Overview
Global change including rapid shifts in historic climate and disturbance regimes is already having profound effects on the composition, structure, and function of forest ecosystems. Adaptive forest management aims to promote ecosystem resilience and adaptation in the face of these changing conditions. Effective adaptive management requires an understanding of how forests are likely to respond to novel conditions under a range of management scenarios (including no management). This is particularly challenging given the decadal-to-centennial time frames over which forest ecosystems evolve. Models are an important conservation tool as they allow us to explore changes in forest ecosystems that may not be realized within our lifetime.
Researchers within the Forest Dynamics Lab work to develop statistical models that can be used to advance understanding of forest responses to novel conditions and predict the outcomes of alternative management strategies. The goal of these models is to better inform adaptive management to conserve forest ecosystems under global change. We value actionable scientific outcomes. To this end, we work actively with forest practitioners and conservation managers to co-develop decision assistance tools informed by model predictions.
While specific models take different forms, they generally seek to approximate forest demographic processes (growth, mortality, regeneration) or their outcomes over time and/or space. We frequently use Bayesian hierarchical approaches given their flexibility in modeling complex ecological processes, their capacity to synthesize multiple datasets, and their ability to quantify and partition different sources of uncertainty. We use a range of forest data types including continuous forest inventory (e.g., Forest Inventory and Analysis), palaeoecological records (e.g., tree rings), functional trait information (e.g., leaf and wood traits), and remotely sensed observations (e.g., LiDAR). While we work almost exclusively with data that has already been collected, we sometimes conduct targeted data collection if it is needed for a given project.
Current research can be broken into three areas as defined below.
Researchers within the Forest Dynamics Lab work to develop statistical models that can be used to advance understanding of forest responses to novel conditions and predict the outcomes of alternative management strategies. The goal of these models is to better inform adaptive management to conserve forest ecosystems under global change. We value actionable scientific outcomes. To this end, we work actively with forest practitioners and conservation managers to co-develop decision assistance tools informed by model predictions.
While specific models take different forms, they generally seek to approximate forest demographic processes (growth, mortality, regeneration) or their outcomes over time and/or space. We frequently use Bayesian hierarchical approaches given their flexibility in modeling complex ecological processes, their capacity to synthesize multiple datasets, and their ability to quantify and partition different sources of uncertainty. We use a range of forest data types including continuous forest inventory (e.g., Forest Inventory and Analysis), palaeoecological records (e.g., tree rings), functional trait information (e.g., leaf and wood traits), and remotely sensed observations (e.g., LiDAR). While we work almost exclusively with data that has already been collected, we sometimes conduct targeted data collection if it is needed for a given project.
Current research can be broken into three areas as defined below.
Advancing models of forest dynamics
Models of forest dynamics are used to predict changes in forest ecosystems over time and space by approximating demographic processes including growth, mortality, and fecundity at an individual or population scale. These predictions can be used to quantify and assess forest outcomes under different management scenarios (e.g., carbon storage and sequestration, species compositional shifts). Ongoing research in this area is focused on integrating models of forest dynamics within spatial-temporal statistical frameworks that allow for improved uncertainty quantification and the synthesis of multiple forest datasets to inform predictions (inventory, LiDAR, LiDAR). Further, we are working to develop decision frameworks that allow for managers to account for uncertainty in future forest conditions when they are considering potential adaptive management strategies.
