Search

Mathematical modelling of riots and protests is now becoming a powerful tool in providing a better understanding of dynamics of social unrest with the eventual goal to ensure a sustainable development of the human society. Currently, however, most of the existing studies in the considered research area are based on either non-spatial or spatially implicit models, whereas in a large number of cases dynamics of social protests clearly exhibit spatial heterogeneity. To bridge the existing gap, here we explore spatial-temporal patterns of social protests using a reaction-diffusion modelling framework. Our model variables are: the number of protesters and the cumulative amount of damage made as an outcome of the protest. The system has been studied analytically as well as by means of extensive numerical simulation in one dimensional and two-dimensional space. We show that the proposed model exhibits a variety of dynamical regimes including stationary patterns with round hot spots as well as complex labyrinthine-like structures. The system also predicts the various types propagating waves of protests with regular and irregular fronts as well as a patchy spread, where protests spread in space via irregular motion and interaction of separate patches of high numbers of protestors without formation of any continuous front, the number of protestors between patches being nearly zero. We reveal the structure of the model's parameter space, identifying the range of key parameters for which particular dynamical regimes are possible. Along with reaction-diffusion model, considering continuous space, we also consider the model of protests on discrete networks of different nature. We find that Turing instability can lead to pattern formation on networks, which opens up an exciting possibility to explore it as a generation mechanism in a large number of social unrest contexts. The networkbased model also shows a variety of non-Turing patterns including both stationary and non-stationary complex dynamics, where different nodes behave differently. Finally, we consider a realistic network to model the yellow vests movement in France.

Maximising the durability of crop disease resistance genes in the face of pathogen evolution is a major challenge in modern agricultural epidemiology. Spatial diversification in the deployment of resistance genes, where susceptible and resistant fields are more closely intermixed, is predicted to drive lower epidemic intensities over evolutionary timescales. This is due to an increase in the strength of dilution effects, caused by pathogen inoculum challenging host tissue to which it is not well-specialised. The factors that interact with and determine the magnitude of this spatial effect are not currently well understood however, leading to uncertainty over the pathosystems where such a strategy is most likely to be cost-effective. We initially use a spatially explicit model, incorporating seasonality and localised reservoirs of inoculum, to explore disease dynamics within landscapes containing a mixture of fields planted with either susceptible or resistant cultivars. When the spatial diversification of these fields is maximised, with lower aggregation of similar fields, the overall intensity of the landscape scale epidemic is reduced. The strength of this spatial effect however depends strongly on the pathogen dispersal characteristics, any fitness cost(s) of the resistance breaking trait, the efficacy of host resistance, and the length of the timeframe of interest. The conclusions drawn from this initial work, about how multi-strain disease dynamics respond to the scale of spatial diversification in a multi-host landscape, allow us to construct a general spatially implicit model that captures these fundamental dynamics. This new model features a novel method for incorporating spatial structure using an intuitive spatial aggregation metric that can be easily estimated from spatially explicit landscape data. The model is simple enough to be amenable to mathematical invasion analysis, while being flexible enough that questions of resistance durability can be thoroughly explored. In particular, results demonstrating interaction between spatial heterogeneity and cultivar cropping ratio are presented, an investigation that was not easily possible in our earlier more complex model. These results indicate that optimal spatial deployment strategies depend on a variety of factors, and may not necessarily be constant over time. Overall, these models allow us to make general predictions of the types of system for which spatial diversification is most likely to be cost-effective, paving the way for potential economic modelling and pathosystem specific evaluation. In addition, this approach for capturing detailed spatial structure and multi-species interactions within simple mathematical models could be applied to a wide variety of ecological and evolutionary systems. This study highlights the importance of studying the effect of genetics on landscape scale spatial dynamics within host-pathogen disease systems, as well as providing new mathematical tools to do so.

One of the key concerns in spatial analysis and modelling is to study and analyse similarities or dissimilarities between places over geographical space. However, ”global“ spatial models may fail to identify spatial variations of relationships (spatial heterogeneity) by assuming spatial stationarity of relationships. In many real-life situations spatial variation in relationships possibly exists and the assumption of global stationarity might be highly unrealistic leading to ignorance of a large amount of spatial information. In contrast, local spatial models emphasise differences or dissimilarity over space and focus on identifying spatial variations in relationships. These models allow the parameters of models to vary locally and can provide more useful information on the processes generating the data in different parts of the study area. In this study, a framework for localising spatial interaction models, based on geographically weighted (GW) techniques, has been developed. This framework can help in detecting, visualising and analysing spatial heterogeneity in spatial interaction systems. In order to apply the GW concept to spatial interaction models, we investigate several approaches differing mainly in the way calibration points (flows) are defined and spatial separation (distance) between flows is calculated. As a result, a series of localised geographically weighted spatial interaction (GWSI) models are developed. Using custom-built algorithms and computer code, we apply the GWSI models to a journey-to-work dataset in Switzerland for validation and comparison with the related global models. The results of the model calibrations are visualised using a series of conventional and flow maps along with some matrix visualisations. The comparison of the results indicates that in most cases local GWSI models exhibit an improvement over the global models both in providing more useful local information and also in model performance and goodness-of-fit.

Avian malaria (Plasmodium spp.) is a popular model system to study the ecology and evolution of parasite-host-vector interactions in the wild. These studies have historically focused mostly on the avian hosts and the malaria parasites. Knowledge regarding the role of vectors is essential to our understanding of these wild systems, but has only very recently started to accumulate. This thesis aimed to contribute to this field by assessing mosquito-malaria-host associations for British mosquitoes and the role of mosquito ecology in shaping these parasite systems in a British woodland study site, using molecular, field ecology and statistical modelling methodologies. From the 12 mosquito species or species groups found, I showed that the Cx.pipiens/torrentium mosquito group is likely to have a major role in avian malaria transmission in Great Britain, while Cs. annulata may be transmitting P. circumflexum. I also demonstrated a positive spatial association between mosquito density per host and avian malaria prevalence, in accordance with theoretical expectations for malaria transmission. Findings here provide evidence that avian malaria transmission in British woodlands is limited mainly to June-August, being preceded by relapse of previous infections or, alternatively, by maintenance of chronic blood parasitaemia through the colder months; this agrees with theoretical expectations and findings elsewhere for temperate climates. This thesis also described local-scale spatial heterogeneity and seasonal variation in adult mosquito abundance within a British woodland where avian malaria is endemic, with differing patterns found between species or species groups. Spatially, variation in adult mosquito abundance was associated with microclimatic and landscape variables such as distances to mosquito breeding sites, microclimate and canopy height; seasonally, variation in mosquito abundance was associated with temperature and rainfall, alongside calendar date. The heterogeneity in mosquito parameters and associations with environmental variables found at a site where avian malaria is endemic highlights the need to anticipate such complexity when trying to understand Plasmodium transmission. By doing so, we further extend the potential of these parasite systems to improve our knowledge regarding the ecology and evolution of parasite-host-vector associations.

