Contextual Framing

Historical introduction: Minoan Crete and Santorini among volcanoes of the world


Mt. Tambora on Sumbawa Island in Indonesia exploded in April 1815 and caused 60,000 to 100,000 deaths on Sumbawa and neighbouring islands alone (Oppenheimer 2003; 2011). It is the most disastrous volcanic eruption in recent history. The voluminous sulphur emissions from this eruption degraded the Earth’s climate and 1816 is famously known as the ‘Year without a Summer’ because of severe climate abnormalities that caused average global temperatures to drop (Self et al. 1984). This resulted in poor harvests and food shortages across the Northern Hemisphere and was probably responsible for an outbreak of diseases and major social upheavals in Europe (D’Arcy Wood 2014). The 1815 eruption also produced voluminous tephra, a fragmental material deposited by falling through the air. The tephra-affected area was probably over 500.000 km2 and destroyed or inflicted severe damages to forests, agriculture, livestock and buildings across the whole region (Self et al. 1984; Oppenheimer 2011; Kandlbauer & Sparks 2014). Moreover, a giant raft of volcanic rock, light enough to float on the surface of water, clogged the nearby sea for two to three years after the eruption.

The eruption of Mt. Tambora provides some clues about the various potential local, regional and global impacts of ancient, catastrophic volcanic events on environment and society. One such eruption was that of Santorini volcano (also called the Thera eruption, the Minoan eruption or the Late Bronze Age eruption) in the Aegean Sea, which shook the Mediterranean world approximately 3,600 years ago. This eruption occurred when the Minoan civilisation on Crete and on the Aegean Islands was at its height. Tephra from Santorini has been found as far away as the Black Sea indicating a minimum fallout area exceeding two million square kilometres. Tephra fall entirely covered the Minoan settlement of Akrotiri on Santorini (Doumas 1983), but also affected communities and agricultural areas on nearby islands and the coast of Crete, probably in combination with seismic shocks and tsunamis. A generation or so after the event, the palaces and major towns on Crete and the other islands were suddenly destroyed and Minoan society collapsed (Driessen & Macdonald 1997; Driessen 2019). Eventually, the Mycenaean civilisation of Mainland Greece took advantage and expanded geographically, politically and economically.

While many studies highlight that a volcanic eruption can effectively affect the trajectory of human society (e.g. Sheets and Grayson 1979; Sheets 2004; Grattan & Torrence 2007; Sheets 2008; Oppenheimer et al. 2018; Torrence 2019), the causal sequence, if any, between the Santorini event and the breakdown of Minoan society is not yet resolved. In particular, several scholars point out that not much tephra fell on the island of Crete, the centre of Minoan civilisation, and that a volcanic tsunami would not have been strong enough to cause large destructions (Perianez & Abril 2014). Perhaps even more puzzling, there is a time lapse of several decades between the eruption and the downfall of Minoan society. However, recent disaster studies stress that evaluating the response of a society impacted by a natural hazard should go beyond solely physical hazard assessment and event-focused reactions (e.g. Oliver-Smith and Hoffman 2002; Wisner et al. 2004). In other words, disasters are inherently multidimensional and they unfold over time, often even over considerable periods of time (Oliver-Smith 2002). As such, it is essential to appraise the social and cultural determinants of their nature and extent.

Arguably, the Santorini case epitomises the difficulty to unravel the response, and its underlying mechanisms, of a human society subjected to a catastrophic natural event such as a large volcanic eruption. It provides one of the most intriguing examples where the complex relationships and interactions between volcanic hazards and socio-cultural traits may have shaped the level and duration of social disruption. Shedding light on this problem necessitates the recognition of culture, including the interplay of belief and knowledge systems of various kinds, as a critical component mediating the volcanic disaster and exacerbating or mitigating its impact (e.g. Lavigne et al. 2008; Bachri et al. 2015; Torrence 2018).

