Introduction

An estimated 25 % of the world’s land area is degraded (FAO 2011). Responses from the international community include the Changwon Initiative of the United Nations Convention to Combat Desertification that aims to achieve land degradation neutrality by 2030 and the Aichi Declaration of the Convention on Biological Diversity that set a goal of restoring 15 % of degraded lands by 2020. An estimated two billion ha of degraded forest land needs restoration (Minnemayer et al. 2011; Seddon et al. 2014) and the international response was the Bonn Challenge to restore 150 million ha by 2020. Recently the New York Declaration increased the challenge to restore 350 million ha of degraded forest land by 2030. One attempt to locate degraded forest land (Hansen et al. 2010) used remotely sensed data to partition the global landscape into areas where restoration could occur in remote, mosaic, or wide-scale fashion (Minnemayer et al. 2011). Many of the current forest restoration opportunities exist in the Tropics and Temperate Zones, mostly for mosaic restoration in areas of moderate human pressure (between 10 and 100 people km2) (Fig. 1).

Fig. 1
figure 1

An estimated 2 billion ha globally of deforested and degraded forest land present the greatest opportunities for restoration in mosaic landscapes with moderate human pressure (between 10 and 100 people km2). Lesser opportunity occurs for remote restoration in unpopulated areas (density < 1 person km2 within a 500 km radius) or wide-scale restoration (in areas with <10 people km2) (Source Minnemayer et al. 2011)

Full size image

The heightened perception in international policy arenas of the importance of forests and trees outside forests (woodlands, savannas, on farms) is leading to a change in the way we approach future landscapes (Menz et al. 2013; Zomer et al. 2008). A renewed awareness of the importance of forests for human well-being stems from efforts to mitigate and adapt to climate change (Biagini et al. 2014); from an awareness of the importance of forests in sustainable development (Macqueen et al. 2014; Minang et al. 2015; Ordonez et al. 2014); and from the sheer magnitude of deforestation and degradation we have achieved (FAO 2011; Minnemayer et al. 2011; Zalasiewicz et al. 2010). Given the scope of the global restoration challenge in the twenty-first century, approaches that focus on restoring functioning landscapes are the most likely to succeed (Minnemayer et al. 2011; Stanturf et al. 2014a). Landscapes are socio-ecological systems (SES) (Liu et al. 2007; Ostrom 2009), mosaics of land cover and land use caused by the interplay of ecological possibilities and socio-economic constraints (Lamb et al. 2012; Parrott and Meyer 2012).

The nature of future landscapes will be determined by the success of interventions to restore degraded lands including degraded forests and deforested areas. The uncertainty introduced by a rapidly changing global environment, however, questions the usefulness of success criteria based on present or past ecosystem conditions (Hobbs 2013). Incomplete understanding of the effects on landscapes from temporal and spatial changes in climate and socio-ecological systems has produced a cascade of uncertainty (Wilby and Dessai 2010) resulting in “spatiotemporal chaos” (Pielke et al. 2013) that precludes realistic predictions of where and when significant changes will occur. My objectives are to outline the uncertain future as it affects landscape restoration and to suggest ways to integrate landscape restoration with climate change mitigation and adaptation. My emphasis is on landscapes where forests, woodlands, and agroforestry are important if not dominant land uses. I draw on the adaptation literature, which is oriented toward avoiding degradation caused by global change. My focus, however, is on the large land area already deforested or degraded, and the landscapes that may become degraded under altered climate. Inasmuch as many of the same strategies and methods apply to both current and future restoration needs (Keenan 2015; Millar et al. 2007), the distinction is not sharp.

Uncertain future

Future uncertainty stems from changing climate and the timing of significant departures from current conditions (Mora et al. 2013; Rummukainen 2012); social system responses to drivers of global change, which include but are not restricted to changing climate (Arneth et al. 2014; Wilby and Dessai 2010); and ecosystem responses to changes in coupled socio-ecological systems (Keskitalo 2011; Liu et al. 2007; Seidl and Lexer 2013). Coupled ocean/atmospheric general circulation models (AOGCM) have provided the scientific basis for concerns about climate change. These models, however, provide only a coarse-resolution view of global climate and each model differs in how it represents the physical forces driving climate, resulting in different levels of skill in modeling historic or present climate, especially short-term extremes (Becker et al. 2013).

Population increases (Gerland et al. 2014), changing consumption patterns as a consequence of rising standards of living, and migrations are the demographic drivers of land-use change (Lambin and Meyfroidt 2011; Pretty 2013). Climate change impacts on humans will elicit responses that will indirectly impact ecological systems by altering the way land and natural resources are utilized (Chapman et al. 2014). One downside of a more integrated world is the facilitated movement of pests and invasive plants (Bradley et al. 2011; Logan et al. 2003; Sturrock et al. 2011).

Examples of the potential effects of the interacting drivers of change can be seen today in the Tropics where biodiversity and the potential for mosaic restoration are high (Fig. 1) (Bellard et al. 2014; Dirzo et al. 2014; Visconti et al. 2011; Williams 2013). Generally, these areas are in countries with the highest population growth vulnerable to climate change and a rural populace that is susceptible to food insecurity (Dixon et al. 2003; McMichael et al. 2006; Patz et al. 2005; Thornton et al. 2014; WHO 2014). Farmers dependent on rain fed agriculture and who lack the capital to intensify (Pretty and Bharucha 2014) may respond to lower crop yields because of climate change through extensification of agriculture at the expense of forests (Seto et al. 2012; Zabel et al. 2014).

