Theoretical Ecology Lab Tea

Spring 2019 schedule

Note: Priority is given to graduate students. A ⊛ symbol next to the speaker's name means that approval is pending for a week and graduate students can still claim the slot.

Titles and abstracts

Wed. January 9^th – 12:30pm

Towards a general framework of adaptive processesLuca Rade

Any cohesive system in an unpredictably changing environment has an adaptive process to maintain fitness. There must be a mechanism for processing the change in the environment's fitness function, exploring and selecting possible adaptations, maintaining the integrity of the rest of the system, and preventing over-adaptation to transient change. By examining genetic evolution, the immune system, biological learning, and artificial learning as adaptive processes, I am seeking to formalize the underlying principles of adaptive processes and the axes along which such systems differ. This is a work in progress and the talk is primarily aimed at soliciting sources and ideas based on the initial framework.

Back to schedule

Wed. January 16^th – 12:30pm

Agroecological theory: a complex systems approachTheresa Ong

Back to schedule

Wed. February 6^th – 12:30pm

Infection and reinfection in epidemic models in networksRenato Pagliara Vasquez

We study the SIRI (Susceptible-Infected-Recovered-Infected) epidemic model for reinfection in which the susceptibility of individuals to first-time infections might be different than their susceptibility to secondary infections. The SIRI model generalizes the basic SIS and SIR models and allows for the study of systems in which the susceptibility of agents changes irreversibly after the first exposure to the infection. Using a mean-field approximation, we derive a set of ODE's describing the spreading dynamics on arbitrary networks. We use the resulting network SIRI model to explore the role of network topology and agent heterogeneity in the transient and long-term dynamics at the level of the group. We show the network SIRI model has four different dynamical regimes, including a bistable regime that takes place when secondary infections are more likely than primary infections, and where there exists a critical manifold of initial conditions that separates solutions where the infection dies out and solutions where the infection spreads through the network. We show that when the infection spreads through the network, solutions exhibit a resurgent epidemic in which the probability of infection across the network initially decreases before ramping up towards an endemic equilibrium.

Back to schedule

Wed. February 13^th – 12:30pm

Pattern formation in opinion dynamicsDaniel Cooney

We will discuss the Hegselmann-Krause model for opinion dynamics and review recent work on the clustering of opinions in the presence of noisy decision-making. Emphasis will be placed on the potential bistability of patterned and uniform states and on the "spikiness" of clustered opinions.

Back to schedule

Wed. February 20^th – 12:30pm

Coexisting attractors in the context of cross-scale population dynamics: measles in London as a case studyAlex Becker

Patterns of measles infection in large urban populations have long been considered the paradigm of synchronized nonlinear dynamics. Indeed, recurrent epidemics appear approximately mass-action despite underling heterogeneity. However, using a subset of rich, newly-digitized mortality data (1897-1906), we challenge that proposition. We find that sub regions of London exhibited a mixture of simultaneous annual and biennial dynamics, whilst the aggregate city-level appears firmly annual. Using a simple stochastic epidemic model and maximum likelihood inference methods, we show that we can capture this observed variation in periodicity. We identify strong agreement between theory and data, indicating that both changes in periodicity and phase coupling between regions can follow relatively simple rules. Our analysis underlines that multiple attractors can coexist in a strongly-mixed population, and follow theoretical predictions.

Back to schedule

Tue. February 26^th – 02:00pm

Oceanographic dynamics drive inequality through critical multi-level links in the Humboldt squid fisheryLaura Elsler

Environmental dynamics can amplify inequality and poverty for fishers dependent on marine species use for their livelihoods. Traders, fishers and other groups in the social system are affected by environmental dynamics. In small-scale fisheries, distribution of benefits is often influenced by relationships between fishers and traders. Those relationships are not included in traditional models for fisheries management. A considerable gap remains in understanding the mechanisms through which benefits are distributed by both trade relationships and environmental dynamics and thus affect inequality and poverty in fisheries. Here we find that driven by oceanographic variability and change the links between target species, traders and fishers, have contradictory effects on income inequality in the Mexican Humboldt squid fishery. We demonstrate that development programs based on models ignoring the multi-level between oceanographic dynamics, fishers and traders may exacerbate already severe income inequality between those groups. In particular, during cold oceanographic conditions fishers benefit from high catches but they are trapped in a traders monopoly that sets low prices. In contrast, catch volumes and trader competition decrease during warming oceanographic conditions. Our quantitative analysis shows that in this context, increasing local market prices, as a management strategy to develop the squid fishery, reinforces existing income inequalities. This research provides an example how to integrate multi-level links in a yet simple modeling framework to improve prediction accuracy and guide contextualised management strategies. We anticipate our model to be a starting point to expand existing fishery models with social and trade relationships that interact with variables relevant to fishery management.

