Zombie fires in peatlands disappear from the surface, smoulder underground during the winter, and ‘come back to life’ in the spring. They can release hundreds of megatonnes of carbon into the atmosphere per year and are believed to be caused by surface wildfires. Here, we propose rate-induced tipping (R-tipping) to a subsurface hot metastable state in bioactive peat soils as a main cause of Zombie fires. Our hypothesis is based on a conceptual soil-carbon model subjected to realistic changes in weather and climate patterns, including global warming scenarios and summer heatwaves. Mathematically speaking, R-tipping to the hot metastable state is a genuine nonautonomous instability, due to crossing an elusive quasi-threshold, in a multiple-timescale dynamical system. To explain this instability, we provide a framework combining a special compactification technique with concepts from geometric singular perturbation theory. This framework allows us to reduce an R-tipping problem due to crossing a quasi-threshold to a heteroclinic orbit problem in a singular limit. We identify generic cases of tracking–tipping transitions via: (i) unfolding of a codimension-two heteroclinic folded-saddle-node type-I singularity for global warming and (ii) analysis of a codimension-one saddle-to-saddle hetroclinic orbit for summer heatwaves, in turn revealing new types of excitability quasi-thresholds.

The Lower Old Red Sandstone in southern Ireland is hosted in the Early Devonian Dingle Basin, which lies immediately south of the Iapetus Suture on the Dingle Peninsula, County Kerry. The basin developed as a post-Caledonian pull-apart structure prior to Acadian deformation, which in turn was followed by end-Carboniferous Variscan deformation. Detrital zircon U–Th–Pb geochronology is complemented by mica Ar–Ar and apatite U–Pb geochronology to gain a comprehensive understanding of the provenance of the Lower Devonian Lower Old Red Sandstone of the Dingle Basin and assess contributions of major tectonic components (e.g. Laurentia, Ganderia). Sedimentary rocks in the Lower Old Red Sandstone have similar detrital zircon age distributions, which are dominated byc.1.2 Ga zircons as well as late Neoproterozoic grains. This indicates a dominant contribution of detritus of Laurentian affinity as well as contributions from westerly and southerly derived Ganderian detritus. Caledonian uplift of the area north of the Iapetus Suture would have facilitated a large contribution of (peri-)Laurentian material. The Upper Old Red Sandstone on the Dingle Peninsula has a distinctly different detrital zircon character including few late Neoproterozoic zircons and abundant zircons ofc.1.05 Ga age, indicating sediment derivation only from Laurentia and no recycling from the Lower Old Red Sandstone.Supplementary material:The full detrital U–Pb zircon and apatite analytical dataset and the revised detrital mica age dataset are available athttps://doi.org/10.6084/m9.figshare.c.3971862

We present a modelling framework for the spreading of epidemics on temporal networks from which both the individual-based and pair-based models can be recovered. The proposed temporal pair-based model that is systematically derived from this framework offers an improvement over existing pair-based models by moving away from edge-centric descriptions while keeping the description concise and relatively simple. For the contagion process, we consider the susceptible–infected–recovered (SIR) model, which is realized on a network with time-varying edges. We show that the shift in perspective from individual-based to pair-based quantities enables exact modelling of Markovian epidemic processes on temporal tree networks. On arbitrary networks, the proposed pair-based model provides a substantial increase in accuracy at a low computational and conceptual cost compared to the individual-based model. From the pair-based model, we analytically find the condition necessary for an epidemic to occur, otherwise known as the epidemic threshold. Due to the fact that the SIR model has only one stable fixed point, which is the global non-infected state, we identify an epidemic by looking at the initial stability of the model.

We present preliminary results on an all-Ireland network modelling approach to simulate the spreading the Severe Acute Respiratory Syndrome CoronaVirus 2 (SARS-CoV-2), commonly known as the coronavirus. In the model, nodes correspond to locations or communities that are connected by links indicating travel and commuting between different locations. While this proposed modelling framework can be applied on all levels of spatial granularity and different countries, we consider Ireland as a case study. The network comprises 3440 electoral divisions (EDs) of the Republic of Ireland and 890 superoutput areas (SOAs) for Northern Ireland, which corresponds to local administrative units below the NUTS 3 regions. The local dynamics within each node follows a phenomenological SIRX compartmental model including classes of Susceptibles, Infected, Recovered and Quarantined (X) inspired from Science 368, 742 (2020). For better comparison to empirical data, we extended that model by a class of Deaths. We consider various scenarios including the 5-phase roadmap for Ireland. In addition, as proof of concept, we investigate the effect of dynamic interventions that aim to keep the number of infected below a given threshold. This is achieved by dynamically adjusting containment measures on a national scale, which could also be implemented at a regional (county) or local (ED/SOA) level. We find that – in principle – dynamic interventions are capable to limit the impact of future waves of outbreaks, but on the downside, in the absence of a vaccine, such a strategy can last several years until herd immunity is reached.