Madleen Grohganz

Tipping points are thresholds at which a system, often abruptly and irreversibly, transitions from a stable state to a contrasting one. Crossing such critical boundaries poses a risk to Earth system stability and may have catastrophic consequences. This is especially relevant, as current climate change is destabilizing Earth subsystems, potentially bringing them closer to tipping points. Thus, it is important to be able to detect approaching tipping points in the Earth’s system, which can be achieved through calibration on palaeo-records. Recently, new deep learning (DL) methods have been established that are able to confidently and quantitatively identify different types of critical transitions characterised by their abruptness and (ir)reversibility. Based on this, we develop a new (simplified) DL classifier focusing on the quantitative detection of catastrophic tipping points (fold bifurcations) in the Earth system. Our approach reduces computational demand and improves performance, especially for short timeseries. We first test the new classifier's performance on synthetic data and subsequently on different existing Cenozoic proxy records. Our DL results are compared to the results from previous studies applying generic early warning signals (EWS), which can detect approaching transitions qualitatively but cannot distinguish bifurcation types (abruptness and (ir)reversibility of the transition). Our DL classifier enables us to identify how abrupt and (ir)reversible an approaching transition is, which is important for tipping point risk assessment and mitigation. Results are generally consistent between generic EWS from previous studies and our DL approach and fit with what is known from the geological context. We note that some results are dependent on the length of the classifier used and the time interval investigated before the bifurcation. We implement an out-of-distribution (OOD) detection method to reduce the misclassification of non-catastrophic bifurcations as catastrophic tipping points. Combined with the binary DL classifier, this approach enables reliable, quantitative detection of catastrophic tipping points in Earth system records.

Tipping points are thresholds at which a system, often abruptly and irreversibly, transitions from a stable state to a contrasting one. Crossing such critical boundaries poses a risk to Earth system stability and may have catastrophic consequences. This is especially relevant, as current climate change is destabilizing Earth subsystems, potentially bringing them closer to tipping points. Thus, it is important to be able to detect approaching tipping points in the Earth’s system, which can be achieved through calibration on palaeo-records. Recently, new deep learning (DL) methods have been established that are able to confidently and quantitatively identify different types of critical transitions characterised by their abruptness and (ir)reversibility. Based on this, we develop a new (simplified) DL classifier focusing on the quantitative detection of catastrophic tipping points (fold bifurcations) in the Earth system. Our approach reduces computational demand and improves performance, especially for short timeseries. We first test the new classifier's performance on synthetic data and subsequently on different existing Cenozoic proxy records. Our DL results are compared to the results from previous studies applying generic early warning signals (EWS), which can detect approaching transitions qualitatively but cannot distinguish bifurcation types (abruptness and (ir)reversibility of the transition). Our DL classifier enables us to identify how abrupt and (ir)reversible an approaching transition is, which is important for tipping point risk assessment and mitigation. Results are generally consistent between generic EWS from previous studies and our DL approach and fit with what is known from the geological context. We note that some results are dependent on the length of the classifier used and the time interval investigated before the bifurcation. We implement an out-of-distribution (OOD) detection method to reduce the misclassification of non-catastrophic bifurcations as catastrophic tipping points. Combined with the binary DL classifier, this approach enables reliable, quantitative detection of catastrophic tipping points in Earth system records.

Tipping points are thresholds at which a system, often abruptly and irreversibly, transitions from a stable state to a contrasting one. Crossing such critical boundaries poses a risk to Earth system stability and may have catastrophic consequences. This is especially relevant, as current climate change is destabilizing Earth subsystems, potentially bringing them closer to tipping points. Thus, it is important to be able to detect approaching tipping points in the Earth’s system, which can be achieved through calibration on palaeo-records. Recently, new deep learning (DL) methods have been established that are able to confidently and quantitatively identify different types of critical transitions characterised by their abruptness and (ir)reversibility. Based on this, we develop a new (simplified) DL classifier focusing on the quantitative detection of catastrophic tipping points (fold bifurcations) in the Earth system. Our approach reduces computational demand and improves performance, especially for short timeseries. We first test the new classifier's performance on synthetic data and subsequently on different existing Cenozoic proxy records. Our DL results are compared to the results from previous studies applying generic early warning signals (EWS), which can detect approaching transitions qualitatively but cannot distinguish bifurcation types (abruptness and (ir)reversibility of the transition). Our DL classifier enables us to identify how abrupt and (ir)reversible an approaching transition is, which is important for tipping point risk assessment and mitigation. Results are generally consistent between generic EWS from previous studies and our DL approach and fit with what is known from the geological context. We note that some results are dependent on the length of the classifier used and the time interval investigated before the bifurcation. We implement an out-of-distribution (OOD) detection method to reduce the misclassification of non-catastrophic bifurcations as catastrophic tipping points. Combined with the binary DL classifier, this approach enables reliable, quantitative detection of catastrophic tipping points in Earth system records.

