Assemblage theory and disaster risk management

I Introduction

This article, and the analytical framework it develops, builds on work that links disaster studies and human geography (Gaillard and Mercer, 2013). To this end, it develops the conversation between disaster risk studies and assemblage theory (AT) opened up by Donovan (2017) in this journal and by others elsewhere (e.g. Angell, 2014; Gillard et al., 2016; Grove, 2013; Grove and Adey, 2015; Marks, 2019). The article begins with an overview of contemporary disaster studies literature and some of the contradictory understandings of disasters, hazards, vulnerability and risk that are present within it (Kelman, 2018). We go on to argue that although it is correct to argue that disasters are not natural, this statement in itself does not actually answer the question of nature in disasters (Chmutina and Von Meding, 2019; Dynes, 2000) in a satisfactory manner. We argue that there is a danger that by arguing that disasters are 100 per cent not natural and/or socially constructed, one deterministic view – that of viewing disasters as completely natural events – is replaced by another that excludes the uncertainty of the material environment and its significant relationships with the social (Donovan, 2017). To provoke further theoretical engagement, we ask: if disasters are not natural, then what are they? The article answers this question by discussing the emergence of the ‘Disaster Risk Management (DRM) Assemblage’ analytical framework in which disasters are conceptualised as ‘more-than-natural’ but also ‘more-than-human’ (Braun and Whatmore, 2010; Donovan, 2017). From the outset, we propose that the idea of a DRM Assemblage can be used in multiple ways. It may be used as an analytical tool to aid the analysis of disaster risk and its management, as proposed by Donovan (2017). The idea can also be used to describe multiple DRM Assemblages which are composed not only of the apparatuses of power which attempt to manage disaster risk in particular places but also of the phenomena that they seek – and fail – to ‘manage’.

The DRM Assemblage – as an analytical tool – is conceptualised as emerging from different strands of human geography and disaster scholarship. It was developed as the ‘DRR assemblage’ in Donovan (2017), but in line with the recent (2017) disaster risk terminology adopted by the UN (UNGA, 2016, 2017), we prefer ‘DRM’ assemblage here. ¹ We enhance the conceptualisation of the framework by first exploring the emergence of flat ontologies with respect to debates over the nature of scale, causality and agency in human geography (Castree, 2005; Demeritt, 2002). We then unpack our understanding of the ‘virtual space’ of assemblages, starting from an adoption of Adam and Groves’ (2007) reconceptualisation of the Deleuzo-Guattarian virtual as ‘futures-in-the-making’. We argue that a greater focus on the futures of assemblages can help to address the ‘question of nature’ posed above as well as some of the common critiques of assemblage thinking in geographical scholarship. To complete the conceptualisation, we link the futures of AT to evental geographies and geographies of emergency and crisis governance. In addition to the DRM Assemblage as an analytical tool, we propose the term can be used to describe the assembled apparatuses of governance which seek to govern more-than-human life – again linking to established understandings of emergency governance in human geography (Adey et al., 2015). To help conceptualisation, we propose that Donovan’s (2017) six components of DRM Assemblages – explored in depth in that piece – can be used to help researchers frame their analysis of disaster risk and its management in all of its complexity:

governance and governmentality in disasters;

We develop this work by exploring how the DRM Assemblage may be deployed in praxis by proposing four methodological principles of research that draws upon the DRM Assemblage as an analytical tool. With these principles in mind, we propose that the broad category of ‘more-than-human methodologies’ are particularly relevant to studies drawing upon the DRM Assemblage (Dowling et al., 2015).

