Creative practice as a potential tool to build drought and flood resilience in the Global South

Global South communities are increasingly exposed and vulnerable to natural hazards such as floods and droughts. Preparing for future extremes requires including diverse knowledges, elevating under-represented voices, thinking out of the box for possible solutions, enhancing communication between diverse groups, and instigating organisational and behavioural change. In this study we investigate whether artistic and creative processes could support Global South communities in improving their preparedness to extremes. A literature review of 267 journal articles published between 2000 and 2018 showed that 5 there is a growing body of research on using creative practice in environmental issues. Art and creativity are for example often used for raising awareness of climate change and for encouraging behavioural change in relation to health issues. Research using creative practice to increase resilience to natural hazards, such as earthquakes, volcanic eruptions, tsunamis, floods and droughts is, however, still very rare. An opportunity exists to better understand the application of the wide range of creative methods currently used for environmental and health-related issues also to enhance resilience to droughts and floods. We tested 10 this in a pilot project in South Africa, in which we designed storytelling workshops to create community narratives about impacts of and preparedness for future drought. These narratives were filmed and edited and shared both with the community and governance actors. Although this was a short pilot project and the effects have not been thoroughly researched, we see potential for these kind of methodologies. We noticed that the approach allowed people to imagine future droughts and that it opened up conversation about potential adaptation measures. Based on the literature and pilot study we call for more research 15 on the use of creative practice in building resilience to extreme events. It is especially important to investigate how the use of creative methods compares to other methods, and how effective creative practice is to make a difference, either in people’s own behaviour or in communicating people’s needs to decision makers.

, which is consistent with reviews focusing on art and climate change (Galafassi et al., 2018) and art and vulnerable populations (Coemans and Hannes, 2017). Most of the papers focus on topics 125 related to health (21% of total no. of papers) and climate change (18%); just over 15% discuss the use of creative practice in topics of environmental management and resource access. Papers on using creative practice in relation to natural hazards and disasters (volcanic eruptions, earthquakes, tsunamis, drought, and flooding) only make up 1-3% of the total sample each.
Recent review articles on the use of different art-based methods in environmental and health research found that most artbased research is carried out in the United States, Canada, the United Kingdom and France (Nurmis, 2016;Galafassi et al., 130 2018;Coemans and Hannes, 2017). Research on arts and health is often also done in Africa (Teti et al., 2018) and water-related creative practice research is concentrated in water-scarce regions in Africa and Australia (Fantini, 2017). In this study, we searched for papers focusing on Global South and vulnerable communities. The resulting selection was spread across the globe (Fig. 2). Most research is done in Africa (29% of total no. of papers), followed by Oceania (25%) and Asia (16%). There is also a lot of work with indigenous communities and vulnerable groups in North America (16%), but only a small amount in 135 Europe (3%).  The creative methods and art forms used are very diverse (Fig. 3). Photography is the most-used method (mentioned 63 times), followed by music & song (59 times), other forms of visual art (48 times), drama (46 times), storytelling (43 times), and video & film (37 times). Art forms such as dance and poetry are used less (17 and 12 times, respectively). One reason for the large amount of papers on photography is that it is often used as a participatory research methodology. Photovoice (or 140 Photo-Elicitation Methods or Camera-User-Study) is often used to explore communities' view on HIV / AIDs (e.g. Jacobs and Harley, 2008;Mitchell et al., 2005;Umurungi et al., 2008;Wood, 2012;Fournier et al., 2014;Adegoke and Steyn, 2017), environmental issues (e.g. Belcher and Roberts, 2012;Bennett and Dearden, 2013), conservation (e.g. Beh et al., 2013), water use and governance (e.g. Fantini, 2017;Bisung et al., 2015), and hazards and disasters (e.g. Yoshihama and Yunomae, 2018; Schumann et al., 2018). This means that photography is mostly used as a research method asking participants to develop new 145 material. Papers on music and song, on the other hand, mostly study existing traditional songs and music on a variety of topics (e.g. Stone, 2003;Saroli, 2005;Wu, 2016;Grant, 2018;Dirksen, 2019 We also looked at the degree of co-creation between the researchers and communities, because we were interested to see how much the researchers were involved in the creative practice, including initiating, supporting, guiding or even leading 150 the creative practice, and how much of the creative practice was pre-existing in the community prior to the research or was completely carried out by the community without researcher involvement. The degree of co-creation is very variable between 6 https://doi.org/10.5194/gc-2020-11 Preprint. Discussion started: 11 March 2020 c Author(s) 2020. CC BY 4.0 License.  papers. In some cases, the methodology is given to a community by researchers (for example in participatory photography), but the resulting product was made without the researcher's involvement (e.g. Belcher and Roberts, 2012). In other cases, the participants communicated to the researchers what the greatest environmental threats to their community are, and then 155 artists develop this into an artistic product (e.g. Steiner, 2015). There are also many examples of researchers observing and documenting creative practice traditionally used in Global South communities to pass down traditional knowledge on natural hazards or environmental issues, for example using storytelling (e.g. Swanson, 2008;Cashman and Cronin, 2008;Johnson and Beamer, 2013;Troll et al., 2015;Fepuleai et al., 2016). These have a low degree of co-production, because they work with existing material, often in a form of participant observation. Others build on traditional creative practice to develop new stories 160 (e.g. Fuertes, 2012;Somerville, 2014;Ayala et al., 2016;Fernández-Llamazares and Cabeza, 2017). These often have a high degree of co-production with the researcher guiding the process.
The aim of the art-based research also strongly affects the degree of co-production. In many papers, the goal of developing an artistic product is communication, for example to raise awareness of an environmental issue and its impact on vulnerable communities. There is a wealth of projects aiming at raising awareness of the general public on climate change and its impacts 165 (Nurmis, 2016;Galafassi et al., 2018). Papers on the topic of climate change generally have a lower degree of co-production (62% medium or high). In other research, the goal of the creative practice is instigating some kind of action, for example to  Table 1.
bring about behavioural change in relation to health (for example HIV / AIDS). Papers on the topic of health generally have a higher degree of co-production (90% medium or high).
Only rarely is the effect of the creative practice evaluated. Fantini (2017) highlights that creative participatory methods such However, a comparison with a control case, in which other, non-creative methods were used or in which no activity was done at all, is hardly ever done.

