Final Report Summary - FORECOFUN-SSA (Assessing climate change impacts over large areas of primary forests in southern South America)
Concern about climate change is increasing due to the many uncertainties that exist regarding the possible impacts on ecosystems in the 21st century and beyond. Human activities, e.g. logging and land use changes, are driving changes in forest ecosystems worldwide, among others altering their biogeochemical functioning and threatening biodiversity. Predicting the impacts of climate change on forests is, therefore, one of the major challenges in global change research. Forest responses to climate change vary from region to region of the world. This requires us to analyze climate change impacts on forest functioning specifically for each region. Little is known on the impacts of climate change over large areas covered by primary temperate rainforests in southern South America (SSA; 37-43 °S). This project contributed towards understanding the impacts of climate change on forest functioning of primary temperate rainforests of SSA. To this end, a dynamic forest modelling framework was developed to analyze multiple, interacting effects of climate on forests at broad spatial scales.
An extensive field campaign was conducted for obtaining data on the structure and composition of temperate rainforests of SSA. These data further focused on identifying tree species traits that can account for major variations in the dominance of tree species in SSA along large-scale climatic gradients. The sampling design was based on advanced geospatial analyses of species distributions. The final database represents the largest and most detailed geospatial database to date for tree species traits and distributions in SSA. First, the analysis of the database was focused on deriving a small set of bioclimatic factors associated with ecological or physiological processes that can be used to predict species distributions. Second, the database was used to constrain parameters of dynamic forest models. Third, the database was used to link plant traits and bioclimatic variables to global and regional scales.
Models of forest dynamics were used to simulate forest composition in SSA at 1) the stand scale, to assess forest functioning under climate change scenarios, and 2) at the regional scale, to assess forest composition and species distributions. Model testing also involved uncertainties related to the spatial extrapolation to simulate range dynamics. The latter was addressed for temperate rainforests of a data-rich area (i.e. the Pacific Northwest of North America, PNW) were the model was comprehensively tested for stand-scale analyses.
Information on tree species presence, tree species traits and forest structure was obtained in 607 sites distributed in SSA. Traits information was obtained for a total of 46 tree species and 456 specimens. In parallel a species presence dataset was developed including 8831 records documenting the distribution of 71 tree species. The researcher contributed to a database that includes 21 plant traits from over 400’000 species-site combinations worldwide. This provided the first global quantification of the stronger correlation of mean annual temperature with plant traits than mean annual precipitation. In parallel, the analyses conducted in SSA documented patterns with implications for our understanding of the geographic distributions of tree species in this region, such as the role of the ratio of summer to annual rainfall as the most powerful predictor of tree species presence. In addition to the scientific results noted above, this research provided the most comprehensive parameter set for running simulations of forest dynamics in SSA, currently including 30 tree species.
The parameter set was used to perform the first spatially explicit, dynamic simulation of a forested landscape (20 km2) in SSA. Simulations showed that anthropogenic changes in fire regimes impact the resilience of the threatened conifer Pilgerodendron uviferum in this region. Furthermore, model testing efforts provided the first simulation of forest composition along broad-scale climatic gradients in SSA, showing that Andean species-rich forests are replaced eastwards by mono-specific woodlands and eventually by steppe, in agreement with field observations. Using the same modelling framework, research conducted in the PNW demonstrated the potential of using dynamic forest gap models to predict regional species distributions. However, results also illustrate a trade-off between predicting species distribution ranges (generality) while still capturing local forest composition accurately (specificity). Finally, the researcher developed a modelling framework to assess climate change impacts at local scales. Results from these simulations showed that drier climate will alter forest structure, leading to decreases in aboveground biomass by up to 73% of current values in primary old-growth forests in SSA.
As part of the training activities the researcher participated in internships in partner institutes, scientific workshops and soft-skill courses. Skills developed by the researcher were put into practice during the execution of this research. As part of the activities for continuing a career in research, the researcher successfully applied for a returning grant to his country of origin (Chile). Currently he occupies an Associate Professor position at Universidad Austral de Chile (Valdivia, Chile).
The expected final results and their potential impact and use are summarised below:
• Traits database for tree species of SSA: Plant traits, both morphological and functional, are essential to understanding and predicting the adaptation of ecosystems to global change. This database improves the empirical basis for such predictions. It is expected in the short-term to provide public access (via a website) for scientific and educational use.
• Validated parameter set for 30 tree species of SSA: Dynamic forest models commonly have many parameters that are uncertain or even completely unknown. The parameter set developed in this research contributes towards informing dynamic forest models in data-poor areas such as SSA. It is expected that this set allows for the prediction of ecosystem responses to global change in remote areas of the world. Additionally it provides a starting point for increasing the knowledge in rare and less-studied tree species that will reduce uncertainty in model predictions.