Historical background

Since at least 2000 BCE, the island of Crete (Fig. 1) was home to an elaborate palatial culture that focussed on a series of complex monumental structures that have been called ‘palaces’ and which were essentially public court-centred buildings with ceremonial, religious, economic and probably also political-administrative functions. Primarily meant to enhance social cohesion of the surrounding urban settlements and regions through a series of ritual practices, these so-called ‘palaces’ epitomise what is known since Arthur Evans’ excavations at Knossos as the Minoan civilisation. Built immediately after 2000 BCE, the first generation of these palace buildings was violently destroyed, either by human action and/or earthquake in the 18th c. BCE. Even though ignored for a long time in archaeological literature, knowledge on the following Middle Minoan III period, roughly between 1675 and 1600 BCE, significantly advanced thanks to a series of recent studies (cf. Macdonald & Knappett 2014). While showing an increased pace of development at Knossos, they also illustrate at least a temporary decline in older palatial centres such as Malia and Phaistos while Knossos-inspired constructions are constructed elsewhere (Galatas, Kommos, Sissi, Palaikastro). These features, combined with other indications within the material culture (Wiener 2016), may provide some support for the claim of an increasing Knossian supremacy during this period. This would evidently have been accompanied by an agricultural overspecialisation and an intensification in the complexity of hierarchical and administrative structure, rendering the system potentially unstable and hence vulnerable (Hamilakis 1996; Haggis 2002). All palatial buildings, including those at Malia and Phaistos, seem again to have been damaged late in the 17th c. BCE, in early Late Minoan IA in ceramic terms (henceforth LM IA) (Fig.1). Aegean archaeologists traditionally suggest that this destruction was caused by a massive earthquake, which is also attested on the island of Thera or Santorini, an island that like many other Aegean islands, had a material culture quite close to that of Minoan Crete. As on Crete, at the site of Akrotiri on Santorini, this so-called Seismic Destruction, acted as a catalyst of social, economic, religious and political change. On both islands, it provoked a new, more generalised, building boom and immediately preceded yet another phase of prolific artistic production, overseas trading and ritual elaboration (Doumas 1983; Palyvou 2005). This phase on Crete is known as that of the ‘New Palaces’ and the period is known as the Second or New Palace period, which is mainly, in ceramic terms, of Late Minoan I style.

Late Minoan I represents both the acme of Minoan civilisation and the prologue to its ruin 150 years later. Somewhere during this long period, Santorini volcano erupted. The precise absolute date of this event remains a matter of considerable discussion since traditional dating methods, primarily based on synchronisms with Pharaonic Egypt, indicate a date that is up to 100 years later than the date(s) suggested by dendrochronology, ice cores, speleothems and C-14 (Hoflmayer 2012; Wiener 2009; 2015). Partly because magmatic CO² that biases C14 dating (Holdaway et al. 2018), certain adjustments of scientific dating seem now to close the temporal gap and a date around the middle of the 16th c. BCE (ca. 1550-1540 BCE) now appears as more likely (Friedrich et al. 2006; Heinemeier & Friedrich 2009; Cherubini et al. 2013; Cherubini et al. 2014; Ehrlich et al. 2018; Pearson et al. 2018). It is not unlikely that some palatial centres on Crete, such as those at Kommos, Galatas, Zominthos and Sissi, were destroyed or abandoned at or around the moment that the volcano erupted, in Late Minoan IA.

Fig. 1 – The Aegean, chronological table and major settlements on the island of Crete, including those with tephra depositions.

During Late Minoan IB (henceforth LM IB), between 1520 and 1450 BCE, fire destroyed most Cretan settlements and palaces and afterwards site numbers dwindle. Only Knossos shows during Late Minoan II, between 1450 and 1400 BCE, signs of renaissance and prosperity, but now with a Mycenaean veneer, especially materialised by the adoption of a new language, Greek, for its administration in Linear B, and entirely new burial customs. The Late Minoan IB destructions can really be considered as the collapse of the Minoan poly-palatial system and most of the conspicuous material correlates that characterised the acme of Minoan society disappear together with these ‘palaces’.

The question is: what caused the downfall of this culture?

Since the initial discovery of antiquities on Santorini and the subsequent geological explorations of the island by Ferdinand Fouqué in 1866, the eruption has been linked to the mythological destruction of Atlantis (Fouqué 1870; Figuier 1872, Nicaise 1885) and already in 1909, K.T. Frost suggested that the Atlantis myth could reflect the demise of Minoan Crete, the civilisation discovered by Sir Arthur Evans from 1900 onwards. Other myths as that of the Flood of Deukalion or Talos have likewise been connected to this event, as have been the Seven Plagues of Egypt (Ellis 1999; Zellinga de Boer & Sanders 2002). The archaeological evidence, however, paints a less dramatic picture. Following the recognition of LM IB fire destructions in the archaeological record, potential culprits were looked for. Because these destructions were absent from Knossos but very visible elsewhere, Arthur Evans blamed them on 'the tyrannous lust of domination on the part of the lords of Knossos' (Evans 1922: 348; 1928: 308). In 1939, John Pendlebury (1939) considered the fire destructions to be 'a deliberate sacking on the part of the enemies of the most powerful cities of Crete'. Only in the late 1950s were LM IB fire destruction deposits also found at Knossos showing that the site had not escaped the problems that the other settlements had faced (Hood 1962). By then, Michael Ventris had deciphered the Knossian Linear B tablets, proving that, by Late Minoan II, immediately after the fire destructions, the administration at Knossos was run by Mycenaean Mainlanders. From then onwards, the latter were considered as the agents of the LM IB destructions, progressively expanding their domination over the island and the Aegean (Hood 1986). Other scholars, however, blamed a series of devastating earthquakes as responsible for the LM IB destructions (Pichler & Schiering 1980; Warren 1991; cf. Jusseret & Sintubin 2017).