175
In our search, we found only two papers focusing on drought and/or water scarcity (Table 1). Barontini et al. (2017) used arts as a communication tool. They documented traditional irrigation techniques to cope with water scarcity in the Mediterranean and, together with other researchers and students, developed a travelling exhibition for students and the general public. In this example, the exhibition was used to raise awareness and understanding of historical water conservation techniques. In contrast, Rigby et al. (2011) used arts as a tool to change behaviour and coping capacity. They investigated the use of creative and artistic 180 practice in response to drought and discuss how encouraging Aboriginal arts in Australia can increase resilience to drought as it enhances the connection with the land. They mention a whole suite of art forms used traditionally to highlight the Aborigines connection to land, e.g. painting, printing, photography, film, theatre, music and dance. This research did not develop new artistic products or ask participants to engage in new forms of creative practice. Instead, they studied how traditional art can help people cope with drought by reconnecting them to the land.

185
Five papers focusing on flooding were identified (Table 1). They ranged from studies on using community workshops to merge scientific and local knowledge of flooding (Ikeda et al., 2016) to a researcher/poet developing poems from interview transcripts of older people's lived experience of flooding (Miller and Brockie, 2015). Three studies used participatory creative methods: participatory theatre to help young people cope with mental health issues related to flooding post-disaster (Contr-  We have classified these seven papers on drought and flooding (Table 1) into in three categories, following the three dimensions introduced in Sect. 1: i) the goal of the creative practice (raising awareness, instigating action, or both), ii) the doer of the creative practice / the creator of the end product (completely led by the community, an academic or artist, or co-created between community and academics / artists), and iii) the audience of the creative practice / the end product (the participants 195 themselves, other community members, decision makers, the general public, or researchers). Figure 4 shows that the papers generally fall into two categories. Firstly, those discussing creative practice aimed at communicating the impacts of drought or 9 https://doi.org/10.5194/gc-2020-11 Preprint. Discussion started: 11 March 2020 c Author(s) 2020. CC BY 4.0 License.
Compared to the wider literature on environmental issues and health described above, these studies on floods and droughts 200 show limited co-creation ( Fig. 4), leaving an opportunity to explore this aspect further. Additionally, despite the mention of how the artistic products could travel to other communities and could be used to inform decision making, this has not yet been investigated in these studies (Fig. 4). This shows that there is a gap in the use of co-created creative practice both for awareness raising / communication and for instigating action, especially with a broader audience. In the next sections, we show an example case study of how this gap might be filled.

