Using machine learning algorithms to predict groundwater levels in Indonesian tropical peatlands.

Tropical peatlands play a vital role in the global carbon cycle as large carbon reservoirs and substantial carbon sinks. Indonesia possesses the largest share (65 %) of tropical peat carbon, equal to 57.4 Gt C. Human perturbations such as extensive logging, deforestation and canalization exacerbate water losses, especially during dry seasons, when low precipitation and high evapotranspiration rates combine with the increased drainage to lower groundwater levels. Drying and increasing temperatures of the surface peat exacerbate ignition and wildfire risks within the peat soils. As such, it is critically important to know how groundwater levels in peatlands are changing over space and time. In this study, a multilinear regression model as well as two machine learning algorithms, random forest and extreme gradient boosting, were used to model groundwater level over the study period (2010-12) within a peat dome impacted by drainage canals and multiple wildfires in Central Kalimantan, Indonesia. Although all three models performed well, based on overall fit, spatial modeling of groundwater level results revealed that extreme gradient boosting (R2 = 0.998, RMSE = 0.048 m) outperformed random forest (R2 = 0.997, RMSE = 0.054 m) and multilinear regression (R2 = 0.970, RMSE = 0.221 m) near drainage canals, which are key fire ignition risk locations in the peatlands. Our study also shows that, on average, elevation and precipitation are the most important parameters influencing groundwater level spatiotemporally.

The road to recovery: a synthesis of outcomes from ecosystem restoration in tropical and sub-tropical Asian forests.

Current policy is driving renewed impetus to restore forests to return ecological function, protect species, sequester carbon and secure livelihoods. Here we assess the contribution of tree planting to ecosystem restoration in tropical and sub-tropical Asia; we synthesize evidence on mortality and growth of planted trees at 176 sites and assess structural and biodiversity recovery of co-located actively restored and naturally regenerating forest plots. Mean mortality of planted trees was 18% 1 year after planting, increasing to 44% after 5 years. Mortality varied strongly by site and was typically ca 20% higher in open areas than degraded forest, with height at planting positively affecting survival. Size-standardized growth rates were negatively related to species-level wood density in degraded forest and plantations enrichment settings. Based on community-level data from 11 landscapes, active restoration resulted in faster accumulation of tree basal area and structural properties were closer to old-growth reference sites, relative to natural regeneration, but tree species richness did not differ. High variability in outcomes across sites indicates that planting for restoration is potentially rewarding but risky and context-dependent. Restoration projects must prepare for and manage commonly occurring challenges and align with efforts to protect and reconnect remaining forest areas. The abstract of this article is available in Bahasa Indonesia in the electronic supplementary material. This article is part of the theme issue 'Understanding forest landscape restoration: reinforcing scientific foundations for the UN Decade on Ecosystem Restoration'.

Effects of distance from canal and degradation history on peat bulk density in a degraded tropical peatland.

Over recent decades, the combination of deforestation, peat drainage and fires have resulted in widespread degradation of Southeast Asia's tropical peatlands. These disturbances are generally thought to increase peat soil bulk density through peat drying and shrinkage, compaction, and consolidation. Biological oxidation and fires burning across these landscapes also consume surface peat, exposing older peat strata. The prevalence and severity of deforestation, peat drainage and fire are typically greater closer to canals, built to drain peatlands and provide access routes for people. We compared bulk densities of 240cm peat profiles from intact forests and degraded peatlands broadly, and also assessed differences between degraded peatlands near-to-canals (50-200m from the nearest canal) and far-from-canals (300+ m from the nearest canal). The effects of vegetation type and fire frequency on bulk density, irrespective of the distance from canal, were also investigated. Mean bulk density values ranged between 0.08 and 0.16gcm-3 throughout the 240cm peat profiles. Drainage of peat near-to-canals increased bulk density of peat above the minimum water table depth. Degradation by deforestation and fire also increased bulk densities of upper peat strata, albeit with greater variability. Peat sampled further from canals experienced less intense water table drawdowns, buffering them from drainage effects. These areas were also more commonly forested and burnt less frequently. Differences in bulk densities below minimum water table levels are less clear, but may reflect lowering of the current peat surface in degraded peatlands broadly. These results clearly show that important differences in bulk density exist across degraded peatlands that are spatially dependent on distance from canals and disturbance history. These landscape features should be taken into account when designing future bulk density sampling efforts and peatland restoration programs, or when extrapolating from existing sources in order to make accurate inferences from them.