Evidence that Pacific tuna mercury levels are driven by marine methylmercury production and anthropogenic inputs.

Pacific Ocean tuna is among the most-consumed seafood products but contains relatively high levels of the neurotoxin methylmercury. Limited observations suggest tuna mercury levels vary in space and time, yet the drivers are not well understood. Here, we map mercury concentrations in skipjack tuna across the Pacific Ocean and build generalized additive models to quantify the anthropogenic, ecological, and biogeochemical drivers. Skipjack mercury levels display a fivefold spatial gradient, with maximum concentrations in the northwest near Asia, intermediate values in the east, and the lowest levels in the west, southwest, and central Pacific. Large spatial differences can be explained by the depth of the seawater methylmercury peak near low-oxygen zones, leading to enhanced tuna mercury concentrations in regions where oxygen depletion is shallow. Despite this natural biogeochemical control, the mercury hotspot in tuna caught near Asia is explained by elevated atmospheric mercury concentrations and/or mercury river inputs to the coastal shelf. While we cannot ignore the legacy mercury contribution from other regions to the Pacific Ocean (e.g., North America and Europe), our results suggest that recent anthropogenic mercury release, which is currently largest in Asia, contributes directly to present-day human mercury exposure.

Spatial risk of Haemaphysalis longicornis borne Dabieshan tick virus (DBTV) in China.

The uncertainty and unknowability of emerging infectious diseases have caused many major public health and security incidents in recent years. As a new tick-borne disease, Dabieshan tick virus (DBTV) necessitate systematic epidemiological and spatial distribution analysis. In this study, tick samples from Liaoning Province were collected and used to evaluate distribution of DBTV in ticks. Outbreak points of DBTV and the records of the vector Haemaphysalis longicornis in China were collected and used to establish a prediction model using niche model combined with environmental factors. We found that H. longicornis and DBTV were widely distributed in Liaoning Province. The risk analysis results showed that the DBTV in the eastern provinces of China has a high risk, and the risk is greatly influenced by elevation, land cover, and meteorological factors. The risk geographical area predicted by the model is significantly larger than the detected positive areas, indicating that the etiological survey is seriously insufficient. This study provided molecular and important epidemiological evidence for etiological ecology of DBTV. The predicted high-risk areas indicated the insufficient monitoring and risk evaluation and the necessity of future monitoring and control work.

Testing for concordance between predicted species richness, past prioritization, and marine protected area designations in the western Indian Ocean.

Scientific advances in environmental data coverage and machine learning algorithms have improved the ability to make large-scale predictions where data are missing. These advances allowed us to develop a spatially resolved proxy for predicting numbers of tropical nearshore marine taxa. A diverse marine environmental spatial database was used to model numbers of taxa from ∼1000 field sites, and the predictions were applied to all 7039 6.25-km2 reef cells in 9 ecoregions and 11 nations of the western Indian Ocean. Our proxy for total numbers of taxa was based on the positive correlation (r2 = 0.24) of numbers of taxa of hard corals and 5 highly diverse reef fish families. Environmental relationships indicated that the number of fish species was largely influenced by biomass, nearness to people, governance, connectivity, and productivity and that coral taxa were influenced mostly by physicochemical environmental variability. At spatial delineations of province, ecoregion, nation, and strength of spatial clustering, we compared areas of conservation priority based on our total species proxy with those identified in 3 previous priority-setting reports and with the protected area database. Our method identified 119 locations that fit 3 numbers of taxa (hard coral, fish, and their combination) and 4 spatial delineations (nation, ecoregion, province, and reef clustering) criteria. Previous publications on priority setting identified 91 priority locations of which 6 were identified by all reports. We identified 12 locations that fit our 12 criteria and corresponded with 3 previously identified locations, 65 that aligned with at least 1 past report, and 28 that were new locations. Only 34% of the 208 marine protected areas in this province overlapped with identified locations with high numbers of predicted taxa. Differences occurred because past priorities were frequently based on unquantified perceptions of remoteness and preselected priority taxa. Our environment-species proxy and modeling approach can be considered among other important criteria for making conservation decisions.

