Maizuru Fisheries Research Station, Kyoto University

My research interest lies in the intersection of two distinct fields of science: ecology and statistics. I use and develop novel statistical models to address a broad range of topics in ecology, including population and community dynamics, abundance estimation, species distribution modeling, evolution of biodiversity, and conservation. Notably, I often prefer adopting statistical models with latent variables (i.e., hierarchical models) which account for nuisance observation processes to obtain better inference about underlying ecological processes.

Population dynamics of marine invertebrates

I study patterns and processes of population dynamics by analyzing population trajectories obtained from long-term field observations. This approach enables us to understand how biological and environmental factors interact to influence population growth, how these effects vary over space and time, and how they result in patterns and properties of population trajectories.

Key words: Density dependence, Population regulation, Population synchrony, Taylor’s power law, Spatial scale, State-space models, Intertidal rocky shore, Barnacles (Chthamalus challengeri / C. dalli), Marine copepod (Tigriopus japonicus).

Inference of community dynamics subject to observation errors

Dynamics of ecological communities, especially of sessile organisms, are often summarized by transition probabilities that characterize the frequency of transitions among discrete ecological states. Accurate estimation of transition probabilities is difficult, however, when a specific type of observation error occurs. I address this issue by developing some novel classes of hierarchical models that explicitly incorporate both ecological and observation processes.

Key words: Markov community models, Multistate dynamic site occupancy models, Classification error, Kernel smoothing, Spatial autocorrelation, Intertidal rocky shore.

Inference of ecological populations and communities based on environmental DNA

Molecular analysis of environmental DNA is an emerging methodology that improves our ability to infer occurrence of species. I study how we can reliably use this approach to infer ecological states and dynamics, especially by developing new statistical methods to estimate population abundance and species diversity based on measurements of environmental DNA.

Key words: Tracer models, Quantitative PCR, Quantitative echo sounder, Japanese jack mackerel (Trachurus japonicus), Maizuru Bay, Metabarcoding, MiFish primer, Multispecies site occupancy models, False-negative detection error.

Macroscale species abundance of woody plants

Although species abundances are one of the fundamental properties of biodiversity, we still lack a clear understanding of them at large spatial scales. I use a joint modeling approach to unveil geographical patterns of species abundances of woody plants that provide important perspectives on ecology, evolution, and conservation of biodiversity.

Key words: Vegetation survey, Woody plant communities, Ecological big data, The unified neutral theory of biodiversity and biogeography, Speciation rates, Red listing, Data integration, Hierarchical modeling.

Sequential Bayesian prediction of menstrual cycles

I developed a statistical framework that provides a sequentially updated prediction for the latent phase of menstrual cycles and upcoming menstruation based on basal body temperature time series.

Key words: State-space models, Sequential Bayesian filtering and prediction, Menstrual cycle length, Periodic phenomena.