RESPONSES OF POPULATIONS FOR COPING WITH AN EXTREME EVENT: SAINT-MARTIN AND SAINT-BARTHÉLEMY IN THE WAKE OF HURRICANE IRMA* RESPOSTAS DA POPULAÇÃO PARA ENFRENTAR EVENTOS EXCEPCIONAIS: O EXEMPLO DAS ILHAS DE SÃO MARTINHO E DE SÃO BARTOLOMEU À PASSAGEM DO FURACÃO IRMA

The intense hurricane season of 2017 in the North Atlantic brought a number of major hurricanes including Hurricane Irma (cat. 5), which landed on the French islands of St. Martin and St. Barthélemy. These two French Overseas Collectivities, separated by a distance of some 20 kilometres, have very different territorial and social characteristics. The study reported in this article is part of a scientific feedback conducted a few weeks after Irma. The social approach to the disaster conducted within this research framework aims to characterise the behaviour and implications of the inhabitants in the experience of the event, placed in a social and territorial context thanks to narrative interviews administered in situ. The comparative approach reveals similar behaviour before and during the hurricane, but individual and collective social and territorial factors guided very distinct post-event reactions.


Introduction
Harvey, Irma, Maria, a few of these successive hurricanes, marked the intense 2017 hurricane season in the North Atlantic, which was devastating and considered "among the nine most 'energetic' seasons since 1851" (Chauvin et al., 2017, p. 1). Hurricane Irma, although it did not cause the greatest number of victims (124 deaths), was the "exceptional" event. It was the first major hurricane (category 5) to land on an island in the Lesser Antilles (Météo-France, 2017). Irma hit the northern islands of Saint-Martin and Saint-Barthélemy, surpassing the reference event, Hurricane Luis (category 4), which hit these islands in 1995. Irma was responsible for 11 deaths on Saint-Martin and the cost of the damage amounted to €2 billion for the two islands. These French overseas territories, which are only 20 kilometres apart and have both common and distinct features, were hit by this highenergy event. The post-disaster phase is regularly the subject of expert appraisal and feedback missions (RETEX), the objectives of which aim to analyse the management of the event. These missions are often the prerogative of expert institutions or organisations, but scientific feedback is less frequent. Furthermore, investigations into the understanding of the phenomenon and the way in which the crisis was managed most often focus on the point of view of institutional or private players involved in collective management (local authority, emergency services, network managers, etc.). But few studies are dedicated to the involvement of the 'great forgotten' population (Fenet and Daudé, 2020, p. 1), which is the approach proposed in this article. Indeed, less than two months after the passage of Hurricane Irma, a multidisciplinary scientific RETEX was initiated, part of which dealt with the experiences and involvement of the populations of Saint-Martin and Saint-Barthelemy. We believe that the populations that have suffered from the phenomenon also represent key players in crisis management. An analysis of the experience of the event and therefore of the behaviour of the population in the face of the event provides information on their involvement in crisis management (from preparation to post-crisis), which contributes to individual and, more widely, territorial recovery. Behaviour in the face of risk and a crisis depends on numerous factors such as "the characteristics of individuals, social processes and cognitive mechanisms" (Etner, 2011, p. 17). They are also linked to "hostile environmental pressure" (Provitolo, 2004, p. 15). From interviews that were semidirective and based on the temporality of the event, the role and involvement of the population in crisis management (from preparation to post-crisis) had to be determined in the context of an intense climatic event. Beyond the analysis of reactions, the comparison of two territories helps to explore the contribution of territorial factors (geographical context, politics, attachment to the territory, etc.) in behaviours. Taking an interest in the territory proves to be a determining factor in the characterisation of territorial vulnerabilities, which define the response capacities of the territory and its populations. The presentation of territories and of hurricane Irma which hit the two islands in September 2017 will provide a portrait of territorial vulnerabilities. The hurricane is the starting point for a scientific experience feedback, so we will then present the methodology for collecting testimonies. Finally, the last part, an analytical description of the results, will show the similarities and differences in the involvement of the populations of the two islands.