That the Santorini eruption was responsible for damage on Crete was, however, already recognised by Arthur Evans in 1922, who blamed some of the earlier, LM IA destructions at Knossos on the eruption (Jusseret 2014). This line of thought was categorically picked up by Spyridon Marinatos in 1939, who, having excavated the coastal villa of Amnissos on Crete, assumed it was devastated by the eruption and its accompanying tidal waves (Marinatos 1939; already Schoo 1937, however). Marinatos explicitly linked the LM IB fire destructions of Cretan sites with the eruption, believed to have been one of the largest in the history of mankind. He was followed in this by several archaeologists (e.g. Platon 1972) while others saw the eruption as causing a climate change, responsible for the Minoans’ downfall (Tsonis et al. 2010). The opening of the excavations at Akrotiri on Santorini from 1967 onwards by the same Marinatos, however, clearly confirmed that the eruption happened when Late Minoan IA ceramic styles, characterised by vegetal and floral decoration, were in use, while, at the moment of the LM IB fire destructions on Crete, a marine ceramic style was popular (Doumas 1983). The discovery of Theran ash in clear LM IA contexts in different sites on Crete (Fig. 1) and Rhodes from the late 1980s onwards, settled the chronological difference between the moment of the eruption and that of the fire destructions on Crete (e.g. Molloy et al. 2014; Driessen 2019).

Using all available evidence as well as theoretical principles of emerging disaster studies, Driessen & Macdonald (1997) recognised a pattern of modifications in the Cretan archaeological record after the eruption. By the time of the LM IB destructions, for example, many sites saw their internal circulation pattern changed with accesses to houses made more difficult, houses being subdivided, and enclosure walls being constructed. Agricultural and artisanal production was seen as receiving more attention while religious and artistic changes were noted. They also stressed certain features of the LM IB destructions: hence, in some settlements, only the palatial buildings burnt down in LM IB while the settlement was simply abandoned and elsewhere too, sites were abandoned and left empty. Some of the fires that brought down palatial buildings appeared to have been carefully prepared and in certain sites, the malicious destruction of some fine pieces of art took place. Moreover, many LM IB destruction contexts include hoards of bronze vessels or precious items, hidden beneath floors or vases, showing advance warning of disaster. All these features were considered as reflecting a crisis before the LM IB destructions, caused by humans. Since more detailed pottery studies had shown that the LM IB destructions were not a one-time event but on ongoing and long-term process, an internal crisis of Minoan society was suggested, triggered off by the Santorini eruption and its collateral damage. Their hypothesis was contested by some (Christakis 2014; Warren 2001) while incorporated in more general schemes of breakdown processes by others (Knappett et al. 2011). Using computational network analysis, Knappett and colleagues suggest that the Minoan collapse at the end of LM IB was one result of an incremental increase in exchange costs required to maintain commercial ties following the disappearance of Santorini as a key node in the network, with the remaining sites concentrating their exchange/trade efforts into a smaller number of stronger links at the expense of weaker ones.

Destruction, Adaptation and Resilience: The Contribution of Disaster Studies

Following pioneering work by O’Keefe et al. 1976 and White 1974, disaster studies were professionalised by scholars such as Oliver-Smith (1996; 2013; 2017), Oliver-Smith & Hoffman (2002), Blaikie et al. (1994), Wisner et al. (2004) and Garcia-Acosta (2018). The impact of disasters on human societies evidently depends on the magnitude, duration and frequency of the phenomenon, on its impact on natural resources, on the pre-existing adaptive strategies of the affected human population and on their size and distribution. But as increasingly emphasized in disaster research, disasters (including volcanic ones) are multidimensional events because they occur at the conjunction of a human population and a potentially destructive agent. Oliver-Smith (1996: 305) describes it as an event that involves a “combination of a potentially destructive agent(s) from the natural and/ or technological environment and a population in a socially and technologically produced condition of environmental vulnerability” (see also Oliver-Smith & Hoffman 1999; Rodriguez et al. 2007; Faas & Barrios 2015). Barrios (2015) evokes the notion of the unpredictable environment as a trickster (see also Ullberg 2015). Disasters sweep across every aspect of human life, affecting environmental, social, economic, political and biological conditions. Negotiating land-use planning with the environment and human communities is a key factor in disaster prevention. Indeed, disasters must be understood as the product of a society, rather than as natural phenomena. Oliver-Smith (2013: 281) is certainly right when claiming that: "By situating the cause of disasters in society and in society-environmental relations, social vulnerability highlights the fundamental tensions that exist in our environment and in our social relations" Bubandt (2005) has coined the notion of “vernacular security”, showing that security is also socially situated rather than an analytical and global category. Others have questioned the role of expert knowledge or that of security in disaster reconstruction (Barrios 2010; Revet 2009, 2013).

