Matthew T. Farr
Zipkin Quantitative Ecology Lab
Department of Integrative Biology
Ecology, Evolutionary Biology, and Behavior Program
Michigan State University
The application of quantitative methods in ecology and conservation is the principal driver for my dissertation research. I develop hierarchical models to parse out the complexities of ecological systems into processes that can be described using multi-level statistical and mathematical models. I utilize the flexibility of a Bayesian statistical framework and rigorous computer programming to implement hierarchical models. The estimates from these models inform wildlife management, and the model development provides a quantitative framework for future ecological and conservation research.
Below is a list of my current projects. Please go to my GitHub profile for available code. To see other projects in the Zipkin Lab please go to the Zipkin Lab Code Archive or directly to the Zipkin Lab’s repositories on GitHub.
Multi-species modeling reveals variable responses of African carnivores to management alternatives
I am working with David Green and the Holekamp lab to quantify the impact of anthropogenic disturbance and management alternatives on the carnivore community within the Masai Mara National Reserve (MMNR), Kenya. Carnivore communities in the Serengeti-Mara ecosystem, including the MMNR, are among the most diverse in the world, but human-wildlife conflict threatens the continuation of this community. To understand the impact of anthropogenic disturbance on the carnivore community, we compare two disparate management regions within the MMNR that are managed by separate entities. The Mara Triangle experiences minimal disturbance while the Talek region contains high frequency of human-wildlife conflict. Using a hierarchical multi-species distance sampling model we estimate the community wide and species-specific effects of the Talek region on carnivore abundance and compared species’ abundances and group sizes between regions.
Combining multiple observation processes using an integrated species distribution model
I am developing an integrated species distribution model that combines presence only data from opportunistic surveying with distance sampling. Species distribution modeling (SDM) uses ecological covariates and occurrence data to predict a species’ distribution. However, SDMs often return biased estimates due to imperfect detection and sampling bias, but model estimates can be improved by integrating more structured sampling into the model. I am working with David Green and the Holekamp lab to demonstrate the use of ISDMs by combining presence only data and distance sampling to evaluate human disturbance and apex predator decline on black-backed jackal distribution.
Disentangling data discrepancies and deficiencies with integrated population models
The Zipkin lab is collaborating with Todd Arnold to use an integrated population model (IPM) to estimate population size and vital rates of the American woodcock (AMWO) using harvest data, band recovery data, and Singing Ground Surveys. We use an IPM framework to integrates multiple datatypes from AMWO and analyze them in conjunction. The results from our IPM reconcile discrepancies in inference from past studies that analyzed these datatypes independently.
Orangutan population dispersion and habitat shifts
in a heterogeneous natural landscape
Department of Integrative Biology, Michigan State University
Graduate Teaching Assistant – IBIO 365 Biology of Mammals Laboratory, 2017
Biological Science Program, Michigan State University
Graduate Teaching Assistant – BS 171 Molecular and Cell Biology Laboratory, 2016 - 2017
Department of Forestry and Natural Resources, Purdue University
Undergraduate Teaching Assistant – FNR 210 Natural Resource Information Management (ArcGIS), 2014