Genetic differentiation in phenotypic traits among populations from heterogeneous environments is often observed in common-garden studies on forest trees, but data on adaptive variation in Scots pine (Pinus sylvestris L.) in Scotland are limited. As a result, current seed transfer guidelines are based on earlier molecular marker studies and do not take into account environmental or adaptive genetic variation. An analysis of spatial variation in climate showed substantial differences in temperature and precipitation among the native Scots pine sites in Scotland. To investigate whether differentiation in response to environmental variation has occurred in Scotland, a glasshouse-based common-garden trial of ~3,360 seedlings from 21 populations and 84 open-pollinated families was established in 2007. At the beginning of the 2nd growing season, timing of bud flush showed evidence of genetic differentiation among populations, with those from cooler origins generally flushing earlier. Variation was also found among families within populations, suggesting that the trait is genetically controlled. Populations and families showed different levels of variability in this trait which could be partly due to variable levels of temporal climate fluctuation in different parts of Scotland. Chlorophyll fluorescence was used to examine drought response in three-year old seedlings from five populations on sites that experience contrasting levels of annual rainfall. It was found that the response was not related to rainfall, but possibly to more complex moisture variables that also take into account additional factors such as evaporation. Also, photosynthetic capacity in response to cold winter temperatures varied significantly among eight populations that were kept outdoors, and the largest reduction was seen in seedlings from the mildest, most maritime coastal site. The following spring, height growth and needle flush started earlier in seedlings from cooler locations. Earlier studies on genetic diversity of native pinewoods have shown high levels of selectively neutral variation in this predominantly outcrossing conifer, and a mating system analysis with a limited number of microsatellite markers supported this pattern. Together, these data suggest that despite significant historic population size decrease, environmental gradients have resulted in genetic differentiation among native pinewoods. In order to minimise the risk of planting poorly-adapted stock and to maximise the success of replanting programmes, it is important that the origins of planting stock are carefully considered in management guidelines for the species.

This project predominantly investigated the implications of spatial heterogeneity in the ecological processes of competition and infection. Empirical analysis of spatial heterogeneity was carried out using the lepidopteran species Plodia interpunctella. Using differently viscous food media, it was possible to alter the movement rate of larvae. Soft Foods allow the movement rate of larvae to be high, so that individuals can disperse through the environment and avoid physical encounters with conspecifics. Harder foods lower the movement rate of larvae, restricting the ability of individuals to disperse away from birth sites and avoid conspecifics encounters. Increasing food viscosity and lowering movement rate therefore has the effect of making uniform distributed larval populations more aggregated and patchy. Different spatial structures changed the nature of intraspecific competition, with patchy populations characterised by individuals experiencing lower growth rates and greater mortality because of the reduced food and space available within densely packed aggregations. At the population scale, the increased competition for food individuals experience in aggregations emerges as longer generational cycles and reduced population densities. Aggregating individuals also altered the outcome of interspecific competition between Plodia and Ephestia cautella. In food media that allowed high movement rates, Plodia had a greater survival rate than Ephestia because the larger movement rate of Plodia allowed it to more effectively avoid intraspecific competition. Also the faster growth rate, and so larger size, of Plodia allowed it to dominate interspecific encounters by either predating or interfering with the feeding of Ephestia. In food that restricts movement, the resulting aggregations cause Plodia to experience more intraspecific encounters relative to interspecific, reducing its competitive advantage and levelling the survival of the two species. Spatial structure also affected the dynamics of a Plodia-granulosis virus interaction and the evolution of virus infectivity. Larval aggregation forced transmission to become limited to within host patches, making the overall prevalence of the virus low. However potentially high rates of cannibalism and multiple infections within overcrowded host aggregations caused virus-induced mortality to be high, as indicated by the low host population density when virus is presented. Also aggregated host populations cause the evolution of lower virus infectivity, where less infective virus strains maintain more susceptible hosts within the aggregation and so possess a greater transmission rate. The pattern of variation in resistance of Plodia interpunctella towards its granulosis virus was found using two forms of graphical analysis. There was a bimodal pattern of variation, with most individuals exhibiting either low or high levels of resistance. This pattern was related to a resistance mechanism that is decreasingly costly to host fitness.

This thesis consists of two self-contained essays on topics in applied macroeconomics. In the first chapter, I study how heterogeneous sensitivities to weather fluctuations and interregional production networks impact the measurement of weather shocks’ impact on economic activity in the United States. I start the analysis by building a general equilibrium model where the impact of weather fluctuations on productivity is state-sector dependent, and networks expose sectors to weather shocks from other regions through the use of intermediate inputs. Then, I quantify the relevance of these mechanisms, combining the model’s predictions with annual data on sectoral GDP and average temperatures by state from 1970 to 2019. My estimates show that models that do not consider these characteristics underestimate the aggregateimpact of weather fluctuations by at least a factor of 3. In particular, when the whole economy faces an unexpected increase in temperature of 1 Celsius degree, the contraction in economic activity increases from -0.13 to -0.37 percent once heterogeneity is considered and -1.14 percent when networks are included. In the second chapter, I propose a new methodology to disentangle between terms of trade movements caused by global shocks and those resulting from country-specific terms-of-trade fluctuations. This methodology extends the so-called maximum-share approach in two ways. Firstly, a global shock is identified as the shock with the highest explanatory power on the forecast error variance of a set of exogenous variables. This is in contrast to the typical approach of using only one variable as a source of information to identify a shock. Secondly, country-specific terms-of-trade shocks are identified as shocks that satisfy two conditions: (i) maximum explanation power on terms-of-trade variability and (ii) orthogonality to global shocks, allowing me to isolate the main drivers of terms of trade that are not related to global fluctuations. I apply this methodology to data on ten small open economies(SOEs) and show that global shocks contribute - on average- to 33 percent of their business cycle fluctuations. The contribution of global shocks to terms-of-trade variability is close to 20 percent, meaning that around 80 percent of terms-of-trade movements have country-specific origins. Interestingly, on average, country-specific terms-of-trade shocks are responsible for less than 10 percent of SOE business cycle variability. These results help to reconcile current estimates on the importance of terms of trade and suggest an intensive evaluation of the origins of terms-of-trade movements by policymakers before any intervention.

Urban flooding has become a growing threat to human society worldwide. Urban flood hazards have occurred more frequently, resulting in a growing population exposed to flood hazards and increasing flood damage to human communities. Thus, there is an urgent need to develop an urban pluvial flooding risk framework that can be used to evaluate the impacts of the flood and to provide helpful guidance on flood mitigation. The first chapter is a general introduction of this dissertation. It also reviews the existing studies related to physical and social factors affecting urban pluvial flooding. The research gaps are highlighted, and the aim of this dissertation is described. There is a lack of studies on investigating the spatial heterogeneity of the risk factors of urban pluvial flooding and their interplay impacts, and discussing how these may affect the urban flood mitigation strategies. The second chapter investigates the spatial relationship between the urban pluvial flood locations and the controlling factors through a case study in the City of Cincinnati. Three controlling factors are investigated, including precipitation, impervious area and topography. Two comparisons are conducted, one is depression-based which utilizes the random sampling method and the other comparison analysis is flooded-location-based. The results show that topography and precipitation are more important in the prediction of urban pluvial flooding risk than impervious areas. There is spatial heterogeneity of the three controlling factors as well as their interplay impacts. The third chapter develops an integrated approach for assessing urban pluvial flooding risk by combining two components, physical exposure and social vulnerability at catchment level. An exposure index (EI), a social vulnerability index (SoVI) and a composite pluvial flood risk index (PFRI) are developed. The results demonstrate the spatial heterogeneity of exposure and social vulnerability of urban pluvial flood, as well as the composite pluvial flooding risk across the study area. The impacts of different combinations of exposure and social vulnerability on urban pluvial flood can provide helpful insights in formulating tailored strategies in mitigating urban pluvial flood risk. The fourth chapter is focused on the mitigation of urban pluvial flood. In this chapter, a Storm Water Management Model (SWMM) is developed to compare the impacts of detention basins and their retrofits, and low impact development practices on erosive flows and urban flooding in two urbanized subcatchments under different return period storm events, including 1-year, 2-year, 5-year, 50-year, and 100-year. The results show that detention basin and detention basin retrofits are more effective in reducing peak flow and delaying peak time. The integration of detention basin retrofits and LID practice can contribute to 70.61% to 94.07% peak flow reductions in the designated storms. The last chapter discusses major findings and conclusions from the three studies conducted in this dissertation. This dissertation reveals the spatial heterogeneity of the physical and social factors affecting urban pluvial flooding and investigates the performances of different stormwater management practices on urban flooding.

