Theoretical Ecology Lab Tea

The Theoretical Ecology Lab Teas are designed to be informal meetings for members of the research groups of Simon Levin, Steve Pacala, Henry Horn, and Andy Dobson to give talks on their current research and receive feedback from their audience. The talks are usually 30 minutes, including the question and answer sessions, scheduled on Wednesdays at 1:00 PM. Additionally, other members of the Princeton University community and visitors are welcome to attend and to give presentations.

Talk schedules and email lists are maintained by Adi Livnat and Katie Hampson. Please contact alivnat@princeton.edu or khampson@princeton.edu to have your name added to the labtea email list so that you can receive reminders about upcoming lab teas.

To view previous schedules and summaries, go to:
    Fall 2002     Spring 2003
    Fall 2003     Spring 2004
    Fall 2004     Spring 2005
 


 

Fall 2005
 
 
Wednesday, September 14, at 1:30 PM
Josh Weitz
Wednesday, September 21, at 1:00 PM
Parviez Hosseini
Wednesday, September 28, at 1:00 PM
Sam Flaxman - Cornell University
Wednesday, October 5, at 1:00 PM
Akiko Satake
Wednesday, October 12, at 1:00 PM
Dunia Lopez-Pintado - Columbia University
Wednesday, October 19, at 1:00 PM
Jeremy Lichstein
Wednesday, October 26, at 1:00 PM
Drew Purves
Wednesday, November 2, at 1:00 PM
(Fall break)
Wednesday, November 9, at 1:00 PM
Juliet Pulliam
Wednesday, November 16, at 1:00 PM
Sergey Kryazhimskiy
Wednesday, November 23, at 1:00 PM
(Thanksgiving break)
Wednesday, November 30, at 1:00 PM
Heather Leslie
Wednesday, December 7, at 1:00 PM
Marissa Baskett
Wednesday, December 14, at 1:00 PM
Yiqi Luo


Titles and abstracts
(posted approximately one week before the talk):


Wednesday, November 16 @ 1:00 PM

Sergey Kryazhimskiy

Patterns in the evolution of flu

There is a lot of excitement now around a potential pandemic of a new flu strain. It may seem that the currently circulating strains are not of concern anymore. Unfortunatly, this is not the case. My talk is about the evolution (also refered to as the 'antigenic drift') of the H3N2 and H1N1 flu subtypes that are still around in the human population. I will explain how I am trying to study the drift using clustering techniques, what potential benefits and pitfalls this approach has and what rather puzzling observations it allows to make.

[back to schedule]


Wednesday, November 9 @ 1:00 PM

Juliet Pulliam

Dynamical consequences of repeated pathogen introduction and implications for control

Host-pathogen systems are well understood from a theoretical perspective, and the phenomenon of a rapidly-spreading pathogen 'burning itself out' by quickly depleting the susceptible population after its initial introduction is well known. Reintroduction of the pathogen into a population following an initial epidemic will produce dynamics distinct from the initial introduction into a na´ve population, and I will use a simple SIR model to show that repeated introductions can be considered a special case of a well-understood type of dynamical behavior. We will see that reintroduction of a pathogen into a partially immune population produces epidemics with smaller peaks but extended duration, which may facilitate spread to surrounding populations. I will explore the consequences of this change in dynamics between initial and subsequent introductions by showing that this change may have been pivotal in the spread of Nipah virus in peninsular Malaysia in 1999. Finally, I will discuss the implications of these dynamics for designing Nipah virus prevention and control strategies.

[back to schedule]


Wednesday, October 26 @ 1:00 PM

Drew Purves

Crown shape and competition for canopy space

The structure of the forest canopy is crucial to the dynamics and functioning of forests, both because it determines the understory light environment, which in turn determines population dynamics; and because it determines the growth rates of canopy trees, which in turn determines carbon fluxes and storage. Current models of forest canopies are inadequate, because they do not capture the most fundamental aspect of tree growth, namely that trees are plastic in their growth and so compete for, and fill, all the available space in the canopy. A couple of years ago I began to try to provide a model formulation that would capture this plastic growth. Ever since, this apparently esoteric and highly applied project has continued to suggest much deeper and broader implications, largely because of the work of others in the Pacala lab (including Steve himself). I'll try to explain what it is about this modeling effort that I've found so exciting: the fact that the model formulation was worked out by foresters 40 years ago, but that none of them thought to tell the broader ecological community; that it is mathematically identical to the latest and greatest formulation for below-ground competition arrived at by plant ecologists only recently; that (counter-intuitively) its greater biological realism makes it easier to implement and possible analyze mathematically; that it leads it to a ridiculously simple mathematical technique to predict the long-term dynamics of forest IBMs without modeling any individuals; and the fun (and heartache) I've had parameterizing the model for every species in the US, via the powerful (but occasionally troublesome) hierarchical Bayesian methods that we heard about from Jeremy last week. It will be interesting to see whether any of you come to share any of this enthusiasm, or whether I just have an unhealthy obsession with virtual trees.

