Geology topics

Katharine M. Banner

Publications and source records attributed to Katharine M. Banner.

10 recordsLinked to original sources

USGS2025

Bayesian model selection to investigate meaningful spatial scales

Ecologists and other statistical practitioners with access to high-resolution spatial data lack guidance on best approaches for discerning meaningful spatial scales for environmental covariates which is necessary when spatial factors influence environmental processes. Recently developed methods have attempted to automate investigating spatial scales for covariates by evaluating models for which potential explanatory variables are derived from concentric circles of increasing size centered at survey locations. However, these methods make a strong assumption on the inclusion of the covariate and do not help discern whether a covariate should be included in the model. We present an approach that utilizes researcher guidance to create informative priors on the model space that, along with parallelizable Reversible Jump MCMC techniques, enables efficient estimation of posterior model probabilities to assist with the choice of meaningful spatial scales for environmental covariates.

Authorea↗

USGS2025

Leveraging an observed-data likelihood improves the use of machine learning labels in a Bayesian hierarchical model for bioacoustic data

Classification of massive datasets by machine learning (ML) algorithms is promising for many scientific domains, especially wildlife monitoring programs that rely on passive acoustic surveys for detecting species. However, treating ML-predicted class labels (e.g., species identity) as truth biases inferences of focal parameters within common modeling frameworks. One solution is to model the misclassification process explicitly using human-validated true-class labels for a subset of observations. Validation by experts can present a substantial bottleneck in otherwise efficient workflows that use ML predictions. Bioacoustics practitioners seek guidance on both the quantity and process for selecting ML-labeled data to validate by an expert. We derive an alternative model formulation that jointly models human-validated and ML-predicted class labels with an observed-data likelihood (ODL) and use empirically informed simulations motivated by a real-data application to explore different probability designs for selecting class labels for validation. Simulation results suggest that with smaller validation sets the ODL formulation increases computational speed and reduces estimation error compared to a default MCMC data augmentation routine. Our methodology is transferable to applications that treat predictions from classification algorithms as the response variable of interest.

Annals of Applied Statistics↗

USGS2024

Clustering and unconstrained ordination with Dirichlet process mixture models

Assessment of similarity in species composition or abundance across sampled locations is a common goal in multi-species monitoring programs. Existing ordination techniques provide a framework for clustering sample locations based on species composition by projecting high-dimensional community data into a low-dimensional, latent ecological gradient representing species composition. However, these techniques require specification of the number of distinct ecological communities present in the latent space, which can be difficult to determine in advance. We develop an ordination model capable of simultaneous clustering and ordination that allows for estimation of the number of clusters present in the latent ecological gradient. This model draws latent coordinates for each sample location from a Dirichlet process mixture model, affording researchers with probabilistic statements about the number of clusters present in the latent ecological gradient. The model is compared to existing methods for simultaneous clustering and ordination via simulation and applied to two empirical datasets; JAGS code to fit the proposed model is provided in an appendix. The first dataset concerns presence-absence records of fish in the Doubs river in eastern France and the second dataset describes presence-absence records of plant species in Craters of the Moon National Monument and Preserve (CRMO) in Idaho, USA. Results from both analyses align with existing ecological gradients at each location. Development of the Dirichlet process ordination model provides wildlife managers with data-driven inferences about the number of distinct communities present across monitored locations, allowing for more cost-effective monitoring and reliable decision-making for conservation management.

Methods in Ecology and Evolution↗

USGS2024

Joint spatial modeling bridges the gap between disparate disease surveillance and population monitoring efforts informing conservation of at-risk bat species

