Search USGS⌕ Search

Geology topics

Kathryn M. Irvine

Publications and source records attributed to Kathryn M. Irvine.

At least 37 records · Page 2Linked to original sources
USGS2019

Statistical power of dynamic occupancy models to identify temporal change: Informing the North American Bat Monitoring Program

Dynamic occupancy models provide a flexible framework for estimating and mapping species occupancy patterns over space and time for large-scale monitoring programs (e.g., the North American Bat Monitoring Program (NABat), the Amphibian Research and Monitoring Initiative). Challenges for designing surveys using the dynamic occupancy modeling framework include defining appropriate derived trend parameters, and providing usable tools for researchers to conduct project-specific sample size investigations. We present a simulation-based power analysis framework for dynamic occupancy models that allows for the incorporation of the underlying environmental space (i.e., as covariates) within a specific study region to inform sample size estimation. We investigate two definitions of temporal trend: (1) a gradual, sustained (linear or nonlinear) change over a period of many years, and (2) an abrupt increase or decrease between two time periods. We draw upon pilot data collected following NABat protocols to inform assumed data generating values in a demonstration of our approach. Due to the complicated parameter structure of dynamic occupancy models, we emphasize the importance of visualizing simulated changes over time based on different parameter settings prior to conducting a power analysis. Our simulations revealed that the linearity of short-term trends (five years in our investigation) conferred higher power with lower sample size than longer trends where occupancy probabilities approached zero (ten years in our investigation). We provide an example of how to use our tools to conduct customized investigations using questions posed by NABat, and in doing so, we shed light on general guidelines that can be applied to programs monitoring species occupancy for other taxa. Importantly, we created an R package to execute our approach for informing program-, species-, and study-specific investigations aimed at identifying changes in species occupancy.

Ecological Indicators↗
USGS2019

Wetland drying linked to variations in snowmelt runoff across Grand Teton and Yellowstone national parks

In Grand Teton and Yellowstone national parks wetlands offer critical habitat and play a key role in supporting biological diversity. The shallow depths and small size of many wetlands make them vulnerable to changes in climate compared with larger and deeper aquatic habitats. Here, we use a simple water balance model to generate estimates of biophysical drivers of wetland change. We then examine the relationship between wetland inundation status and four principal drivers (i.e., temperature, precipitation, evapotranspiration, and runoff) spanning varying meteorological conditions over an 8-year time series from Grand Teton and Yellowstone national parks. We found that a higher percentage of surveyed wetlands were dry in years characterized by lower snowmelt runoff. While runoff-based models were most supported, wetland drying was also related to variations in April to June precipitation and temperatures. Our work shows that wetland drying was widespread across both parks, but sub-regional variations were best described at the hydrologic subbasin-level. Documenting the varying responses of wetlands to meteorological drivers is a necessary first step to identifying which subbasins are most sensitive to recent change and contemplating how future change may alter the distribution of wetlands and their dependent taxa.

Idaho, Montana, Wyoming↗
USGS2019

Assessing spatial and temporal patterns in sagebrush steppe vegetation communities 2012-2018: Grand Teton National Park

Visual cover class data were collected on over 80 species across 30 permanent sampling frames in sagebrush steppe vegetation communities in Grand Teton National Park from 2012 to 2018. In this report, temporal and spatial patterns in species composition were assessed and used to inform potential sampling strategies for future monitoring. Specifically, the viability of a reduction in sampling effort was evaluated based on the similarity in species composition within each frame over time and among frames within each year. Using distance-based ordination techniques, we found little to no evidence of differences in species composition within each frame over time. Furthermore, there was little evidence of heterogeneity in species composition among frames within each year, though there was some evidence of differences in composition between the two principle sagebrush community types (sagebrush dry shrubland and sagebrush-bitterbrush) aggregated across frames. Based on these results, we propose that a reduction in sampling effort is viable and suggest a new monitoring schedule.

Wyoming↗
USGS2018

Occupancy modeling species–environment relationships with non‐ignorable survey designs