Evaluación de la concordancia entre la riqueza de especies pronosticada, priorizaciones pasadas y la designación de áreas marinas protegidas en el oeste del Océano Índico Resumen Los avances científicos en la cobertura de datos ambientales y los algoritmos de aprendizaje automatizado han mejorado la capacidad de predecir a gran escala cuando hacen falta datos. Estos avances nos permiten desarrollar un representante con resolución espacial para predecir la cantidad de taxones marinos en las costas tropicales. Usamos una base de datos espaciales de diversos ambientes marinos para modelar la cantidad de taxones a partir de ∼1000 sitios de campo y aplicamos las predicciones a las 7039 celdas arrecifales de 6.25­km2 en nueve ecorregiones y once países del oeste del Océano Índico. Nuestro representante para la cantidad total de taxones se basó en la correlación positiva (r2=0.24) de la cantidad de taxones de corales duros y cinco familias de peces arrecifales con diversidad alta. Las relaciones ambientales indicaron que el número de especies de peces estuvo influenciado principalmente por la biomasa, la cercanía a las personas, la gestión, la conectividad y la productividad y que los taxones de coral estuvieron influenciados principalmente por la variabilidad ambiental fisicoquímica. Comparamos la prioridad de las áreas de conservación a nivel de las delimitaciones espaciales de provincia, ecorregión, nación y fuerza del agrupamiento espacial basado en nuestro total de especies representantes con aquellas especies identificadas en tres reportes previos de establecimiento de prioridades y con la base de datos de áreas protegidas. Con nuestro método identificamos 119 localidades aptas para tres cantidades de taxones (corales duros, peces y su combinación) y cuatro criterios de delimitación espacial (nación, ecorregión, provincia y grupo de arrecifes). Las publicaciones previas sobre el establecimiento de prioridades identificaron 91 localidades prioritarias de las cuales seis fueron identificadas por todos los reportes. Identificamos doce localidades que se ajustan a nuestros doce criterios y se correspondieron con tres localidades identificadas previamente, 65 que se alinearon con al menos un reporte anterior y 28 que eran nuevas localidades. Sólo 34% de las 208 áreas marinas protegidas en esta provincia se traslaparon con localidades identificadas con un gran número de taxones pronosticados. Hubo diferencias porque en el pasado se priorizaba frecuentemente con base en las percepciones no cuantificadas de lo remoto y prioritario de los taxones preseleccionados. Nuestra especie representante del ambiente y nuestra estrategia de modelo pueden considerarse entre otros criterios importantes para tomar decisiones de conservación.

The spatial distribution of heat related hospitalizations and classification of the most dangerous heat events in California at a small-scale level.

Many studies have explored the impact of extreme heat on health, but few have investigated localized heat-health outcomes across a wide area. We examined fine-scale variability in vulnerable areas, considering population distribution, local weather, and landscape characteristics. Using 36 different heat event definitions, we identified the most dangerous types of heat events based on minimum, maximum, and diurnal temperatures with varying thresholds and durations. Focusing on California's diverse climate, elevation, and population distribution, we analyzed hospital admissions for various causes of admission (2004-2013). Our matching approach identified vulnerable zip codes, even with small populations, on absolute and relative scales. Bayesian Hierarchical models leveraged spatial correlation. We ranked the 36 heat event types by attributable hospital admissions per zip code and provided code, simulated data, and an interactive web app for reproducibility. Our findings showed high variation in heat-related hospitalizations in coastal cities and substantial heat burdens in the Central Valley. Diurnal heat events had the greatest impact in the Central Valley, while nighttime extreme heat events drove burdens in the southeastern desert. This spatially informed approach guides local policies, prioritizing dangerous heat events to reduce the heat-health burden. The methodology is applicable to other regions, informing early warning systems and characterizing extreme heat impacts.

A spatial model of YAP/TAZ signaling reveals how stiffness, dimensionality, and shape contribute to emergent outcomes.