Deconstructing a disaster highlights the unequal distribution of vulnerability to risks and invites us to put into practice, as Oliver-Smith (2013) suggests, a political ecology of disasters. Just as it is necessary to deconstruct disasters and the complexity of the forms of vulnerability, the current challenge is also to look at what is done at the local level, instead of giving too much attention to what is not done at the national political level. Thus, societal organisation is a fundamental element in the way that groups respond to disasters; it defines the vulnerability of the group exposed to the physical hazard (e.g. Oliver-Smith & Hoffman 2002; Sheets 2012; 2015). As disasters develop, they reveal the connections and the interpenetrations of natural forces, power structures and social arrangements, but also cultural values, belief systems (Oliver-Smith 2002; Hoffman & Oliver-Smith 2002; Hermesse 2016). To what extent disasters are remembered (or not!) by local people remains a question of debate (see Ullberg 2010; Simpson 2013). Narratives can be vague, ambiguous, they take place in a perceptual world organized through language and symbolic forms (see Skinner 2000; 2004) but also social practices (Falk 2011). As it has been shown more recently, ontologies (Moreau 2014; 2017; see below), unveil a society’s pattern of vulnerability and resilience. The impact of environmental hazards on human groups is concomitantly conditioned by geological variables and social features (Riede 2015; 2018), both of which will be of special interest in this research project. Disaster studies have shown how such events produce long-term effects and can act as catalysts for political, economic, social and psychological actions (Gibbs 2000; 2003; Leach 1994; Dynes & Turney 1994). Temporally speaking, parameters such as the speed of onset, the recurrence and the duration of the catastrophic event can also leave traces in the archaeological and environmental record. The duration of the event is important because it is directly linked to different buffering approaches (Riede 2018: 2), usually comprising risk management strategies in two phases: the emergency and the rehabilitation phase (Driessen 2019: 5-7). Both imply a whole gamut of human behaviours ranging from physical storage and resource distribution to modifications in social networks and group mobility, variations in terms of economic intensity but also alterations in belief systems and associated practices. What these coping mechanisms have in common is that they are highly culture-specific and hence variable (Hermesse 2014). In other words, addressing how ancient societies dealt with environmental hazards can constitute a very rewarding angle of inquiry into their sociopolitical, cultural and socioecological conditions, categories that can all merge in the broad notion of ontologies, as was well demonstrated by Descola (2005; 2011) but also Latour (1999; 2015) and Danowski & Viveiros de Castro (2017).

To give some examples: crisis studies have convincingly argued that violence, antisocial behaviour and community conflicts are in fact rather rare in the emergency period, but frequent during the rehabilitation period because of the discriminatory nature of the recovery process, when underprivileged minority groups find it much more difficult to regain their former level of subsistence and unleash their grievances. Conflict then usually arises because of problems with the allocation of resources for rehabilitation and the assignment of blame (Brändström 2016). Where the allocation of resources is concerned, many authors have underlined the role of famine as an accelerator of historical change, with effects on demography, economy, politics, society and culture (Vanhaute 2015). Frequent cultural responses to reduced harvest yields include a reduction in population size, a change in the distribution of human groups (including their mobility patterns), a diversification of production and the conversion of food into direct and indirect storage (Halstead & O’Shea 1989). The allocation of blame after a volcanic disaster has received considerable attention in scientific literature (references in Black 1981; Hansell et al. 2006; Johrendt 2007; Doocy et al. 2013). 'Acts of God' may represent an easy out for modern-day insurance companies, but disasters in ancient societies were closely connected to religion and ritual (Douglas 1995; Driessen 2015). Since there is a widespread tendency to explain disasters in terms of the sins of the people or its leaders, victimization and scapegoating are common human reactions following disasters. Such blame is often directed against specific, individual groups, especially during the recovery process. In contrast, other case studies illustrate a greater cohesion amongst local groups, which, in political terms, sometimes translates into moves towards regionalism, decentralization and the formation of new groups (Kirschenbaum 2003; Rodriguez et al. 2007; Stor et al. 2016). Blame is also an element, which may have played a role in the post-disaster phase when it comes to issues of authority and leadership, as new strong figures or charismatic persons may have replaced earlier institutionalized and more collective or corporate lines of command (McCaughey et al. 1994; Rodriguez et al. 2007). Whatever the particulars, the potential for radical change resulting in uncommon authority structures and innovative social organization exists. That disasters can also cause psychological distress is illustrated by a study by Adams & Adams (1984), who describe how, during the seven months following the Mt St Helens eruption, there was a 18.6 % increase in mortality rate, a 21 % increase in emergency room visits, a 198 % increase in stress-related illnesses, a 235 % increase in mental illnesses, a 25.5 % increase in sick leave, a 45.5 % increase in domestic violence, and a 37.5 % increase in aggression in the area where the ash had fallen (cf. Shore et al. 1986).

