Browsing by Subject "hierarchical model"

Cell division is a biological process fundamental to all life. One aspect of the process that is still under investigation is whether or not cells in a lineage are correlated in their cell-cycle progression. Data on cell-cycle progression is typically acquired either in lineages of single cells or in synchronized cell populations, and each source of data offers complementary information on cell division. To formally assess dependence in cell-cycle progression, I develop a hierarchical statistical model of single-cell measurements and extend a previously proposed model of population cell division in the budding yeast, Saccharomyces cerevisiae. Both models capture correlation and cell-to-cell heterogeneity in cell-cycle progression, and parameter inference is carried out in a fully Bayesian manner. The single-cell model is fit to three published time-lapse microscopy datasets and the population-based model is fit to simulated data for which the true model is known. Based on posterior inferences and formal model comparisons, the single-cell analysis demonstrates that budding yeast mother and daughter cells do not appear to correlate in their cell-cycle progression in two of the three experimental settings. In contrast, mother cells grown in a less preferred sugar source, glycerol/ethanol, did correlate in their rate of cell division in two successive cell cycles. Population model fitting to simulated data suggested that, under typical synchrony experimental conditions, population-based measurements of the cell-cycle were not informative for correlation in cell-cycle progression or heterogeneity in daughter-specific G1 phase progression.

By concentrating materials and increasing the speed with which rainfall is conveyed off of the landscape, nearly all forms of land use change lead to predictable shifts in the hydrologic, thermal, and chemical regimes of receiving waters that can lead to the local extirpation of sensitive aquatic biota. In Central Appalachian river networks, alkaline mine drainage (AlkMD) derived from mountaintop removal mining for coal (MTM) noticeably simplifies macroinvertebrate communities. In this dissertation, I have used this distinct chemical regime shift as a platform to move beyond current understanding of chemical pollution in river networks. In Chapter Two, I applied a new model, the Hierarchical Diversity Decision Framework (HiDDeF) to a macroinvertebrate dataset along a gradient of AlkMD. By using this new modeling tool, I showed that current AlkMD water quality standards allow one-quarter of regional macroinvertebrates to decline to half of their maximum abundances. In Chapter Three, I conducted a field study in the Mud River, WV to understand how AlkMD influences patterns in aquatic insect production. This work revealed roughly 3-fold declines in annual production of sensitive taxa throughout the year in reaches affected by AlkMD. These declines were more severe during summer base flow when pollutant concentrations were higher, thereby preventing sensitive organisms from completing their life cycles. Finally, in Chapter Four I described the idea of chemical fragmentation in river networks by performing a geospatial analysis of chemical pollution in Central Appalachia. In this work I showed that the ~30% of headwaters that remain after MTM intensification over the last four decades support ~10% of macroinvertebrates not found in mined reaches. Collectively my work moves beyond the simple tools used to understand the static, local consequences of chemical pollution in freshwater ecosystems.