Discussion
Native tree plantations were shown to be more effective in accumulating above-ground and soil carbon stocks than natural regeneration during the first 50 years, but their higher implementation (tree plantations: US$2,788 versus natural regeneration: US$1,250) and land opportunity (tree plantations: US$324 per year versus natural regeneration: US$106 per year; Molin et al. 2018) costs make them less cost-effective for carbon farming. It is important to note that our work assessed high-diversity plantations of native tree species rather than monoculture tree plantations, to which natural regeneration has often been compared (Lewis et al., 2019). Several independent factors have contributed to this outcome. Exploring them is key to understanding how the cost-effectiveness of different restoration approaches may vary within human-modified tropical landscapes.
Most of the plantations included in the present study were found in riparian buffers − a privileged environmental condition for biomass accumulation due to the reduced water deficit and higher soil fertility − and in landholdings dominated by intensive agriculture, where soils usually display high nutrient and clay content. The studied tree plantations may therefore have benefited from higher nutrient and water availability than second-growth forests, which usually regenerate on slopes and sandy soils distant from watercourses. Water deficit and low soil fertility are known to limit tropical forest successional development (Jakovac, Pena-Claros, Kuyper, & Bongers, 2015; Martins, Marques, dos Santos, & Marques, 2015; Lourens Poorter et al., 2016; Zermeno-Hernandez, Mendez-Toribio, Siebe, Benitez-Malvido, & Martinez-Ramos, 2015), and may have reduced the biomass accumulation rates of second-growth forests in our study. Water limitation, in particular, may have played a critical role for the differential performance of restoration approaches, since the study region has a seasonal climate with annual water deficits of 20 mm or more depending on geographical features (Alvares, Stape, Sentelhas, Gonçalves, & Sparovek, 2013). In addition, tree plantations were fertilized, weeded, and planted at a regular spacing, which allow an efficient occupation of the deforested area by trees and enhance their growth, resulting in a higher accumulation of biomass per area (P.H.S.; Brancalion et al., 2019; Ferez et al., 2015). Our observation that tree plantations initially accumulate more carbon than second-growth forest corroborates previous results obtained in southern Costa Rica (Karen D. Holl & Zahawi, 2014) and Queensland, Australia (L. P. Shoo et al., 2016), but contradict the findings of Lewis et al. (2019) based on commercial forestry plantations.
The predominance of soil properties over land use effects on soil carbon stocks in the tropics has been reported by Powers, Corre, Twine, and Veldkamp (2011) and is confirmed in our study landscape. Greater biomass productivity and carbon inputs are expected to increase soil carbon stocks (Jandl et al., 2007; Karlen & Cambardella, 1996), even though changes in soil microbial communities may affect this causal dependency (Fontaine, Bardoux, Abbadie, & Mariotti, 2004). As a result, intensive forest management, afforestation, and reforestation are commonly associated with increased soil carbon stocks (Don, Schumacher, & Freibauer, 2011; Guo & Gifford, 2002). We do not report such an association between above-ground biomass and soil carbon stocks, as plantations displayed soil C stocks comparable to naturally regenerated forests, and the temporal accumulation of above-ground biomass in restored forests was not accompanied by a similar increase in soil carbon stocks (Fig. 1 and 2). Variations in soil properties among study plots may have obscured the relationship between soil carbon stocks and restored forest age at landscape level, as also reported by Mora et al. (2018) and Martin et al. (2013). Strikingly, none of the restoration management types recovered soil carbon stocks comparable to reference forests. Our results thus corroborate a global meta-analysis in tropical regions that found that second-growth forests stored 9% less soil carbon than primary forests (Don et al., 2011). The differences between forest types was in our case remarkably higher, with reference forests having approximately 50% higher soil carbon stocks at similar soil clay content. Taken as a whole, our results show that both above-ground biomass and soil carbon were enhanced in plantations compared to second growth forests, but that 60 years of stand development was not sufficient for the restored areas to recover stocks comparable to reference forests.
The inclusion of restoration costs in the analysis reversed the priority order of restoration approaches for carbon farming, and natural regeneration emerged as the most cost-effective solution. In other words, the cost reduction allowed by natural regeneration compared to planting more than compensated for the slower rate of accumulation of biomass, even in the unfavorable environmental conditions for tree growth in which second-growth forests regenerated in the study region. It is also important to note that we included in our analysis the direct financial costs of passive restoration (i.e., natural regeneration; Zahawi, Reid, & Holl, 2014) and considered both implementation and land opportunity costs. Natural regeneration may show even higher cost-effectiveness in regions where these costs are reduced, such as in landscapes not dominated by agriculture, where land rental prices are usually low. In addition, natural regeneration can be assisted (Shono, Cadaweng, & Durst, 2007), potentially at much lower costs than tree planting over the whole area. Assisted regeneration has the potential to enhance the growth performance of spontaneously regenerating trees, thus enhancing the biomass accumulation potential with lower implementation costs. We note that our results may be affected by a site selection bias, possibly resulting in an over-estimation of natural regeneration success. As was highlighted recently (Reid et al., 2018), natural regeneration is commonly conducted at sites closed to secondary forests remnants, which could bias conclusions of restoration practice comparisons. Additional studies would be needed to adequately represent the Atlantic Forest, a 1.3 million km² ecosystem with several biogeographical zones and socioeconomic conditions that would affect the comparisons made in this study.
Carbon market currently values a carbon credit approximately 5 US$ (Hamrick & Goldstein, 2016) for 1 ton of CO2 (273 kgC), i.e. 54.6 kgC.US$-1. Here, we report average total cost-effectiveness values for above-ground carbon of 9.1 kgC.US$-1 and 15.1 kgC.US$-1, for plantations and naturally regenerated forests, respectively. The market price of 5 US$ thus underestimates by more than a factor of 3 the actual price of carbon accumulation in restored forests. This clearly demonstrates that the revenues potentially obtained by trading carbon credits do not adequately cover the basic costs of both active and passive restoration. Overall, our results suggest that carbon markets as they are today offer a very low potential to up-scale restoration efforts in the Atlantic Forest. Other complementary revenue sources like those resulting from timber and non-timber forest products’ exploitation and payments for watershed services have been proposed to make tropical forest restoration financially viable (P.H.S.; Brancalion et al., 2017), which could be bundled with carbon farming for more favorable financial results. Notwithstanding, carbon farming will continue to be one of the major demands of environmental organizations, private companies, and governments supporting forest restoration in tropical regions, and finding the most cost-effective restoration approaches for this objective remains as a critical research challenge.