Climate change affects the structure and the multiple functions of forests, challenging forest managers to use innovative tools to evaluate and develop new viable alternatives for a sustained provision of goods and services. Particularly, the role of forest management today is to sustain the health and resilience of forest ecosystems in a way that the multitude of goods and services provided enhance the well-being of people [1
]. This role has been supported by numerous studies, which have generated knowledge around multiple forest ecosystem services (ESs) and their interrelations [6
], as well as how they evolve under climate change and interact with other natural and human-induced disturbances [7
]. In the realm of knowledge transfer, an effective way to deliver scientific information to forest practitioners is by providing user specific computer-based tools [9
]. Nevertheless, transfer to practice is one of the shortcomings identified in forest management tools developed by researchers [10
Most common software applications for projecting the evolution of forests are forest simulators—an implementation of mathematical models in computer programs that are able to predict the consequences of different courses of action in forestry [11
]. Thus, the triptych of assessing changing factors in forest stands is the combination of past knowledge, present observation and future projection. A number of different modeling approaches, along with computer tools, have been developed over the years, addressing at least one of these aspects. Early empirical models of growth and yield are still powerful instruments for forest managers [12
]. This approach uses data obtained from forest inventories to model statistical relationships between stands and tree attributes [13
]. Empirical models have low modelling complexity and high accuracy, which makes them adequate to address traditional forest management objectives [14
]. However, they are poor in explaining the underlying mechanisms [13
] and in predicting the harvesting effect on ecosystem structure and functions. Moreover, they lack the ability to make long term prediction under climate change scenarios, or extrapolate the growth of the trees for conditions different to those observed in the past, as their equations are fitted to historical data [14
On the other hand, the process-based models focus on the eco-physiological processes and their responses to external dynamics. These models are more flexible, and can explain the cause–effect, but are less able to predict forest yield [16
]. Process-based models are suitable for considering climate change and ecological objectives in forest planning [16
], as they embed ecological processes influenced by climate.
Another type of models are succession, or gap models (e.g., [17
]), which explicitly assess the impacts of temperature, water, and nutrients on the growth and development of trees, with the main goal to study and project the structural and compositional dynamics of forest ecosystems influenced by the environment. Therefore, these models (e.g., JABOWA, FORET [17
]) are able to assess the impacts of global change on long-term dynamics of forest structure, biomass, and forest composition [17
]. The hybrid models integrate elements of the previous modelling paradigms (i.e., environmental conditions and reliable growth estimations), and are used to predict forest dynamics (mortality, growth, regeneration, etc.) at different spatial scales [20
]. Specifically, the underlying idea of hybrid models is to benefit from the predictive ability of the calibrated data of empirical approaches, as well as from the explicit environmental dependence of process-based formulations, in order to offer reliable support in forest management and planning [21
]. The gap models and the hybrid models can be used in both short term timber-related management, and for long term plans, in the light of climate change [12
]. Yet, although forest simulators are considered to be the fundamental tools to be applied in forest management planning, they are rarely used by forest practitioners.
Current approaches to forest management supporting tools integrate forest data, growth models and decision algorithms into a decision support system (DSS) [24
]. Many authors have been contributing to different aspects of state of the art forest management DSSs by exploring new software architectures, objectives, or spatial scales [24
]. From the architectural point of view, the modular design is the predominant approach. Scale-wise, DSSs are divided into stand level, forest or landscape level systems, and systems for regional or national assessment [26
]. In regard to forest management objectives, traditional, timber-related management has been gradually replaced by more holistic approaches since the late 1980s due to the rise of new ideas such as forest ecosystem management, sustainable forest management, and adaptive forest management [27
]. More recently, the Millennium Ecosystem Assessment (MA) [2
] has brought to attention the concept of ecosystem services. First conceived in the 1990s [28
], this notion aimed to highlight the interrelation of ecosystems and human well-being. Nowadays, it is the main framework for sustainable ecosystem management and policy-making [29
Under this framework, forest communities are viewed as complex systems of interconnected ecosystem services (ESs) influenced by external factors such as natural disasters, human-induced disturbances, and climate change. Forest decision support tools (DSTs) need to embrace this approach by applying scientific knowledge in projecting changing factors and estimating their impacts on ESs. Respectively, we identified three major aspects to be considered in modern DS tools: climate change, ability to estimate multiple ESs, and risk integration.
In terms of climate change consideration, a plethora of process-based and hybrid models have been developed over the years. Several authors (e.g., [21
]) have reviewed the current state of these models, reporting the main strengths and weaknesses of each one. In their overview of the models used in Europe, Fontes et al. (2011) [21
] indicated that in most of the cases these models are even-aged and single-species, and able to evaluate biomass and carbon storage as well as drought as a natural disturbance. Sparingly, fire risk, storms, and soil erosion are taken into consideration. A recent review by Morán-Ordoñez et al. [33
] showed that most studies only evaluate a single ecosystem service, and calls for more integrative approaches that allow a more complete view of the ecosystems. In this regard, process-based models are more versatile and can be easily adapted for multiple ES assessment [21
A number of studies propose combining forest growth and ES models to assess the provision of multiple services. For example, Wikstrom et al. (2011) [26
] presented a DSS able to estimate recreation values, carbon sequestration, and habitat suitability by coupling growth models and ES models; Garcia-Gonzalo et al. [34
] extended the SADfLOR DSS [24
] to include trade-off analysis between timber production, cork, and carbon. In the matter of risk integration in forest management, first attempts were made in the early 1980s in North America. As a case in point, Martell [35
] and Reed [36
] analyzed the effects of fire risk on the optimal forest stand rotation. More recent approaches at both stand and landscape levels have examined the reciprocal interaction of natural risk impact and management regimes (e.g., [37
]). From the perspective of integrated systems, Hanewinkel et al. [39
] discussed the possibility of incorporating mechanistic and empirical storm risk models, as well as an empirical fire risk model into growth simulators, for assessing the impact of disturbances on forests. Later, Reyers et al. [8
] discussed six case studies in Europe where climate-sensitive growth and yield models were combined with risk assessment models in order to evaluate the joint impact of climate change and disturbances on forest production. Most of the integrated models presented in their study are able to assess ESs other than production, such as biodiversity and recreation.
