Introduction

During this century, the resilience of many ecosystems is likely to be challenged by climate change and other anthropogenic drivers such as land-use change, pollution, and overexploitation of natural resources (Settele et al 2014). Mountain ecosystems are considered particularly sensitive and vulnerable to climate change (Körner et al 2005; Valdivia et al 2010; Field et al 2014), and an increasing number of people depend on their upstream and downstream services (Huddleston et al 2003). Mountain regions in developing countries are recognized to be among the most exposed to the impacts of climate change (McDowell et al 2013). There is growing concern about how to cope with and adapt to climate change in fragile mountain ecosystems, particularly in countries with substantial mountainous areas (Grêt-Regamey et al 2012). The recently published research priorities report of the Programme of Research on Climate Change Vulnerability, Impacts and Adaptation (PROVIA 2013) also stresses the importance of understanding adaptation to climate change in specific human and natural systems such as mountain regions.

Understanding adaptation in coupled human and natural systems requires understanding of the degree and nature of observed and projected climate impacts, current and future vulnerability, and the adaptive capacity of a country or region (Field et al 2014). As documented in the Intergovernmental Panel on Climate Change's (IPCC's) fifth assessment report (Field et al 2014), the scientific community has increasingly addressed the topics of impacts, vulnerability, and adaptive capacity, and its commitment to investigating these topics has been influential at all policy levels. Adaptation strategies in fact are often based on, or triggered by, scientific knowledge on climate systems, impacts, vulnerabilities, and adaptation (Biesbroek et al 2010). Scientific information and knowledge can lead to an effective response to adverse impacts of climate change, thus reducing vulnerabilities and enhancing adaptive capacities (Patwardhan et al 2009). This can be especially helpful in those areas most exposed to the impacts of climate change such as mountain regions.

McDowell et al (2014), in a review of adaptation actions in mountain regions, highlighted that synthesis work on climate change impacts and adaptation actions is paramount for increasing the effectiveness of adaptation policy. However, this review also revealed that there is limited documentation of the impacts of climate change on human systems and adaptation in mountain regions. Biesbroek et al (2010), in a comparative synthesis of national adaptation strategies in 10 European countries, found that only 2 countries, Germany and Spain, specifically prioritize or plan to prioritize mountain zones in their adaptation policy.

In their analysis of the types of knowledge needed to support adaptation policies in general, Hanger et al (2012) pointed out the importance of impact and vulnerability assessments, socioeconomic analyses of adaptation, and assessment of the costs of adaptation measures, as well as the need to reduce uncertainty. Similarly, Patwardhan et al (2009) emphasized that understanding current and future vulnerability and its interactions with climate stresses requires more studies on exposure, sensitivity, adaptive capacity—all 3 components of vulnerability as defined in the IPCC's fourth assessment report (Pachauri and Reisinger 2007)—and their interaction with environmental, social, and economic conditions.

Numerous studies have analyzed national adaptation policies in European and other developed countries and the mechanisms by which they were formulated (Gagnon-Lebrun and Agrawala 2006; Massey and Bergsma 2008; Swart et al 2009), but they only briefly discuss what scientific contributions shaped such documents (Swart et al 2009). A more global comparison focused on needs and gaps in adaptation research in mountain countries is currently lacking (McDowell et al 2014; Huggel et al 2015). Although important progress has been made in investigating the interface between science and policy in climate adaptation (Vogel et al 2007; Dilling and Lemos 2011; Hanger et al 2012; Lemos et al 2014), less attention has been devoted to analyzing scientific contributions to adaptation policy with the aim to address research needs and gaps (Hanger et al 2012; Huggel et al 2015).

In this study we analyzed the scientific knowledge that supported the development of nation- or statewide adaptation policies in 8 locations (7 countries and 1 US state) that represent different mountain regions and socioeconomic conditions. We performed a comparative analysis of Austria, Bhutan, Colombia, Germany, Nepal, Peru, Switzerland, and Washington State, USA, which we chose through purposive sampling in July 2012. We opted for a case-study design in order to obtain a contextualized and comparative analysis of the selected locations (McDowell et al 2014). We also looked for any mention in each policy of scientific knowledge specific to mountain areas and possible issues associated with the complex topography of these areas, conducted a questionnaire survey, and reflected on knowledge gaps from our own expert natural science perspective. We are aware that the development of adaptation strategies and measures is a multidimensional process, ideally involving multiple actors from science, governmental and nongovernmental institutions, and civil society, and that in addition to exploring the natural science knowledge base, it is important to explore gaps in the social sciences and analyze how climate change adaptation policies are negotiated and agreed upon at different levels of social organization. But that goes beyond the scope of the present paper and would require a different research approach.