II Reclaiming Hazards for the Social: Vulnerability, Unnatural Multi-hazards and Complexity

In the 1970s and 1980s, disaster-focused researchers in the broad field of political ecology argued that, at the time of their writing, ‘most risk research…and efforts at hazard reduction, have assumed a severity and geography of risk based primarily upon hazard agents’ (Hewitt, 1992: 38). In response, they conducted research to illustrate that many of the disasters which were being studied were influenced significantly by political and economic processes (Hewitt, 1983; Watts, 1983; White, 1974; Lewis 1981, 1984) and thus were in fact not ‘natural disasters’ at all (O’Keefe et al., 1976). The United Nations’ Sendai Framework for Disaster Risk Reduction of 2015 is testament to the fact that this understanding of socially constructed vulnerability and disaster risk has cut through, at least at the international level and in a theoretical sense (UNDRR, 2015). However, the predominance of the hazards paradigm still exists at national levels (Briceño, 2015).

The flagship conceptual frameworks that emerged out of this disasters literature are the well-established ‘pressure and release model’ and associated ‘access model’ (Blaikie et al., 1994; Wisner et al., 2003). These authors suggest that the models are best used to describe ‘composite events’, triggered by a definite and singular hazard, as opposed to ‘complex emergencies’ – generally understood as the combination of environmental hazards and violent conflict/state fragility (Wisner et al., 2003: 91). Different strands of contemporary disaster research have since argued that such ‘composite’ disasters exist only in simplistic imaginations of how disasters unfold.

The study of multi-hazard interactions suggests the idea of single, natural hazard triggers is simplistic. In reality, most disasters are characterised by multiple hazard processes triggering a series of secondary – and beyond – hazards (Gill and Malamud, 2014, 2016, 2017). Vulnerability and the impacts of disasters also interact with multiple hazards, particularly where recovery is prolonged (Pescaroli and Alexander, 2018). Furthermore, critical work in political ecology suggests that the idea of ‘natural’ hazards fails to consider how human actions actively contribute to the lethality/magnitude of hazards (Mustafa, 2005) or how scientific expertise in the modelling and mapping of hazards can also contribute to the effective, or not, management of hazard events (Donovan, 2017). We emphasise that hazards must always be understood in relation to both the knowledges that represent them and those at risk, but that this understanding should not render the hazardous process as unable to act in unexpected ways: there are epistemic, stochastic and social uncertainties that influence hazards (Donovan, 2019). Hazards are both natural and unnatural; humanly defined and modelled, but also more-than-human.

The conceptualisation of disasters as composite events has been problematised further by studies that have shown that more than 50 per cent of deaths from ‘natural-hazard related disasters’ occur in conflict affected and/or fragile contexts (Peters, 2017; Peters and Budimir, 2016). Frameworks that are unable to account for the complexity of disasters in areas affected by conflict cannot then be drawn upon to understand the majority of disaster-related deaths (Collins, 2019; Marktanner et al., 2015; Siddiqi, 2018). Even disasters which occur in peacetime tend to have long-lasting impacts following the initial ‘event’ (Cutter, 2018); assuming the event itself is even a singular event in space and time – rarely a valid assumption in reality (Pescaroli and Alexander, 2018).

Given these drawbacks, understandings that speak more to complexity and the non-linear interaction of ‘socially constructed vulnerabilities’ and ‘natural’ hazards have started to flourish within the study of disasters (Fekete and Fiedrich, 2018). One such conceptual framework is ‘social/socio-ecological systems’ (SES) (Adger, 2006; Shukla et al., 2017) and more recently cascading disasters – with the latter essentially representing the linking up of a plurality of SES (Pescaroli and Alexander, 2015). In these framings, it is the intersection and interaction of both social and biophysical/technological processes that determine the vulnerability – or resilience/adaptive capacity (Smit and Wandel, 2006) – of places (Cutter, 1996: 537). Such approaches, particularly cascading disasters (Pescaroli and Alexander, 2016), rest upon the ecological and essentialist concept of ‘panarchy’ (Holling, 2001; Holling et al., 2002). However, the essentialist and geometric conception of scale within panarchy is at odds with the understandings of scale on which contemporary political ecology and critical disaster studies rest (Blackburn, 2014; Briassoulis, 2017). Furthermore, critics suggest these approaches are unable to reflect accurately ‘the social’ in disasters due to the tendency of ecological concepts to quantify, or omit, unquantifiable components of disaster risk such as power and the political implications of violent conflict (Ahmed and Kelman, 2018; Hinkel, 2011). Ultimately, many analytical approaches tend to oversimplify through their ontological basis. Understanding disasters, then, requires engagement not only with concepts but also with epistemology and ontology.

