Introduction

Forests cover approximately 42 million km2 on Earth, which represents one-third of the world’s land area (Bonan 2008). They play a critical role in the regulation of climate and hydrological cycles, preservation of biodiversity and soil resources and provision of food, medicines or recreational values. A continuous forest cover loss during the last millennia (Pongratz et al. 2008) has compromised local to global processes involved in providing multiple ecological, social and economic services to human societies (Bongaarts 2007). Nevertheless, despite deforestation is still prevalent in some areas, according to Song et al. (2018) a forest transition could be beginning on Earth (i.e. when a region moves from losing to gain net forest cover; sensu Rudel et al. 2005). First forest transitions started during the last centuries in many countries of the northern hemisphere (e.g., Álvarez-Martínez et al. 2014). These transitions have been mainly driven by spontaneous revegetation, but also by local afforestation programs as societies undergo economic development, industrialization and urbanization (Mather 1992; Rudel et al. 2005). Nowadays, the utility of forests for counteracting the effects of droughts, floods, land degradation and, specially, climate change (Doelman et al. 2020), as well as global economic dynamics, is being reflected in the interest of other countries and supranational entities (e.g. FAO or WWF) in actively increasing forest cover (Rudel et al. 2020). Thus, new forest transitions usually imply more large-scale afforestation and faster changes in forest cover (Meyfroidt and Lambin 2011). However, the urgency to plant trees does not necessarily guarantee long-term environmental and societal benefits (Chazdon and Brancalion 2019). Tree cover can take many forms, with completely different effects on ES supply and biodiversity (Lamb 2014; Veldman et al. 2015).

Despite the wide range of benefits potentially provided by forests, afforestation has been traditionally focused on achieving only few of them. Most of the afforestation programs developed during the last century have the explicit target of timber production, replacing previously-cleared natural forests (Lamb 2014) as well as reducing the pressure over natural remaining stands (Sedjo 1999). However, these practices usually incorporate fast growing species in monocultures, which generates undesired effects and trade-offs between timber production and other regulating ES (Brockerhoff et al. 2013). These trade-offs have favored wood production at the expense of native biodiversity, hydrological regulation or socioeconomic status of local communities (Lamb et al. 2005; van Dijk and Keenan 2007; Gerber 2011). More recently, the use of monocultures to mitigate climate change has also created other negative effects such as impoverishing biodiversity, especially when afforestation is aimed solely at carbon sequestration (Grainger et al. 2009). Complementarily, there are growing evidences that natural forests provide multiple benefits not delivered by tree plantations (Alexander et al. 2011), which has led to a change in attitude during the last decades.

Nowadays, a growing number of national restoration and rehabilitation programs aim at restoring biodiversity while enhancing ES provisioning rather than simply maximizing productivity or carbon sequestration (Brockerhoff et al. 2013; Chazdon et al. 2016; Waring et al. 2020). However, this poses a huge challenge on the forestry industry as methods developed to establish industrial timber plantations continue to be chosen simply because the species and methodologies are well-known (Lamb 2018). As a consequence, some of the most ambitious strategies for large-scale afforestation have failed to supply their environmental targets (Hua et al. 2018), such as the afforestation program aiming at controlling desertification and soil erosion in China, with an investment of 100 billion US$ and 76 M ha of land afforested (Cao et al. 2011). This means that the paradigm has definitely not shifted yet, and tree planting for climate change mitigation, among many other environmental purposes, is still accomplished following traditional approaches. While new evidences in Europe call for reviewing the main past forest management strategies (Naudts et al. 2016), other studies still prioritize productivity criteria as a way of maximizing carbon sequestration (see in Ennos et al. 2019).

