Abrupt Climate Change (ACC) is one of the biggest unknowns of the modern climate system, posing a critical risk to ecosystems and human lives. Climate models struggle to simulate ACCs, underestimating the amplitude of climate variability at decade-plus timescales, and at identifying tipping points with significant accuracy. Paleoclimate records are therefore critical in assessing the magnitude, spatial footprint and impact of ACCs. However, the detection of ACCs in individual paleoclimate studies exceeds its prevalence in statistical summaries. Could this reflect confirmation bias in individual studies, statistical limitations in syntheses, or a language barrier between methodologies? We propose a distinction between an “impact sphere” where the climate impact of ACC is of unusual magnitude, and a “detection sphere” where the climate impact of ACC is detectable but not unusual. This framework will aid paleoclimate inferences and climate modelling of ACCs and help discussions on ecosystem disturbance and societal upheavals, both past and future.

Identifying ACCs is particularly important in the Holocene [1]. Once thought to be relatively quiescent, significant low-frequency climate variability is now recognised as a common feature of the Holocene [2]. With near-modern boundary conditions and high data density, Holocene paleoclimate records of ACCs are a potential goldmine of information for understanding tipping points in the climate system. Further, multi-annual to multi-centennial changes in temperature and rainfall variability are frequently implicated in societal upheavals and transitions via freshwater availability and crop growth [3].

Commonly invoked drivers of Holocene ACCs include freshwater pulses to the North Atlantic, solar variability and clusters of volcanic eruptions. The largest ACC of the Holocene, which occurred 8.2 kyr BP (the 8.2 ka event) is associated with freshwater pulses to the North Atlantic from icesheet melt and/or a glacial lake outburst. Large singular and clusters of volcanic eruptions are now identified as significant drivers of multi-annual to decadal scale climate anomalies [4]. However, internal variability in the climate system is also frequently implicated in Holocene ACCs. ACCs may therefore be forced or unforced.

The 4.2 ka event (4.26 to 3.97 kyr BP) [5], labelled a ‘Global Megadrought’, is implicated in societal upheavals across the northern hemisphere mid-latitudes, drawing substantial interest from paleoclimate and archaeological fields. However, it has no smoking gun driver so it is unclear whether it could be a forced ACC or an example of internal climate variability. While the 4.2 ka event has been identified in many paleoclimate records, statistically rigorous syntheses don’t identify it as having an unusually large magnitude or a coherent spatial signal in the Eastern Mediterranean/Middle East [6], the Indian Ocean [7], the East Asian Summer Monsoon region [8] or globally [9,10]. This holds across both high-resolution, selective syntheses, and low-resolution, inclusive syntheses. What is behind the discrepancy? Are individual studies hunting for excursions and running the risk of confirmation bias and positive results bias, or are different methodologies talking different languages?

Most climate events are likely to have a well-defined signal around some geographic centre of origin. From this centre, the amplitude of the signal decays with increasing distance, although teleconnections can transmit the signal to other regions with minimal decay. These phenomena are apparent in syntheses of the 8.2 ka event [9–11].

We propose two spheres of influence based on signal decay. The narrower “impact sphere” occurs where a climate anomaly has a temporally well-defined, spatially coherent, and anomalously large magnitude signal. The broader “detection sphere” occurs where a climate anomaly can be detected at the right time, but the magnitude of signal is comparable to other local climate variability at the same temporal scale. Outside the two spheres the signal can be considered part of the noise.

As examples, both the 8.2 ka and 4.2 ka events are evident in speleothem stable isotope records from Madagascar, as wet and dry anomalies respectively [12,13]. Yet neither event is of unusual magnitude relative to other proximal centennial scale variability [13,14]. In the broader Indian Ocean domain, the 4.2 ka event is detectable in the second principal component of a pan-regional synthesis, but it is secondary to the secular Holocene trend driven by orbital forcing and a pan-tropical climate shift at 4.0 kyr BP associated with changing El Niño-Southern Oscillation variability [7]. These detectable but moderate signals fit within the ‘detection sphere’ of these events.

Statistically rigorous syntheses tend to focus on the “impact sphere”, while individual studies often focus on the “detection sphere”. For individual studies, searching for and identifying the local impacts of well-known “events” can quickly lead down the path of confirmation and positive result biases. Detecting an anomaly in a distant region that co-occurs with a known event doesn’t guarantee causation, especially if that region has large centennial scale climate variability, anomalous events, and/or there is no a priori expectation for how the event should manifest locally.

Recognizing both spheres is vital to understanding ACC climate dynamics. As a community, we know very little about internal climate variability on multidecadal to centennial timescales. Understanding both the impact and detection spheres of these events will help guide paleoclimate modelling studies and aid studies that focus on the ecosystem and societal impacts of ACCs.

When discussing potential ecosystem and societal impacts of ACCs [3,15] we should be mindful of which sphere the study site is in, and rigorously evaluate alternative hypotheses to avoid false positives. Detection alone doesn’t imply causation for ecosystem or societal changes. Is it likely that a climate anomaly of ‘normal’ amplitude is capable of outsize effects on ecosystems or societies? On the other hand, perhaps even small, not unusual, climate events are capable of outsize impacts, if other factors are at play, such as system structure and complexity. Regardless, we must avoid climate determinism and collapse narratives surrounding ACCs. The role of climate change in societal upheavals, both past and future, must therefore consider the magnitude and unusualness of a climate anomaly as well as the system state, its resilience, and its adaptation potential.