Genetic gain for important agronomic traits can be accelerated in plant breeding by better understanding the environmental effects on field experimental plots over the growing season. Today, aerial image remote sensing via unoccupied aerial vehicles (UAVs) offers cost-effective collection of high throughput phenotypes (HTPs) with high temporal and spatial resolution. However, utilizing HTP to evaluate the genetic merit of tested accessions in a modern breeding program requires effective integration of the captured imagery with genome-wide marker data, experimental design and geographic information, agronomic phenotypic data, and soil and weather data. Furthermore, plant breeders require timely predictions of genetic effects for selecting accessions to advance; therefore, the data should be readily integrated into a quantitative genetics statistical framework. Hence, this dissertation presents four chapters: (1) a database schema for storing genome-wide marker data, (2) a web-database platform for managing plant breeding programs and their experiments, (3) a web-database tool for reliably processing aerial imagery into HTP, and (4) a statistical approach integrating HTP with genomic data to better resolve genetic effects over spatio-temporal environmental effects. In (1), a NoSQL data model is presented within the Chado database schema, utilizing the NoSQL and relational capabilities of PostgreSQL to link the genome-wide marker data and the plant breeding experimental data, respectively. Benchmarking demonstrates computation of a genomic relationship matrix (GRM) and a genome wide association study (GWAS) for datasets involving 1,325 diploid Zea mays L. (maize), 314 triploid Musa acuminata (banana), and 924 diploid Manihot esculenta (cassava) samples genotyped with 955,690, 142,119, and 287,952 genotype-by-sequencing (GBS) markers, respectively. In (2), Breedbase illustrates a web-database platform enabling plant breeders around the world to manage their breeding program data in a standardized process. Importantly, the Breeding API (BrAPI) allows open access and interoperability to the data. Then (3) focuses on ImageBreed as a web-database tool for processing aerial image phenotypes into HTP. Multi-spectral or color imagery from UAVs or from fixed camera systems can be uploaded, processed into orthophotomosaics if required, designated into geospatially referenced plot-polygons, and then summarized into vegetation indices (VIs) or convolutional neural network (CNN) HTP. In (4), the normalized difference vegetation index (NDVI) collected on several years of Genomes-to-Fields (G2F) hybrid maize (Zea mays L.) field experiments is used to improve genomic prediction for grain yield, grain moisture, and ear height. The proposed approach enables greater understanding of spatial heterogeneity in the field and improves the estimation of genetic effects. To conclude, continued aggregation of genomic and image data, coupled with statistical approaches, will enable plant breeders to better understand the stability of genetic effects across space and time. Future research into latent genetic spaces embedded in ground rover lidar point clouds and aerial imagery is an exciting avenue to understanding the permanent environment and genetic stability of accessions.

Heart failure is a broad pathology manifestation categorized by an inability of the heart to successfully pump blood throughout the vast vessel network of the body. Within the United States, heart failure is projected to increase by approximately 46% from 2012 to 2030. Modalities of heart failure are generally related to wall mechanics that are impacted following myocardial infarction events. Interplay exists between the wall mechanics, responding cell populations, and the spatial heterogeneities in the resultant scar. This interplay directs the myocardium towards heart failure modalities governed by overly stiff or compliant states. It is essential to elucidate details underlying the progression, maturation, and remodeling of the collagenous infarct scar to uncover therapeutic avenues and improve outcomes. We developed in vitro platforms that enable these desired insights into infarct progression. Our first platform generates continuous collagen gels possessing spatial fiber alignment heterogeneity utilizing a cost-efficient magnetic microsphere methodology. Spatial heterogeneities in the fiber architecture and cell response were evaluated. The second platform mechanically and electrically stimulates engineered heart tissue constructs and enables the evaluation of spatial heterogeneities in construct mechanics following a simulated infarction. Additionally, to avoid losing sight of the in vivo reality, we evaluated the interplay of the cellular detection of mechanics and spatial strain distributions throughout an infarcted myocardial wall. Histologically evaluated mechanosensor intensities were correlated to in vivo strain data collected via ultrasound imaging of the myocardium of a murine infarction animal model.

The emergence of antibiotic resistance in microbial populations is a complex evolutionary process dependent on a large number of factors, both mechanistic and external. We have modeled spatial heterogeneity as one such external factor and have found that it can greatly affect the times to resistance. We formed various models governed by simple atomistic rules in which antibiotic resistance can be treated as an emergent phenomenon and studied using classical tools from statistical mechanics, including a classical mean first passage time calculation and a kinetic Monte Carlo simulation. We have found that spatial heterogeneity can speed or slow the emergence of resistance, depending on the specific system and the relative rates of mutation and migration in spatially-extended systems. Finally, we developed a robust Bayesian inference pipeline to precisely estimate the relevant parameters from turbidostat evolution experiments. We believe that this work establishes environmental spatial heterogeneity as a critical factor in understanding the emergence of antibiotic resistance and may the groundwork for drug dosing strategies spatially-optimized to slow the emergence of resistance.

The McMurdo Dry Valleys (MDV) of Antarctica are on the threshold of widespread landscape scale change due to increasing temperature and solar radiation and altered hydrology: buried ice is melting, the soil active layer is thickening, thermokarst features are developing along streams, water tracks are expanding, and lake levels are rising. These changes will impact the microbial communities found in each of the affected habitats. The purpose of this work is to first, understand the spatial distribution of soil bacteria in the MDV, specifically investigating the scale-dependent effects of environmental heterogeneity, and second, to perform surveys and coupled experiments to document and assess the impacts of these changes on the associated microbial communities. Our results suggest that this region harbors highly adapted, endemic communities that are susceptible to changing environmental conditions.

This research attempts to understand the relationship between different impact factors and the residential land transaction price of Hangzhou, China. Exploiting various online open data sources and conducting detailed data preprocessing, this study builds a spatial econometric model to identify factors which significantly influence the transaction prices over the year 2012 – 2016. Furthermore, the model exams the spatial heterogeneity of different factors across the city. Adopting points of interest (POI) data from Gaode, one of the largest online map service companies in China, the research first defines the centers of different types of locational factors with K-means clustering. The nearest distances to these urban centers are computed accordingly to be set as potential explanatory variables. Collinearity issue is detected and variable selection is modified by variance inflation factors (VIF) calculation and principal component analysis (PCA). From this, land area, plot ratio, distances to major waterbodies, education centers, company centers, government centers, medical centers, tourism centers, opened subway stations and future subway stations which will be opened within three years are selected as independent variables for regression modeling. Finally, the non-stationarity of spatial distributions of estimated coefficients and pseudo t-values of ten explanatory variables indicate that Geographically Weighted Regression (GWR) is more suitable to analyze the spatial variation of impact factors with land price compared to conventional Ordinary Least Square (OLS) regression. The result of GWR model with fixed Gaussian spatial kernels indicates the spatial non-stationary relationship between residential land transaction price and the selected impact factors. The negative impact of most factors is larger in less developed areas than in highly developed areas. With Inversed Distance Weighting (IDW) interpolation improving the data visualization of coefficients spatial distribution, we can identify that land price change is more sensitive to the selective variables in the periphery regions of the city. Distance to job centers shows stronger influence on land price in the south especially the surrounding areas near the information industry agglomeration identified by K-means clustering. Moreover, two types of subway station factors are both contributing to the land price changes. The stations which will be opened within 3 years are influencing more areas in the city compared to the opened stations, while the magnitude of the coefficient is smaller. It reflects the anticipation of developers about land premium potential. The different spatial patterns of influences from the selected variables on land price implies that urban planners should acquire better understanding about local contexts and conduct location-specific strategies of land price evaluation and land use polices.