Understanding the consequences of past afforestation programs is paramount to meet the growing demands of ecosystem restoration and rehabilitation and guide forest transitions in the upcoming future. This could be achieved through the revision of the scientific literature synthesizing project results while analyzing the main drivers of observed effects and the ES arising from new tree cover trajectories (Wilson et al. 2017). However, this is a challenging task, as some afforestation programs have been never monitored (e.g. Cortina et al. 2011) or they are monitored using inappropriate comparisons (e.g. tree plantations are frequently compared to natural forests, when plantations are mostly established on intensive or degraded land cover types; Stephens and Wagner 2007). In order to avoid negative biases when assessing the efficiency of a tree plantation, the benchmark and controls used should be chosen based on their specific target (e.g. an ecosystem function) or, at least, the land use that is being replaced by the afforestation (Paquette and Messier 2010).

In this work, we have selected Spain as the study area as it has been one of the EU countries with larger areas of land afforested during the last century. In fact, this large-scale afforestation effort was considered rather unique in the world at those times and lead its forest transition (Traugott 1964). As in the rest of Europe, afforestation has been mainly addressed towards an increase of timber production and it has been an important agent of change for the distribution of forest species (McGrath et al. 2015). In this sense, Spain is representative of most of the large-scale afforestation programs of many other developed countries mainly focused on forest productivity. Additionally, the Spanish afforestation policy also sought improving the provision of other regulating ES, such as erosion or hydrological regulation because of the many reservoirs built during those years (Vadell et al. 2016). This is an important asset, as the effectiveness of past afforestation programs could be used to inform current policy agenda that tries meeting multiple goals. In this context, this study was set around two major objectives: (i) assessing the general efficiency of past afforestation policy on the provision of ES and other socio-ecological attributes, and (ii) determining whether the efficiency of the afforestation actions depended on the type of ecosystem against which the plantation is compared. In order to answer these questions we tested the following hypothesis: (i) past afforestation policy improved provisioning ES and socioeconomic aspects but failed on improving regulating ES and biodiversity conservation, and (ii) the efficiency of afforestation depended on the type of vegetation cover (e.g. bare land, pasture, shrubs or natural forests) used to evaluate it.

Materials and methods

We developed a conceptual framework to analyze the heterogeneous set of studies obtained from a systematic literature review (Fig. 1). As the Spanish afforestation policy has not been systematically monitored, this was necessary to get an objective assessment of restoration goals. Specifically, we evaluated the main outcomes of several afforestation programs by looking at their effects over a group of socio-ecological attributes sensitive to afforestation actions and related to environmental and social dynamics. The selected afforestation actions were independently considered as samples of a semi-quantitative meta-analysis on which we assessed the change in the benchmark (i.e. a socio-ecological attribute) by means of qualitative response types and sensitivity analyses. Response types were directly extracted from comparisons against other land cover classes made in the selected studies. Each land cover class was classified in different vegetation types according to a degradation-maturity gradient (Table 1). This allowed contrasting not only the efficiency of the afforestation to restore a degraded stand, but also how good afforestation actions were in relation to the ecological properties of a reference condition.

Fig. 1
figure 1

Flow diagram showing the methodological steps and conceptual framework followed in the study

Full size image
Table 1 Categories in which the vegetation types used as controls to assess the afforestation effects were divided
Full size table

Through this paper we use the term “afforestation” for the deliberate establishment of any tree-dominated vegetation over previous non-forested land and the terms “planted forest” and “plantation” interchangeably, regardless of the method of establishment (cf. FAO 2011).

Study area

The Spanish afforested area occupies nowadays over 12% of its territory (> 5 M ha), with over 60% of tree cover afforested between 1940 and 1984. The Spanish afforestation policy was articulated by the central public administration towards the regulation of hydrological and erosion processes and timber production. In fact, timber production and regulating-oriented afforestation represented, respectively, 36 and 64% of the total afforested land (Vadell et al. 2016). This was one of the most important differences with respect to other European countries such as Germany, France or United Kingdom, where timber production prevailed as the main focus of afforestation due to the shortage resulting from the Second World War (Mather 1978; Gadant 1991; Oosthoek 2013; Polley et al. 2014). However, implemented techniques did not differ much, as Spain adapted even‐aged plantation forestry, formulated in the forestry schools of France and Germany, to its complex physical environment in order to satisfy production and regulating goals simultaneously. This interest was mainly shown by the planting of non‐native timber trees (e.g. Eucalyptus spp., Monterrey pines or poplar hybrids) in existing woodlands, marginal pastures or agricultural lands and by the establishment of large-scale monospecific plantations (mainly Pinus spp.) on mountain degraded catchments and slopes (Castroviejo et al. 1985).