YAP/TAZ is a master regulator of mechanotransduction whose functions rely on translocation from the cytoplasm to the nucleus in response to diverse physical cues. Substrate stiffness, substrate dimensionality, and cell shape are all input signals for YAP/TAZ, and through this pathway, regulate critical cellular functions and tissue homeostasis. Yet, the relative contributions of each biophysical signal and the mechanisms by which they synergistically regulate YAP/TAZ in realistic tissue microenvironments that provide multiplexed input signals remain unclear. For example, in simple two-dimensional culture, YAP/TAZ nuclear localization correlates strongly with substrate stiffness, while in three-dimensional (3D) environments, YAP/TAZ translocation can increase with stiffness, decrease with stiffness, or remain unchanged. Here, we develop a spatial model of YAP/TAZ translocation to enable quantitative analysis of the relationships between substrate stiffness, substrate dimensionality, and cell shape. Our model couples cytosolic stiffness to nuclear mechanics to replicate existing experimental trends, and extends beyond current data to predict that increasing substrate activation area through changes in culture dimensionality, while conserving cell volume, forces distinct shape changes that result in nonlinear effect on YAP/TAZ nuclear localization. Moreover, differences in substrate activation area versus total membrane area can account for counterintuitive trends in YAP/TAZ nuclear localization in 3D culture. Based on this multiscale investigation of the different system features of YAP/TAZ nuclear translocation, we predict that how a cell reads its environment is a complex information transfer function of multiple mechanical and biochemical factors. These predictions reveal a few design principles of cellular and tissue engineering for YAP/TAZ mechanotransduction.

Mapping and assessing natural soundscape quality: An indicator-based model for landscape planning.

Natural soundscape quality (NSQ) has been recognized as an essential cultural ecosystem service that contributes significantly to human health and well-being. It also stands as an indispensable component of environmental quality, especially for landscape aesthetic quality. However, an assessment tool for NSQ in landscape planning and environmental impact assessments is still absent. Therefore, this paper aims to address this gap by proposing an indicator-based model for assessing and quantifying NSQ in the Geographic Information System. The model characterizes NSQ based on Calmness and Vibrancy, and employs several indicators, sub-indicators, and respective metrics as proxies to quantify and map them spatially. The evaluation criteria of the model correspond to the general public's preferences for soundscape features. The case study results in Springe municipality, Germany, show that the relative values of NSQ are high in green spaces, including forests, grasslands, and shrublands, whereas they are low in open farmlands. The multiple natural sounds yield higher NSQ scores than the individual ones. The same soundscape compositions in forests and in urban parks exhibit higher NSQ scores than in other land cover types. In addition, the shares of relative values show similar distribution patterns among Calmness, Vibrancy, and NSQ according to land cover types and soundscape compositions. The evaluation results align with public values and preferences for soundscape features. Unlike subjectivist approaches, our user-independent methodology is easily transferable and reproducible. The results are comparable and communicable among the assessed areas. These endow the indicator-based model with the potential to be applied at various planning and management scales. The findings can help to incorporate soundscape evaluation into landscape planning and management systems, supporting sustainable landscape development, and providing valuable information for policy-, plan- and decision-making.

Innovations in Public Health Surveillance for Emerging Infections.

Public health surveillance is defined as the ongoing, systematic collection, analysis, and interpretation of health data and is closely integrated with the timely dissemination of information that the public needs to know and upon which the public should act. Public health surveillance is central to modern public health practice by contributing data and information usually through a national notifiable disease reporting system (NNDRS). Although early identification and prediction of future disease trends may be technically feasible, more work is needed to improve accuracy so that policy makers can use these predictions to guide prevention and control efforts. In this article, we review the advantages and limitations of the current NNDRS in most countries, discuss some lessons learned about prevention and control from the first wave of COVID-19, and describe some technological innovations in public health surveillance, including geographic information systems (GIS), spatial modeling, artificial intelligence, information technology, data science, and the digital twin method. We conclude that the technology-driven innovative public health surveillance systems are expected to further improve the timeliness, completeness, and accuracy of case reporting during outbreaks and also enhance feedback and transparency, whereby all stakeholders should receive actionable information on control and be able to limit disease risk earlier than ever before.

