The Rio Doce is one of the main Brazilian rivers and its watershed has a long history of land degradation and unplanned water use. The Rio Doce watershed presents high rates of deforestation followed by erosion, sedimentation and eutrophication of their rivers (Consórcio ECOPLAN - LUME, 2010). In November 2015, the collapse of the Fundão dam – in the municipality of Mariana, the state of Minas Gerais – spilled 40–62 million m3 of mining tailings in the Rio Doce. This incident has brought casualties, destroyed human settlements, impacted people's livelihoods and drastically polluted the Rio Doce (Meira et al., 2016; Neves et al., 2016). The accident was the last straw in a long-lasting degrading process, which affected the provisioning of ecosystem services for more than 1 million people (Fernandes et al., 2016). It has been estimated a loss of ecosystem services of about US$5.21 billion per year in the region (Garcia et al., 2017). Actually, forest restoration has been proposed as one of the main mechanism able to ensure water quality and quantity in the Rio Doce watershed, mainly in spring areas (Consórcio ECOPLAN - LUME, 2010; Fernandes et al., 2016; Morais et al., 2012).

Coincidently, the disaster took place almost in parallel to the UNFCCC Conference in Paris, where Brazil announced an ambitious commitment to restore 12 million hectares by 2030, which has been ratified in September 2016. The feasibility of this commitment has been discussed in previous studies (Rajao and Soares-Filho, 2015), but a set of national level policies, including recent legislation and national plans, back up such commitment (Scarano, 2017). The first one is the Native Vegetation Protection Law (the so-called New Forest Code; NVPL hereafter) that establishes the proportion of land within rural properties that must be maintained under protection or should be restored with native vegetation (Brancalion et al., 2016). The second in the National Climate Change Adaptation Plan (NAP), which has a strong focus on water resource management and actions related to ecosystem-based adaptation. Lastly, there is the National Policy for the Restoration of Native Vegetation (PROVEG) that announces a commitment to restore 12 million hectares by 2030. All these commitments have forest restoration as the most challenging and ambitious goal in the coming years. In particular, NVLP establishes the full protection of riparian vegetation of rivers, springs and other water sources in rural properties as areas of permanent preservation, hereafter APPs (Brancalion et al., 2016). The potential benefits and problems of the full implementation of the NVPL are still debated in the ecological literature (Metzger, 2010; Soares-Filho et al., 2014). The NVPL guaranteed amnesty for the past deforesters, as well as reduced the area of reserves in private properties compared to the former forest code (Soares-Filho et al., 2014). On the other hand, it could have some benefits like creating one of the world's major forest trading market (Soares-Filho et al., 2016). It is expected that the balance between such effects will depend on the region of its implementation and the ecological parameter studied. In the Rio Doce watershed case, water quality could be one of the main indicators of the effects of forest restoration in the region, as water quality reveals multiple integrated processes. However, it remains unclear how compliance to NVPL by restoring the existing environmental debt nearby the watercourses would drive changes in water quality of the Rio Doce watershed and how the national commitments can promote forest restoration in the region.

In this paper, we examine how compliance to the NVPL and consequent riparian forest cover restoration can influence the resilience and resistance of the Rio Doce watershed by using the water quality as a proxy and how it contributes to the national restoration commitments. We have particularly focused on watershed resistance and resilience during the rainy and the dry seasons, and discussed potential ecological and socioeconomics benefits that might result from compliance with the law. We defined resilience as the ability of the system to return to a hypothetical natural condition of water quality, while watershed resistance was defined as the ability to keep the water quality close to its average temporal behavior (see Box 1). Additionally, we discussed how multiple policy instruments could promote water quality by increasing forest restoration. Finally, we estimated the indirect contribution of the implementation of the NVPL in the Rio Doce watershed to Brazilian global commitments of forest restoration.

