehabilitation and conservation efforts can be strengthened by integrating local ecological knowledge and traditional land-use systems into restoration strategies.
However, quantitative ecological assessments of tembawang vegetation in relation to watershed rehabilitation remain limited, particularly in the Sepauk Sub-watershed of West Kalimantan. Previous studies on tembawang have predominantly emphasised its socio-cultural functions, structure and composition, traditional management practices and economic benefits for local communities (Salafsky 1994, Ripin et al. 2014, Pradityo et al. 2016, Roslinda et al. 2017, Silvianingsih et al. 2020, Roslinda et al. 2022), while comprehensive ecological assessments of vegetation structure across growth stages remain limited, particularly in the Peninsung Catchment. The novelty of this study lies in exploring the potency of a multistage vegetation characteristic of tembawang in the Peninsung Catchment, for rehabilitating the critical land which is ecologically suitable in achieving sustainable watershed management.
Materials and Method
Study area
The Sepauk Sub-watershed is one of the upstream sub-watersheds of the Kapuas watershed, covering an area of 122,025 hectares. It is administratively part of the Sepauk District in the Sintang Regency and extends into a small portion of the Sekadau Regency, located in the West Kalimantan Province. The Sepauk subdistrict covers an area of 1,825.70 km², which constitutes 8.44% of the total area of the Sintang Regency, making it the fifth-largest area within the regency. The site is located approximately 300 km from Pontianak, the capital city of West Kalimantan (BPS-Central Bureau of Statistic Sintang Regency 2023).
This research was conducted in February 2024 in the Peninsung Catchment, Sepauk Sub-watershed, West Kalimantan. The Peninsung Catchment covers an area of 8,111.36 ha and includes 10 villages. The critical land in this catchment is predominantly classified as being in the moderately critical level, covering 5,771.57ha (71.15%). The location was chosen purposively, based on two criteria: (1) the representation of critical land in the catchment and (2) the presence of tembawang.
The research locations are presented in Fig. 1.
Tembawang, also known as agroforest tembawang, is a land-use system practised by the local Dayak people. In the study area, there were four types of tembawang that originated from ex-settlements known as field huts, which were built following their shifting cultivation pattern. Based on the tenurial classification proposed by Soeharto (2010), the study sites consisted of Communal Tembawang, Young Heritage Tembawang and Old Heritage Tembawang. Based on field information, these tembawang types were estimated to be > 80 years, 50–60 years and 60–70 years old, respectively. A visual representation of tembawang T1, T2, T3 and T4 in the study sites can be seen in Fig. 2.
Data collecting
The vegetation analysis was performed across four tembawang sites (T1, T2, T3, T4), which ranged in size from 0.5 hectares to 0.7 hectares. We used a minimum sampling intensity of 30% from the total area of 2.2 hectares to obtain detailed results. Employing a combination of line and nested plot sampling methods, four plots were established per site, resulting in 16 total plots. Data collection utilised nested quadrats measuring 20 × 20 m (trees), 10 × 10 m (poles), 5 × 5 m (saplings) and 2 × 2 m (seedlings). The assessment covered all growth stages. Species identification was carried out in situ based on local vernacular nomenclature provided by landowners.
Data Analysis
Vegetation data analysis was conducted using the PAST (Palaeontological Statistics) software version 4.04 (Hammer et al. 2001). The composition and similarity of vegetation communities across the study sites at each growth stage (seedling, sapling, pole and tree) were evaluated using Cluster Analysis. The clustering was performed using the Unweighted Pair Group Method with Arithmetic Mean (UPGMA) algorithm, based on the Bray-Curtis similarity index.
