Social Sciences & Humanities Open 8 (2023) 100557 Contents lists available at ScienceDirect Social Sciences & Humanities Open journal homepage: www.sciencedirect.com/journal/social-sciences-and-humanities-open Regular Article Space-time cube approach in analysing conflicts in Africa Adams Osman a,*, Alex Barimah Owusu b, Kofi Adu-Boahen a, Emmanuel Atamey a a Department of Geography Education, University of Education, Winneba, Ghana b Department of Geography and Resource Development, University of Ghana, Legon, Ghana A R T I C L E I N F O A B S T R A C T Keywords: In terms of conflict types and occurrences, there is a mesh of old, new, and concurrent conflicts which coexist and Emerging hotspots are affecting each other over space and time. Existing conflict studies are unable to fully explicate the space-time Political geography elements, hence this generates static and two-dimensional hotspots/coldspots. Using Africa as a case, this study Spatial statistics used three-dimensional space-time cube, with conflict occurrences grouped into bins where space is mapped Conflict Peace horizontally and time is mapped vertically for analysis. Analysis of conflict based on the three dimensional space- Africa time cube produced four main categories of hotspots namely consecutive, sporadic, oscillating, and new hotspots. Furthermore, the causes of conflicts in Africa varied significantly across each time-based hotspot, providing insight into why straightjacket solutions have been unsuccessful. Conflict managers can learn from the patterns of time-based hotspots which helps to see conflicts as three-dimensional entities needing with three levels of orientation that focus on type, space, and time instantaneously. 1. Background Africa is that civil war tends to cluster within 15% of a state’s territory, with different kinds of political violence exhibiting high rates of repe- Globally, conflict are on the rise with Africa facing fresh waves of tition (Raleigh et al., 2018). Wars are shaped by the collateral damage extremism, rioting, and protest. Between 2018 and 2019, Africa expe- suffered by belligerent parties in neighbouring areas, as well as rienced about 36% rise in conflict occurrences, with a total of 21600 spatial-temporal spillovers (Kibris, 2021). Mineral resource locations in conflicts (Allison, 2020). Climate change, catastrophes, religious low-cost extraction countries are hotspots for conflicts (Raleigh, 2014). extremism, poverty, corruption, foreign meddling, poor education, While border regions/towns are clusters for insurgent crime-related ethnic tension, unemployment, declining wages, rising commodity pri- incidents (Johnson & Braithwaite, 2017). These insights are useful for ces, weak institutions, and resource competition contribute to the the development of military strategies, consolidation of peacekeeping emergence and perpetuation of conflicts in Africa (Adaawen et al., 2019; operations, and solutions targeted at the regional, national, and zonal Cabot, 2017; Petrova, 2022). Consequently, conflicts have sunk the levels (Mack et al., 2021). continent into its current socioeconomic distresses, which causes tens of In spite of the useful information provided by most spatial analysis thousands of deaths per year. Conflicts have also created hunger, techniques on conflicts, they are less time sensitive, so a sequence of malnutrition, internal and external displacement, degraded social discrete outputs is employed to demonstrate the impact of time, pre- structures, conflict traps, and economic stagnation on the continent venting the discovery of time-based clusters. Space-time cube analysis (Burke et al., 2009; Manotas-Hidalgo et al., 2021). has a superior advantage over absolute and relative location mapping, Conflict events, causes, and consequences exist within space and spatial autocorrelation, Getis Ord G, and local Anselin Moran’s I by time, providing possibilities for analysing past trends and future patterns using three-dimensional model to describe space horizontally and time for mitigation purposes. Existing analytical approaches have relied on vertically (Li et al., 2010). The space time cube aggregates discrete spatial analysis (absolute and relative location mapping, spatial auto- events into bins, resulting in highly accurate representations of correlation, Getis Ord G, and local Anselin Moran’s I), for distribution spatio-temporal data in time-based clusters such as new, consecutive, mapping and clustering of conflict zones as hotspots or coldspots across persistent, oscillating, and sporadic hotspots or coldspots (Purwanto the continent (Kotsadam & Østby, 2019; Raleigh et al., 2018; O’Loughlin et al., 2021). The importance of the spatial-time cube has been crucial in & Raleigh, 2008). A significant finding from such geospatial research in understanding the spread of