French west indies, tourist lands and lands at risk
The territorial context as a vector of vulnerability Saint-Martin and Saint-Barthelemy, brought back into the forefront after the passage of hurricane Irma, are still perceived as exceptional tourist destinations. While these sister islands, "so close and so different" (Hartog and Chardon, 1996) share a common destiny, some of their distinct social and territorial characteristics help to explain crisis management preparedness and post-disaster territorial recovery. The characterisation of territories based on development indicators provides information on the vulnerability of the territory (D'Ercole, 2003).   IGN (1969IGN ( , 1984IGN ( , 1999IGN ( , 2017, Gherardi, 2020, 2021). de urbanização em São Martinho desde1969 (Fonte: RGE alti 2016, BD Topo IGN 2017, aerial photo IGN (1969, 1984, 1999, 2017, Gherardi, 2020, 2021).  (1969,1984,1999,2017), Gherardi, 2020, 2021 (1969,1984,1999,2017), Gherardi, 2020, 2021 (Sanguin, , 1982 on the northern islands already envisaged the tourist future of Saint-Martin and Saint-Barthelemy, a trend which is emerging on the scale of the Caribbean basin (Urunuela, 2004). The changes have been felt since the 1970s when the last agricultural activities were in decline.

Fig. 1 -Uso do solo e fases
At the beginning of the 1980s, even before the law of tax exemption, these islands had already started a "tourist and residential revolution" (Sanguin, , p. 65, 1982 based on an upper-class, American-dominated clientele, rejecting mass tourism and aiming for luxury tourism (Magnan, 2008;Theng, 2014). Saint-Barthelemy operates selective tourism facilitated by transport infrastructures which are insufficiently sized to accommodate boats and large aircraft, but which has the disadvantage of creating a partial dependence with the international airports of Juliana (Saint-Martin) and Pointe-à-Pitre (Guadeloupe). This same dependence falls to Saint-Martin which is moreover in competition with Sint-Maarten to capture tourist flows. The construction and real estate sector derives from tourism and provides a large number of jobs. The impact of Irma on these sectors of activity has followed a similar logic in the two islands but with different temporalities (Defossez et al., submitted). The curve of tourism has fallen while that of construction and public works has increased the time of the (structural) reconstruction, which is almost completed  identity (Sanguin, 1982;Redon, 2006). Economically, the island is considered to be dominated by the metropolitans (Duvat, 2008). Saint-Barthelemy has also attracted a significant proportion of metropolitans described as "neo-St. Barts" (Chardon and Hartog, 1995: 275) and West Indians. Nearly 40 nationalities (mainly European) make up the foreign population (15% of the island's population in 2011), of which more than half are Portuguese (IEDOM, 2015), settled by construction and public works activities (Chardon and Hartog, 1995). Very low atmospheric pressure reached up to 916hPa, with maximum sustained winds recorded at around 290 km/h (Cangialosi et al., 2018) and modelled at over 350 km/h (Cécé et al., 2020). Cyclonic surges reached 2 and 3 metres accompanied by waves of 3 to 10 metres, with maximum heights mainly affecting rocky and urbanised coasts already eroding, accentuated by the anthropisation of the coastline (Pillet et al., 2019;Rey et al., 2019). Considerable structural damage has been estimated at more two billion euros, and 11 deaths add to the trauma for Saint-Martin.

Interviewing populations identified as playing a role in crisis management
The feedback widely used after a disaster is destined to learn, to draw lessons from crisis management in order to improve strategies in the face of a future event. The process of analysing the cinematics of the event and the management of the crisis focuses more on the institutional and managerial players, and less on the point of view of the populations (Goutx, 2012) who are affected by the event but who also participate in its management at their own level (Campbell, 2009;Weiss et al., 2010). Determining the knowledge, risk awareness and involvement of the population in the complete crisis management process (from anticipation to recovery) has guided this approach based on semi-directive interviews set within a temporal framework (before -during -after), with reactions depending on the temporal phases of the disaster (Provitolo et al., 2015). In our view, the place of the population in crisis management processes is based on a necessary involvement that is partly independent of institutional management. It is also based on psychosocial and territorial determinants. However, Although it is understood that post-disaster reconstruction is a long-term process, the primary objective was to ask people about their short-term experiences, from a few days before the hurricane (as soon as its direction The interview, which was introduced by a general phrase "tell us how you experienced the hurricane, before, during and after", was very well received, with a return rate approaching 100%. In Saint-Martin, no prior contact was made, so the presence of the interviewers in the island's daily life was enough to build trust and encourage people to give themselves up. In Saint-Barthelemy, the same approach was taken, in addition to the publication of an advertisement in the Saint-Barth newspaper offering volunteers the opportunity to share their experience with us.
While each testimonial is unique and depends on the intrinsic nature of the individual, the behaviour and experience show similarities, particularly in the phases preceding the impact. Disparities, and even inequalities, are even more obvious in the emergency and recovery phases, both within a given territory and, above all, between the two islands.
Individual and collective behaviours explained by territorial vulnerabilities and capacities but also by the determinants of knowledge and risk awareness. before Irma passed through, two safe shelters were closed because they were too exposed to sea submersion... fortunately, they were largely damaged or even partially destroyed (photo 1).
Finally, the reactions in the phase preceding the hurricane remained fairly similar between the two territories, even if the Saint Barts seemed more involved in preparing for the event (fig. 6). Even though the populations had prepared themselves on the whole, the intensity of the event exceeded expectations. The risk may have been underestimated, particularly with the reference event (which was either really less intense or was forgotten), but also due to the lack of experience, and finally due to the performances (Weiss et al., 2010). In this phase, although individual mechanisms prevail, they are also based on collective considerations, particularly with the dissemination of information on the approach to the phenomenon and the associated safety instructions.