We verified the behavior of water quality in the Rio Doce watershed by using the temporal dataset provided by IGAM that is fully available at the site (http://portalinfohidro.igam.mg.gov.br/). We have extracted the raw values of nine limnological variables and then calculated the water quality index (WQI) for the 64 sampling points. We have established two metrics of stability: watershed resilience (a) and watershed resistance (b) based on the WQI temporal behavior. Watershed resilience was defined as the ability of the system to return the WQI value to a natural condition, where the hypothetical WQI value is maximum (100). Thus, as WQI increase as time goes by (positive slope values) reveals high resilience. Slope value zero reveals that WQI values did not change through the sampling period (no/low resilience) while negative slope values mean a decrease in the WQI (degradation). Watershed resistance was defined as the ability to keep the WQI close to its mean temporal behavior, determined as the coefficient of variation (CV). High CV values reveal a great fragility to move from the mean behavior and a low resistance. Thus, we are describing watershed resilience and resistance as a property of the system upstream each sampling point that was correlated to the native vegetation in APPs in the same area.

The Rio Doce watershed lost historically 71% of its native vegetation cover (Fig. 1). Such deforestation has also affected riparian areas that by law are required to be protected within APPs. The deficit in APPs is of ca. 716364 hectares of native vegetation, which is equivalent to 71% of the total APP due area (Fig. 1). This percentage was based on the riparian vegetation in hydric APPs as predicted by the NVPL in the whole watershed. Our estimates do not include hilltops and legal reserves. Most of the environmental debt is concentrated in the lower and middle Rio Doce (Fig. 1), nearby the most populous city in the region: Governador Valadares.

Fig. 1.

Environmental debt in the Rio Doce watershed based on the areas of permanent protection (APPs) predicted by the Native Vegetation Protection Law (NVPL). The environmental debt was estimated by forestry, urban and anthropic uses in the riparian area and it sums 71%. Points in the map represent the 64 areas monitored by IGAM (Instituto Mineiro para a Gestão das Águas) that were used in the present study to determine water quality behavior in the last decades.

(0.46MB).

Average WQI for the Rio Doce watershed, considering all sampling points in all periods, was 65.2±4.97 (mean±SD). We observed an increased WQI during the dry season (WQI=68.5±5.42; mean±SD) when compared to the rainy season (WQI=61.6±5.01). However, some regions in the watershed, such as the region downstream the Caratinga river had WQI values lower than 50 (48.5 during dry season; 45.2 during rainy season; 47 for all samplings), which means poor conditions in which conventional water treatment is not enough to allow human use (Lumb et al., 2011). On the other hand, sampling points close to springs had the highest WQI values, such as the observed for the Maquiné river (74.2 during the dry season; 74.3 during the rainy season; and 74.3 for all samplings). Table S3 shows complete results.

Percentage of native vegetation cover in APPs did not provide any significant change in the magnitude of WQI values (Fig. 2a). However, the percentage of native riparian vegetation increased watershed resilience during the dry season. Although resilience was almost three-fold higher during the rainy season, we did not observe a significant effect of riparian vegetation on resilience during this period. Considering all sampling periods, those samples holding at least 50% of upstream native vegetation in APPs (as required by NVPL) increased their WQI, on average, twice as faster than observed in other samples (Fig. 2). Considering the dry season, this pattern is even stronger; samples in places holding at least 50% of their upstream native vegetation cover increased their WQI three-fold faster than other regions (slope values: 0.24 versus 0.08). We also found that the higher the amount of native vegetation cover the higher the resistance of the watershed during the rainy season. Taking only those samples in which upstream riparian vegetation was lower than 50%, their coefficient of variation was 19% higher than the values observed in samples holding at least 50% of the riparian vegetation required by the NVPL.

Fig. 2.

