Our study area encompasses all forestlands in Maryland and Pennsylvania. Maryland and Pennsylvania contain an estimated 0.99 and 6.72 million hectares of forestland that corresponds to 36.59% and 57.32% of total land area within each state, respectively (USDA Forest Service, 2019). Forestland (defined as land of at least 10% canopy cover by trees of any size and that will be naturally or artificially regenerated, Burrill et al., 2021) within both states is predominantly privately owned, with roughly one quarter managed by each state government and minor federal landholdings (Table 1; Figure 2) including a National Forest covering over 200,000 hectares in Pennsylvania. Hardwood forests dominate both states, especially the oak/forest type group in both states and maple/beech/birch forests in Pennsylvania (Table 2). However, Maryland also contains sizable areas of loblolly/shortleaf pine and oak/pine forests. Descriptions of each forest type group are listed in Appendix D of Burrill et al. (2021).

Maryland and Pennsylvania have large quantities of aging forests (i.e., reaching their commercial rotational age), with almost half of total forest area being over 80 years old (Figure 3). The current age distribution of forests in both states is a result of prior land use and management regimes throughout the early parts of the 20^th century that prioritized the clearing of forested landscapes for agriculture and human settlements, forest regrowth of depleted agricultural lands, management intensification, legacy effects of diameter-limit-cuts, and the disruption to historic natural disturbance regimes (Millers et al., 1989; Otto, 1989). Presently, forests in both states ensure a steady timber supply in the near term. However, with over 63.3% and 72.1% of forestlands in Maryland and Pennsylvania, respectively, being over 60 years old and nearing their commercial rotation ages, future timber supplies are dependent upon continued successful regeneration or thinning in older stands. Together, Maryland and Pennsylvania provide a unique study area to explore the role of climate-smart forestry in quantifying the C mitigation potential of forests and HWPs.

We used the Carbon Budget Model of the Canadian Forest Sector (CBM-CFS3; Kurz et al., 2009; Kull et al., 2019), an operational scale carbon model that uses spatially referenced inventory data and empirically derived growth-yield curves to simulate forest carbon dynamics through time utilizing guidelines and carbon pools established by the Intergovernmental Panel on Climate Change (IPCC). The model incorporates both human activities and natural disturbances to simulate forest C dynamics on annual timesteps. Process-based equations simulate carbon dynamics between soil, dead organic matter, and forest processes such as litterfall, while disturbance matrices represent the impacts of specific disturbance events, through the transfer of carbon between pools, carbon transfer to the forest product sector, and carbon emissions to the atmosphere (Kurz et al., 2009). The CBM-CFS3 has had wide applications within the United States (Dugan et al., 2018b, 2021), Canada (Kurz et al., 2013), and internationally (Pilli et al., 2013, 2017, 2022; Olguin et al., 2018; FERS, 2021), with thorough ground truthing (Shaw et al., 2014) and characterization of model uncertainty (Metsaranta et al., 2017). It is also the core model of Canada’s National Forest Carbon Monitoring, Accounting and Reporting System (Kurz and Apps, 2006; Kurz et al., 2018; ECCC, 2023).

Primary data inputs to CBM-CFS3 included a detailed forest inventory, growth-yield relations to estimate forest productivity, and estimates of harvest yields and intensity, land-use change, and natural disturbances. Inventory data are categorized by a series of forest classifiers defining relevant characteristics such as spatially referenced boundaries, ownership, forest type, site productivity, or reserve status. Allometric equations are used to predict tree volume-to-biomass relationships (Boudewyn et al., 2007). For this study, forest inventory, growth-yield curves, and harvest data were estimated from the USDA Forest Service Forest Inventory and Analysis (FIA) program, which we accessed through the FIA DataMart (USDA Forest Service, 2019) using the rFIA package (Stanke et al., 2020) in the R programming environment (R Core Team, 2020). rFIA enables data exploration and user-defined spatio-temporal queries and estimation of the FIA database (FIADB). Methodologies derived from Bechtold and Patterson (2005) and Pugh et al. (2018) were used to estimate each state’s forest inventory by a predetermined list of classifiers (Supplementary Table S1). Natural disturbance history was estimated from both the FIADB and LANDFIRE (USGS, 2016) datasets to better constrain initial belowground and soil carbon parameters during what the modeling framework refers to as the model spin-up period (Kurz et al., 2009). Estimates of merchantable volume and corresponding biomass from FIADB were used to calibrate the model’s allometric volume-to-biomass assumptions to match forest type groups and growth conditions in Maryland and Pennsylvania.

To estimate empirical growth-yield curves, a Gompertz growth function (Eq. 1) was used to model plot-level relationships between merchantable timber volume and average stand age. This growth model is a common exponential function used to estimate various forest attributes while not assuming symmetry within the curve unlike other logistic functions (Fekedulegn et al., 1999). The Gompertz growth curve takes the following form:

where, α is the upper asymptote, β is the growth displacement, and k is the growth rate or slope at time t.