To cope with an intense event: confine yourself
The violet weather vigilance "confine yourself" led the     (moving, job loss, returning to work, etc.) had a good "theoretical" knowledge of the hurricane, and after an initial phase of strong winds, the calm weather gave the interviewees the opportunity to go outside for a few moments to reinforce the protection of the openings, but above all it allowed them to take shelter elsewhere or to take in neighbours in danger (some ground floors took on water while the top floors had no roof). However, some of those interviewed seem to have perceived these different phases less, like the residents of the "Grand-Case" coastline, for whom the eye of the hurricane (period of calm) was not felt. The hurricane, which was perceived as very intense, gave an impression of violence and a combination of phenomena.
Some people spoke of the shaking of the house similar to an earthquake, while others heard the sea packets comparing them to a tsunami. In spite of the frightening experience, all feel lucky to be alive after such an intense phenomenon. Moreover, the death toll (11 officials) convinced almost none of them and rumours multiplied. As communication was impossible, the end of the phenomenon with the "grey" vigilance could not be announced. The populations relied on the end of the wind and the outside noises that emerged.

Emergency management reveals individual and collective, citizen and institutional capacities
The impact gives way to desolation, to observation, and it was necessary to clean up the area, particularly the roads, and so everyone began this clearing phase spontaneously, but in an organised manner. The Saint-Barthelemy "Collectivité" (local institution) had prepositioned machines and vehicles that would be used for this operation. Unanimously welcomed and thanked, the Portuguese building and public works companies played a key role, as they immediately and constantly became involved with everyone. Waste collection areas were made accessible free of charge. For some, 24 to 48 hours afterwards it was possible to circulate on part of the island. The other major concern was to communicate with relatives and neighbours. All of them took care to check whether their closest neighbours were now safe, and then waited anxiously for the possibility of communicating with the outside world. This opportunity presented itself 1 to 2 days later, with the installation by the local authority of Irma WIFI terminals, installed towards the airport. The management of the emergency phase by the local authority was unanimously appreciated, even though the president of the local authority (collectivité) announced the following day that he did not need the Some of them also took part in necessity robberies in shops that had already been vandalised, making necessity goods "available". But the insecurity seemed to be growing and violent. It should be remembered that on this territory, precarious and high standing neighbourhoods are found side by side. Also in some neighbourhoods where the organisation into gated communities is frequent, the inhabitants have organised themselves into a "militia" to ward off possible robberies with violence.
Sometimes it is neighbours or acquaintances who are identified as looters, leaving a feeling of mistrust. In spite of everything, the population has organised itself to start clearing the gates but has not been able to count on institutional help. The community was considered to be absent without having anticipated this emergency phase and the interviewees felt that the state had intervened too late without, for example, the pre-positioning of law enforcement and rescue forces. The imminent arrival of Hurricane José, announced as soon as Hurricane Irma was over, contributed in part to the impossibility of action.
These factors, external to the organisation of the postcrisis period, gave rise to a feeling of abandonment, often evoked in the event of a major event (Dourlens, 2003).
They also reveal, according to the interviewees, a lack of anticipation and organisation of institutions. Supplies were set up by institutions or NGOs (Non Governemental Organization), which were considered to be disorganised or sometimes totally unsuitable (canned food to be cooked when the electricity networks were out of order, no information on the places and times of distribution).
Faced with this, some of the interviewees have adapted by recovering, for example, wood from damaged buildings as cooking wood for food. But these substitutes were worthwhile in the short term. Fortunately, the island has benefited from a great deal of organised solidarity thanks to local associations (distribution of food, clothing, etc.) and reinforcements sent to repair the networks (electricity for example).
In both islands, once the emergency phase is over, the time has come to rebuild, renovate and repair housing, at their own expense or with the help of insurance companies.
The insurance penetration rate is lower than the national values, with 2/3 of insured properties on Saint-Barthelemy and half on Saint-Martin. There are several explanations, the main one for Saint-Martin refers to the last major hurricane (Luis) and the low compensation payments that followed, but also to fragile financial situations that do not allow a share of the budget to be allocated to insurance.
The recourse to insurance does not always preserve the possibility of finding decent housing as soon as possible, which is part of the population's expectations. The time taken to process applications was considered to be very uneven: for some very fast and for others the wait was several months later A last common element is the trauma and life changes observed by the interviewees (fig. 3), identified in other cases of natural disasters (David et al., 1996).
The traumas took place over several months, or even years in some cases. The setting up of a psychological unit immediately after Irma was not very successful, as the people interviewed still felt that they were in the thick of it. Speaking out is still perceived as a difficulty, a challenge that not everyone is ready to take on. It is sometimes the trigger or catalyst for situations that are in the process of change with positive or negative life changes (divorce, marriage, pregnancy, professional reconversion, moving house, etc.). These situations are discussed in a personal capacity, while the subject of voluntary departures (temporary or definitive) in the days following the hurricane is addressed by the interviewees in a collective dimension. As soon as possible, the two islands organised the departures of women and children who wished to leave the island by air. A feeling of frustration and injustice was born from these departures. First of all among those who wanted to leave the island with a lack of understanding of the organisation in the order of departure. Then, among the people who stayed, because they had no place to take refuge or simply because they did not want to leave the territory to which they were attached and/or native.
The impression of being abandoned was expressed, and very quickly replaced, at least in the speeches, by an opportunity to "sort out" the inhabitants, leaving only those attached to the island.