Literature search

We performed a systematic review of scientific literature to identify quantitative evidence on the impacts of afforestation on the provision of ES and other socio-ecological attributes. We used two literature databases and focused on publications before October 2019. We searched in the Scopus database for studies including the term “afforestation” in the title, abstract or keywords. We limited the affiliation country to Spain, getting 961 published works. We also used Dialnet, one of the widest scientific literature databases of iberoamerican languages, in order to get access to a larger range of Spanish studies. We included the terms “repoblación” or “repoblación forestal” or “reforestación” in their title, abstract or keywords. We limited the affiliation country to Spain, getting 1441 publications. The studies had to meet all the following criteria to be included in the analyses: (i) assessing the effect of at least one afforestation action on a component of the socio-ecological system (e.g. water quality, nutrient cycling, human population dynamics, etc.), and (ii) evaluating the effect by means of one or more indicators showing the deviation of the afforested plot in comparison to the control plot (i.e. a plot with different land cover or vegetation type). We initially scanned all titles and abstracts of selected studies and then read the complete papers when initial scans indicated a potential relevance for the study. In total 75 peer-reviewed studies were retained for further analysis (Appendix S1): 53 from Scopus, 18 from Dialnet and 4 additional publications obtained from cross-references of the previous ones.

Beyond this, each afforestation identified was considered as a case study (CS) whether it has been independently compared to any land cover typology regardless of the type or spatial scale of the comparison or type of study. These and other features as the date, typology and spatial location of the afforestation actions were also collected in order to characterize each CS (Fig. 2). Then, we focused on analysing the effects of the afforestation actions related to the different nature of the land cover typologies used to evaluate them. Thus, each CS was divided in as many observations as different land cover typologies were identified in it (Case Study Observations; CSO). Land cover typologies were categorized in three vegetation types (Table 1): native forest ecosystems (F), shrubland and grassland ecosystems (SG), and degraded ecosystems (D). While F assesses whether the afforestation would reach the ecological potential of the reference ecosystem, SG and D inform whether the afforestation improves the features of the restored (non-forest) land cover. In total, we evaluated 128 CS that produced 162 CSO.

Fig. 2
figure 2

A Location of the case studies (CS) and number of CS per, B afforestation date, C tree species and afforestation objective (light grey- logging; dark grey- hydro-forestry restoration) and D type of study included in the review

Full size image

Data extraction

We extracted and analysed measures of different variables (i.e. indicators; Table S2) related to the following socio-ecological attributes: biodiversity (i.e. abundance and diversity of particular species), ES and socioeconomic aspects.

The ES were classified according to the widely applied Common International Classification of Ecosystem Services (CICES) V5.1 (Haines-Young and Potschin 2018). Only a small number of studies referred explicitly to the concept of ES. The identification and classification of the reported indicators were therefore largely made on the basis of expert criteria refined by pair revision among authors (Table S2). The value of the indicator was derived from text, figures or tables of the studies included in the review. One of the CICES’ advantages is that it contains a nested hierarchical structure from Section to Division, Group and Class. In our case, this property has been used to make the aggregation of the indicators at each level more flexible in order to have enough data points and statistical power for further analyses. Thus, indicators were grouped in 11 categories; some of them were renamed to adapt it better to the meaning of the evaluation (Table 2). Furthermore, 4 extra categories were added with the purpose of extending the analysis to other interesting components and properties of the socio-ecological system. This were biodiversity, landscape connectivity, social structure and economic structure (Table 2).