III Flattening Ontologies in Geography

Theories relating to SES and cascading disasters are at odds with the understandings of scale prevalent in human geography and urban political ecology (UPE). In such interpretations, scale is generally conceptualised as relational and co-produced by socio-ecological phenomena in ways that serve the needs and wants of the powerful, often at the expense of marginalised groups (Brenner, 2003; Neumann, 2009; Smith, 2010; Swyngedouw and Heynen, 2003; Walker, 2005). While such interpretations are now commonplace in human geography (Miller and McGregor, 2019), tensions remain over the idea that powerful human actors are invariably responsible for the production of these relational scales and interactions (Castree and Braun, 2001; Gandy, 2008; Head, 2009; Smith and Doel, 2010). This tension emerges partly as a result of the fact that UPE is theoretically rooted in Marxist geographies and the associated dialectical understandings of the relationship between capitalist society and nature that these geographies provoke (e.g. Smith and O’Keefe, 1980). Despite the more contemporary and hybridised interpretations of these dialectical approaches which can usefully dissolve the false society/nature dualism (Royle, 2017), these approaches still tend to overemphasise the dominance of capitalist relations in specific contexts (Gabriel, 2014), where other more powerful relations – such as geopolitics, uncertain scientific understandings and cultural imaginaries – may be of equal importance to the analysis (Collier and Ong, 2005; Grove, 2009).

It is these tensions and assumptions that were, and remain to be, contested by advocates of ‘flat ontologies’ in geography (Marston et al., 2005). Partly as a result of debates over these tensions, there has been a degree of cross-fertilisation between UPE approaches and assemblage-based flat ontologies (Castree, 2002; Holifield, 2009). For example, Ranganathan (2015) identifies areas where UPE and assemblage-based approaches can productively engage with each other to understand the flows and fixities of urban flooding in Bangalore, India. Here, capitalist modes of production interact with legacies of colonialism, changing cultural understandings of storm drains and the agency of water itself (Ranganathan, 2015). We largely agree with Ranganathan’s approach but also advocate for more engagement and synthesis with the oft-neglected field of disaster studies, which has a lot to offer – and learn from – these debates. Such a synthesis is well placed to unpick how these relations and interactions emerge in situated socio-material assemblages (McFarlane, 2011; Neisser and Müller-Mahn, 2018). Such interactions are conceptualised as acting not across hierarchical scales but rather between sited assemblages where power is differentially distributed across scales as outcomes or effects of these interactions (Agrawal, 1995; Escobar, 2001; Legg, 2009). Disaster studies have a lot to offer such inquiries, particularly the idea that while powerful human actors are generally better positioned to resist the impacts of, recover from, and adapt after, disasters (Blackburn and Pelling, 2018; Pelling, 2012), it is difficult to suggest disaster events – such as volcanic eruptions – emerge in ways that always serve the needs of the powerful (Bennet, 2005), even if their impacts are frequently manipulated in that way.

IV AT: Reassembling ‘the Virtual’ as Futures-in-the-Making

In AT, the virtual can be understood as the possibility spaces which emerge out of the behaviour of – and relations within – assemblages (DeLanda, 2016). Debates over the precise conceptualisation of the Deleuzo-Guattarian understandings of virtuality remain alive and have been covered in depth elsewhere (Buchanan, 2017; DeLanda, 2005, 2006; Deleuze, 1994; Groves, 2010, 2019; Massumi, 1992). An in-depth synthesis of these discussions is beyond the scope of this article, where we choose to adopt the well-established understanding that the interactions of socio-material assemblages can result in one of many potential outcomes, with some outcomes materialising more often than others (DeLanda, 2016; Lane et al., 2013; McConnell and Dittmer, 2017). To ground our understanding of AT in the contemporary geographical literature, we follow Adam and Groves (2007)’s concept of ‘futures-in-the-making’ – a rewording of the problematic Deleuzo-Guattarian concept of ‘the virtual’ (Adam and Groves, 2007: 175). For Adam and Groves (2007: 196), futures-in-the-making are real, despite not being material – reflecting the DeLandan and Deleuzo-Guattarian understanding of the virtual as real but not actual (DeLanda, 2005). Futures-in-the-making are expressive assemblage components (Adam and Groves, 2007: 196), which, together with actual material and expressive components, may be thought together as heterogeneous assemblages through which space is territorialised (Groves, 2017: 32).