Osteostracans are considered the sister group of jawed vertebrates and are phylogenetically the earliest branching vertebrate lineage to exhibit true paired fins with endoskeletal support. This means they take centre stage in understanding the jawless‐to‐jawed transition and the origin of paired appendages. Unfortunately, their macroevolutionary significance is obscured by current rudimentary understanding of their taxonomy and phylogenetic relationships. This is especially the case with the notoriously problematic wastebasket taxon ‘ Cephalaspis ’. Previous systematic studies of British ‘cephalaspids’ have relied largely on qualitative assessments of continuous traits, but it remains unclear to what extent these traits can reliably distinguish morphospecies. Here, we perform a geometric morphometric analysis to assess the morphological disparity of ‘cephalaspid’ osteostracans from the Midlands Valley and the Anglo‐Welsh Basin. Using this approach, we find that continuous traits cannot distinguish current/existing taxa within and across faunas, emphasizing the importance of discrete characters to support robust diagnoses. We revise ‘cephalaspid’ taxonomy accordingly. We establish four new genera: Platyrhakis , Saigheadaspis , Ymersaspis and Rochtarge , and propose the synonymy of five species with existing taxa. We estimate the phylogenetic affinity of these new genera using parsimony searches with ‘rogue’ taxon removal and Bayesian analyses. We find that while several groups are statistically well‐supported, they are united by highly homoplastic characters, highlighting the problematic nature of current osteostracan morphological datasets.

Tipping points are thresholds at which a system, often abruptly and irreversibly, transitions from a stable state to a contrasting one. Crossing such critical boundaries poses a risk to Earth system stability and may have catastrophic consequences. This is especially relevant, as current climate change is destabilizing Earth subsystems, potentially bringing them closer to tipping points. Thus, it is important to be able to detect approaching tipping points in the Earth’s system, which can be achieved through calibration on palaeo-records. Recently, new deep learning (DL) methods have been established that are able to confidently and quantitatively identify different types of critical transitions characterised by their abruptness and (ir)reversibility. Based on this, we develop a new (simplified) DL classifier focusing on the quantitative detection of catastrophic tipping points (fold bifurcations) in the Earth system. Our approach reduces computational demand and improves performance, especially for short timeseries. We first test the new classifier's performance on synthetic data and subsequently on different existing Cenozoic proxy records. Our DL results are compared to the results from previous studies applying generic early warning signals (EWS), which can detect approaching transitions qualitatively but cannot distinguish bifurcation types (abruptness and (ir)reversibility of the transition). Our DL classifier enables us to identify how abrupt and (ir)reversible an approaching transition is, which is important for tipping point risk assessment and mitigation. Results are generally consistent between generic EWS from previous studies and our DL approach and fit with what is known from the geological context. We note that some results are dependent on the length of the classifier used and the time interval investigated before the bifurcation. We implement an out-of-distribution (OOD) detection method to reduce the misclassification of non-catastrophic bifurcations as catastrophic tipping points. Combined with the binary DL classifier, this approach enables reliable, quantitative detection of catastrophic tipping points in Earth system records.

Teeth are a key innovation that underpinned the adaptive radiation of jawed vertebrates; however, their evolutionary origin must lie with the diverse tooth‐like structures of living and fossil jawless vertebrates. Most previous studies have focussed on the extinct stem‐gnathostomes that phylogenetically intercalate the living jawed and jawless vertebrates. The only two extant jawless cyclostome lineages, the lampreys and hagfish bearing keratinous toothlets, have long been overlooked, though they possess complex (but unmineralised) toothlets that some have interpreted as precursors to the teeth of jawed vertebrates. Regardless of whether the toothlets of cyclostomes are homologous or convergent on the teeth of jawed vertebrates, they have the potential to offer unparalleled molecular developmental insights into the evolutionary origin of teeth. To that end, we provide a synthesis of classical literature on cyclostome toothlet structure and development, as a basis for informing future molecular studies, to which we add new insights from X‐ray microtomography of three parasitic lamprey species spanning the breadth of the lamprey crown group. Based on detailed morphological analysis we describe their toothlet replacement mechanism at tissue level and uncover a relationship between toothlet size and the number of replacement cones. All examined species reveal the presence of replacement toothlets, suggesting this replacement mode is a conserved feature of the lamprey crown group. We discuss these results in comparison to hagfish, and conclude that toothlet replacement is a symplesiomorphy of cyclostomes. By describing lamprey toothlet development and replacement and comparing it with gnathostome teeth, this study lays the foundation for research into the development and evolution of teeth and tooth‐like structures across vertebrate lineages.