[back to schedule]


Wednesday, October 19 @ 1:00 PM

Jeremy Lichstein

Hierarchical Bayesian analysis of tree allometries

Hierarchical statistical models allow us to estimate parameters for entities (e.g., species) for which we have little data, or even no data at all, if there is a known hierarchical (e.g., phylogenetic) relationship among the entities. If closely related species for which we do have lots of data have similar parameter values, then we might expect the same to be true for the data-poor species. Hierarchical modeling provides a framework for quantifying the preceding sentence. I will present analyses of height and leaf mass as a function of diameter for all 339 tree species recognized by the US Forest Service inventory program. Tree height is relatively easy to measure, there are lots of data for lots of species, and the analysis works quite nicely. In contrast, most forest researchers are not keen on blowing an entire day just to weigh the leaves of a single tree, and those who do attempt this foolish enterprise are not very good at it. So, the data are few and crappy, and the analysis is a mess. This is where the Bayesian part comes in. If you choose the right Bayesian priors, you can get some very sweet looking results, which admittedly is troubling. I am hoping to conclude with an open discussion about the pros and cons of Bayesian analysis, but this will require that you hang on my every word as I try to explain what I did and why.

[back to schedule]


Wednesday, October 12 @ 1:00 PM

Dunia Lopez-Pintado

Diffusion in Complex Social Networks

This paper studies the problem of spreading a product (an idea, cultural fad or technology) among agents in a social network. An agent obtains the product with a probability that depends on the spreading rate of the product as well as on the behavior of the agent's neighbors. This paper shows, using a mean field approach, that there exists a threshold for the spreading rate that determines whether the product spreads and becomes persistent or it does not spread and vanishes. This threshold depends crucially on the connectivity distribution of the network and on the mechanism of diffusion.

[back to schedule]


Wednesday, October 5 @ 1:00 PM

Akiko Satake

An agent-based model for deforestation

We develop an agent-based model for deforestation to explore how agents make decisions associated with deforestation and how these decisions come to influence macro patterns of land-use over time. We assume that a forest is composed of many land parcels arranged in a regular square lattice in a one-dimensional space. Each parcel is in either a forested or a deforested state. When utilities of forested and deforested states are given, landowners make a decision to increase the net present value of their land. The net present value of the land is the weighting average of the current utility and the utility to be received in a future. By analyzing equilibrium patterns, we show that when landowners largely discount the future utilities, the deforestation rate is very high and an entire forest becomes to be deforested. We give the exact condition representing when deforestation occurs, and show that the condition is independent of the landowner's expectation on deforestation rate occurring in a future.

[back to schedule]


Wednesday, Septmeber 28 @ 1:00 PM

Sam Flaxman


Ideal free distribution models are commonly employed by both theoreticians and empiricists to predict distributions of animals in patchy environments. These models have undergone nearly constant development and extension since their original formulation over three decades ago, with the result that there are now myriad variations whose predictions have not been rigorously tested. I will show how a payoff structure that is common to many ideal free distribution models causes such models to make predictions that are paradoxical (and erroneous) for certain empirical situations. These predictions arise because many of the models lack sufficient variables to encapsulate natural kinds of variation in patches. A simple revision of the common payoff structure eliminates the erroneous predictions, and I provide empirical data that strongly support the proposed revision.

[back to schedule]


Wednesday, Septmeber 21 @ 1:00 PM

Parviez Hosseini

The potential role of disease in grassland community structure in California

Our project seeks to develop a general, dynamic mathematical theory of vector-transmitted pathogen spread in multi-host communities that explicitly incorporates the abiotic context. Using barley yellow dwarf virus (BYDV), we will build a theory based on a bottom up understanding of how anthropogenic environmental change might affect pathogen dynamics and their host communities, complementing much current research in human and animal systems. We will independently both parameterize and test mathematical models on this pathogen system. Preliminary model predictions suggest that BYDV may currently be a key species affecting the community dynamics of California grasslands. Thus we plan on elucidating how perennial and annual grasses are differentially effect, and are affected by, pathogen dynamics. We will also experimentally examine several sets of sub-communities in the field that closely match communities of general theoretical interest, and also replicate our tests across multiple phylogenetic lineages. By using these multiple lineages, we can better understand how these life history traits affect pathogen dynamics and community structure. By using these sets of sub-communities, we will be able to test how well model predictions and experimental results match, strengthening the basis for using mathematical model to understand how pathogens affect communities. We will also examine how vector preference for different hosts, and dependence on different hosts, may vary seasonally and effect pathogen dynamics, and thus community structure.

[back to schedule]


Wednesday, Septmeber 14 @ 1:30 PM

Josh Weitz

Size and scaling of pedator-prey dynamics

[back to schedule]



Last updated 01/05/05
pulliam@princeton.edu