The reliability of the resulting dendrograms was verified through the Cophenetic Correlation Coefficient (CCC). Furthermore, the Importance Value Index (IVI) was calculated to determine the dominance of each species within the community. For the tree and pole stages, the IVI was determined by the sum of Relative Density (RD), Relative Frequency (RF) and Relative Dominance (RDo), resulting in a maximum value of 300%. For the seedling and sapling stages, the IVI was calculated as the sum of Relative Density and Relative Frequency only, with a maximum value of 200%. These IVI values were subsequently used as the primary data input for calculating biodiversity parameters, including the Shannon Diversity Index (H'), the Simpson Diversity Index (1-D) and the Evenness Index (E).
The classification of Diversity and Evenness indices follows the criteria adjusted from Odum (1971) and Magurran (2004). The Shannon index (H') was classified as low (H' < 1), moderate (1 ≤ H' ≤ 3) or high (H' > 3). The Simpson Diversity Index (1-D) was categorised as low ( 1-D < 0.33), moderate (0.33 ≤ 1-D ≤ 0.66) and high (1-D > 0.66). Finally, the Evenness Index (E) was classified into three levels: low (E < 0.4), moderate (0.4≤ E ≤ 0.6) and high (E > 0.6).
The potency of tembawang in critical land rehabilitation was analysed descriptively based on its vegetation structure, environmental performance and its role in reducing runoff and erosion in the hydrological process.
Results and discussion
The Important Value Index (IVI) and Dominance Species
The IVI of a species is a quantitative metric used to indicate the level of dominance of a species within a vegetation community (Soerianegara and Irawan 2008). A high IVI of species indicates a composite score for high relative species density and frequency at the seedling and sapling stages, as well as relative density, frequency and dominance at the pole and tree stages (Sarmiento et al. 2022,Santoso et al. 2023). Relative density is a quantitative measure used to describe the relative importance of a particular plant species within a community compared to other species. Relative Frequency describes the distribution of each vegetation. Additionally, relative dominance describes the degree of control or influence a species has over the physical space and resources in an ecosystem, measured by the basal area of trees and poles. Data collection in plots revealed 55 species in tembawang T1, T2, T3 and T4 at each growth stage and IVI of species presented in Suppl. material 1.
The Importance Value Index (IVI) analysis reveals a dynamic transition in species dominance across the four study sites from the seedling to the tree stage (Fig. 3). This vertical shifting reflects the interplay between natural forest succession and deliberate multi-species management within the tembawang agroforestry system.
Fig. 3 graphically illustrates the structural and phytosociological dynamics of the Tembawang agroforestry system across the four study sites (T1–T4), which were quantitatively evaluated using the Importance Value Index (IVI). The grouping configuration on the horizontal axis captures the successional movement across four distinct growth stages (seedling, sapling, pole and tree), while the vertical axis measures the relative ecological footprint and competitive pressure exerted by the primary dominant species within each site.
We found a total of 55 species (30 families), consisting of 14 species at the seedling stage, 34 at the sapling stage, eight at the pole stage and 24 at the tree stage in tembawang. The highest IVIs of each stage are presented in Table 1.
Table 1 shows that, at the seedling stage, the forest floor is heavily characterised by locally managed fruit species, rubber and a local herb. We found the domination of Lansium domesticum Corr. with an IVI of 117.36 at site T3, Hevea brasiliensis (Willd.ex A.Juss.) Mull.Arg. with an IVI of 100.31 at Site T4, Durio zibethinus with an IVI of 93.27 at Site T2 and Durio Zibethinus, Artocarpus integer and Syzygium polyanthum with an IVI of 42.22, respectively.
As the vegetation transitions into the sapling stage, species composition becomes more diverse and dispersed. In Site T1, the woody plant Macaranga gigantea takes temporary dominance (IVI = 40.47), while the fruit trees Artocarpus integer (Thunb.) Merr dominate Site T2 (IVI = 55.02) and Lansium domesticum Corr. dominates Site T3 (IVI = 31.11), while Hevea brasiliensis (Willd.ex A.Juss.) Mull.Arg. dominates Site T4 (IVI = 53.33). The relatively lower maximum IVI values during the sapling phase indicate a highly competitive environment where resource sharing prevents a single species from completely monopolising the lower canopy.
