Introduction
Tropical forest restoration is a central strategy in the global efforts to mitigate climate change (Bernal, Murray, & Pearson, 2018; Griscom et al., 2017; Houghton, Unruh, & Lefebvre, 1993; Pugh et al., 2019; Silver, Ostertag, & Lugo, 2000). Regenerating forests in the lowland tropics have high rates of carbon accumulation in aboveground biomass. On average, regenerating forests recover over 100 Mg of biomass per hectare within 20 years (L. Poorter et al., 2008) and reach biomass stocks comparable to old-growth forests within 66 to 80 years (Martin, Newton, & Bullock, 2013). Protecting young regenerating forests in Neotropics over the next 40 years could accumulate an amount of carbon sufficient to offset emissions from fossil fuel burning and industrial processes across all Latin American and Caribbean countries over the 1993-2014 period (Robin L. Chazdon et al., 2016).
The Bonn challenge, launched in 2011, is a global commitment of countries, NGOs, and private companies that aims to restore 350 M ha of degraded forests by 2030 (R. L. Chazdon et al., 2017; Grassi et al., 2017; K. D. Holl, 2017). The success of large-scale restoration programs will rely crucially on their economic viability over both short- and long- time scales, as well as the benefits they may provide to human wellbeing in relation to alternative land uses (Pedro H. S. Brancalion et al., 2019). Restoring large areas of tropical forests requires knowledge about which restoration approach provides the best returns on investment for accumulating carbon and other expected benefits, such as reducing extinction risk, improving water supplies, and increasing food security (Pedro H.S. Brancalion et al., 2018; Crouzeilles et al., 2017; Luke P. Shoo et al., 2017).
Natural regeneration is widely recognized as a core restoration strategy to mitigate climate change due to its potential to accumulate biomass at large spatial scales and at lower costs compared to tree plantations (R. L. Chazdon & Guariguata, 2016; Evans et al., 2015; Gilroy et al., 2014; Lewis, Wheeler, Mitchard, & Koch, 2019). This potential is already well demonstrated worldwide through several examples of forest transitions, where the abandonment of marginal agricultural lands led to the widespread expansion of forest cover through natural regeneration (Aide et al., 2013; Nanni et al., 2019; Rudel et al., 2005). As a spontaneous, uncontrolled reassembling process, natural regeneration trajectories can vary remarkably even among stands with similar local biophysical conditions and disturbance regime (Arroyo-Rodríguez et al., 2016; Mesquita, Massoca, Jakovac, Bentos, & Williamson, 2015; Norden et al., 2015). This variation has to be accounted for in restoration planning (Pedro H. S. Brancalion, Schweizer, et al., 2016; Uriarte & Chazdon, 2016).
In contrast with natural regeneration, tree plantings permit tighter control on the characteristics of the restored forest stand. Tree planting has often been established by restoration practitioners when ecosystem resilience is expected to be insufficient to support effective natural regeneration (K. D. Holl & Aide, 2011), or where rapid recovery of forest structure is required for legal compliance (Pedro H. S. Brancalion, Schweizer, et al., 2016; Chaves, Durigan, Brancalion, & Aronson, 2015; Rodrigues et al., 2011). Intensive management practices, such as planting trees at regular spacing, fertilizing soil, and controlling weeds and leaf-cutter ants, can promote rapid accumulation of above-ground biomass in tree stands (Rubilar et al., 2018; Wheeler et al., 2016). However, plantation management is costly, and restoration projects do not provide sufficient revenues to offset high implementation and maintenance costs (P.H.S.; Brancalion et al., 2019). In addition, most low-resilience sites where tree planting is required to support restoration have been historically used for intensive agricultural production and therefore have high land opportunity costs (Laurance, Sayer, & Cassman, 2014) and altered biophysical conditions that reduce the potential for natural regeneration (R.L. Chazdon, 2014).
Comparing restored forests established by tree planting and natural regeneration (i.e. second-growth forests) within the same geographic region is key to guide the selection of appropriate restoration approaches for different socioecological contexts and expected benefits (Luke P. Shoo et al., 2017). As stated above, tree planting and natural regeneration are usually found in contrasting biophysical conditions within landscapes. Consequently,in situ comparisons cannot be used to develop general guidelines about the relative costs and benefits of restoration practice (Reid, Fagan, & Zahawi, 2018), to provide a basis for guiding decisions regarding how to restore a given piece of land. Rather, such comparisons have the potential to highlight the actual efficiencies of “real-world” restoration practices, as they are determined by environmental, social and economic constraints. Such comparisons can help to guide decisions regarding how to invest limited resources for implementing forest restoration approaches across a range of socio-ecological conditions.
At the global scale, natural regeneration was shown to be more effective (Crouzeilles et al., 2017) or present similar outcomes (Meli et al., 2017) than tree plantations for recovering forest biodiversity and structure. At the local scale, however, L. P. Shoo, Freebody, Kanowski, and Catterall (2016) found that restoration plantations 1-25 yr old in the Australian wet tropics had a faster recovery of wood volume than natural regeneration sites, and César et al. (2017) found similar results in the Atlantic Forest of Brazil for 7-15 yr old forests undergoing restoration. These previous works focused mainly on above-ground carbon, with no assessment of the cost-effectiveness of restoration approaches. Cost-effectiveness assessments, rather than the evaluation of costs and effectiveness independently, are essential for guiding decision-making in restoration (Birch et al., 2010; Molin, Chazdon, Ferraz, & Brancalion, 2018; Strassburg et al., 2019). Moreover, the relative effect of tree planting and natural regeneration for soil carbon sequestration remain unknown.
Here, we assessed the carbon accumulation (above-ground and soil), implementation and land opportunity costs of forests established by natural regeneration (i.e, passive restoration) and high-diversity native tree plantations (i.e, active restoration). Our study was based on chronosequences (10-60 yr) in human-modified landscapes of Brazil’s Atlantic Forest. Brazil’s Atlantic Forest is a highly degraded biodiversity hotspot with low remnant native forest cover (only 12-28% of the original Atlantic Forest remains today, depending on remote sensing resolution; Rezende et al., 2018; Ribeiro, Metzger, Martensen, Ponzoni, & Hirota, 2009). Deforestation and fragmentation resulted mainly from the expansion of profitable agricultural and forestry commodities such as sugarcane, cattle ranching, and eucalypt plantations (Joly, Metzger, & Tabarelli, 2014), making cost-effectiveness a fundamental topic for the upscaling of restoration in this biome (Pedro Henrique Santin Brancalion, Viani, Strassburg, & Rodrigues, 2012). This work aimed to compare current management practices in order to guide the ambitious restoration programs recently established for the Atlantic Forest (Molin et al., 2018; Soares et al., 2014; Strassburg et al., 2019). Restoration practitioners and organizations worldwide have to make hard decisions on where and how to invest in restoration to enhance carbon accumulation while promoting biodiversity and other ecosystem services (Luke P. Shoo et al., 2017). The solutions identified for carbon farming through restoration with native trees in the Atlantic Forest of Brazil may be readily transferrable to other tropical forest regions with comparable economic and environmental contexts, and hopefully will contribute to more effective use of the limited resources currently available for forest restoration.