Biomass growth rates across climates and over time
Planted forests and natural regeneration were found to be the FLR types that have been most widely studied across the globe, representing 45% and 32% of all the data points collected, respectively, and covering all climates and continents. Both FLR activities demonstrated regional trends, but the variability of natural regeneration across regions, as well as stand age heterogeneity, precluded us from drawing stronger conclusions for this interpretation. We explored, however, the distribution of total biomass growth rates of species planted in planted forests. Our data show that the average growth rate is significantly different between climates (p-value < 0.05, F (1, 44) = 4.062), consistently increasing from colder to warmer climates, with a stronger relationship when Eucalyptus is excluded (R2 = 0.99 and p-value < 0.001 with Eucalyptus; R2 = 0.94 and p-value < 0.01 without Eucalyptus; Fig. 4). Eucalyptus was found to maintain a relatively consistent growth rate from temperate to tropical climates, dry or humid, evidencing its efficient biomass productivity across regions.
Planted forests, and especially commercial timber plantations, are typically managed to maximize productivity in short periods of time, i.e., during the first 20-year period after establishment [30, 33, 44] and therefore, longer rotation periods would result in lower average sequestration rates (as reflected in CO2FIX carbon modelling program [45]). On the other hand, removal factors produced from our research on natural regeneration show that biomass growth can be as high or higher in the 20–60-year period after establishment (Fig. 3a). Crouzeilles et al. [46] showed similar results in naturally regenerated forests and determined that the diversity among natural forest species and the frequently low initial tree density associated with unassisted natural regeneration favors slow initial biomass accumulation rates of low wood density trees, while succeeding trees establish more consistently and at higher density and with higher wood density once the forest is more established [47]. Agroforestry, in contrast, shows minimum growth after 20 years, suggesting that this FLR category uses fast growing species to maximize the efficiency of this system to provide protection, and/or produce fodder, fruit, or timber, among other products.
Applicability of this database
Accounting for the benefits of forest landscape restoration
The loss and degradation of forest habitat diminishes the ability of the landscape to capture atmospheric CO2 and results in the decline of goods and services upon which a significant portion of the global population depend [10, 11], impeding livelihoods and the ability to adapt to a changing climate. Key ecosystem services provided by forest ecosystems include regulation of water quality and quantity [10, 48, 49], regulation of climate [50, 51], protection of biodiversity and soils [12, 13, 52], and provision of food and goods [10, 11, 13]. While forest landscape restoration is not intended as an alternative to conserving forests [53], it can recover multiple benefits lost by deforestation and forest degradation [10, 11, 13, 54] by restoring forest health and ecological functionality at the landscape scale [13, 19, 54].
Despite the multitude of benefits offered by FLR, it is the ability to sequester carbon through the increase of standing biomass in the landscape what has driven many efforts to expand FLR efforts worldwide [19]. Our set of CO2 removal factors can be a valuable resource for countries, practitioners and policy makers that need to associate reliable carbon capture numbers with current and planned FLR activities, and thereby help boost global FLR efforts. Where maximizing CO2 removal is a priority, the factors provided in this study help identify the most efficient FLR options for capturing carbon in each region, which is shown in this study to be planted forests and woodlots. These monocultures may allow for fast sequestration and potential long-term storage of carbon [10], but can have negative implications for water availability, biodiversity, and other ecosystem functions [15, 55,56,57] that offer a great range of long-lasting socioenvironmental benefits [13, 54, 58]. Globally, forest species richness has been demonstrated to increase productivity to the extent that the economic value of the forest has been estimated to be over five times the costs of its conservation [28, 29].
Planning FLR actions is approached at the landscape scale, encompassing entire watersheds, diverse land uses, and communities and their livelihoods [13, 19, 54], seeking to modify poor land use practices that led to the loss of forest habitats and landscape fragmentation and to enhance human well-being [12, 13, 19]. Successful FLR approaches must therefore take into consideration the needs and priorities of local communities, balancing the full range of benefits offered by the various FLR options rather than aiming to maximize one [10, 19], and accounting for the viability of FLR activities in the geographic and biophysical context [13, 53]. Accordingly, ensuring successful long-term FLR efforts needs to include parallel efforts to develop a sustainable forest sector and sustainable biomass energy production activities [15]. This entails meeting the long-term needs of populations while introducing campaigns that educate and show communities the value of FLR efforts for the often-overlooked critical ecosystem services they provide. National definitions and circumstances can dictate what activities are officially classified as forestry or agriculture, often influencing which activities are included under national FLR restoration pledges. Where agroforestry is officially considered by governments to be an agricultural practice, it may be excluded from the restoration pledges. However, agroforestry is a valid agriculture-based FLR activity [19] that affords significant socioeconomic and biophysical benefits [54, 58]. The value of this study comes in both providing data for current FLR pledges as well as contributing to the body of knowledge on the mitigation potential of agroforestry activities to support meeting restoration-related goals.
The need for robust and specific CO2 removal factors
Global commitments such as the Bonn Challenge and NDCs will require reporting on progress made toward established goals. The research undertaken in this study and resulting removals rates offer a useful resource for generating credible estimates of CO2 sequestration achieved for countries who made restoration pledges under these commitments and may help inform ongoing efforts by providing a way to compare the relative impact of different FLR activities. Our study also provides pledgers with key data needed to fill current knowledge gaps on removals from FLR activities across regions and climates, and to support their reporting under the Bonn Challenge Barometer of Progress that is currently under development [59]. Further, the CO2 removal rates developed in this study may also help analysts to better validate and compare assessed climate impacts as reported by pledgers.