White-Nose Syndrome (WNS) is a wildlife disease that has decimated hibernating bats since its introduction in North America in 2006. As the disease spreads westward, assessing the potentially differential impact of the disease on western bat species is an urgent conservation need. The statistical challenge is that the disease surveillance and species response monitoring data are not co-located, available at different spatial resolutions, non-Gaussian, and subject to observation error requiring a novel extension to spatially misaligned regression models for analysis. Previous work motivated by epidemiology applications has proposed two-step approaches that overcome the spatial misalignment while intentionally preventing the human health outcome from informing estimation of exposure. In our application, the impacted animals contribute to spreading the fungus that causes WNS, motivating development of a joint framework that exploits the known biological relationship. We introduce a Bayesian, joint spatial modeling framework that provides inferences about the impact of WNS on measures of relative bat activity and accounts for the uncertainty in estimation of WNS presence at non-surveyed locations. Our simulations demonstrate that the joint model produced more precise estimates of disease occurrence and unbiased estimates of the association between disease presence and the count response relative to competing two-step approaches. Our statistical framework provides a solution that leverages disparate monitoring activities and informs species conservation across large landscapes. Stan code and documentation are provided to facilitate access and adaptation for other wildlife disease applications.

Journal of Agricultural, Biological and Environmen↗

USGS2022

Statistical assessment on determining local presence of rare bat species

Surveying cryptic, sparsely distributed taxa using autonomous recording units, although cost-effective, provides imperfect knowledge about species presence. Summertime bat acoustic surveys in North America exemplify the challenges with characterizing sources of uncertainty: observation error, inability to census populations, and natural stochastic variation. Statistical uncertainty, if not considered thoroughly, hampers determining rare species presence accurately and/or estimating rangewide status and trends with suitable precision. Bat acoustic data are processed using an automated workflow in which proprietary or open-source algorithms assign a species label to each recorded high-frequency echolocation sequence. A false-negative occurs, if a species is actually present but not recorded and/or all recordings from the species are of such poor quality that a correct species identity cannot be assigned to any observation. False positives for a focal species are a direct result of the presence and incorrect identification of a recording from another species. We compare four analytical approaches in terms of parameter estimation and their resulting (in)correct decisions regarding species presence or absence using realistic data-generating scenarios for bat acoustic data within a simulation study. The current standard for deciding species presence or absence uses a multinomial likelihood-ratio test p value (maximum likelihood estimate [MLE]-metric) that accounts for known species misidentifications, but not imperfect detection and only returns a binary outcome (evidence of presence or not). We found that the MLE-metric had estimated median correct decisions less than 60% for presence and greater than 85% for absence. Alternatively, a multispecies count detection model was equivalent to or better than the MLE-metric for correct claims of rare species presence or absence using the posterior probability a species was present at a site and, importantly, provided unbiased estimates of relative activity and probability of occurrence, creating opportunities for reducing posterior uncertainty through the inclusion of meaningful covariates. Single-species occupancy models with and without false-positive detections removed were insufficient for determining local presence because of substantially biased occurrence and detection probabilities. We propose solutions to potential barriers for integrating local, short-term and rangewide, long-term acoustic surveys within a cohesive statistical framework that facilitates determining local species presence with uncertainty concurrent with estimating species–environment relationships.

Ecosphere↗

USGS2022

Coupling validation effort with in situ bioacoustic data improves estimating relative activity and occupancy for multiple species with cross-species misclassifications

The increasing complexity and pace of ecological change requires natural resource managers to consider entire species assemblages. Acoustic recording units (ARUs) require minimal cost and effort to deploy and inform relative activity, or encounter rates, for multiple species simultaneously. ARU-based surveys require post-processing of the recordings via software algorithms that assign a species label to each recording. The automated classification process can result in cross-species misidentifications that should be accounted for when employing statistical modelling for conservation decision-making. Using simulation and ARU-based detection counts from 17 bat species in British Columbia, Canada, we investigate three strategies for adjusting statistical inference for species misclassification: (a) ‘coupling’ ambiguous and unambiguous detections by validating a subset of survey events post-hoc, (b) using a calibration dataset on the software algorithm's (in)accuracy for species identification or (c) specifying informative Bayesian priors on classification probabilities. We explore the impact of different Bayesian prior specifications for the classification probabilities on posterior estimation. We then consider how the quantity of data validated post-hoc impacts model convergence and resulting inferences for bat species relative activity as related to nightly conditions and yearly site occupancy after accounting for site-level environmental variables. Coupled methods resulted in less bias and uncertainty when estimating relative activity and species classification probabilities relative to calibration approaches. We found that species that were difficult-to-detect and those that were often inaccurately identified by the software required more validation effort than more easily detected and/or identified species. Our results suggest that, when possible, acoustic surveys should rely on coupled validated detection information to account for false-positive detections, rather than uncoupled calibration datasets. However, if the assemblage of interest contains a large number of rarely detected or less prevalent species, an intractable amount of effort may be required, suggesting there are benefits to curating a calibration dataset that is representative of the observation process. Our findings provide insights into the practical challenges associated with statistical analyses of ARU data and possible analytical solutions to support reliable and cost-effective decision-making for wildlife conservation/management in the face of known sources of observation errors.