A drastic ecological shift occurs at the pole stage, marking a phase of absolute commodity dominance. Except for Site T1, which is dominated by Aporosa arborea (IVI = 78.44), all other locations (Sites T2, T3 and T4) are dominated by Hevea brasiliensis (Willd.ex A.Juss.) Mull.Arg., with IVIs of 140.98, 160.02 and 161.38, respectively. This 'Rubber Phase' represents an economic intensification period within the tembawang lifecycle. From an ecological standpoint, this intense single-species pressure in the mid-canopy explains the sharp decline in the diversity indices during this specific growth stage, as the canopy space is occupied by rubber poles.
At the final tree stage, the canopy structure matures into a complex, multi-layered forest system where native woody fruit trees regain their ecological dominance. In Sites T1, Artocarpus integer (Thunb.) Merr and Durio Zibethinus dominated with IVIs of 106.4 and 93.78, while at Site T2, Durio zibethinus continued to dominate with an IVI of 57.27. At site T3, the domination of indigenous fruit trees is still dominated by Lansium domesticum Corr. (IVI = 108.63) and Durio zibethinus (IVI = 62.79). Concurrently, another indigenous fruit tree Mangifera laurina Blume stabilises the overstorey canopy in Site T3 (IVI = 74.15). While Hevea brasiliensis (Willd.ex A.Juss.) Mull. Arg. remains dominant in Site T4 (IVI = 75.08), its importance value is significantly diluted compared to its dominance in the pole stage. This multi-layer recovery at the tree stage confirms that, as the tembawang system matures, it naturally limits monoculture-like dominance, allowing diverse, large-stature fruit trees to establish a stable, permanent top canopy.
The Important Value Index (IVI) is widely used to evaluate the relative dominance and ecological importance of species within vegetation communities. Higher IVI values indicate a greater contribution of a species to community structure and ecosystem functioning (Mueller-Dombois and Ellenberg 1974). At the research site, the tembawang was dominated by fruit-bearing plant species, a pattern traceable to the historical practice of shifting cultivation. This practice led to the planting of several types of vegetation, particularly fruit trees, which serve as indicators of land tenure (Peluso 2005). Lansium domesticum Corr. was identified as the dominant species at both the seedling and tree stages, found at T3 (Communal Tembawang), indicating ecological adaptability. The high IVI of the tree stage reflects its abundance, wide distribution and large basal area, indicating that mature individuals are well established and structurally dominant in the stand. Concurrently, the high IVI at the seedling stage signifies natural regeneration.
In addition, the rubber tree was one of the important species in tembawang due to its economic value. This was confirmed by Penot (2021) and Rafdinal et al. (2021b), who stated that the local community in West Kalimantan had planted rubber trees over the past two decades due to their economic value and low maintenance requirements. Meanwhile, Mulyoutami et al. (2009) identified that the species typically retained in traditional agroforestry were those with high economic potential, such as fruit-bearing tree species and rubber.
The Diversity Indices of Tembawang
Species diversity is an important metric for determining the stability of a community. Diversity indices in tembawang quantified using the Shannon Index (H'), Simpson Index (1-D) and Evenness (E), are shown in Table 2.
Table 2 shows the species diversity in the study sites, as measured by the Shannon Index (H'), mostly falls into the moderate category (1 < H' < 3), with the exception of the pole stage in Sites T3 and T4, which shows low diversity (H' < 1). Furthermore, the Evenness Index (E) across all stages is consistently high (E > 0.6), indicating a relatively stable distribution of individuals amongst species in the tembawang system.
