Rates and drivers of aboveground carbon accumulation in global monoculture plantation forests

Database representativeness of global monoculture plantations

Our database includes 4756 measurements of carbon in aboveground live tree biomass in monoculture plantations, collected from 829 distinct sites across the globe (Fig. 1). These sites were primarily in Asia (59%), Europe (16%), North America (15%), and South America (4%), with the remainder located in Oceania and Africa (~5%). The dataset included 240 species from 96 distinct genera of tree; however, 33 genera were poorly represented (n < 3). Across all observations, mean stand age was 21.3 years (median of 18 years, range of 1–98 years) and mean aboveground carbon stock was 47.0 MgCha⁻¹ (median of 36.4 Mg Cha⁻¹, range of 1.0–284.6 MgCha⁻¹).

The typology and spatial distribution of global plantation forests are poorly characterized due to (a) proprietary holdings of plantation forest data, and (b) difficulties in differentiating planted from natural forests. However, the 2020 Global Forest Resources Assessment (GFRA) of the United Nations Food and Agriculture Organization (FAO)¹⁶ suggest that China (84.7 Mha; 29%), the USA (27.5 Mha; 9%), Russia (18.8 Mha; 6%), Canada (18.2 Mha; 6%), Sweden (13.9 Mha; 5%) and India (13.2 Mha; 5%) hold the largest extents of global planted forests. Data from these countries were represented well within our database, with observations primarily within China (32%), the USA (12%), Russia (11%), Canada (2%), Sweden (1%), and India (3%). Data from Brazil (4%), the United Kingdom (5%), and Japan (5%) also accounted for significant portions of our database.

Our database captures the genera of tree crop species that dominate plantations globally. Comprehensive data on plantation species by location and percent of global plantation forests is difficult to obtain, but 2006 FAO data suggest that Pinus, Cunninghamia, Eucalyptus, Populus, Acacia, Larix, Picea, Tectona, Castanea and Quercus species account for approximately 70% of global planted area¹⁷. These same ten genera accounted for 81% of our observations and we parameterized nonlinear growth functions for seven of these genera with greater than 100 observations in our database.

Drivers of aboveground carbon accumulation rates

To explain variation in biomass accumulation, we tested major biological (genus, endemism, and plant traits), human (prior land use and management practices), and environmental (biome) drivers that may control growth (Supplementary Table 1). Almost all drivers explained significant variation in aboveground carbon accumulation rates (Table 1). The effects of individual genera on growth were highly variable, with an almost three-fold difference between the slowest growing (Robinia, t = −3.43, P = < 0.001) and fastest growing (Eucalyptus, t = 3.09, P = < 0.001) genus. Note that although Robinia is often considered a fast-growing species¹⁸, our Robinia observations derived from dryland forest areas of China. Aboveground carbon accumulation rates were 127% higher in plantations with exotic rather than endemic species (t = 7.26, P = < 0.001). Of the five plant traits tested (leaf type, leaf compoundness, leaf phenology, nitrogen fixation capacity, and wood density), leaf type, nitrogen fixation capacity, and wood density all significantly explained variation in carbon accumulation rates. Prior land use significantly affected the rate of aboveground carbon accumulation, with rates on formerly harvested land (t = −3.95, P = < 0.001) and pasture (t = −4.16, P = < 0.001) roughly 75 and 66% of rates on former croplands, respectively. Fertilizer use resulted in a relatively minor increase (4%) in the rate of aboveground carbon accumulation (t = 1.0, P = 0.3) relative to unfertilized stands, whereas thinning decreased the rate of carbon accumulation by 24% (t = −8.11, P = < 0.001). Finally, the effect of individual biomes on carbon accumulation was variable but ranged less than two-fold.