A good example from the literature which explains these ideas of territorialisation in relation to futures-in-the-making is Davis and Groves’ (2019) research on post-Olympics urban planning in London. They show how formalising processes lead to the planning assemblage being coded predominantly by the ‘economic rationale’ of developers, as opposed to the everyday realities and routines of the local people (Davis, 2019). They argue that this process showed how the ability to anticipate and territorialise socio-ecological imaginaries is differentially distributed among actors in assemblages and thus how the territorialisation of the imagined futures of the powerful deterritorialises the imagined futures of the marginalised (Davis and Groves, 2019: 27–30). A synthesis of this approach with disaster studies would work in a similar way but also pay attention to how disasters tend to emerge not as imaginations per se – this would be rather counterintuitive – but more as unforeseen disasters-in-the-making (Pelling et al., 2020). In the DRM Assemblage, it is not only the planned – and often periodic – results of decision-making that are the subject of analysis but also the unintended collective consequences of human decisions, which can be transformative (DeLanda, 1991; 1997: 16–17). Such an engagement with this politics of disaster could also pay attention to the ongoing dominance of the hazard paradigm in the imagined futures of governments (UNDRR, 2019).

Scholarly work on the political implications of the 1999 Marmaris earthquake provides a good example of such an engagement. Drawing on the literature cited above which seeks to denaturalise disaster imaginaries, Pelling and Dill (2010) explore the ways in the earthquake impacts – particularly the damage done in Istanbul – opened up a number of potential futures in Turkey. They explain that while the disaster did lead to the materialisation of an altered social contract, the government closed down other – more transformative and unsettling for the government – futures-in-the-making through economic sanctions and political suppression (Pelling and Dill, 2010). Drawing upon AT specifically, Angell (2014) analyses the same disaster to show how the risk of earthquakes was reassembled as a natural and existential threat to the population by the government to reinforce the need for national governments to lead DRM initiatives. This reassembly ran counter to popular narratives within civil society that ‘earthquakes don’t kill people, buildings do’, or that the disaster was not natural but the result of poor governance and risky, outsourced urban development (Angell, 2014). Angell explores the reverberations of the disaster through space and time in relation to contestations over urban development by showing that the government deployed imaginations of future earthquakes to pass a ‘Transformation of Areas under Disaster Risk law’ through parliament. The law was criticised for bypassing legal obstacles raised against previous urban renewal projects and was also perceived to be a useful political opportunity for the government to create a profitable opportunity for the construction sector (Angell, 2014). AT, then, provides a stronger theoretical basis for established ideas in disaster studies such as disasters representing windows of opportunity for political change or tipping points. Investigating these ideas also dovetails well with the motivations behind action research agendas in disaster studies (Yadav et al., forthcoming): the desire to effect change through research and empower the marginalised to shift the dynamics of the futures-in-the-making. These acts are also collaborative between the earth itself (the geopower of the earthquake) and human society – but not in a linear and predictable way. The relationship is rather mediated through the assemblage as it is territorialised and reconfigured by, for, and through different groups.

