FAQ about Gen3sis

Ge3sis is a simulation model. Simulation models represent a simplification of the reality and their development is done along three main lines of abstraction

1. the agent
2. the biological mechanisms
   
3. the landscape dynamic

A first line of abstraction is the characterisation of the agent, which might depend on the mechanisms and the type of emergent pattern that the model should explore. The development of an agent-based model requires the definition of the most indivisible particle, whose interaction will together lead to the emergent pattern of the ecosystem. A second line of abstraction is the interpretation of processes acting on agents to generate sub-models, which should reflect the distinct hypotheses that are evaluated. In the context of biodiversity, the mechanisms may include speciation, dispersal, extinction, trait evolution and ecological interactions acting on the species agents. The sources for defining the structural variation of sub-models are typically the literature containing published textual presentation of various hypotheses as described previously. A third line of abstraction is the numerical simplification of the landscape over which interactions between agents play out. The landscape can be considered as a network of sites that is either static or dynamic and represents a spatial network of nodes connected through vertices. The vertices of the network will determine the species ability to move and/or to speciate between cells, while the properties of the nodes might influence local agent dynamics.

Philosophically, the design of models should be the crystallisation of the scientific knowledge of a system rather than an exercise of fitting datasets to each other. While causation can hardly be tested without manipulation, simulation models provide the closest tool available to researchers for manipulating large complex spatial systems with a strong historical inertia such as ecosystems. So far, hypotheses on the emergence of spatial gradients in biodiversity have been largely verbal and their expectations not compared. The adoption of process-based models of biodiversity could provide a leap forward in the integration of knowledge in ecology and evolution. As in Richard Feynman’s famous dictum “What I cannot create, I do not understand”, the best available confirmation of the understanding of the causes of a phenomenon is the ability to build a simulation model from first principles and reproduce realistic higher-level empirical patterns. Conversely, process-based simulations places causal hypotheses on a leading role, as an a priori abstraction of the inner workings of the system that is used to build the model. Those models are primarily built to understand processes, but their expectation can be then compared to data qualitatively and quantitatively. Thus, predictions from mechanistic models are entirely derived from fundamental theoretical principles, so that tests of prediction of such models represent a potentially more direct validations of the underlying theory used to build them. The simulations from the different sub-models over a static or dynamic landscapes produce ˗ independently from the biological data ˗ emerging patterns. Through the comparison with empirical patterns, sub-models can be ranked in their ability to simultaneously reproduce empirical patterns. The development and application of simulation models, where bottom-level mechanisms act on agent over realistic landscapes, could foster a causal understanding of the processes shaping biodiversity.

Biodiversity has not evolved in a static world, it has evolved in response to dynamic changes in earth’s geology and climate over millions of years. Yet spatially-explicit paleoenvironmental dynamics, such as plate tectonics, mountain building, and changes in earth’s temperature, are often not considered when modelling the evolution of biodiversity. Therefore, we still lack a comprehensive understanding of how geodynamics and global climate change shape and interact with different ecological and evolutionary processes. Recent advances in reconstructions of Earth history from biological and lithological indicators (Scotese 2021), oxygen isotope ratios (Song et al 2019), as well as global climatic models (Valdes et al 2021) and paleo-landscape evolution models (Salles et al 2020) allow us to simulate the evolution of biodiversity in Gen3sis with realistic reconstructions of the Earth’s climate and geology over the Phanerozoic era (540 Ma- present) in order to understand how these environmental dynamics over deep-time have led to the extraordinary biodiversity we observe across the planet today.