To date, the Forest Land chapter of the IPCC Guidelines [20] has been the most widely used source of tier 1 removal factors where in-country data are scarce or not available. Yet these Guidelines only provide removal factors for natural forests and commonly planted species in forest plantations. This leaves out agroforestry and mangrove restoration FLR activities, common around the globe (Fig. 1), which also play a significant role in removing CO2 from the atmosphere (Tables 1, 2) and provide multiple ecosystem service benefits [58, 60]. Furthermore, a significant additional shortcoming of the removal rates offered by the IPCC [20] is that, while offering a data range in some cases, they do not offer estimates of uncertainty nor basic standard statistical information and thus, the accuracy of these commonly used CO2 removal rates is unknown.
Langner et al. [21] used pan-tropical biomass maps [23, 24] to estimate alternative tier 1 removal rates of tropical forests and determined that IPCC’s biomass growth rates were up to 35% higher than what their map-based method estimated. Nonetheless, when they compared intact forests only, their results were similar to IPCC default rates. Intact forests, however, do not represent the state of most tropical forests [4, 9, 21]. The removal factors for planted forests and naturally regenerated forests from the IPCC Guidelines [20] and our study are comparable, yet as in previous studies [21], we find them consistently higher than our rates (on average, 41% higher in planted forests and 38% higher in naturally regenerated ones). Although the IPCC does not provide natural regeneration rates for Central America, its biomass growth rates in naturally regenerated forests are higher for Africa and South America compared to the other regions, as in our study.
Overall, the comparison of the suite of FLR CO2 removal rates developed in our study with the current IPCC removal rates shows that we include a broader range of FLR activities (agroforestry and mangrove restoration in addition to plantations and natural regeneration, currently represented in the IPCC Guidelines), climates (IPCC plantations defaults are only for tropical climates where as we also include boreal and temperate data), and regions (we provide tier 1 defaults for all regions and climates of the world), along with 95% confidence intervals and goodness of fit (R2), which the IPCC Guidelines are currently lacking. Our study also represents a more comprehensive and updated compilation of data on tree growth rates, with removals rates derived from data on biomass increment from over 330 published studies and reports, whereas the IPCC 2006 defaults included less than 100 studies (including the 2003 IPCC Guidelines). Further, this study benefits from over 10 years of additional research on biomass carbon sequestration (more than 36% of the studies used in the development of our growth curves were published post-2006).
Remaining gaps and limitations
While our set of removal factors (Tables 1, 2) are robust and more comprehensive than those available in the IPCC Guidelines [20], there are still gaps and limitations in data availability. FLR data are available across the globe (Fig. 1), yet a large proportion of FLR studies are focused in tropical regions [15, 33, 61]. As elaborated in the Results section of this paper, these include limited data for some planted species in dry climates, data abundance but high variability of teak and eucalyptus plantations, a lack of data on natural regeneration of African dry forests, scarce data on natural regeneration of European forests, and a limited data available for agroforestry in Europe, North America, and Oceania that has prohibited the construction of specific removal factors for different agroforestry activities. Practitioners seeking to produce CO2 removal estimates of agroforestry practices in the United States can also do so using the USDA Comet-Farm [27] platform. CO2 removal rates offered by the IPCC Guidelines [20] represent only the aboveground biomass pool, while those produced through our study include both aboveground and a calculated estimate of belowground growth. They therefore exclude other relevant carbon pools in the ecosystem such as soil carbon and, of a lesser size, litter and dead wood carbon pools [62]. While this may underestimate the carbon sequestration potential of FLR activities, the majority of carbon in forest ecosystems is typically stored in living biomass. However, FLR activities such as some restored mangrove forests or forests established in organic soils would have significant soil carbon stocks [63, 64]. Studies seeking to report total ecosystem removal rates would need to include these additional pools, which would require evaluating the availability and applicability of national data, soil carbon maps where available (e.g., soil mangrove carbon map by Sanderman et al. [65]), or using default stocks provided by the IPCC Guidelines [20], with the same limitations mentioned above.
Both IPCC tier 1 defaults and the removal factors developed through our study represent broad, regional estimates that do not necessarily account for important management and biophysical conditions that can significantly impact CO2 removals rates. A wide range of factors such as former land use, topography, soil type and quality, microclimate, management practices, and proximity to pollinators and seed sources, can have a significant impact on the success (and subsequent CO2 removal) of FLR activities [13, 66,67,68,69]. Thus, the removal factors from our study and the IPCC defaults cannot be used as an alternative to site-specific data needed for carbon offset projects seeking to participate in market schemes that transact carbon credits.
Lastly, this study presents the CO2 sequestration potential of the assessed FLR activities and does not account for emissions associated with them (e.g., energy inputs). This may be a particularly important shortcoming where associated emissions are significant. For example, silvopastoral agroforestry activities that intensify cattle management can lead to higher methane emissions from enteric fermentation. Where this is the case, it would be necessary to apply additional approaches or tools that account for the net climate impacts of the FLR activities undertaken to ensure complete accounting. In these cases, tools like FAO’s EX-Ante Carbon-balance Tool (EX-ACT) [70], which uses tier 1 defaults, could be used to estimate net emissions or removals.