British Columbia↗

USGS2020

The use of Bayesian priors in Ecology: The good, the bad and the not great

Bayesian data analysis (BDA) is a powerful tool for making inference from ecological data, but its full potential has yet to be realized. Despite a generally positive trajectory in research surrounding model development and assessment, far too little attention has been given to prior specification. Default priors, a sub‐class of non‐informative prior distributions that are often chosen without critical thought or evaluation, are commonly used in practice. We believe the fear of being too ‘subjective’ has prevented many researchers from using any prior information in their analyses despite the fact that defending prior choice (informative or not) promotes good statistical practice. In this commentary, we provide an overview of how BDA is currently being used in a random sample of articles, discuss implications for inference if current bad practices continue, and highlight sub‐fields where knowledge about the system has improved inference and promoted good statistical practices through the careful and justified use of informative priors. We hope to inspire a renewed discussion about the use of Bayesian priors in Ecology with particular attention paid to specification and justification. We also emphasize that all priors are the result of a subjective choice, and should be discussed in that way.

Methods in Ecology and Evolution↗

USGS2019

Evidence of region‐wide bat population decline from long‐term monitoring and Bayesian occupancy models with empirically informed priors

Strategic conservation efforts for cryptic species, especially bats, are hindered by limited understanding of distribution and population trends. Integrating long‐term encounter surveys with multi‐season occupancy models provides a solution whereby inferences about changing occupancy probabilities and latent changes in abundance can be supported. When harnessed to a Bayesian inferential paradigm, this modeling framework offers flexibility for conservation programs that need to update prior model‐based understanding about at‐risk species with new data. This scenario is exemplified by a bat monitoring program in the Pacific Northwestern United States in which results from 8 years of surveys from 2003 to 2010 require updating with new data from 2016 to 2018. The new data were collected after the arrival of bat white‐nose syndrome and expansion of wind power generation, stressors expected to cause population declines in at least two vulnerable species, little brown bat ( Myotis lucifugus ) and the hoary bat ( Lasiurus cinereus ). We used multi‐season occupancy models with empirically informed prior distributions drawn from previous occupancy results (2003–2010) to assess evidence of contemporary decline in these two species. Empirically informed priors provided the bridge across the two monitoring periods and increased precision of parameter posterior distributions, but did not alter inferences relative to use of vague priors. We found evidence of region‐wide summertime decline for the hoary bat ( = 0.86 ± 0.10) since 2010, but no evidence of decline for the little brown bat ( = 1.1 ± 0.10). White‐nose syndrome was documented in the region in 2016 and may not yet have caused regional impact to the little brown bat. However, our discovery of hoary bat decline is consistent with the hypothesis that the longer duration and greater geographic extent of the wind energy stressor (collision and barotrauma) have impacted the species. These hypotheses can be evaluated and updated over time within our framework of pre–post impact monitoring and modeling. Our approach provides the foundation for a strategic evidence‐based conservation system and contributes to a growing preponderance of evidence from multiple lines of inquiry that bat species are declining.

Oregon, Washington↗

USGS2019

North American Bat Monitoring Program regional protocol for surveying with stationary deployments of echolocation recording devices: Narrative version 1.0, Pacific Northwestern US

