Diversity and evolution of ecosystems: From genomes to the biosphere

In this thesis, I present research on biological systems at three different scales of space and time: biodiversity of ecological systems, the dynamics of repetitive elements and their diversity in the genome, and the development of phylogenetic trees in evolution. The unifying theme is the interplay between ecology and evolution, expressed within an ecosystem, within genomes, and over the evolutionary history of life. Part I concerns biodiversity on the ecological scale. I study the “Kill the Winner” (KtW) hypothesis, a proposed solution to the biodiversity paradox questioning why many competitors can coexist in a single niche. The original KtW model is deterministic and expressed in terms of continuous biomass concentrations, and appears to predict the coexistence of species. Here I present a stochastic individual-level model for the KtW paradigm, representing populations as finite integers. We find an extinction cascade and a monotonic loss of diversity in time due to the stochasticity, thus failing to explain diversity in the presence of stochasticity. To solve this problem, we couple the coevolution of predators and prey with the KtW model, and show that the diversity of the stochastic system can arise from the constant population flux induced by the emergence of new mutants, although there are undoubtedly contributions from the spatial variations in populations too. Our results suggest that diversity reflects the dynamical interplay between ecological and evolutionary processes, and is driven by how far the system is from an equilibrium state. Part II consists of three projects on the dynamics and diversity of repetitive DNA elements on the genomic scale. The first project is to develop a statistical mechanical model for the interaction between two types of DNA transposons, known as LINE and SINE. These mobile genetic elements are respectively autonomous and non-autonomous: SINE steals the machinery of LINE to complete its migration, and thus acts as a parasite. We have found that the demographic noise due to the discreteness of element copy numbers leads to noisy oscillations on the evolutionary time scale, in a similar way to that resulting in the predator-prey quasi-cycles in ecology. By viewing these DNA elements as predators and prey, we have shown that the dynamics in the genome can fruitfully be analyzed using the analogy to ecological models. In the second project, we look for the predicted quasi-cycles of LINE and SINE in the genomic history of the ancient fish coelacanth. We analyze the periodicity of the age distribution recorded in the genome by the molecular clock, and also develop a theoretical model to examine under what conditions can the cycles be recorded. Our analyses provide a procedure for future research work, but the conclusion is that the rapid deterioration of DNA transposons due to mutations means that the observational window is restricted to the last 50 million years, which is not long enough to conclude that the predicted oscillations are present. In the third project, we further explore the analogy of a genome to an ecosystem and DNA elements to organisms. We use the metric known as the rank-abundance distribution (RAD) from ecology to study the diversity of junk DNA “species”. We have found that the RADs for all the 46 examined species can be reasonably fit by a power law, with very similar exponents. This universal RAD can help identify the underlying microscopic evolutionary processes of these DNA species. Our work demonstrates that applying ecological methods to study genomic elements may provide novel insights for genome functions and evolution. Part III focuses on the development of phylogenetic trees on the evolutionary scale. The topology of phylogenetic trees has been found to obey a universal scaling law. The exponent lies in between the two extreme cases of completely balanced binary trees and completely imbalanced ones. We seek evolutionary processes that can generate the observed topology, and here study in particular the effect of niche construction on the large-scale structure of phylogenetic trees. In contrast to the conventional natural selection framework, which treats the environment independently of the organisms under selection, the niche construction theory views the feedback of organisms on their environment as a crucial and explicit process in evolution. We present a coarse-grained statistical model of niche construction coupled to simple models of speciation, and show that the resultant phylogenetic tree topology can exhibit a scale-invariant structure, through a singularity arising from large niche construction fluctuations. These results show in principle how the scaling laws of phylogenetic tree topology can emerge from rather general assumptions about the interplay between ecological and evolutionary processes