Pattern Formation in Mesic Savannas.

We analyze a spatially extended version of a well-known model of forest-savanna dynamics, which presents as a system of nonlinear partial integro-differential equations, and study necessary conditions for pattern-forming bifurcations. Homogeneous solutions dominate the dynamics of the standard forest-savanna model, regardless of the length scales of the various spatial processes considered. However, several different pattern-forming scenarios are possible upon including spatial resource limitation, such as competition for water, soil nutrients, or herbivory effects. Using numerical simulations and continuation, we study the nature of the resulting patterns as a function of system parameters and length scales, uncovering subcritical pattern-forming bifurcations and observing significant regions of multistability for realistic parameter regimes. Finally, we discuss our results in the context of extant savanna-forest modeling efforts and highlight ongoing challenges in building a unifying mathematical model for savannas across different rainfall levels.

Geographically varying associations between mentally unhealthy days and social vulnerability in the USA.

OBJECTIVES: A growing body of research has incorporated the Social Vulnerability Index (SVI) into an expanded understanding of the social determinants of health. Although each component of SVI and its association with individual-level mental health conditions have been well discussed, variation in mentally unhealthy days (MUDs) at a county level is still unexplored. To systematically examine the geographically varying relationships between SVI and MUDs across the US counties, our study adopted two different methods: 1) aspatial regression modeling (ordinary least square [OLS]); and 2) locally calibrated spatial regression (geographically weighted regression [GWR]). STUDY DESIGN: This study used a cross-sectional statistical design and geospatial data manipulation/analysis techniques. Analytical unit is each of the 3109 counties in the continental USA. METHODS: We tested the model performance of two different methods and suggest using both methods to reduce potential issues (e.g., Simpson's paradox) when researchers apply aspatial analysis to spatially coded data sets. We applied GWR after checking the spatial dependence of residuals and non-stationary issues in OLS. GWR split a single OLS equation into 3109 equations for each county. RESULTS: Among 15 SVI variables, a combination of eight variables showed the best model performance. Notably, unemployment, person with a disability, and single-parent households with children aged under 18 years especially impacted the variation of MUDs in OLS. GWR showed better model performance than OLS and specified each county's varying relationships between subcomponents of SVI and MUDs. For example, GWR specified that 69.3% (2157 of 3109) of counties showed positive relationships between single-parent households and MUDs across the USA. Higher positive relationships were concentrated in Michigan, Kansas, Texas, and Louisiana. CONCLUSIONS: Our findings could contribute to the literature regarding social determinants of community mental health by specifying spatially varying relationships between SVI and MUDs across US counties. Regarding policy implementation, in counties containing more social and physical minorities (e.g., single-parent households and disabled population), policymakers should attend to these groups of people and increase intervention programs to reduce potential or current mental health illness. The results of GWR could help policymakers determine the specific counties that need more support to reduce regional mental health disparities.

Mapping and dynamic monitoring of military training-induced vegetation cover loss using Sentinel-2 images and method comparison.

Sustainable management of the US Army installations is critical for military training and readiness of forces. However, monitoring military training-induced vegetation cover disturbances using remote sensing data is challenging due to the lack of methodology for optimizing the selection of spectral variables or predictors and spatial modeling methods. This study aimed to propose and demonstrate a methodological solution for this purpose. The study was conducted in the Fort Riley installation in which three training areas were selected to map and monitor the training-induced vegetation cover loss. Sentinel-2 images and field observations of percentage vegetation cover (PVC) were combined at a spatial resolution of 10 m by 10 m to map PVC and its dynamics by comparison of two predictor selection methods and five spatial modeling algorithms based on a total of 304 spectral variables from the images before and after the training. Results showed that overall, the correlation-based predictor selection method reduced the relative root mean square error (RRMSE) of PVC predictions by 4.44% than the random forest (RF)-based predictor selection. Machine learning methods including RF, neural network, and support vector machine overall reduced the RRMSE of PVC predictions by 42.83% compared with multiple linear regression and k-nearest neighbors. Combining correlation-based predictor selection and RF modeling, coupled with leave one out cross validation, provided the greatest potential of increasing the accuracy of monitoring the vegetation cover loss. The findings provided useful implications to develop a near real-time system of monitoring military training-induced vegetation cover loss.

