Deviating from the Plan: Assessing the Impact of Forest Management Delays on Ecosystem Function

Forests are under increasing stress due to changes in disturbance regimes, such as wildfire and pest or disease outbreaks, an increase in more severe and prolonged drought, and changes in land use. These stressors are already having an observable impact on forests in the western United States. Many forests within the western US are managed by the US Forest Service. Forest management is important as a tool for increasing a forest\u27s ability to withstand or recover from these stresses. Additionally, because of the forest’s influence on interactions between the land surface and the atmosphere, forest management has implications for future energy, water, and carbon cycles. However, management is driven by socio-economic, political, and ecological needs which can influence the timing of management activities. Forests are dynamic ecosystems, and changes to the timing of management through delays could lead to unanticipated impacts on a forest’s structure, productivity, and ecohydrological function. Land surface models (LSMs) are one tool used to investigate land surface processes and land-atmosphere interactions. LSMs represent vegetation dynamics in different and increasingly sophisticated ways. While the fidelity of plant biophysical and biogeochemical process representation has increased in many of these models, the representation of forest management is still very simplistic. Until recently, the temporal aspects of management have rarely been included in studies using LSMs. Here, we addressed this challenge by including the temporal details of representative timber harvest activities from the western USA within LSM simulations. We hypothesized that changes in the timing of management activities can have long term impacts on the structure and functioning of a forest. To test this hypothesis, we quantified vegetation management activities in the western USA and investigated the role specific project characteristics have on potential project delays. As a proof-of-concept, we used this data to inform the timing of single point scale logging simulations using the Functionally Assembled Terrestrial Ecosystem Simulator of the Community Land Model (CLM-FATES) for a ponderosa pine dominated forest in southern Idaho. A challenge encountered in simulating realistic forests was a bias towards smaller-diameter trees (i.e., \u3c 50 cm diameter at breast height), relative to observations. In overcoming this challenge, we expanded on current work within the CLM-FATES and greater LSM community to better parameterize the model for temperate, evergreen forests. We developed methods to generate multiple parameter ensembles and simulated these ensembles under different climate forcing and coexistence conditions. Over the course of this work, we developed significant and important overarching findings about critical facets of simulating managed forest ecosystems. First, we found that environmental regulations (here the type of NEPA analysis required for a project), the length of time to complete that analysis, and the type of management activity had the biggest impact on the probability of project implementation. Second, from logging simulations, we found that the timing of treatments can have long term impacts on the resulting forest size-structure, but timing has less of an impact on the long-term functioning of the forest as indicated by the model. Third, more complex ecosystems – as represented by the addition of an additional plant functional type – can lead to more realistic distributions of tree size classes, although this added ecosystem complexity does not appear to assist in identifying optimal sets of model parameters in CLM-FATES. This work makes an important contribution to deploying sophisticated, demography based LSMs in western US forests by demonstrating how models can now capture legacies of human interventions and that calibration of model parameters is complex and constrained by the existing structure and makeup of these systems. These results highlight the importance of representing different aspects of human systems in ecosystem models as well as highlighting the tension between a need to improve model calibration without increasing model complexity