Systems Ecology @ MIT

Understanding change in ecological systems under environmental uncertainty

Our research program

Systems ecology has been established as a holistic approach to understand emergent behavior using formal procedures of systems thinking, synthesis, and mathematical modeling.

Ecological systems are formed by the collection of interacting species (be they plants, mammals, insects, bacteria, viruses, etc.) in a given place (be it a human host or natural habitat) and time, exhibiting complex emergent behavior. Each local system forms a feedback process with the physical and chemical environment and regulates different ecosystem services and functions, such as soil formation, water purification, CO2 sequestration, and human health, among others. Hence, understanding the species composition of ecological systems and its transformations has tremendous potential for bio-conservation, bio-technologies, and bio-medicine.

However, the species composition of ecological systems is in continuous change, reacting and adapting to environmental variations (e.g., pH, temperature, humidity, species richness), in what appears, a strongly context-dependent manner. Importantly, the challenge in identifying regularities governing changes in the species composition of ecological systems resides on the fact that no two systems are the same, assembled under the same initial conditions, subject to identical environmental contexts, nor we have the technical capacity to infer all the exact factors forming the environmental contexts under which ecological systems evolve.

Our research is centered on explaining and predicting the emerging species composition of ecological systems under environmental uncertainty (unknown changing environments). For this purpose, our works is based on a probabilistic systems analysis rooted on the notion of structural stability and uses tools as varied as: population dynamics, structural stability, probability, statistical mechanics, information theory, homology, matrix theory, network theory, geometry, causal inference, and empirical dynamic modeling.

In particular, we believe that change (and lack of change) under known conditions can be systematically understood by investigating the probability of a system to retain its initial composition despite perturbations to its dynamics---what is known as structural stability. Thus, we develop tractable model-driven and scalable data-driven methodologies to estimate the fraction of phenomenological environmental conditions (i.e., a set of variables tied to specific values defining the dynamics of a system) compatible with a given composition of an ecological system. Following this general systems framework, we investigate transformation rules predicting when changes in the species composition of ecological systems are likely to happen, transformation principles dictating how changes in the species composition are likely to happen, and transformation theories explaining why changes in the species composition are likely to happen. While our work is theoretical, we work closely with experimental and field ecologists to corroborate hypotheses of change in ecological systems.