Effects of percentage native vegetation in APPs on water quality magnitude (a), watershed resilience (b) and watershed resistance (c) in the Rio Doce watershed. Water quality index (WQI) was determined as proposed by the National Sanitation Foundation – USA and by using the IGAM open dataset (portalinfohidro.igam.mg.gov.br). Water quality magnitude (mean value), watershed resilience (slope value) and watershed resistance (1/coefficient of variation value) were determined through the temporal behavior of each sampling. Percentage native vegetation in APPs was determined based on the land use upstream each sampling point. Significant correlations were observed for the resilience of WQI in the dry season and the resistance of WQI in the rainy season. For full description and stats of the analysis performed, see Supplementary Material.

(0.27MB).

Costs to restore the whole vegetation debt of the watershed were estimated around US$ 3.6 billion or 59% of the ca. 6.1 billion dollar fund proposed by Fundação Renova, the consortia formed by the companies Samarco, Vale and BHP after the Fundão dam disaster.1 The implementation of the APPs required by the NVLP on the Rio Doce watershed will contribute to achieving 6% of the national target established by the Brazilian government, although this watershed represents less than 0.001% of the country area. Consequently, this restoration effort would promote sequestration of 14GtCO2yr−1ha−1. Finally, we found that restoration at this scale would increase watershed resilience four folds its actual condition and decrease 25% its coefficient of variation.

Compliance with the Brazilian NVPL through legal protection or restoration of riparian vegetation could increase both resistance and resilience of Rio Doce watershed. Despite the history of forest, soil, and water degradation in the watershed, coupled with the recent environmental tragedy, our results echo previous studies showing that restoration is an important mechanism to ensure water quality (Sáenz et al., 2016). In the Rio Doce watershed, we demonstrated that forest restoration might ensure water provisioning by increasing the resistance and resilience of water quality, during the rainy and dry season, respectively.

Our analysis demonstrated that increases in the percentage of riparian vegetation cover decreased the variability in water quality during the rainy season. Considerable changes in rainfall dynamics are predicted for the Rio Doce watershed area, including an increase in the occurrence of extreme rainfall events and a slight increase in total rainfall amounts (PBMC, 2013). Previous studies demonstrated that rainfall changes in the Atlantic forest will be an important driver of change in the behavior of aquatic ecosystems (Pires et al., 2017, 2016). During the rainy season, the surface drainage carries out several compounds to water bodies, making the riparian vegetation extremely important in buffering such effect (Klapprath and Johnston, 2000). On the other hand, during the dry season, restored riparian vegetation enhanced watershed resilience, as shown by the slight increase in water quality over the last decade (see positive slope values, Fig. 2). We suggest that under drought conditions, evapotranspiration increases vegetation demand for water, which causes a negative water balance decreasing river flow and water availability (Trisurat et al., 2016). This impact may be offset by the positive effects of forested areas in protecting soils from erosion and then increasing nutrient retention, which can increase water quality. In this way, forest restoration can have a stabilizing role in determining how changes in rainfall will actually influence water provisioning in the Rio Doce watershed.

Previous studies have reported successful initiatives of forest restoration in Atlantic forest with positive effects on multiple components of riverine systems (Attanasio et al., 2012). In terms of the accident impacts, vegetation plays a significant plus role in the recovery of biodiversity in the Rio Doce watershed. While some studies point out for the fact that tailings where composed of mud deposits, others claim that it holds high levels of toxicity (ANA, 2015b; Fernandes et al., 2016). The event has transformed the benthic region of the river in a big “desert” by suppressing the organic matter in there, the main energy source for aquatic organisms in rivers (Wallace et al., 1997). Increased riparian vegetation would increase the production of terrestrial organic matter that will reach the aquatic ecosystem, as allochthonous material (Dosskey et al., 2010; Wallace et al., 1997), recomposing the necessary conditions for the proper reestablishment of benthic organisms. Riparian vegetation restoration may also increase the incidence of organisms searching for suitable habitats for oviposition, increasing the organisms’ recruitment (Beltrão et al., 2009).