There are also important contributions to the theorising of possible futures in human geography by scholars who focus on emergency governance and the politics of possibility and potentiality (e.g. Amoore, 2013; Cooper, 2011). We recognise a distinction between such studies and ‘critical disaster studies’, a body of literature that exists within geography, other disciplines, and in its own right. The reason for this distinction seems to stem partly from different disciplinary histories, with critical disaster studies emanating from more practice-based approaches linked with humanitarianism (Alexander, 1997, 2013; Lechat, 1990) and political ecology (Hewitt, 1983), while geographies of emergency and crisis governance have their roots in scholarly work on political economy and political philosophy (Agamben, 2005, 2009; Foucault, 2007, 2012). Each area of work certainly focuses on risk, but each tends to speak of it in different temporal terms, and much of the emergencies and crisis literature (though not all) is focused on the developed world. Critical disaster studies have largely endeavoured to show how past decisions, political economy and environmental degradation led to the creation of risk and subsequently an emergent past or present disasters, most often in developing contexts (Collins et al., 2015; Oliver-Smith, 1999). The scholarly work on the governance of future emergencies and crises has paid more attention to the ways in which both emergent disaster assemblages, and the prior imagination of these disaster assemblages, influence, and have influenced, the ways in which governments seek to influence the conduct of their populations (O’Grady, 2014, 2018). Other distinctions in research foci revolve around the types of risks and hazards considered. Critical disaster studies have historically been more preoccupied with meteorological and geophysical hazards (Burton, 1978; Collins et al., 2017), while the geographical study of emergency governance has tended to focus on political/state emergencies (Anderson, 2020), national defence and terrorism (De Goede, 2012), health emergencies (Adey and Anderson, 2012) and technological/infrastructural emergencies (Lakoff and Collier, 2010).

In recognising these distinctions, this article can be read as an attempt to further develop the conversation and possible convergence of these disciplines and also outline work which has already sought to bridge these gaps (Barnett, 2020; Blackburn and Pelling, 2018; Grove, 2014b; Grove and Adey, 2015). We argue that critical disaster studies would benefit from paying more attention to the work imagined disasters – and attempts to mitigate them – do in the realm of politics and governmentality, as is done in geographies of crisis and emergency governance. This type of critical work on the practices of disaster risk reduction (DRR) such as early-warning systems, vulnerability assessments and technology-based risk mitigation measures – too often presented as politically neutral and purely ‘science-led’ – certainly exists but has thus far only had a marginal impact on policy and practice (Borie et al., 2019; Donovan and Oppenheimer, 2015; Farías, 2014; Mustafa et al., 2015). Pointing critique in the other direction, the geographical study of crises and emergency governance could learn from critical disaster studies by theorising further how the imagined crises and emergencies that are acted upon by governments are also directly related to present decision-making and also display qualities which demand the integration of knowledge from outside the social sciences (Donovan et al., 2019). The former is particularly important so that geographical studies of emergency and crisis governance and governmentality do not reproduce the narrative that disasters – particularly those which emerge in relation to geophysical and meteorological phenomena – are, or have to be, inevitable, exceptional or ‘natural’ (Kelman, 2020).

V Disaster Assemblages: Reconciling Process and Outcome

In AT, disasters can be considered as an actualisation of one of many possible futures of an assemblage of expressive and material components: as ‘disasters-in-the-making’. The materialised disaster represents a disaster assemblage – ‘characterized by complex ideas, physical processes, physical-human interactions, human cultures and technologies that experience a varying power distribution in time’ (Donovan, 2017: 51). It has been argued above that the territorialisation of assemblages can be constrained by historical development trajectories but also that the (de)territorialisation caused by disaster assemblages can disrupt and transform these trajectories through reverberations of Geopower (Donovan, 2017; Grosz, 2008). There are however key theoretical tensions which arise out of this conceptualisation of disasters. These tensions can be summarised as between understanding disasters as potentially transformational but also as moments where dominant power relations persist, and between conceptualising disasters as outcomes, or events, rather than processes (Manyena, 2012). The latter also reflects a dilemma facing assemblage theorists, apparent in Delanda’s conceptualisation of assemblages as ‘real things’ which can be analysed, while simultaneously in a constant state of becoming (Buchanan, 2017). On these tensions, Guggenheim (2014) argues that while disasters are reflective of ‘normal’ social development processes – or normal social development processes experienced at a higher intensity – treating them as such belies the fact that they are also inherently related to non-social/human processes, often experienced as ‘events’, and do represent ruptures originating outside of social processes. The latter understanding of disasters has been associated with a Badiouan understanding of events by Cloke et al. (2017), who considered the 2011 Christchurch earthquake through a Badiouan lens. In a similar vein to Pelling and Dill (2010) and Angell (2014), they argue that the earthquake created a possibility space – through a rupturing of previously territorialised power relations – for new subjectivities and political assemblages to emerge, as well as giving ‘pre-quake political visions traction/momentum in a way that was not previously possible’ (Cloke et al., 2017: 74). They argue that the Badiouan framing of the earthquake allows us to consider disasters as ‘continuing events’ which are both ‘mappable moments’ in time but also continually reterritorialised according to how people relate to and recollect ‘the event’.