Due to limitations of using stand age as a predictor to estimate merchantable volume in uneven-aged stands following harvest events, a modified growth-yield table was derived by utilizing methods developed and validated in Pilli et al. (2013) for calculating annual growth increments of unevenly aged systems following harvests. This is derived from the assumption that younger cohorts of trees move toward canopy dominance via canopy gap dynamics inherent to uneven-aged managed forests common in this region of the US. This methodology outlines that, following the removal of a specific proportion of merchantable volume, the growth curve continues to approach the same asymptote established in the unmodified growth-yield table. Using merchantable timber volume as a function of stand age and productivity class, MAI of each productivity curve can be modified using a basic exponential function (Supplementary Equation S2) to then reapproach the original asymptote.

Harvest removals were estimated as an average annual removal of merchantable timber in cubic feet between 2007 and 2019, converted to metric tons of carbon using methodologies and specific gravities reported by Smith et al., 2006. To assign a harvest type and intensity to each record of volumetric removal, stand age at the time of removal was calculated by taking the mid-point average between time t₁ and t₂ (Bechtold and Patterson, 2005) where t₁ is the year the unharvested stand was measured and t₂ is the repeat interval year measurement post-harvest. In collaboration with state partners, harvest type and intensity were determined heuristically for each forest type based upon state-level management documentation, peer-reviewed literature, and expert input. A complete list of harvest types and intensities prescribed to each forest type group as a product of stand age can be found in Supplementary Table S3.

Longer-term averages from 2007 to 2019 were used to estimate annual area targets for all LUC and natural disturbance events including wind, fire, disease, and insects. Averaged annual LUC rates by ownership and forest type group were derived by overlaying a geospatial forestland ownership dataset (Sass et al., 2020), the Protected Areas Database of the U.S. (PAD-US), a national geodatabase of protected areas (USGS, 2018), and the National Land Cover Database (NLCD), a remotely-sensed data product used to characterize land cover and land cover change (Wickham et al., 2021). Wind disturbance events were calculated using the LANDFIRE Historic Disturbance dataset (USGS, 2016), a remotely-sensed data product provided by the USGS that estimates annual disturbance events. Annual averages for wildfire disturbances were derived from the LANDFIRE Historic Disturbance dataset (USGS, 2016) and validated through annual reports from the National Interagency Fire Center (NIFC). Annual prescribed fire acres were estimated from reports provided by the Maryland DNR Forest Service and Pennsylvania DCNR Bureau of Forestry and scaled to represent treatments on forestlands only. Annual acreages of insect and disease disturbance were derived from National Insect and Disease Detection Survey (USDA Forest Service, 2020), a spatial data product produced by USDA that collects and reports data on forests insects, diseases, and other disturbances. For more information on all input and activity data, see Supplementary material 1.2. A complete list of BAU parameters can be found in Supplementary Table S2.

To calculate and assess carbon stored by and GHGs emitted from forest products, we developed separate HWP models (CBM-HWP-MD for Maryland and CBM-HWP-PA for Pennsylvania) using the Abstract Network Simulation Engine (ANSE) model framework. ANSE is a carbon estimation tool developed by the Canadian Forest Service and employed in annual reporting of Canada’s national GHG inventory. Both the CBM-HWP-MD and CBM-HWP-PA models facilitate modeling, tracking, and calculation of embodied carbon stored by and emitted from HWPs. Furthermore, both models were adapted to and parameterized with US-based data and estimates and region-specific data where available.

Specific disturbance actions implemented in the CBM-CFS3 (particularly, though not exclusively, harvest events) transfer carbon in the form of metric tons of carbon directly into the CBM-HWP-MD and CBM-HWP-PA models. This transferred carbon is then partitioned among various wood product streams (Supplementary Figure S4), based on current practices in the forest products sector. Carbon is either exported in the form of roundwood exports or produced commodities, retained for domestic commodity use, or immediate use domestically for mill residues, energy, and additional commodity production. Maryland and Pennsylvania have distinct product ratios split by wood type (Supplementary Figure S5). Each commodity stream has a corresponding half-life that determines the in-use residence time of carbon in a specific product stream before being allocated to an end-of-life path (i.e., recycled, burned for energy, or sent to a landfill). Landfills also have corresponding half-lives by landfill type and location, and from which carbon eventually decomposes and is emitted to the atmosphere as either carbon dioxide (CO₂) or methane (CH₄). Furthermore, each state specific model tracks inherited wood products, or products in-use, prior to model simulation starting from 1950 onwards.

When HWPs are substituted for alternative, more emissions-intensive products (e.g., concrete, steel), that change in production is assumed to have associated displaced emissions, or substitution benefit (Geng et al., 2017). For example, when additional wood products are manufactured relative to BAU, we assume those additional products will be used in place of other non-wood materials, resulting in a corresponding substitution benefit and reduction in GHG emissions (Leskinen et al., 2018). Substitution benefits are applied only to saw logs, composite panels, and bioenergy products. Inversely, a decrease in wood product manufacturing results in increased emissions (or negative substitution benefits; Myllyviita et al., 2021).