Teeth are a key vertebrate innovation; their evolution is generally associated with the origin of jawed vertebrates. However, tooth-like structures already occur in jawless stem-gnathostomes; heterostracans bear denticles and morphologically distinct tubercles on their oral plates. We analysed the histology of the heterostracan denticles and plates to elucidate their morphogenesis and test their homology to the gnathostome oral skeleton. We identified a general model of growth for heterostracan oral plates that exhibit proximal episodic addition of tubercle rows. The distal hook exhibits truncated lamellae compatible with resorption, but we observe growth layers to be continuous between denticles. The denticles show no evidence of patterns of apposition or replacement indicating tooth homology. The oral plates and dermal skeleton share the same histological layers. The denticles grew in a manner comparable to the oral plate tubercles and the rest of the dermal skeleton. Our test of phylogenetic congruence reveals that the distribution of internal odontodes is discontinuous, indicating that the capacity to form internal odontodes evolved independently several times among stem-gnathostomes. Our results support the ‘outside-in’ hypothesis and the origin of teeth through the spread of odontogenic competence from extra-oral to oral epithelia and the subsequent co-option to a tooth function in gnathostomes.

Early vertebrate evolution has been characterized as a gradual shift from passive to more active feeding modes. However, this evolutionary scenario is contingent on poorly constrained inferences of the feeding ecology of extinct stem‐gnathostomes. Heterostracans are among the earliest members of the gnathostome stem‐lineage. Pteraspidiform heterostracans possessed an oral apparatus composed of rod‐like plates that has been interpreted as being used for either passive suspension feeding or mechanical food processing. Direct tests of the suspension feeding interpretation are challenging and so we tested hypotheses of mechanical function using a combination of microstructural and finite element analysis. Our results demonstrate a negative relationship between simulated minimum principal stress (mostly compressive stress) and bone volume fraction (a proxy for internal microstructure); the higher the stress, the higher the bone volume fraction. This relationship is clearest in the oral plate shaft. The hook, where a load would be applied, shows the highest bone volume fraction values. Our results are compatible with the adaption of skeletal microstructure to mechanical function in which bone adaptively remodels under applied load to become denser to withstand increased stress. On this basis we reject the suspension feeding hypothesis in favour of a mechanical function for oral plates, such as deposit feeding or scavenging, in support for which we observe wear patterns on the aboral surface and the distal tip of the hook compatible with repeated abrasive polishing.

Attempts to explain the origin and diversification of vertebrates have commonly invoked the evolution of feeding ecology, contrasting the passive suspension feeding of invertebrate chordates and larval lampreys with active predation in living jawed vertebrates. Of the extinct jawless vertebrates that phylogenetically intercalate these living groups, the feeding apparatus is well-preserved only in the early diverging stem-gnathostome heterostracans. However, its anatomy remains poorly understood. Here, we use X-ray microtomography to characterize the feeding apparatus of the pteraspid heterostracan Rhinopteraspis dunensis (Roemer, 1855). The apparatus is composed of 13 plates arranged approximately bilaterally, most of which articulate from the postoral plate. Our reconstruction shows that the oral plates were capable of rotating around the transverse axis, but likely with limited movement. It also suggests the nasohypophyseal organs opened internally, into the pharynx. The functional morphology of the apparatus in Rhinopteraspis precludes all proposed interpretations of feeding except for suspension/deposit feeding and we interpret the apparatus as having served primarily to moderate the oral gape. This is consistent with evidence that at least some early jawless gnathostomes were suspension feeders and runs contrary to macroecological scenarios that envisage early vertebrate evolution as characterized by a directional trend towards increasingly active food acquisition.

Attempts to explain the origin and diversification of vertebrates have commonly invoked the evolution of feeding ecology, contrasting the passive suspension feeding of invertebrate chordates and larval lampreys with active predation in living jawed vertebrates. Of the extinct jawless vertebrates that phylogenetically intercalate these living groups, the feeding apparatus is preserved only in the early diverging stem-gnathostome heterostracans and its anatomy remains poorly understood. Here we use X-ray microtomography to characterise the feeding apparatus of the pteraspid heterostracan Rhinopteraspis dunensis (Roemer, 1855). We show that the apparatus is composed of thirteen plates arranged approximately bilaterally, the majority of which articulate from the postoral plate. Our reconstruction of the apparatus shows that the oral plates would have been capable of movement within the dorso-ventral plane, but their degree of movement was limited. The functional morphology of the apparatus in Rhinopteraspis precludes all proposed interpretations of feeding except for suspension/deposit feeding and we interpret the apparatus as having served primarily to moderate the oral gape. This is consistent with evidence that at least some early jawless gnathostomes were suspension feeders and runs contrary to macroecological scenarios that envisage early vertebrate evolution as characterised by a directional trend towards increasingly active food acquisition.