The relationships between people and natural environments in coupled socio-ecological systems (SES) are complex. As complex adaptive systems with humans as one of the central drivers, coupled SES exhibit non-linear behavior, multiple stable-states, path dependence, and highly dynamic webs of connectivity across domains and scales. It should come as no surprise therefore that efforts to push these relationships towards greater sustainability are challenging to design and execute, particularly in resource constrained contexts with multiple competing actors. In this dissertation I examine conservation efforts in Zambia and Namibia from perspectives that vary along two gradients. Theoretically, my analyses are based upon epistemologies ranging from highly qualitative, interpretative and relational explanations of project outcomes, to highly quantitative models of spatially structured processes. Empirically, I draw upon ethnographic methods that seek to develop a grounds-eye view of reality, to satellite-sensed data which capture large scale patterns at the expense of context and detail. These two poles are bridged by an information framework that employs the meta-language of vision to describe and bound different forms of understanding, with an emphasis on rich articulations of context to enable dialogue across knowledges.

Part I of the dissertation engages a long-standing debate about community-based natural resource management (CBNRM), which together with protected areas has been one the core conservation strategies across countries in southern Africa. Since the advent of CBNRM in the mid-1980s, scholars and practitioners have sought to explain the uneven performance of CBNRM programs. Most CBNRM assessments examine the underlying principles of community-based conservation, the local social and ecological contexts, and connections with larger political and historical patterns. I argue that analysis of the potential and pitfalls of CBNRM also requires an understanding of the institutional history and internal dynamics of projects that implement CBNRM reforms. Drawing upon theory and methods from development ethnography and public policy, I examine the rise and fall of CONASA, a second-generation CBNRM project in Zambia that operated in the early 2000s. CONASA was constituted from a merger of organizations and discourses to create continuity with previous projects. Its ambitious suite of activities included support for household livelihoods, community-based resource management, policy analysis, advocacy, and conservation enterprises at local, national, and transboundary levels. While individual activities were largely successful, CONASA's hybrid origins and logframe-centric management created fissures between its holistic design and daily operations, and hindered its ability to develop a broader narrative and maintain key alliances. This study illustrates the importance of understanding the interplay between project design and operational context in order to fully appreciate the possibilities and limitations of project-mode conservation.

While a ground's eye view highlights context and nuances of process, the bird's eye view reveals pattern and emergent behavior across scales. In Part II, I present a new spatial modeling method for location data from orbiting satellites to analyze the spatiotemporal patterns in movement data. Advances in GPS technology have created both opportunities in ecology as well as a need for analytical tools that can deal with the growing volume of data and ancillary variables associated with each location. Time Local Convex Hull (T-LoCoH), is a home range construction algorithm that incorporates time into the construction and aggregation of local kernels. Time is integrated with Euclidean space using an adaptive scaling of the individual's characteristic velocity, enabling the construction of utilization distributions that capture temporal partitions of space as well as contours that differentiate internal space based on movement phase and time-use metrics. I test T-LoCoH against a simulated dataset and provide illustrative examples from a GPS dataset from springbok in Namibia. The incorporation of time into home range construction expands the concept of utilization distributions beyond the traditional density gradient to spatial models of movement and time, opening the door to new applications in movement ecology.

Hulls also provide a means for characterization of the interactions between individuals, which are central to many aspects of population biology, including competition, predation, reproduction, resource optimization, and disease transmission. Classic measures of association are based on the intersection of static home range estimates and counts of matched locations, but the advances in GPS technology have created new possibilities for characterizing dynamic spatiotemporal properties of association. Hull metrics of association for both spatially and temporally overlapping hulls reveal the intensity of interaction both in raw terms as well as corrected for revisitation, as well as time lags. Similarly, a metric for the proportion of enclosed points constructed from the combined set of locations reveal the distribution of sharing across space and time for the entire time period. Illustrative examples of the methodology are provided using elephant tracking data from Tembe Elephant Park in South Africa. Plotting association metrics on a map, over time, and against other hull properties reveal novel patterns that can be basis for further study. Local hulls offer a promising approach for characterizing the spatiotemporal properties of association and exploring their covariates.

Parts I and II of the dissertation both illustrate ways to pull out and characterize pattern in complex systems, but at scales and methods appropriate for the data and question. The challenge in seeing the 'whole picture' of complex coupled SES is not only to draw upon multiple disciplines and perspectives, but to do so in ways that enable dialog. Central to bridging nodes of knowledge is aggregation of detail into accessible and meaningful abstractions, embedding contextual information to inform new connections and enable extensions into new systems and domains, and making visible the spatial and social vantage points through which knowledge is produced and disseminated.