Framework for data analysis

Data were collected from national or state climate change adaptation policies available in a language known by the authors (English, German, or Spanish) and through questionnaires submitted to policy-makers responsible for these policies. Data were gathered from policy documents in an inductive way: the documents were read and then information was systematically selected (Altheide et al 2008; Bryman 2012). The information was selected based on the objectives of the present paper, namely to find out what scientific knowledge contributed to the strategies and what research needs could be identified. The research needs were identified based on explicit statements in the policy documents we reviewed, additional findings from an in-depth qualitative document analysis of those documents (Altheide et al 2008), and the informants' responses to the questionnaire.

Finally, we looked for recurrent topics influencing the scientific knowledge relevant to the strategies. In this context, we recognized that the approach to adaptation policy development influences the way scientific knowledge contributes to the policy (Addor 2015). In particular, the choice between a top-down or bottom-up approach dictates how scientific knowledge supports the development of the strategy. A top-down approach starts from a scientific perspective (assessment of global climate change and its impacts) and proceeds to assess vulnerabilities at a regional or sectoral level. A bottom-up approach starts from the local socioeconomic context and vulnerabilities (Swart et al 2009; Addor et al 2015). A third possibility is to use a top-down approach but attempt to integrate different degrees of participation at the local level; we labeled this approach inclusive.

To analyze the contribution of scientific knowledge to climate-change adaptation strategy, as well as remaining research needs, we classified research contributions in 3 broad areas (adapted from Swart et al 2009):

In this context, we defined vulnerability, following the IPCC's fourth assessment report (Pachauri and Reisinger 2007: 883), as “the degree to which a system is susceptible to, and unable to cope with, adverse effects of climate change, including climate variability and extremes.” We are aware that more recent IPCC reports have shifted away from this conceptualization to a more risk-based approach (Field et al 2014). In this new conceptualization, vulnerability and exposure are 2 separate components of risk to climate extremes and climate-related events. However, at the time the adaptation policies were developed, the prevalent vulnerability paradigm followed the IPCC'S fourth assessment report.

The documents were screened to identify topics pertaining to mountain areas. The focus was mainly on natural science topics, partly because of our academic background and partly because social science topics rarely occurred in the documents in relation to mountains.

Results

Contributions from science and perceived research needs

This section focuses in turn on the 3 categories of scientific knowledge mentioned earlier: climate system research, impact assessments, and vulnerability assessments. Results are summarized in Table 1 with additional information provided in Supplemental material, Appendix S1; ( http://dx.doi.org/10.1659/MRD-JOURNAL-D-15-00016.S1) and further tables.

TABLE 1 

Summary of the contribution of scientific knowledge on climate system research to countries' adaptation policies, research needs identified, and mountain-specific issues associated with this body of knowledge.^a) (Table continued on next page.)

a) 

HISTALP, Historical Instrumental Climatological Surface Time Series of The Greater Alpine Region; SENAMHI, Peruvian National Service for Meteorology and Hydrology; CRU, Climatic Research Unit of East Anglia.

b) 

More details on the GCMs and RCMs are provided in Supplemental material, Appendix S1; ( http://dx.doi.org/10.1659/MRD-JOURNAL-D-15-00016.S1).

c) 

Research needs include gaps identified in the policy documents themselves, in replies to the questionnaire administered during this study, and based on the authors' assessment. Mountain-specific issues were identified during qualitative document analysis.

Climate system research: This area refers mainly to climate models and their results (Table 1). General circulation models (GCMs) are numerical climate models that provide a comprehensive representation of the global climate system at relatively course spatial resolution. In order to obtain results at the regional or local scale, 2 downscaling methods are available: (1) dynamical downscaling (or regional climate models [RCMs]), where a climate model of higher resolution (RCM) is run over a limited area, driven by output from a GCM, and (2) statistical downscaling, where local climate features based on observations are empirically related to large-scale climate features taken from a GCM. Few of the countries analyzed for this study calculated their own RCM. Most countries used combined results from multiple GCMs informed by the Special Report on Emissions Scenarios (SRES) (Nakicenovic and Swart 2000) and Solomon et al (2007) (which contains a useful description of the scenarios discussed in this paper).