Using methods from Smyth et al. (2017), we calculated and applied state-specific displacement factors for softwood and hardwood saw logs and composite panels. The calculated displacement factors rely on LCA data for the emissions associated with the extraction, raw material transport, and manufacture of both the HWPs and the assumed alternative materials and Smyth et al. (2017) data on relative product weights in different end use products. As an additional component of the displacement factor calculations, we used state-level RPA mill residue and primary timber product data (core tables 10 and 5) and Howard et al. (2017) data on wood product market share percentages in the US to calculate weighted lumber and composite panel production and use for each state. Maryland had displacement factors of 2.045 (softwood saw logs), 2.681 (hardwood saw logs), 2.682 (softwood composite panels), and 1.972 (hardwood composite panels). Pennsylvania had displacement factors of 2.032 (softwood saw logs), 2.692 (hardwood sawlogs), 2.682 (softwood composite panels), and 1.932 (hardwood composite panels). We applied a conservative and linearly decreasing displacement factor for bioenergy for both states, starting at 0.47 in 2022 (Smyth et al., 2017) and reaching zero by 2040 to account for Maryland’s net-zero target by 2045 and the Biden Administration’s goal of reach a carbon pollution-free power sector by 2035. Calculations reflect current wood and alternative product extraction, raw material transport, and manufacture; they do not assume increased carbon efficiencies in wood (or alternative product) production throughout the modeling period, nor the adoption of carbon capture and storage, as these technologies are not currently widely adopted within the region.

For scenarios that result in lower levels of harvest relative to BAU, we apply a leakage factor to estimate an assumed increase in harvest activities occurring outside the study area compensating for a decrease in timber supply (Murray et al., 2004). We assume demand for wood products remains constant despite reductions in harvest and assume a portion of that demand will be met via imports of additional wood (i.e., leakage). Further, we assume all remaining product demand not met by wood imports will be met by an increased use of non-wood materials in place of wood. Determination of leakage rates are dependent upon the degrees of assumed regional collaboration with estimates ranging from 63.9% (Gan and McCarl, 2007) to 84.4% (Wear and Murray, 2004). In this study we assumed a leakage factor of 63.9% due to the multi-state nature of the project, meaning that the remaining 36.1% of reduced harvest rates are subject to additional emissions from non-wood materials. Leakage is only assumed to result from reduced in-state harvest whereas increased in-state harvests are assumed to result in increased wood utilization rather than reductions in out-of-state harvest.

State-specific trade and commodity data from Resource Planning Act (RPA) assessments (USDA Forest Service, 2021), US Commodity Flow Surveys (US Department of Transportation, Bureau of Transportation Statistics, US Department of Commerce, and US Census Bureau, 2020), US International Trade Commission export data (US International Trade Commission, 2021), and published peer-reviewed data (Howard and Liang, 2019) when available, or US averages from the same sources, were used to adapt and parameterize both HWP models. FAOSTAT data (FAO, 2021) were utilized to determine the commodity distributions of exported roundwood. Softwood products were parameterized and modeled separately from hardwood products, as the two wood types differ in exports and commodities produced as well as their associated product half-lives and displacement (Dymond, 2012; Howard et al., 2017). Published data were used to calculate softwood- and hardwood-specific half-lives for Maryland and Pennsylvania sawn wood and veneer products, while we relied on literature estimates for other products (Smith et al., 2006; Skog, 2008). To calculate substitution benefits, we coupled region-specific data (USDA Forest Service, 2021), US consumption rates (Howard et al., 2017), product weights (Smyth et al., 2017), and LCA data (Bala et al., 2010; Dylewski and Adamczyk, 2013; Meil and Bushi, 2013; Puettmann et al., 2016; Puettmann and Salazar, 2018, 2019; Hubbard et al., 2020; Puettmann, 2020), following methods developed by Smyth et al. (2017). Landfill CO₂ and CH₄ emissions rely on IPCC defaults for methane generation (k) and landfill half-lives for wet, temperate climates (Pingoud et al., 2006). See Supplementary material Appendix S1 for more details on substitution and leakage calculation methods and Appendix S2 and Appendix S3 for site-level modeling parameters.

The management, disturbance, and wood utilization data collected as described above were used to parameterize our models for the historical period of 2007–2019. Simulations run for a total of 93 years in which the first 13 years use observed activity data and the remaining 80 years use scenario-based activity data to cover both the historical period (2007–2019) and forward-looking projections (2020–2100), providing seamless comparisons between past and future trends. Simulation projections use the most recent decadal (2007–2019) averages for harvest yields, natural disturbances, LUC, and wood use determined by the activity data above. Alternative management scenarios were then developed by changing selected parameters from the BAU scenario to represent possible alternate forest management practices or priorities. All other disturbance and climate data were assumed to be constant in all scenarios except for specific climate change scenarios. Scenarios were grouped for analysis into six broad categories representing similar management practices or objectives: (1) altered rotations; (2) tree planting; (3) maintain forest health and regeneration; (4) climate change impact; (5) no harvest activities; and (6) bioenergy (Table 3). A portfolio scenario was also constructed by combining multiple individual scenarios to represent a concurrent suite of climate-smart forestry activities.

