A 15-Year Multi-Temporal Assessment of Tropical Peatland Restoration in Sumatra: Landscape Dynamics, Hydrological Drivers, and Vegetation Outcomes at Kebun Raya Sriwijaya (2010-2025)

2.1 Study area

This study was conducted at KRS, a botanical garden located in Ogan Ilir Regency, South Sumatra Province, Indonesia (3°9'31.56"S; 104°32'44"E). The site covers approximately 100 ha and was situated within the Sungai Musi-Sungai Blidah Peat Hydrological Unit (KHG.16.03-07.01), which spans approximately 31,167 ha [12]. The broader hydrological unit represents a functionally connected peatland landscape where surface and subsurface water flows are hydrologically interdependent.

Land-use history and surrounding landscape. Before its designation as a botanical garden, the site was an active oil palm plantation from the 1990s until 2010. Between 2010 and 2015, the site remained in a transitional state with no active restoration, during which time existing drainage canals from the plantation period remained open. To the northeast of the KRS boundary, a coal haul road became visible in 2021 satellite imagery, indicating ongoing extractive land use adjacent to the site. Between 2021 and 2024, a parcel of land previously characterized by shrubland immediately northeast of the boundary was converted into an oil palm plantation. The drainage canals from this adjacent plantation are oriented perpendicular to the KRS boundary, directing water flow toward the botanical garden.

Hydrological infrastructure. In 2020, KRS management completed the installation of hydrological infrastructure designed to regulate water levels within the site. This infrastructure includes: (i) a perimeter canal surrounding the majority of the garden area, (ii) inlet and outlet gates that control water exchange between the perimeter canal and the external drainage network, and (iii) multiple retention ponds distributed across the site to store water during wet periods and release it gradually during dry periods.

2.2 Satellite imagery analysis

Landscape transformation at KRS was analyzed using multi-temporal satellite imagery sourced from Google Earth Pro. Four image acquisitions representing different stages of restoration were selected based on cloud-free conditions and temporal relevance: September 2014 (pre-restoration), February 2017 (early restoration period), July 2021 (mid-restoration), and November 2024 (late restoration). Pixel-based contrast analysis was performed to interpret land cover types, linear features (canals, roads, paleochannels), and boundary changes.

2.3 Peat depth measurement

Peat depth was measured in 2023 and 2025 at 12 sampling points (TS1 to TS14, excluding TS7 and TS11) distributed across KRS. Each sampling point was located within a separate permanent plot of 20 m × 20 m (Figure 1), with three replicates taken per point. Although no permanent above-ground markers (e.g., subsidence poles) were installed at the exact sampling points to enable precise repositioning, all measurements were consistently taken within the same 12 designated permanent plots during both surveys. A standard peat auger was used to extract peat cores vertically until resistance indicated the mineral substrate or dense root layer. Points were categorized based on management history:

• Undisturbed secondary forest (no clearing for ≥10 years): TS9 to TS10 and TS12 to TS14 (n = 5 points).
• Actively managed areas (regular weeding, planting, foot traffic): TS1 to TS6 and TS8 (n = 7 points).

Table 1. Vegetation type of permanent water table monitoring wells at Kebun Raya Sriwijaya (KRS)

2.4 Hydrological data collection

Post-intervention monitoring was conducted from September 2020 to August 2021 at five permanent wells (4-inch polyvinyl chloride, 200 cm depth, 50 cm above peat surface) distributed across different vegetation types. Each well was positioned at a minimum distance of 2 meters from the nearest drainage canal to avoid direct surface runoff effects. Table 1 describes the location, vegetation type, and represented zone for each well. The five wells were distributed across different vegetation types and hydrological zones within the 100-ha site: managed area, canal riparian (W1); managed area, flood-prone transitional area (W2); secondary forest, flood-prone transitional area (W3); secondary forest, canal riparian (W4); and secondary forest, flood-prone transitional area (W5). While the well locations cover the main vegetation types, we acknowledge that the 2-meter distance from canals may limit the representativeness of areas closer to drainage features. Well locations are shown in Figure 1.

Depth to water table was measured manually on the first Wednesday of each month. Monthly rainfall was recorded using a standard ombrometer. The 12-month monitoring period was limited by logistical constraints; this dataset represented the only continuous water table record available. For the pre-intervention period (2014–2019), hydrological conditions were inferred from fire events (2015, 2019), regional rainfall patterns, and satellite-observed surface wetness. Longer-term monitoring is ongoing and will be reported in future communications.