The SRES emission scenarios most often used were A1B, followed by A2, B1, and finally B2. In most cases at least 2 SRES scenarios were used, usually 1 each from the A and B scenario families. The reasons for considering multiple scenarios included the ability to (1) explore the impacts of a wide range of greenhouse-gas emission concentrations and (2) assess the impacts of business-as-usual fossil fuel emission scenarios as well as more optimistic scenarios of reduced emissions and the consequent reduced temperature increases. Moreover, the choice of the A2 scenario seems to have been due to its closer agreement with current emission trends.

Because none of the SRES scenarios had a probability assigned, the use of multiple GCMs and SRES scenarios (as done by all countries in the study) made it possible to assess ranges of uncertainties inherent in the emission scenarios and the climate models. Switzerland additionally used the emission scenario RCP3PD (CH2011 2011), which was developed for the Coupled Model Intercomparison Project Phase 5 of the latest IPCC assessment (Stocker et al 2013), illustrating an emission concentration pathway that stabilizes carbon dioxide-equivalent concentration near 450 parts per million by the end of the century, and as such corresponds to a limitation of global warming to less than 2°C over the preindustrial level (van Vuuren et al 2007).

All countries had a clear need for high-resolution spatial climate model output, as reflected in their considerable use of both statistical and dynamic downscaling approaches (RCMs). The need for high-resolution models was of particular importance in mountain settings. Small-scale and complex topography significantly influence meteorological processes and lead to a high spatial variability in climate variables as well as regional differences in the rate of change of these variables. In addition to high-resolution climate model data, it is also important to have reliable long-term climate observations in order to validate the climate models and to predict future trends (Kotlarski et al 2014). In countries with extensive and remote mountain areas, such as Bhutan, Colombia, Nepal, and Peru, we infer a clear need for reliable long-term climate observations and a robust scientific baseline. This is typically a challenging task, as the maintenance of a sufficiently dense observational network for reliable long-term monitoring is costly and challenging (in terms of technical issues and accessibility) (Salzmann et al 2012, 2015). This is also reflected in the research needs identified by the informants or mentioned in the policy documents, namely limitations in coverage, availability, and accessibility of current and past climate data (Bhutan, Colombia, Nepal, and Peru) and limitations in regional climate scenarios for specific regions of the world (Nepal and Bhutan).

Austria, Switzerland, and Washington State benefited from numerous reliable and long-term climate records and many RCM-based assessments, allowing a more robust analysis of the climate change signal versus the natural climate variability. A higher-resolution image of the mountain topography is useful for a better projection of climate impacts at scales relevant to planners and managers (local to regional), although uncertainty is not per se reduced with higher resolution (Hallegatte 2009). In Peru, the lack of meteorological and hydrological data was recognized as particularly important for mountain regions. In Bhutan, corrections for elevation led to mismatching outputs of the 2 GCMs used, as reported in the policy document.

Impact assessments: This area refers to assessments of the observed and projected impacts of climate change on natural and human systems (Table 2). The majority of the countries employed qualitative assessments or index/indicator-based assessments. Qualitative assessments often refer to expert judgments based on results of climate system dynamics assessments, combined with knowledge of previous impacts in the country, usually related to climate variability and extremes (OcCC 2007). Index-based assessments blend climate system information in the form of indices (calculated values used to describe the state of and changes in the climate system) with proxies for public health, ecosystem integrity, and other key factors (eg Fischer and Schär 2010; Buytaert et al 2011). Impact assessments are usually (eg in Austria, Bhutan, Switzerland, and Washington State) carried out at the sectoral level, which means exploring impacts of climate change on the different sectors of the economy. Washington State is the only one of the entities studied to have developed its own impact models for several economic sectors.

TABLE 2 

Summary of the contribution of impact assessments to countries' adaptation policies, research needs identified, and mountain-specific issues associated with this body of knowledge.^a)

a) 

Research needs include gaps identified in the policy documents themselves, in replies to the questionnaire administered during the study, and based on the authors' assessment. Mountain-specific issues were identified during qualitative document analysis.

In Nepal, Bhutan, and Peru, local perceptions also constituted part of the formal impact assessment. Typically, interview-based studies evaluate how local people perceive changes in their environment, as related to climate change (Byg and Salick 2009; McDowell and Hess 2012).

In Austria, Germany, Peru, and Washington State, impacts were assessed specifically for mountain regions or mountain ecosystems. In Bhutan's National Adaptation Plan for Action, a chapter is dedicated to assessments of impacts on glaciers and the consequences of these impacts on people's wellbeing and security. In Switzerland, impacts and their consequences in the Alpine regions are highlighted throughout the adaptation strategy. However, in Colombia and Nepal, there was no assessment of impacts in mountain regions or in sectors relevant to their welfare and security.