At the seedling stage, the four study sites exhibited varying levels of species richness, with the number of species (S) ranging from 3 to 6. Sites T1 and T2 recorded the highest richness with six species each, followed by Site T4 with four species and Site T3 with three species. This variation in richness directly influenced the diversity indices. Site T1 demonstrated the highest ecological complexity with a Shannon Index (H') of 1.75 and a Simpson Index (1-D) of 0.82, placing it in the moderate diversity category. Site T2 also maintained a moderate diversity status (H' = 1.51), while Site T3 recorded the lowest values (H' = 0.96; 1-D = 0.57), falling into the low diversity category. Furthermore, the Evenness Index (E) across all sites remained high, with values of 0.96 (T1), 0.75 (T2), 0.87 (T3) and 0.82 (T4). These results (E > 0.6) suggest that, despite the differences in species richness, the individuals at the seedling stage are distributed relatively uniformly amongst the existing species, indicating a stable early-stage regeneration process within the tembawang system.
At the sapling stage, the four study sites showed a marked increase in species richness, with the number of species ranging from 8 to 14. Site T3 recorded the highest richness with 14 species, followed by Site T4 with 11 species, while Sites T1 and T2 each recorded eight species. This elevated species richness at T3 and T4 significantly contributed to their diversity profiles. All study sites were categorised as having moderate diversity, with the Shannon Index (H') ranging from 1.92 to 2.53. Site T3 exhibited the highest complexity (H' = 2.53), followed by Site T4 (H' = 2.16). These findings were further corroborated by the Simpson Index (1-D), which showed high values between 0.83 and 0.91, indicating a low probability of species dominance. Furthermore, the Evenness Index (E) across all sites remained consistently high, with values of 0.86 (T1), 0.85 (T2), 0.89 (T3) and 0.79 (T4). According to the criteria adjusted from Odum (1971), these values (E > 0.6) demonstrate that the sapling community is structurally stable, with individuals distributed relatively evenly amongst the species, ensuring a resilient foundation for the next successional stage.
At the pole stage, the study sites experienced a noticeable decline in species richness, with the number of species (S) ranging from 2 to 5. Site T1 recorded the highest richness with five species, followed by Site T2 with three species, while Sites T3 and T4 each contained only two species. This low species richness significantly impacted the diversity indices. The Shannon Index (H') for Sites T1 (1.58) and T2 (1.04) remained in the moderate diversity category. However, Sites T3 and T4 both recorded a value of 0.69, placing them in the low diversity category (H' < 1). Similarly, the Simpson Index (1-D) for T3 and T4 reached its lowest point at 0.50, indicating a higher degree of species dominance compared to earlier stages.
In contrast to the low diversity, the Evenness Index (E) reached its peak during this stage. Site T1 recorded an evenness of 0.97, Site T2 was 0.95, while Sites T3 and T4 both achieved a perfect value of 1.00. These results represent a very high evenness level (E > 0.6). This unique condition, characterised by low diversity paired with perfect evenness, suggests that, while only a few species are present at the pole stage, their individuals are distributed with absolute uniformity across the sites, likely due to specific traditional management or natural selection processes in the tembawang system.
At the tree stage, the study sites exhibited a significant recovery in species richness, with the number of species (S) ranging from 4 to 13. Site T2 demonstrated the highest richness with 13 species, followed closely by Site T4 with 11 species, while Sites T1 and T3 recorded five and four species, respectively. This increase in species richness contributed to a more robust diversity profile across the sites. The Shannon Index (H') for Sites T2 (2.38) and T4 (2.23) firmly placed them in the moderate diversity category, showing a substantial upward trend compared to the pole stage. Sites T1 (1.46) and T3 (1.35) also maintained a moderate diversity status. These findings were supported by the Simpson Index (1-D), where Sites T2 and T4 achieved high values of 0.89 and 0.87, respectively, indicating a high probability of interspecific encounters and low dominance.
The Evenness Index (E) remained consistently high across all locations, with values of 0.86 (T1), 0.83 (T2), 0.96 (T3) and 0.85 (T4). According to the criteria of Odum (1971), these results (E > 0.6) represent a high evenness level. The exceptionally high evenness in Site T3 (0.96), despite its lower species count, indicates a very stable and uniform distribution of individual trees. Overall, the tree stage highlights the maturation of the tembawang system, fulfilling its role as a structurally complex and diverse forest-garden model.