• Predictive tool for forest composition and dynamics of primary forests in SSA: This research evaluated the performance of three dynamic forest models in SSA. In addition, calibration schemes were elaborated to reduce parameter uncertainty. The experience gained contributes towards applying the models over large extents of primary forests in SSA. This research contributed to determine climate change threats on ecosystem services of primary forests in SSA and produced a high quality baseline data for future research.
Project public website
http://www.fe.ethz.ch/research/standdynamics/FORECOFUN-SSA/FORECOFUN-SSA
An extensive field campaign was conducted for obtaining data on the structure and composition of temperate rainforests of SSA. These data further focused on identifying tree species traits that can account for major variations in the dominance of tree species in SSA along large-scale climatic gradients. The sampling design was based on advanced geospatial analyses of species distributions. The final database represents the largest and most detailed geospatial database to date for tree species traits and distributions in SSA. First, the analysis of the database was focused on deriving a small set of bioclimatic factors associated with ecological or physiological processes that can be used to predict species distributions. Second, the database was used to constrain parameters of dynamic forest models. Third, the database was used to link plant traits and bioclimatic variables to global and regional scales.
Models of forest dynamics were used to simulate forest composition in SSA at 1) the stand scale, to assess forest functioning under climate change scenarios, and 2) at the regional scale, to assess forest composition and species distributions. Model testing also involved uncertainties related to the spatial extrapolation to simulate range dynamics. The latter was addressed for temperate rainforests of a data-rich area (i.e. the Pacific Northwest of North America, PNW) were the model was comprehensively tested for stand-scale analyses.
Information on tree species presence, tree species traits and forest structure was obtained in 607 sites distributed in SSA. Traits information was obtained for a total of 46 tree species and 456 specimens. In parallel a species presence dataset was developed including 8831 records documenting the distribution of 71 tree species. The researcher contributed to a database that includes 21 plant traits from over 400’000 species-site combinations worldwide. This provided the first global quantification of the stronger correlation of mean annual temperature with plant traits than mean annual precipitation. In parallel, the analyses conducted in SSA documented patterns with implications for our understanding of the geographic distributions of tree species in this region, such as the role of the ratio of summer to annual rainfall as the most powerful predictor of tree species presence. In addition to the scientific results noted above, this research provided the most comprehensive parameter set for running simulations of forest dynamics in SSA, currently including 30 tree species.
The parameter set was used to perform the first spatially explicit, dynamic simulation of a forested landscape (20 km2) in SSA. Simulations showed that anthropogenic changes in fire regimes impact the resilience of the threatened conifer Pilgerodendron uviferum in this region. Furthermore, model testing efforts provided the first simulation of forest composition along broad-scale climatic gradients in SSA, showing that Andean species-rich forests are replaced eastwards by mono-specific woodlands and eventually by steppe, in agreement with field observations. Using the same modelling framework, research conducted in the PNW demonstrated the potential of using dynamic forest gap models to predict regional species distributions. However, results also illustrate a trade-off between predicting species distribution ranges (generality) while still capturing local forest composition accurately (specificity). Finally, the researcher developed a modelling framework to assess climate change impacts at local scales. Results from these simulations showed that drier climate will alter forest structure, leading to decreases in aboveground biomass by up to 73% of current values in primary old-growth forests in SSA.
As part of the training activities the researcher participated in internships in partner institutes, scientific workshops and soft-skill courses. Skills developed by the researcher were put into practice during the execution of this research. As part of the activities for continuing a career in research, the researcher successfully applied for a returning grant to his country of origin (Chile). Currently he occupies an Associate Professor position at Universidad Austral de Chile (Valdivia, Chile).
The expected final results and their potential impact and use are summarised below:
• Traits database for tree species of SSA: Plant traits, both morphological and functional, are essential to understanding and predicting the adaptation of ecosystems to global change. This database improves the empirical basis for such predictions. It is expected in the short-term to provide public access (via a website) for scientific and educational use.
• Validated parameter set for 30 tree species of SSA: Dynamic forest models commonly have many parameters that are uncertain or even completely unknown. The parameter set developed in this research contributes towards informing dynamic forest models in data-poor areas such as SSA. It is expected that this set allows for the prediction of ecosystem responses to global change in remote areas of the world. Additionally it provides a starting point for increasing the knowledge in rare and less-studied tree species that will reduce uncertainty in model predictions.
• Predictive tool for forest composition and dynamics of primary forests in SSA: This research evaluated the performance of three dynamic forest models in SSA. In addition, calibration schemes were elaborated to reduce parameter uncertainty. The experience gained contributes towards applying the models over large extents of primary forests in SSA. This research contributed to determine climate change threats on ecosystem services of primary forests in SSA and produced a high quality baseline data for future research.
Project public website
http://www.fe.ethz.ch/research/standdynamics/FORECOFUN-SSA/FORECOFUN-SSA