This study's analysis of research needs found a clear concern in at least 3 countries about the lack of impact assessments at the local to regional scale. In Bhutan and Nepal, the research needs were mainly related to climate system dynamics. In Peru and Switzerland, the absence of socioeconomic impact assessments was a reason for concern. In Germany and Switzerland, information on interconnections between impacts and cross-sectoral assessment of impacts was identified as a significant knowledge gap. In Washington State, current data resources and scenarios were recognized as inadequate to assess impacts at scales relevant to planners and decision-makers.

Vulnerability assessments: This area refers to assessments of how the impacts of climate change will affect human and natural systems (Table 3). Half of the countries analyzed in this research based their vulnerability assessment on concepts outlined in the IPCC's fourth assessment report (Pachauri and Reisinger 2007), which defines vulnerability as a function of exposure, sensitivity, and adaptive capacity (Füssel and Klein 2006; Pachauri and Reisinger 2007). Colombia, Germany, and Nepal took a more quantitative approach to vulnerability assessment, combining indices and proxies for exposure, sensitivity, and adaptive capacity to derive a measure for vulnerability. In Nepal, vulnerability maps were produced by using geographic information system (GIS)-based products.

TABLE 3 

Summary of the contribution of vulnerability assessments to countries' adaptation policies, research needs identified, and mountain-specific issues associated with this body of knowledge.^a)

a) 

Research needs include gaps identified in the policy documents themselves, in replies to the questionnaire administered during the study, and based on the authors' assessment. Mountain-specific issues were identified during qualitative document analysis.

Switzerland's vulnerability assessment, unlike the others, was not based on the IPCC's fourth assessment report's definition but on risks at the district level. This requires combining the climate hazard areas with the impact areas, where impact areas are assessed for different climate scenarios and socioeconomic situations (Holthausen et al 2013).

Peru uses a semiquantitative approach, blending a qualitative vulnerability assessment with physical vulnerability maps. Bhutan and Nepal include local perceptions in their vulnerability assessments, in particular to evaluate adaptation capacities and sensitivities.

With regard to gaps in vulnerability assessment, some informants identified a lack of regional, local, and socioeconomic assessments. In the case of Bhutan and Nepal, the informants and the policy documents underlined gaps with regard to community perception, autonomous adaptation, and vulnerability assessment through participative methods, in addition to the already mentioned gap in climate systems knowledge.

The mountain aspect in the vulnerability assessments mainly relates to assessing vulnerability to human and natural systems in sectors specific to mountain areas, such as water management, forest management, and mountain-related hazards (slope instabilities, rock and snow avalanches, glacier lake outbursts). Bhutan devoted a chapter to impact and vulnerability assessments related to glacier changes. Germany and Washington State each had a short mountain-related section (focused on the region in Germany's case and on the ecosystems in Washington State's case).

Discussion

Some of the study countries had a full adaptation strategy in place and were moving toward implementation and monitoring, whereas others were still in, or had just started, the process of collecting information that could be used to develop a strategy. Looking at knowledge about climate system dynamics and associated climate impacts, Germany and Washington planned to use more integrated and quantitative forms of assessment, whereas Bhutan, Colombia, Nepal, and Peru were mainly concerned with broadening their data coverage.

The scarcity of spatial and temporal data is a common drawback in climate and climate adaptation research, exacerbated in mountain areas where the climate is more variable and physical access and infrastructure often less developed, representing a challenge for long-term environmental monitoring (see for example Salzmann et al 2012). Quality assurance procedures at various points of the data chain are also greatly needed in mountain areas, together with standardized methods to account for missing data (Negi et al 2012). Efforts made to treat available observational data so that they become reliable and long-term time series are ongoing (eg Schwarb et al 2011), and the use of alternative data sets such as reanalysis data, interpolated data sets, or satellite-based products are an option when applied with care and prior evaluation, in particular for climate variables such as precipitation (eg Scheel et al 2011; Salzmann et al 2015).

In addition to scarcity of data, a clear lack of scientific capacity, funding, and institutional capacity is recognized for developing countries. Local universities and research centers often lack academic and technological skills that are essential for translating data outputs (from observations and models) into viable knowledge that supports policy-making (Huggel et al 2009, 2015; Salzmann et al 2009, 2014).

To bridge these gaps in natural science knowledge, some developing countries have pursued an approach that draws on the perceptions and perspectives of local people to estimate vulnerability dimensions. The assumption is that a better understanding of the adaptive capacity of individuals and communities (as a dimension of vulnerability) helps to reduce the knowledge gaps at the local and regional level (Nakashima et al 2012). Combining local knowledge (also termed traditional or indigenous knowledge) with other science findings for the purpose of climate adaptation planning is increasingly popular and a thriving field of scientific activity (Reyes-García et al 2015). Countries with abundant natural science knowledge to contribute to their adaptation strategies may also benefit from integrating local knowledge in adaptation planning in full consideration of the different social and institutional contexts (Leonard et al 2013).