The moderate to low diversity indices (H' = 0.69 – 2.38) observed across the growth stages in the Peninsung Catchment are consistent with previous vegetation assessments in West Kalimantan's tembawang systems. Prior studies by Artuti Ekamawanti and Widiastuti (2016) and Sumaharni and Kalima (2015) also reported that tembawang values frequently fall into the low-to-moderate range (H' < 2.0), particularly during mid-successional phases. Similar patterns have also been documented in traditional agroforestry systems in other tropical regions, where species diversity tends to be lower than that of natural forests due to selective management practices and the dominance of economically important species (Scales and Marsden 2008, Valencia et al. 2014). This ecological phenomenon is driven by human intervention and strategic agricultural filtering, where local communities selectively manage high-value canopy species, while suppressing non-commercial timber. This aligns with Marjokorpi and Ruokolainen (2003), who confirmed that the tembawang is managed properly to ensure that seedlings and saplings of non-utilised tree species are systematically eliminated from the surroundings of intentionally planted fruit trees. The intense focus on cultivating Hevea brasiliensis (Willd. ex A. Juss.) Mull. Arg. and specific edible resources permanently shape the community, leading to specialised dominance that limits the overall Shannon Index.
Cluster Analysis and Ecological Similarity Patterns
Using the PAST 4.04 software, based on species’ IVIs, a hierarchical Cluster Analysis was performed across the four study sites (Fig. 4). The clustering configuration was systematically modelled using the Unweighted Pair Group Method with Arithmetic Mean (UPGMA) algorithm built upon the Bray-Curtis similarity index. This multivariate approach allows for a rigorous comparison of ecological similarity patterns across growth stages, from the seedling to the tree. The mathematical reliability of each resulting dendrogram was strictly validated using the Cophenetic Correlation Coefficient (CCC) to ensure an accurate, undistorted representation of the original pairwise similarity matrices. The detailed analysis of each growth stage is elaborated in the following sub-sections.
Seedling Stage Similarity
The structural integrity of the dendrogram was supported by a high Cophenetic Correlation Coefficient (CCC) value of 1.00, indicating that the clustering configuration closely represented the original pairwise ecological distance matrix derived from the seedling stage data.
The topology of the seedling dendrogram reveals a symmetrical division into two main distinct clusters. The first primary cluster consists of Sites T3 and T4, which are linked at a similarity threshold of approximately 0.42. This indicates a moderate sharing of seedling species composition between the two plots. In the context of tembawang management, this early-stage clustering suggests that T3 and T4 experience comparable environmental filters or uniform seed rain dynamics that guide their initial species recruitment.
Conversely, the second major cluster groups Sites T1 and T2 together, exhibiting a lower similarity level of around 0.33. This configuration implies that the initial recruitment of young fruit-tree seedlings and understorey wild resources in T1 and T2 follows a slightly more heterogeneous path compared to the T3–T4 group. The definitive basal bifurcation between the (T3–T4) group and the (T1–T2) group at the lowest similarity baseline underscores that the distinct ecological trajectories of these tembawang sites are already established at the forest floor level, setting the foundation for the vertical strata development in later successional stages.
Sapling Stage Similarity
At the sapling stage, the reliability and structural integrity of this resulting dendrogram were validated by a high Cophenetic Correlation Coefficient (CCC) value of 0.8874, confirming that the cluster configuration highly reflects the original pairwise similarity matrix. The dendrogram distinguishes two primary clusters at a similarity threshold. The first major cluster is formed by Sites T3 and T4, which exhibit a remarkably close ecological affinity with a similarity level of approximately 0.80. This high compositional overlapping suggests that, during the transition to the sapling phase, both T3 and T4 share highly comparable environmental filters or human intervention trajectories, leading to a homogeneous recruitment of young woody species.