While such a Badiouan understanding of disasters as events can be productive, Deleuzian understandings of events (e.g. Patton, 2002) can be of equal utility in the analysis of disasters and also speak more to the aforementioned tension between understanding disasters as transformative events but also as processes through which unjust power relations are reterritorialised. Such a Deleuzian reading is particularly illuminating when read alongside critical studies of disasters as outcomes of development processes (Collins, 2018) and evental geographies. In a conceptual piece which both synthesises and differentiates the philosophies of Badiou, Deleuze and Heidegger, Shaw (2012: 622–623) suggests that an ‘evental geography’ should be attentive to what sort of transcendental powers hold assemblages together and also to how ‘geo-events’ can – through a form of creative destruction – de-anchor the integrity of these powers by forcing reconfigurations of how nation states, institutions and individuals manage more-than-human life (see also Donovan, 2020). Building upon Shaw’s geo-event, it could be said here that disasters-in-the-making are the ‘inexistent objects’ that are simultaneously created and held back by the seemingly stable, stratified and transcendental power relations that constrain development trajectories and hold people’s imaginations of the socio-material world in place (2012: 622). This synthesis of evental geographies with critical disasters studies helps us to understand how these seemingly transcendental and stratified power relations try to hold back the infinite contingency of the world before, during and after disaster events but also how they contribute to the emergence of the disasters-in-the-making which periodically disrupt and deterritorialise those very same power relations and development trajectories (Shaw, 2012: 622). This understanding reflects the conceptualisation of assemblage put forward by Legg (2011), where Deleuzian assemblages and Foucauldian apparatuses are considered dialectically. Legg suggests that while Foucauldian assemblages appear as those expressions of power that seek to manage the emergence of more-than-human life through omniscient foresight and the enforcement of certain types of development, they also get muddled and mix things up, producing new subjectivities – and disasters-in-the-making – which, through their emergence and ongoing reverberations, force governments to reconsider ‘the new’ and thus the configuration of their DRM Assemblage/apparatus (Legg, 2011: 130–131).

Thus, we propose the DRM Assemblage as both the assembled apparatuses that seek to manage risk and emergence in a given location and as an analytical tool which can be used to analyse these assemblages/apparatuses in relation to disasters-in-the-making. Such an analysis allows researchers to consider critically how risk management techniques have emerged over time (Donovan and Oppenheimer, 2014, 2015), while remaining attentive to the localised conditions of what risk management might mean in a particular place (Woods, 2015; Zeiderman, 2012, 2016). This synthesis of disaster scholarship and AT can help to unpick why some disasters-in-the-making resemble Deleuzian events that reshape power relations through their territorialisation (Beck and Gleyzon, 2016), while some do not (Kelman, 2011; Siddiqi, 2013, 2014). The DRM Assemblage – as an amalgamation of AT and disaster studies – echoes calls in the geographical literature to not fetishise the description of the aleatory at the expense of critically analysing the racialised (Anderson et al., 2019), gendered (Kinkaid, 2019), sexualised (Seymour, 2013) and uneven territorialisation of space and disasters (Grove, 2014a; Wachsmuth et al., 2011).