|
Examples of this work include:
Itter, M. S., & Finley, A. O. (2024). Toward improved uncertainty quantification in predictions of forest dynamics: A dynamical model of forest change. bioRxiv, 2024-07. doi.org/10.1101/2024.07.22.604669
Itter, M. S., & Finley, A. O. (2024). Toward improved uncertainty quantification in predictions of forest dynamics: A dynamical model of forest change. bioRxiv, 2024-07. doi.org/10.1101/2024.07.22.604669
Joint species distribution models for forest conservation
Joint species distribution models (JSDMs) model the occurrence or abundance of species as a function of fixed environmental effects and residual species correlations. They differ from species distribution models (SDMs) in that they jointly (simultaneously) predict a collection of species (a natural community) at a location accounting for the statistical dependences among co-occurring species. JSDMs are frequently used to predict how the spatial distribution of species may shift under climate change. Ongoing research in this area applies JSDMs to inform conservation management of forest and plant communities under global change. Specifically, we are using a Bayesian JSDM to identify spatial locations suitable for the introduction of blight-resistant, hybridized chestnut in eastern North America. Suitability is determined based on the predicted abundance of historically associated species under climate change scenarios. We recently started work on a new project sponsored by the Northeast Climate Adaptation Science Center applying a JSDM to identify native plant species predicted to maintain viable populations under climate change in the presence of common invasive species. These predictions will be used to provide conservation managers with location-specific lists of climate-adapted species to inform management of climate resilient plant communities.
|
Examples of this work include:
Itter, M. S., Kaarlejärvi, E., Laine, A. L., Hamberg, L., Tonteri, T., & Vanhatalo, J. (2024). Bayesian joint species distribution model selection for community‐level prediction. Global Ecology and Biogeography, 33(5), e13827. doi.org/10.1111/geb.13827
Kaarlejärvi, E., Itter, M., Tonteri, T., Hamberg, L., Salemaa, M., Merilä, P., Vanhatalo, J. & Laine, A. L. (2024). Inferring ecological selection from multidimensional community trait distributions along environmental gradients. Ecology, e4378. doi.org/10.1002/ecy.4378
Itter, M. S., Kaarlejärvi, E., Laine, A. L., Hamberg, L., Tonteri, T., & Vanhatalo, J. (2024). Bayesian joint species distribution model selection for community‐level prediction. Global Ecology and Biogeography, 33(5), e13827. doi.org/10.1111/geb.13827
Kaarlejärvi, E., Itter, M., Tonteri, T., Hamberg, L., Salemaa, M., Merilä, P., Vanhatalo, J. & Laine, A. L. (2024). Inferring ecological selection from multidimensional community trait distributions along environmental gradients. Ecology, e4378. doi.org/10.1002/ecy.4378
Forest demography under global change
Forest demographic rates including growth, mortality, and regeneration are already being modified by changes to the historic climate and disturbance regimes to which species are adapted. These demographic shifts have the potential to modify regional forest ecosystems in terms of their composition, structure, and function. We are actively working to better understand and predict changes in forest demographic rates and their impact on regional forests. Existing work in this area includes analyzing how tree growth rates vary in response to climate extremes and disturbance, how demographic responses to climate vary as a function of seed source (to better understand assisted migration potential), and predicting tree mortality rates under climate change scenarios. The knowledge gained through these analyses is subsequently used to refine models of forest dynamics to better predict forest change and inform adaptive management.
Examples of this work include:
Itter, M. S., Vanhatalo, J., & Finley, A. O. (2019). EcoMem: An R package for quantifying ecological memory. Environmental Modelling & Software, 119, 305-308. doi.org/10.1016/j.envsoft.2019.06.004
Itter, M. S., D'Orangeville, L., Dawson, A., Kneeshaw, D., Duchesne, L., & Finley, A. O. (2019). Boreal tree growth exhibits decadal‐scale ecological memory to drought and insect defoliation, but no negative response to their interaction. Journal of Ecology, 107(3), 1288-1301. doi.org/10.1111/1365-2745.13087
Examples of this work include:
Itter, M. S., Vanhatalo, J., & Finley, A. O. (2019). EcoMem: An R package for quantifying ecological memory. Environmental Modelling & Software, 119, 305-308. doi.org/10.1016/j.envsoft.2019.06.004
Itter, M. S., D'Orangeville, L., Dawson, A., Kneeshaw, D., Duchesne, L., & Finley, A. O. (2019). Boreal tree growth exhibits decadal‐scale ecological memory to drought and insect defoliation, but no negative response to their interaction. Journal of Ecology, 107(3), 1288-1301. doi.org/10.1111/1365-2745.13087