Active restoration accelerates the carbon recovery of human-modified tropical forests

The carbon gain in restored logged forest There is currently great interest in the capacity of global forest to store carbon and hence contribute to the mitigation of climate change in the coming decades. In a study of Southeast Asian tropical forest, Philipson et al. show that active restoration of logged forests generates higher rates of carbon accumulation than naturally regenerating forest. To estimate the economic feasibility of restoration treatments, they modeled the carbon price required to offset the cost of restoration, finding that the highest prices seen in recent years would be needed to approach those that could offset restoration costs. These results are important for tropical forest policy, establishing the importance of restoration for the carbon recovery potential of tropical forests. Science, this issue p. 838 Restoration of logged tropical forests will be incentivized by carbon prices, consistent with the 2016 Paris climate agreement. More than half of all tropical forests are degraded by human impacts, leaving them threatened with conversion to agricultural plantations and risking substantial biodiversity and carbon losses. Restoration could accelerate recovery of aboveground carbon density (ACD), but adoption of restoration is constrained by cost and uncertainties over effectiveness. We report a long-term comparison of ACD recovery rates between naturally regenerating and actively restored logged tropical forests. Restoration enhanced decadal ACD recovery by more than 50%, from 2.9 to 4.4 megagrams per hectare per year. This magnitude of response, coupled with modal values of restoration costs globally, would require higher carbon prices to justify investment in restoration. However, carbon prices required to fulfill the 2016 Paris climate agreement [$40 to $80 (USD) per tonne carbon dioxide equivalent] would provide an economic justification for tropical forest restoration.

T ropical forests contain 55% of global stores of aboveground forest carbon (1), but the size of these stocks is declining rapidly because of forest loss and degradation (2). Across the tropics, primary forests continue to be degraded by numerous human impacts, such as timber harvesting, agriculture, and fire. Consequently, more than half of all tropical forests have some human impact, even though many disturbed sites retain high tree cover (3). Loss of tropical forest cover is particularly acute in Southeast Asia, which has the highest deforestation rate in the tropics and where the intensity of logging is increased by the high densities of commercially important dipterocarp trees (4). These forests, which are no longer pristine, may still support numerous ecosystem services, including timber production, sequestration of carbon, maintenance of biodiversity, and hydrological services (5)(6)(7)(8).
Despite their ecological value, degraded forests remain vulnerable to conversion to agroecosystems possessing substantially lower carbon stocks and biodiversity (5,9,10). Alternatively, carbon stocks in degraded tropical forests can recover, particularly if accelerated by active restoration and if financial compensation mechanisms encourage avoided deforestation projects. However, these mechanisms require verification of aboveground carbon density (ACD) baseline values and recovery rates (11), which are currently lacking. Mean and maximum ACD values are higher in Southeast Asian forests than in other tropical forests, and the highest values occur in Malaysia (12). For example, in the Malaysian state of Sabah, where selective logging has been one of the main forms of habitat degradation, unlogged lowland forests show consistently high ACD values averaging over 200 Mg C ha −1 , whereas in logged forests, ACD varies from 60 to 140 Mg C ha −1 (10). The difference in ACD between logged and unlogged forests in Sabah shows the potential carbon gain if logged forests were allowed to recover, which is estimated at 362.5 Tg C (10). At current carbon prices [typically between $2 and $10 (USD) per tonne carbon dioxide equivalent (tCO 2 e)], the potential value of this sequestered carbon would total between $0.725 billion and $3.625 billion for Sabah alone (13)(14)(15). Similar values could be calculated for any territory that has a comprehensive map of forest carbon and information on land-use history, but such data are not widely available across the globe.