Ecosystems and species have adapted to dynamically changing climatic conditions for millennia (Jackson and Overpeck 2000; Millar 2014). Plant populations will respond to future moderate shifts in habitat by tolerating and persisting, migrating, or dying out (Aitken et al. 2008). The nature and rapidity of changing conditions caused by anthropogenic forcing could overwhelm natural adaptation processes, including the dispersal rates needed for species to adapt (Burrows et al. 2014; Corlett and Westcott 2013; Jump and Peñuelas 2005). Predicting the effects of climate change on natural systems, in particular long-lived woody species such as forest trees is important for restoration because abrupt changes will increase the area requiring restoration interventions and ecosystem responses to changed conditions influences what interventions are needed, adaptive, and affordable. At the species level, attempts to project the nature of future landscapes will face a variety of sources of uncertainty, including

  • How will a species respond to climate change within its present range (Pacifici et al. 2015; Park et al. 2014)?

  • What other changes in addition to direct climate effects will occur in a species’ present location and how vulnerable will it be to indirect effects (Chapman et al. 2014; Pacifici et al. 2015)?

  • Where might other locations be with a suitable future climate (Vos et al. 2008)?

  • How will a species respond to dispersing or being moved to a new location with a suitable future climate (Breed et al. 2013; Fitzpatrick and Hargrove 2009; Lunt et al. 2013; Rout et al. 2013)?

  • What effect will an introduced species have on the receiving ecosystem (Hewitt et al. 2011; Laikre et al. 2010; McLachlan et al. 2007; Pedlar et al. 2012)?

Views of the future

Climate and weather are inherently variable and vegetation has adapted to the regional and local range of multi-decadal as well as inter-annual temperatures, precipitation, wind, etc. (Nicotra et al. 2010; Valladares et al. 2014). The most recent projections are for a warmer 2100 world (IPCC 2012) characterized by decreases in cold days and nights and increases in unusually warm days and nights. Land areas will experience increases in the frequency, duration, and/or intensity of warm spells or heat waves. Average precipitation is projected to decrease in some regions and increase in others with more frequent heavy precipitation events, or more of the total precipitation occurring as heavy events. Higher latitudes and tropical regions will be particularly affected, as well as winter precipitation in mid-latitudes in the Northern hemisphere.

Most climate models project changes 50–100 years in the future with an implication that change will be gradual and linear, but projections of extreme events suggest more abrupt shifts will occur along the trajectory (Cai et al. 2014; Leadley et al. 2014). Knowing when to expect significant departure from historic climatic conditions is critical for deciding whether to alter current management strategies, such as species selection and spatial prioritization of target areas for restoration and in forests, rotation length. Mora et al. (2013) examined the timing of departures from current climates, roughly equivalent to the novel or no-analog climates of Williams et al. (2007). The projected near-surface temperature of the average location would exceed the historic range of variability by 2047 under the business-as-usual RCP85 scenario (Van Vuuren et al. 2011) and by 2069 under the RCP45 (rapid mitigation scenario). The earliest occurrence of unprecedented climate occurred in the Tropics; even though the magnitude of warming was small the current annual and seasonal temperature variability was low. These novel climates could give rise to no-analog plant communities that may be transient within the lifespan of long-lived forests tree species (Williams and Jackson 2007).

In the very near-term (0–30 years), on-going pressures from human demographic shifts, land use change, and introduced pests from globalization of trade will greatly affect ecosystems (Cochrane and Laurance 2008; Kiage 2013; Lambin and Meyfroidt 2011; Laurance et al. 2014). The long-term synergistic effect of human and climate induced change will be significant (Barnosky et al. 2012; Hughes et al. 2013; Leadley et al. 2014; Schröter et al. 2005). For example, forests are important to the terrestrial carbon cycle and measures to mitigate climate change have focused on maintaining and increasing carbon stocks by increasing C-sequestration and reducing emissions. Restoring future landscapes likely will play an increasing role in climate change mitigation and adaptation. Approaches to restoration are not without controversy, particularly in terms of objectives or desirable endpoints (Stanturf et al. 2014a, b) and vegetation patterns altered by climate change will add additional opportunities for debate. For example, afforestation (planting trees on land that was used for purposes other than forests) is a restoration practice and by increasing forest area and carbon stocks, achieves climate change mitigation goals. Creating closed canopy forest on grassy biomes may compromise ecosystem services (Veldman et al. 2015). Under future climates and altered fire regimes, forest/grassland transitions may shift, adding to the complexity of restoration decisions. On one hand, altered climate may favor grassland and cause degradation of forest or woodland and attempting to restore forest or woodland would go against ecological forces. On the other hand, future climate may favor woody species and afforestation would be adaptive to future conditions but may be opposed by the public as degrading current grasslands.

Mitigation and adaptation

Concerns about climate change and loss of biodiversity underscore the importance of forest land cover and trees in the landscape for mitigation and adaptation strategies (Mayaux et al. 2013; Rudel 2013). Mitigation aims at causes of climate change, the emission of green house gases (GHG) and their accumulation in the atmosphere. Mitigation interventions either reduce the sources, or enhance the sinks for GHG (IPCC 2003). Mitigation has been regarded as an international issue (Locatelli et al. 2011); the benefits of mitigation accrue globally, over the long-term because of the inertia of the climate system. Adaptation is local in nature, focusing on the effects of climate change on natural and social systems. Integrating landscape restoration with climate change mitigation and adaptation means squarely facing the diversity of ecological conditions and socio-cultural contexts (Liu et al. 2007). Adaptation strategies must be robust under a broader range of potential climatic conditions than those faced by managers in the past (Hallegatte 2009). Adaptation is essentially about managing the effects of climate change; therefore it is a continuing process (Stein et al. 2013). Inasmuch as many interventions are long-term commitments, restoration decisions including species selections will be made under uncertainty.