The water balance in mountain regions describes the relations between precipitation, snow accumulation and melt, evapotranspiration and soil moisture and determines the availability of water for runoff in downstream areas. Climate change is affecting the water budget of mountains at a fast pace and it has thus become a priority for hydrologists to quantify the vulnerability of each hydrological component to climate change, in order to assess the availability of water in the near future. However, our incomplete understanding of mountain hydrology implies that our knowledge about the future water supply of billions of people worldwide is limited. In this thesis, I use the ecohydrological model Tethys-Chloris (T&C) to (1) explore the responses of forests to the increasing atmospheric CO2, (2) quantify the major drivers of ecohydrological processes and their vulnerability to climate change across the European Alpine environments, and (3) partition the water budget into blue (hydrological) and green (biological) water fluxes and quantify the sensitivity of each component to temperature and precipitation change at the pan-Alpine scale. The first part of the thesis focuses on vegetation parameterization in ecohydrological models. It is a common practice to apply static vegetation parameters, although recently several studies have questioned this approach, showing that vegetation may adjust to climate change at shorter timescales than previously thought. This implies that traditional model approaches using temporally constant parameters might be biased. Recent evidence suggests that one such example of vegetation plasticity may be related to the increasing atmospheric CO2 concentrations. Through numerical simulations with T&C, I show that plasticity in key vegetation parameters can explain the changes in water and carbon vegetation fluxes in 20 forest sites in the northern Hemisphere; changes that otherwise cannot be explained. In the second part of the thesis, I explore the key drivers of Alpine ecohydrology. Applying T&C on three case studies, I quantified the drivers of ecohydrological fluxes and explored the vulnerability of different Alpine ecosystems to climate change. By correlating the spatial distribution of ecohydrological responses with that of meteorological and topographic attributes, I computed spatially explicit sensitivities of net primary productivity, transpiration, and snow cover to air temperature, radiation, and water availability to evaluate their absolute and relative importance. The results demonstrate the sharp differences between different parts of the Alps, thus highlighting the need for a high-resolution assessment of the Alpine water budget. The third part of the thesis addresses the ecohydrological sensitivity to climatic variables across the Alps. I collected a dataset from meteorological and environmental agencies and universities from six countries and combined it with new distributed products of meteorological forcing, soil properties, vegetation and snow cover to perform and validate large-scale, high-resolution ecohydrological simulations of the entire Alpine region for a period of three years (2001-2003). The focus in these simulations is on the partitioning of the pan-Alpine water budget into runoff and evapotranspiration. These simulations allowed us to quantify the sensitivity of each component of the pan-Alpine water budget to climate change and show that even during major heatwaves, Alpine vegetation keeps high transpiration rates, thus amplifying the decrease in streamflow. These unique simulations permitted vi for the first time a very detailed understanding of water sinks and sources at the scale of the Alps, and a correlation analysis revealed the potential “hot spots” of changes in the water cycle in space and time under climate change. This study also shows that recent advances in ecohydrological modeling, combined with large scale datasets and new computational capabilities, can bridge the gap between coarse-scale Earth system models and detailed catchment analyses, providing more reliable, high-resolution climate change impact assessments in the world’s mountain regions.

ABSTRACTSpatial Heterogeneity in Modeling Environmental and Human Health Impacts of ChemicalsbyMengya TaoThe number of chemicals that humanity is using for producing goods and services is rapidly increasing, while our understanding of their environmental and human health impacts improves slowly. Life cycle assessment (LCA) is one of the tools that evaluate environmental and human health impacts of chemicals. Traditional LCAs often rely on the models that represent broad spatial boundaries at regional, national, or global scales. However, the use, release, fate, and transport of chemicals, which are collectively referred to as biophysical processes, may vary substantially within those boundaries. This misalignment in spatial attributes between LCA models and the biophysical processes that determine environmental and human health impacts is one of the major sources of uncertainties in LCA. This dissertation is an attempt to close the gap between the spatial resolutions of the models used in LCA and the biophysical processes relevant for understanding the environmental and human health impacts of chemicals.This dissertation covers three topics that collectively addresses the aforementioned objectives: (1) measuring spatial variabilities in LCI, (2) modeling the fate of chemicals in the environment at the spatial resolution that matches with the underlying biophysical processes, (3) building a systematic release framework to estimate chemical releases that supports the fate modeling.First, I demonstrated that spatial disparities in state-specific LCI for four major crops in the USA can lead to two to fourfold differences in characterized results for most impact categories. The differences, however, increase to over an order of magnitude for freshwater ecotoxicity and human health non-cancer. Among the crops analyzed, winter wheat shows higher variability partly due to a larger difference in yield. As a result, the use of national average data derived from top corn and soybean producing states significantly underestimates the characterized impacts of corn and soybean in the states where land conversion from cotton to corn or soybean actually took place. Secondly, I developed a spatially explicit and time-dependent multimedia fate modeling framework, ChemFate, that can be incorporated into regional LCIA. ChemFate consists of four multimedia fate models: (1) organoFate, a model for non-ionizable organic chemicals, (2) ionOFate, a model for ionizable organic chemicals, (3) metalFate, a model for metals, and (4) nanoFate, a model for nanomaterials. ChemFate is able to not only provide predictions for four different classes of chemicals, but also incorporate dynamic emissions and dynamic environmental conditions. The dynamic capability of ChemFate supports the model to simulate with real regional climatic data and produce better model performance. Thirdly, I built a comprehensive release framework, OrganoRelease, to estimate the release of organic chemicals from the use and post-use of consumer products with limited information. OrganoRelease connects 19 unique functional uses and 14 product categories across 4 data sources and provides multiple pathways for chemical release estimation. The results can be used as input for methods estimating environmental fate and exposure.