In that context, it is important to recognize that local knowledge is based on its own values and epistemology, which may be independent of findings from natural sciences (Berkes et al 2000; Berkes 2009; Ford et al 2012). In terms of climate impacts, there is a higher level of confidence if both local knowledge and scientific studies detect the same phenomena, as for instance reported for changing water resources in the Andes of Peru (Bury et al 2011). In turn, inconsistencies or discrepancies between the 2 knowledge sources may indicate higher uncertainties and be related to different spatial or temporal scales. For instance, local knowledge refers to the effects of climate variability rather than climate change, as long-term changes are often difficult to perceive at the local level. An exception relevant for mountain countries may be glaciers and snow, whose changes have been reflected in local knowledge in the Andes, Himalaya, and other mountain regions (Byg and Salick 2009; Carey 2010; Brugger et al 2013).

Top-down inclusive approaches were preferred in the majority of the countries analyzed in this study. Swart et al (2009) suggested that a reason for this preference could be that the strategies tend to respond to concerns about global climate change and thus the adaptation policies are developed at the highest possible scale of competence, which is usually the national level. A problem with this approach is that the propagation of uncertainty along a model chain can quickly become important (Ludwig et al 2013). Several informants reported the challenges of obtaining robust climate signals from the climate projections in their mountain areas, and this was also pointed out in some of the policy documents. This could constitute a limitation in the assessment and subsequent choice of adaptation options. Moreover, RCMs and other downscaling methods do contain uncertainties and errors, which may still be too high for effective adaptation planning at a level relevant for managers and planners (Wilby and Dessai 2010). Bhave et al (2014) suggested blending top-down and bottom-up approaches in order to deliver locally relevant adaptation policies, whereas Ludwig et al (2013) suggested a new approach in which adaptation strategies are developed based on current and future risks and then evaluated under future scenarios in order to test their robustness.

All 3 categories of scientific knowledge analyzed here have informed the development of adaptation strategies and policies. However, in the third area, vulnerability assessment, some studies have highlighted that relying too much on climate system dynamics and climate impacts might not be fully representative of future changes (Hallegatte 2009; Wilby and Dessai 2010). Knowledge of vulnerabilities, and associated exposures and coping capacities of coupled human and natural systems, is relevant to understand future changes (Field et al 2014). However, uncertainties in this body of knowledge can be larger than those in RCMs and need to be taken increasingly into account when formulating adaptation policies.

In response to this, a burgeoning body of literature has suggested that alternative paths to adaptation exist in the form of no-regret strategies (which yield benefits even without climate change), flexible strategies (which keep the cost of making wrong predictions as low as possible), “soft” strategies (institutional and financial solutions such as insurance, early warning systems, and capacity building) (Hallegatte 2009), and adaptive management (solutions based on the ability to change management practices in response to new scientific knowledge and insights) (Pahl-Wostl 2007). These approaches can be especially relevant in mountain regions, where the interactions of a complex topography with several interrelated social and environmental vulnerabilities exacerbate uncertainties (Beniston 2005).

Conclusions

Our analysis of national adaptation policies has shown that, at an initial stage, efforts were made to understand climate dynamics and develop emission scenarios. Observational records offered important sources of information against which to compare the output of climate models. Progress in the use of RCMs and other downscaling techniques allowed a number of countries to identify a more robust climate signal in their mountain areas, where topography often limits the performance of climate models. In these (mostly industrialized) countries, this stage has almost reached a saturation point and research efforts are now being devoted to impact and vulnerability assessments. These countries are quickly moving toward improvement in quantitative vulnerability and risk assessments, improved impact predictions at smaller scales, and the integration of socioeconomic issues. These improvements are highly relevant for mountain areas where the interaction of multiple climatic stresses and diverse socioeconomic circumstances often makes adaptation more challenging than it is in lowland areas (McDowell et al 2014). Observational records are also being improved in Colombia, Nepal, and Peru, and there is hope that this will trigger more research funding and activities.

The following main conclusions can be drawn from this comparative case study analysis:

The discussion of different types of knowledge raises the fundamental question of what knowledge production paradigms can contribute to delivering sustainable adaptations to growing environmental stressors in mountain regions. This is still an open question that needs full consideration by the research community.