Modeling Ant Feeding Interactions: With this research, in collaboration with the David Hughes Lab, we propose a novel penalized hidden Markov model that penalizes rapid state switching through a Bayesian ridge-like prior on continuous-time Markov chain transition rates. We show that this smooths the rate of estimated behavioral state switching and results in better 1-step ahead prediction than the non-penalized HMM. We demonstrate the utility of this penalized approach to stochastic process modeling on ant trophallaxis (feeding interaction) data. Pre-print available soon.
Cont-LIMNOS: LAGOS is a multi-scaled database system and a set of tools to study lake water quality at macroscales. Within the LAGOS umbrella, researchers have created and shared: versioned water quality databases, software and code for analysis of geographic and tabular data, procedures and policies for team science, and research publications. My contribution to this collaborative effort is to explore an extension of extrapolation cut-offs to a generalized, multivariate setting.
Thesis: At New Mexico State University (NMSU), I worked with the National Alliance for Advancing Biofuels and Bioproducts, a consortium of universities and research institutions aimed at commercializing algae into biofuel, to directly impact the field by advancing cultivation methods. My contribution was threefold: study environmental conditions that enhanced growth and lipid accumulation of our study species, observe the aquatic communities of production systems, and make processes involved more efficient. My work included utilizing mesocosm experiments and mathematical modeling to understand how to optimize parameters for maximized growth and lipid accumulation.
Honor’s Capstone Thesis: Influenced by a colleague’s network analysis project I began exploring the role of detritus in an ecosystem. Other work suggested that detritus had a vital role in the stability of a food web; however, more recently published food webs in the literature failed to include this link. My project modified 12 previously published food webs to include a realistic link back to detritus from every species. Representing the webs with binary adjacency matrices, and using matrix algebra applications I found the abundance of indirect links through each pair of nodes at differing length of indirect pathways. I hypothesized that when links back to detritus are included in preexisting food webs, the abundance of indirect links would increase. All 12 food webs exhibited an increase in indirect pathways when modified to include links returning energy towards detritus. Increase in indirect links with detritus has similarities to other modifications; however, with detritus there is a full six orders of magnitude increased despite same level of connectance modified.
Upper Green River Biopreserve: My job with Dr. Albert Meier’s group was my introduction to being a scientist and gave me exposure to a broad array of projects and techniques in restoration and conservation. Projects underway at the Preserve included biological surveys of fish, amphibians, reptiles, birds, mammals, mussels, and other endangered species. Other projects, include restoration of riparian corridors, bottomland hardwood forest, native grasslands and barrens, and control of stream bank erosion. I served as a team member under graduate students and professors in a large effort to complete restoration projects that would increase the health of the ecosystem.
Chicago Botanic Gardens NSF-REU: My internship through the Research Experience for Undergraduates program at the Chicago Botanic Gardens gave me experience beyond the scope of my coursework and an opportunity to begin an independent project. My work explored the use of arachnids as an indicator species to determine the health of an area of oak woodlands. As a continuation of a 2005 study, we hoped to verify the observed increase in diversity over time in restored areas of the forest. We hypothesized that the restored plots would exhibit a more diverse spider population than the unmanaged control plots. We also predicted a correlation between diversity and age of restoration. My analysis included calculating species diversity, richness, and evenness and comparing each between years. This experience provided me with an early look at multiple aspects of conducting an independent project, and was a vital facet of a developing education in science. I concluded that an exotic species of spider (Ozyptila praticola) I observed primarily in the un-restored (Buckthorn) site suggested that it would be an effective indicator species for disturbed, unhealthy habitats.