Habitat suitability modeling of Descurainia sophia medicinal plant using three bivariate models.

Climate change has caused medicinal plants to become increasingly endangered. Descurainia sophia (flixweed) is at risk of extinction in Fars Province, Iran, due to climate change and modifications of land use. Flixweed is highly valuable because of its medicinal properties. The conservation of this species using habitat suitability modeling seems necessary. In this research, the geographical locations of D. sophia's distribution in southern Iran were recorded and mapped using ArcGIS 10.2.2. Then, ten important variables affecting the growth of D. sophia medicinal plants were identified and prepared as thematic layers. These variables were, namely, "elevation," "slope degree," "slope aspect," "soil physical characteristics (sand, silt, and clay percentage)," "soil chemical properties (EC and pH)," "annual mean rainfall," "annual mean temperature," "distance to roads," "distance to rivers," and "plan curvature." In this study, three bivariate models, including the "index-of-entropy (IofE)," "frequency ratio (FR)," and "weight of evidence (WofE)," were used for mapping the habitat suitability of D. sophia. Moreover, the ROC curve and AUC index were used for evaluating the accuracy of the models. Based on the results, the IofE model ("AUC": 0.93) was the most accurate, while the FR ("AUC": 0.92) and WofE ("AUC": 0.90) models ranked second and third, respectively. The models in this study can be applied as tools for the protection of endangered medicinal plants. Furthermore, the map could assist planners, decision-makers, and engineers in extending study areas. By determining the habitat maps of medicinal plants, their extinction can be prevented. Such maps can also assist in the propagation of medicinal plants.

A pruned feed-forward neural network (pruned-FNN) approach to measure air pollution exposure.

Environmental epidemiology studies require accurate estimations of exposure intensities to air pollution. The process from air pollutant emission to individual exposure is however complex and nonlinear, which poses significant modeling challenges. This study aims to develop an exposure assessment model that can strike a balance between accuracy, complexity, and usability. In this regard, neural networks offer one possible approach. This study employed a custom-designed pruned feed-forward neural network (pruned-FNN) approach to calculate the air pollution exposure index based on emission time and rates, terrain factors, meteorological conditions, and proximity measurements. The model's performance was evaluated by cross-validating the estimated exposure indexes with ground-based monitoring records. The pruned FNN can predict pollution exposure indexes (PEIs) that are highly and stably correlated with the monitored air pollutant concentrations (Spearman's rank correlation coefficients for tenfold cross-validation (mean ± standard deviation: 0.906 ± 0.028) and for random cross-validation (0.913 ± 0.024)). The predicted values are also close to the ground truth in most cases (95.5% of the predicted PEIs have relative errors smaller than 10%) when the training datasets are sufficiently large and well-covered. The pruned-FNN method can make accurate exposure estimations using a flexible number of variables and less extensive data in a less money/time-consuming manner. Compared to other exposure assessment models, the pruned FNN is an appropriate and effective approach for exposure assessment that covers a large geographic area over a long period of time.

Combining spatial autocorrelation with artificial intelligence models to estimate spatial distribution and risks of heavy metal pollution in agricultural soils.