Pilot study
The pilot project CreativeDrought (https://creativedrought.wordpress.com/) aimed to develop approaches to local level preparedness to future drought. According to Biggs et al. (2012), important aspects of resilience building are maintaining diversity and redundancy, managing connectivity, managing slow variables and feedbacks, fostering complex adaptive systems thinking, encouraging learning and experimentation, broadening participation, and promoting polycentric governance systems. In this 210 project, we focused on participation, learning and experimentation, and connectivity, and wanted to explore how creative practice could support these. We designed an approach that allowed members of the community to actively engage with potential futures including sharing of existing local knowledge, experience and strategies ('imagining futures'; Anderson, 2010), and inclusion of scientific information ('calculating futures'; Anderson, 2010). The goal was to create stories about potential future drought impact and preparation/adaptation via a multiple method approach developing text-based narratives, performance, and 215 visuals (video). We used this process to increase dialogue between groups in the community and used the visuals to make the voices of the community members 'travel' to policymakers' circles, where they would not normally be routinely heard (see CS in Table 1 and Fig. 4).

Methods
As case study region we selected a village in Limpopo Province in South Africa. This (anonymous) village was chosen because 220 of its vulnerability to drought and because the villagers get their water supply from a number of different sources and use it for a range of different purposes, i.e. a groundwater borehole for domestic water supply, two reservoirs for irrigation of agricultural land, and a river for washing, bathing, brickmaking, etc. (Rangecroft et al., 2018). The village has a population of c. 2800 (StatsSA, 2017) and is led by a chief and his royal council. Drinking water supply is organised by the Department of Water and Sanitation and irrigation water is regulated by the Department of Agriculture (Makaya et al., in review). 225 We developed and tested an interdisciplinary approach with, as final output, videos of narratives produced by groups of community members in the village (Rohse et al., in prep.). These were produced in small-scale workshops guided by an interdisciplinary team of researchers and local research assistants, in which the participants were asked to reflect imaginatively on future drought scenarios produced by a hydrological model. Focusing on narratives as the creative method was a pragmatic choice related to participants' preference for storytelling compared to other (more visual) methods, for example maps or games, and familiarity with these methods by the research team. Additionally, because we focused on imagining of future events, we could not use methods that are rooted in the current or past situation, such as Photovoice.
The development of the narrative videos happened in three phases. In the first phase, we carried out group narrative interviews on the topic of past drought events (Rohse et al., in prep.). These allowed us to collect rich and contextualised information on past droughts and their impacts on different groups in the community, but also to surface hidden voices and explore the 235 potential for future drought narrative workshops. We used oral history techniques in small group conversations. We did 12 groups interviews with 2 to 7 participants each covering a range of different groups within the village (occupation, gender, age). The interviews were recorded, transcribed, and translated from Venda (the local language) into English.
In the second phase, we set up a hydrological model for the area that could be used to model future scenarios (see Rangecroft et al., 2018). We used the SHETRAN hydrological model developed from available datasets and catchment observation.

240
Topography was based on DEM data, precipitation and potential evapotranspiration input came from climate datasets, geology, soil type, land use, and information on location and amount of water abstraction was derived from field observations, dam level and release data, and discussions with knowledgeable locals. The model was qualitatively validated with water level data and with information about the drought events shared by the participants of the narrative interviews in phase 1. The model was run for a baseline run and for three different scenarios (Rangecroft et al., 2018). The scenarios were designed based on conversa-245 tions with the villagers and with a number of government representatives. The "Warmer Temperatures" scenario was based on an increase of 3C compared to present day temperatures representing a climate change expectations for the region. The "Larger Irrigation Scheme" scenario represents a possible increase in water demand in the future with the area of the irrigation scheme expanded to be twice as large as present day. The "No Dams" scenario was based on the expectation that without maintenance the dams, which were built in the 1960s, might not be available for the community any more in the future. On the model results

250
for the scenarios a drought analysis was performed and the resulting drought characteristics were compared with the baseline scenario to determine the difference between future and historic droughts (Rangecroft et al., 2018). These results were then translated into storylines for easy communication with the community.
In the third phase, we organised 6 workshops, in which we brought together different groups in the community. Our aim was to have around 4 people per group, so 8 per workshop, but due to various reasons attendance was low for some groups. These In these workshops the participants first listened to the storylines of the possible future droughts. They then discussed what the impacts of these droughts would be on them and their community and developed these into 'stories' that were filmed. Next, the groups within each workshop exchanged these stories and had a discussion with the whole group about possible responses and preparedness measures. They then went back to their smaller groups to develop this into a story about coping strategies to future drought, which was also filmed. This resulted in two stories about the future for each group, one on future drought 265 impacts and one on future drought preparedness (which took into account the exchange with the other group).
After the narrative videos were produced, the results of the workshops were discussed in a community forum, shared with community via YouTube, USBs and transcripts, and used in conversations with water management actors.