Even more importantly, the resilience of a society, its ability to anticipate, contend with, and recover from a disaster is an important parameter. From Latin resilire (= bounce off), it denotes the way a community can adapt to, or cope with, both acute and chronic stresses (Cyrulnik 1999). Furedi (2007) has argued that community response to a disaster episode is far more likely to be defined by its vulnerability, its degree to experience harm (Turner et al. 2003; Riede 2015), than by its resilience, and that vulnerability is a cultural construct, a state of mind, rather than a state of being that emerges in response to a disaster, i.e. it is something that precedes, not follows, disaster (also Cardona 2003). Unique resilience profiles are, as shown by Coetzee and colleagues (2017), evident in almost all communities studied.

In the social and natural sciences, the concepts of resilience and vulnerability are now intrinsically related to address the issue of the interplay between environmental hazards and human groups as well as disaster as transformative events (Sheets 2012; 2016; Riede 2018: 8; Driessen 2019: 4). The first “highlights creative responses, cultural continuity and the […] ability of communities to collectively weather shocks and threats” whereas the second concept “draws attention to internal social tensions, the inherent social fragility of many social constellations and the role of hazards as stimulants of cultural change” (Riede 2018: 8). Both perspectives are important and worth investigating, most notably because both resilience and vulnerability inherently involve adaptive behaviours which, very often, have their counterparts in material culture and ultimately in the archaeological record of the communities affected by the environmental hazards.

But resilience and vulnerability cannot be considered in ‘one-size-fits-all’ scenarios – every disaster is different, as are the vulnerability and resilience of each society. Bergstrand and colleagues (2015) have argued that the most vulnerable societies are also the least resilient and that it is much more the type of social system, societal development and infrastructure rather than the exposure to hazard that result in loss and destruction. Hence, the term social vulnerability, which is measured by factors such as economy, demography and infrastructure. Each crisis is by definition multifarious, involving challenges at a variety of levels and scales. Although disasters may affect everybody, certain groups will be more vulnerable – the poor, the elderly, the sick etc. Depending on their location, e.g. close to the shore or within a densely populated area, people will also be affected more. ‘Disasters’, as aptly underlined by Shimoyama (2002), ‘are not [only] natural phenomena but [also] social phenomena’ (our emphasis). We argue that the impact of a disaster and the resilience of a particular society can only be studied successfully if all cultural and environmental variables are considered so that sufficient knowledge can be gathered to characterise the ethos or ontology of a society, to identify its level of vulnerability, within a time-and-place perspective (see below). Paton and colleagues (2001), for example, have shown how the existence of stronger social support networks and community ties effectively facilitates resilience while Tselios and Tompkins (2017) specifically suggest that a greater transfer of political power to subnational tiers of government reduces hazard impacts on populations. By inference, this means that politically decentralised political units should resist better than centralised ones because decentralised decision-making usually reacts more rapidly because of proximity and better knowledge of local actors and resources. Layered decision-making, decentralisation and high local autonomy seem all beneficial when it comes to avert crisis and augment resilience (Edwards 2015: 836). This is because decentralised decision-making usually reacts more rapidly because of proximity and better knowledge of local actors and resources (Miller & Douglas 2016, also Aoki 2016; Ng 2016). García-Acosta (2014) proposes a new notion that articulates practices, actions, objectives, strategies and capacities around the prevention of disasters in a community context. It is the notion of "community social construction of prevention."

Forensic volcanology?

Volcanic explosive eruptions are characterised by multiple hazards that can unfold at different times and in different places. They violently unleash tephra (a fragmented rock) and toxic gases into the atmosphere, often together with searing avalanches known as pyroclastic density currents. On volcanic islands and coastal volcanoes, the brutal entry into the sea of materials released by an eruption may initiate a tsunami. Other hazards, including lava flows, debris avalanches and debris flows can also occur. Because volcanoes are often located in densely populated areas, their eruptions can be particularly disruptive to society; they put lives and assets directly at risk. The destructive force of an explosive eruption materialises most strongly locally. However, for high magnitude events, disturbances can also be felt far away from the volcano. A large volcanic blast has the potential to blanket tens to hundreds of thousands of square kilometres with several centimetres of fine-grained tephra. To illustrate this, the rock material expelled by an eruption with a size comparable to that of the largest volcanic event of the last century (Katmai in 1912 in Alaska) is sufficient to bury the entire UK under 11 centimetres of tephra (Oppenheimer 2011). Even a few millimetres of tephra deposition can initiate a wide range of impacts on both the natural and built environments, with ripple effect in human society. Further, if the eruption injects a massive amount of sulphur gas into the atmosphere, cooling of the climate may ensue and the impact becomes global (Oppenheimer 2011 and references therein). Few places in the world would be immune if a volcano somewhere were to explode in a manner similar to some of the catastrophic eruptions that occurred in prehistoric and historic times.