Table 2 Socio-ecological attributes evaluated in the study and number of studies and CS used to assess each attribute
Full size table

The 163 CSO provided 817 measures of socio-ecological attributes because most studies provided information for more than one ES type, biodiversity and socioeconomic issues.

Response type

All reviewed studies differed widely in temporal and spatial scale, survey design, evaluated indicators, environmental conditions or measurement procedures. To allow suitable comparisons, all CSO were characterized in a qualitative way. For each of the 15 socio-ecological attributes on which there was at least one indicator, we evaluated whether each CSO afforestation action produced a positive (1), negative (− 1) or neutral effect (0). This assessment was done following three major steps which have been explained in detail in Appendix S3. First, each indicator was classified as positive, negative or variable depending on the direction of the effect (i.e. response type of the indicator) on the socio-ecological attribute (see Table S2). Second, each indicator was independently evaluated from its comparison, and then classified in positive (1), negative (− 1) or neutral effect (0). Third, when the CSO presented only one indicator to evaluate the socio-ecological attribute, we used it to determine the effect. Otherwise, all the indicators of the same socio-ecological attribute were compared among them in order to determine the most common effect. For this purpose, the indicators were split up in subcomponents of the socio-ecological attribute in order to consider their global contribution (Table S2). The evaluation was firstly done at the subcomponent level and then aggregated to the selected socio-ecological attribute.

Data analysis

In total, we quantified 409 measures of afforestation actions (CSO) and their effects on the socio-ecological attributes. In order to test the hypotheses of the study we applied two independent statistical analyses for categorical data. To evaluate our first hypothesis, we compared the frequency of response types (positive, negative or neutral) for each socio-ecological attribute with the Chi-square test. The differences between the number of positive and negative cases allowed us estimating the general effect of the afforestation policy on the group of socio-ecological attributes. In this analysis, we only considered the CSO of shrubland and grassland (SG) and degraded (D) vegetation types in order to assess the effect of the afforestation action against to a suitable land cover (i.e. restored ecosystem). To test our second hypothesis, we generated a 3 × 3 contingency table per each socio-ecological attribute in order to relate the response type (positive, negative or neutral) to the vegetation type used to evaluate it (F, SG and D). Then we applied a Fisher’s exact test. This method has been advised when some of the Chi-square test’s assumptions do not meet (i.e. 5 or more expected counts in 80% of cells and no cells with zero expected counts; McDonald 2014).

Sensitivity analysis

Evaluation of the response type depended on expert judgement, i.e. the relative importance of each indicator is ultimately related to author’s decision. In order to reduce the uncertainty of the results and generate a more objective expert-driven data evaluation, we re-analysed the effects of the afforestation actions on the socio-ecological attributes by assigning a set of weights to each indicator instead of a single value. For this purpose, we repeated the “response type” and “data analysis” processes 1500 times. At each iteration, we randomly assigned a value from a specific range (from 0.25 to 1.75 in steps of 0.25) to each indicator. Finally, we looked at the changes of p-value with respect to the results of a single expert-driven process.

Results

Out of the 129 CS compiled, 78% were pines, 6% poplars and 15% Eucalyptus plantations. Likewise, 67% of them were considered as hydro-forestry restoration, while the rest had mainly productive purposes. A total of 62% of the CS corresponded to afforestation actions made between the years 1940 and 1985. Finally, 75% of CSs were younger than 50 years when they were evaluated (Fig. 2).

Effects of the afforestation on the socio-ecological attributes

Afforestation had a significant negative impact on 6 out of the 15 socio-ecological attributes explored (40%; Fig. 3): water provision, hydrological regulation, life cycle—habitat protection, soil properties, social structure and biodiversity. Significant positive effects were only reported in 2 of the 15 socio-ecological attributes (13%; Fig. 3): timber provision and carbon sequestration. Sensitivity analysis showed changes in the p-value of 4 of the 15 socio-ecological attributes (Fig. S4A). However, only in hydrological regulation and social value the level of significance changed to non-significant.