Pearson's correlation coefficient was used to examine the relationship between monthly rainfall and mean depth to water table (n = 12 months, September 2020 – August 2021). For each month, the water table depth was averaged across the five monitoring wells before correlation analysis. Statistical significance was set at p < 0.05. Analyses were conducted using SPSS version 23.

2.5 Vegetation survival assessment

Survival data were obtained from KRS's annual planting records from 2011 to 2023 and annual assessment records from 2013 to 2025. Each planted specimen was georeferenced and monitored annually. The two-year survival rate was defined as the proportion of planted specimens still alive 24 months after planting. A specimen was recorded as dead when no living above-ground tissue was observed during two consecutive annual surveys; specimens that could not be relocated due to missing or damaged tags were excluded from the analysis. For each cohort, the climate condition at maturation (24 months after planting) was assigned based on El Niño–Southern Oscillation (ENSO) phase records from the Indonesian Agency for Meteorology, Climatology and Geophysics [13].

2.6 Collection provenance and International Union for Conservation of Nature Red List status determination

Specimen origins were recorded in KRS's annual log reports (2013–2025). Sources included: (1) Bogor Botanic Gardens seed bank (2013); (2) plant exploration in South Sumatran secondary forests and peat swamps (2015–2025); and (3) village cemetery areas in three villages (2019–2021). Data recorded for each event included date, coordinates, collector name, local name (from local guides), and habitat. Voucher specimens were deposited at the KRS herbarium. The conservation status of each plant species was determined based on the IUCN Red List of Threatened Species. Status categories (Near Threatened (NT), Vulnerable (VU), Endangered (EN), and Critically Endangered (CR)) follow the IUCN Red List Categories and Criteria Version 3.1. Species statuses were verified through the IUCN Red List online database (https://www.iucnredlist.org) accessed in April 2026. For each species, the most recent assessment available was used.

2.7 Ethical procedures and data sources

A standardized ethical protocol was applied to all plant exploration activities, whether in village cemeteries or secondary forests. Formal written requests were submitted to village heads, followed by face-to-face meetings to obtain verbal informed consent. A resident served as a field guide for each collection event, assisting in site location and provision of local plant names. This ensured community participation and respect for local knowledge across all collection sites. To provide transparency regarding the multiple datasets used in this study, Table 2 summarizes the year, sample size, frequency, purpose, and limitations of each data type.

Table 2. Summary of data sources, sample characteristics, and analytical framework used in this study

4.1 Landscape-hydrological change: Beyond canal blocking and rewetting

Over fifteen years, KRS has developed a peatland restoration model distinct from the canal-blocking and rewetting approach dominant in Indonesia and other tropical peatlands. The conventional model, widely used by Indonesia's BRG since 2016, focuses on raising water tables through canal blocking, followed by native revegetation [15]. While effective for reducing carbon emissions, it relies on narrow ecological indicators and treats restoration as a time-bound project [3].

The KRS hydrological strategy diverged fundamentally from conventional peatland restoration practice. Most studies recommend canal blocking as the primary intervention to raise water tables and reduce carbon emissions in degraded tropical peatland [16]. In these approaches, hydrological management serves a single objective: carbon emission reduction. By contrast, KRS maintained the existing drainage network during Phase 1 (2010-2015) to prevent peat desiccation while site planning and nursery establishment proceeded. Over the 12-month monitoring period following infrastructure completion (September 2020–August 2021), the perimeter canal system with inlet-outlet gates and retention ponds appeared to reduce the sensitivity of site hydrology to rainfall variability [17], as indicated by the gradual water table decline despite sharp rainfall reduction (Figure 3). However, longer-term monitoring—particularly during extreme dry or wet periods beyond the 12-month window—remains necessary to confirm the persistence of this buffering capacity.

Unlike conventional canal blocking, which prioritizes carbon outcomes above all else, KRS suggests that hydrological infrastructure may serve multiple purposes [18] without compromising water table elevation. However, this multifunctionality comes at a cost: perimeter canal systems are more expensive and technically demanding than simple canal blocks [19]. The 2024 La Niña flood revealed an additional trade-off [20]: when restoration is located near communities, flood risk becomes a genuine management concern, not merely a carbon accounting footnote. This suggests that future restoration projects in inhabited peatlands must plan for both dry and wet extremes, not a simple return to "natural" water tables—a scenario planning gap evident in most existing hydrological guidelines.