Statistical models supporting inferences about species occurrence patterns in relation to environmental gradients are fundamental to ecology and conservation biology. A common implicit assumption is that the sampling design is ignorable and does not need to be formally accounted for in analyses. The analyst assumes data are representative of the desired population and statistical modeling proceeds. However, if data sets from probability and non‐probability surveys are combined or unequal selection probabilities are used, the design may be non‐ignorable. We outline the use of pseudo‐maximum likelihood estimation for site‐occupancy models to account for such non‐ignorable survey designs. This estimation method accounts for the survey design by properly weighting the pseudo‐likelihood equation. In our empirical example, legacy and newer randomly selected locations were surveyed for bats to bridge a historic statewide effort with an ongoing nationwide program. We provide a worked example using bat acoustic detection/non‐detection data and show how analysts can diagnose whether their design is ignorable. Using simulations we assessed whether our approach is viable for modeling data sets composed of sites contributed outside of a probability design. Pseudo‐maximum likelihood estimates differed from the usual maximum likelihood occupancy estimates for some bat species. Using simulations we show the maximum likelihood estimator of species–environment relationships with non‐ignorable sampling designs was biased, whereas the pseudo‐likelihood estimator was design unbiased. However, in our simulation study the designs composed of a large proportion of legacy or non‐probability sites resulted in estimation issues for standard errors. These issues were likely a result of highly variable weights confounded by small sample sizes (5% or 10% sampling intensity and four revisits). Aggregating data sets from multiple sources logically supports larger sample sizes and potentially increases spatial extents for statistical inferences. Our results suggest that ignoring the mechanism for how locations were selected for data collection (e.g., the sampling design) could result in erroneous model‐based conclusions. Therefore, in order to ensure robust and defensible recommendations for evidence‐based conservation decision‐making, the survey design information in addition to the data themselves must be available for analysts. Details for constructing the weights used in estimation and code for implementation are provided.

Ecological Applications↗
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↗
USGS2018

Impacts of temporal revisit designs on the power to detect trend with a linear mixed model: An application to long-term monitoring of Sierra Nevada lakes

Long-term ecological monitoring programs often use linear mixed models to estimate trend in an ecological indicator sampled across large landscapes. A linear mixed model is versatile for estimating a linear trend in time as well as components of spatial and temporal variationin the case of unbalanced data structures , which are common in complex monitoring designs where limited sampling effort must be optimized over time and space. A power analysis was used to inform a lake chemistry monitoring design, including selecting the most appropriate temporal revisit design. Pilot data from surveys of lakes across large wilderness national parks (Sequoia, Kings Canyon, and Yosemite national parks) were used to obtain variance components for a Monte Carlo power simulation. Using a linear mixed model for a range of temporal revisit designs, sample sizes, and trend magnitudes, we evaluated the power to detect trend, the trend test size, and the relative bias of trend coefficient estimates for four continuous and normally distributed indicators. Contrary to prior research based on large-sample approximations that identified a single panel of sites visited annually as the revisit design generating the highest power, we found that the power to detect a 12-year trend based on the Wald t-test from a linear mixed model may be optimized by obtaining unbalanced data sets with limited to no annual replication. We emphasize the importance of examining variance composition, sample size, and the power and size of the trend test with Monte Carlo simulation when allocating sampling effort over time and space.

Ecological Indicators↗
USGS2017

Statistical design and analysis for plant cover studies with multiple sources of observation errors

Effective wildlife habitat management and conservation requires understanding the factors influencing distribution and abundance of plant species. Field studies, however, have documented observation errors in visually estimated plant cover including measurements which differ from the true value (measurement error) and not observing a species that is present within a plot (detection error). Unlike the rapid expansion of occupancy and N-mixture models for analysing wildlife surveys, development of statistical models accounting for observation error in plants has not progressed quickly. Our work informs development of a monitoring protocol for managed wetlands within the National Wildlife Refuge System. Zero-augmented beta (ZAB) regression is the most suitable method for analysing areal plant cover recorded as a continuous proportion but assumes no observation errors. We present a model extension that explicitly includes the observation process thereby accounting for both measurement and detection errors. Using simulations, we compare our approach to a ZAB regression that ignores observation errors (naïve model) and an “ad hoc” approach using a composite of multiple observations per plot within the naïve model. We explore how sample size and within-season revisit design affect the ability to detect a change in mean plant cover between 2 years using our model. Explicitly modelling the observation process within our framework produced unbiased estimates and nominal coverage of model parameters. The naïve and “ad hoc” approaches resulted in underestimation of occurrence and overestimation of mean cover. The degree of bias was primarily driven by imperfect detection and its relationship with cover within a plot. Conversely, measurement error had minimal impacts on inferences. We found >30 plots with at least three within-season revisits achieved reasonable posterior probabilities for assessing change in mean plant cover. For rapid adoption and application, code for Bayesian estimation of our single-species ZAB with errors model is included. Practitioners utilizing our R-based simulation code can explore trade-offs among different survey efforts and parameter values, as we did, but tuned to their own investigation. Less abundant plant species of high ecological interest may warrant the additional cost of gathering multiple independent observations in order to guard against erroneous conclusions.