Covid-19, forest decline, urban mobility * Corresponding author. E-mail address: aosman@uew.edu.gh (A. Osman). https://doi.org/10.1016/j.ssaho.2023.100557 Received 28 August 2022; Received in revised form 18 February 2023; Accepted 7 May 2023 Available online 24 May 2023 2590-2911/© 2023 The Authors. Published by Elsevier Ltd. This is an open access article under the CC BY-NC-ND license (http://creativecommons.org/licenses/by- nc-nd/4.0/). A. Osman et al. S o c ia l S c i e n c e s & H u m a n i t i e s O p en 8 (2023) 100557 patterns, and people’s life histories (Harris et al., 2017; Kveladze et al., 2.2. Space and time influence on conflict 2015; Purwanto et al., 2021) but its benefits are yet to be experienced in conflict studies. Several methodologies have been explored to study the relationship Conflict studies would benefit from space-time cube analysis espe- between conflict, space, and time. Nearest neighbour analysis, Getis Ord cially in Africa because of the unique overlay of conflicts predating General G., incremental autocorrelation, and Ripley’s K function are colonial periods and cemented by recent geopolitical events. Thus, new used to explore the statistical significance of conflict distributions. Getis conflict patterns cannot be blended with old conflict patterns since the Ord Gi* and Anselin Moran’s I, are used as density and clustering two coexist with separate underlying causes and may require different techniques for identifying conflict hotspots (Griffith, 2021). However, practical solutions (Allison, 2020). Hence, this study sought to assess the the minimal focus on time is one of the limitations of these spatial ap- space-time effect on conflicts in Africa through the following questions; proaches. Space-time cube analysis developed by Hägerstrand (1970) solves for time through a simultaneous analysis of representing space 1 Where are the notable central points of conflicts in Africa? horizontally and time vertically on a cube (Li et al., 2010). The weakness 2 How do space and time simultaneously influence the occurrence of of the cubes is treating space and time as a container for boxing social conflicts in Africa? life and interactions rather than a social construct (Simandan, 2016, 3 What factors influence the various space-time hotspots/coldspots 2020; Thrift, 1996, 2005). However, cubes give a better conceptual and zones in Africa? visual representation of space and time concurrently than any two-dimensional display and InfoVis (Bach et al., 2017). The study’s principal argument is that conflicts in Africa are not only Although the space-time cube is based on Getis Ord Gi* hotspot spatially confined but also time-bound. This is because months before analysis, it generates bins with an identification (ID) showing and after the rainy season, herder-farmer confrontations across the geographic position, as well as a time-step ID (Fig. 1). It also applies the continent heighten (Adaawen et al., 2019; Mack et al., 2021; Petrova, Mann-Kendall statistic to decide if a statistically significant temporal 2022). The legendary Algerian revolt sparked a wave of riots and pro- trend exists based on a time-slice of Z-scores generated by the Getis-Ord tests across the continent (Arab Spring), particularly in Libya, Tunisia, Gi (Harris et al., 2017). The characteristics of a space-time cube enable it and Egypt, and was followed by insurgencies in Nigeria, Mali, and to produce statistically significant patterns such as new, historical, Burkina Faso (Van den Bosch & Raubo, 2017; Aghedo & James Eke, consecutive, persistent, intensifying, sporadic, as well as diminishing 2013). Furthermore, recent military takeovers in Africa have sparked a and oscillating (Bass, 2017). New hotspot/coldspot locations mean it has chain reaction of violence and demonstrations that must be understood never been statistically significant previously, while consecutive hot- from a spatio-temporal perspective (Elischer, 2021). Another relevance spots/coldspots have areas significant in the final time-step but not prior of this work is the ability to trace conflict clusters as time-dependent to the final hotspot/coldspot analysis (Li et al., 2010). An intensifying output rather than static output (Beetar, 2019). Understanding the hotspot/coldspot defines areas that have about ninety percent of the development or re-emergence of conflicts and their varied causalities is time-step intervals being significant areas including the final time step. critical for providing meaningful information and resource mobilisation This results in persistent hotspots/coldspots having ninety percent of the for conflict resolution. time-step intervals with no noticeable movement towards increasing or decreasing over time (Shimazaki & Shinomoto, 2007). Diminishing 2. Literature hotspots/coldspots are areas that were once statistically