Establishing and protecting forests contributes to natural climate solutions (NCS) by maintaining carbon stored in forest biomass and removing carbon dioxide (CO₂) from the atmosphere through growth (Pacala, 2004; Bastin et al., 2019; Hemes et al., 2021). However, forests have a lower albedo than alternative land-cover types, such as agriculture or grasslands, which increases the amount of radiation absorbed at the surface (Jackson et al., 2008; Anderson et al., 2011). Depending on the location, the lower albedo associated with a forested land-cover can offset the climate benefit provided by sequestered carbon (Betts, 2000; Bala et al., 2007; Bonan, 2008). This is especially pronounced at high latitudes where there is a large discrepancy in albedo between a tall, dark forest and a short crop or grassland covered by snow from the late fall through early spring. In temperate latitudes, the albedo difference is less pronounced because snow cover is less frequent (Betts, 2000). As a result, the albedo influence on the radiative energy balance can be similar to the carbon influence (O'Halloran et al., 2012). Land-cover change can also indirectly influence climate by altering evapotransipration and cloud cover (Bala et al., 2007; Ahlswede and Thomas, 2017). Simulated changes in cloud fields linked to land cover change can manifest at great distances downwind in the atmosphere, and are highly model-dependent, making attribution a challenge (Devaraju et al., 2018). Albedo radiative forcing exhibits a global climate sensitivity similar to carbon dioxide (Davin and de Noblet-Ducoudré, 2010). Therefore, to estimate the climate benefits of land-cover and land-management decisions in the temperate region, it is critical to quantify the combined impact of the change in carbon and albedo on global climate.

An alternative NCS to establishing forests for carbon absorption and storage is replacing fossil fuel energy sources with a dedicated bioenergy crop. The above-ground biomass of a bioenergy crop is burned after harvesting for energy and, if below-ground biomass increases carbon accumulated in the soil, there is a net removal of CO₂ from the atmosphere (Anderson-Teixeira and DeLucia, 2011). Crops and grasses, including perennial bioenergy crops such as Miscanthus x giganteus and switchgrass (Panicum virgatum), also have higher albedo values than forests (Bonan, 2008; Miller et al., 2016; Abraha et al., 2021). The direct effect of the higher albedo of perennial bioenergy crops on the atmospheric radiation balance can be large. One study found it to be six times larger than the offset of fossil fuel use (Georgescu et al., 2011). However, it is unknown how the net climate benefits of forests, with high productivity and potential for long-term carbon storage in wood, compare with high-albedo bioenergy crops.

Using observations of net carbon fluxes and albedo from eddy covariance (EC) sites, we compare the influence of managed pine stands to managed bioenergy crop fields on global climate using the radiative forcing metric (Hansen et al., 1997). We focus on comparing a pine plantation with a switchgrass field because they are managed ecosystems with overlapping natural ranges. In the Southeastern United States (SEUS), pine plantations are most commonly planted with loblolly pine, a wide-spread fast-growing native species that can be used for timber or bioenergy (Wear and Greis, 2013). A typical rotation length for a southern pine plantation is between 25 and 35 years (Stanturf et al., 2003; Fox et al., 2007; Wear and Greis, 2013). Switchgrass is a native, C4, perennial grass species that has emerged as a promising bioenergy feedstock with high expected yields for the SEUS (McLaughlin and Adams Kszos, 2005; Parrish and Fike, 2005; Wullschleger et al., 2010). Both species are reported to perform well on marginal lands and can even be inter-cropped (Albaugh et al., 2014). Unlike a pine plantation that is harvested at the end of the rotation with possible thinning at intermediate ages, switchgrass is typically harvested at least once per year. Therefore, including the entire cycle of establishment and harvest for both systems is critical for a robust comparison of radiative forcing.

Here we ask, (1) which ecosystem (pine plantation or bioenergy switchgrass crop) has a lower radiative forcing over a typical pine rotation and (2) how do assumptions about the fate of harvested carbon influence the conclusions? To address these questions, we combined new observations from central Virginia with published data from EC sites in the Eastern U.S. and Canada to model the relationship between age since establishment, net carbon storage (defined using net ecosystem exchange; NEE), and albedo. The age vs. albedo and NEE relationships were used with an established method for calculating relative forcing (Kirschbaum et al., 2011) over a 30-year time horizon that includes one complete harvest rotation of a southern pine plantation (a typical rotation length for the region) and annual harvests of switchgrass.

We describe our new eddy-covariance site and data processing, the synthesis of data from existing eddy-covariance sites, the development of the NEE and albedo age curves over a full rotation for the pine plantation and switchgrass fields, and the modeling of radiative forcing. Throughout, we use the atmospheric sign convention for NEE, whereby negative NEE corresponds to transfer of carbon from the atmosphere into the ecosystem.