VI DRM and the Future – Towards DRM Assemblages

A DRM Assemblage – as an object of study – can be conceptualised as emerging from the relations between those assemblages/apparatuses of governance which are concerned with governing the futures of more-than-human life, disaster assemblages and the socio-material relations between those components which lead to the emergence of disaster risk in a given location. The way it behaves is determined by the interactions between its more-than-human component parts, which determine the possible futures-in-the-making of the DRM Assemblage, and thus the materialisation – or not – of disaster assemblages. In the disaster and development paradigm, sustainable development equates to DRR (Collins, 2009: 218). Thus, in an ideal world, the DRM Assemblage would sit within assembled apparatuses of government to code development – where development is understood as the continued territorialisation of futures-in-the-making (Mathews and Barnes, 2016) – with logics of risk reduction, equity and sustainability. To link this with both the geographical literature on crises and emergency governance and critical disaster studies, respectively, a DRM Assemblage would be coded by logics of precaution and pre-emption (Anderson, 2010) rather than logics of preparedness in the face of future, and unavoidable, catastrophes (Lavell and Maskrey, 2014: 275). At the present moment, DRM Assemblages as actual entities exist, at best, as under-resourced, ineffective and de-territorialised assemblages of actors and policies which are largely unrelated to these dominant and everyday codes of development and government (GNDR, 2018; Jones et al., 2015).

To illustrate how the DRM Assemblage, as an analytical tool, can help to analyse disasters-in-the-making and the way in which the emergence of them interacts with DRM Assemblages – or apparatuses – in-place, we consider the example of the Soufriere Hills Volcano in Montserrat. In 1995, the Soufriere Hills Volcano on Montserrat – a UK Overseas Territory in the Caribbean – began to erupt for the first time in recorded history (Young et al., 1998). Over the following 15 years, it destroyed the capital city (Plymouth) and forced two-thirds of the population off the island (Clay et al., 1999). In 1988, a scientific paper had forecast the eruption as one of the several potential futures (Wadge and Isaacs, 1988), but this was not incorporated into planning by those managing the DRM Assemblage of Montserrat – not least because it was a scientific paper – and after Hurricane Hugo in 1989, development on the island continued to focus on Plymouth. Montserrat’s drive towards economic independence through a strong tourist and music industry, a strong sense of identity around Plymouth, the lack of uptake of scientific knowledge, the challenges of colonial governance and then the earthly forces from the volcano combined into a disaster whose origins can be traced in many different threads of the island’s socio-geological history. There were scientific hints before the eruption that this was a ‘disaster-in-the-making’. However, the intersections between international scientific institutions, the institutions and identities of Montserrat and the UK government, and the vulnerabilities of a colonially administered population prior to the eruption were not territorialised to the extent that decisions to rebuild Plymouth on the slopes of the volcano could be challenged. This demonstrates the importance of transdisciplinary understanding of risk in relation to identity, culture and knowledges: Transformation requires engagement with imagination and experience-in-places.

VII The DRM Assemblage as Method: Understanding and Mitigating Emergent Disasters-in-the-Making

To operationalise the six themes of the DRM Assemblage – listed earlier and taken from Donovan (2017) – in analysis, we propose four methodological principles of assemblage-inspired research into disasters. We then propose some possible research methods which one drawing upon the DRM Assemblage as an analytical framework might use in practice.

VIII Doing Research With the DRM Assemblage

The DRM Assemblage does not demand the use of specific research methods, though we suggest the broad category of ‘more-than-human methodologies’ may work well (Dowling et al., 2016). These methodologies seek to unsettle established research methods by questioning and reconceptualising what it means to do research (Barker and Pickerill, 2019). This process of unsettling – of new ways of doing – may relate to working both within and against power relations that may be colonial, oppressive and/or discriminatory (Sultana, 2020). In this vein, Grove and Pugh (2015) reflect on how the imagined futures of DRR initiatives may, through their materialisation, reterritorialise problematic relations of power if they are coded by what they describe as modernist interpretations of resilience; where the research or programme may be designed to create self-sufficient, empowered citizens in the face of vulnerability generating processes over which these citizens have increasingly little control (Joseph, 2013; Mills-Novoa et al., 2020; Rogers, 2015). In response, they propose that the researcher reflects on ‘becoming resource’. This involves acknowledging that participation itself may not necessarily empower citizens, and also examining how, through reimagining researcher-researched relations and doing research differently, participation can empower both researchers and research subjects to challenge, recode and territorialise alternatives to uneven and unjust power relations (Grove and Pugh, 2015: 10).