Abstract: Surface mining in the Athabasca Oil Sands Region (AOSR), Alberta, Canada creates a large-scale ecosystem disturbance requiring ‘land reclamation’. Mining approvals require that land reclamation returns the sites to an equivalent land capability class, but this goal has proven challenging to quantify. Restoring ‘ecosystem function’ might be more realistic and perhaps, quantifiable by examining spatial patterns of soil functional indices. Cover soil prescriptions that emulate close-to-nature conditions and recreate similar realized niches for the reestablishment of native flora and fauna can increase reclamation success. Surface applications of forest floor mineral-mix (FFM) sourced from upland forest ecosystems and peat mineral-mix (PM) sourced from lowland ecosystems are termed “cover soil” and used as a proxy for replacing native surface soils in upland forest reclamation; yet, there is a wide margin between these surface amendments and native surface soils. The focus of this research first was the spatial characterization of heterogeneity of key soil functions such as nutrient bioavailability, soil respiration, and microbial biomass. We compared reclaimed sites and natural benchmark sites in the AOSR. Secondly, methods for improvement of cover soil performance were tested. In 2013 a field study was initiated that compared four different sites: 1) FFM reclaimed site; 2) PM reclaimed site; 3) type b ecosite recovering from fire; 4) type ab ecosite recovering from timber harvesting. We identified differences in spatial heterogeneity of nutrient profiles and soil respiration. Key bioavailable nutrients such as P were significantly different on PM and showed no heterogeneity. Seasonal patterns of respiration showed variability on natural reference sites and on FFM, indicating that disturbance had not removed belowground function completely. PM reclaimed sites showed no strong seasonal respiration patterns indicating homogeneous belowground function. During 2014 we measured soil microbial biomass (SMB) and soil respiration on six different sites. Two natural benchmarks were added including a 5) mature type a ecosite forest stand and 6) a type a ecosite recovering from fire to further characterize spatial heterogeneity on type a ecosites. The data indicated that there were differences in the pattern of soil respiration. Reoccurring patterns of soil respiration on the Harvested a/b ecosite indicated a linkage of above- and belowground function. The PM reclaimed site showed large-scale spatial patterns in SMB, similar to the type a ecosite affected by severe fire. The FFM site showed smaller scale spatial pattern than sites disturbed by clear-cutting and severe forest fire. We demonstrated that the amount of SMB and its heterogeneity increased with time since disturbance, potentially indicating stages of site recovery on benchmark type a ecosites. Boreal forest soils of by reclamation targeted ecosites are characterized by higher mineral soil fractions and notably they contain pyrogenic carbon (PyC) as a native soil component that affects biogeochemistry. Biochar is a humanmade analog for PyC and its amendment to cover soils used for reclamation might be a suitable method for reestablishing ecosystems that function more similarly to upland forest ecosystems recovering from fire. Observations indicated that tree growth on the tested PM reclaimed sites was lower in comparison to FFM reclaimed sites. Accordingly, methods for cover soil improvement were tested. The effect of admixing subsoil with peat and amendment of peat biochar on bioavailable nutrients, foliar nutrient concentration and stoichiometry, aspen (Populus tremuloides Michx.) growth, soil respiration, root exudation, and soil organic matter (SOM) stability was evaluated in two greenhouse studies. Seedling growth increased with admixing subsoil with and without biochar, and there was an overall positive effect of amendment with biochar in the first study and significant positive effect of biochar amendment on seedling growth and a reduction of soil respiration in the second study. Our findings suggest that seedlings grown on PM and peat-subsoil mixes were potentially affected by nutrient deficiency as well as toxicity. Biochar improved seedling nutritional status and soil organic matter stability was positively correlated with tree growth and increased with biochar amendment. Overall our studies demonstrate that by increasing heterogeneity in key soil functions, as well as assimilation of close-to-nature conditions, PM reclaimed sites can be improved. It is likely that this could be achieved by admixing of mineral subsoil, biochar amendment, and potentially targeted fertilizer application.

This dissertation contains three essays examining new Bayesian econometric methodologies. The first develops a heterogeneous Spatial Autoregressive Model by integrating a finite mixture model structure into the traditional homogeneous specification. The second essay builds upon the first by extending this Spatial Mixture Model structure to the more general Spatial Durbin and Spatial Durbin Error specifications. Additionally, this essay covers the interpretation of these new model specifications. Finally, the third essay develops a predictive based model selection process by integrating cross-validation algorithms into standard Bayesian sampling methods with a focus on explicit out-of-sample prediction. 1.0.1 Embracing Heterogeneity: The Spatial Autoregressive Mixture ModelIn this essay, a mixture distribution model is extended to include spatial dependence of the autoregressive type. The resulting model nests both spatial heterogeneity and spatial dependence as special cases. A data generation process is outlined that incorporates both a finite mixture of normal distributions and spatial dependence. Whether group assignment is completely random by nature or displays some locational "pattern", the proposed spatial-mix estimation procedure is always able to recover the true parameters. As an illustration, a basic hedonic price model is investigated that includes sub-groups of data with heterogeneous coefficients in addition to spatially clustered elements. 1.0.2 Spatial Durbin Mixture ModelsThis essay extends the finite mixture model structure to include Spatial Durbin and Spatial Durbin Error model specifications. The partial derivatives of this heterogeneous spatial model structure are shown to differ between border and interior agents; the designation of which is based on group assignment and first order neighbor designation. As an illustration, individual income based on data from the Panel Study of Income Dynamics (PSID) is examined using the Spatial Durbin Mixture Model specification.Results from the model indicate that returns to income from education are heterogeneous with some agents receiving negative returns on each additional year of human capital development. 1.0.3 Bayesian Predictive Model Selection using Cross-ValidationIn this essay, leave-one-out cross-validation is combined with Bayesian sampling and inference to develop a new model selection process. This process relies both upon standard hypothesis testing, and, in the absence of sufficient evidence, Ockham's Razor to selection a model from a competing set. It is shown that this process outperforms many of the criteria used in empirical work both by selected the true process more often, and selecting a smaller set of models overall.

The biological characteristics of fisheries stocks that are assessed for management considerations are rarely homogeneous over time or space. However, stock assessment scientists largely ignore this heterogeneity in their models. This thesis addresses the effects of spatial and temporal heterogeneity on stock assessment models using Atlantic Weakfish (Cynoscion regalis) as a case study. First, spatial and temporal variation was incorporated into length-, weight-, and maturity-at-age estimates using mixed-effects models (Chapter Two). The resulting heterogeneous weight and maturity parameters then were applied to per-recruit analyses to examine the sensitivity of biological reference points to spatial and temporal variation in life history attributes (Chapter Three). Mixed-effects life history models incorporating spatial and temporal variation revealed distinct regional and annual trends that were not visible from standard homogeneous models. In several instances, the homogeneous modelling approach produced life history estimates that varied significantly from the spatial and temporal means produced by the heterogeneous models. In some cases, this difference was so great that the homogeneous means were much higher or lower than the heterogeneous means for all regions or years. Minimized AIC statistics revealed that spatially and temporally integrated mixed-effects models were more robust and descriptive of Atlantic Weakfish life history than the standard homogeneous models. Per-recruit and biological reference points derived from these life history estimates in Chapter Three were found to be highly sensitive to spatial and temporal variations in weight parameters. In several cases, reference points used as management targets were so significantly different that ignoring spatial and temporal heterogeneity in Atlantic Weakfish life history would likely cause overfishing and decline of Weakfish in certain regions and years.

Climatic trends and anthropogenic interventions during past decades have undeniably impacted the hydrology and water quality of streams at the field, watershed, and regional scales in complex ways. This dissertation has developed a quantitative framework within which first-order estimates of the relative changes in travel time statistics in those landscapes undergoing substantial alterations in land-use land-cover can be quantified. Results from analysis of a sub-basin in the Minnesota River Basin located in the Midwestern United States indicated a significant decrease in the mean travel time of water in the shallow subsurface layer during the growing season under current conditions compared to the pre-1970’s conditions. Highly damped year-to-year fluctuations in the mean travel time were also found, which were attributed to the homogenization of the hydrologic response due to artificial drainage. Regarding the lack of high spatiotemporal resolution hydro-chemical data (in many catchments) required for an accurate estimation of transport time scales, this work has explored whether a lumped system representation via a time-varying stochastic Lagrangian formulation can provide reliable estimates of mean travel time in spatially heterogeneous catchments in the absence of within-catchment observations. This time (variability) for space (heterogeneity) substitution yielded mean travel times that are not significantly biased to the aggregation of spatial heterogeneity under different sampling assumptions. Despite the significant variability of mean travel time at small spatial scales, there exists a characteristic scale above which the mean travel time is not impacted by the aggregation of spatial heterogeneity. Extensive simulations of randomly generated river networks revealed that the ratio between the characteristic scale and the mean incremental area is on average independent of river network topology and the spatial arrangement of incremental areas. As the river network structure (which is typically considered self-similar) might impose a significant control on the in-stream transport time scales and be used to explain some properties of the relating physical processes, we further explored those river networks that do not obey the typical self-similar relationships over a range of spatial scales. We proposed methodologies that can probe into the structure of such outlier basins, called self-dissimilar, in ways that are able to reveal their spatially heterogeneous organization and quantify preferential scales of dissection, which in turn can influence the transport and storage time scales at large-scale catchments with implications for solute transport and water quality.