Exposure represents but one of several processes that underlie disease transmission dynamics in animal and human populations. Infection frequently depends on a number of complex interactions among factors related to the clinical properties of the pathogen (or the magnitude of the dose acquired upon contact) and the immune status of the host. When considering exposure, however, many of these aspects become trivial; the primary consideration is contact between a host and the infectious agent, whether it is harbored by another animal or an environmental reservoir. Contact, in turn, emerges from the space-use decisions of animals over time, potentially resulting in patterns that amplify or dilute the probability of encountering a pathogen. In this sense, the movement behavior of host animals is a fundamental determinant of disease dynamics. Using anthrax as its focal system, this dissertation aims to delve into the exposure process as it relates to the movement behaviors of host animals.

A set of movement trajectories were collected via GPS collars fastened to zebra (Equus quagga) and springbok (Antidorcas marsupialis) in Etosha National Park in Namibia between 2009 and 2010. These data offer insight into the home ranging and habitat selection behaviors that characterize two ungulate species exhibiting susceptibility to anthrax infection and thus, form the basis of the analyses and models developed in this dissertation.

Spatial overlap analysis represents one of the most common methods for evaluating the potential for disease exposure when movement data is available. In the case of an indirectly transmitted pathogen, such as Bacillus anthracis, the overlap between individuals may be less important than other characteristics of individual home range usage. Metrics such as revisitation (the rate at which an animal returns to a specific location) and duration rate (the length of time spent in a specific location) may be more informative, particularly if the locations of locally infectious zones (LIZs) are known. To assess the relative risk faced by zebra and springbok during the anthrax season, I developed a method that reduces the subjectivity in parameter selection when delineating home ranges using the Time Local Convex Hull (T-LoCoH) method. Using a cross-validation-based approach, the resulting site fidelity metrics are more directly comparable. The high values of the two site fidelity metrics imply that similar home ranging behavior among individuals can result in heterogeneous outcomes, contingent entirely upon the presence of a LIZ within an individual's home range.

Much like spatial overlap analyses, habitat selection approaches can offer insight into patterns of potential risk with respect to exposure to disease, particularly in the case of environmentally-borne pathogens. When certain environmental characteristics can be associated with pathogen persistence, niche models can be developed and directly incorporated into the resource selection function framework. I used remotely sensed data on soil, bioclimatic, and vegetation covariates to build such a niche model for anthrax based on soil samples from 40 carcass sites in Etosha National Park harboring viable anthrax spores two or more years after deposition. When this risk layer was applied as a predictor in a step-selection function of zebra, a behaviorally-dependent pattern was evident. When animals were in the foraging state exhibited an avoidance of high risk areas, whereas the same animals were apparently attracted to those higher risk areas when moving in a directed manner. One possible explanation for this pattern is that zebra recognize not only where but also when they are most susceptible to anthrax, and adjust their behavior to reduce their risk.

Another means of exploring the exposure process is through the use of simulation models. Due to the difficulty associated with comprehensive monitoring of susceptible host populations and infectious reservoirs, simulation models represent an ideal approach for extending general rules emerging from limited movement data to landscapes with known qualities. Using the behaviorally-contingent habitat selection framework created in Chapter 3, I explored the relationship between a set of environmental covariate layers and the exposure process whereby individuals encounter LIZs on the landscape. The method reveals that Wetness may represent a reasonable predictor of epidemic dynamics, with movement serving as the mediating process.

The general analytical methods and models applied here serve to elucidate the role of individual movement behavior in the disease exposure process. Rather than analyzing data on case incidence or prevalence, these methods offer insight into the potential contact patterns that might give rise to endemic or epidemic infections. Thus, they reveal the manner by which analysis of host movements, particularly in conjunction with comprehensive (or simulated) data on the spatial distribution of infectious agents on a heterogeneous landscape, might aid in the management of transmission risk before any actual infections occur.

Concerns about the sustainability of wildlife hunting, particularly in Central Africa, have dominated the scientific literature on wildlife utilization. Only more recently have researchers began considering the human dependence on wildlife for both nutritional needs as well as sources of livelihoods. I begin with a systematic literature review of the wildlife harvesting literature, examining in detail the type of sustainability indicators predominating in the literature and their strengths and weaknesses. We find that indicator type, continent of study, species body mass, taxonomic group, and socioeconomic status of study site are important predictors of the probability of reported sustainability. Indicators relying on population-specific biological data are most often used in North America and Europe, while cruder estimates are more often used in Africa, Latin America, and Oceania. Our results highlight both the uncertainty and lack of uniformity in sustainability science. This presents a major gap in our ability to monitor wildlife and its use, since the importance of wildlife for human consumption is at its greatest precisely in the places where indicators used are the weakest. We point to future directions in the field.