Mitigation

Mitigation activities (Table 1) include carbon conservation and increasing sequestration, offsets through substitution for fossil fuels or unsustainably harvested wood, and offsets from use of wood products rather than steel, cement, or plastic (Ravindranath 2007). Land use change, including deforestation and forest degradation, is a major cause of GHG emissions (Cochrane and Laurance 2008; Mahmood et al. 2013; Pielke et al. 2011) and despite the stability of total global forest cover (FAO 2010), deforestation is regionally significant and continues unabated in the Tropics (Kelatwang and Garzuglia 2006; Kim et al. 2015). Attempts to mitigate climate change have addressed forest loss in the Tropics through Reduced Emissions from Deforestation and Degradation (REDD+), an effort to protect and increase forested area (Nepstad et al. 2013). Increasing bioenergy use and efficiency are other ways to offset fossil fuel emissions (Creutzig et al. 2014).

Table 1 Mitigation opportunities relevant to forest landscape restoration
Full size table

Mitigation activity may be situated on the landscape to improve connectivity among patches of intact forests and reduce fragmentation, aiding dispersal, migration, and gene flow among populations of plants and animals (Table 1). New or rehabilitated forest areas around intact forests, especially protected areas, may act as buffers and reduce pressure on native forests. Examples of landscape-scale restoration designs are given in Stanturf et al. (2014a). These include corridor plantings that connect patches of intact forests or as riparian buffers in agricultural fields. Dispersed plantings, usually of species with limited dispersal capability, planted as individuals or in clumps near intact forest remnants may over time be augmented by other species dispersed from the forest. Nucleation and cluster plantings within an agricultural matrix are similar plantings (Corbin and Holl 2012; Schönenberger 2001).

Adaptation

Adaptation emphasizes robust solutions that increase resilience of forest ecosystems (Dumroese et al. 2015). Reconstructing forests (e.g., afforestation) can be done in a way that achieves mitigation, provides biodiversity and other ecosystem services, benefits local communities, and increases adaptive capacities of the forest and local community (Table 2). Restoring degraded forests by altering composition, structure, or ecological processes, singly or in combination (Löf et al. 2012; Stanturf et al. 2014a), can be done with an eye toward future climate (Table 2). Some guiding principles are to maintain or improve ecosystem processes (Bolte et al. 2009a; Janowiak et al. 2014; Keenan 2015; Spittlehouse and Stewart 2004) and to promote species, genetic, structural, and age-class diversity (Millar et al. 2007; Oliver et al. 2012) in order to spread risk (Ando and Mallory 2012; Crowe and Parker 2008; Yemshanov et al. 2013).

Table 2 Adaptation opportunities relevant to forest landscape restoration (Adapted from Bolte et al. 2009a; Janowiak et al. 2014; Keskitalo 2011; Kolström et al. 2011; Lindner et al. 2014; Spittlehouse and Stewart 2004; Stanturf et al. 2014b)
Full size table

Passively responding to climate change by accepting what develops and accommodating to the change can be a reversible adaptation strategy (Hallegatte 2009), in that active approaches can be adopted later if the changes are deemed unacceptable. As a restoration method, a passive response is not cost-free (Zahawi et al. 2014) but may be justified if one or more of the following conditions are met: (1) the risk of degradation is low, (2) the social and ecological values at risk are low, hence the costs of degradation are bearable, (3) the costs of acting are high relative to the benefits, and (4) especially if limited resources must be directed toward higher-valued ecosystems that are at greater risk. After a drought-induced insect outbreak, for example remote areas could be allowed to regenerate naturally without intervention even if an herbaceous or shrub-dominated assemblage develops. Notwithstanding, higher-valued stands could be replanted with species or provenances better adapted to future climate (Bolte et al. 2009a; Keenan 2015).

Three active approaches to reducing vulnerability and increasing adaptive capacity are (Table 2) incremental, anticipatory, or transformational adaptation (Joyce et al. 2013; Kates et al. 2012). Incremental adaptation is a short-term coping strategy (Moser and Ekstrom 2010) that seeks to avoid disruptions and maintain forest ecosystems at their current locations, essentially managing for persistence (Kates et al. 2012; Stein et al. 2013). Incremental adaptations may be extensions of current adaptive practices (Heltberg et al. 2009) that might also reduce vulnerability or avoid loss under altered conditions (Hobbs et al. 2011; Joyce et al. 2009; Kates et al. 2012). Restoration to some measure of historical fidelity (Burton and Macdonald 2011; SERI 2004; Tierney et al. 2009) or range of natural variability (Agee 2003; Keane et al. 2009) are incremental approaches that rely on projections of a stable climate or that the resistance or resilience of healthy ecosystems will enable them to persist under altered climate (Chapman et al. 2014; Stein et al. 2013). Anticipatory adaptations require more substantial adjustments but stop short of transforming the system (Kates et al. 2012). They incorporate some aspects of an incremental approach but are more future-oriented in terms of goals and expectations of altered climate. Transformational adaptations attempt to anticipate climate change and respond in ways that are larger in scale or more intense than anticipatory adaptations, or they are novel by their nature or new to a region or resource system (Hobbs et al. 2011; Joyce et al. 2013; Kates et al. 2012).