BAU results are reported in million metric tons of C (MMT C) or million metric tons of CO₂ equivalency (MMT CO₂e), using 100-year global warming potentials. Annual Net Biome Productivity (NBP) is calculated as gross ecosystem productivity minus C losses from respiration, decomposition, and disturbance (including harvest and fire), where negative values represent a net carbon sink and positive values represent a net carbon source. To assess the relative gains in mitigation potential from alternative management and wood utilization practices, scenario results below are discussed in both actual values and in standardized terms relative to the BAU scenario. This standardized mitigation potential (henceforth, “mitigation”) is calculated by subtracting the annual net C balance of the BAU from the annual net C balance of each scenario, comparing net GHG emissions (CO₂, CH₄, CO, N₂O) for each modeled scenario with net GHG emissions from the BAU simulation. This approach isolates the effect of each management practice to understand and assess the differential impact between an alternative management approach and BAU. Net emissions (or the net C balance) include (1) emissions resulting from disturbances (harvest, LUC, or natural disturbance) or decay processes in the forest ecosystem, (2) emissions resulting from HWP use, mill efficiency, and decomposition, (3) emissions results from the substitution of non-wood materials and bioenergy for other forms of energy derivation, and (4) leakage from reduced harvest rates.

A total of 15 management scenarios were modeled for Maryland and 13 management scenarios for Pennsylvania, along with two climate change impact scenarios and two bioenergy scenarios for both states (Table 3). All scenarios were developed within a participatory framework alongside state partners to represent a variety of plausible or theoretical management activities that might be implemented to respond to forest health concerns, shifting forest management priorities, or policy incentives. Partners were asked to identify scenarios of interest for all land ownership types, as well as the scale and duration of each management practice. The development of scenarios and modeling results and analysis were conducted in a close iterative process with our state partners to validate and infer any model results and parameters. More detailed descriptions of scenarios, methodologies, data acquisitions, and assumptions for each can be found in Supplementary material Appendix S1.

Our results are first partitioned between the BAU forest ecosystem C balance followed by C dynamics in the forest products sector. We then analyze the two linked assessments prior to assessing and quantifying impacts of alternative management scenarios as counterfactuals to the BAU.

Several scenarios exhibited similar trends in both Maryland and Pennsylvania. The portfolio and some tree planting scenarios outperformed all other scenarios in terms of sequestering additional C (Figure 7). Most scenarios focusing on maintaining forest health, structure, and regeneration showed increased mitigation potential relative to BAU in both the forest ecosystem and HWPs. Altered rotations exhibited trade-offs in mitigation potential, increasing carbon storage on forestlands and (temporarily) decreasing carbon in HWP with a subsequent decline in substitution benefits until forest stands equalized under longer rotation lengths. Not all scenarios yielded additional mitigation potential; certain practices such as timber stand improvements treatments reduced C storage and sequestration relative to BAU as anticipated due to increased thinning and prescribed fire practices, which released C from the forest ecosystem in favor of forest health outcomes. However, the additional pulpwood removed comprised predominantly of pulp and paper products provided minimal additional substitution benefits due to the short half-life and quick retirement of such products.

Results from the BAU scenario suggest that forests in Maryland and Pennsylvania are projected to weaken as a net C sink throughout the 21st century and, in Pennsylvania, ultimately become a net C source. Several alternative management scenarios have the potential for additional mitigation benefits, including increasing or maintaining forest area, increasing the ability of forests to regenerate, encouraging sustainable harvesting practices, altering rotations, supporting sustainable wood utilization, and preparing for future climate change impacts (Table 5).

Timber harvest remains the most extensive disturbance across eastern US forestlands in terms of both area and C impacts (Williams et al., 2016; Oswalt et al., 2019). Beginning in the 20th century and continued into the 21st century, the forest C sink primarily remained strong due to forest recovery or regrowth from previous harvests beginning in the mid-19th century (Birdsey et al., 2006; Woodall et al., 2015). Thus, the projected weakening trend in the forest C sink is driven in part by older forests becoming less productive with age (Ryan et al., 2010), which is well documented across US forests (Ryan et al., 1997; Binkley et al., 2002; Powers et al., 2011; Dugan et al., 2017; Curtis and Gough, 2018). However, our results strongly suggest that decisions around forest management, wood use, and policy can significantly alter and potentially improve the role of forests as C sinks in the future. Specifically, our results reinforce that forests can continue to provide a sustainable amount of wood products without hampering the mitigative potential of forests.