Methods in Ecology and Evolution↗
USGS2017

Assessment of imperfect detection of blister rust in whitebark pine within the Greater Yellowstone Ecosystem

We examined data on white pine blister rust (blister rust) collected during the monitoring of whitebark pine trees in the Greater Yellowstone Ecosystem (from 2004-2015). Summaries of repeat observations performed by multiple independent observers are reviewed and discussed. These summaries show variability among observers and the potential for errors being made in blister rust status. Based on this assessment, we utilized occupancy models to analyze blister rust prevalence while explicitly accounting for imperfect detection. Available covariates were used to model both the probability of a tree being infected with blister rust and the probability of an observer detecting the infection. The fitted model provided strong evidence that the probability of blister rust infection increases as tree diameter increases and decreases as site elevation increases. Most importantly, we found evidence of heterogeneity in detection probabilities related to tree size and average slope of a transect. These results suggested that detecting the presence of blister rust was more difficult in larger trees. Also, there was evidence that blister rust was easier to detect on transects located on steeper slopes. Our model accounted for potential impacts of observer experience on blister rust detection probabilities and also showed moderate variability among the different observers in their ability to detect blister rust. Based on these model results, we suggest that multiple observer sampling continue in future field seasons in order to allow blister rust prevalence estimates to be corrected for imperfect detection. We suggest that the multiple observer effort be spread out across many transects (instead of concentrated at a few each field season) while retaining the overall proportion of trees with multiple observers around 5-20%. Estimates of prevalence are confounded with detection unless it is explicitly accounted for in an analysis and we demonstrate how an occupancy model can be used to do account for this source of observation error.

idaho, Montana, Utah, Wyoming↗
USGS2016

Whitebark pine mortality related to white pine blister rust, mountain pine beetle outbreak, and water availability

Whitebark pine ( Pinus albicaulis ) forests in the western United States have been adversely affected by an exotic pathogen ( Cronartium ribicola , causal agent of white pine blister rust), insect outbreaks ( Dendroctonus ponderosae , mountain pine beetle), and drought. We monitored individual trees from 2004 to 2013 and characterized stand-level biophysical conditions through a mountain pine beetle epidemic in the Greater Yellowstone Ecosystem. Specifically, we investigated associations between tree-level variables (duration and location of white pine blister rust infection, presence of mountain pine beetle, tree size, and potential interactions) with observations of individual whitebark pine tree mortality. Climate summaries indicated that cumulative growing degree days in years 2006–2008 likely contributed to a regionwide outbreak of mountain pine beetle prior to the observed peak in whitebark mortality in 2009. We show that larger whitebark pine trees were preferentially attacked and killed by mountain pine beetle and resulted in a regionwide shift to smaller size class trees. In addition, we found evidence that smaller size class trees with white pine blister rust infection experienced higher mortality than larger trees. This latter finding suggests that in the coming decades white pine blister rust may become the most probable cause of whitebark pine mortality. Our findings offered no evidence of an interactive effect of mountain pine beetle and white pine blister rust infection on whitebark pine mortality in the Greater Yellowstone Ecosystem. Interestingly, the probability of mortality was lower for larger trees attacked by mountain pine beetle in stands with higher evapotranspiration. Because evapotranspiration varies with climate and topoedaphic conditions across the region, we discuss the potential to use this improved understanding of biophysical influences on mortality to identify microrefugia that might contribute to successful whitebark pine conservation efforts. Using tree-level observations, the National Park Service-led Greater Yellowstone Interagency Whitebark Pine Long-term Monitoring Program provided important ecological insight on the size-dependent effects of white pine blister rust, mountain pine beetle, and water availability on whitebark pine mortality. This ongoing monitoring campaign will continue to offer observations that advance conservation in the Greater Yellowstone Ecosystem.

Ecosphere↗
USGS2016

Extending ordinal regression with a latent zero-augmented beta distribution

Ecological abundance data are often recorded on an ordinal scale in which the lowest category represents species absence. One common example is when plant species cover is visually assessedwithin bounded quadrats and then assigned to pre-defined cover class categories.We present an ordinal beta hurdle model that directly models ordinal category probabilitieswith a biologically realistic beta-distributed latent variable.Ahurdle-at-zero model allows ecologists to explore distribution (absence) and abundance processes in an integrated framework. This provides an alternative to cumulative link models when data are inconsistent with the assumption that the odds ofmoving into a higher category are the same for all categories (proportional odds). Graphical tools and a deviance information criterion were developed to assess whether a hurdle-at-zero model should be used for inferences rather than standard ordinal methods. Hurdle-at-zero and non-hurdle ordinal models fit to vegetation cover class data produced substantially different conclusions.The ordinal beta hurdle model yielded more precise parameter estimates than cumulative logit models, although out-of-sample predictions were similar. The ordinal beta hurdle model provides inferences directly on the latent biological variable of interest, percent cover, and supports exploration of more realistic ecological patterns and processes through the hurdle-at-zero or two-part specification.We provide JAGS code as an on-line supplement. Supplementary materials accompanying this paper appear on-line.

