Comment on “The global tree restoration potential”

The paper by Bastin et al. (1) is to be welcomed for drawing fresh attention to the potential for expanding forest and tree cover to mitigate global climate change by sequestering CO₂ from the atmosphere. Unfortunately, it fails to acknowledge the huge amount of research in this field during the 1980s and 1990s. This is a common problem because the UN Framework Convention on Climate Change did not begin negotiations on a REDD+ (Reducing Emissions from Deforestation and Degradation) mechanism until 2007, and research to support REDD+ has grown rapidly since then. As Bastin et al. emphasize, time is short. So it is vital that new forest mitigation programs build on preexisting knowledge and do not try to “reinvent the wheel.” Here, we use key achievements of this early research to assess the contributions and limitations of Bastin et al.’s findings.

Bastin et al. find “room for an extra 0.9 billion hectares of canopy cover, which could store 205 gigatonnes of carbon” and relate this to the net amount of carbon transferred into the atmosphere since pre-industrial times. Early estimates, however, focused on the size of a new forest sink needed to sequester a meaningful amount of carbon on a continuing basis. They included (i) 500 Mha to absorb gross emissions in the 1980s of 5 gigatonnes of carbon per year (GtC year^–1) (2), and (ii) 465 Mha with a growth rate of 15 m³ ha^–1 year^–1 to sequester only the net annual rise of 2.9 GtC year^–1 in the atmosphere (3) after uptake by terrestrial and oceanic sinks.

Later studies showed that more than enough degraded tropical land existed to support forest expansion on this scale. Two papers presented to an Intergovernmental Panel on Climate Change (IPCC) conference in January 1990 estimated that in the tropics, (i) 620 Mha of degraded lands were physically suitable for establishing this new “carbonforest,” and another 137 Mha of degraded forests could be restored (4); and (ii) 500 Mha of land could be afforested by 2100, with a further 365 Mha of forest fallows having potential for restoration (5). Both studies were summarized in the First IPCC Assessment Report (6). The comparability of these findings with those of Bastin et al. is remarkable, given that Bastin et al. use very high (≤1 m)–resolution satellite data and cloud-based machine-learning algorithms, whereas early estimates depended largely on United Nations statistics—although one 1990 estimate did use low (≥1 km)–resolution satellite data (5).

Bastin et al. also relate their principal finding to a recent IPCC estimate (7) that 950 Mha of new forests could help to “limit global warming to 1.5°C” by 2050. Because this estimate is based on the current net annual rise in CO₂ in the atmosphere, which is twice as high (~5.7 GtC year^–1) as in the 1980s, it is consistent with the 465 Mha considered appropriate 30 years ago (3).

Bastin et al. do not evaluate the operational feasibility of expanding tree cover in time to tackle global warming promptly. Nor does the recent IPCC report, which discusses planting 950 Mha of new forests in just 30 years (7). Yet one early study (4) argued that to afforest 600 Mha over a 30-year period would require a planting rate 20 times the contemporary rate of 1 Mha year^–1. It anticipated REDD+ by showing that more modest planting rates could suffice if afforestation proceeded in parallel with programs to reduce the rate of tropical deforestation, which is a major source of carbon emissions (4).

Bastin et al. assess the potential to increase carbon density in existing forests, using carbon densities in protected areas as benchmarks, but do not mention a pioneering methodology for making restoration assessments that was devised in the early 1990s and applied to all of tropical Asia. Starting with an FAO map of tropical forest area in Asia in 1980 derived from medium-resolution Landsat satellite data, the distribution of potential forest carbon density was determined by using Geographical Information System modeling to combine forest inventory data (8) with multiple environmental datasets. Using degradation factors developed as a function of human population density for each ecofloristic zone along a moisture gradient, actual carbon density was then mapped to identify the distribution of degraded forests that could be restored. This map agreed well with an alternative map of a global vegetation index quantified using low-resolution satellite data (9, 10).

Commenting on Bastin et al.’s paper, Chazdon and Brancalion (11) wisely stress the need to address “social and environmental issues” in tree restoration. The importance of integrating afforestation with forest conservation was recognized in 1991 (12). Early studies also estimated the costs of afforestation; for example, just to plant a 300 Mha carbonforest over 30 years would cost US$4 billion annually (4). These costs would be offset by income from converting the wood produced in the new forests into energy and harvested wood products (HWP), so these forests would be sustainable carbon sinks. The potential role of HWP in climate change mitigation is still poorly understood.

Increasingly sophisticated tools for feasibility analysis were then developed. The “technical suitability” of land for afforestation was distinguished from the “actual availability” of land on which afforestation might be socially, economically, and politically acceptable (10, 13), and the link between the level of national development (i.e., “developmental time”) and the area of actually available land was recognized in the new concept of “national forest carbon transition functions” (14). Now that climate change is shifting the potential locations of biome boundaries (15), Bastin et al. rightly allow for the influence of “climatic time” on their assessment by including alternative climate change scenarios.

If Bastin et al.’s paper gives new impetus to using forests to mitigate climate change, then the results of this early research can finally be used for the purpose for which they were originally intended.