Theme 4 | Evaluation of Various Forest Management Approaches Under Global Stressors

A great challenge when studying forests is that they evolve at a very slow pace. While a forest fire may burn every tree in a couple of days, the subsequent re-growth will take decades, or even centuries to get back to how the forest was before the fire. Our decisions in forest management are therefore bets that we place on a long, and increasingly uncertain future. In this context, how can we be as certain as possible that our decisions will have a positive impact on forests ?

That’s where computer models come into the picture. Many models aiming to simulate the dynamics of forests over long time periods have been developed in the scientific literature since computers have become widely available. These models have become more realistic as time went by, using the opportunity offered by advancements in forest inventory data, and knowledge about the physiology of trees. Today, computer models can simulate entire forest landscapes for centuries, showing us how they may evolve under the combined effect of natural disturbances (fire, disease, insect epidemic, windthrow, etc.), human disturbances (harvesting, land-use change) and climate change.

To assess the long-term and large scale effects of the management recommendations highlighted in themes 1-3, we will use the LANDIS-II model to simulate the forest landscapes of several of our research sites. The goal here is to measure as precisely as possible how the forest of these landscapes will change according to the different management strategies tested, and to extract further guidance on how to manage Canadian forests at large in the future.

The LANDIS-II model was first released in 2007, and has been updated ever since through the persistent work of the non-profit organization in charge of its development, the LANDIS foundation. The model is open-source and free of use, with all its source code available to anybody wanting to contribute to its development. It is also modular, allowing users to choose among a large catalog of extensions representing the dynamic of forests and disturbances in the model.

One of the most important extensions in LANDIS-II is the one that simulates the “succession” of trees, meaning tree growth  and natural mortality at the stand level (in the absence of disturbances). In our case, we will use the PnET succession extension, which uses equations derived from an ecophysiological model called PnET. The PnET extension is one of the most detailed extensions, requiring many parameters regarding the physiology of the tree species that are simulated (such as evapotranspiration rate, photosynthesis rate, etc.). But this complexity will be crucial to properly simulate the relationship between the functional diversity of forests and their dynamics.

In addition to simulating the succession of trees, we will simulate several natural and human disturbances that often affect Canadian forests using different LANDIS-II extensions : forest fires, insect outbreaks , tree diseases, windthrows, forest cuts, etc. We will then simulate three forest management scenarios : Business As Usual (BAU), with an implementation of current forestry practices; Climate Smart Forestry (CSF), a new type of forestry that focuses on favoring tree species expected to be adapted to future disturbances and climate conditions; and Functional Complex Network (FCN), an approach that will build  on the work of theme 3 and that aims at maximizing the functional diversity of individual forest stands as well as the  functional diversity across the network of forest stands via the dispersal of seeds.

We will begin with the parameterization of the LANDIS-II extensions that we will use. Such a process is a long one, as each extension can require hundreds of different parameters which have to be obtained, verified, and sometimes calibrated in order to produce realistic results. In addition, we will have to characterize the initial composition of the landscapes that will be used as inputs at the beginning of the simulations. This requires working with large datasets of forest inventory data coming from different provinces of Canada, with their own particularities, formatting and sampling methods.

Following the parameterization of LANDIS-II for each study area targeted, we will design  the simulation scenarios that will vary the forest management strategy used  as well as  the scenarios of climate change and disturbances. To do this, we will build on the work carried in the other research themes to elaborate the Functional Complex Network management strategy. We will also study how forest management has been done in the past in our simulated areas to design realistic “Business as Usual” scenarios which will serve as control treatments to compare the FCN and CSF approaches.

Once all our simulations are ready to run, we will use the supercomputing clusters of the Digital Research Alliance of Canada to execute them since they will require extensive computing power. LANDIS-II simulations will generate a large quantity of results that we will carefully analyze and summarize for each scenario. This step will be crucial to detect and interpret how forests are expected to be impacted by the different management strategies, and to derive useful management recommendations.