Subsequent field work was conducted in the humid forest zone of southeastern Cameroon in Central Africa. Cameroon is one of the six Congo Basin countries, and there has traditionally been great concern on the part of environmental conservation organizations over the level of wildlife hunting in the country. The first part of my field work was a pilot study to compare field methodologies for wildlife consumption by rural peoples. For wildlife surveys, I used distance sampling on wildlife transects to determine presence/absence and abundance of wildlife species in four different village sites that represent a gradient of human impact and environmental intactness. To evaluate human use of wildlife relative to economic status, I used household surveys with heads of households to ascertain relative wealth status and other household demographic and economic parameters as they relate to wildlife use and consumption. Finally, I tested methodologies for enumerating hunter activity and catch-per-unit-effort (CPUE) as a potentially useful tool for monitoring the status of hunting sustainability. Results indicate that transect surveys do in fact detect increasing wildlife species in more rural, intact village sites, although sample sizes were too small to enumerate actual wildlife densities. Further, more rural households tend to both hunt and consume more wildlife; wildlife use in rural areas thus forms a more important source of total livelihood than for more urban households.

Although humans have hunted wildlife for millennia, and it remains an important source of animal protein, there is increasing concern that `bushmeat' hunting, particularly in central Africa, is unsustainable. We explore the role that wildlife and alternative meat sources play in the food security of human populations in southeastern Cameroon. We conducted a large, cross-sectional study in 24 village and town sites in southeastern Cameroon to evaluate the role of wildlife in human food security in a gradient from urban to rural households. Rural households are significantly more likely to rely on wildlife for animal protein, whereas urban households rely on significantly more domestic meat. Using generalized linear mixed modeling, we found significant associations between bushmeat hunting and consumption and positive effects on food security, highlighting the importance of wildlife to human security in the Congo Basin. We asked interviewees about most consumed and most preferred wildlife species; interestingly, there is a potential synergy between taste preferences and the more resilient species that are hunted.

These results indicate that wildlife consumption plays an important role in human food security in the humid forest zone of southeastern Cameroon. Disappearance of wildlife would negatively impact the food security situation in the region, particularly in the forms of protein-energy malnutrition and iron deficiency. At present, there is little ability to maintain small animal husbandry due to the poor veterinary services throughout the region. I evaluate the cost-effectiveness of a `Heifer International' model extended to two of the ten provinces in Cameroon that make up the region where wildlife hunting is currently the most important form of animal protein, consisting of the Southern and Eastern provinces ("Regions"), which together have a population of about 1 million people. The Heifer International model would replace the animal protein traditionally taken from wildlife sources with a revolving "micro-loan" of livestock, that must be eventually passed on to neighbors. At a population density of around 8 people/km2, wildlife hunting is believed to be at least four times above maximum sustainable wildlife hunting rates, and therefore supplementary forms of animal protein need to come from elsewhere. Assuming administrative and training costs are included in the prices of the animals as estimated, a `Heifer International' model of small animal husbandry would be a cost-effective way to address protein-energy malnutrition and iron deficiency, as well as wildlife conservation concerns in this part of the world.

We describe a new method for estimating the area of home ranges and constructing utilization distributions (UDs) from spatial data. We compare our method with bivariate kernel and alpha-hull methods, using both randomly distributed and highly aggregated data to test the accuracy of area estimates and UD isopleth construction. The data variously contain holes, corners, and corridors linking high use areas. Our method is based on taking the union of the minimum convex polygons (MCP) associated with the k-1 nearest neighbors of each point in the data and, as such, has one free parameter k. We propose a "minimum spurious hole covering" (MSHC) rule for selecting k and interpret its application in terms of type I and type II statistical errors. Our MSHC rule provides estimates within 12% of true area values for all 5 data sets, while kernel methods are worse in all cases: in one case overestimating area by a factor of 10 and in another case underestimating area by a factor of 50. Our method also constructs much better estimates for the density isopleths of the UDs than kernel methods. The alpha-hull method does not lead directly to the construction of isopleths and also does not always include all points in the constructed home range. Finally we demonstrate that kernel methods, unlike our method and the alpha-hull method, does not converges to the true area represented by the data as the number of data points increase.