The outbreak of white-nose syndrome (WNS) and the growing awareness of the risks to bats from wind power generating facilities have driven radical changes to North American bat conservation. Over the last decade, formerly common species such as the little brown myotis (Myotis lucifugus) and hoary bat (Lasiurus cinereus) have experienced unprecedented mortality rates and are now facing non-trivial extinction risk. In response to this change, federal land management agencies such as the US National Park Service, US Fish and Wildlife Service, US Forest Service, US Bureau of Land Management and state wildlife management agencies such as the Oregon Department of Fish and Wildlife and Idaho Fish and Game have invested in collaborative, interagency bat monitoring to close the gap in information about bat welfare and to inform bat conservation strategies. Bats are notoriously difficult to track and study and there remains a paucity of fundamental information about the seasonal patterns of bat activity and habitat use and population distributions and abundances. Moreover, because bats are so highly mobile and difficult to survey (e.g., nocturnal flight), this information needs to be contextualized at broad regional (e.g., 10,000 km2) and range-wide extents. Delimiting bat populations at local scales (e.g., 100 km2) is very difficult and it is not clear, for example, how a declining trend in local (e.g., a small park unit) patterns of bat activity or relative abundance should be interpreted without broader context. In recognition of these challenges, a plan for coordinated continental-scale monitoring of bats, the North American Bat Monitoring Program (NABat) was developed (Loeb et al. 2015). The centerpiece of the plan is the use of a spatially-balanced randomized master sample of grid-cell sample units from a grid-based sampling frame to provide the architecture for collaboration and the statistical foundation for making inferences about bat populations across broad regions and entire bat geographic ranges. The plan outlines general goals, survey design, and field methods for both summertime acoustic surveys of bats as well as winter and summer counts of bats in hibernacula and maternity colonies but it does not provide field-level protocol and standard operating procedures for consistent and efficient implementation. This regional protocol provides these details for one component of NABat, the deployment of stationary acoustic detectors to record bats during summer, as is called for by the NABat plan. This protocol was written specifically to provide guidance and consistency across the Pacific Northwestern US (N. California [California Department of Fish and Wildlife Northern Region], Idaho, Washington, and Oregon; US Fish and Wildlife Service Region 1 and portion of Region 8 [in Northern California and Klamath Basin]; US Forest Service Region 6 and portions of Regions 1 and 5 in Idaho; and the Upper Columbia Basin, North Coast Cascades, and Klamath Networks of the National Park Service). This region has internal cohesion, sharing a distinct bat faunal assemblage of 15 species (with several additional species occurring on the southern periphery of the region), and a long history of collaborative bat monitoring beginning with the interagency Bat Grid Program which operated from 2003-2010 across Oregon and Washington (US Forest Service Region 6). This protocol will be coordinated and implemented by the Northwestern Bat Hub, on behalf of the collective interagency partnership. The Northwestern Bat Hub is housed on the Oregon State University-Cascades campus and leverages pooled partner funds and resources to maintain a small staff that coordinates and conducts monitoring, provides training and oversight, ensures high-quality data quality and control, and analyzes data and reports on results.

California, Idaho, Oregon, Washington↗

USGS2018

Improving geographically extensive acoustic survey designs for modeling species occurrence with imperfect detection and misidentification

Acoustic recording units (ARUs) enable geographically extensive surveys of sensitive and elusive species. However, a hidden cost of using ARU data for modeling species occupancy is that prohibitive amounts of human verification may be required to correct species identifications made from automated software. Bat acoustic studies exemplify this challenge because large volumes of echolocation calls could be recorded and automatically classified to species. The standard occupancy model requires aggregating verified recordings to construct confirmed detection/non‐detection datasets. The multistep data processing workflow is not necessarily transparent nor consistent among studies. We share a workflow diagramming strategy that could provide coherency among practitioners. A false‐positive occupancy model is explored that accounts for misclassification errors and enables potential reduction in the number of confirmed detections. Simulations informed by real data were used to evaluate how much confirmation effort could be reduced without sacrificing site occupancy and detection error estimator bias and precision. We found even under a 50% reduction in total confirmation effort, estimator properties were reasonable for our assumed survey design, species‐specific parameter values, and desired precision. For transferability, a fully documented r package, OCacoustic, for implementing a false‐positive occupancy model is provided. Practitioners can apply OCacoustic to optimize their own study design (required sample sizes, number of visits, and confirmation scenarios) for properly implementing a false‐positive occupancy model with bat or other wildlife acoustic data. Additionally, our work highlights the importance of clearly defining research objectives and data processing strategies at the outset to align the study design with desired statistical inferences.

Ecology and Evolution↗