Long term recovery
Irma's impact has more broadly questioned crisis perspective what a long reconstruction is. On Saint-Martin, the populations interviewed are tired of the time taken to restore the buildings, the lack of priority in the reconstruction (for example, schools which are not in the priorities) and also compare with Saint-Barthelemy, for whom everything goes faster. The comparison on both sides is based on economic factors, but also, and above all, political factors, with a greater willingness and stronger investment on the part of the Saint-Barthelemy community and the population.

Discussion/Conclusion
The crisis situation, managed at the level of institutions, relies de facto on the populations whose involvement is essential (Weiss et al., 2010) in the different temporalities before, during and after. Individual reactions are based on psycho-sociological determinants, depending on experience and knowledge and awareness of the risk, but also on personal capacities (Sattler et al., 2000;Kouabénan, 2007;Etner, 2011). However, they cannot be totally disconnected from institutional collective management, particularly in the post-crisis phase. While the population is in control of the situation in terms of anticipation (protection of its housing, securing the surroundings of the dwelling, etc.), it is largely up to the institutions after the impact. However, as Saint-Barthelemy shows, collective and individual mechanisms can go in the same direction. The root causes, territorial vulnerabilities influence these mechanisms (Duvat et al., 2021). External factors also disrupt preparation and recovery (Defossez and Gherardi, 2020), which are partly dependent on root causes. Organised robberies in Saint-Martin took place in a deteriorated social context, and then state intervention either ruled out the community's strategies or, on the contrary, allowed the situation to be brought under better control. Beyond crisis management, Hurricane Irma, like any crisis, exacerbated existing problems (Leone and Vinet, 2006) at all levels. At the individual level, with changes in life that were previously envisaged without ever coming to fruition. At the collective level, on the population side, the paradox of solidarity/robbery has degraded social relations; on the institutional side, dysfunctions have been highlighted. At the territorial level (island), socio-economic gaps have become more pronounced, especially on Saint-Martin, which has more marked vulnerabilities than on Saint-Barthelemy. Moreover, the disaster did not only raise questions about the crisis and crisis management, but more generally about risk prevention, with the views measured between Saint-Martin and Saint-Barthelemy explained in part by the political context. On the one hand the will to rebuild on the coastal strip the economy which depends almost exclusively on seaside tourism and on the other hand the necessity to stop the urbanisation of the coastal strips in order not to expose oneself even more to risks.
These dissonances give rise to tensions and contradictory decisions, "individuals can change their behaviour […] and decide, for example, to stop investing in prevention as they consider that it has not been sufficiently effective" (Etner, 2011, p. 25).
Faced with a major "predictable" phenomenon such as cyclones, populations find their place in crisis management preparedness. The collective and individual scales tend to converge, overlap and sometimes contradict each other. The involvement of populations is based on numerous factors of vulnerability and intrinsic capacities, but also on the nature, intensity and scale of the natural phenomenon and on collective (mainly institutional) strategies. What people learn from such an event depends on the actors and scales of decisionmaking and the social commitment to deal with it.