Salles et al., (2020). gospl: Global Scalable Paleo Landscape Evolution. Journal of Open Source Software, 5(56), 2804, https://doi.org/10.21105/joss.02804

Scotese, C. R., Song, H., Mills, B. J. W. & van der Meer, D. G. 2021. Phanerozoic paleotemperatures: The earth’s changing climate during the last 540 million years. Earth-Science Reviews

H. Song, P.B. Wignall, H. Song, X. Dai, D. Chu. 2019. Seawater temperature and dissolved oxygen over the past 500 million years. Journal of Earth Sciences, 30:2, 236-243. https://doi.org/10.1007/s12583-028-1002-2

Valdes, P. J., Scotese, C. R., Lunt, D.J. 2021. Deep Ocean Temperatures through Time. Climates of the Past. 17, 1483–1506. https://doi.org/10.5194/cp-17-1483-2021

To run a gen3sis simulation you will need the R package for gen3sis itself, an input landscape, and a simulation config. You can either create your own landscapes or try some of our prepared landscapes here. The same for creating a simulation config. You can create your own config from scratch or browse our published configs here. Once you have your config and input landscape in place you are ready to run a simulation.

All our projects use gen3sis in one way or another to determine the processes and mechanisms that drive biodiversity patterns on different spatial and temporal scales. Past and ongoing projects span a multitude of model systems, from plants to tetrapods, and cover systems from the terrestrial to freshwater and the marine realm. By combining the framework with inputs from various disciplines, such as with geological and paleoclimatic reconstructions or genetic data, we can address a manifold of questions which have been fascinating natural scientists ever since.

In the following we would like to feature four previous or ongoing projects of the group which made use of the gen3sis framework to address diverse research questions.

Natural ecosystems are experiencing unprecedented rates of disturbance due to human activity, contributing to the rapid biodiversity loss we are facing today. Understanding how communities recover from disturbances such as intensive land use or pollution is crucial to understand the spatio-temporal impacts of human activities on community composition and associated functions. This project uses simulated landscapes representing different levels of landscape connectivity and different intra- and interspecific interaction scenarios within gen3sis to investigate respective effects on recovery trajectories following a simulated disturbance event.

Tropical moist forests harbor much of the world’s biodiversity, but this diversity is not evenly distributed globally, with tropical moist forests in the Neotropics and Indomalaya generally exhibiting much greater diversity than in the Afrotropics. Here, we assess the ubiquity of this “pantropical diversity disparity” (PDD) using the present-day distributions of over 150,000 species of plants and animals, and we compare these distributions with a spatial model of diversification combined with reconstructions of plate tectonics, temperature, and aridity. Our study demonstrates that differences in paleoenvironmental dynamics between continents, including mountain building, aridification, and global temperature fluxes, can explain the PDD by shaping spatial and temporal patterns of species origination and extinction, providing a close match to observed distributions of plants and animals.

In the Anthropocene, global changes are causing an erosion biodiversity at a rate unmatched since the last mass extinctions. The processes that shaped current levels and distribution of biodiversity are still not clearly understood, which is hampering our ability to forecast the future of biodiversity over large spatial scales. Moreover, given the strong modification of landscapes globally under anthropogenic pressures, it is urgent to understand the processes underlying natural ecosystems before those systems are too far away from their natural states. Despite the general agreement that extant biodiversity is the result of past environmental changes in combination with extant ecological constraints, we lack a general consensus on the causes of biodiversity gradients. Textbooks prepared for students still present multiple alternative hypotheses, and the science is not settled. One of the main reason for the lack of consensus is that there was no existing framework in which the expectations from the different hypotheses can be contrasted. The process-based modelling approach developed in this project will provide a new framework to compare biodiversity hypotheses with multiple empirical patterns and allow a new means of contrasting among many plausible mechanisms. Another reason for the difficulty to clearly ascertain the factors that generate and maintain biodiversity is the necessity to take a transdisciplinary approach. So far, only a fraction of simulation models considers deep-time paleo-environmental conditions to model the emergence of ecosystem structure. The development of paleo-landscapes is key to simulate and understand the emergence of global biodiversity gradients. The proposed work addresses long-standing unresolved fundamental questions in ecology and evolution, but will also foster collaborative interactions among different research fields.