The three active adaptation strategies differ in their future orientation but share similar objectives of maintaining vigor at stand level by favoring genotypes that are adapted to local conditions; resisting pathogens; managing herbivory to ensure adequate regeneration; encouraging species and structural diversity at the stand-level, landscape-level, or both; and providing connectivity and reducing landscape fragmentation (Bolte et al. 2009a; Janowiak et al. 2014; Keenan 2015; Lindner et al. 2014; Spittlehouse and Stewart 2004). The approach to reaching these objectives, however, may differ among the strategies (Table 3). Restoration focused on resilient forests under future climate conditions may utilize any of the strategies or some combination on the landscape (Bormann and Kiester 2004; Park et al. 2014). The appropriateness of any of these adaptation strategies will depend on the actual rate and nature of climate change and the vulnerability of species or ecosystems of interest (Park et al. 2014).

Table 3 Comparison of the features of incremental, anticipatory, and transformation adaptation strategies
Full size table

Incremental adaptation

Incremental adaptation focuses on historic fidelity using native species from local sources that may adapt to changing climate through natural evolutionary mechanisms (Table 3). This strategy has a low degree of novelty (Perring et al. 2014) and is likely to be motivated by the ecological restoration paradigm (Stanturf et al. 2014b). Nevertheless, assisted population migration may be a feature of incremental adaptation to the extent of moving seed sources climatically or geographically within the current range of a species (Breed et al. 2013; Janowiak et al. 2014; Williams and Dumroese 2013). Favoring species and genotypes better adapted to future conditions provides a safety margin (Bolte et al. 2009a; Hallegatte 2009; Janowiak et al. 2014; Keenan 2015). This is most easily accomplished by outplanting or sowing germplasm from a wide geographic range (Janowiak et al. 2014; Williams and Dumroese 2013). Maintaining minor species in natural regeneration systems is another safety margin provided at little or no “cost.” For example, species currently in low abundance but potentially adapted to even the most extreme future conditions could be maintained or favored by silvicultural practices (Dumroese et al. 2015).

Unique habitats and species of concern can be conserved within existing natural or protected areas, or by enlarging existing areas (Janowiak et al. 2014). Other adaptive activities include favoring multiple species plantings at the stand-level (Gamfeldt et al. 2013; Kelty 2006; Lockhart et al. 2008) and developing structure/age diversity in the landscape (Millar et al. 2007; Oliver et al. 2012). Avoiding consequences of climate change can be accomplished by reducing rotation length or planting species or varieties that grow rapidly to maturity (Park et al. 2014). Short-rotation bioenergy plantings have an additional mitigation benefit (Agostini et al. 2015) and may be combined with a slower growing species in a nurse crop system for restoration (Löf et al. 2014; Stanturf et al. 2009, 2014a).

Establishing new forests or restoring degraded forests must balance sustainability under current climate conditions and adaptability to future climates; thus choice of species, stand structure, and management regime may require trade-offs (Seidl and Lexer 2013). Lowered productivity may result because adaptation to future conditions could be sub-optimal under current climate (Hallegatte 2009; Keenan 2015). Restoration that strives for rapid revegetation and quick site capture (Pichancourt et al. 2014; Stanturf et al. 2001) can avoid negative effects of accelerated soil erosion or invasion by non-native plants (D’Antonio and Vitousek 1992; Janowiak et al. 2014). The greatest opportunities for incremental adaptation exist where active forest management already occurs and adequate infrastructure and technical capacity exists (Guldin 2013; Spittlehouse and Stewart 2004).

Anticipatory adaptation

Anticipatory adaptation is future-oriented and the underlying paradigm is functional restoration, which emphasizes restoration of abiotic and biotic processes rather than fidelity to historic structure or composition (Stanturf et al. 2014a, b). While the starting point for anticipatory adaptation is the same suite of incremental activities described above (Janowiak et al. 2014; Spittlehouse and Stewart 2004), there is greater tolerance for novelty and for non-native species that are functional analogs to native species (Davis et al. 2011; Hobbs et al. 2009; Lugo 2009). Novelty is a matter of degree (Hobbs et al. 2013); replacing a maladapted genotype of a native species with a better-adapted provenance (incremental adaptation) is an example of low degree of novelty. Replacement with a non-native species with desired functional traits, or a genetically modified clone of a native species, would constitute a greater degree of novelty and fall into anticipatory adaptation (Table 3). Neo-native or hybrid ecosystems could arise spontaneously as assemblages of native species in new combinations or by intentionally moving communities of native species to a new location in anticipation of climate change (Hobbs et al. 2013; Perring et al. 2013; Rout et al. 2013). Extending the historic range of species in advance of extirpation is another anticipatory adaptation (Williams and Dumroese 2013). New refugia for sensitive species can be identified and established within the climatically stable portion of geographic range or beyond (De Frenne et al. 2013; Dobrowski 2011; Janowiak et al. 2014; Keppel and Wardell-Johnson 2012).

Transformational adaptation

Transformational adaptation may be planned or arise spontaneously (Alig et al. 2004; Joyce et al. 2013). Assisted migration of a species far beyond its historical range (Lunt et al. 2013; McLachlan et al. 2007; Pedlar et al. 2012), introduction of non-native species (Davis et al. 2011), or genetic modification to restore keystone species (Jacobs et al. 2013; Seddon et al. 2014) are transformational adaptations. Prominent ecologists and conservationists recently called for intervention ecology, a transformational approach to restoration of degraded ecosystems (Hobbs et al. 2009, 2011; Sarr and Puettmann 2008) that acknowledges the dynamic nature of ecosystems, the prospect for even more rapid change under altered climate, and the infeasibility of complete restoration (Clement and Junqueira 2010; Hobbs 2013). Considerable planning (especially in advance of extreme events; Beatty and Owen 2005; Hallegatte 2009; Stanturf et al. 2007), experimentation, and monitoring will be required for transformation to be successful (Joyce et al. 2009).