The shortcoming of a great extent of the existing decision support tools is a lack of management-oriented approaches, or of development in close relation with managers [10
]. There is also a deficiency of integrated software solutions to address multiple ESs, including different sources of risk and uncertainty (e.g., forest fires) in the light of climate change. The majority of tools address ecosystem services at a limited level, and are mainly focused on services related to biomass production (e.g., timber and carbon sequestration) [33
]. Although research-oriented software, including forest simulators (see [26
]), may provide a comprehensive impact on ESs in a climate change context, they are applicable for use within the scientific community. As a result, forest managers will, most likely, use only a few DS tools that they feel comfortable with [41
]. Thus, new approaches, adapted for management purposes and developed in close cooperation with forest managers, are needed to fill the gap between research and forestry practice.
The challenge herein is to design a simple to use yet powerful forest DS tool that is able to assess the provision of ecosystem services in the context of global changes and incorporate risk assessment. In the present study we address this challenge by presenting a decision support tool (DST) for forest management planning that considers climate change and other sources of uncertainty, such as forest fires and storms, and delivers information on the provision of multiple goods and services. The development of the software followed a user-centered design, with forest managers being identified as the target audience. We outline the methodology in terms of the architecture of the embedded elements of the system, and we present the results through an example of stand simulation.
4. Discussion and Conclusions
In this paper we presented a user-centered decision support tool able to efficiently simulate and visualize the future of forest stands and assess multiple ecosystem services under different management options and climatic scenarios. The main challenge was to couple a forest simulator that is sensitive to climate change with multiple ecosystem services models, while assuring the usability of the system. Forest managers were defined as potential end users and were engaged in the development stage. The tool may be efficiently used to project the growth of stands under different climate scenarios and/or management alternatives. Yet, the user must be aware that there is uncertainty in the outputs when performing long-term projections (e.g., uncertainty in climate scenarios, uncertainty related to pest attacks and fire events).
We chose the SORTIE-ND forest dynamics model for a number of reasons: the model is climate-change sensitive, it is able to simulate mixed and uneven-aged stands, and it has been parameterized for 59 tree species in 11 different study areas around the world, including the main species in the montane forests of the Pyrenees (northeastern Spain). The latter makes it accessible for a broader range of users. In addition, it is distance-dependent, and the individual tree location is kept constant during the whole simulation, which can be of interest for future applications (e.g., the use of this system for training future foresters to make thinnings and observe their effects). Although the chosen model is able to simulate other species than the ones selected in the present work, and can be used in different geographical areas by using proper parameters, however, some ecosystem services models embedded in the system (e.g., mushroom production) are restricted to a specific spatial extent, which makes them inappropriate to be extrapolated to other regions.
The architectural paradigm, chosen for the software development, is the most suitable for integrating third-party software into an existing application, as is the case with integrating the SORTIE-ND simulator into our system. We followed a modular implementation, imbedded in a three tier architectural pattern.
Through a use case scenario, we illustrated the workflow of the system, but also defined the requirements, and identified the drawbacks and the added values. The scenario was tested on a stand located in northeastern Spain, composed of tree species calibrated for the SORTIE-ND model. In order to run a simulation with different species and/or for other geographical regions, parameters for these species have to be provided. One of the advantages of the developed system is its ease of use and clarity in the performed tasks, which was achieved by considering usability rules at the very beginning of the project and by performing evaluations subsequently. As indicated by Gordon et al. [10
], one of the oversights usually done by researchers that jeopardizes the use of decision support tools in practice is the lack of communication with potential end users at different stages of the development of the tools. We collaborated with the Montnegre Forest Owners Association for the user requirements stage, and also conducted a usability test on a sample of potential end users in the Forest Science and Technology Centre of Catalonia, using the SUS scale method. The results showed a score of 77.8 out of 100, which translates into “good”. The overall satisfaction questionnaire revealed a favorable result of 100%, where all the participants agreed (with responses “agree” and “strongly agree”) with the statements:
Overall, I am satisfied with the ease of completing the task(s).
Overall, I am satisfied with the amount of time it took to complete the task(s).
Most of the comments related to the complimentary questions were focused on the functionality of the system, which, at the time, was not yet applied. Suggestions, such as uploading files or generating tabular data, were implemented at a later stage of development. Since the session was focused mainly on the usability evaluation of the system, we concluded that efficiency, effectiveness, and satisfaction of interacting with the software was achieved to a sufficient level.
To sum up, the developed tool can be efficiently used by forest management practitioners for traditional objectives, as well as for more holistic purposes under climate change uncertainty and multiple ESs focus. While we specifically focused on practitioners target group, the tool has the potential to be used in facilitating education in the field of forest management. In terms of future work, we design to add more flexibility to the system, and expand the audience by integrating various forest dynamic models at different spatial scales, as well as by adopting more ecosystem services models for more geographical regions. We also aim to incorporate an optimization-based decision support module, thus providing an integrated approach to decision-making practices.