We compared the climate mitigation potential of two ecosystems in the mid-latitudes of the eastern U.S using the radiative forcing metric. First, we found that the radiative forcing from albedo and carbon associated with the establishment of a pine plantation rather than a switchgrass field were on the same order of magnitude and that the pine had a slightly net positive (warming) radiative forcing when averaged over the 30-year analysis time horizon. This finding is the result of site-level observations of NEE and albedo from eddy-covariance studies across eastern North America, including observations from a newly established cluster of towers with co-located pine, switchgrass, and clear-cut systems. Therefore, (1) our results are not specific to a single location, (2) our results include the temporal dynamics of NEE and albedo following the establishment of the managed ecosystem, and (3) the carbon radiative forcing includes both the gains of carbon through photosynthesis and the loss of carbon through autotrophic and heterotrophic respiration.

Second, we found that the fate of harvested carbon influenced the direction the overall climate forcing, whereby the slight positive radiative forcing when only considering NEE and albedo measurement either increased in magnitude or changed sign when considering harvest fate. Specifically, we examined the end members for the potential fate of harvested carbon, where we assumed that all of the carbon from either loblolly pine or switchgrass was released back into the atmosphere or held in storage over the 30-year analysis horizon. The combined 30-year mean radiative forcing was negative (i.e., net cooling by the pine plantation) in only one of the scenarios, where switchgrass harvested carbon was returned to the atmosphere (e.g., bioenergy combustion) and pine harvested carbon was stored (e.g., in wood structures) (Figure 4C). Within that scenario 10.2% of the realizations had a net positive ΔF_net, resulting from uncertainty in the NEP and albedo relationships. Overall, this highlights the importance of managing the fate of harvested carbon after it is removed from the field. In addition, it illustrates the strength of the radiative forcing from a change in albedo, even in lower latitude ecosystems. Future work can (1) refine the literature values used for harvested carbon to evaluate whether management to increase productivity and harvested carbon can alter the balance between (ΔF_c) and (ΔF_α), (2) evaluate how the offset of fossil fuel combustion through the use of harvested carbon as bioenergy influences the net radiative forcing outcome and (3) investigate how ecosystems can be managed for albedo in addition to carbon climate benefits.

Our results may be sensitive to some assumptions embedded in the construction of the 30-year albedo and NEE time series. First, our study assumes that the switchgrass field had a 10-year rotation before it was replanted, resulting in three 10-year rotations to compare with a 30-year pine rotation. Ideally, we would explore the implications of longer periods between switchgrass plantings, as switchgrass can remain productive for over 20 years (Parrish and Fike, 2005). However, we could not find any published studies examining the carbon balance and albedo of switchgrass fields older than 8 years. Future work should prioritize observations of switchgrass fields over 10 years of age. Second, we assumed that the starting point for each time series was the pine clear cut. This starting point did not have a strong influence on the albedo dynamics of the switchgrass since the switchgrass and clearcut albedo values were similar. However, including the clearcut as the initial starting point for the switchgrass does add a loss of carbon that may not occur if using an alternative starting point, such as a cropland. We do not expect this to change our results substantially because the same alternative starting point would be also applied to the pine ecosystem in the comparative analysis. Finally, to build the albedo and NEE models we used data from other sites that did not have co-located measurements like the Virginia site. This is was necessary to generate NEE and albedo models across the full rotation of both systems as it would be challenging to build the models using only the Virginia site. However, since the pine ecosystem observations at the Virginia site fell along the NEE curve but the switchgrass was below (thus absorbed less carbon than the other sites), an analysis that only used data from the Virginia site would shift the net radiative forcing in the negative direction (favoring pine), but provide a less general result.

Our findings differ from those reported by two other studies that use similar approaches for calculating radiative forcing of pine forests in temperate latitudes. Each of these studies found that ΔF_c exceeded ΔF_α with a combined negative radiative forcing by pine ecosystems (Betts, 2000; Kirschbaum et al., 2011). There are four primary causes of the differences. First, our study used a productive bioenergy grass ecosystem with carbon accumulation as the baseline rather than a carbon-neutral baseline as used in Betts (2000) and Kirschbaum et al. (2011). As a result the difference in carbon accumulation at the end of our analysis was only 70 tC ha⁻¹ compared to 200 and 170 tC ha⁻¹ in Betts (2000) and Kirschbaum et al. (2011), respectively. Second, our study used time-integrated NEE flux measurements from eddy-covariance sites across eastern North America to determine the (ΔF_c) rather than observations of biomass change from a single site (Kirschbaum et al., 2011) or literature values of biomass and soil carbon (Betts, 2000). By using NEE to quantify carbon storage, our study also includes the loss of carbon through heterotrophic respiration. Consequently, our carbon storage rates were lower for the pine than switchgrass ecosystems early in succession, a difference that would be absent from an analysis that focused only on vegetation carbon. Third, the albedo contrast between the pine and grass systems was larger in our study (9%) compared to Betts (2000) (5%) and Kirschbaum et al. (2011) (7%). This creates a larger radiative forcing from albedo, particularly when compared to Betts (2000). Finally, each study differs in the time horizon of analysis. However, our 30-year horizon was less than the 50-year in Betts (2000) but longer than the 20-year horizon used in Kirschbaum et al. (2011), indicating that horizon is not the primary cause of the difference. Overall, the difference in carbon accumulations in both the pine and grass ecosystems is likely the most critical difference between our work and previous work.