Species extinctions and loss of ecosystem services caused by climate change are likely more susceptible to extreme events and climate variability than changes in climate means (Seppälä et al. 2009). Extreme events are inherent in climate variability (Rummukainen 2012); as disturbances they shape ecosystems (Sprugel 1991; Turner 2010). Increases in frequency, intensity, and duration of heat waves and heavy precipitation are expected under climate change. Droughts are projected to increase in different parts of the world, including central and southern Europe, central North America, Central America and Mexico, northeast Brazil, southern Africa, Australia, and Southeast Asia (Dai 2011; IPCC 2012; Rummukainen 2012; Thornton et al. 2014). Many wind-related disturbances such as tornadoes, thunder storms, and derechos are significant for forest ecosystems (Peterson 2000); they occur, however, at a small scale and are not yet represented in either GCMs or RCMs (Regional Climate Model) (Diffenbaugh and Field 2013; Hawkins et al. 2014; Kunkel et al. 2013; Mora et al. 2013).

Extreme events can create a window of opportunity for transformation (Pelling and Dill 2010), temporarily lowering institutional and social barriers to change (Nelson et al. 2007). Prolonged drought, insect outbreaks, wildfire, and wind disturbances that reach the level of a natural disaster (Stanturf et al. 2014b; Van Aalst 2006) provide impetus for transformational adaptation. Extreme events are expected to increase in frequency and intensity under climate change (Allen 2009; Allen et al. 2010; Cai et al. 2014; Meehl et al. 2005; Reichstein et al. 2013); even so, windows for transformational adaptation associated with extreme events likely will be narrow because the usual reaction is to restore to the familiar (Travis 2010).

A way forward

At the landscape level, little or no ability exists to reduce exogenous drivers of global change. Socio-ecological systems, however, are amenable to reducing vulnerability and the negative effects of global change (Füssel 2007; Heltberg et al. 2009; Sarewitz et al. 2003). Outcome vulnerability, a top-down analysis, and contextual vulnerability, a bottom-up analysis, are two different perspectives (Pielke et al. 2013; Weaver et al. 2013). The top-down approach works best when addressing well-constrained problems where probabilities can be attached to the likelihood of different outcomes (Lempert et al. 2004). Top-down climate change vulnerability assessments lead either to decision-making ahead of evidence or alternatively, deferring decisions until uncertainty is reduced by improved climate projections (Pielke et al. 2013). Although the latter “wait and see” approach defers decisions and potential costs of adaptive measures; it risks the danger that the climate system will reach or exceed a tipping point of irreversible change before convincing evidence emerges resulting in greater costs of sustaining SES and mitigating the damage (Adams 2013; Lenton et al. 2008).

Decision-making under uncertainty is a necessary part of life; managing risk, reducing vulnerability, and focusing on increasing adaptive capacity may be a better way to cope with climate change trends that currently may be undetectable but are nonetheless real (Kates et al. 2012; Pielke et al. 2013). Bottom-up analysis yields an assessment of contextual vulnerability (Fig. 2) by examining the multiple stressors affecting socio-ecological systems. Changes in exogenous drivers include climate, political and institutional structures, and socioeconomic structures (Fig. 2). The lingering effects of historical events that confer an exceptional nature on a landscape are recognized as landscape legacies including ecological memory (Parrott and Meyer 2012; Turner 2010).

Fig. 2
figure 2

Contextual vulnerability is a bottom-up approach that includes exogenous (political and institutional structures, economic and social institutions, and climate) and endogenous (biophysical, socioeconomic, institutional, and technological) drivers of land use change in landscapes that have legacies and ecological memory (Adapted from Pielke et al. 2013)

Full size image

Modeling capabilities will improve over time, but because people will change behavior in light of projections and new experience (e.g., extreme events), thus coupled SES are unlikely to become truly predictable (Liu et al. 2007; Parrott and Meyer 2012). Despite this uncertainty, our accumulated silvicultural and ecological knowledge provides the ability to act now and react in the future as more information becomes available (Keenan 2015; Lindner et al. 2014; Park et al. 2014; Pinkard et al. 2015). In the meantime, adapting to climate change requires improved coping strategies or reduced exposure to known threats (Moser and Ekstrom 2010; Wilby and Dessai 2010). Regardless of adaptation strategy (Tables 2 and 3), improved coping ability will require strategic and institutional flexibility and structured feedback (monitoring and evaluation) to facilitate course changes (Choi et al. 2008; Dow et al. 2013; Hobbs et al. 2006).

Integrate landscape restoration with climate change mitigation and adaptation

A central premise has been that the magnitude of degraded and deforested landscapes is best approached at a landscape scale. Landscape restoration incorporates consideration of all land uses, not just forests, which adds considerable complexity (Lamb et al. 2012; Lindenmayer et al. 2008; Sayer et al. 2013). Land use dynamics reflect current demand for land; future demand will increase for agriculture, forestry, energy, and conservation resulting in more intense competition (Lambin and Meyfroidt 2011; Smith et al. 2010). In addition to large areas of already degraded landscapes, the effects of climate change on SES pose additional threats to future forests. Integrating attempts to restore landscapes and mitigate and adapt to climate change may synergistically increase adaptive capacity.