4.2 Vegetation and conservation outcomes: From co-benefit to core mandate

In conventional peatland restoration, biodiversity outcomes are treated as co-benefits—secondary indicators measured if resources permit, but rarely central to project design [21]. Standard indicators focus on hydrology (water table depth) and carbon (emission factors) [22], while species recovery, when monitored at all, is limited to a handful of planted native trees. KRS inverts this priority entirely. As a botanical garden, conservation is a core mandate, not a co-benefit. The living collection of 477 cultivated specimens (117 species, 49 families) is maintained regardless of restoration outcomes, and 12 native Sumatran species with IUCN Red-listed status were deliberately collected, propagated, and planted as part of the garden's multifunctional mandate. Crucially, prioritizing ex-situ conservation increases the complexity of peatland restoration—shifting the goal from merely restoring degraded peat to actively building an ecosystem—a shift not addressed in conventional restoration frameworks [23].

The most distinctive finding—that five of these 12 species were obtained from village cemeteries, not forests—would not emerge from a conventional restoration monitoring framework. Conventional projects do not typically collect from cemeteries because their mandate is rewetting, not botanical exploration. This finding has practical implications beyond KRS: culturally protected landscapes across Sumatra (cemeteries, sacred groves, village forests) may harbor threatened species that formal conservation areas have failed to protect—a possibility that restoration guidelines have largely ignored. For restoration practitioners, this means community-engaged plant exploration can identify local genetic sources that are both culturally appropriate and ecologically valuable—an approach conventional canal-blocking projects rarely consider [24].

The presence of 12 IUCN Red-listed native species at KRS accounts for 2.22% of Sumatra's reported forest and wetland flora [14]. This percentage is not claimed as comprehensive ex-situ coverage of the region's threatened biodiversity. Rather, its significance lies elsewhere: these species now grow on a degraded peatland landscape where they were previously absent. Each individual, though currently maintained as an ex-situ living collection, occupies its native peatland habitat and will mature to produce seeds and other propagules. These propagules can be harvested, multiplied in nurseries, and deployed for large-scale reintroduction into surrounding degraded peatlands—a restoration supply function that conventional canal-blocking projects rarely provide. In this sense, the 2.22% figure is not an endpoint but a baseline: a nucleus of threatened species from which restoration material can be generated. Future work should prioritize seed orchard development and nursery capacity expansion at KRS to realize this potential.

The marked increase in survival rates following infrastructure completion confirms that hydrological regulation buffers planted collections against climate variability. During the pre‑intervention period, survival rates were consistently lower for cohorts that matured under La Niña conditions—the 2015 cohort showed 40.0% survival when assessed in 2017, while the 2019 cohort showed 45.5% when assessed in 2021—compared to those that matured under El Niño conditions—the 2013 cohort showed 56.8% survival when assessed in 2015 and the 2017 cohort showed 63.6% when assessed in 2019. Under regulated conditions following the 2020 infrastructure completion, subsequent cohorts (planted in 2022 and 2023) exhibited two-year survival rates exceeding 95.7% (assessed in 2024 and 2025, respectively), indicating that hydrological regulation had effectively buffered the site against climate extremes.

4.3 Botanical-garden governance: Institutional permanence, failure tolerance, and landscape vulnerability

One of the most distinctive features of the KRS model is its ability to absorb and learn from failure. The 2015 and 2019 fires were not explained away as external anomalies but treated as operational setbacks that informed infrastructure investment. This capacity stands in sharp contrast to conventional project-based restoration, where donor reporting requirements incentivize low-risk site selection, proven methods, and selective reporting of success. A systematic review of peatland restoration literature found that failures—particularly fires and planting mortality—are substantially underreported, creating a biased evidence base that favors over-optimistic conclusions about restoration effectiveness [25].

KRS provides a rare longitudinal account of failure and learning over 15 years—something most restoration studies lack due to their project-based time horizons [21]. The five-year gap between the first fire and completion of water retention infrastructure represents real costs (lost specimens, carbon emissions, institutional credibility), but because KRS has no project closure deadline, it could absorb these costs and continue operating. This institutional feature—permanent mandate and patient capital—is not an accident; it is a deliberate design choice that project-based restoration cannot replicate. By contrast, a typical 3-5 year restoration project would have closed, reported partial success, and moved on. The case suggests a distinction that restoration policy may need to consider: project-based funding models incentivize failure avoidance, while institutionally permanent models may better incentivize learning from failure. However, this inference is drawn from a single case (KRS). Comparative studies across multiple restoration sites with different institutional arrangements would be needed to confirm this as a general mechanism.