Our analysis does not account for the indirect effects of land cover change such as a changes in evapotranspiration and biogenic volatile organic compounds (BVOC) emissions. Previous work using climate models has found that the indirect effect of evapotranspiration, via changes in planetary albedo from altered cloudiness, can offset the direct effects of a change in albedo (Bala et al., 2007; Spracklen et al., 2008; Pongratz et al., 2011; Ahlswede and Thomas, 2017). However, at the Virginia site, we found evapotranspiration was similar between the pine and switchgrass field, so changes in atmospheric water content and cloudiness should be minimal (Ahlswede, 2021). However, this result may not apply to all locations where decisions to establish pine or switchgrass occur, especially those with greater water stress. Therefore, additional work is necessary to examine the potential importance of evapotranspiration differences as a driver of indirect climate mitigation.

Furthermore, forests emit more BVOCs than crops and grasses, directly and indirectly influencing radiative forcing (Guenther et al., 2012; Unger, 2014). Forests in the SEUS are known for releasing large amounts of BVOCs, creating, for example, the famous Blue Ridge haze. The combination of BVOCs and anthropogenic oxidants in the SEUS creates high levels of secondary organic aerosol during the summer, which has a detectable cooling effect on the region (Goldstein et al., 2009). Since loblolly pine is a strong source (Stroud, 2005; Hu et al., 2015), and switchgrass a weaker source (Graus et al., 2013) of BVOCs, including the aerosol effect would potentially alter our findings by decreasing radiative forcing of the pine forest. Recent analyses in boreal forests found that biogenic aerosols contribute toward the net cooling effects of forest when weighed with carbon and albedo (Arvesen et al., 2018; Kalliokoski et al., 2020). Thus future analyses in the SEUS that weigh the net climate effects of land cover change may be more accurate if the influence of biogenic aerosols is included.

Decisions to use land-based climate solutions involve trade-offs when establishing different ecosystems for mitigation potential. Our study is a specific example that quantifies the trade-off between two managed ecosystem types in the eastern U.S. However, our conclusions likely apply to other pine plantation vs. cropland/grassland comparisons because the switchgrass albedo is typical of similar ecosystem types and our representation of uncertainty in both the NEE and albedo for the switchgrass was large enough to include variation across a wide region. Our results may not extend to a decision to establish a deciduous forest that has a higher average albedo than a pine plantation. Furthermore, forests and timber production have other indirect benefits that we did not account for here such as the substitution of timber-based building materials for emission-intensive products (O'Halloran and Bright, 2017; Kalliokoski et al., 2020). These indirect climate and ecosystem services may take higher precedence, given the small differences in the direct climate mitigation potential between the two ecosystems illustrated here. In conclusion, while the pine ecosystem absorbs more carbon than switchgrass, the difference in albedo was large enough to result in similar climate mitigation potential at the 30-year horizon between the two systems, whereby the direction and magnitude of radiative forcing depends on the fate of harvested carbon. This conclusion is a critical consideration if the expansion of either pine plantations or switchgrass is a component of a regional land-based climate mitigation strategy.

The original contributions presented in the study are included in the article/Supplementary Material, further inquiries can be directed to the corresponding author/s.

BA, TO'H, and RT conceived the analysis, conducted the measurements, and edited the article. BA performed the analyses and wrote the first draft of the article. All authors contributed to the article and approved the submitted version.

This work was supported by the USDA National Institute of Food and Agriculture (2017-68002-26612 and 2015-67003-23485) and McIntire Stennis project 1025966. We also received support from the Global Change Center and Fralin Life Sciences Institute at Virginia Tech and Clemson University Experiment Station.

The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.

All claims expressed in this article are solely those of the authors and do not necessarily represent those of their affiliated organizations, or those of the publisher, the editors and the reviewers. Any product that may be evaluated in this article, or claim that may be made by its manufacturer, is not guaranteed or endorsed by the publisher.