Fig. 3
figure 3

Frequencies of case study observations (CSO) effects (positive in white, negative in black and neutral in grey) on the socio-ecological attributes considered in this study. We only considered the CSO of restored type of controls (SG shrubland and grassland ecosystems and D degraded ecosystems). Significant differences in the frequencies of response types within a socio-ecological attributes are tested with a Chi-squared test (d.f. = 2) and indicated by asterisks (*p < 0.05, **p < 0.01, ***p < 0.001 and n.s. p > 0.05)

Full size image

In the supply of provisioning and cultural ES, the number of positive CSOs were larger than the sum of negative and neutral effects (68 and 75% of positive cases respectively; Fig. 3A, C). The rest of the groups accumulated 40% (regulating ES; Fig. 3B), 40% (socioeconomic aspects; Fig. 3D) and 31% (biodiversity; Fig. 3E) of positive CSOs.

The effect of vegetation types used to evaluate afforestation efficiency

There were not significant differences in afforestation efficiency between the three vegetation types used to assess it in the groups of provisioning and cultural ES (Fig. 4A, C, respectively). In opposition, there were significant differences in afforestation efficiency between the three vegetation types used to assess it in 50% of regulating ES: erosion (p = 0.005), hydrological regulation (p = 0.024), soil properties (p < 0.001) and carbon sequestration (p = 0.007). In these cases, positive effects of afforestation were lower when the vegetation type in the control plot was natural forest (F). Furthermore, excluding carbon sequestration, positive responses were higher when afforestation was compared to degraded ecosystems (D) than in shrubland and grassland ecosystems (SG; Fig. 4B). There were not significant differences when evaluating socioeconomic aspects or biodiversity (Fig. 4D, E respectively). Finally, the sensitivity analysis showed only a slight difference in the p-value of hydrological regulation (from p < 0.05 to non-significant in 869 iterations; Fig. S4 B).

Fig. 4
figure 4

Mosaic plots for the effect of a case study observation (CSO) on socio-ecological attributes (P positive in white, N negative in black and 0 neutral in grey), the vegetation type used in this CSO is also shown (F forest ecosystems, SG shrubland and grassland ecosystems and D degraded ecosystems). Mosaic plots are grouped in types of socio-ecological attributes (A provisioning ecosystem services, B regulation ecosystem services, C cultural ecosystem services, D socioeconomic aspects and E biodiversity) representing the counts in contingency tables by tiles. The size of the tile is proportional to the cell count. Significant differences in the response types among vegetation types within a socio-ecological attribute are tested with a Fischer’s exact test and indicated by asterisks (*p < 0.05, **p < 0.01, ***p < 0.001 and n.s. p > 0.05). The number of CSO used to evaluate each socio-ecological attribute is also presented under each mosaic plot

Full size image

Discussion

In this work, we performed a meta-analysis identifying the main effects of past large-scale afforestation programs on the provision of ES and other socio-ecological attributes. Our first hypothesis was confirmed as the afforestation programs improved provisioning ES but not regulating ES, biodiversity conservation or socioeconomic aspects. Our second hypothesis was also confirmed since the efficiency of afforestation actions depends on the type of vegetation used for comparison.