Information on spatial distribution and potential sources of heavy metals in agricultural lands is very important for human health and food safety. In this study, pollution degree of lead (Pb), cadmium (Cd), and nickel (Ni) in Yüksekova Plain, located on the border in the southeastern part of Turkey, was evaluated by geoaccumulation index (Igeo), modified contamination factor (mCdeg), and Nemerow pollution index (PINemerow) combined with spatial autocorrelation using deep learning algorithms. A total of 304 soil samples were collected from two different depths (0-20 and 20-40 cm) in the study area, which covered 17.5 thousand ha land. Covariates were determined for spatial distribution models of Pb, Cd, and Ni by factor analysis (FA). Spatial distribution models for surface soils were developed using pedovariables (silt, sand, clay lime, organic matter, electrical conductivity, pH, Ca, and Na) determined by the FA and Igeo and mCdeg values by artificial neural network (ANN) and adaptive neuro-fuzzy inference system (ANFIS) models. The estimation success of models for different depths was assessed by root mean square error (RMSE), mean absolute percent error (MAPE), and Taylor diagrams. The RMSE and MAPE values showed a strong correlation between heavy metal contents and the covariates. The RMSE values of ANN-Ni0-20, ANN-Ni20-40, ANN-Pb0-20, ANN-Cd0-20, and ANN-Cd20-40 models (0.01240, 0.07257, 0.0039, 0.00045, 0.00044, and 0.04607, respectively) confirmed the success of the models. Likewise, the MAPE values between 0.2 and 8.5% indicated that all models were very good predictors. In addition, the Taylor diagrams showed that the estimation performance of ANFIS and ANN models are compatible. The IgeoNi and IgeoPb values in both models at both depths indicated that strongly to extremely polluted (4-5) areas are quite high in the study area, while the IgeoCd values revealed that unpolluted areas are widespread. The mCdeg index value showed a moderate to high contamination at the first depth, while very high contamination at the second depth in most of the study area. Spatial distribution of PINemerow revealed that moderate pollution (2-3) is common in both soil depths of the study area. The PINemerow of subsurface layer was between 0.91 and 1 (warning limit class) in a small part of the study area. The results showed that vertical mobility of heavy metals is closely related to pedovariables. In addition, the ANN and ANFIS models are capable of exhibiting the heterogeneity in the spatial distribution pattern of high variation in the data. Thus, the locations with extreme contamination have been accurately determined. The pollution indices calculated considering the commonly used international reference values revealed that heavy metal pollution in some part of the study area reached the detrimental levels for human health and food safety. The results suggested that the pollution indices were more successful than simple heavy metal concentrations in interpreting the pollution risk levels. High-resolution spatial information reported in this study can help policy makers and authorities to reduce heavy metal emissions of pollutants or, if possible, to eliminate the pollution.

Geonomics: Forward-Time, Spatially Explicit, and Arbitrarily Complex Landscape Genomic Simulations.

Understanding the drivers of spatial patterns of genomic diversity has emerged as a major goal of evolutionary genetics. The flexibility of forward-time simulation makes it especially valuable for these efforts, allowing for the simulation of arbitrarily complex scenarios in a way that mimics how real populations evolve. Here, we present Geonomics, a Python package for performing complex, spatially explicit, landscape genomic simulations with full spatial pedigrees that dramatically reduces user workload yet remains customizable and extensible because it is embedded within a popular, general-purpose language. We show that Geonomics results are consistent with expectations for a variety of validation tests based on classic models in population genetics and then demonstrate its utility and flexibility with a trio of more complex simulation scenarios that feature polygenic selection, selection on multiple traits, simulation on complex landscapes, and nonstationary environmental change. We then discuss runtime, which is primarily sensitive to landscape raster size, memory usage, which is primarily sensitive to maximum population size and recombination rate, and other caveats related to the model's methods for approximating recombination and movement. Taken together, our tests and demonstrations show that Geonomics provides an efficient and robust platform for population genomic simulations that capture complex spatial and evolutionary dynamics.

Soil Properties and Moisture Synergistically Influence Nontuberculous Mycobacterial Prevalence in Natural Environments of Hawai'i.