Continuous and discrete unobserved heterogeneity have been widely used in modeling discrete choice models. In this dissertation I investigate how these modeling strategies can be used to capture and model spatial heterogeneity or locally varying coefficients for different latent structures. In the first chapter, I outline the main advantages and disadvantages of both continuous and discrete spatial modeling strategies. Then I conduct a simulation experiment in order to understand the ability of both approaches to retrieve the true representation of the spatially varying process under small sample size situations. The results show that the data requirement to achieve lower bias in the continuous case is substantial compared with the discrete case. I have also found that, as the number of individuals per spatial unit increases, both models are able to identify the regional-specific estimates. However, the discrete case is able to retrieve the true spatial heterogeneity surface with lower bias and better coverage. In the second chapter, I show the Rchoice package for R that allows estimating models with individual heterogeneity for both cross-sectional and panel data. In particular, the package allows binary, ordinal and count response, as well as continuous and discrete covariates. This chapter is a general description of Rchoice and all functionalities are illustrated using real databases. The last chapter shows how continuous and discrete spatial heterogeneity models can be applied in order to analyze whether monetary subjective well-being eval- uations vary across space using a cross-sectional dataset from Chile. The results show that focusing just on the average estimates of compensating variations veils useful local variation. Moreover, the discrete approach shows some weak superiority over the continuous case in terms of model fit and interpretation.

This study has two primary focuses. The first is on the uneven distribution of high quality land across space. Of particular interest is how the spatial heterogeneity of land affects the location of households across regions, the development of land for urban and agricultural purposes within regions and the spatial structure of urban land use patterns within a given city. A second focus is how interbasin water transfers, which ameliorate the uneven distribution of water across regions, affect inter and intraregional land-use patterns. More specifically, the question this research asks is, what are the consequences of water transfers across regions when the preferences of households and the productivity of agricultural appears to favor the land in arid locations? And when land quality varies spatially, how does that alter the development patterns of cities and regions generally?This research develops a framework for analyzing the implications of interbasin water transfers on interregional migration and intraregional land use patterns. The model employed in Chapter 2 is a novel synthesis of the two dominant models in the urban economics literature, namely, the two region core-periphery model and the monocetric city model. However the model is modified to account for spatial disparities in the quality of land across regions. In particular, a scenario is explored where one region has a greater degree of natural amenities and thus, ceteris paribus, is preferred by households, as well as a comparative advantage in agricultural production. In addition, there are agglomeration externalities in the urban labor force, which, if sufficiently strong, leads to a concentration of all households in a single region. A second modification is that the more attractive region lacks water resources and, in order to satisfy household and agricultural water demand, must import water from the other region. This setup focuses on the tension between what is termed the amenity premium for households and the productivity premium for the agricultural sector, both of which compete for land in the same region. Chapter 3 introduces public infrastructure into the previous model. The infrastructure, which is endogenous and defined by the demand for water in each region, is financed through a flat tax. In addition, we consider that there are transport costs in the distribution of agricultural output across regions. This assumption allows for three different trade regimes. Autarky, in which each region produces agriculture solely for the local population. Incomplete specialization, in which the more productive region produces all local supply and any excess is sold to the other region to supplement local output. Complete specialization, where all agricultural production is concentrated in a single region.Chapter 4 turns to the issue of the heterogeneity in the quality of land within cities. The monocentric city modeling framework has developed a robust specification of urban spatial structure. In particular, it has shown how commuting costs play a crucial role in the spatial variation of housing rents, building heights and household living spaces. However, the model has been less capable in explaining the structure of cities at a more local scale. For instance, in cities with a dominant Central Business District (CBD) there does tend to be a decline in both building heights and housing rents as one moves further from the CBD. However, in a given neighborhood building heights and rents may not decline monotonically. There could be a host of reasons why this would be the case: historical factors, zoning regulations, idiosyncratic development patterns. Chapter 4 proposes that one such explanation is the heterogeneity of developable land over the space of a city.

Stream environments are spatially and temporally heterogeneous. Mainstem rivers are often wide, sunlit, warm and productive, while tributaries are shaded by riparian trees, unproductive, and remain cool in summer. Within mainstem rivers themselves there is substantial spatial heterogeneity in habitat structure and physical conditions, such as water temperature. River environments also change dramatically with season. Organisms that live in the riverine environment respond to and take advantage of such heterogeneous environments by moving between microhabitats or shifting their phenology. I studied the life cycle of a riverine mayfly, Ephemerella maculata (Ephemerellidae), in a northern California river system, its responses to spatial and temporal heterogeneity, and how its movements connect stream and riparian food webs in space and time. I discovered that E. maculata migrates between the mainstem and tributaries of rivers during its life cycle, thereby linking food webs in these two habitats, and enhancing predator growth in unproductive tributaries. The resource subsidy from productive but warm rivers to cool, unproductive tributaries associated with the mayfly migration increase the growth of stenothermic predators like juvenile salmonids in otherwise food-limited, cool thermal refuges, and increase their resilience to future warming. Furthermore, I examined the resilience of E. maculata to changes in water temperature using field surveys and lab rearing experiments. I discovered that different life stages of E. maculata have different thermal responses, and they shift their phenology depending on the water temperature, allowing each life stage to occur in the most desirable thermal condition. Therefore, as long as the natural seasonal pattern of the water temperature is sustained, E. maculata can resist temperature changes by shifting their phenology. Finally, I have shown that thermal spatial heterogeneity of rivers desynchronizes mayfly emergence timing, prolonging the subsidy period to riparian predators, and changing the predators’ responses to this subsidy.

Genetically modified (GM) crops are crops in which single or multiple genes have been introduced artificially in order to obtain certain characteristics that are difficult to obtain through conventional breeding. Even though farmers have the right to freely choose what types of crops to grow, pollen mediated gene flow from GM crops to non-GM crops can limit the possibility for crops to coexist on a same landscape, resulting in economic losses that depend on the institutional arrangements and the type of property rights in place. Although it is well known that spatial variability affects the degree of cross-contamination between GM and non-GM crops, no spatial analysis has been carried out to investigate how heterogeneity of landscapes influences the possibility for GM and non-GM crops to coexist. We aim with this research to analyze how spatial variability affects land allocation between GM and non-GM corn crops through a model composed of two parts: the first one simulates spatial units based on landscape criteria through Voronoi diagrams, and the second one reallocates the land between buffers, the GM and the non-GM crop based on cross-contamination and initial assignment of property rights. The model identifies coexistence clusters based on the deviation from an initial land allocation. We show that increasing spatial variability reduces the possibility of acceptable coexistence of crops and increases the economic losses. The economic impact from the assignment of property rights depends on the parameters that drive profitability differences (average market prices, yields and production costs). We show that buffer zones enforced to reduce cross-contamination result in less coexistence in heterogeneous spatial situations. We also elicit the economic value of unobserved factors that create a competitive advantage for certain farmers necessary for alternative crops to coexist on the same landscape.

