Biology subjects

Itter, M. S.

Publications and source records attributed to Itter, M. S..

6 recordsLinked to original sources

bioRxiv2026

Integration of individual and population data to improve predictions of size-structured forest dynamics

Integral projection models (IPMs) are a powerful tool for predicting structured population dynamics under global change. Inverse calibration approaches allow IPMs to be fit using widely available population density data increasing potential applications. Yet population density data alone may not provide sufficient information for IPMs to accurately identify the environmental factors driving demographic rates leading to poor predictions under variable future conditions. We construct a Bayesian dynamical IPM framework to integrate population density with individual growth data to better predict size-structured population dynamics. The framework pairs an IPM process model with data models that control for individual growth variability and a mismatch in the resolution and precision of population density and individual growth data. The model is applied to a combination of experimental and simulated forest population data to assess its ability to predict size-structured population density and estimate underlying demographic rates. Predictions of size-structured population density were similar regardless of whether the dynamical IPM was provided both population density and individual growth data (integrated model) or population density data alone (population model). The population model, however, did not identify annual growth effects driven by weather variables including vapor pressure deficit leading to poor estimates of mean size-structured growth. Simulation trials under which the integrated model was provided varying numbers of individual growth records indicated that 10 records were sufficient for the model to estimate annual growth effects with near equivalent inference when 30 or more records were applied. Results highlight the potential for inversely calibrated IPMs to correctly predict structured population dynamics while incorrectly estimating underlying demographic rates. Integrating individual demographic data resolves this issue allowing for inference on growth responses to variable environmental conditions, thereby improving the ability of IPMs to predict structured population dynamics under global change. While individual demographic rate data is often limited, simulation results indicate that only a small number of individual records are needed for valid inference.

ecology↗

bioRxiv2025

Leveraging Light Detection and Ranging (LiDAR) to Elucidate Forest Structural Conditions that Influence Eastern Whip-poor-will Abundance

Eastern North American forests are degraded due to land use history and are threatened by numerous factors that further reduce their structural complexity, which contributes to population declines of many taxa. As such, many agencies and their conservation partners are employing habitat centric conservation efforts. Increased availability of airborne Light Detection and Ranging (LiDAR) data provides an opportunity to quantify fine-scale structural habitat characteristics for forest wildlife. One such species of conservation concern, the eastern whip-poor-will (Antrostomus vociferus), requires diverse forest structural conditions to meet its breeding season habitat requirements. We used airborne LiDAR data and autonomous recording units (ARUs) to identify elements of forest structure that influence whip-poor-will breeding season abundance in Pennsylvania, USA. Specifically, we applied a machine-learning classifier for whip-poor-will song to audio recordings obtained from 851 ARUs that were deployed in forested landscapes and then created daily detection histories to estimate whip-poor-will relative abundance. Whip-poor-wills were detected at 334 survey locations (41%). Abundance exhibited positive linear relationships with percent forest cover and percent oak forest and a negative linear relationship with percent impervious cover. Whip-poor-will abundance was also influenced by forest structure, with abundance exhibiting a quadratic relationship with two LiDAR-derived covariates; canopy heterogeneity and height within 300 m. Using these results, we predicted whip-poor-will abundance and habitat management potential. Whip-poor-will conservation in our study region will depend on public and private land efforts that maintain heavily forested, oak dominated landscapes that are managed using practices that increase canopy height diversity among and within stands.

ecology↗

bioRxiv2025

Connecting growth and yield models to continuous forest inventory data to better account for uncertainty

Models of forest growth and yield are frequently used to inform adaptive management decisions aimed at increasing forest resilience or promoting long-term carbon storage. Despite the increasing ecological detail represented in growth and yield models, there remains large variability (uncertainty) in predictions of forest dynamics under global change. Quantifying this uncertainty and accounting for it when making management decisions is integral to sustainable management in the face of changing conditions. However, the structure and complexity of modern growth and yield models make it challenging to quantify uncertainty and propagate it to predictions of forest dynamics under alternative management strategies. To address this challenge, we develop a Bayesian dynamical model informed by continuous forest inventory data that supports the quantification and propagation of uncertainty in predictions of forest dynamics at a stand scale. The model predicts the temporal evolution of the size-species distribution using a matrix projection process model approximating growth, mortality, and regeneration. Disturbance is integrated through its effect on the size-species distribution within a stand providing a flexible framework to represent adaptive management. We apply the model to long-term inventory data from the Penobscot Experimental Forest in Maine, USA to predict multi-decadal biomass dynamics under five alternative management strategies. Predictions are used to identify the management strategy maximizing live aboveground biomass growth and yield over the model period. We conclude by discussing the benefits and challenges of connecting the model to large-scale inventory data and how its predictions can be used to better inform adaptive management decisions.

