Periodic Reporting for period 3 - ROCKS-PARADOX (Dissecting the paradox of stasis in evolutionary biology)
Periodo di rendicontazione: 2024-01-01 al 2025-06-30
There is something disconcerting about the current state of knowledge on rates of morphological evolution across different timescales: Why do most species in the fossil record exhibit negligible morphological change when contemporary populations often respond rapidly to selection? The ROCKS-PARADOX project will address this fundamental question – known as the paradox of stasis – along mutually reinforcing lines of enquiry, by merging theory and data across paleontology and evolutionary biology.
The prevalence of stasis and other patterns of change are hard to evaluate without knowledge of evolution on timescales unattainable by studies of contemporary populations (microevolution) and comparative species-data (macroevolution). The ROCKS-PARADOX project will address this by analyzing the world’s largest collection of data on within-lineage evolution – spanning decadal to million-year timescales – using a statistical framework (developed by the project) where new and already established mathematical models of evolution are implemented.
The ROCKS-PARADOX project will also assess the effects of genetic constraints and evolvability on evolution beyond microevolutionary timescales. To do this, we will break new ground by estimating quantitative genetic parameters from fossil samples using machine-learning algorithms on a collection of thousands of fossil clonal organisms (bryozoans) from a rich and highly-resolved stratigraphic section spanning 2.3 million years.
The ROCKS-PARADOX project will bridge our current understanding of phenotypic evolution within lineages across timescales into a single cohesive theoretical framework, and open up new avenues for how fossil data can be collected and analyzed to inform questions within evolutionary biology. The project will develop new methodology with broad applications, including long-awaited tools for high-throughput phenotyping.
The prevalence of stasis and other patterns of change are hard to evaluate without knowledge of evolution on timescales unattainable by studies of contemporary populations (microevolution) and comparative species-data (macroevolution). The ROCKS-PARADOX project will address this by analyzing the world’s largest collection of data on within-lineage evolution – spanning decadal to million-year timescales – using a statistical framework (developed by the project) where new and already established mathematical models of evolution are implemented.
The ROCKS-PARADOX project will also assess the effects of genetic constraints and evolvability on evolution beyond microevolutionary timescales. To do this, we will break new ground by estimating quantitative genetic parameters from fossil samples using machine-learning algorithms on a collection of thousands of fossil clonal organisms (bryozoans) from a rich and highly-resolved stratigraphic section spanning 2.3 million years.
The ROCKS-PARADOX project will bridge our current understanding of phenotypic evolution within lineages across timescales into a single cohesive theoretical framework, and open up new avenues for how fossil data can be collected and analyzed to inform questions within evolutionary biology. The project will develop new methodology with broad applications, including long-awaited tools for high-throughput phenotyping.
I have completed the modeling framework, which has been implemented in the software (evoTS). evoTS is openly and freely available on both the **Comprehensive R Archive Network (CRAN)** (https://cran.r-project.org/web/packages/evoTS/index.html(si apre in una nuova finestra)) and **GitHub** (https://github.com/klvoje/evoTS(si apre in una nuova finestra)). I have begun developing adequacy tests for the models introduced in evoTS and integrating them into a software named adePEM. Progress on this work is well underway, and I anticipate completion within the next few months. Two team members, both PhD students, are actively analyzing data and rigorously testing the newly implemented adequacy tests.
A PhD student in the project team has analyzed numerous phenotypic time-series utilizing the new softwares evoTS and adePEM. We are currently in the process of drafting a manuscript to present the initial outcomes of this research.
Another member of our project team, a PhD student, has embarked on an investigation into how we can deduce the dynamics of the adaptive landscape over time by examining fossil time-series data. This student is actively exploring explicit predictions generated by previously proposed explanations for the paradox of stasis.
We have successfully developed and published machine learning tools for efficiently extracting phenotypic trait data from images of bryozoans. Together with other phenotyping pipelines, these tools will contribute to the successful achievement of efficiently estimating quantitative genetic parameters from fossil samples/populations.
A member of our project team, a postdoctoral researcher, is currently utilizing computer vision and other machine learning tools to estimate evolvability and constraints within a fossil lineage of bryozoans spanning 2 million years (Steginoporella magnifica). We anticipate developing a manuscript for this aspect of the project within the next year.
A PhD student in the project team has analyzed numerous phenotypic time-series utilizing the new softwares evoTS and adePEM. We are currently in the process of drafting a manuscript to present the initial outcomes of this research.
Another member of our project team, a PhD student, has embarked on an investigation into how we can deduce the dynamics of the adaptive landscape over time by examining fossil time-series data. This student is actively exploring explicit predictions generated by previously proposed explanations for the paradox of stasis.
We have successfully developed and published machine learning tools for efficiently extracting phenotypic trait data from images of bryozoans. Together with other phenotyping pipelines, these tools will contribute to the successful achievement of efficiently estimating quantitative genetic parameters from fossil samples/populations.
A member of our project team, a postdoctoral researcher, is currently utilizing computer vision and other machine learning tools to estimate evolvability and constraints within a fossil lineage of bryozoans spanning 2 million years (Steginoporella magnifica). We anticipate developing a manuscript for this aspect of the project within the next year.
The modeling framework I have developed (evoTS) incorporates new models of evolutionary dynamics within lineages, including multivariate versions of these models. This modeling framework, which I have exclusively developed, contains specific models tailored for analyzing evolutionary trait dynamics. Therefore, I consider the software's development to represent a novel methodology that wouldn't have been realized without the support of my ERC project. This software has already contributed to gaining new insights into the behavior of evolution in the fossil record, as these models have not been applied to empirical data previously. Given the novelty of these models, it remains uncertain whether they truly provide statistically sound descriptions of observed evolution. Additionally, the creation of statistical tests to assess potential violations of key model assumptions during the fitting of a particular model to data constitutes groundbreaking work. The introduction of a deep-learning-based morphometric characterization tool for cheilostome bryozoans, known as DeepBryo has established new and more efficient pipelines for studying the phenotypes of bryozoans. This organism serves as the model for a significant portion of the research assessing the importance of genetic constraints on phenotypic evolution.