Next, to examine the relative importance of the drivers, we ran three “full models” on subsets of the data. Data on management practices were highly limiting, so we ran models with and without subsets of management practices. These models identified stand age, genus, prior land use, and ordinated plant trait data as statistically significant effects across all three models (Table 2). Ordinating the plant trait data produced two axes that explained approximately 53% of the trait data variance. The first axis primarily accounted for the leaf type, leaf phenology, and nitrogen fixation data, whereas the second axis primarily represented the wood density data (Table 2 and Supplementary Fig. 4). For the model that excluded management practices (Full Model 1, Table 2), we identified large positive effects for planting of exotic tree species as well as for most of the biomes (relative to boreal conifer forests). Conversely, we found strong negative effects on the rate of carbon accumulation for the coniferous genera (e.g., Picea and Cryptomeria relative to Acacia) as well as prior land uses/disturbances of fire, harvest, and pasture (relative to cropland). Unexpectedly, we found a non-significant effect for planting density in Full Model 2 (Table 2). We hypothesize that the effect of planting density may be masked due to size-density trade-offs, particularly for monoculture plantations in which stocking is likely to be optimized for maximal production (see Supplementary Discussion). Finally, for the model including the use of fertilizer (Full Model 3, Table 2), we found a significant positive effect of fertilizer use on growth (t = 3.3, P = 0.001). The individual effect of planting exotic species was not significant in this model (t = 0.4, P = 0.4), whereas planting exotics had a strong positive effect on rate of carbon accumulation in the other two models.

Nonlinear growth functions

We modeled aboveground carbon accumulation using the Chapman-Richards growth function, which is widely used and provides biologically meaningful growth parameters^19,20,21. Prior to fitting the function, we aggregated data by plant functional type (e.g., tropical broadleaf species) and genus. Relationships between aboveground carbon and stand age varied across plant functional types and genera (Figs. 2–5). Of the four plant functional types considered, tropical broadleaf forests had the fastest growth rate (k = 0.24), which was roughly 2.5 times that of the next fastest growing plant functional type (temperate broadleaf forests) (Fig. 2, 3). Compared to growth rates, asymptotic growth limits varied less and ranged from 73.9 (boreal needleleaf) to 121.0 (tropical needleleaf) MgCha⁻¹. Growth rates across the nine genera of tree differed even more than across plant functional types, with the fastest rate (Acacia, k = 0.33) roughly 9-fold greater than the slowest rate (Picea, k = 0.04) (Fig. 2, 3). Broadleaved genera commonly grown on short rotations (Acacia, Eucalyptus, and Populus) had rapid growth rates, roughly three times those of the coniferous genera (Fig. 2, 4). Lastly, Picea had the highest asymptotic aboveground carbon, which was three times that of Acacia, the genera with the lowest asymptotic aboveground carbon.

The results of the model validation procedure suggest large scatter in the data that is not fully captured by growth functions based solely on age. Normalized RMSE values ranged from 0.74 (Acacia) to 2.58 (Picea), suggesting that model uncertainty ranged from roughly 74% in the best case to 258% of mean aboveground carbon in the worst case. Despite the high uncertainty, the Chapman-Richards growth function is biologically appropriate and is an improvement over commonly used linear growth rates, which do not reflect how forests develop with time, and logarithmic growth functions, which often start with biologically impossible negative intercepts.

Annualized aboveground carbon accumulation rates

Although the Chapman-Richards function best captures stand development through time, practitioner and policy audiences frequently employ annual linear rates to inform reforestation planning. We therefore derived annualized aboveground carbon accumulation rates, which varied by genus of tree, plant functional type, and biome (Fig. 6). Broadleaved tropical genera (Eucalyptus and Acacia) had the highest accumulation rates (7.78 ± 0.20 MgCha⁻¹yr⁻¹ for Eucalyptus) and drove the high mean growth rates for the tropical broadleaf plant functional type and the tropical biomes (Fig. 6a). All four coniferous genera had roughly similar carbon accumulation rates, with the highest mean rate found for Cryptomeria (2.76 ± 0.17 MgCha⁻¹yr⁻¹). Across plant functional types, differences in aboveground carbon accumulation rates were less pronounced with the exception of tropical broadleaf plantations having the highest rate (6.25 ± 0.17 MgCha⁻¹yr⁻¹; Fig. 6b). Plantations in tropical & subtropical grasslands, savannas, & shrublands had growth rates (8.18 ± 0.43 MgCha⁻¹ yr⁻¹) roughly four times those of the slowest growing biome (temperate coniferous forests; 1.62 ± 0.14MgCha⁻¹yr⁻¹) (Fig. 6c).