Results
The first phase provided very useful information in preparation for the second and third phases. It gave understanding of historic imagine and talk about the future. We also found that there was seemingly little intergenerational exchange of drought coping strategies, although there were some ambiguous testimonies on this point. And importantly, we tested which creative method would resonate most with the community. We talked about visual methods like artistic maps or other methods like board games 280 for the community to interact with potential future changes in water availability and use. However, from the start it was clear that the idea of 'stories' was most resonant with the community. The participants of phase 1 and the village elders of the royal council indicated their interest in developing stories.
In phase 2, the information collected in phase 1 and during catchment observation was used to set up and test the model.
Because the model was used for the development of scenarios, which were then used as starting point for discussion in the 285 workshops, accuracy of model results was not our main aim. We wanted the model to represent the past droughts relatively well so that we could trust the modelled potential futures. From phase 1, we found that farmers and elderly men and women talked mostly about the 1983 drought, whereas the young people, married mothers and civic group discussed the impacts of the 1992 and 1994-95 droughts on the community (Rangecroft et al., 2018).
The model was then used to extrapolate and calculate several scenarios that were mentioned by community members and 290 government representatives. Instead of predicting the future, we were exploring plausible futures. Droughts were calculated and compared between the scenario and the baseline. These were transformed into storylines including information on the duration and severity of future droughts compared to previous experience (e.g. more severe than has been experienced in the past 40 years or twice as long as the drought in the early 1980s). We used both a climate change scenario and scenarios related to human activities (i.e. increased water use for irrigation and decreased water availability due to lack of dam maintenance), but found that the latter were more difficult to communicate so we ended up mostly using the climate change scenario in the workshops.
The workshops generated very rich information on potential future drought impacts and possible coping and preparedness strategies. These did not necessarily develop into fully-formed stories, but they did clearly communicate strong emotions and allowed for imagining positive options. Compared to the first phase, in which narrative-style answers were also used but where 300 people found it hard to engage with uncertain futures and referred mostly to God or their own death, in the workshops the narrative approach supported by data from the model scenarios allowed participants to use their imagination and exchange ideas. Some participants now made the link between water use for irrigation and water available for domestic purposes, which did not happen in the phase 1 group interviews (Rohse et al., in prep.). We found that the intergenerational exchange was very powerful, with older farmers willing to share traditional techniques and younger farmers eager to learn. The exchange 305 between participants with different professions also worked well, although there was already an awareness of the needs of different groups in the community, mostly because these are not strictly separated with for example livestock farmers also having a small plot to grow vegetables and being domestic water users as well. We also found that there were very different preparedness and coping strategies brought forward in the different workshops, including individual actions (digging for water in the river bed, selling livestock, saving food), community actions (collective maintenance of the irrigation system, drilling a 310 new borehole), and government support (providing food, fodder, drinking water, and jobs).
The approach allowed participants to use their imagination and consider future drought events, their impacts and preparation, and to exchange ideas between different intergenerational groups and across different professional occupations. The research assistants who carried out the bulk of the facilitation in the local language, reported that whilst some participants were a bit concerned with how long the workshops were, there was general enthusiasm for the topics discussed and participants had many 315 ideas to contribute, and valued the opportunity to have a platform to exchange and learn from their peers, in particular those from the younger generation.
The recorded and edited narratives (Fig. 5) were given back to the community with the idea that these could be used in schools and community gatherings. The narratives were also a useful tool for creating space for conversations with government representatives about local perspectives on drought management. Whilst on the outset the local policymakers we interacted with 320 were uncertain about the videos and their content, through careful facilitation, they acknowledged the value of the videos in sharing the concerns of the community. The videos proved to be stimulating prompts for conversations on drought preparedness and on the current situation facing the community. In addition, the policymakers found some of images useful as illustrations of the specific conditions facing the villagers, as explained it is sometimes difficult to get reliable information on villagers' situations.