Back to schedule

Wed. February 27^th – 12:30pm

Implications of diverging media in polarized societies: A model of information cascades and their effect on social network structureChris Tokita

Cascades are ubiquitous phenomena in biological and social systems. Information cascades occur in social systems when individuals base their behavior on the social information provided by the behavior of others. In essence, these cascades are the rapid spread of a behavior through a group. Common examples include startle cascades in fish schools or the spread of fads in human groups. While cascades are theoretically and experimentally well explored, most approaches ask how the social organization (i.e., social network structure) affects the likelihood and size of cascades. In this talk, I'll present very preliminary work on a cascade model that will ask the opposite: how might cascades reshape social networks? I will frame this model within the context of contemporary politics, where cascades take place among ideologically-polarized individuals reliant on increasingly divergent, partisan media sources. While this project is in very early stages, I hope to gather feedback and suggestions on the model and approach.

Back to schedule

Wed. March 6^th – 12:30pm

"Brave leaders" and collective action problems: Explaining costly cooperation in greater ani groupsChristie Riehl

Collective action problems are ubiquitous in nature. When initiating a collective action is costly, why should a group member pay this cost instead of hanging back and waiting for someone else to take the lead? In this talk, I'll propose a game-theoretic approach to understanding the evolutionary stability of cooperative nesting in the greater ani, a tropical bird. Anis nest in social groups in which several females lay their eggs into a single, shared nest. Social nesting increases individual fitness -- females in large groups have higher reproductive output than females in small groups -- but it also comes with a cost. Each female in the breeding group ejects others' eggs from the shared nest until she lays her own first egg, so early-laying females pay a significant cost to initiating the communal clutch. I suspect that a game-theoretic model known as the brave leader game (a variant of the snowdrift game) might be useful in explaining the evolutionary stability of this behavior in anis, and could potentially yield new insights into the selective pressures shaping group size. This is still in the brainstorming stages; the purpose of this talk is to explain the general assumptions of the model and the characteristics of the ani system, and to get feedback on whether this is a reasonable approach.

Back to schedule

Wed. March 13^th – 12:30pm

Spatial property rights as a management strategy for "leaky" common-pool resources, with an application to fisheriesAlice Lin

Classically, a common-pool resource is inevitably over-extracted because individuals take more than the social optimum. In fisheries management, one way of avoiding this outcome is assigning private property rights, partitioning the space into smaller territories. Ideally, each fisher would behave optimally within their own territory. However, fish are able to move across territories, and the extent to which this occurs provides a continuum between private and public property regimes. Using adaptive dynamics and numerical simulations, we investigate the equilibrium behavior of fishers along this continuum by varying the extent to which the resource leaks across territory lines. Although extraction level is concave as a function of fish mobility, our results demonstrate that privatization can still be effective when fish mobility is low relative to territory size.

Back to schedule

Wed. March 20^th – 12:30pm

Spring break - no Lab Tea

Back to schedule

Wed. March 27^th – 12:30pm

Incentivizing hospital infection control (with prosociality)Sarah Drohan

I will be talking about my recent paper that can be found here. Additionally, as part of my dissertation I will be adding a prosociality element on which I would appreciate feedback. The abstract for the paper is below: "Healthcare-associated infections (HAIs) pose a significant burden to patient safety. Institutions can implement hospital infection control (HIC) measures to reduce the impact of HAIs. Since patients can carry pathogens between institutions, there is an economic incentive for hospitals to free ride on the HIC investments of other facilities. Subsidies for infection control by public health authorities could encourage regional spending on HIC. We develop coupled mathematical models of epidemiology and hospital behavior in a game-theoretic framework to investigate how hospitals may change spending behavior in response to subsidies. We demonstrate that under a limited budget, a dollar-for-dollar matching grant outperforms both a fixed-amount subsidy and a subsidy on uninfected patients in reducing the number of HAIs in a single institution. Additionally, when multiple hospitals serve a community, funding priority should go to the hospital with a lower transmission rate. Overall, subsidies incentivize HIC spending and reduce the overall prevalence of HAIs."