Comparing our accumulation rates against those of naturally regenerating forests helps situate our findings but should be done with the understanding that planted and naturally regenerating forests are functionally different systems and carbon accumulation rates are only one metric of comparison¹⁰. Requena-Suarez and colleagues²² estimate carbon accumulation rates in younger (<20 years) naturally regenerating secondary forests of approximately 1.1–3.6 MgCha⁻¹yr⁻¹. Similarly, Cook-Patton and colleagues found annualized mean aboveground carbon accumulation rates to range from 0.1–6.0 Mg Cha⁻¹yr⁻¹ for natural regeneration²³. Using the same age classes and spatial aggregations (i.e., the FAO Ecozones), we determined a range of 0.9–8.2 MgCha⁻¹yr⁻¹ for aboveground carbon accumulation in young plantations. While the lower rates are similar, the high end of the accumulation rates in plantation forests is roughly 1.4–2.3 times greater than those of naturally regenerating forests.

The IPCC also reports mean annual carbon accumulation rates for plantations in Table 4.10 of the 2019 Refinement to the 2006 IPCC Guidelines for National Greenhouse Gas Inventories²⁴. However, data are limited (n = 86) and reported for only three genera: Eucalyptus (n = 22; range: 1.4–11.8 MgCha⁻¹yr⁻¹), Pinus (n = 17; range: 1.2–9.4 MgCha⁻¹yr⁻¹), and Tectona (n = 7; range: 0.9–7.1 MgCha⁻¹yr⁻¹). Our carbon accumulation estimates for these three genera align well with the IPCC data, and our database greatly expands (by a factor of ~60) the scope and breadth of carbon accumulation observations in global monoculture plantations.

Key predictors of above carbon accumulation

Genus, prior land use, and plant traits explained variation in aboveground carbon accumulation rates. The significance of genus and plant traits suggests that biological drivers primarily explain variation in plantation carbon accumulation at global scales. However, species choice can also be considered a key management decision (i.e., a human factor). Endemism of tree species, another decision-point in species selection, was found to be highly significant until we accounted for fertilizer use. This suggests that if nutrient limitations within plantations are lifted, endemic species may perform as well as exotic species; however, additional work is needed on this topic.

Prior land use, which may be a proxy for site productivity, was also a significant predictor of carbon accumulation. Plantations established on former croplands had higher carbon accumulation rates relative to former pastures, clear-cuts (of both prior rotations and native vegetation), or areas that had burned. This finding parallels the results of others, which show secondary forest growth to be higher on former croplands than pasture, and negative growth effects associated with frequency of fire²⁵. While our data do not elucidate a mechanism for this result, hypothesized factors may include soil fertility and/or competition from native vegetation. Our findings further indicate that monoculture plantations established on former crop or pasture lands may accumulate carbon faster than areas that were formerly forested but clear-cut. This finding has key implications for siting of future plantations, suggesting that clear-cutting of intact forests will not only adversely impact biodiversity, but may also result in lower growth rates than establishing plantations on former croplands or pastures.

Although our results suggest that management practices are important for explaining variation in growth, data limitations constrained our ability to examine the relative effects of management practices. Our interpretation is that while management practices are important for understanding variation in carbon accumulation in plantations, they are difficult to generalize and may therefore have limited utility in predicting carbon accumulation across broad scales.

Nonlinear accumulation of aboveground carbon in plantation forests

Our use of nonlinear growth functions provided biologically meaningful understanding of aboveground carbon accumulation in monoculture plantations. For example, we estimated rapid carbon accumulation rates for genera commonly grown on short rotations for pulpwood (e.g., Eucalyptus, Acacia, Populus) versus slower carbon accumulation rates for coniferous species that are commonly grown for timber (e.g., Picea, Cunninghamia, Pseudotsuga). Although this is an expected result, our parameterized growth functions are valuable for multiple reasons (Supplementary Tables 2, 3). First, they are significant improvements over annualized mean aboveground carbon accumulation rates, which do not account for how carbon accumulation varies with time. This distinction is critical for accurately assessing time-dependent carbon accumulation within the forest sector. For example, the delayed carbon accumulation rates that occur during stand initiation (e.g., ~5 years) represent half of the time window under which many public and private-sector programs have committed to reduce emissions (e.g., by 2030). Assumption of immediate and sustained rates of carbon accumulation within forestry projects may systematically overestimate climate benefits at decadal scales.