4 Reflections & perspectives
Through mapping existing literature on using creative practice to build resilience to droughts and floods by Global South communities, we identified a clear research gap. The seven studies we found either focused on creative practice as research tool, for raising awareness of the general public, or for instigating action by the participants. There was relatively little cocreation and little evidence of how the end products could travel or be used in decision making. The 267 studies with a wider 330 focus than solely drought and flooding selected in our literature mapping exercise, showed a similar pattern with most artbased climate change communication focused on raising awareness with little co-production, and most creative practice in the field of health focused on encouraging behavioural change with high co-production. Our case study on developing narratives to increase preparedness to future drought in southern Africa was most similar to the second type, but we felt that there was potential to explore the middle space between these two contrasting types. By filming and editing the stories developed by the 335 community, we could use them as a discussion starter with decision makers. Unfortunately, due to the short nature of the pilot project funding, we have not been able to evaluate the effectiveness of communicating via stories compared to more traditional ways of communicating. This is a common feature among much of the research on art and creativity in environmental and health-related issues. Most papers are merely describing a methodology and promoting a potential beneficial method without clear evaluation of its efficacy.

340
Both in the reviewed literature and in our case study, we encountered many barriers and practical challenges to using creative practice in resilience-related research in the Global South. These practical challenges include: language: in cases where the researcher does not speak the local language or is not from the area being researched, knowing the full breadth of traditional art-based methods or translating the details of creative practice activities and intended when using other more verbal methodologies, such as interviews or archival research.
time: creative methods often take a lot of time, both from the participants and from the researchers. Researchers need to gain in-depth insights of their case studies and if translators are involved (see language), they need to be well-briefed.
Also, if art-forms are used that the community is not familiar with, a thorough explanation is needed.
unfamiliarity: participants are not always comfortable in engaging in creative practice and some decision makers tend to 350 prefer quantitative outputs with a specified (un)certainty instead of contextualised stories, photographs, songs or other art products (Owens, 2005).
To overcome these challenges, good facilitation is crucial. Ideally, the research team is interdisciplinary and including local researchers and the work is guided and facilitated by creative practitioners / artists speaking the local language and familiar with participatory art-based research (see 'social volcanology';Donovan, 2010). Some scholars have critiqued the use of art-based 355 methods, for example by noting that art distances the problem (Miles, 2010) or by questioning whether art-based methods can actually achieve any change (see Nurmis, 2016). Apocalyptic climate change art, for example, can lead to fatalistic views that are not helpful for instigating action and behavioural change (Nurmis, 2016).
Whether creative practice leads to action improving resilience to droughts and floods in marginalised communities is an unresolved question. It is widely recognised that improvements in flood and drought resilience need to come both from com-360 munities and government as a shared responsibility between public and private partners (Trim, 2004;Amaratunga et al., 2009).
However, Global South communities often face many challenges and barriers for actions include lack of resources (land and financial resources), unemployment, and lack of information (such as drought early-warning) and creative practice on its own is unlikely to be able to solve these. However, if the creative practice is part of a larger programme that integrates local and scientific knowledge and combines community-focused activities with activities aimed at decision makers, it may have greater 365 impact potential. McMillen et al. (2016) showed, based on an example from Hawai'i, that arts-based approaches to community resilience have alternating phases of being more and less important, in relation to socio-ecological shifts over time. We feel that globally we are now living in a time that asks for more creativity in relation to environmental issues and disasters. Traditional methods for natural resource management are either forgotten (Janif et al., 2016) or inadequate in the Anthropocene (McDowell and Hess,370 2012; Kareiva and Fuller, 2016;Head, 2016). Adaptation based on experience might have worked in the past. For example, the process of sharing oral history accounts can contribute to community resilience (Osterhoudt, 2018), but how does it apply to future events? There is evidence that damage from natural hazards decreases after repeated events due to adaptation (for floods: Kreibich et al., 2017). Memory of a previous event has been found to decrease damage in the next event (Viglione et al., 2014), but this memory decreases over time (Lopez et al., 2017). A big question is how to increase resilience if the previous extreme 375 event was very long ago or when future events are outside of previous experience? Alternative ways are needed that require out-of-the-box thinking to imagine how the future might be different from the past and what actions are needed to prepare for this future. Creative methods might be able to enhance this process, especially if they are embedded in or making use of traditional ecological knowledge (e.g. Laidler, 2006).
However, much more research is needed on various aspects of using creative practice to build resilience to future drought 380 and flooding.
-Firstly, we believe that more knowledge on traditional ways of using art and creativity for coping with environmental issues is useful. Indigenous knowledge systems have had to deal with climatic and anthropogenic change (McMillen et al., 2016;Gibson and Gordon, 2018) and might show ways to use art and creativity for natural resource management under changing conditions (e.g. Berkes et al., 2000;Whyte, 2018). An interesting example is Aboriginal use of art for 385 connecting to the land in Australia (Rigby et al., 2011;Zurba and Berkes, 2013).
-Secondly, research is needed on evaluating the three dimensions of creative practice (goal, doer, audience; see Sect. 1).
Is the aim to communicate information or awareness, or to instigate action? And who needs to have more awareness or take action: the participants of the creative practice or others, such as the general public, researchers or decision makers?
It is especially important to evaluate whether the methodology of the creative practice fits the intended outcomes.