Back to schedule

Wed. April 3^rd – 12:30pm

Contagion dynamics on simplicial complexes: derivation, simulations, inferenceBernat Guillen Pegueroles

I will talk about the second chapter of my thesis which is an extension of some common tools in network analysis to simplicial complexes (with higher order interactions). I will showcase some simulations (with a next reaction type algorithm adapted to simplicial complexes) and models and argue that this way of modeling may be a good link between simple contagion processes and complex contagion processes. I will also show a proposed tool for inference of the simplicial complex structure based on cascade data in a cascading process.

Back to schedule

Wed. April 10^th – 12:30pm

The ecology of information: measuring and modeling life’s hidden currencyAndrew Hein (NOAA/UC Santa Cruz)

Interconnectedness is a defining feature of ecological systems. This feature is often studied under the rubric of food webs, networks of populations that are connected by feeding links. I will discuss a different kind of network that may be equally important for the function of ecosystems: the network of information flow among organisms. I’ll discuss a case study from a coral reef ecosystem, where we have measured the nature of information flow and shown how it affects population processes and nutrient cycling. I will also talk about some theoretical work that seeks to integrate sensory information into coupled population models.

Back to schedule

Tue. April 16^th – 04:00pm

Maximum information entropy: a foundation for ecological theoryJohn Harte (UC Berkeley)

The maximum information entropy procedure (MaxEnt) is both a powerful tool for inferring least-biased probability distributions from limited data and a framework for the construction of complex systems theory. From physics to economics, from forensics to medicine, this powerful inference method has enriched science. The maximum entropy theory of ecology (METE) describes remarkably well the widely observed patterns in the distribution, abundance and energetics of individuals and species in relatively static ecosystems, but fails in rapidly-changing disturbed ecosystems. I describe the static theory and then show preliminary results from an extension of the theory to ecosystems undergoing change in response to disturbance (DynaMETE).

Back to schedule

Wed. April 17^th – 12:30pm

Optimal group size for collective insurancesFernando Santos

The risk of extreme weather events discourages investment in high-return technologies that would enhance smallholder farms productivity. This situation is evident in developing countries, where extreme weather events are more frequent and their effects more pronounced given the prevalence of fragile infrastructures. As a result, low-income farmers often find themselves in so-called poverty traps. Insurances are a possible solution for this conundrum. Nonetheless, different type of insurance products have specific caveats and benefits that influence their attractiveness and consequence take-up rates: indemnity-based insurances require experts in the field and are sensitive to information asymmetries and moral hazard, which ultimately results in pricey premiums; index-based insurances, whereby compensations are paid based on objective weather indexes, suffer from basis risk, i.e., the risk that a loss is suffered and no compensation is paid given that the contracted index was not reached. Collective Index Insurances (CII) appear as a promising alternative. These are index-based insurances contracted by groups, with lower transaction costs and with the potential to alleviate basis risk through informal transfers within the collectives. I will present a new (work-in-progress) model where CIIs are studied from the perspective of evolutionary game theory. Individuals decide whether to take part in groups that subscribe collective insurances. Their decision is impacted by, e.g., insurance premium, basis risk, group compositions (in terms of individuals with different risk exposure) or risk-aversion levels. We show that, under different scenarios, there exists an optimal group size that favor the take-up rates of collective index insurances.

Back to schedule

Wed. April 24^th – 12:30pm

Shape analysis: an application to conservation predictionBernat Guillen Pegueroles

In this talk (corresponding to the 3rd chapter of my thesis) I will present some metrics, adapted from Topological Data Analysis, that could be helpful for evaluating the impact of the shape of a domain in the solutions to Competition-Diffusion Equations. I will show how these metrics are computed, why they may be relevant, and some preliminary positive results for a biodiversity predictor based on domain shape.