We stress that there are many important correlating factors, such as human population density, sanitation and urbanization level in unprotected areas that could explain the great residual effect of forest cover on water quality resilience and resistance and they should be taken into account. Also, several restoration technics have been applied in the Atlantic forest and they show considerable differences in their success rates (Rodrigues et al., 2009). The success of restoration initiatives prescribed by the NVLP may be extremely dependent on the context in which they take place. We suggest that different ways to restore forest areas should be evaluated to ensure ecosystem functioning and services in the Rio Doce watershed. In urban areas, the role of vegetation to reduce the effect of runoff may be supplanted by large sewage untreated discharge in aquatic environments (ANA, 2015a; Mulligan, 2009; Tong and Chen, 2002). Thus, initiatives that integrate multiple strategies of forest restoration and the particular sanitation issues of each area in the watershed would enhance the recovery processes of the Rio Doce river basin throughout its whole length.

Municipalities inserted in the Rio Doce basin have human development indices 11% lower than the average observed in Brazil and can be classified as a poor region (see Atlas do Desenvolvimento Humano no Brasil; http://atlasbrasil.org.br/2013/). We consider that implementation of NVPL can increase income provision and promote the local economy (Brancalion et al., 2014; Ferraz et al., 2014). In this way, the implementation of NVPL will regularize the legal situation of landowners, increasing the monetary value of their land and creating new markets for forest restoration. The establishment of the APPs predicted by the NVPL in the Rio Doce watershed will promote carbon sequestration of nearly 14GtCO2yr−1ha−1 while contemplating 6% of the total commitment for forest restoration established in the Brazilian NDCs (Fig. 3). Together, it would make the Rio Doce restoration a key example of ecosystem-based adaptation (Scarano, 2017; Scarano and Ceotto, 2015) by increasing income generation for populations extremely vulnerable to climate change and ensuring water quality in one of the most threatened Brazilian region (Fig. 4). Moreover, it could drive the country to reach the proposed sustainable development goals (SDGs), as the forest restoration in the Rio Doce watershed integrates multiple aspects, such as water security, biodiversity conservation and human dimension (Fig. 4).

Fig. 3.

Potential benefits of the implementation of the NVPL on carbon sequestration, forest areas and watershed resilience and resistance, as well as the contribution to the Brazilian NDCs. Values were estimated based on the current condition and the restoration of all areas predicted by the NVPL as APPs by considering the reforestation in Atlantic forest areas.

(0.19MB).

Fig. 4.

Overall scenario for the implementation of the Brazilian laws and national plans on biodiversity, ecosystem services (water quality) and human dimension, considering multiple factors such as climate change, sanitation and land use. Small boxes around biodiversity, ecosystem services and human dimension terms refer to the sustainable development goals attended by the implementation of these laws. Together, it could be considered a good model for the establishment of ecosystem-based adaptation strategies (EbA) face to climate changes and the sustainable development goals (SDGs) in the country.

(0.23MB).

We conclude that the restoration of riparian vegetation predicted by the NVPL may be one of the most effective mechanisms to ensure water quality for human provisioning in the Rio Doce watershed. Although our results suggest that large-scale, watershed-scale restoration could accelerate water quality improvement, restoration experience in Brazil is mostly confined to small spatial scales (Rodrigues et al., 2009). Additionally, the interaction between land use and climate changes may drastically affect the human well-being by regulating several services ecosystems provide, including water resources, in the coming years (Maneta et al., 2009; Scarano and Ceotto, 2015). Changes in water provisioning may affect several human activities and it is a challenging question, mainly in vulnerable areas such as the Rio Doce watershed area. Thus, we suggest that the full implementation of the NVPL will bring ecological, social and economic benefits and it will enable ecological restoration in one of the main Brazilian river basins. The potential success of the NVPL for the Rio Doce watershed would be strengthened if combined with other regulating factors and if several social sectors are part of the initiative, sharing the benefits arising from such endeavor.