Second, plantation forestry is an attractive mitigation strategy for climate change given the economic benefits that can accrue to landowners, which may help incentivize restoration of forest cover²⁶. Accounting for temporal variation in growth is important for accurately modeling economic returns from plantation forests. For example, forest management decision-making such as identifying optimal rotation lengths (including for joint management of timber and carbon) employs nonlinear modeling of stand growth²⁷. Although a range of growth models exist (e.g., see the FORMODELs database), they tend to be developed for site to regional level scales. Our dataset and growth functions complement these models by providing an open-source and freely available framework for modeling carbon accumulation in the dominant monoculture plantation types seen across the globe.

Caveats and potential limitations

Several important caveats and limitations to our analyses exist, which we discuss here to facilitate appropriate use of our results. First, we lack clear understanding of the global distribution and type of monoculture plantations, making it challenging to assess whether our dataset represents the full range of variation in carbon accumulation seen across the globe. However, comparing our dataset to the most comprehensive overview of monocultures (i.e., the 2020 Global Forest Resources Assessment) suggests that most of our data are from those countries and regions that dominate global plantation forestry. Furthermore, grouping the data by biome, genus, and age class and examining the number of observations per site shows balance within the dataset (mean number of observations per site = 2, median = 1, max = 15, or less than 0.5% of all data; see Supplementary Information). To the best of our knowledge, our dataset is the most comprehensive compilation of carbon accumulation in monoculture plantations in the public domain.

Our asymptotic growth limit estimates (particularly the A parameter) should be interpreted with caution. The data we compiled are primarily from plantations managed on economically driven rotation ages and do not represent the maximum carbon stocks that could be achieved if these plantations were left to grow in perpetuity. Rather, the asymptotic values we present are fixed effects estimates for the population-level mean aboveground carbon stocks across all sites. Differences in site class are captured within our random effects estimates. For example, when accounting for these site level effects, the maximum predicted aboveground carbon value of our Pinus model was 164 MgAGCha⁻¹, roughly double that of our population level Pinus asymptote (see Supplementary Information). However, these random effects cannot be estimated for sites in which we have no training data, and fixed effects estimates of the population-level parameters are most appropriate for approximating aboveground carbon accumulation at new sites. Further, the potential risks of biased asymptotic growth limits can be mitigated by restricting use of our parameters to near decadal scales (e.g., ~ 20–30 years). Thus, our growth parameters are perhaps most relevant for the next few climate-critical decades.

The Chapman-Richards growth function is one of many within the literature and our parameter estimates may be sensitive to the selection of this function. A key criticism of the Chapman-Richards function is instability in the model parameters¹⁹. However, others have examined the risk of parameter instability when applying Chapman-Richards growth functions to aggregated datasets and have shown that, for the relatively constrained system of even-aged monoculture plantations, the Chapman-Richards growth function performs well²⁸. Rather, parameter instability in the asymptotic growth limit estimate is a greater issue for multi-species and multi-aged natural forests, as well as using the parameterized functions to model growth beyond the age range of the data. Further, others have concluded that the biological basis of growth functions have been overstated and users should focus instead on plausible growth functions that best meet their needs, as well as parameter estimation strategies²⁹. Here, we elected to use a theoretical growth function that is among the most widely used within forestry and empirically derived parameter estimates using a global dataset. By making our dataset publicly available, we anticipate that others will investigate the effects of alternative functional forms on the parameter estimates.