Forest landscapes are a linked socio-ecological system (Locatelli et al. 2011; Ostrom and Cox 2010). Potential synergies from linkages among mitigation, forest adaptation, and social adaptation to climate change can be illustrated at a local level by a community adjacent to a protected area that participates in a carbon benefit scheme such as REDD+ (Fig. 3). A mitigation planting connects the protected area to remnant forests outside through a corridor, enhancing connectivity. Forest adaptation measures are crucial to ensuring permanence of carbon fixed in mitigation forests (Galik and Jackson 2009; Hurteau et al. 2008) and may increase carbon sequestration in native forests (inside and outside the protected area) through improved forest management. The local community benefits from payments for carbon and other ecosystem services. In addition to payments, the carbon project has provided other benefits in the interim including improved agricultural seed and technical assistance to increase crop yields that reduce pressure on native forests (Blay et al. 2008; Schelhas et al. 2010).

Fig. 3
figure 3

Linkages among mitigation, forest adaptation, and community adaptation to climate change illustrate a linked socio-ecological system in a landscape, with reciprocal benefits from mitigation and adaptation (Adapted from Locatelli et al. 2011)

Full size image

Restoration planning and evaluation

Many natural resources organizations have well-established strategic and operational planning frameworks to guide future activities (Oliver et al. 2012). Planning methods that incorporate risk and assume a stationary climate will be challenged by the uncertainty of future climate (Bettinger et al. 2013). The following presents some features that could be incorporated to better adapt to climate change (Fig. 2).

Building a conceptual model of the landscape as a complex system means bringing diverse interests to a common understanding of the system and the wider context of exogenous environmental and social variables. Formally modeling the system may be attempted but if resources are insufficient to construct a quantitative model, a simple diagram may suffice. The starting point is evaluating current conditions and identifying significant landscape components and their sensitivity to current climate stimuli (Millar 2014). The next step is determining what changes in climate would have a significant effect on important landscape components (Daron et al. 2015; Pielke et al. 2013). These steps are iterative and may involve linking climate models and resource models, realizing this increases the uncertainty of projections (Maslin and Austin 2012; Wilsey et al. 2013). The estimated likelihood of these changes occurring should accompany recommendations to decision makers and stakeholders (Pielke et al. 2013).

Projections of future climate at an appropriate scale are necessary (Daron et al. 2015) and an ensemble of several (>10) AOGCMs gives better estimates of future conditions than using one or more “best” models (Mote et al. 2011). Current AOGCMs are better than older models at hindcasting historical temperature and precipitation at finer temporal and spatial scales (Sakaguchi et al. 2012; Wilsey et al. 2013). Publically available downscaled projections reduce the technical capacity needed to incorporate climate change projections into restoration planning (Girvetz et al. 2009; Groves et al. 2012). Nevertheless, expertise is needed to interpret model output and dynamical downscaling (rather than statistical) will be necessary to provide adequate spatial resolution in tropical areas where meteorological data are sparse or lacking.

Well-defined expectations are a hallmark of successful restoration (Stanturf et al. 2014a, b) and include in addition to the desired endpoint, the mechanism and trajectory of change (Burton 2014; Dey and Schweitzer 2014; Toth and Anderson 1998). Historic conditions as endpoints (e.g., reference sites) or regaining historical trajectories of ecosystem development as a guide (SERI 2004; Suding et al. 2015) may not be adapted to future conditions (Balaguer et al. 2014; Millar 2014; Stanturf et al. 2014b; Stein et al. 2013). A diversity of forest and non-forest conditions may be best suited to meet multiple social needs (Oliver et al. 2012; Sayer et al. 2013). Restoration is the opposite of degradation and indicators of successful restoration can be the reverse of degradation indicators (Stanturf et al. 2014b). Positive restoration indicators may be increasing such as forest area maintained or enlarged or decreasing such as encroachment or fragmentation (Table 4). Some indicators are relevant at several scales (landscape, stand, and species) or singly. The indicators in Table 4 relate to current conditions (reversing past degradation) and many apply to adaptation to future climate but additional indicators are needed that focus on traits and adaptations of individual species to altered climate and to the non-forest elements in the landscape.

Table 4 Restoration indicators, ▲ indicates an increase and ▼ a decrease in an indicator but both are positive outcomes (Adapted from Stanturf et al. 2014a)
Full size table

Trajectories and intermediate states are components of defined expectations (Toth and Anderson 1998). Key variables can be used for a temporal monitoring and evaluation system (Hutto and Belote 2013) within a spatial hierarchy (species, stand, landscape), stratified by significant habitats or landforms (Herrick et al. 2006; Sayer et al. 2007). The monitoring system would document successful adaptation and detect emerging threats. New information arising would be evaluated to determine if additional or innovative actions were needed because adaptation had not met expectations or altered conditions had changed vulnerability. Additional approaches to reduce risk from climate change and other impacts will likely arise and merit adoption (Kates et al. 2012).

A major question remains: when, or rather how quickly to act? Scenario planning captures the uncertainty surrounding the impacts of decisions and choices (Peterson et al. 2003). Characterizing when climate is likely to become sufficiently different so that action is needed is one way to set trigger points for pursuing more aggressive strategies (Travis 2010). Incremental adaptation is the default mode in restoration and initially a “no-regrets” approach (Heltberg et al. 2009) is the rational choice. Managers need to consider how actions affect the adaptive capacities of landscape components (Millar et al. 2007; Stephens et al. 2010). Managing for a portfolio of stand composition and structures across the landscape provides the flexibility to intervene and adapt to future conditions (Ando and Mallory 2012; Crowe and Parker 2008; Millar et al. 2007; Yemshanov et al. 2013). Stanturf et al. (2014a) provided a summary of available restoration methods; summarized by restoration objectives and starting points.