Carbon sequestration rates in degraded tropical forests are highly variable. Long-term studies indicate that postlogging carbon recovery rates are between 0.30 and 4.3 Mg C ha −1 year −1 in Southeast Asia (11) and 0.04 and 2.96 Mg C ha −1 year −1 in Amazonia (16), whereas naturally regenerating pasture and abandoned agricultural land accumulate ACD at a mean rate of 3.05 Mg C ha −1 year −1 in the Neotropics (17). These rates could be enhanced by implementing active restoration measures, which include tree planting, cutting of climbers such as lianas, and liberation of sapling trees from competing vegetation by thinning. Enrichment planting is especially important in the logged forests of Southeast Asia because selective logging affects the ecologically and economically important dipterocarp trees that are dispersal limited and mast fruit at irregular intervals (18,19). Such measures are, however, expensive to implement; for example, enrichment planting of lowland forest in Sabah costs~$1500 to $2500 ha −1 over the implementation period, consistent with estimates of restoration costs for other tropical forests globally (table S1). Carbon offset schemes could provide a potential financing mechanism for these restoration costs, but evidence of restoration treatment efficacy with respect to ACD recovery in degraded forests exists for very few sites globally (table S1). The likelihood that such measures will be adopted is critically dependent on the operational costs over the lifetime of the stand relative to the additional value in terms of enhanced ACD accumulation.
Here, we report estimates of the response of ACD accumulation rates to active restoration using a combination of climber cutting and enrichment planting in a logged tropical forest over decadal time scales. We compare the fiscal benefits of this restoration across a range of potential carbon prices. Using detailed information on logging history and repeated in situ measurements from 257 forest plots from three different plot networks in Sabah, Malaysia, we compared recovery rates for naturally regenerating forest with recovery rates for areas that had been actively restored (20). During the 30 to 35 years after logging, naturally regenerating forest accumulated aboveground carbon at a rate of 2.9 Mg ha −1 year −1 ( Fig. 1 and fig. S2) [confidence interval (CI): 2.1 to 3.7], whereas those areas with active restoration recovered at the considerably higher rate of 4.4 Mg ha −1 year −1 ( Fig. 1 and fig. S2) (CI: 3.6 to 5.2). These values suggest that the reduction in ACD associated with a single logging event would be recovered to the ACD of unlogged forest (mean of 203 Mg ha −1 ) ( Fig. 1  and fig. S2) (95% CI: 157 to 247) through natural regeneration after~60 years, but that this could be reduced to 40 years if restoration treatments are applied.
We validated the distribution of ACD spanned by our plots using a fine-spatial-resolution ACD map of the study landscape that was generated using an airborne LIDAR (light detection and ranging) survey in 2016 and calibrated with independent ground surveys (10,21). Remote estimates showed a mean ACD of naturally regenerating forests in 2016 to be 135 Mg ha −1 (Fig. 2) (CI: 123 to 148), whereas that of forest that had been subjected to restoration treatments was 166 Mg ha −1 (Fig. 2) (CI: 152 to 176), confirming a substantial difference in ACD in response to restoration (Fig. 2). We inferred remote estimates of ACD accumulation rates based on the 2016 LIDAR-derived carbon map and baseline ACD values derived from plot data (the intercepts in Fig. 1), which gave values of 3.5 and 4.8 Mg C ha −1 year −1 without and with restoration treatments, respectively, consistent with the results from ground surveys alone. Together, these results suggest that our estimates of carbon recovery are robust and scalable across the whole study area.
To estimate the economic feasibility of applying restoration treatments, we modeled the carbon price required to offset the cost of res-toration, assuming carbon credits are released every 5 years for a project life span of 30 years and adjusting for the time value of money through nominal discount rates of 1, 5, and 10% ( Fig. 3 and fig. S3). Carbon prices on the voluntary market fluctuate widely, and our analyses suggest that only values close to the top of those seen in recent years (around $10 per tCO 2 e) approach the minimum value required to offset the cost of restoration by tree planting and maintenance (13-15). Accounting for variation in ACD recovery rates Philipson   Transverse Mercator (in meters). (Right) Violin plots indicating the distribution of ACD (Mg ha −1 ) from logged forest allowed to regenerate naturally (left, red outline), actively restored (middle, blue outline), and from primary unlogged forest (right, green outline). The data presented on the right correspond to the full study area shown on the left and are independent of those used in the analysis of forest regrowth (Fig. 1).
