My primary interests are in the development and application of quantitative methods to confront ecological
data to answer questions about how ecosystems change across space and time, and to quantify the uncertainty
of these changes. Most of my work is focused on understanding change in forest ecosystems, including
changes in structure, community, and productivity, and learning how these changes relate to abiotic factors
like climate. Datasets that inform us about these changes vary in spatial and temporal resolution and scale,
and developing coherent models that account for these differences in scale is a particularly exciting aspect
of my research. The quantitative tools I use most often include differential equations, statistics – especially
Bayesian methods, and scientific computing.
Mapping forest change using mathematics and data
Assumptions about past forest change play an important role in ecosystem models used to predict future
interactions among earth systems in a changing climate. Quantifying past forest change and its uncertainty
will help us better constrain and validate these models. As a part of the Paleoecology Observatory Network, I develop space-time models to infer past forest composition using a network of fossil pollen records and historical public land survey forest data. We calibrate the model using settlement-era fossil pollen counts and public land survey data. Then, we use calibration results in our prediction model to go back further in time. Check out our QSR paper on the pollen-vegetation calibration model here.
Reconstructing Forest Net Primary Productivity
Forests play an important role in the terrestrial carbon cycle. At a global scale, forests act as a carbon sink,
mitigating the effects of climate and land-use change by approximately 30%. However, changes in forest
carbon allocation in response to factors such as climate is not well understood. I am currently working in collaboration with others to develop a model to estimate past forest aboveground net primary productivity (aNPP) by quantifying aboveground biomass increment using tree-ring data.
White Spruce mortality using retrospective data
This research project was motivated by the forest industry need to obtain timely estimates of spruce mortality in the mixedwood boreal forest. Retrospective sampling looks at tree-rings to determine the history of the forest. Tree ring formation is influence by climate, and trees that grow together experience the same climatic conditions. This allows us to identify ‘marker years’, which are especially good or bad years, enabling us to determine the year that a tree died. By doing this for all the dead trees in your sample area you can then reconstruct the stand history. What makes this project really interesting is that we used our data collected from over forty locations throughout Alberta to relate local competiion Trembling Aspen and White Spruce to mortality.
Mixedwood Boreal forest species coexistence
The aim of this project is to develop a model that allows us to predict and better understand the mixedwood Boreal population dynamics. To do this, we use a coupled integral projection model that incorporates demographic components which capture the survival, growth, and fecundity of both of Aspen and Spruce, and incorporates individual interactions within and between species. Analysis of the model allows us to better understand coexistence, and to hypothesize about the different local niches occupied by each species.
Efficient integration schemes for the integral projection model
After encountering computational difficulties as a result of the structure of one of the models I study (the IPM), I was motivated to better understand why my chosen quadrature routine was failing. As it turns out, I am not the first person to run into computational problems with the IPM, but recognized the need for a thorough analysis of the efficiency of quadrature schemes for this type of models. The manuscript describing this work evaluates the ability of point-based and cell-based integration schemes for IPM which often exhibit a variety of problematic conditions (such as discontinuities), and makes recommendations on the appropriateness of each method. This work is a collaborative effort with Matthew Emmett from the Lawrence Berkeley Lab.
Climate reconstruction using tree-ring proxies
This project was initiated during the North American Dendroecological Fieldweek, with the goal of answering the question: can closed-canopy trees be used as proxies to build climate reconstructions? Typically when interested in using tree-rings as proxies for climate, scientists look for trees living in extreme environments without neighbours to ensure that the chosen individuals are climate limited. When trees grow closely together, their interactions that are a result of competition for resources affect the growth signal. But there is hope – if you can successfully separate the growth signal into a climate-driven component and a competition component, you may be able to use this climate-driven component as a proxy record. In this study we show that for Quercus prinus on the east coast of the US, this is possible, and we use these proxy records to reconstruct precipitation for a period beginning 150 years prior to the first date of available weather data for that region. This work is a collaborative effort led by Valerie Trouet at the LTRR at the University of Arizona.
PyDendro: a new software package that simplifies tree-ring data processing
There are some really great dendrochronology software packages available, but many of them lack flexibility because they are closed-source and can therefore not be easily customized, and it can be difficult to understand the mathematics behind the algorithm. The goal of PyDendro is to provide an open-source tool for dendrochronologists that allows them to better understand and have more control over the data processing routine. This software package is still in the development phase, and is a collaboration with Matthew Emmett from the Lawrence Berkeley Laboratory.