ecology↗

bioRxiv2024

Toward improved uncertainty quantification in predictions of forest dynamics: A dynamical model of forest change

Models of forest dynamics are an important tool to understand and predict forest responses to global change. Despite recent model development, predictions of forest dynamics under global change remain highly variable reflecting uncertainty in future conditions, forest demographic processes, and the data used to parameterize and validate models. Quantifying this uncertainty and accounting for it when making adaptive management decisions is critical to our ability to conserve forest ecosystems in the face of rapidly changing conditions. Dynamical spatio-temporal models (DSTMs) are a particularly powerful tool in this setting given they quantify uncertainty associated with process-based models of forest demography, the parameters upon which those models depend, and the forest data used to inform them. Further, DSTMs propa-gate this uncertainty to predictions of forest dynamics allowing for its formal integration within adaptive management decision frameworks. A major challenge to the application of DSTMs in applied forest ecology has been the lack of a scalable, theoretical model of forest dynamics that generates predictions at the stand level--the scale at which management decisions are made. We address this challenge by integrating a matrix projection model motivated by the McKendrick-von Foerster partial differential equation for size-structured population dynamics within a Bayesian hierarchical DSTM informed by forest inventory data. The model provides probabilistic predictions of species-specific demographic rates and changes in the size-species distribution over time. The model is applied to predict long-term dynamics (60+ years) within the Penobscot Experimental Forest in Maine, USA, accounting for uncertainty in inventory observations, process-based predictions, and model parameters for nine Acadian Forest species. We find that variability in inventory observations associated with heterogeneous stand conditions drives uncertainty in predictions of forest dynamics. We conclude with a discussion of how the initial DSTM can be refined and extended to better represent forest dynamics under global change and inform adaptive management.

ecology↗

bioRxiv2023

Inferring ecological selection from multidimensional community trait distributions along environmental gradients

Understanding the drivers of community assembly is critical for predicting the future of biodiversity and ecosystem services. While trait-based frameworks are often used to this end, correlations among traits and the assumption of unimodal trait-abundance distributions confound detection of the underlying processes. To overcome these caveats, here we quantify multidimensional trait distributions of communities (community trait niches), which we use to identify ecological selection types shaping communities along environmental gradients. We find that directional, stabilizing, and divergent selection all modify community trait niches in over 3600 boreal forest understory plant communities, and selection on a particular trait may change from one type to another over time. Our results provide novel and rare empirical evidence from a natural system for divergent selection. The results also show that while higher trait diversity (measured as trait niche volume) is generally associated with higher species richness, high resource availability may enable tighter niche packing. Jointly our approach provides a framework for identifying key traits under selection and facilitates the detection of processes underlying community dynamics.

ecology↗

bioRxiv2022

Identifying and separating the processes underlying boreal forest understory community assembly

Joint species distribution models (JSDMs) are an important conservation tool for predicting ecosystem diversity and function under global change. The growing complexity of modern JSDMs necessitates careful model selection tailored to the challenges of community prediction under novel conditions (i.e., transferable models). Common approaches to evaluate the performance of JSDMs for community-level prediction are based on individual species predictions that do not account for the species correlation structures inherent in JSDMs. Here, we formalize a Bayesian model selection approach that accounts for species correlation structures and apply it to compare the community-level predictive performance of alternative JSDMs across broad environmental gradients emulating transferable applications. We connect the evaluation of JSDM predictions to Bayesian model selection theory under which the log score is the preferred performance measure for probabilistic prediction. We define the joint log score for community-level prediction and distinguish it from more commonly applied JSDM evaluation metrics. We then apply this community log score to evaluate predictions of 1,918 out-of-sample boreal forest understory communities spanning 39 species generated using a novel JSDM framework that supports alternative species correlation structures: independent, compositional dependence, and residual dependence. The best performing JSDM included all observed environmental variables and multinomial species correlations reflecting compositional dependence within modeled community data. The addition of flexible residual species correlations improved model predictions only within JSDMs applying a reduced set of environmental variables highlighting potential confounding between unobserved environmental conditions and residual species dependence. The best performing JSDM was consistent across successional and bio-climatic gradients regardless of whether interest was in species- or community-level prediction. Our study demonstrates the utility of the community log score to quantify differences in the predictive performance of complex JSDMs and highlights the importance of accounting for species dependence when interest is in community composition under novel conditions.

ecology↗