BAU results emphasize an important tradeoff between higher carbon stocks in older forests and diminishing rates of C sequestration as forest age (Harmon, 2001; Coomes et al., 2012; Yuan et al., 2019). However, the portfolio scenario supports that management can optimize the benefits received from a host of ecosystem services (D’Amato et al., 2011; Bradford and D’Amato, 2012; Bradford et al., 2013; Catanzaro and D’Amato, 2019; Littlefield and D’Amato, 2022) such as C sequestration in forests recovering from disturbance (Zhao et al., 2022), C storage in older forests (McGarvey et al., 2015; Curtis and Gough, 2018), and a steady stream of timber that stores additional carbon in HWP (Skog, 2008; McKinley et al., 2011) and drive the forest bioeconomy (Sohngen et al., 1999; Puddister et al., 2011; Antikainen et al., 2017). The cumulative mitigation potential realized by the portfolio scenario demonstrates the power of coordinated climate-smart forestry action, showing that the C benefits accumulated from certain practices can offset potential C losses from other management activities. The portfolio scenario results also suggest more complex interactions between management practices because contributions of some mitigation actions are not additive.

Results from the no harvest activities scenario emphasize the importance of quantifying emissions from the forest products sector. This scenario quickly increases sink strength in the forest ecosystem in both states but is accompanied by a substantial increase in emissions compared to BAU, driven by net emissions from HWPs, substitution, and leakage. Ultimately, the no harvest activities scenario performs second or third lowest in both states. This scenario signifies an important trade-off between immediate sequestration potential and future sequestration potential, ignoring the larger effects on local- to regional-economies (Antikainen et al., 2017) and potential carbon loss from increased disturbance risk associated with older forests (Kurz et al., 2008). As demonstrated with increased substitution emissions, the lack of timber supply forces other sectors (particularly construction) to increase use of non-wood materials such as concrete and steel, which have greater embodied emissions, significantly impeding GHG emission targets (Geng et al., 2017; Hildebrandt et al., 2017).

Balancing trade-offs and opportunities by managing forest landscapes as a heterogeneous patchwork provides a potential pathway to ensure forest ecosystems remain future carbon sinks (Wear and Coulston, 2015) while optimizing the co-benefits received from forests varying by age, structural complexity, and management goals (Nevins et al., 2021). These co-benefits include C storage and sequestration, habitat creation, reduced risk of disturbance, and enhanced resilience and adaptive capacity from both stand- to landscape-level management goals (Littlefield and D’Amato, 2022) as well as optimizing both short- and long-term mitigation potential (Petersson et al., 2022; Schulze et al., 2022). Specific to the context of Maryland and Pennsylvania, particular emphasis should be placed on forest regeneration following disturbances to ensure a strong future C sink, demonstrated by the scenarios in the maintaining forest health, structure, and regeneration category. The rate at which forests recover from disturbance is a strong driver of future forest C sink strength (Curtis and Gough, 2018; Peichl et al., 2022) in addition to providing valuable timber resources.

Protecting and encouraging natural regeneration is another key strategy for promoting forest diversity, in terms of age classes, species composition, and stand structure. Restocking efforts can promote a diversity of locally adapted species that are competitive under future climate conditions which, vary in mature size, shade tolerance, and canopy dominance as well as wood densities and hydrological strategies increasing forest resilience (Folke et al., 2004; Seidl et al., 2014; Anderegg et al., 2018). In addition to promising C gains in both the restocking and control deer browse scenarios, these efforts to promote increased biodiversity and structural complexity of forests also reduce disease and insect risk, further protecting future C storage and sequestration potential (Seidl et al., 2016, 2017). Ecosystem sequestration increases in the restocking scenario in the latter half of the century driven by maturing stands that have been restocked; however, the impact is not particularly significant due to a modest amount of forest stands targeted as restocking. Although control deer browse and restocking scenarios may be cost prohibitive for some, their expected C gains support these practices as a key tool for ensuring future forest productivity (Keefe et al., 2012; Buma and Wessman, 2013; Catanzaro and D’Amato, 2019; Dey et al., 2019).

The majority of forestland in the eastern US is privately held, with high potential for climate mitigation due to high percentages of working forestlands (Williams et al., 2016; Oswalt et al., 2019). Forest management planning and behavior on private lands to intentionally incorporate climate-smart management actions have strong potential for enhancing C mitigation, as demonstrated by the reduced DLC, control deer browse, and restocking results, while providing additional adaptation benefits (Köhl et al., 2010; Lafond et al., 2014; Lesser et al., 2019). However, significant uncertainties remain around forest policies that discourage harvesting or encourage lengthening rotation ages and their potential long-term effect on climate mitigation potential (Huntington et al., 2019). Ideally, policies should target not just management but the entire forest sector creating more flexibility to enhance possible forest outcomes (Sjolie et al., 2013). Potential policy effects on future investment in forest resources signals an important tradeoff where forest policy strategies could potentially incentivize continued investment in climate-smart activities or inversely, disincentive future investments (Pohjola et al., 2018). The relevancy and applications of these policies could affect both the rate and magnitude of future forest investments as well as discourage future harvests from extending rotations (Sohngen and Brown, 2011) potentially impeding future investments (Roberge et al., 2016). Where, as demonstrated by the afforestation, reduced deforestation, and silvopasture scenarios, increasing or maintaining forest extent, especially on private forestlands, remains one of the most cost effective and viable options to increasing C stocks and sequestration potential at the regional scale in the eastern US (Catanzaro and D’Amato, 2019; Doelman et al., 2020). This not only strengthens forest C sinks but has the additional benefit of reducing the timber demand on other forestlands, freeing up forests for a variety of both passive and active management approaches. Ultimately, this protects forestland by ensuring sufficient forests remain in operation to meet current and future wood demands while allowing landowners and managers to set aside areas of high priority for conservation (Pirard et al., 2016). The current decline in forest area only increases pressures on forests to meeting growing timber needs in addition to other essential supporting and regulating services (Jurgensen et al., 2014).