Nontuberculous mycobacteria (NTM) are opportunistic pathogens that cause chronic pulmonary disease (PD). NTM infections are thought to be acquired from the environment; however, the basal environmental factors that drive and sustain NTM prevalence are not well understood. The highest prevalence of NTM PD cases in the United States is reported from Hawai'i, which is unique in its climate and soil composition, providing an opportunity to investigate the environmental drivers of NTM prevalence. We used microbiological sampling and spatial logistic regression complemented with fine-scale soil mineralogy to model the probability of NTM presence across the natural landscape of Hawai'i. Over 7 years, we collected and microbiologically cultured 771 samples from 422 geographic sites in natural areas across the Hawaiian Islands for the presence of NTM. NTM were detected in 210 of these samples (27%), with Mycobacterium abscessus being the most frequently isolated species. The probability of NTM presence was highest in expansive soils (those that swell with water) with a high water balance (>1-m difference between rainfall and evapotranspiration) and rich in Fe-oxides/hydroxides. We observed a positive association between NTM presence and iron in wet soils, supporting past studies, but no such association in dry soils. High soil-water balance may facilitate underground movement of NTM into the aquifer system, potentially compounded by expansive capabilities allowing crack formation under drought conditions, representing further possible avenues for aquifer infiltration. These results suggest both precipitation and soil properties are mechanisms by which surface NTM may reach the human water supply. IMPORTANCE Nontuberculous mycobacteria (NTM) are ubiquitous in the environment, being found commonly in soils and natural bodies of freshwater. However, little is known about the environmental niches of NTM and how they relate to NTM prevalence in homes and other human-dominated areas. To characterize NTM environmental associations, we collected and cultured 771 samples from 422 geographic sites in natural areas across Hawai'i, the U.S. state with the highest prevalence of NTM pulmonary disease. We show that the environmental niches of NTM are most associated with highly expansive, moist soils containing high levels of iron oxides/hydroxides. Understanding the factors associated with NTM presence in the natural environment will be crucial for identifying potential mechanisms and risk factors associated with NTM infiltration into water supplies, which are ultimately piped into homes where most exposure risk is thought to occur.

Spatial factor modeling: A Bayesian matrix-normal approach for misaligned data.

Multivariate spatially oriented data sets are prevalent in the environmental and physical sciences. Scientists seek to jointly model multiple variables, each indexed by a spatial location, to capture any underlying spatial association for each variable and associations among the different dependent variables. Multivariate latent spatial process models have proved effective in driving statistical inference and rendering better predictive inference at arbitrary locations for the spatial process. High-dimensional multivariate spatial data, which are the theme of this article, refer to data sets where the number of spatial locations and the number of spatially dependent variables is very large. The field has witnessed substantial developments in scalable models for univariate spatial processes, but such methods for multivariate spatial processes, especially when the number of outcomes are moderately large, are limited in comparison. Here, we extend scalable modeling strategies for a single process to multivariate processes. We pursue Bayesian inference, which is attractive for full uncertainty quantification of the latent spatial process. Our approach exploits distribution theory for the matrix-normal distribution, which we use to construct scalable versions of a hierarchical linear model of coregionalization (LMC) and spatial factor models that deliver inference over a high-dimensional parameter space including the latent spatial process. We illustrate the computational and inferential benefits of our algorithms over competing methods using simulation studies and an analysis of a massive vegetation index data set.

Bayesian spatial models for voxel-wise prostate cancer classification using multi-parametric magnetic resonance imaging data.

Multi-parametric magnetic resonance imaging (mpMRI) has been playing an increasingly important role in the detection of prostate cancer (PCa). Various computer-aided detection algorithms were proposed for automated PCa detection by combining information in multiple mpMRI parameters. However, there are specific features of mpMRI, including between-voxel correlation within each prostate and heterogeneity across patients, that have not been fully explored but could potentially improve PCa detection if leveraged appropriately. This article proposes novel Bayesian approaches for voxel-wise PCa classification that accounts for spatial correlation and between-patient heterogeneity in the mpMRI data. Modeling the spatial correlation is challenging due to the extreme high dimensionality of the data, and we propose three scalable approaches based on Nearest Neighbor Gaussian Process (NNGP), reduced-rank approximation, and a conditional autoregressive (CAR) model that approximates a Gaussian Process with the Matérn covariance, respectively. Our simulation study shows that properly modeling the spatial correlation and between-patient heterogeneity can substantially improve PCa classification. Application to in vivo data illustrates that classification is improved by all three spatial modeling approaches considered, while modeling the between-patient heterogeneity does not further improve our classifiers. Among the proposed models, the NNGP-based model is recommended given its high classification accuracy and computational efficiency.