Trade-offs between ES provisioning

Past afforestation policy produced a significant positive effect on timber provision (Fig. 3A) and carbon sequestration (Fig. 3B). This result is mainly related to the dominant use of fast-growing species (Vadell et al. 2016), which also led to trade-offs with regulating ES (Fig. 3B). These monospecific plantations use larger quantities of water (van Dijk and Keenan 2007) and they hamper natural vegetation colonization and species settling by sprouting or seedling (Romero-Alcaraz and Ávila 2000; Chirino et al. 2006). This produces a depletion of biodiversity and associated ecological functions. Regarding hydrological regulation, successful plantations usually improved hydrological responses at the catchment scale. However, the common use of hard techniques for hydric control and soil preparation (e.g. slope terracing) to reduce costs and to plant in constrained environments contributed to increase the number of negative CSO (Ortigosa et al. 1990). In many cases, these techniques also reactivated slope erosion processes and soil losses (Romero-Diaz et al. 2010). In others they simply produced negative effects on physical and chemical soil properties, together with a lower quality of the organic matter provided by the planted species to the topsoil (Ternan et al. 1996; Cristina Zancada et al. 2003).

Cultural ES improved with afforestation (Fig. 3C) by means of increasing social values in areas without a specific target or previous economic activity that avoids stakeholder interests to keep previous land cover and use types (Derak and Cortina 2014). Furthermore, afforestation made on severely degraded areas improved people’s perception about water flow, erosion regulation and aesthetic values (García Pérez et al. 1995). However, this group of ES should be carefully considered due to the highly context-dependence of their indicators (Derak and Cortina 2014; Fig. S4 A).

Finally, past afforestation policy produced negative effects on socioeconomic aspects (Fig. 3D), although economic benefits might have been underestimated in our analysis by assessing only two CSO with Eucalyptus plantations. Among the most remarkable negative effects on the society domain, large-scale afforestation programs were considered as a catalyst of rural exodus in several Spanish mountain depressed regions (Chauvelier 1990; Fernández-Muñoz 2002). In these areas, the demand for land to afforest caused deep socioeconomic changes in the former agrarian uses that led to the abandonment of traditional activities. Conversely, land use intensification in lowlands promoted landscape mosaics in where productive plantations usually produced higher incomes than the replaced land use (e.g. Rodríguez-Loinaz et al. 2013). However, low-income farmers were unable of modernizing their production system and frequently also forced to shut down (Marey-Pérez and Rodríguez-Vicente 2009).

Afforestation programs versus regulating ES

Regulating ES are related to functions arising from ecosystem stability, maturity and structural diversity (Odum 1969), which are also positive correlated with biodiversity patterns (Balvanera et al. 2006; Rey Benayas et al. 2009). Similarly, our results indicate that regulating ES were dependent on the vegetation type used to assess the efficiency of afforestation (Fig. 4B). Firstly, since the provision of regulating ES is closely related to the maturity of the ecosystems, vegetation typologies that are more advanced in the ecological succession are significantly more restrictive. This means that the success of an afforestation action decreases from control plots with degraded ecosystems (D) to shrubland and grassland ecosystems (SG) or native forests (F). Life cycle—habitat protection and biodiversity (Fig. 4B, E respectively) showed the same pattern, although the great number of negative cases under SG and D categories prevented finding significant differences.

The significant differences encountered for regulating ES and biodiversity when comparing to F and D vegetation types are in agreement with previous studies (Stephens and Wagner 2007; Bauhus et al. 2010; Brockerhoff et al. 2013). We observed that plantations improved the benchmarks in degraded and intensively managed land, but they usually provide lower values of regulating ES compared to native forests. Moreover, afforestation actions did not improve regulating ES when compared to SG vegetation types. This confirms that afforested land may follow slower or undesired restoration trajectories that would not reach the functional levels of the reference condition (F) or even previous stages of ecological succession (SG) (Chazdon et al. 2016). In other words, the observed failures of some afforestation actions for improving the provisioning of regulating ES can be related to the application of non-effective restoration techniques but also to their negative impacts on biodiversity and community dynamics. Furthermore, Mediterranean related ecosystems are highly hampered by climate and other abiotic constraints and forest could not be always the reference target of ecological restoration (Cortina et al. 2006). Afforestation may then represent a perturbation producing a regression in some functional properties (Chirino et al. 2001; Maestre et al. 2003).