Doing disaster research differently could involve synthesising, and in many cases prioritising, local, traditional or indigenous knowledge (Kelman et al., 2012). In many cases, like AT, such knowledges are relational and dissolve hierarchies and essences between humans and non-humans, societies and environments and so forth (Rai and Khawas, 2019). The potential of AT to act as a boundary object between scientific and/or Western understandings of disaster risk and other place-specific, alternative knowledges is a fruitful avenue for further DRM Assemblage research (Mercer et al., 2007).

Another way of doing DRM Assemblage research relates to ensuring the stories and voices that speak through the research are those which understand place-specific disaster risk best: the ones who live with that risk every day (Delica-Willison and Gaillard, 2012; Moezzi and Peek, 2019). One well-trodden methodological path which helps researchers to bring out these at-risk voices is the interview. Dowling et al. (2015) explore how more-than-human geographers have ‘enriched the interview’ through combining these with go-along tours, photography and video work (Dowling et al., 2017) and how researchers supplement interview data with data gathered through social media, ethnographic diaries and observing relations between people and things. Other work has been done to enrich the analysis of texts and place this analysis in the context of more-than-human worlds (Doel, 2016; Nimmo, 2011). Such analysis can be particularly useful to historicise the emergence of disaster risk (Adamson et al., 2018; Walshe et al., 2020). Research that uses these types of methods, and which embodies the four methodological principles of the DRM Assemblage outlined above, can contribute to the imagination of alternative, less vulnerable and more resilient, futures. This process of imagination may territorialise new power relations – between researchers, decision makers and participants – in transformative ways.

IX Conclusions

As the Sendai Framework has emphasised, the role of science – including social science – in DRR is and should be increasing. This is not simple. Experts bring different types of knowledge and resources into a risk reduction context, and they interact with stakeholders in a wide range of ways, through reports, warning systems, research projects and assessments that can have significant impacts across a variety of institutions. Placing hazard and risk assessment at the heart of development is critically important in reducing disaster risk but needs to be done sensitively and with awareness of the power dynamics and diverse ontologies that are inherent in any local/national context – factors which we have sought to argue that research into DRM Assemblages is well placed to unpick and potentially reconfigure. This article has shown that in AT, disasters are understood as neither socially constructed nor naturally occurring; rather, they are seen as possible disasters-in-the-making, materialising through the uneven relationships between more-than-human phenomena – be they the uneven relations between geophysical forces which trigger landslides or the uneven power relations between city planners and slum residents. By focusing on the six components of the DRM Assemblage, research using the framework is directed to understanding, critiquing and potentially challenging the ways in which diverse techniques and technologies of DRM attempt to manage uneven relationships of a more-than-human life (Anderson, 2012; Donovan, 2017). A focus on ‘root causes’ is replaced by a focus on how place-specific political, scientific, economic and social imaginations become dominant futures-in-the-making and how these imagined futures interact with uncertain more-than-human hazards to lead to the continued territorialisation of inequalities and vulnerabilities in disaster events (Barnett, 2018, 2020; Granjou et al., 2017; Grove, 2014b). The way in which these vulnerabilities might be understood and addressed is explored not only through transdisciplinary hazard assessments and radical disaster studies (Gaillard, 2019) but also through the literature on feminist ethics of care and sustainability (de La Bellacasa, 2017; Kinkaid, 2019). It is this latter aspect of the praxis of the DRM Assemblage where future work should focus and where the radical potential of the approach will be witnessed.