In the span of human life, eruptions can change a landscape from forest to desert, degrade the global climate (the 1815 eruption of Tambora in Indonesia and more recently, the 1991 eruption of Mt. Pinatubo in the Philippines), induce great floods, and even bury entire cities (the Late Bronze Age eruption of Santorini which destroyed Akrotiri, the AD 79 eruption of Vesuvius which ruined Herculaneum and Pompeii, the 1902 eruption of Mt. Pelée in Martinique which devastated St. Pierre). Through the environmental havoc they create, they can deviate the cultural trajectory of society, its structure and functioning. Although such change is most likely to happen near the volcano where (almost) complete obliteration occurs, as on the island of Sumbawa, there is abundant archaeological and anthropological evidence that cultures that are more distal can also be impacted profoundly by a large volcanic eruption. Sheets (1979) and Dull et al. (2001) provides a fascinating description of the intersections between the voluminous eruption of Illopango in El Salvador, possibly in the first half of the sixth century, and society. Popocatépetl volcano in México is another example where an eruption precipitated social disturbances. The archaeologists Plunket and Urunuela (2006) believe that this eruption probably served as a catalyst to accelerate deep social transformation that was already on its way. In Ecuador, the ca. 2900 BCE and ca. 1200 CE eruptions of Cuicocha and Quilotoa, respectively, have been linked to modifications in agricultural practices in the Andes (Mothes 1998). In Melanesia, the work led by Torrence (e.g. 2009; Torrence 2016) highlights the complex relationships between the Papuans and volcanic disturbances of varying magnitude and frequencies. Around 13,000 years ago, Laacher See volcano in present-day Germany erupted and spread tephra across an estimated 225,000-km2 area. This event triggered significant demographic fluctuations, which in turn resulted in a number of material culture transformations (Riede 2007; 2008).

Undoubtedly, volcanoes and their eruptions are crucibles of changes. However, it often proves difficult to establish firmly how these unfold and to disentangle causation from correlation. This is partly due to poor, or lack of, distal records of eruption-induced environmental disturbances, the trigger that may ultimately alter the trajectory of a society. Besides, the social response to these more or less severe perturbations in the physical environment can take various forms, which are inherently difficult to apprehend as they are shaped by the vulnerability and resilience of the affected population. Vulnerability and resilience are themselves complex traits; they are shaped by socio-cultural variables and evolve with time. Further, the effect of an eruption on a cultural group’s worldview may be long lasting, or may have acted as catalysts for social change not apparent until many years after the volcanic event (Torrence 2019). The long and still on-going debate surrounding the role played by the Late Bronze Age eruption of Santorini in the downfall of Minoan culture is an emblematic illustration of these difficulties (Driessen 2019). On the one hand, we currently lack substantial evidence of the environmental impacts of the eruption in the region and, on the other hand, we have only limited knowledge of the vulnerability and resilience of the Minoan culture that was affected. Overall, various interpretations of how past volcanic eruptions affected human society have emerged but, often, these have not been put to the test.

While being an essential component for appraising the significance of a past volcanic eruption as an agent of environmental and social changes, the destructive potential of the event as inferred from various measurements must be matched with impacts on the atmosphere, environment, climate and society. Knowledge of the magnitude, sulphur loading of the atmosphere and location of an explosive volcanic event of the past helps to infer the potential it had to degrade the global climate. When the eruption is strong enough, the sulphuric acid particles that derive from the sulphurous gases injected into the mid atmosphere may eventually be trapped in glaciers and ice sheets, leaving a readable chemical fingerprint. These data can sometimes be cross-checked with tree-ring anomalies, the latter being able to reveal temporary climate changes. For some of the very large volcanic events, historical accounts suggest a connection between climate alteration and crop failure, food shortage and even famine thousands of kilometres from the eruptive centre (Oppenheimer 2011).

There is temptation to use the magnitude of an explosive eruption to speculate on the distal environmental impacts from tephra (e.g. Riede 2018; Sevink et al. 2019). However, this metrics is too coarse and certainly captures neither the multiple nature of the tephra hazard nor the complex interconnections of their impacts on the environment (Ayris & Delmelle, 2012). In order to improve our understanding of the range of inter-relationships between a large volcanic eruption and distal societies, one needs to assess the types of physical and chemical volcanic hazards, and their likely intensity, which can exist at several tens or hundreds kilometres away from the volcano. Typically, this requires careful field observations in order to identify tephra layers preserved in soil profiles or in archaeological excavations. Modelling of tephra dispersion and deposition may complement the task, although in most cases insufficient data are available for parameterizing correctly the volcanic conditions at the source (Bonnadona et al. 2015; Biass et al. 2019). While often regarded as the primary driver of environmental impact, the thickness of a tephra deposit at some distance from the volcanic source is only one aspect of the hazard intensity; grain size, surface composition, hardness and even colour are the other tephra properties that may strongly dictate impacts on vegetation, soil and water (Ayris & Delmelle 2012; Arnalds 2013; Craig et al. 2016). For example, our recent work in Ecuador demonstrates that a few millimetres of fine tephra can be far more damaging to crops than a thicker but coarser deposit (Guevara 2015). The presence of salts on the tephra surface also considerably increases the potential for injury to plants (Ayris & Delmelle 2012). When fluoride is associated to the tephra, its deposition on soil and in water bodies may represent a serious health hazard to both animals and humans (e.g. Thorarinsson 1979; Cronin et al. 2000; Cronin et al. 2003; Weinsten & Davison 2004).