However, conventional restoration projects often define their boundaries by administrative limits, implicitly assuming that interventions within those boundaries will produce measurable outcomes while external factors beyond the site can be safely ignored. The KRS case challenges this assumption. Two external landscape dynamics directly affected restoration outcomes at KRS despite optimal internal management. First, a coal haul road appeared along the northeastern boundary in 2021 (Figure 2(C)), signaling ongoing extractive land use adjacent to the site. Second, and more consequentially, a parcel of land previously characterized by shrubland was converted into an oil palm plantation, with drainage canals oriented perpendicular to the KRS boundary, directing water flow toward the botanical garden (Figure 2(D)). The 2015 and 2019 fires similarly originated outside KRS from adjacent land clearance and agricultural burning, not from within the garden itself.

These observations lead to an uncomfortable conclusion: even a well-designed, permanently institutionalized restoration site remains vulnerable to landscape-level drivers beyond its control—a finding that complicates the implicit assumption of most site-scale restoration studies that internal interventions are sufficient. The perimeter canal and water gates that successfully buffered KRS from rainfall variability (Section 3.1) cannot buffer drainage from adjacent oil palm plantations or fires from neighboring land. Landscape-scale restoration has been advocated for decades, yet most tropical peatland projects remain site-scale due to land tenure complexity and governance fragmentation. KRS demonstrates that without parallel investment in landscape governance—beyond the site boundary—the biophysical outcomes of site-scale restoration may be fundamentally capped, a limitation that most published restoration evaluations fail to acknowledge.

4.4 Supporting evidence and its implications

4.4.1 Interpreting anomalous peat depth increases

The peat depth measurements presented in Section 3.3.1 required cautious interpretation. Several points showed increases of +83 to +293 cm over two years—values far exceeding the maximum plausible vertical accretion rate for tropical peatlands (typically <2 cm yr⁻¹). These anomalies likely reflect spatial heterogeneity across the study area, compaction artifacts from repeated augering, and the inherent uncertainty associated with repeat measurements at non-fixed points within permanent plots. Therefore, these data should not be interpreted as evidence of net peat growth or long-term carbon accumulation at KRS. Instead, they serve two limited purposes: illustrating the high spatial heterogeneity of peat depth within a 100-ha site, and providing a methodological caution for restoration monitoring—without permanent markers and co-located elevation surveys, repeated peat depth measurements are unlikely to yield reliable estimates of change over short timeframes.

Beyond methodological artifacts, however, an alternative ecological interpretation deserves consideration. The 15-year restoration period at KRS was punctuated by alternating El Niño (drought) and La Niña (wet) events, which repeatedly stimulated boom-and-bust cycles in understory vegetation—particularly the dominant fern Dicranopteris linearis. This species produces rigid, hollow, highly lignified tissues that decompose slowly and may retain their physical form after death, creating a loosely packed, porous organic layer [26]. Under this hypothesis, the measured depth increases do not represent compacted, well-humified peat, but rather transient, low-density accumulation of undecomposed necromass—material that registers as "peat" in auger measurements but contributes little to long-term carbon storage. This interpretation would explain both the magnitude and reversibility of depth changes observed between 2023 and 2025. Confirmation requires additional data (bulk density, organic matter composition, permanent elevation markers), but if correct, this hypothesis carries an important implication: standard peat depth measurements alone cannot distinguish meaningful carbon accumulation from transient, low-density necromass buildup—a caution particularly relevant for restoration projects claiming rapid carbon recovery. Pending confirmation with permanent markers, elevation surveys, and bulk density measurements, the peat depth data presented here should be interpreted as a methodological caution rather than evidence of ecosystem change.

4.4.2 Socio-economic outcomes and political constituencies

Conventional restoration projects increasingly include "community engagement" components, but these typically take the form of consultation or compensation for lost access—rarely delivering direct economic benefits to communities [27]. Carbon credit revenues primarily flow to project developers and intermediaries, not to forest-adjacent communities, making the latter the least significant beneficiaries relative to their direct dependence on and interaction with forest ecosystems [28]. KRS operates on a different logic. Because the botanical garden has a permanent mandate beyond restoration, it can invest in livelihood linkages impractical for time-limited projects: (1) a natural dye program supplying local artisans, (2) eco-tourism generating local economic activity, and (3) environmental education reaching thousands of participants.

These outcomes suggest a hypothesis that warrants further investigation: that multifunctional restoration models may build local political constituencies for restoration [27]. At KRS, local artisans who benefit from dyes, tourism vendors who depend on visitors, and students who learn about peatlands all have a tangible stake in keeping the garden (and therefore the restored peatland) functional. However, this interpretation is based on garden records and artisan reports, not on formal interviews or surveys. We therefore present this as a case-based insight from KRS, not a generalizable mechanism. Dedicated social science research—including stakeholder interviews and economic valuation—would be required to test whether such political constituencies indeed emerge and whether they contribute to long-term sustainability.