The allocation of forestland to various management goals and practices, such as bifurcating forests into ecological reserves and high productivity plantations based on site-specific considerations, provides a potential way to balance ecosystem service trade-offs and maximize the co-benefits humans rely on from forests (Bradford and D’Amato, 2012; Catanzaro and D’Amato, 2019). However, rarely is the delineation of forestlands into a binary paradigm of reserved lands vs. working lands easy or necessarily appropriate, considering the complex relationships between forest owner decision making, active and passive management strategies, legacies of disturbance, ecological complexity, variation in the vulnerability of sites and species to the projected impacts of climate change, and related policy implications. For example, areas of high risk to carbon loss from disturbance can be targeted for timber stand improvement practices to ensure carbon storage and stability into the future (Bachelet et al., 2015). Alternatively, areas of low forest productivity and regeneration can be targeted for restocking through enrichment plantings to boost both sequestration rates and future timber supply (Ontl et al., 2020) in part reducing timber demand of forests of higher ecological complexity and biodiversity (Köhl et al., 2020).

Balancing the trade-off between the current C storage and future C sequestration in the ecosystem relies primarily on relationships between growth, harvest removals, and recovery from disturbance, which are dependent upon both current and past legacies of management on forest structure and species composition driven by landowner goals (Seidl et al., 2014). Therefore, there is no one-size-fits-all approach to forest management for meeting carbon-related goals (Catanzaro and D’Amato, 2019). The preferred approach for any forest will vary on a case-by-case basis and be determined by climate implications, site-specific considerations including risks from natural disturbances, and forest owner values (Fahey et al., 2010; Catanzaro and D’Amato, 2019). At landscape scales, a full suite of climate-smart activities can be implemented simultaneously, as in the portfolio scenario, to optimize the mitigation potential of management activities under future climate uncertainties (Millar et al., 2007) and promote carbon and other ecosystem service benefits.

The climate change impact scenarios likely do not capture all the nuanced ways in which climate change may affect forest productivity, mortality, and demographic processes. Evidence of increased tree growth rates due to climate change is limited (Dow et al., 2022). Although the climate change growth scenario showed slight increases in growth under future climate conditions, it resulted in minimal C gains. Some studies suggest that climate change will increase drought-related mortality in forests and may ultimately decrease productivity, counteracting or exceeding any potential growth benefits observed within our prescriptive approach to future growth (Kurz et al., 2008; Allen et al., 2010, 2015). The climate change disturbance scenario emphasizes that increases in the severity and extent of natural disturbance drastically decreases ecosystem C stocks. The climate change impact scenarios suggest that increases in disturbance intensity and severity could counteract any projected increases in future growth caused by climate change resulting from carbon fertilization, warming temperatures, or longer growing seasons (Kurz et al., 2008). Additionally, evidence suggests that risk from wildfire, insect and pathogen, and drought disturbances pose a severe future threat to carbons storage (Allen et al., 2015; Jolly et al., 2015; Brando et al., 2019; Walker et al., 2019; Asaro et al., 2023). Therefore, focusing exclusively on C sequestration and storage cannot supplant other management goals such as enhancing forest health and resiliency without potential catastrophic repercussions for future ecosystem resilience. To do so would be contradictory to climate-smart forestry principles, which emphasize promoting future adaptation and resilience, and reducing catastrophic disturbance risk, while meeting current human needs. For example, the TSI scenario projected a weakening of the C sink in the near term relative to BAU due to increased removals and emissions from thinning and prescribed burns, but solely considering the C impacts does not fully account other co-benefits provided by these practices. Forest resilience treatments (which also consist of thinning and prescribed fire) can reduce future wildfire risk (Shinneman et al., 2012), increase forest restoration and health benefits (Brown et al., 2004), improve habitat creation (Tingley et al., 2018), and achieve desired species mixes or stand structural diversity (Brockway and Lewis, 1997). Resilience treatments can reduce catastrophic disturbance risk by increasing a forest’s adaptive capacity or an ecosystem’s ability to tolerate large-scale disturbances (Nicotra et al., 2015; Beever et al., 2016; Rogers et al., 2017) and future climate change induced decreases in resilience (Clark et al., 1999; Zhu et al., 2018; Hill et al., 2023). Further increases in wood utilization and efficiency of small diameter stems removed during resilience treatments can generate additional climate benefit by storing additional carbon and boosting substitution benefits (Mohammad, 2023).