Recognizing that the impact of a volcanic eruption on society is the result of the complex interactions and intersections between people, environment, volcanoes and volcanic activity implies not only to describe the volcanic hazards but also the conditions of the natural and cultural environments in which they occur. This is to say that the exposure, vulnerability and resilience of the natural and human assets vary greatly from one place to another (Blong 1984; Aspinall & Blong 2015). Thus, obtaining information on the climate, ecosystem type, agricultural system, social, political and belief systems and eruption timing is essential to attribute impacts to volcanic hazards. However, discipline-bound forensic observations cannot encapsulate this complex realm of interactions. This is true for contemporaneous eruptions but even more so when dealing with past events. For example, the traces of previous agricultural practices may not be apparent in a modern environment, or the tephra deposit laid down distally by an eruption may have been thick enough for causing damage to agriculture but too thin for being preserved as a visible layer in soil profiles. Nevertheless, the impacts from past volcanic eruptions are recorded in diverse ways in archaeological assemblages that can be recovered through various methods (see below). Moreover, memories and records of eruptions in oral or inscribed forms offer a mine of information that can be decrypted and interpreted in terms of the nature of the volcanic hazard and its potential link to the identified environmental effect (Cashman and Cronin 2008; Fast 2008). Thus, it is only through the combination of volcanological environmental, archaeological and anthropological studies that we will able to unravel fully the changes brought about in the environment and society by a catastrophic volcanic event see also Riede 2018). Arguably, little effort has been made to adopt this approach and there is ample room for considerable progress. The reconstruction of the Long Island volcano eruption that occurred in Papua New Guinea some 300 years ago (Blong 1982) probably represents the most comprehensive attempt to confront observations from volcanology and anthropology.

Anthropological approaches to volcanic disasters

That disasters have considerable impact on the behaviour and psychology of people and religious systems has been shown through a variety of historical studies. For example, emperor Leo III of Constantinople believed that the eruption of Santorini in 726 CE reflected the wrath of God, provoking the destruction of icons from Byzantine churches and the start of the first iconoclasm. It is precisely those impacts, which turn a natural event into a disaster. The disaster unfolds as its effects emerge. A natural disaster by definition cannot exist on its own.

The idea of making greater use of local or indigenous knowledge and taking it into account when studying so-called ‘natural’ disasters and volcanic eruptions in particular has been around for a long time, but this objective is far from being achieved. Research remains fragmented and researchers have little opportunity to connect their findings with how these disasters are actually perceived by indigenous people. A better communication between the natural sciences and the human and social sciences would make it possible to meet this challenge. To capture the different types of human behaviour associated with volcanic eruptions, this project builds on two paths taken by the natural and social sciences in recent years, which have led to significant advances. The first step is to give the ‘other’, by which we mean here especially animals and plants, unique powers to anticipate, manage and respond to disasters. For example, biologists now know that some plants anticipate droughts by producing more flowers or by changing their sex or even their genes, that trees communicate with each other when a stressful situation arises, or when an earthquake is coming (see Hallé 2004; 2013). Other biologists indicate that some birds (e.g. golden-wings warblers) predict storms, being able to perceive ultrasound frequencies of less than 20Hz, etc. (Bondaz et al. 2015). Anthropologists who have long worked with indigenous peoples also report such knowledge and their ability to anticipate disasters by observing the fauna and flora around them. Blaan people from Mindanao, for example, an indigenous group that inhabits an ancient volcano, insist that they can anticipate earthquakes by listening to the Butan bird or by observing chickens hopping on roof tops (Laugrand et al. 2018a: 187). Alangans from Mindoro prepare for a typhoon if paypalis birds fly in flocks, or if they see their pigs gathering their food (Laugrand 2015). The present project will have the advantage of confronting and bringing together these different types of knowledge. The second step is the proposal formulated by comparative anthropologists for an effective model that will make it possible to better understand the cognitive drivers of human engagement with the environment and to test the compatibilities and incompatibilities of intercultural dialogue.

