Attractor identification and control in Boolean models of plant-pollinator networks
Restricted (Penn State Only)
- Author:
- Fateminasrollahi, Fatemehsadat
- Graduate Program:
- Physics
- Degree:
- Doctor of Philosophy
- Document Type:
- Dissertation
- Date of Defense:
- June 07, 2023
- Committee Members:
- Dezhe Jin, Major Field Member
Reka Albert, Chair & Dissertation Advisor
Jorge Sofo, Major Field Member
Katriona Shea, Outside Unit & Field Member
Nitin Samarth, Program Head/Chair - Keywords:
- Boolean modeling
mutualistic systems
complex systems
plant-pollinator communities
network theory - Abstract:
- Ecological and biological systems consist of numerous interlinked components that interact and exchange information; such interactions give rise to emergent, collective behaviors that are of interest for ecologists and life scientists. The study of the relationship between the interactions and dynamics of individual components and the emergent dynamics of the system is important because it can lead to the development of control methods to manipulate the collective dynamics. In turn, these control methods can be used for ecological community management or restoration, or for therapeutic medical applications. One promising method to gain a deeper insight into such complex systems is to model the interactions among elements using a network and couple it with a predictive dynamical model. The analysis of such dynamical models provides us with a platform to advance our knowledge of the intricate behaviors exhibited by ecological and biological networks, and it has wide-ranging implications across various domains, spanning conservation efforts, the development of community management strategies, and drug target identification in the context of drug design. The innate challenge that arises when analyzing these models is the large size of the system and the non-linearity of the dynamical processes. Recently, a new approach has been developed by Jorge Gómez Tejeda Zañudo and collaborators that focuses on the stable motifs in the network; stable motifs are minimal positive feedback loops that maintain a specific state regardless of the state of the rest of the components in the system. By characterizing the stable motifs and the conditions that lead to their lock-in, this method can identify the system’s dynamic repertoire, predict the outcome of specific interventions and suggest management and control methods. In this dissertation, the main focus is on mutualistic plant-pollinator networks, and specifically on their description by a well-established predictive dynamical model developed by Colin Campbell and collaborators. The study of such systems is of ecological significance as pollinator species face considerable degradation across the world. The loss of pollinator species has a dramatic negative effect on crops as the majority of food crops require pollination to survive. The examination of the reliability and stability of these communities holds great significance for agricultural management and ecological preservation endeavors. There is a great need for measures and methods to predict the magnitude of any cascading effects of species extinction, and for prevention and restoration strategies to maintain the communities. I contribute to this field of study by making it possible for the first time to apply stable motif analysis to plant-pollinator communities. I transform the equations of the existing model by changing threshold functions into suitable logical functions of plant-pollinator networks so that stable motif analysis can be applied to it. I then extend the classical stable motif analysis and introduce a novel method based on stable motifs that determines the stable communities of large plant-pollinator systems efficiently. This method relies on a new concept called the network of functional relationships among stable motifs; I show that these relationships can be leveraged to identify stable communities and accelerate the process significantly. Put into the ecological context, stable motifs can be intuitively interpreted as small groups of species in which species can maintain a specific survival state. I show how such groups of species and the relationships of these groups determine the final community outcomes in plant-pollinator networks. Once the stable communities are characterized, I study their reaction to perturbation and analyze the behavior of the system in the case of species extinction. I extend Boolean modeling concepts, so far only defined for functions of a specific logical form, to the plant-pollinator Boolean threshold functions and introduce a new algorithm to measure the cascading effects caused by species extinction. I then use the information gained from stable motifs to first identify the species whose extinction leads to massive catastrophe in the community and next suggest restoration measures that can be incorporated in ecological sciences. In chapters 1 and 2, I introduce the mutualistic plant-pollinator networks, the Campbell et al. Boolean model of community formation, and the key concepts of Boolean modeling respectively. In chapter 3 I present my contributions to the methodological advancements in the field of Boolean modeling and computational ecology. The methods in this chapter are presented in the context of plant-pollinator networks, but are general and can be implemented in other types of Boolean networks. Chapter 4 describes the properties of the alternative stable states available to the same group of species. Chapter 5 describes the response of plant-pollinator communities to the extinction of a species; specifically, whether there will be cascading effects. This chapter also proposed multiple damage prevention and community restoration measures. The analysis results in these two chapters rely heavily on the concept of stable motifs and the methods introduced in chapter 3. I demonstrate that stable motifs successfully pinpoint the crucial species and this method outperforms the previous well-established measures. Finally, in chapter 6 I study network control in Boolean networks that have a modular structure. In general network control means that by externally fixing the state or the dynamics of a group of nodes, the system as a whole will converge into a desired state or attractor. In this analysis, I aim to identify methods that identify control targets, relying solely on the properties of the network. Taking advantage of the fact that many ecological and biological networks are composed of smaller densely connected modules, I propose a novel module-based method to localize the search for control targets - nodes that if externally controlled, the system will converge into a desired dynamical outcome (e.g., a rich and bio-diverse stable community) or move away from the unwanted dynamical outcome (i.e., full collapse of the community). In this analysis, I study a large ensemble of biologically inspired synthetic Boolean networks to capture the properties of these systems across different levels of modularity. I show that it is considerably more efficient and advantageous to localize the search for control targets in networks with clear modular structure. Chapter 7 presents conclusions and possible future research directions.