Clotilde Hardy PhD thesis

ammonitePhylogenetic distribution of extinctions in the fossil record: contributions in the understanding of the dynamics of the biodiversity

Funding: ministry grant

Supervisors: Pascal Neige and Emmanuel Fara

Beginning of the thesis: october 2009

Defended the 12 december 2012



The disappearance of species has become a major scientific and societal concern over the last decades, and conservation biologists are devoting much effort to understand current extinctions and their potential consequences. Palaeontologists contribute to the debate by putting the current erosion of biodiversity into a deep-time perspective. These two approaches to the study of extinction are currently getting closer as they both increasingly incorporate the same factor: phylogeny. This conceptual and methodological convergence has already revealed that most current extinction risks and past extinctions are phylogenetically non-random: taxa in some lineages are consistently more extinction-prone than others. This phylogenetic clustering is frequent at several spatial and temporal scales, suggesting that extinction-related key traits (or combination of traits) are themselves phylogenetically conserved.

However, there is currently no consensus on how to quantify the phylogenetic signal of extinction (which is treated as a binary variable). Our first aim is to find the most appropriate method to test and quantify the phylogenetic distribution of extinctions. We investigated several statistics proposed in the literature (Moran’s I, D) or inspired by existing methods (Pearson’s f and two indices R²mtl and R²sel based on the multidimensional scaling of phylogenetic distance matrices).

These indices can measure phylogenetic correlation by taking the entire phylogeny into account at once (Moran’s I, D, R²mtl and R²sel) and/or can be computed for various levels of the phylogenetic hierarchy and be displayed as correlograms (Moran’s I, Pearson’s f). In all cases, significance was computed by randomizing the extinction state across species. Results from simulations show that if all indices can determine the presence or absence of a phylogenetic clustering of extinctions, the index D associated to Pearson’s f correlograms is the most appropriate technique for investigating the phylogenetic signal.

As a case study, we investigate the phylogenetic distribution of extinctions on ammonites (fossil cephalopods) from the Early Jurassic. Our dataset has one of the best temporal resolutions available for such a remote geological interval, and it concerns a diverse clade with an excellent fossil record. The implementation of all indices shows that ammonite extinctions were significantly clustered during most chronozones and sub-chronozones of the Pliensbachian-Toarcian interval. This result suggests a phylogenetic conservatism of the susceptibility to extinction signal. Such a phylogenetic pattern is robust in the face of potentially confounding factors (polytomies, species used in biostratigraphy or clades showing no extinction). Interestingly, one of the zones showing a non-significant phylogenetic signal corresponds to a second-order biotic crisis at the end of the Pliensbachian.

This work provides new insights on the dynamics of ancient marine biodiversity from a phylogenetic perspective. It also helps to the recognition of research prospects, especially the correlation between the phylogenetic clustering of extinctions and biological traits through the study of morphological disparity and of the geographic distribution of ammonite species.



phylogeny, phylogenetic conservatism, extinction, fossil cephalopods, ammonites


Gilles Escarguel, université Claude Bernard, Lyon 1
Loïc Villier, université Pierre et Marie Curie, Paris
Thierry Rigaud, université de Bourgogne
Matthew Wills, University of Bath
Christian Meister, Muséum d'histoire naturelle de Genève
Emmanuel Fara, université de Bourgogne
Pascal Neige, université de Bourgogne (supervisor)