Back to schedule

Thu. April 25^th – 04:30pm

Evolutionary rescue is determined by differential selection on demographic rates and density dependenceAnna Vinton (Yale/International Institute for Applied Systems Analysis (IIASA))

Current rates of climate change and habitat loss are expected to either lead to populations adapting and persisting, or submitting to extinction. For many species ecological models predict accelerating extinction risk due to climate change. Traditionally ecological models make extinction predictions based on how environmental change alters intrinsic growth rate (r). However, these models often ignore the potential for populations to undergo evolutionary rescue, or to avoid climate-induced extinction via adaptive evolution. Moreover, in natural populations the environment may impose selective pressure on specific demographic rates (such as birth and death rates) rather than directly on r (the difference between the birth and death rates). Therefore, when we consider the potential for evolutionary rescue, populations with the same r can have very different abilities to persist amidst a changing environment. Theoretical studies often do not differentiate between the major demographic rates that are selected upon, or in the way selection is acting on density dependence. I will discuss how this changes the propensity for population persistence via evolution using stochastic birth-death models. Furthermore I will introduce my work investigating how spatial complexity comes into play.

Back to schedule

Wed. May 1^st – 12:30pm

How range residency changes encounter ratesRicardo Martinez-Garcia

Pairwise encounter rates link individual movement strategies to intra- and inter-specific interactions, and therefore represent a key nexus between animal movement and population and community processes. Despite this paramount importance, most encounter models assume that individuals perform ballistic or Brownian motion and occupy the environment uniformly. Mounting empirical evidence suggests that these assumptions are violated for most species. Instead, individual movement shows home ranging behavior with two key features: 1) they inhabit a subset of the population range and 2) space use within the home range is non-uniform. This ubiquity of range residency indicates a need to incorporate it into encounter models. I will discuss a first step in that direction, presenting analytical expressions for pairwise encounter rates based on three different movement models (Ornstein-Uhlenbeck, Reflected Brownian Motion and Brownian Motion) that incorporate neither, one, or both main features of home ranging. I will conclude by discussing how these results extrapolate to larger population sizes and possible extensions to construct a more realistic encounter theory grounded on empirically supported movement assumptions.

Back to schedule

Wed. May 8^th – 12:30pm

Daniel Cooney

Back to schedule

Thu. May 23^rd – 12:30pm

Evolutionary dynamics of multi-agent learningDaan Bloembergen (Centrum Wiskunde & Informatica)

In this talk I will discuss the link between classical reinforcement learning algorithms and the replicator dynamics from evolutionary game theory. This link has been utilized in the past decade(s) to better study and understand the complex dynamics that arise in systems in which multiple autonomous agents simultaneously learn to optimize their behavior through experience. Starting from the seminal work by Börgers & Sarin in 1997, who first formalized this link, I will provide an overview of work done in the area. In addition I will highlight some applications, such as the comparative analysis of learning algorithms, and parameter tuning. The talk will be based on my 2015 survey paper on this topic: "Bloembergen, Tuyls, Hennes & Kaisers (2015). Evolutionary dynamics of multi-agent learning: A survey. Journal of Artificial Intelligence Research, 53, 659-697".

Back to schedule

Links to previous schedules

1. Fall 2000
2. Spring 2001
3. Fall 2001
4. Spring 2002
5. Fall 2002
6. Spring 2003
7. Fall 2003
8. Spring 2004
9. Fall 2004
10. Spring 2005
11. Fall 2005
12. Spring 2007
13. Fall 2007
14. Spring 2008
15. Fall 2008
16. Spring 2009
17. Fall 2009
18. Spring 2010
19. Fall 2010
20. Spring 2011
21. Fall 2011
22. Spring 2012
23. Fall 2012
24. Spring 2013
25. Fall 2013
26. Spring 2014
27. Fall 2014
28. Spring 2015
29. Fall 2015
30. Spring 2016
31. Fall 2016
32. Spring 2017
33. Fall 2017
34. Spring 2018
35. Fall 2018