Increased wood utilization has the potential co-benefit of slowing forestland conversion by supporting a more robust sustainable economy for wood products, whereas stopping harvest activities provides inverse incentives for landowners to potentially convert land (Sohngen et al., 1999). However, establishing safeguards to minimize negative externalities to forests of high biological value and biodiversity is advisable, understanding the complex interactions of multi-faceted management goals across the landscape (Clay and Cooper, 2022). Wood as a basis for a sustainable bioeconomy provides additional livelihood support for rural landowners (Puddister et al., 2011), incentivizing sustainable management in terms of C emissions and removals (Köhl et al., 2020), and a transition toward a sustainable bioeconomy supported by renewable resources instead of fossil fuels (Rockstrom et al., 2009). There is a significant trade-off between avoided GHG emissions from wood utilization and substitution and the near-term reduction in forest C sink from harvest (Soimakallio et al., 2016), but assessing net C balance within forest boundaries fails to acknowledge the key linkages between forest C dynamics, forest product-based C pools, and the important benefits of transitioning away from fossil fuel-based materials (Köhl et al., 2020).

Innovative wood utilization only serves to further boost the carbon mitigation potential of forests (McKinley et al., 2011). New technologies and uses increasing product half-lives bolster the C storage and substitution benefits of forest products, increasing displaced emissions (and so reducing embodied carbon emissions) from other more emissions-intensive materials or energy sources (Skog, 2008; McKinley et al., 2011; Xie et al., 2023). The construction industry presents a prime opportunity to maximize both the C storage and substitution benefits of HWPs, due to large emissions associated with concrete and steel production (Churkina et al., 2020; Zhang et al., 2020). Additionally, proper maintenance of landfills to ensure anaerobic conditions that slow decay of solid wood products offers additional opportunities to accumulate C stocks (Barlaz, 1998; Geng et al., 2017). Improved landfill management could involve the reduction of methane emission by diverting wood waste away from the landfill or capturing landfill methane emissions (Moreau et al., 2023). Lastly, increasing mill efficiency and effective wood recovery rates represent additional opportunities to boost the amount of C stored in forest products by utilizing more material (and the C it contains) in longer-lived products and reducing the amount of mill residue that would otherwise likely be emitted quickly (Lippke et al., 2011, 2012); over 90% of mill residue in both states goes toward short-lived pulp and fuel products, or is unused. However, to optimize both sequestration and storage benefits, forest owners and managers should consider balancing current forest growth rates with removals to not further exacerbate forest C sink declines.

To meet current ecological, economic, and climate goals, no single strategy is right for all landowners and forest types. Forest landscapes need to be managed under a variety of climate-smart approaches, tailored to site-specific conditions and management goals. This study demonstrates that a focus on existing priorities of forest health and stewardship can constitute part of an effective mitigation strategy (Millar et al., 2007), without hindering future demands on timber supply. While simultaneous increasing C sequestration potential to reduce atmospheric GHG levels in-service of 2030 emissions targets (den Elzen et al., 2022).

Significant mitigation opportunities exist through protecting existing forestland from permanent loss, i.e., keeping forests as forest, and planting trees in ecologically appropriate areas to increase forest acreage, two practices already aligned with policy priorities in Maryland and Pennsylvania (MDE, 2021; PDEP, 2021). Emphasis on restoration of degraded or poorly functioning forestlands along with implementing sustainable management practices would provide further C benefits through maintaining both forest health and productivity. More efficient wood utilization and creation of longer-lived wood products can increase C storage in HWP. Policies promoting bioenergy derivation from waste materials provide another potential climate mitigation strategy, especially if waste materials are diverted from landfill streams. Emerging carbon capture and storage technologies have added potential to couple with bioenergy, further increasing substitution benefits shifting production away from other more carbon intensive energy sources (Shahbaz et al., 2021; Petersson et al., 2022). Additionally, decreasing logging operation emissions, while not explicitly modeled in this study, would further enhance statewide carbon balances converging toward net-zero emissions targets. Coupling a reduction of fossil fuel use in management activities along with implementing best practices to reduce the impact of logging only further increases the potential of carbon benefits (Fahey et al., 2010). This does not preclude the impact that forest management activities focused on resiliency and adaptive capacity can have in altering net C balances. Further, the increased responsible use of forest products may result in additional climate benefits, ensuring net benefits for environment and society (Clay and Cooper, 2022).