Restoration interventions

Cost-effective and low-cost methods are preferred in order to restore as much area as possible but without sacrificing benefits that could be secured by using more intensive methods, at least in some parts of the landscape (e.g., Stanturf et al. 2001). Passive restoration is a low-cost (but not necessarily free) method and useful where appropriate (Zahawi et al. 2014). Importantly, the long time needed for passive restoration to affect visible change might be seen as failure and cause premature termination of the effort or interpreted as abandonment and an invitation to encroachment (Zahawi et al. 2014). Significant opportunity costs of passive versus active restoration are incurred by the delay in delivering ecosystem services and foregone benefits (e.g., Stanturf et al. 2001).

Native recolonization of non-forest land or natural regeneration in degraded forests are lower-cost alternatives to more intensive planting methods but require adequate seed sources, advance regeneration, or sprouts on-site or within effective dispersal distance (Ashton et al. 2014; Chazdon 2008; Vieira and Scariot 2006). Low-intensity methods are unreliable, however, if threatened by ungulate herbivory or competing vegetation. Recolonization and natural regeneration rely on locally adapted genetic material, thus losing the opportunity to introduce new provenances or species better adapted to changing climatic conditions. Natural regeneration can be augmented by planting or sowing a desired species mix or stem density and in the event, non-local material more adapted to changing climate can be introduced (Stanturf et al. 2014a). If local climate is likely to remain stable or change slowly, however, species may have sufficient phenotypic plasticity or genetic variation to acclimate or adapt through natural selection (Corlett and Westcott 2013; Keenan 2015; Park et al. 2014). Conversely, species already in decline or at risk because of isolation or under attack by introduced pests are poor candidates to survive climate change and likely will require active adaptation (Aitken et al. 2008; Jump and Peñuelas 2005).

Moving individuals of a species to locations where it has not occurred in the Holocene remains controversial (e.g., Laikre et al. 2010). Variously termed assisted migration, managed relocation, or assisted colonization, this strategy may be anticipatory or transformational, depending upon the distance a species is moved (Dumroese et al. 2015). Assisted migration is considered replacement in functional restoration (Stanturf et al. 2014a). Controversy aside, a key question is when and where to move species. Answers will vary over time as climate and landscape conditions change (Williams and Dumroese 2013). Methods are needed to identify species at risk, identify suitable habitat, and prioritize translocations. Assessing contextual vulnerability (Fig. 2) requires combining genetic information, bioclimatic models, and downscaled climate projections, augmented by historical records and experimental evidence from provenance testing (Rout et al. 2013; Williams and Dumroese 2013). This approach is information intensive and best applied for a defined region; a similar approach can be used to broadly identify where species at risk are likely to be found (Watson et al. 2013). In highly degraded ecoregions (the target of forest landscape restoration) likely to undergo significant climate shifts, a combination of assisting out-migration of species at risk and in-migration of climate-adapted species will be needed.

Implementing a decision to translocate a species requires using seed sources with the lowest risk of maladaptation and inbreeding or outbreeding depression over time (Breed et al. 2013; Williams and Dumroese 2013). A framework using climate change projections and genetic and environmental information was developed by Breed et al. (2013) to guide decisions about deploying provenances under various levels of uncertainty. Predictive provenancing is suitable for high-value species because of the cost of acquiring necessary data from provenance growth trials and population genetics (Crowe and Parker 2008). Uncertainty of climate projections, however, suggests that the risk of failure increases beyond 2050 (Williams and Dumroese 2013). Admixture provenancing (mixtures of seeds drawn from large populations across different environments) is the suggested strategy when genetic data are lacking and climate conditions are uncertain (Breed et al. 2013; Williams and Dumroese 2013). Sites to test provenancing strategies are needed (Breed et al. 2013) and degraded forest landscapes provide ample widely dispersed habitats for experiments.

Long-distance assisted migration raises questions, including whether it is worth the cost, financially and in terms of potential ecological risks at the introduction site (Hewitt et al. 2011). Rout et al. (2013) provided a quantitative approach to these questions as well as questions about the costs associated with the failure of the introduction and extinction risk of the source population. Extensions to their method could include the probability of dispersal and spontaneous colonization by the target species. Even in situations where insufficient information makes it uncomfortable to apply probabilities to stochastic events, they suggested that decisions based on an explicit, structured framework are preferable to ad hoc decision-making (Rout et al. 2013).

Implementing change

Implementing change is not free or immune from resistance by entrenched interests and institutional and behavioral barriers that favor the status quo (Kates et al. 2012). There are limits to adaptation by social actors, whether the focus is on individuals, businesses, or governments (Dow et al. 2013; Moser and Ekstrom 2010). The uncertainties about risks and benefits are compounded by the differing perceptions of risk by different actors (Klinke and Renn 2002) that influences their willingness to act (Dow et al. 2013), which is also determined by their ability to implement change and the amount of influence they have to overcome barriers (Moser and Ekstrom 2010). Nevertheless, climate variability and extreme events will require institutions to be proactive and flexible (Dow et al. 2013; Gupta et al. 2010). Adaptation should reduce risks to things of value and keep damage to tolerable levels (Dow et al. 2013).

