As a theoretical and applied ecologist, my research investigates how global change, particularly habitat disturbance and climate change, drives changes in biodiversity in complex landscapes. I approach this question from the perspective of spatial ecology, mixing studies of ecological pattern and process, across multiple spatial and temporal scales, with the ultimate goal of understanding and predicting the spatial structure of biodiversity at landscape to regional scales in complex, human-altered ecosystems.

In service of this broad goal, my research program falls into three main areas. First, I develop new spatial scaling theory to translate biodiversity data up, down, and across space. Second, I use this scaling theory and other quantitative methods to predict species extinctions risks and rates due to habitat loss and climate-driven range contraction. Third, I collect high-resolution, large scale data to study bat and bird populations along disturbance gradients, with a particular focus on developing new computational tools and workflows for analyzing large data sets. This work relies on a synthesis of tools from theoretical and mathematical ecology with computational techniques from the emerging practice of data science.

Scaling biodiversity data up, down, and across space

A core goal of my research is to develop mathematical theory and models that predict the relationships between species abundance and diversity across spatial scales. Such theory can be used to "upscale" plot-scale data to estimate species richness at large scales, to "downscale" regional diversity to estimate species loss following habitat loss and climate change, and to predict turnover in species composition, or beta diversity, across space.

Focusing on scaling biodiversity data up and down, I have recently combined global range maps, small plot census data, and body size scaling rules to infer the likely landscape scale biodiversity impacts caused by the construction of three hydroelectric dams in northern Borneo (Kitzes and Shirley 2016). Using a new theoretical form of the species-area relationship, I have also contributed to efforts to upscale small plot data to estimate regional tree and arthropod diversity in Panama (Harte and Kitzes 2015). Recent work with a graduate student has explored extended macroecological frameworks for biodiversity scaling (Wilber et al. 2015)

Focusing on scaling across space, I have several ongoing projects investigating the theoretical basis of species turnover and beta diversity. After contributing to empirical work documenting the failure of a previous theory to predict these patterns (McGlinn et al. 2015), I developed a mathematical approach that uses data from small plot censuses to infer the shape of a species' pair correlation function, a spatial measure used frequently in cosmology (Kitzes and Harte in submission). I have also developed an extension of a well-known stochastic process model that predicts the relationship between plot area and species aggregation (Kitzes in prep).

In support of this work, I created and continue to lead the development of the open source Python package macroeco, which supports the evaluation and prediction of diversity patterns in ecological communities (Kitzes and Wilber 2016). In addition to low-level Python routines, this package was developed with external command line and graphical interfaces to support its use by researchers who are not programmers

Predicting species extinction following global change

Among biodiversity impacts, species extinction is particularly significant from both an ecological and a policy perspective. Much of my work has focused on exploring the use of spatial theory and models for predicting extinction rates due to habitat loss and climate-driven range contraction.

The species-area relationship (SAR) is a widely used, general model for predicting species extinction risks and rates. After identifying several shortcomings of applying classic species-area relationships to extinction prediction (Harte and Kitzes 2012), I developed extended SAR frameworks that provide probabilistic extinction estimates based on an explicit minimum viable population size (Kitzes and Harte 2013) and that distinguish between immediate and time-delayed extinctions, known as extinction debt (Kitzes and Harte 2015). I have also explored the interactions of statistical SAR theories with taxonomic boundaries (Harte et al. 2013).

Investigating species populations in fragmented habitats, I have used stochastic metapopulation models parameterized from body size scaling relationships to uncover basic principles for reserve network design (Kitzes and Merenlender 2013). I have also assisted in developing a new approach to biodiversity conservation in patchy, production forest landscapes that is based on the maintenance of temporary, rotating refugia (Ramage et al. 2013a, 2013b). Focusing on global-scale conservation outcomes, I contributed to a widely cited perspective on global tipping points (Barnosky et al. 2012) and served as a lead author for the biodiversity chapter of the recent UNEP Global Environmental Outlook (GEO-5) report.

Understanding changes in biodiversity along disturbance gradients

Aside from the amount of available habitat and its spatial configuration, the quality of habitat in complex human-disturbed landscapes can vary greatly. My field research involves surveying biodiversity along disturbance gradients using autonomous acoustic sensors, which are able to efficiently and rapidly gather diversity and activity data at large spatial and temporal scales.

I have used acoustic recording devices to study bat and bird populations in urban and agricultural landscapes in northern California. In a two-year study, I demonstrated that bat activity levels are depressed by 50\% near large highways and that this effect is consistent across species (Kitzes and Merenlender 2014). A second study in vineyard landscapes found that site-scale remnant vegetation significantly increased local bat activity, while surrounding landscape features had negligible local effects (Kelly et al. 2016). Bird call data from these studies is currently being analyzed by graduate students at the University of California, Berkeley.

Many researchers who use acoustic sensors analyze the resulting data by hand, a time consuming process that limits the scope of sensor-based field studies. To support the expansion of this acoustic research, I developed and released an open source software package, BatID, that quickly and automatically identifies recorded bat calls to the species level.

Other interdisciplinary research

In addition to the three core areas described above, I have also conducted research on global models of sustainable resource consumption and on frameworks for reproducible research.

With a team of ten researchers, I recently used spatial data on land cover and human appropriation of net primary productivity along with a global input-output analysis (Kitzes 2014) to create the first spatially-explicit global biodiversity model that links specific economics activities to species population declines (Kitzes et al in press). This work builds on my long history of global-scale modeling in the context of sustainability science and human ecology, including extensive work on "ecological footprint" accounts (Kitzes et al. 2008, 2009a, 2009b, 2009c) and contributions to global analyses of nitrogen footprints (Leach et al. 2012).

At the Berkeley Institute for Data Science, I am involved in several projects that aim to improve reproducibility in computational research. I am the lead editor of a book, titled The Practice of Reproducible Research, that will be the first major work to comprehensively document and teach the core, cross-disciplinary practices of reproducible data-intensive research.