Linking remote sensing data to the estimation of pollination services in agroecosystems.

Wild bees are key providers of pollination services in agroecosystems. The abundance of these pollinators and the ecosystem services they provide rely on supporting resources in the landscape. Spatially explicit models that quantify wild bee abundance and pollination services in food crops are built on the foundations of foraging and nesting resources. This dependence limits model implementation as land-cover maps and pollination experts capable of evaluating habitat resource quality are scarce. This study presents a novel approach to assessing crop pollination services using remote sensing data (RSD) as an alternative to the more conventional use of land-cover data and local expertise on spatially explicit models. We used landscape characteristics derived from remote sensors to qualify nesting resources in the landscape and to evaluate the delivery of pollination services by mining bees (Andrena spp.) in 30 fruit orchards located in the Flemish region of Belgium. For this study, we selected mining bees for their importance as local pollinators and underground nesting behavior. We compared the estimated pollination services derived from RSD with those derived from the conventional qualification of nesting resources. We did not observe significant differences (p = 0.68) in the variation in mining bee activity predicted by the two spatial models. Estimated pollination services derived from RSD and conventional characterizations explained 69% and 72% of the total variation, respectively. These results confirmed that RSD can deliver nesting suitability characterizations sufficient for estimating pollination services. This research also illustrates the importance of nesting resources and landscape characteristics when estimating pollination services delivered by insects like mining bees. Our results support the development of holistic agroenvironmental policies that rely on modern tools like remote sensors and promote pollinators by considering nesting resources.

National Exposure Models for Source-Specific Primary Particulate Matter Concentrations Using Aerosol Mass Spectrometry Data.

This paper investigates the feasibility of developing national empirical models to predict ambient concentrations of sparsely monitored air pollutants at high spatial resolution. We used a data set of cooking organic aerosol (COA) and hydrocarbon-like organic aerosol (HOA; traffic primary organic PM) measured using aerosol mass spectrometry across the continental United States. The monitoring locations were selected to span the national distribution of land-use and source-activity variables commonly used for land-use regression modeling (e.g., road length, restaurant count, etc.). The models explain about 60% of the spatial variability of the measured data (R2 0.63 for the COA model and 0.62 for the HOA model). Extensive cross-validation suggests that the models are robust with reasonable transferability. The models predict large urban-rural and intra-urban variability with hotspots in urban areas and along the road corridors. The predicted national concentration surfaces show reasonable spatial correlation with source-specific national chemical transport model (CTM) simulations (R2: 0.45 for COA, 0.4 for HOA). Our measured data, empirical models, and CTM predictions all show that COA concentrations are about two times higher than HOA. Since COA and HOA are important contributors to the intra-urban spatial variability of the total PM2.5, our results highlight the potential importance of controlling commercial cooking emissions for air quality management in the United States.

Environmental Injustices of Leaks from Urban Natural Gas Distribution Systems: Patterns among and within 13 U.S. Metro Areas.

Natural gas leaks in local distribution systems can develop as underground pipeline infrastructure degrades over time. These leaks lead to safety, economic, and climate change burdens on society. We develop an environmental justice analysis of natural gas leaks discovered using advanced leak detection in 13 U.S. metropolitan areas. We use Bayesian spatial regression models to study the relationship between the density of leak indications and sociodemographic indicators in census tracts. Across all metro areas combined, we found that leak densities increase with increasing percent people of color and with decreasing median household income. These patterns of infrastructure injustice also existed within most metro areas, even after accounting for housing age and the spatial structure of the data. Considering the injustices described here, we identify actions available to utilities, regulators, and advocacy groups that can be taken to improve the equity of local natural gas distribution systems.