Institutional inertia includes the generally ponderous, top-down, hierarchical nature of bureaucratic processes (Game et al. 2014). Policy sluggishness and regulatory inflexibility inhibit innovation (Keenan 2015) and guidelines, best practices, and tacit or explicit rules of thumb that are based on practices and experience proven successful in the past but may become maladaptive under future conditions (Kates et al. 2012). Nevertheless, aversion to using non-native species (Shackelford et al. 2013), stressing collection from local sources (Breed et al. 2013; McKay et al. 2005), or the debate about assisted migration (McLachlan et al. 2007; Pedlar et al. 2012; Williams and Dumroese 2013) are flashpoints in climate change adaptation. Within organizations, norms emerge from frames of reference that will be maintained, even in the face of new evidence that runs counter to these norms (Berkhout et al. 2006). Organizations appear more receptive to climate change adaptation when core business or core competencies are engaged (Berkhout et al. 2006). Individuals may respond in similar fashion. For example, Blennow et al. (2012) found that personal experience of perceived local climate change or its effects explained attitudes of private forest owners toward adaptation. Forestry professionals in Germany were likewise receptive to undertaking adaptive measures (Yousefpour and Hanewinkel 2015). In contrast, conservation professionals were less receptive (Hagerman and Satterfield 2014). Similarly, studies of individuals and their attitudes toward climate change have shown that political persuasion is more predictive than knowledge (Kahan 2015).

Social resistance to anticipatory and transformational adaptation is likely if actions contravene popular beliefs and values (Moser and Ekstrom 2010). For example the naturalness paradigm still dominates policy and public perception of restoration goals (Stanturf et al. 2014b). Native species, historic structure, and natural disturbance regimes are widely held tenets of ecological restoration (Burton and Macdonald 2011; Hobbs 2013) as well as close-to-nature silviculture (Brang et al. 2014). Another facet of the naturalness paradigm is the aversion to deploying genetically modified organisms (GMO) into native ecosystems (Laikre et al. 2010; Strauss and Bradshaw 2004). The ability to develop plant materials better adapted to new climatic conditions will be severely constrained if GMOs cannot be widely deployed (Dumroese et al. 2015).

The uncertain costs of climate change adaptation have been identified as a barrier (Moser and Ekstrom 2010) and what we know of the costs of large-scale restoration is daunting. For example, the 20+ year project to restore the Everglades ecosystem in Florida, USA has cost an estimated $13.4 billion. Guldin (2013) provided another example; he speculated on the financial hurdle of changing species composition of forests in the southern USA, based on a history of planting degraded land and converting naturally regenerated stands to pine plantations. Briefly, a century was required to convert a fourth of the 100 million ha of forest for commercial purposes; the approximately $US16 billion spent to convert 250,000 ha annually on private land was under reasonably certain conditions (Guldin, 2013). This estimate did not account for the millions invested in research and development by public and private entities (e.g., Stanturf et al. 2003). Generally, there is little credible information on average costs of restoration (Holl and Howarth 2001). The cost of implementing the Bonn Challenge of restoring 150 million ha by 2020 (Minnemayer et al. 2011) is estimated at US$18 billion per year, with a return of US$84 billion per year to the global economy (Menz et al. 2013). Meeting global biodiversity conservation needs is estimated to cost US$76.1 billion annually (McCarthy et al. 2012).

Stratagems to overcome barriers

In order to move beyond incremental adaptation, we need “risky” research that pushes the boundaries of knowledge and accepted practice. Silvicultural and ecological knowledge of important relationships underlying successful adaptation to current climate can be the basis for speculative experiments that test the limits of important variables (Park et al. 2014). For example, warming experiments would provide needed information on the thermal tolerance of valued species. No doubt many of these experiments will fail, but even a few successful outcomes would provide needed information on species adaptation and appropriate management systems (Bormann and Kiester 2004).

More information about available provenances is needed as well as breeding programs to develop entirely new material. Modifying seed transfer zones and common garden experiments with commercial tree species are underway in the USA and Canada (Park et al. 2014; Williams and Dumroese 2013) and some European countries have considerable experience with non-native species (Bolte et al. 2009a; Isaac-Renton et al. 2014). Past efforts have concentrated on current climate or small increases in average temperature (Bolte et al. 2009b; Park et al. 2014) and more speculative experiments are needed to match the higher limits of projected warming (>5 °C). Moving provenances greater distances will have limited success for species that have limited genetic variation (Aitken et al. 2008) or have been decimated by non-native pests (Jacobs et al. 2013). Developing new climate-resilient plant material, through traditional breeding or genetic modification (Borland et al. 2015); will provide needed adaptations to warmer and drier conditions as well as defenses against novel pests (Dumroese et al. 2015).

Current species distribution models can be improved by incorporating more information on a host of species traits (Fitzpatrick and Hargrove 2009; Keenan 2015; Travis et al. 2013) and even better if more information is available at the genotype or provenance level (Bolte et al. 2009a, b). Better models will improve prioritizing species and habitats targeted for assisted migration (McKenney et al. 2007; Pedlar et al. 2012; Williams and Dumroese 2013). The ability to design adapted species mixtures in neo-native and novel ecosystems will depend upon trait-based approaches with species and provenances. Because climate sensitivity is likely to be greatest in the regeneration phase, ameliorating harsh conditions for seedling survival and establishment using underplanting or nurse crops could be included in speculative experiments testing provenancing strategies (Breed et al. 2013; Park et al. 2014).

Future landscapes pressured by radically different climate, twice as much humanity to feed and clothe, and greater global interconnectedness that facilitates the spread of innovation as well as threats to natural systems, requires an unprecedented effort to restore past degradation and avoid future maladaptation. Perceptions of the future based on projections of the past are no longer tenable. Producing significant research results is important for academic advancement and science-based management but does not guarantee that the information produced will enter into policy formulation or management decisions. Methods to improve science-based adaptation include transdisciplinary teams with embedded managers and stakeholders (Pennington et al. 2013), communities of practice (Barlow et al. 2011; Hendriks et al. 2012) or knowledge hubs (Bidwell et al. 2013). In some cases, establishing research sites with an eye toward demonstrating how innovations compare to current best practices can yield unexpected benefits (Gardiner et al. 2008).