suggests that implementing restoration uniformly across the logged forest landscape would require carbon prices 2-to 10-fold greater than those that currently exist in the voluntary carbon market (Fig. 3). Independent reports suggest that carbon prices in the range of $40 to $80 per tCO 2 e by 2020 are required to fulfill the obligations of signatories to the Paris Climate Agreement for maintaining a global temperature rise of less than 2°C (22), and a value in this range would be sufficient to offset the costs of tropical forest restoration in our model (Fig. 3). We report the long-term gains in tropical forest ACD after restoration of logged forest using interventions such as enrichment planting, climber cutting, and liberation thinning. These methods contribute differentially to ACD recovery. Climber cutting is likely effective because lianas compete with trees and substantially reduce carbon accumulation in tropical secondary forests (23)(24)(25). Enrichment planting eliminates the constraints of dispersal limitation for large canopy trees and may have resulted in a more uniform distribution of trees than in areas regenerating naturally, thus filling canopy gaps more quickly and reducing light competition from other species (18). The tree species that were planted, mostly in the Dipterocarpaceae family, have a potential for rapid growth rates (26,27) and include the tallest trees recorded in the tropics (28), yielding high biomass when mature. Our findings support previous claims that rates of ACD accumulation in formerly logged Southeast Asian tropical forests are among the highest in the tropics [e.g., (11)] and reinforce the value of logged forests with respect to carbon storage potential in addition to maintenance of biodiversity and other ecosystem functions and services (5-8, 10, 29). We also show that targeted restoration treatments, initiated an average of 9 years after logging, generate substantially higher rates of ACD over the following two decades, which has important implications for the conservation and management of logged forests.
The breakeven carbon price can be estimated for any combination of ACD accumulation rate attributable to restoration and restoration costs for a specific set of economic assumptions (Fig. 4). Restoration programs have a lower breakeven carbon price if they achieve a higher additional ACD accumulation rate (over and above natural regeneration) or if the costs are reduced (Fig. 4). To our knowledge, only two other studies have reported both the costs and additional carbon benefits of tropical forest restoration (30)(31)(32). In Uganda, for lands dominated by grasses that were degraded by agricultural encroachment, an additional ACD gain of 1.62 Mg ha −1 year −1 was achieved through protection from fire and tree planting at a cost of $1200 ha −1 , and abandoned pastures in Costa Rica gained an additional ACD of 1.17 or 2.48 Mg ha −1 year −1 through tree planting at costs of $297 or $1100 ha −1 , respectively, depending on planting strategy. The mean cost of tropical forest restoration (except Australia) across more than 50 published examples was $1596 ha −1 (95% CI: $1338 to $1854 ha −1 ) (Fig. 4 and table S1). Most    (see table S1 for data and details).
of these examples were derived from projects in tropical developing countries where restoration costs are less than $5000 ha −1 , whereas restoration costs of Australian forests are in the range of $6000 to $15,000 ha −1 (table S1). This review of previously published costs suggests that the range of carbon prices available on the voluntary market during 2017 and 2019 would be sufficient to incentivize investment in widespread active restoration in about half the settings where restoration costs have been reported, as long as the additional ACD gains from this investment are equivalent to those achieved by the three case studies in Fig. 4. Conversely, current carbon prices may be insufficient to support restoration in the logged forests of Southeast Asia despite the high rates of recovery reported in this study, unless financing to accept a low (1 to 3%) nominal discount rate is available ( fig. S3). An additional constraint is that the infrastructure and labor force required to implement this large-scale restoration across the global tropics are lacking in many sites, particularly in Southeast Asia, where mast fruiting necessitates greater investment in seedling nurseries (33). Under these circumstances, an alternative approach is to implement generic low-cost measures such as climber cutting, combined with selective tree planting in accessible parts of the degraded forest landscapes where the density of mature trees is insufficient to ensure adequate natural regeneration. This strategy may be attractive to investors in the carbon market, even at current carbon prices, and would leverage recent investments in a new generation of space-borne sensors designed to deliver global high-resolution maps of forest biomass (34)(35)(36). Varying the type and intensity of restoration treatments according to the residual ACD of the stand has the potential to reduce the net costs of implementation, help bridge the gap to financial sustainability, and therefore enable much larger areas of forest to be restored.
Carbon stocks and future carbon sequestration are not the only valuable services provided by forest ecosystems (37), and climate change mitigation is not the single goal of restoration, particularly for local stakeholders. The multiple co-benefits of restoration, such as biodiversity conservation (8,38), flood protection, provision of clean drinking water, and support for the livelihoods of local communities and stakeholders, provide additional justification for legislation and financing mechanisms that incentivize tropical forest restoration.