Conversely, Sites T1 and T2 group together to form the second primary cluster, sharing a similarity level of around 0.40. Within this group, T1 and T2 are closely linked, indicating a moderate degree of shared species composition. The structural bifurcation between the (T3–T4) group and the (T1–T2) group highlights a clear ecological divergence in the tembawang system's understorey layer. This separation implies that, while local fruit species selectively dominate the seedling and sapling stages in T1 and T2, different successional drivers or specific commodity enrichments dictate the early-stage community configuration in T3 and T4.
Pole Stage Similarity
The resulting dendrogram of pole stage demonstrated acceptable model robustness, backed by a Cophenetic Correlation Coefficient (CCC) value of 0.7851, validating its accuracy in representing the original distance matrix. The cluster topology at the pole stage highlights a highly pronounced and close ecological grouping between Sites T3 and T4. These two sites are linked at a very high similarity threshold of approximately 0.86. This extreme compositional overlapping correlates directly with the previous Importance Value Index (IVI) findings, where both T3 and T4 were found to be dominated by Hevea brasiliensis (Willd.ex A.Juss.) Mull. Arg. poles (IVI > 160.00). This indicates that managed rubber intensification creates a uniform vegetative filter in the mid-canopy layer, forcing these two distinct locations into an almost identical ecological state.
In contrast, Sites T1 and T2 display a lower similarity connection, branching out at a lower threshold of around 0.47. At the base of the dendrogram, the (T3–T4) monoculture-like cluster remains distinct from the (T1–T2) multi-species fruit garden group. This clear structural division demonstrates that human-induced economic prioritisation, specifically the choice between establishing rubber-dominant plots versus maintaining heterogeneous native fruit trees, serves as the primary driver behind the mid-canopy community divergence within the tembawang agroforestry system.
Tree Stage Similarity
At the final mature tree stage, the structural integrity of this mature canopy dendrogram is robust, as confirmed by a high representative Cophenetic Correlation Coefficient (CCC) value of 0.9549. This correlation demonstrates that the hierarchical branching accurately represents the true ecological distances and pairwise similarities from the field data.
The topology of the tree stage dendrogram clearly separates Site T4 from the other three locations, branching out independently at the lowest similarity baseline of approximately 0.10. This extreme isolation indicates that the mature overstorey canopy of T4 possesses a unique species composition that diverges drastically from the rest of the catchment. This pattern directly aligns with the IVI data, where T4 remains strongly influenced by mature Hevea brasiliensis (Willd.ex A.Juss.) Mull. Arg. structures, while other sites have successfully transitioned into a diverse matrix of indigenous fruit trees.
Meanwhile, the remaining locations form a separate multi-layered cluster, with Sites T1 and T2 showing a close ecological affinity at a similarity threshold of around 0.50. This pair is subsequently linked to Site T3 at a lower similarity level of approximately 0.24. The strong clustering between T1 and T2 at the tree level highlights a shared traditional management legacy that specifically favours large-stature fruit trees, primarily Durio zibethinus. Overall, the structural configuration at the tree stage proves that, as the tembawang system matures, it creates distinct ecological footprints, where site-specific management histories dictate whether the overstorey remains commodity-orientated (Site T4) or evolves into a species-rich traditional orchard (Sites T1, T2 and T3).
The potency of vegetation in tembawang for critical land rehabilitation
The synthesis of the Important Value Index (IVI), biodiversity indices, vegetation stratification and cluster analysis demonstrates the ecological potential of the tembawang agroforestry system for supporting critical land rehabilitation in the Peninsung Catchment. The vegetation characteristics observed in tembawang, particularly its multilayer vegetation structure, species composition and continuous regeneration across growth stages, indicate important ecological functions that contribute to ecosystem stability.
