Date Published: June 7, 2019
Publisher: Public Library of Science
Author(s): Eric S. Walsh, Tara Hudiburg, Paul Pickell.
Avian cavity nesters (ACN) are viable indicators of forest structure, composition, and diversity. Utilizing these species responses in multi-disciplinary climate-avian-forest modeling can improve climate adaptive management. We propose a framework for integrating and evaluating climate-avian-forest models by linking two ACN niche models with a forest landscape model (FLM), LANDIS-II. The framework facilitates the selection of available ACN models for integration, evaluation of model transferability, and evaluation of successful integration of ACN models with a FLM. We found selecting a model for integration depended on its transferability to the study area (Northern Rockies Ecoregion of Idaho in the United States), which limited the species and model types available for transfer. However, transfer evaluation of the tested ACN models indicated a good fit for the study area. Several niche model variables (canopy cover, snag density, and forest cover type) were not directly informed by the LANDIS-II model, which required secondary modeling (Random Forest) to derive values from the FLM outputs. In instances where the Random Forest models performed with a moderate classification accuracy, the overall effect on niche predictions was negligible. Predictions based on LANDIS-II simulations performed similarly to predictions based on the niche model’s original training input types. This supported the conclusion that the proposed framework is viable for informing avian niche models with FLM simulations. Even models that poorly approximate habitat suitability, due to the inherent constraints of predicting spatial niche use of irruptive species produced informative results by identifying areas of management focus. This is primarily because LANDIS-II estimates spatially explicit variables that were unavailable over large spatial extents from alternative datasets. Thus, without integration, one of the ACN niche models was not applicable to the study area. The framework will be useful for integrating avifauna niche and forest ecosystem models, which can inform management of contemporary and future landscapes under differing management and climate scenarios.
The structure and composition of forest ecosystems are expected to shift with climate-induced changes in precipitation, temperature , fire , carbon mitigation strategies [3,4], and biological disturbances . Specifically, climate change induced declines in tree species occurrence , shifts in forest carbon stocks , increases in forest mortality events , and increases in forest burn area  have been predicted. Forest composition and structure are integral to biodiversity , and the climate induced changes are likely to have wildlife biodiversity implications  especially for avifauna . For example, moderate to high severity fires can create open forest habitat, adequate snag density, and minimal mid-story vegetation for primary avian cavity nesters such as woodpeckers . Though climate models predict increases in area burned or fire intensity, which may increase habitat suitability for woodpeckers , tree species composition shifts via climate change may pose adaptation constraints on them . Integrating the feedbacks between wildlife, habitat, and forest processes in a modeling framework could improve our understanding of climate induced wildlife habitat changes and subsequent climate-forest adaptive management.
Ecological modeling focused on the effects of climate change and management scenarios (e.g., fire, carbon mitigation, harvest) on forest resiliency will need to account for the effects of these dynamics on wildlife habitat. Thus, to implement climate adaptive management strategies aimed at increasing or preserving wildlife species, modeling efforts will need to include the coupled response of vegetation and wildlife to climate change. We evaluated a framework for integrating ACN ecological niche and forest landscape models to improve ACN climate change niche modeling and provide ecological modelers with a means to account for wildlife measures in biogeochemical forest modeling.
The presented framework for the integration of ACN and forest landscape models based on the transfer of existing niche models is viable. Transferability was hindered by limitations such as model training region and study area landscape similarities in addition to forest landscape model output variables. We addressed these limitations through the criteria of selecting appropriate niche models, evaluating training and study area landscape similarities, and secondary modeling of niche model inputs from the forest landscape model outputs. The framework proved useful when niche models are not easily transferable to a landscape due to data constraints. LANDIS-II estimated spatially explicit landscape information (e.g., biomass distributions) that were unavailable from other datasets, and the framework included the process of comparing habitat suitability maps and underlying variables. This increased the application of avian niche models across a broad landscape improving habitat conservation information for land managers. Finally, this framework provides a process to ascertain species responses to climate change and management scenarios while providing forest ecosystem modelers with a means to account for wildlife species responses.