Journal of Agricultural, Biological, and Environme↗
USGS2016

A goodness-of-fit test for occupancy models with correlated within-season revisits

Occupancy modeling is important for exploring species distribution patterns and for conservation monitoring. Within this framework, explicit attention is given to species detection probabilities estimated from replicate surveys to sample units. A central assumption is that replicate surveys are independent Bernoulli trials, but this assumption becomes untenable when ecologists serially deploy remote cameras and acoustic recording devices over days and weeks to survey rare and elusive animals. Proposed solutions involve modifying the detection-level component of the model (e.g., first-order Markov covariate). Evaluating whether a model sufficiently accounts for correlation is imperative, but clear guidance for practitioners is lacking. Currently, an omnibus goodnessof- fit test using a chi-square discrepancy measure on unique detection histories is available for occupancy models (MacKenzie and Bailey, Journal of Agricultural, Biological, and Environmental Statistics, 9, 2004, 300; hereafter, MacKenzie– Bailey test). We propose a join count summary measure adapted from spatial statistics to directly assess correlation after fitting a model. We motivate our work with a dataset of multinight bat call recordings from a pilot study for the North American Bat Monitoring Program. We found in simulations that our join count test was more reliable than the MacKenzie–Bailey test for detecting inadequacy of a model that assumed independence, particularly when serial correlation was low to moderate. A model that included a Markov-structured detection-level covariate produced unbiased occupancy estimates except in the presence of strong serial correlation and a revisit design consisting only of temporal replicates. When applied to two common bat species, our approach illustrates that sophisticated models do not guarantee adequate fit to real data, underscoring the importance of model assessment. Our join count test provides a widely applicable goodness-of-fit test and specifically evaluates occupancy model lack of fit related to correlation among detections within a sample unit. Our diagnostic tool is available for practitioners that serially deploy survey equipment as a way to achieve cost savings.

Ecology and Evolution↗
USGS2016

Semi-arid vegetation response to antecedent climate and water balance windows

Questions Can we improve understanding of vegetation response to water availability on monthly time scales in semi-arid environments using remote sensing methods? What climatic or water balance variables and antecedent windows of time associated with these variables best relate to the condition of vegetation? Can we develop credible near-term forecasts from climate data that can be used to prepare for future climate change effects on vegetation? Location Semi-arid grasslands in Capitol Reef National Park, Utah, USA. Methods We built vegetation response models by relating the normalized difference vegetation index (NDVI) from MODIS imagery in Mar–Nov 2000–2013 to antecedent climate and water balance variables preceding the monthly NDVI observations. We compared how climate and water balance variables explained vegetation greenness and then used a multi-model ensemble of climate and water balance models to forecast monthly NDVI for three holdout years. Results Water balance variables explained vegetation greenness to a greater degree than climate variables for most growing season months. Seasonally important variables included measures of antecedent water input and storage in spring, switching to indicators of drought, input or use in summer, followed by antecedent moisture availability in autumn. In spite of similar climates, there was evidence the grazed grassland showed a response to drying conditions 1 mo sooner than the ungrazed grassland. Lead times were generally short early in the growing season and antecedent window durations increased from 3 mo early in the growing season to 1 yr or more as the growing season progressed. Forecast accuracy for three holdout years using a multi-model ensemble of climate and water balance variables outperformed forecasts made with a naïve NDVI climatology. Conclusions We determined the influence of climate and water balance on vegetation at a fine temporal scale, which presents an opportunity to forecast vegetation response with short lead times. This understanding was obtained through high-frequency vegetation monitoring using remote sensing, which reduces the costs and time necessary for field measurements and can lead to more rapid detection of vegetation changes that could help managers take appropriate actions.