Lessons learnt: from afforestation programs to landscape planning

Our study encompasses a great variability in bioclimatic, topographical, geological and socioeconomic conditions because of the large environmental heterogeneity of the Iberian Peninsula (Rey Benayas and Scheiner 2002). Most of the revised afforestation actions focused on timber production as underlying goal, in opposition to a more comprehensive ecological functioning restoration. This is also the case for most of the large-scale afforestation programs carried out in developed countries during the twentieth century (Zanchi et al. 2007; Tal and Gordon 2010; Bradshaw 2012; McGrath et al. 2015). Therefore, an extrapolation of the main results can be reliable at a pan European or at least Mediterranean scale (Rudel et al. 2005), especially for mountainous regions (Fig. S5). Moreover, these lessons learnt can be particularly interesting for countries that are currently experiencing similar forest transitions (e.g. Vietnam and Thailand; Wilson et al. 2017). In this sense, the sensitivity analysis carried out precluded expert biases, reinforcing the consistency of our results. In fact, similar worldwide studies (e.g. Gerrand et al. 2003; Tak et al. 2007; Cao et al. 2011) give support to our initial hypothesis that the investment on afforestation for timber production is not an effective strategy for improving regulating ES and biodiversity at a large-scale (Bennett et al. 2009). However, recovering the full list of species of primary forests (i.e. alpha diversity) in a restored forest is not necessary to enhance many regulating services (Lamb 2018). Reliable win–win strategies to improve the provisioning of multiple ES should be chosen to warrant multifunctionality at a landscape scale in an affordable manner. For example, by using different species mixtures at different spatial extents (Pasari et al. 2013) and fostering the conservation of key ecosystem components.

Consequently, large-scale afforestation programs should be integrated into broader spatial planning instruments, such as Green Infrastructure Network schemes (GIN; Benedict and McMahon, 2012), in order to: (i) optimize biodiversity and multifunctionality at the landscape scale and (ii) control trade-offs among bundles of ES. In this sense, commercial forest plantations can be potentially incorporated to the GIN strategy to generate provisioning ES (e.g. timber production) if their spatial extent, location and techniques applied are suitable also for other purposes related to sensitive regulating ES. If not, additional measures should be implemented when afforestation trade-offs are detected (e.g. green filters of natural vegetation in peripheral areas of fast growing tree plantations; Thompson et al. 2011). Conversely, afforestation projects focusing on recovering species-richness or provisioning of regulating ES could be promoted through passive restoration of native vegetation or active restoration of particular vegetation traits.

Conclusions

Since the beginning of the twentieth century, a huge amount of money and effort has been addressed towards the afforestation of degraded territories. Afforestation methods and strategies have shifted as scientific knowledge and society perspectives evolve and, thus, the analysis of main outcomes of past afforestation policies can provide useful information to tackle future challenges. Our study shows the difficulties of ensuring the provisioning of regulating ES when timber provisioning or carbon sequestration are prioritized at broad scales. In these cases, the productivity target usually conditions the design and management of tree plantations and clear trade-offs could appear as a consequence of biodiversity losses but also because of the impacts on soil properties and the entire ecosystem functioning and dynamics. In fact, past large-scale afforestation programs only improved the provision of regulating ES when compared to highly degraded or intensively managed areas. Thus, our study shows that new forest restoration and rehabilitation programs should also consider fostering successional trajectories towards an ideal state that allows recovering important biophysical processes and ecosystem components related to the provision of regulating ES and biodiversity patterns. As a consequence, any new large-scale afforestation program should be implemented under more comprehensive landscape approaches that allow reducing undesired ES trade-offs and getting multiple objectives while managing different type of ecosystems. These approaches allow defining appropriate mechanisms to assemble active and passive restoration actions, tree commercial planting and other alternatives in cost-effective ways that enable achieving multifunctional landscapes and ensuring biodiversity conservation under uncertain scenarios of Global Change.