Ecological forestry is an increasingly important tool for forest management and restoration efforts looking to incorporate the structures and spatial patterns created by natural disturbances into management practices. In northern Lower Michigan, jack pine (Pinus banksiana Lamb.) plantations are managed as habitat for the endangered Kirtland's Warbler (Dendroica kirtlandii Baird, KW) and have been criticized for creating homogenized stand structures, reducing natural jack pine regeneration, and excluding other threatened species. Historically, jack pine forests experienced stand-replacing wildfires on intervals of 26 to 69 years, and little is known about the impact these natural disturbances had on ground-flora composition and structure. If management plans are to incorporate ecological forestry principles, it is important to understand the disturbances and ecological processes that influence vegetation patterns and development. Ground-flora vegetation can indicate disturbance and environmental stress in an ecosystem and provide insight into ecological processes for forest managers. Ground flora composition, structure, and diversity were examined at 12 wildfire-regenerated jack pine sites either < 8 years post-wildfire (young) or > 22 years post-wildfire (mature) in northern Lower Michigan. Percent cover of woody and herbaceous vegetation ( < 1 m tall) was collected within nested 1-m^2 quadrats, in conjunction with information of the overstory composition and structure, and fuel loadings. Semivariograms were also used to examine the spatial autocorrelation in ground-flora species richness and Shannon's Index of Diversity. Young stands were dominated by blueberry (Vaccinium spp.) and sedges (Carex spp.), while mature stands were dominated by eastern teaberry (Gaultheria procumbens), mosses, and lichens. On average (± SD), mature stands had significantly higher species richness, with 13.4 ± 3.3 species per m^2 in young stands and 17.4 ± 4.8 species per m^2 in mature stands (P < 0.01). Gradient analyses suggest ground-flora composition and structure shift as stands develop over time, and these changes are related to changes in the overstory and fuel structure. Young stands showed strong positive correlation with seedling density and a negative correlation with overstory basal area, while the mature stands showed a positive correlation to overstory basal area. The mature stands also appeared to cluster into three groups along the first canonical axis, presumably along a productivity gradient (as indicated by the strong, positive relationship with live herbaceous cover). This suggests ground-flora composition and structure is variable within the wildfire-regenerated stands despite having similar environmental characteristics. Spatial analyses suggest that the variability of species richness and diversity is expressed at relatively small spatial scales, as the sill and range values were low for most of the sites. These results suggest that the ground flora developing in wildfire-regenerated jack pine forests can be variable and shift in terms of species dominance and functional groups over time. The changes in the ground-flora composition and diversity appear to be driven by stand structure and fuel loadings, factors influencing light availability and forest floor conditions. Restoration and management practices emulating natural disturbance patterns could enhance flora diversity and composition, as well as promote ecosystem services such as habitat for additional threatened bird species.

The accuracy of transpiration simulation by current land surface models has been limited by several uncertainties. This work focuses on effects of representing fine-scale spatially varying soil moisture, vegetation characteristics, radiation environment, as well as three-dimensional root water uptake processes on larger-scale energy fluxes at the plot- and watershed-scales. The study location is a northern temperate mixed forest with spatially heterogeneous canopy, where intermediate disturbance experiment removed canopy of aspen and birch trees. An initial analysis of empirical data on temporal dynamics of soil water content under the canopies with varying degrees of disturbance demonstrates that crown-scale canopy heterogeneity leads to detectable differences in soil water status, attributed predominantly to a decreased root water uptake in areas of thinner canopy. In order to infer larger-scale implications of crown-scale variations in canopy biomass and soil moisture conditions, an ecohydrologic model tRIBS + VEGGIE is used as an integrating tool to explicitly resolves spatially varying canopy biomass, radiative forcing, and soil moisture. Numerical experimentation shows that root-canopy controls of water uptake exhibit two opposing effects on soil moisture spatial variability. Further, variations in canopy light environments introduce non-linear effects into large-scale response of transpiration to soil moisture conditions: they result in smaller spatially aggregated transpiration and lower water stress as compared to traditional representations. To alleviate the constraints of commonly used heuristic one-dimensional approach for root water uptake in moisture-limited conditions, a novel formulation based on hydraulic representation of three-dimensional uptake process is designed. When mimicking soil drought at the scale of a single root system, the formulation demonstrates plants’ ability to compensate the suppressed root water uptake in water-stressed regions by increasing uptake density in moister regions. Pilot, exploratory experimentation at the plot-scale demonstrates that transpiration and timing of water stress onset are sensitive to different representations of lateral spread. The simulation scenarios with a higher degree of root overlapping exhibit smaller spatial heterogeneity of soil moisture and a later onset of plot-scale water stress. It is argued that hydraulics-based modeling of small-scale root water uptake can be used to guide improvements in the representation of drought effects in land-surface models.

It is well known that the interactions between diffusion and spatial heterogeneity could create very interesting phenomena. In this thesis, using the classical Lotka-Volterra competition system, we will illustrate the combined effects of dispersal and spatial variation on the outcome of the competition. We first show that, with the total resources being fixed at exactly the same level, a heterogeneous distribution of resources is usually superior to its homogeneous counterpart in the presence of diffusion. Then we study the more general case when both species have heterogeneous carrying capacities, but still with the same amount of total resources. Limiting behaviors of co-existence steady states as the dispersal rates tend to 0 or infinity are also obtained. In the end, we investigate the much broader situations - including different strengths and distributions of the resources, and with different competition abilities. Stability properties of semi-trivial and co-existence steady states are characterized under various circumstances.

Oak savannas of the Midwestern U.S. are among the most imperiled North American plant communities. The 478-km2 Oak Openings region of Northwestern Ohio is one of the few landscape-scale savanna systems remaining in the Midwest. Despite conversion of large portions of the Oak Openings for human land uses, the region still supports high levels of floristic diversity. However, regional patterns of Oak Openings plant diversity within the modern landscape are not well understood. My research objectives were 1) to determine the current extent and distribution of Oak Openings plant communities, 2) to quantify multiscale patterns of plant species richness within the context of the surrounding landscape, and 3) to build predictive species distribution models of rare plants to evaluate regional patterns in habitat suitability. First, using multi-seasonal Landsat images, I determined that

Natural ecosystems are exposed to numerous factors that determine their structure and dynamics. Although change is typically inherent in to biological systems, very few factors change a natural ecosystem as drastically as the arrival and successful establishment of non-native species. The impact of the Mediterranean grass (Schismus barbatus) in the southwestern U.S. exemplifies such dynamics. In my dissertation, I empirically and theoretically explore two components of Schismus invasion. First, native shrubs may promote Schismus invasion if Schismus benefits from the favorable growth conditions that shrubs provide. Second, Schismus may threaten ecosystem level properties if Schismus precludes establishment of shrubs. Thus Schismus may compromise the maintenance of Islands of fertility that favor growth of native plant community. To meet these objectives, I explore Schismus performance with respect to the location and size of one of the most abundant native shrubs in the Sonoran and Mojave Deserts, the Bur-sage shrub, Ambrosia dumosa. As a first approximation to the study system, I describe the natural distribution of Schismus with respect to the location of native shrubs (Chapter 1). Then, I examine factors that control Schismus-Ambrosia interaction using a three-year field experiment (Chapter 2), and subsequent characterization of soil nutrients changes around Ambrosia shrubs in the context of Schismus invasion (Chapter 3). I found that Schismus barbatus responds to the spatial heterogeneity formed by Ambrosia. This response is strongly controlled by water availability (Chapters 1 and 2). Ambrosia creates a dichotomous landscape formed by nutrient rich areas below their canopies and nutrient poor areas at inter-shrub spaces (Chapter 2). In this landscape, performance of Schismus is favored around Ambrosia during years with good precipitation. Nevertheless, at any given growing season, shading and water interception by shrub canopies are the most influential factors on Schismus performance. Given that germination was consistently low below Ambrosia canopies, the consequences of Schismus-Ambrosia interactions might not be as straightforward as predicted by the individual effect of Ambrosia on Schismus . In fact, Ambrosia might actually reduce realized Schismus densities. Conversely, inter-specific competition with Schismus regulates Ambrosia seedling survival. Thus, high-density stands of Schismus might generate a negative feedback on the dynamics of favorable areas associated with shrubs; ultimately limiting Schismus abundance. I design a modeling framework to test this hypothesis and present several predictions about the long-term demographic consequences of reciprocal interactions between Schismus and Ambrosia (Chapter 4).

