Kathryn M. Irvine

Many bioacoustic data workflows rely on manual review (i.e., validation) of a subset of call files to provide information to statistical models that account for misclassification by automated algorithms. Because manual review can be prohibitively expensive, simulation can be a valuable tool to aid the design of studies that use validation. This package provides user-friendly functions to reduce the programming burden of simulation studies that compare validation sampling designs. Simulations assume the count-detection model, which is a realistic model for bioacoustic data, especially for bats. For more information, see Oram et al. (2025) .

Ecologists and other statistical practitioners with access to high‐resolution spatial data often 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 circular extents 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 Markov chain Monte Carlo techniques, enables efficient estimation of posterior model probabilities to assist with the choice of meaningful spatial scales for environmental covariates.

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.

In ecology and related sciences, missing data are common and occur in a variety of different contexts. When missing data are not handled properly, subsequent statistical estimates tend to be biased, inefficient, and lack proper confidence interval coverage. Missing data are often grouped into three categories: missing completely at random (MCAR), missing at random (MAR), and missing not at random (MNAR). We review each category and compare their benefits and drawbacks. We review several approaches to handling missing data including complete case analysis, imputation, inverse probability weighting, and data augmentation. We clarify what types of variables should accompany imputation methods and how those variables are influenced by the analysis methods. Additionally, we discuss missing data that lack a formal basis for measurement and hence are fundamentally different from MCAR, MAR, and MNAR missing data. Throughout, we introduce concepts and numeric examples using both simulated data and data from the United States Environmental Protection Agency's 2016 National Wetland Condition Assessment. We conclude by providing five considerations for ecologists and other scientists handling missing data.

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.

After the onset of the COVID-19 pandemic, scientific interest in coronaviruses endemic in animal populations has increased dramatically. However, investigating the prevalence of disease in animal populations across the landscape, which requires finding and capturing animals can be difficult. Spatial random sampling over a grid could be extremely inefficient because animals can be hard to locate, and the total number of samples may be small. Alternatively, preferential sampling, using existing knowledge to inform sample location, can guarantee larger numbers of samples, but estimates derived from this sampling scheme may exhibit bias if there is a relationship between higher probability sampling locations and the disease prevalence. Sample specimens are commonly grouped and tested in pools which can also be an added challenge when combined with preferential sampling. Here we present a Bayesian method for estimating disease prevalence with preferential sampling in pooled presence-absence data motivated by estimating factors related to coronavirus infection among Mexican free-tailed bats (Tadarida brasiliensis) in California. We demonstrate the efficacy of our approach in a simulation study, where a naive model, not accounting for preferential sampling, returns biased estimates of parameter values; however, our model returns unbiased results regardless of the degree of preferential sampling. Our model framework is then applied to data from California to estimate factors related to coronavirus prevalence. After accounting for preferential sampling impacts, our model suggests small prevalence differences between male and female bats.

We reviewed current (2008-2023) protocols for monitoring American alligator nests in Everglades National Park (ENP) using a qualitative statistical review framework that emphasizes connecting measurable objectives with evaluation of survey design elements. Our review outlined the statistical assumptions that, if severely violated, would prevent defensible inferences from being drawn regarding the three alligator nesting metrics monitored in ENP’s systematic reconnaissance flight (SRF) surveys: nest effort (number of nests), nest distribution, and nest fate. After evaluating the current survey and inference methods, we offer potential adjustments or augmentations to the design and models currently used to better account for potential error sources and align the stated objectives with statistical inferences.

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.

Estimating the abundance of unmarked animal populations from acoustic data is challenging due to the inability to identify individuals and the need to adjust for observation biases including detectability (false negatives), species misclassification (false positives), and sampling exposure. Acoustic surveys conducted along mobile transects were designed to avoid counting individuals more than once, where raw counts are commonly treated as an index of abundance. More recently, false‐positive abundance models have been developed to estimate abundance while accounting for imperfect detection and misclassification. We adapted these methods to model summertime abundance and trends of three species of bats at multiple spatial scales using acoustic recordings collected along mobile transects by partners of the North American Bat Monitoring Program (NABat) from 2012 to 2020. This multiscale modeling spanned individual transect routes, larger NABat grid cells (10 km × 10 km), and across the entire extent of modeled species ranges. We estimated relationships between species abundances and a suite of abiotic and biotic predictors (landcover types, climatological variables, physiographic diversity, building density, and the impacts of white‐nose syndrome [WNS]) and found varying levels of support between species. We present clear evidence of substantial declines in populations of tricolored bats (Perimyotis subflavus) and little brown bats (Myotis lucifugus), declines that corresponded in space and time with the progression of WNS, a devastating disease of hibernating bats. In contrast, our analysis revealed that similar population‐wide declines probably have not occurred in big brown bats (Eptesicus fuscus), a species known to be less affected by WNS. This study provides the first abundance‐based species distribution predictions and population trends for bats in their summer ranges in North America. These models will probably be applicable to assessing wildlife populations in other monitoring programs where acoustic data are used or where false‐negative and false‐positive detections are present. Finally, our abundance framework (as a spatial point pattern process) can serve as a foundation from which more sophisticated integrated species distribution models that incorporate additional streams of monitoring data (e.g., stationary acoustics, captures) can be developed for North American bats.

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.