Lastly, we did not incorporate changes in belowground biomass and soil organic carbon into our analyses, but discussion of carbon flows to and from these pools is warranted. For belowground biomass, we excluded this pool given little data and constraints associated with synthesizing across studies with highly variable sampling protocols³⁰. Instead, we recommend using root:shoot ratios, which are available for particular forest types and regions³¹, as well as through spatially explicit maps at global scales³². These ratios allow carbon stocks in belowground biomass to be modeled as a function of the aboveground carbon values obtained from our models. Soil organic carbon, often the largest carbon pool in forested ecosystems, is impacted by forest management practices in varied ways³³. Generally speaking, only the forest floor and the upper soil horizons (e.g., O and A horizons) are expected to be impacted, with recent reviews estimating that <10% of total soil organic carbon stocks might be impacted³³. The magnitude of impact will depend upon the type and intensity of management practice (in particular, harvesting and site preparation), with more intense soil disturbance inducing greater soil losses.

Future assessments of the climate change mitigation potential of monoculture plantations would do well to incorporate these belowground carbon pools. Moreover, additional empirical data is needed to improve the representativeness and balance of the dataset we have compiled here. Specifically, we recommend that future studies focus on i) notably data-poor regions (e.g., Africa), ii) repeatedly measured plots, iii) better data sharing, and iv) standardized metrics (e.g., age, biomass, site index) to facilitate synthetic understand of carbon accumulation in plantations³⁴. In doing so, we can improve our understanding of how carbon accumulates in plantation forests across the globe and facilitate assessments of the full climate impacts of plantation forests (including both short-lived and long-lived harvested wood products). Lastly, although we did not directly account for mortality here, it will be important to consider how shifts in climate regimes and disturbance patterns present new risks to these systems, particularly for slow-growing plantation species that are investments over approximately century-scale time frames.

Potential applications of our findings

Monoculture plantations are only one pathway of many (e.g., natural regeneration, assisted natural regeneration, agroforestry, diverse plantations) for restoring forest cover^2,35. However, they represent the dominant approach underlying restoration commitments and thus merit careful consideration⁴. When managed sustainably and integrated into a broader landscape, plantation forestry can help reduce pressures on mixed-use forests (e.g., concessions) and meet demand for harvested wood products efficiently⁵. Substitution of sustainably produced timber for carbon intensive products such as concrete or steel can also reduce embodied carbon and expand carbon stocks beyond what exists on the landscape^36,37. However, emissions from harvesting activities, decomposing harvest residues, and short-lived wood products may reduce or negate these benefits³⁸. To accurately assess the climate benefits of plantations, enthusiasm for the high early growth rates we quantified here must be tempered by careful life cycle accounting. Furthermore, accounting for potential changes in the belowground biomass and soil carbon stocks will give a more holistic picture of the climate change mitigation potential of plantations.

Our results also have important implications for species choice. Exotic monocultures generally have poor biodiversity outcomes compared to other forest types³⁹. Concerning declines in biodiversity⁴⁰ coupled with global momentum to conserve biodiversity (e.g., the UN Decade on Ecosystem Restoration) encourage interventions to the biodiversity and climate crises in tandem. Using endemic species in monocultures instead of exotics could improve biodiversity outcomes, and the higher carbon accumulation rates that we observed in exotic monocultures disappeared after accounting for fertilizer use. This suggests that careful nutrient management in endemic plantations could help to optimize both climate and biodiversity benefits. However, further research on the trade-off between fertilizer use versus exotic species would be of high value, since fertilizer can negatively impact the local environment⁵ and result in additional climate emissions⁴¹.

We also identified traits linked to higher carbon accumulation rates, which can be used to select additional species not commonly used within plantation forestry. A handful of species and genera dominated our dataset and these were generally not endemic to the places where they were grown. Identifying additional endemic species to use in lieu of exotic species would improve the site-level biodiversity value of plantation forests⁴², and investing in plantation polycultures would help to resist trends towards global biotic homogenization⁴³. Biodiverse forests are also likely to be more resilient to disturbances such as pests or natural disasters, which is key for the durability of carbon sequestered in natural ecosystems, particularly under a changing climate⁴⁴. Moreover, species-rich forests may store more carbon than species-poor forests due to complementary resource use (niche complementarity)^45,46; however, competitive and mutualistic interactions among species may vary^47,48. Additional work is urgently needed to better understand how diversifying plantations could improve climate and biodiversity outcomes.

Source: Ecology - nature.com