Applied Vegetation Science↗
USGS2015

Establishing conservation baselines with dynamic distribution models for bat populations facing imminent decline

Aim Bat mortality rates from white-nose syndrome and wind power development are unprecedented. Cryptic and wide-ranging behaviours of bats make them difficult to survey, and population estimation is often intractable. We advance a model-based framework for making spatially explicit predictions about summertime distributions of bats from capture and acoustic surveys. Motivated by species-energy and life-history theory, our models describe hypotheses about spatio-temporal variation in bat distributions along environmental gradients and life-history attributes, providing a statistical basis for conservation decision-making. Location Oregon and Washington, USA. Methods We developed Bayesian hierarchical models for 14 bat species from an 8-year monitoring dataset across a ~430,000 km 2 study area. Models accounted for imperfect detection and were temporally dynamic. We mapped predicted occurrence probabilities and prediction uncertainties as baselines for assessing future declines. Results Forest cover, snag abundance and cliffs were important predictors for most species. Species occurrence patterns varied along elevation and precipitation gradients, suggesting a potential hump-shaped diversity–productivity relationship. Annual turnover in occurrence was generally low, and occurrence probabilities were stable among most species. We found modest evidence that turnover covaried with the relative riskiness of bat roosting and migration. The fringed myotis ( Myotis thysanodes ), canyon bat ( Parastrellus hesperus ) and pallid bat ( Antrozous pallidus ) were rare; fringed myotis occurrence probabilities declined over the study period. We simulated anticipated declines to demonstrate that mapped occurrence probabilities, updated over time, provide an intuitive way to assess bat conservation status for a broad audience. Main conclusions Landscape keystone structures associated with roosting habitat emerged as regionally important predictors of bat distributions. The challenges of bat monitoring have constrained previous species distribution modelling efforts to temporally static presence-only approaches. Our approach extends to broader spatial and temporal scales than has been possible in the past for bats, making a substantial increase in capacity for bat conservation.

Oregon, Washington↗
USGS2015

Estimating the phenology of elk brucellosis transmission with hierarchical models of cause-specific and baseline hazards

Understanding the seasonal timing of disease transmission can lead to more effective control strategies, but the seasonality of transmission is often unknown for pathogens transmitted directly. We inserted vaginal implant transmitters (VITs) in 575 elk ( Cervus elaphus canadensis ) from 2006 to 2014 to assess when reproductive failures (i.e., abortions or still births) occur, which is the primary transmission route of Brucella abortus , the causative agent of brucellosis in the Greater Yellowstone Ecosystem. Using a survival analysis framework, we developed a Bayesian hierarchical model that simultaneously estimated the total baseline hazard of a reproductive event as well as its 2 mutually exclusive parts (abortions or live births). Approximately, 16% (95% CI = 0.10, 0.23) of the pregnant seropositive elk had reproductive failures, whereas 2% (95% CI = 0.01, 0.04) of the seronegative elk had probable abortions. Reproductive failures could have occurred as early as 13 February and as late as 10 July, peaking from March through May. Model results suggest that less than 5% of likely abortions occurred after 6 June each year and abortions were approximately 5 times more likely in March, April, or May compared to February or June. In western Wyoming, supplemental feeding of elk begins in December and ends during the peak of elk abortions and brucellosis transmission (i.e., Mar and Apr). Years with more snow may enhance elk-to-elk transmission on supplemental feeding areas because elk are artificially aggregated for the majority of the transmission season. Elk-to-cattle transmission will depend on the transmission period relative to the end of the supplemental feeding season, elk seroprevalence, population size, and the amount of commingling. Our statistical approach allowed us to estimate the probability density function of different event types over time, which may be applicable to other cause-specific survival analyses. It is often challenging to assess the cause of death, or in this case whether the reproductive event was an abortion or live birth. Accounting for uncertainty in the event type is an important future addition to our methodological approach.

Wyoming↗
USGS2015

Meteorological variables to aid forecasting deep slab avalanches on persistent weak layers

Deep slab avalanches are particularly challenging to forecast. These avalanches are difficult to trigger, yet when they release they tend to propagate far and can result in large and destructive avalanches. We utilized a 44-year record of avalanche control and meteorological data from Bridger Bowl ski area in southwest Montana to test the usefulness of meteorological variables for predicting seasons and days with deep slab avalanches. We defined deep slab avalanches as those that failed on persistent weak layers deeper than 0.9 m, and that occurred after February 1st. Previous studies often used meteorological variables from days prior to avalanches, but we also considered meteorological variables over the early months of the season. We used classification trees and random forests for our analyses. Our results showed seasons with either dry or wet deep slabs on persistent weak layers typically had less precipitation from November through January than seasons without deep slabs on persistent weak layers. Days with deep slab avalanches on persistent weak layers often had warmer minimum 24-hour air temperatures, and more precipitation over the prior seven days, than days without deep slabs on persistent weak layers. Days with deep wet slab avalanches on persistent weak layers were typically preceded by three days of above freezing air temperatures. Seasonal and daily meteorological variables were found useful to aid forecasting dry and wet deep slab avalanches on persistent weak layers, and should be used in combination with continuous observation of the snowpack and avalanche activity.

Montana↗