A Step-by-Step Guide to Initialize and Calibrate Landscape Models: A Case Study in the Mediterranean Mountains

Abstract

The use of spatially interactive forest landscape models has increased in recent years. These models are valuable tools to assess our knowledge about the functioning and provisioning of ecosystems as well as essential allies when predicting future changes. However, developing the necessary inputs and preparing them for research studies require substantial initial investments in terms of time. Although model initialization and calibration often take the largest amount of modelers’ efforts, such processes are rarely reported thoroughly in application studies. Our study documents the process of calibrating and setting up an ecophysiologically based forest landscape model (LANDIS-II with PnET-Succession) in a biogeographical region where such a model has never been applied to date (southwestern Mediterranean mountains in Europe). We describe the methodological process necessary to produce the required spatial inputs expressing initial vegetation and site conditions. We test model behaviour on single-cell simulations and calibrate species parameters using local biomass estimations and literature information. Finally, we test how different initialization data—with and without shrub communities—influence the simulation of forest dynamics by applying the calibrated model at landscape level. Combination of plot-level data with vegetation maps allowed us to generate a detailed map of initial tree and shrub communities. Single-cell simulations revealed that the model was able to reproduce realistic biomass estimates and competitive effects for different forest types included in the landscape, as well as plausible monthly growth patterns of species growing in Mediterranean mountains. Our results highlight the importance of considering shrub communities in forest landscape models, as they influence the temporal dynamics of tree species. Besides, our results show that, in the absence of natural disturbances, harvesting or climate change, landscape-level simulations projected a general increase of biomass of several species over the next decades but with distinct spatio-temporal patterns due to competitive effects and landscape heterogeneity. Providing a step-by-step workflow to initialize and calibrate a forest landscape model, our study encourages new users to use such tools in forestry and climate change applications. Thus, we advocate for documenting initialization processes in a transparent and reproducible manner in forest landscape modelling.

Impact of climate change on forest resources: Case of Quercus rotundifolia, Tetraclinis articulata, Juniperus phoenicea, J. oxycedrus, J. thurifera and Pinus halepensis

“…specific species-areas and harvesting regimes, post-harvest planting), the spatial level (i.e. the resolution of the cells) is user-defined, making it very flexible and adaptive to a large variety of simulation experiments (Suárez-Muñoz et al, 2021). In LANDIS-II, the terrain is divided into ecoregions, which are sub-regions sharing similar climatic conditions and soil characteristics.…”

“…The aim is to define for each forest species composing the Ourika Forest massif, certain characteristics describing its behaviour with respect to well-defined ecological and climatic conditions. The identification of these characteristics is a combination of two sources of information, namely the bibliography as the first source (Pausas et al, 2004;Valladares, 2005;Niinemets et al, 2006;Serrada et al, 2008;Kattge et al, 2020;Suárez-Muñoz et al, 2021), local data (www.try-db.org/TryWeb/ home.php) and expert opinion as a second source to complete the bibliography.…”

Effects of outliers on remote sensing‐assisted forest biomass estimation: A case study from the United States national forest inventory

“…Applications of NFI data include ecological indicators of climate change (Horn et al, 2018; Knott et al, 2020; Will‐Wolf et al, 2017; Zhu et al, 2012); below‐ground mycorrhizal associations with tree species (Averill et al, 2022; Carteron et al, 2022; Jo et al, 2019); invasion biology of plants, pests and pathogens (Baer & Gray, 2022; Fei et al, 2019; Iannone et al, 2016); and biodiversity patterns and processes (Chirici et al, 2012; Fei et al, 2018; Winter et al, 2012; Zeller et al, 2018), among others. In addition to characterizing observed ecological patterns and processes, NFI data have also supported the parameterization of ecological models including spatially explicit landscape models (Purves et al, 2007; Suárez‐Muñoz et al, 2021; Wang et al, 2013), tree growth models (Lichstein et al, 2010; Giebink, DeRose, et al, 2022; Heilman et al, 2022) and species distribution models (Crookston et al, 2010; Iverson et al, 2019; Prasad et al, 2020), which aim to predict forest responses to changing climate and disturbances.…”

“…In addition to characterizing observed ecological patterns and processes, NFI data have also supported the parameterization of ecological models including spatially explicit landscape models (Purves et al, 2007;Suárez-Muñoz et al, 2021;Wang et al, 2013), tree growth models (Lichstein et al, 2010;Heilman et al, 2022) and species distribution models (Crookston et al, 2010;Iverson et al, 2019;Prasad et al, 2020), which aim to predict forest responses to changing climate and disturbances.…”

Forest Landscape Model Initialization with Remotely Sensed-Based Open-Source Databases in the Absence of Inventory Data

“…Thus, the most commonly used FLMs require information about tree species and age classes. Collecting such information of acceptable accuracy over large areas is a complex task, often requiring a combination of data from different sources [19]. With the increase in the study area size, it becomes more difficult to collect field data, and sometimes inaccessible due to the geographical characteristics and low transport accessibility of the territory.…”