Meeting global climate goals benefits from linking results of studies such as this to various policy levers and incentives, as well as communications and outreach. Tax programs, stewardship and management plans, voluntary offset projects, and cost-share programs have the potential to incentivize desired forest management behaviors, ensuring healthy and vigorous forests. Continued enrollment of forestland in sustainable forest management plans along with other policy vehicles may help to ensure the future viability of the forest C sink (Fahey et al., 2010). The incorporation of new data will help decisionmakers continually refine GHG emission targets and goals and better understand trade-offs, risks, and uncertainties associated with specific climate mitigation activities. Likewise, forest managers will benefit from incorporating the best possible information to inform flexible data-driven decision-making regarding forest management (Novick et al., 2022) to enhance the role of forests to meet GHG emission targets. The creation of flexible decision-making processes guided by the best available information may allow for an incorporation of refined management practices on a continual basis to adapt and mitigate climate change.

Uncertainties exist in the assumptions and simplifications necessary to model uneven-aged management through growth-yield curves, post-harvest stand dynamics, and late-stage successional dynamics (Ekholm et al., 2023). Theoretical curves of stand development may not capture multi-aged cohort dynamics of certain forests due to the complexity of forest ecosystems. Additionally, uncertainties remain about the long-term relationship between stand age and biomass represented in the form of growth-yield curves; statistical functions used to estimate the relationship between stand age and volume have strong implications for modeled results (Gustafson et al., 2020). Continued research will help to understand future forest growth, recruitment, and mortality under a changing climate with complex management interactions. With empirically driven frameworks, the chosen growth function used to model the relationship between stand-age and volume can lead to inherent biases in predicting growth, especially within older forest stands. Dependent upon the function, models may either over or under predict MAI in older stands due to increased uncertainty from underrepresentation of old trees in forest inventories. Future research will benefit from the development of improved data and modeling tools to understand tree to stand-level dynamics of volume and biomass and their relationship to stand age and other structural attributes to more accurately model forest growth within forest carbon models. Additionally, error associated with allometric equations to predict volume-to-biomass estimation remains dependent upon the sample size of trees but can oftentimes be ignored when using national inventories (Lin et al., 2023).

Modeling spatially stochastic events remains challenging but can be somewhat ameliorated through the usage of a spatially referenced model and the aggregation of forest records (Metsaranta et al., 2017). Furthermore, estimation methods for LUC can mischaracterize recently harvested forest pixels (instead classifying them as non-forest land use) and resulting in a potential overestimation of LUC rates; however, oftentimes remotely sensed inputs can be assigned an uncertainty of 5%–10%. Our analysis was designed to account for this as much as possible, but uncertainties still exist, and net C balance results are sensitive to assumed LUC trends. Capturing fine-scale management data, especially for private lands, remains challenging. However, through direct expert opinions from project stakeholders, we developed harvest regimes heuristically. Future research is needed to assess the sensitivity of future C sink-source strength to model inputs such as LUC rates, harvest rates, and growth as these inputs are prescriptive.

Further, quantifying relationships between substitution benefits, accounting of bioenergy emissions, and leakage across multiple timeframes remains challenging when the competing sectors are under pressure to actively decarbonize (Pingoud et al., 2018; Soimakallio et al., 2022). Some recent evidence suggests that substitution benefits are likely to be overestimated or oversimplified (Harmon, 2019); while this study has sought to provide reliable, state-specific substitution benefits by incorporating state- and region-specific wood use data and LCAs and allowing for the realities of leakage to dampen assumed product substitution needs, uncertainties in future product demand (which we assume to be constant) do contribute to overall substitution value uncertainties. In contrast, we did not consider substitution benefits from carbon capture emitted during woody biomass bioenergy derivation. Potential shifts in product types and half-lives of products have strong ramifications on future carbon storage (Xie et al., 2023), necessitating additional research to refine the relationship between carbon stored across HWP product streams, mill efficiencies, and the usage of both forest and mill residues. We assumed leakage rates for decreases in harvest rates; however, we did not estimate potential leakage rates associated with increases in harvest rates (i.e., reduced out-of-state harvest to compensate for increased in-state harvest); while this is how leakage is typically applied, it comes with an assumption in our case that all additional harvest goes toward additional in-state wood use, spurred by either policy or cost incentives. It is also worth noting that estimation of methane emissions from landfills, which are accounted for in this study, pose potential risks for improper accounting dependent upon assumptions which may result in smaller displacement factors (Chen et al., 2018).

Future forest C balances can be impacted in unforeseen ways with projected forest dynamics, climate change impacts, and disturbance regimes through 2100 which introduces uncertainty in our results. While some scenarios were designed to address this uncertainty prescriptively (such as the climate change scenarios), other uncertainties remain regarding fluctuations in socio-economic conditions. For example, knowledge of future rates of harvest and LUC remain highly uncertain. It is important to address socioeconomic factors driving forest management and the forest products sector when evaluating mitigation potential of scenarios due to their significant effects on both the results and feasibility of implementing carbon mitigation actions (Alig et al., 2004; Zhao et al., 2022). Additionally, our results do not incorporate changes to radiative forcings caused by a shift in surface albedo or other global teleconnection processes that may influence regional to global C balances (Swann et al., 2012).