Overcoming conceptual hurdles to accurately represent trees as cohorts in forest landscape models

Abstract

Forecasting the response of vegetation to climate change and human actions at global to landscape scales often relies on system models. Many such models represent the composition and demography of vegetation using age class cohorts of species or vegetation types. Forest landscape models (FLM) simulate spatially explicit forest dynamics in response to disturbance and abiotic drivers and constraints at landscape spatial and temporal scales (105 – 108 ha, hundreds of years), and many represent trees as species-age cohorts on a grid of cells to make the model tractable at landscape scales. However, it has become evident that complications arise when calculating cell-level summaries involving multiple cohorts because one cannot assume that all cohorts simultaneously occupy the entire physical space of a cell. A fundamental assumption of a widely used FLM (LANDIS-II) is that cohorts are homogeneously distributed across the grid cells that they occupy. It is only implicitly assumed that the stems (trunks) and crowns (foliage) of the individual trees that make up the cohorts physically occupy some proportion of the cell's area with a non-clustered distribution, but the model does not track those proportions. In reality, when cohort biomass is reduced by disturbance or turnover (self-thinning), gaps are created that are at least temporarily unoccupied. In current versions of LANDIS-II, such gaps are ignored. We describe a new approach for cohort-based FLMs to estimate those proportions (including open space) and apply them to compute more realistic estimates of light attenuation and cell-level estimates of cohort and total cell biomass. This approach accounts for gap dynamics in a FLM where stems are not explicitly modeled, allowing cohorts to partially fill cells. These improvements allow more accurate simulation of forest “gap” dynamics in cohort-based landscape models to produce more accurate absolute biomass estimates at species, cell and landscape scales.