Differences in house-level Anopheles densities (HLAD) are likely to affect the intensity of Plasmodium transmission and malaria. Spatial patterns of house-level An. gambiae and An. funestus densities in a 70 km2 area of western Kenya with year-round, intense Plasmodium transmission were characterized during one wet and one dry season to examine their spatial correlation structure and to test for global spatial autocorrelation. Traditional measures of spatial dispersion (k of the negative binomial distribution, Taylor's power law, Lloyd's index of mean crowding, Iwao's index of patchiness) and spatial statistics were used. Results indicated that both mosquito species had indices and parameters characteristic of non-random distributions and aggregated populations. To understand environmental factors associated with the HLAD spatial heterogeneity, vegetation indexes (from Landsat TM satellite images), house construction, house occupant characteristics, and geographical context (e.g. house density, distance to water bodies) were evaluated and multivariable linear models developed. Results for the wet season indicated that lakeshore distance and land cover characteristics were associated with variation in HLAD. In the dry season, geographical context variables, as well as land cover, were associated with HLAD, but the direction of this association differed by species. Models with interaction terms were significant and explained up to 39% of HLAD's variance. Spatial statistical methods were used to examine HLAD clustering patterns across time and season. Overlap was detected between high density or hot spot clusters of An. gambiae and An. funestus for both wet seasons that were studied. In the dry season, however, there were few hot spots and no overlap for either species. The low density or cold spot clusters, were similar for both species in both seasons. These findings provide new insight into how An. gambiae and An. funestus abundances are associated with ecological spatial heterogeneity, and how this might affect sampling programs, measures of population size, and interpretation of population and species dynamics. Results suggest different complex dynamics underlying each An. gambiae and An. funestus house-level densities. Sampling frames focused on individual or groups of houses may be more appropriate for classifying malaria exposure risk than those based on village-level estimates.

A study was conducted at the Savannah River Site near Aiken, SC to: 1) characterize the spatial patterns of soil and forest floor variables (moisture, pH, soil phosphate, forest floor and soil carbon and nitrogen, and soil available nitrogen), 2) assess the spatial patterns of the plant community, and 3) investigate spatial relationships among the variables and between the variables and woody vegetation. Spatial soil and litter samples were collected on five 0.25 hectare plots, and relationships were explored using Pearson's correlation tests, canonical correlation analysis, variogram modeling and kriging. The average range of spatial autocorrelation for the forest floor variables was >45 m, while that for soil variables was 12 m. Woody stem basal area exhibited spatial autocorrelation at ranges of less than 12 m, and was only weakly correlated with forest floor and soil resource patterns. Few strong spatial correlations among the forest floor and soil variables were observed. The means and variances of the variables were low, and differences in resource levels probably had little impact on the spatial pattern of vegetation. Results indicate a weak, differential effect of species group on litter quality, a weak relationship between large pine trees and soil nitrogen patterns, and a general homogeneity of the stands.

The endangered monophagous coxella weevil, Hadramphus spinipennis (Coleoptera: Curculionidae) is confined to two outlying islands of the Chatham Island group. This study was carried out on Mangere Island. The native forest, which used to cover Mangere, has been burned and replaced by grassland. A metapopulation of weevils inhabits 11 distinct host plant (A. dieffenbachii, Apiaceae) patches. Adult weevils feed on the foliage and flowers of A. dieffenbachii. Eggs are laid from September until May. Larvae hatch after two to three weeks and feed on the root parenchyma. Development from egg to adult takes between six months and a year. Pupation takes place in the soil next to the roots. Adults are flightless. The weevil responds to host plant specific olfactory cues. Mangere was visited six times between summer 1994 and summer 1997. To assess weevil-host plant interaction and to investigate if the weevil causes local extinction of its host plant, the population dynamics of H. spinipennis and its host plant were investigated in one discrete patch. Population estimates for the weevils were obtained with a capture-recapture study using Pollock's robust design. Location, phenology and size of A. dieffenbachii plants were recorded once a year. Weevil numbers more than quadrupled over three consecutive summers and survival and recruitment rates increased. Plant numbers halved over the same period. By the fourth summer the plant population had collapsed and no weevils were found. Overexploitation, particularly via root feeding, caused host plant extinction. Weevil intra-patch movement was studied to determine the mobility of the weevils at varying food levels and the extent to which they leave a host plant patch. Over 90% of the weevils stayed within 0-6 m of where they were caught the previous day. Intra-patch movement followed a random walk model. High survival between capture occasions (during one visit) suggested that the weevils had a low tendency to leave a patch. An annual census was used to estimate weevil and host plant abundance for six patches and to assess if local weevil populations are regulated by density-dependence. The total weevil population almost tripled over three consecutive summers while plant numbers increased by 11%. No density-dependent regulation of local weevil populations was detected. After a weevil density of between 18 and 33 per plant was exceeded, three patches collapsed. During each visit all patches were searched for marked weevils to assess inter-patch movement before and after the collapse of a local patch. Inter-patch movement was never detected before the collapse of a patch, but subsequently weevils were found in all neighbouring patches. The weevils' low tendency to disperse in the presence of food, yet high dispersal ability in its absence, can be explained by the natural habitat. Weevils and host plants were originally confined to rugged coastal areas, where the risk of inter-patch dispersal is likely to be greater than the disadvantage of intra-specific competition. After the collapse of a local patch, however, high dispersal power is needed to reach a new one. The rate of change in weevil numbers was not spatially correlated. Therefore, stability of the weevil metapopulation is probably maintained by spatial heterogeneity acting in a 'shifting mosaic' mode.

Wildlife management and natural resource policy decisions are increasingly being made at the landscape level. Understanding the relationship between modification and the pattern of land classes may minimize potential impacts and enhance the complement of wildlife species. Twenty-four expressions were selected for landscape analysis that describe the spatial heterogeneity, fragmentation, edge characteristics, and connectivity of pattern. Metric relationships were characterized across a variety of landscapes. Cluster analysis organized the metrics into classes quite different from the classification categories used in the literature. Cluster membership reflected the number of land classes, the amount and distribution of forest cover, number of forest patches, patch position, patch shape, patch radius, and edge length. Cartographic modeling was used to determine how modification influenced landscape pattern. The models depicted spatial relationships resulting from proposed landscape changes. Timber harvest schemes with a few large units and those in clustered arrangements led to less fragmentation than those schemes with several small units or those dispersed across the landscape. The placement of roads had either an invasive or partitioning effect on landscape pattern. Discriminant analysis rated the effectiveness of pattern expressions for environmental assessment. Metric effectiveness differed among the timber harvest, road expansion, and deforestation modification schemes. The utility and limitations of each expression was discussed. Sensitivity analyses examined the effects of changing spatial scale on pattern description. Scale influence was dependent upon landscape complexity, distribution of land classes, and the size and shape of those classes. The loss of ability to detect localized variability, to differentiate among spatial patterns, and to represent boundary detail accompanied the use of large pixels (420m²). There was evidence that spatial scale influences habitat evaluation. Semivariogram analysis assessed the constancy of expression behavior during changes in scale and presented the limits of tolerance for using large pixels in pattern analyses. The variability observed suggested that pattern misrepresentation occurred at coarse resolution levels. The successful application of landscape analysis depends on the ability to quantify pattern. By analyzing and understanding selected aspects of landscape pattern, I have examined how wildlife management can be enhanced through a knowledge of the landscape.