This self-reinforcing political economy represents a mechanism that conventional restoration logic models have largely overlooked, and may prove critical for long-term sustainability—a hypothesis that warrants further investigation. However, this model sacrifices efficiency and scalability relative to conventional rewetting: building a botanical garden is more expensive and slower than blocking canals. The 15-year timeline from clearing to >95% survival is not replicable in a 3-5 year project—a temporal mismatch rarely discussed in restoration policy debates.

4.5 What Kebun Raya Sriwijaya sacrifices: Complementarity, not competition

Relative to conventional canal-blocking restoration, the KRS model sacrifices three things—trade-offs that multifunctional restoration advocates rarely quantify. First, efficiency. Building a botanical garden is more expensive and slower than blocking canals. The 15-year timeline to achieve >95% two-year survival for the most recent cohorts (planted in 2022–2023) is not replicable in a 3–5 year project, contrasting sharply with the rapid emission reduction targets assumed by carbon-focused restoration frameworks. Second, scalability. Indonesia needs to restore over 1.2 million hectares of degraded peatland. Developing botanical gardens across all of them is neither feasible nor desirable—a point often overlooked by proponents of multifunctional restoration who emphasize benefits without acknowledging scale constraints. Third, simplicity. The KRS model is harder to evaluate, harder to fund, and harder to replicate than canal blocking, requiring diverse expertise (botany, hydrology, education, tourism management, community engagement) and sustained political support—requirements that exceed the capacity of most conventional restoration projects.

The way forward is not to choose one model over the other—as much of the current restoration literature implicitly assumes by prioritizing carbon outcomes—but to recognize their complementarity. Conventional rewetting can be deployed at scale across remote peatlands where the primary objective is carbon emission reduction. The botanical garden model can be deployed at strategic nodes—near cities, on degraded plantation land, within existing institutional frameworks—to generate multifunctional outcomes that conventional restoration cannot. This complementary relationship has received limited attention in the peatland restoration literature and, to our knowledge, has not been explicitly articulated previously.

Synthesizing the findings across all domains, the KRS model offers three original contributions that distinguish it from conventional approaches. First, institutional permanence enables outcomes that project-based models cannot achieve. The capacity to survive the 2015 and 2019 fires, delay infrastructure investment, and continue operating after the 2024 flood—all without a project closure deadline—represents a category of restoration governance not captured by existing project-based logic models. Second, multifunctionality can be a design principle, not an afterthought. By embedding restoration within a botanical garden mandate, KRS generated outcomes—threatened species from village cemeteries, natural dye livelihoods, environmental education for thousands of participants—that are less likely to emerge from carbon- or hydrology-driven projects based on available evidence from this case. Third, even well-executed site-scale restoration remains vulnerable to external landscape drivers. Adjacent oil palm drainage, fires originating on neighboring land, and extractive land uses persistently affected site conditions despite optimal internal infrastructure, revealing fundamental limits that restoration science has under-theorized.

Thus, KRS does not replace conventional restoration but expands the available toolkit for peatland restoration practitioners and policymakers. The 15-year record provides preliminary empirical proof of concept for this expanded toolkit—a contribution that, to the best of our knowledge, has not been previously offered in tropical peatland restoration literature.

4.6 Limitations

Several limitations should be acknowledged. Hydrological data: continuous water table monitoring was available for only 12 months (September 2020–August 2021) due to COVID-19 disruptions; pre-intervention conditions (2014–2019) were therefore inferred from fire events, rainfall patterns, and satellite observations rather than direct measurement. Longer-term monitoring is ongoing. Peat depth measurements: large increases between 2023 and 2025 at several points exceed plausible accumulation rates, likely reflecting spatial heterogeneity or measurement artifacts, not net peat growth. Future studies should employ permanent marked points and co-located elevation surveys. Single-site case study: KRS is one botanical garden on one peatland in South Sumatra, so generalization requires caution. However, this study aims for analytical generalization—developing a model to be tested elsewhere. Replication at other Indonesian peatland botanical gardens would strengthen the evidence base. Socio-economic data: findings on natural dye, eco-tourism, and education are based on garden records and artisan reports, not independent economic evaluation. A formal economic impact assessment is a valuable next step. Carbon balance not directly measured: we did not measure greenhouse gas fluxes (CO₂, CH₄); carbon implications are inferential. Direct flux measurements using chamber or eddy covariance methods would strengthen future work.

This work is supported by the Regional Research and Innovation Agency of South Sumatra Province, Indonesia and the Regional Research and Innovation Agency of Ogan Ilir Regency, Indonesia.