Rangelands cover about half of Earth's land surface, encompass considerable biodiversity, and provide pivotal ecosystem services. However, rangelands across the globe face degradation due to changes in climate, land use, and management. Moreover, since herbivory is fundamental to rangeland ecosystem dynamics, shifts in the distribution of herbivores lead to overgrazing and desertification. To better understand, predict, and prevent changes on rangelands it is important to monitor these landscapes in a timely and efficient manner. Remote sensing can be a viable tool for measuring such change. However, the high spatial and temporal variability of rangeland vegetation, high reflectance from soil background and senesced vegetation during prolonged parts of the year, present challenges to the application of remote sensing in these ecosystems. The goal of my dissertation is to address these challenges and advance the application of remote sensing and geographic information system (GIS) to quantify vegetation and herbivores on rangelands across the world. My dissertation aims to address the connections among three main components of rangelands: the landscape, herbivores and human factors. I first develop a method to characterize the rangeland landscape by measuring and mapping detailed vegetation variables in Etosha National Park, Namibia. Etosha is a 22,270 km2 semiarid savanna, which encompasses great diversity of flora and fauna. I then examine how landscape variables affect the movement patterns of a large mammalian herbivore that is a keystone species in Etosha, the African elephant (Loxodanta africana). Finally, I develop tools to monitor how herbivory affects the productivity of rangelands conservation easement in California.

In the first chapter, I outline the importance of rangelands and the threats these ecosystems face. I review the main challenges of measuring change processes on rangelands and describe some of the remote sensing based approaches that have been used to address these challenges. In the second chapter, I show that time series analysis of Moderate Resolution Imaging Spectroradiometer (MODIS) vegetation indices can produce excellent results in predicting detailed field measurement of vegetation in Etosha. Using three innovative approaches I improve the prediction of both woody and herbaceous vegetation on the landscape, providing good measurements of vegetation cover, density, and biomass over large spatial extents. First, I develop field methodology that combines visual estimation of vegetation cover and vegetation type together with accurate field measurements. Second, by integrating time series of remote sensing data over six years and consolidating this information with partial least square regression, I achieve accurate models of vegetation measurements. Third, by using four different MODIS-based vegetation indices: Normalized Difference Vegetation Index (NDVI), Enhanced vegetation index (EVI), Leaf Area Index (LAI), and Fraction of Photosynthetically Active Radiation (FPAR), I measure different vegetation forms - grasses, shrubs, and trees, and thereby provide valuable information for monitoring and conserving Etosha's savanna vegetation. An exciting result from this chapter is the ability to transfer the application of these models in space, to other parts of the reserve, and in time, to other seasons and years. This emphasizes the validity of the models I have developed for predicting vegetation measurements and the ability to use these models in other locations.

In the third chapter, I use the detailed vegetation maps I have created for Etosha National Park to understand resource selection of African elephants. I show how landscape variables affect both the speed and the direction of elephants' movements. Elephants prefer to move into areas with higher grass and shrub biomass, but lower tree biomass. Moreover, elephants prefer to be closer to water sources and, interestingly, to roads. Elephants' resource selection is influenced by the sex and the age of the individual. Importantly, temporal variation significantly influences the movement in response to the landscape: elephants choose different resources at different times of the day, which illustrates the behavior underpinnings of their resource selection. Moreover, they respond differently to resources at different times of the year, which highlights the ecological importance of these resources to the elephants. This chapter provides valuable information on how to manage resources in a manner that will promote the conservation of these magnificent keynote animals.

In the fourth chapter, I use MODIS satellite data to monitor the effects of grazing on rangeland conservation easements in California, using as a study case the Simon Newman Ranch, a conservation property own by The Nature Conservancy. I use time series information of three vegetation indices to measure Residual Dry Matter (RDM), which is the dry grass matter left on the ground in the fall, at the end of the grazing season. RDM is a measure of grazing pressure; moderate levels of RDM are correlated with the health of rangeland ecosystem, soil stability, water retention and biodiversity of native plants and wildlife. Therefore RDM levels are used by The Nature Conservancy and other land managers as a conservation easement compliance measure. I develop a rapid, easy to use, efficient and robust methodology to predict RDM in the fall using spring maximum and annual sum of vegetation index values. My results show that MODIS-based Leaf Area Index (LAI) is the best measure of dry grass biomass. Most importantly, I demonstrate that MODIS data can be efficiently used by range managers and conservationists to estimate RDM easement compliance.

In summary, in this dissertation I develop the use of quantitative spatial tools to measure both vegetation and herbivores on rangelands and to characterize landscapes on large spatial scales. I conduct interdisciplinary research connecting landscape ecology, remote sensing science and wildlife ecology. I demonstrate how freely available MODIS satellite imagery and open source software can be used by conservation managers to understand vegetation patterns and wildlife distribution in relatively easy, cost efficient, rapid and robust ways. The tools I develop in this dissertation identify and quantify change in rangelands. My results inform targeted management and conservation practices, and contribute to improve monitoring and to the understanding of these imperiled ecosystems.

Author summary