Human behaviour towards volcanoes has varied widely throughout history and across societies and cultures. Urban and demographically important civilizations do not behave in the same way as small societies do. Human reactions and attitudes vary according to the ontological patterns that predominate in cosmologies, i.e. according to systems of thought and practice. Anthropologist Philippe Descola (2005) has identified four main schemes (naturalist, animic, analogy, totemism; Fig. 2) that structure all societies and oppose each other, even though there is a wide variety of cultures within each of these four patterns. This theoretical framework appears as relevant for thinking about volcanic disasters (Moreau 2017). It makes it possible to consider the variety of cultural specificities and cognitive, social and political factors. Each ontology refers to a dominant scheme among the four:

Fig 2 – The four main ontological schemes as defined by Descola (2005)

Since the advent of modernity in the 16th c., Western societies have been part of the naturalist system. The rigorous knowledge they have accumulated is built on the premise that culture is distinct from nature, so that all phenomena, such as disasters and volcanic eruptions are perceived as results of identifiable mechanics, often predictable and explicable according to a so-called modern rationality. In naturalistic cosmologies, to use the terms of Descola (2005; 2011) and Bruno Latour (1999), volcanoes are objects that lend themselves to a multitude of scientific measurements, to the extent that we are convinced that science can protect humans living around them by providing them with prevention and even prediction tools. Within such a system, humans question the risks and potential damage, they mobilize experts, techniques and quantifiable data.

But other societies favour other interpretations or intelligibilities. Thus, some animic cultures consider volcanoes as subjects and not as objects. These societies do not distinguish or rather separate nature from culture as clearly as naturalism does. In such systems, volcanoes and mountains are considered persons as having an interiority comparable to that of humans (Povinelli 1995). This means that volcanoes are given a consciousness, a memory, a sociality, an agency, a capacity of reaction, etc. In these societies, humans can communicate with the volcanoes near which they live. The latter are often associated with deities who, in exchange for the favours they offer to humans (fertile land, etc.), also expect them to respect them, give them offerings and sometimes even sacrifices. Here, humans interact with volcanoes as they do with other humans, other animals and deities. In the face of a ‘disaster’, the question is therefore who is responsible, who is to blame and what is the motive for this volcanic anger? What are the motivations of the non-humans involved, the rules that have been broken and in what way compensation can be made? These contrasts make dialogues difficult since all the potential points of view are almost immeasurable, but no less fruitful.

A third scheme that is very present particularly in areas where volcanoes abound is analogy, a system in which humans establish a multitude of relationships and connections between the observations they make at different levels. In these societies, humans connect the microcosm to the macrocosm; they consider that if humans do not respect the social rules and prohibitions that exist in their universes, mountains and volcanoes are likely to take revenge by sending misfortunes, disease, etc. (Cruikshank 2006; Hermesse 2016). Equilibrium is the rule of cosmos, and a misbalance is always a threat. To individual disorder thus responds another disorder. To avoid volcanic eruptions, some consider it essential to follow prohibitions that may imply taboo food or social practices as long as they do not generate the anger of the deities who live in volcanoes or who are embodied by these volcanoes. The question here therefore focuses on the links between the disorders observed and the ability of humans to restore a harmonious situation. Like animism, humans seek to grasp the motives that caused the disaster and identify regulatory principles. A fourth scheme is totemism, which in fact is quite rare. In this cosmological configuration, the question is how the disaster happened, under what circumstances and what resources are threatened.

These four schemes are co-present in all cultures but usually one predominates over the others. They also overlap on occasion. In any case, they reveal four explanatory modes: the search for causes in naturalism, for motives in animism, for principles in analogy and for reasons in totemism (Moreau 2017: 359). This said, an additional element that explains the different ways of understanding volcanic eruptions is the presence of religions, such as the Abrahamic ones (Christianity, Islam etc.), which convey very different representations and ask for contrasting gestures (Grim 2001). Sometimes, scientists refer to the presence of “a sacred ecology” (Berkes 2008), yet these categories raise new problems (Laugrand 2016). Thus, volcanoes are sometimes associated with bad-tempered divinities, with expressions of divine will or with forces of nature. The religions involved are themselves often transformed or coloured by local indigenous traditions, as is the case in the Andes (Bouysse 1984; 1992), or in Melanesia, among the Orokaiva (Belshaw 1951) or in Polynesia among the Tonga (Quesada 2005), or even in Japan (Augendre 2004). Sébillot (1887) has commented upon this phenomenon since the end of the 19th century but his analysis remains incomplete.

In short, in a very large number of societies, humans can literally interact and ‘talk’ to volcanoes. They live with them and ensure their kindness through gestures and rituals. As naturalists, we can only think of ‘protecting ourselves’ from volcanoes by using scientific tools and experimenting with the effectiveness of predictive measures, but others live with volcanoes. In all cases, volcanic disasters are predicted with varying degrees of success. There are thus avenues for both ontological and interdisciplinary dialogue.