390
-Thirdly, we want to encourage more research on co-production during the creative process. How can artists and researchers work together with local communities to ensure mutual learning? Can / should decision makers be included in the creative process and not only be receivers of the end product? We feel that the use of participatory art is currently underexplored. As Rathwell and Armitage (2016) noted: "collaborative art making is a platform for knowledge coproduction, whereby novel ideas or products emerge from different ways of knowing".

395
-Fourthly, more evidence is needed on the effectiveness of creative methods. Only in (mental) health research, where art-based methods are used therapeutically, there is some evidence for their effectiveness (e.g. Stuckey and Nobel, 2010;Van Lith et al., 2013). We call for empirical research comparing creative methods to more traditional methods of doing research, communicating information, or instigating action in relation to droughts and flood. For example, it is as yet unclear whether narrative workshops, like those used by Rangecroft et al. (2018) and Rohse et al. (in prep.), have more 400 effect than traditional workshops, as used by Ikeda et al. (2016). Research is needed on the drivers of any observed change: is the reason for change that the members of a community are coming together and exchanging ideas, or is it the creative practice allowing participants to think outside the box? One of the methodological challenges in evaluating effects and drivers is that change can happen many months after an intervention. Longitudinal studies and being embedded in the communities would be needed (Donovan, 2010).

405
-And finally, we suggest research to investigate how creative practice can be embedded in a holistic strategy for building resilience. Can creative practice support communities in Global South settings to communicate their worries or traditional methods of dealing with environmental issues more effectively? How can art-based non-structural methods be combined with more traditional structural measures to mitigate drought and flooding? Like Whittle et al. (2012), we argue that creative side of resilience cannot be separated from the physical and socio-economic aspects of resilience.

410
Our literature mapping and case study have indicated that there is a potential for artistic and creative methods to be used more for building resilience to drought and flooding. Currently, research on this topic is limited, especially on creative coproduction of the end product and on audiences beyond the participants themselves and the general public. Several advantages of using creative practice are mentioned: it has been suggested to surface hidden voices, communicate issues on a deeper, more emotional level, travel to wider audiences, increase engagement and behavioural change. Potential challenges are language barriers between researcher and participants, time investment of both researchers and participants, and possible unfamiliarity of participants with artistic or creative methods or products.
Based on the literature mapping and pilot study we call for more research on the use of creative practice in building resilience to extreme events. It is especially important to investigate how the use of creative methods compares to other methods, 420 and how effective creative practice is at bringing about change, either in people's behaviour or in measures implemented by decision makers. We also think future research should address the question how to combining creative methods with more traditional scientific methods and decision making. For decision makers a combination of quantitative, qualitative, and creative information might be most successful in supporting marginalised communities in coping with drought and flooding.
Data availability. The literature review database (Sect. 2 and Appendix A) will be made openly available upon publication of the article.

425
Video supplement.  Galafassi et al., 2018). When searching these databases we used a combination of keywords describing different sectors, different art and creative forms, specific regions in the Global South, and words like 'participatory', 'indigenous', 'community'. The search process was iterative, with search terms adapted when they did not yield the expected results. Like Coemans and Hannes (2017), we found that searching by specific art type works better than using 'art' in general. This yielded many papers on the 'state of the art' in certain research field. The same holds for geographic area: 435 most researchers do not use the keyword 'Global South' in their titles or keywords, but rather mention the specify region or country/countries. We focused our search on Low and Middle Income Countries and on vulnerable (indigenous) groups within High Income Countries. We limited the search to papers in which art was used as research process and art was developed as product from the research. Articles on art therapy and projects using art therapeutically were excluded. This required a