One of the primary ecological mechanisms underlying this rehabilitation potential is the establishment of a multilayered vegetation structure (Winarni and Susanto 2025). The continuous presence of vegetation from the seedling and sapling stages to pole and mature tree stages reflects strong successional continuity and ecosystem stability. Such vegetation stratification has been recognised as an important indicator of watershed sustainability due to its role in maintaining ecological stability and hydrological functions (Wang et al. 2013, Maridi et al. 2014, Shao et al. 2020, Li et al. 2021). In Site T3, for example, high species richness at the sapling stage (up to S = 14) indicates the presence of a dense understorey layer that functions as an important biological buffer within the degraded Peninsung Catchment area. This lower vegetation layer helps reduce the kinetic energy of rainfall, minimises surface runoff, prevents nutrient loss and enhances water infiltration into the soil.
The canopy layer plays a crucial role in intercepting rainfall by retaining water on leaves and branches before it reaches the ground as throughfall or stemflow. This interception process reduces the kinetic energy of raindrops, thereby minimising soil surface disturbance and lowering erosion rates (van Noordwijk et al. 2004, Marwah 2012). According to Afrianda and Yasin (2023), such protective vegetation cover is effective in suppressing soil erosion and reducing the expansion of degraded land. In addition, understorey vegetation contributes to decreasing surface runoff and improving soil stability (van Noordwijk et al. 2008). The ecological function of tembawang is further strengthened by the accumulation of litter biomass, which helps reduce runoff, increase infiltration and diminish erosion and sedimentation processes (Rafdinal et al. 2021a).
In addition its ecological functions, tembawang also possesses important socio-economic and cultural values that support long-term rehabilitation sustainability. Successful critical land rehabilitation requires not only ecological effectiveness, but also local community acceptance and participation. In West Kalimantan, conservation efforts involving tembawang have proven highly relevant because this traditional agroforestry system is closely linked to the cultural identity and livelihoods of Dayak communities (Marjokorpi and Ruokolainen 2003). Therefore, tembawang represents a locally managed agroforestry system that combined ecologically functions with socio-cultural importance, supporting its potential contribution to critical land rehabilitation in this landscape.
Conclusions
The vegetation characteristics of tembawang in the Peninsung Catchment demonstrate that this traditional agroforestry system has ecological potential for supporting critical land rehabilitation in this catchment. Across all growth stages, the vegetation structure reflects a dynamic, yet stable successional process, characterised by the dominance of ecologically and economically important species, as indicated by high Important Value Index (IVI) values, including Durio zibethinus (IVI = 93.27), Lansium domesticum Corr. (IVI = 117.36), Artocarpus integer (Thunb.) Merr. (IVI = 106.40) and Hevea brasiliensis (Willd. ex A.Juss.) Müll.Arg. (IVI = 161.38).
The diversity analysis further showed that most study sites exhibited moderate species diversity, with Shannon Index (H') values ranging from 1.04 to 2.53 and Simpson Index (1-D) values ranging from 0.72 to 0.91, accompanied by consistently high Evenness values (E = 0.75 - 1.00). These results indicate relatively stable vegetation communities and balanced species distribution within the tembawang system. However, the pole stage in Sites T3 and T4 showed low diversity, with H' = 0.69 and 1-D = 0.50, primarily due to the strong dominance of Hevea brasiliensis (Willd. ex A. Juss.) Mull. Arg. with high IVI values reaching 160.02 and 161.38, respectively. This dominance reflects a temporary rubber-centred management phase within the tembawang system, where mid-canopy vegetation becomes structurally simplified before transitioning into a more heterogeneous mature tree community.
The multilayered vegetation structure consisting of seedlings, saplings, poles and trees contributes significantly to ecological functions, such as rainfall interception, erosion control, runoff reduction, litter accumulation and water infiltration. Overall, the study confirms that tembawang possesses ecosystem stability, as reflected by its IVI, diversity indices and multilayer vegetation structure, thereby highlighting its high potential for rehabilitating degraded land.