significant hotspot/coldspot for ninety percent of the time-step intervals but 2.1. Central points of conflicts in Africa decreased in intensity over time (ESRI, 2022). Sporadic hotspots/cold- spots are on-again then off-again hotspots/coldspots with oscillating Early studies on space and conflict discovered no association be- hotspots/coldspots as statistically significant hotspot for the final tween these two phenomena because conflicts were analysed at the time-step interval that has a history of statistically changing its entity as country level before World War II (Lis et al., 2021). By the end of World hotspot or coldspot a prior time step. The last hotspots/coldspots cate- War II, Cold War outplays resulted in the spread of conflicts both gory is historical hotspots/coldspots having its recent period as neither internally and internationally, necessitating new spatial analysis hot/cold, but at least ninety percent of the time-step intervals have been through the advancement of philosophy, geography, and technology statistically significant hotspots/coldspots (ESRI, 2022). (Barnes, 2011, 2022). Consequently, this promoted the study of conflict The ability to detect these hotspot/coldspot patterns provides re- diffusion. According to the diffusion conflict hypothesis, conflicts start at searchers with a competitive advantage in tracking geographical and a key location and then spread to other geographic areas (O’Loughlin & time-based phenomena and dissemination. It was crucial for COVID-19 Raleigh, 2008). The concept further argues that the location of spread mapping across Asia. Bass (2017) used it to track high manatee conflict-prone nations is not only a characteristic but also a cause of (Trichechus manatus latirostris) death rates in Florida, USA. Space-time conflict (O’Loughlin & Raleigh, 2008). cube analysis aided in the collection of baseline ecological data for un- Furthermore, conflict tend to follow borderlines where similar socio- derstanding pilot whale subspecies habitat and migration to advise economic conditions exist between neighbouring countries (Murdoch & conservation management (Betty et al., 2020). Harris et al. (2017) were Sandler, 2002). Thus, conflict spreads through a two-way reflexive also able to map emergent degradation zones inside the Amazon Forest process between the originating conflict location and spread to other using a space-time cube. It enabled Nakaya and Yano (2010) revealed places with similar population characteristics or ideology (Zupančič temporal inter-cluster linkages where transitory clusters appeared in a et al., 2018). Evident is the diffusion of the Afghan war into Pakistan pair of hotspot zones, resulting in the discovery of a novel sort of while the Iraq war triggered the Syrian and the spread of ISIS into Asia “displacement” criminal phenomena. According to this study, the ben- and North Africa. According to Ruggeri et al. (2017), identifying the loci efits of space-time cubes are significant for uncovering new sides of and conflict directions are significant in reducing the time and cost of conflict trends in Africa that earlier studies could not. A shortcoming mobilising peacekeeping forces. with space-time cubes happens to be the complexity of representing its output and difficulty in interpreting its results (Kveladze et al., 2019). Hence most researchers represent the outputs as two-dimensional rather than a three-dimensional output. 2 A. Osman et al. S o c ia l S c i e n c e s & H u m a n i t i e s O p en 8 (2023) 100557 2.3. Causes of conflicts According to resource-conflict theory, escalating conflicts in Africa can be attributed to the richness and reliance on natural resources (Namakula, 2022). It is responsible for thousands of conflict-related deaths every year in Africa (Bruch et al., 2019). High gold and diamond reserves encourage and feed hostilities, as witnessed in Sierra Leone, Liberia, and other countries. Natural resource reliance impairs institutional and democratic processes, making communities more vulnerable to violent conflicts, like the situation in Delta State, Nigeria (Ebiede, 2017; Omorede, 2014). Ac- cording to the theory of rebellion, riots, and demonstrations by people in Africa may be traced to popular agitation against the continent’s economic woes (Engels, 2015). Rising unemployment, declining wages, and rising commodity prices give sufficient motivation to fight, protest, riot, and even join armed groups (Manotas-Hidalgo et al., 2021). Also, the emergence of climate change, accompanied by changes in rainfall patterns and land degradation, is to be blamed for the recurrence of hostilities in many parts of Africa (Sakaguchi et al., 2017). Petrova (2022) observed increased levels of community conflict because of less rainfall in savannah regions and grasslands. In addition, high ethnic diversity and nationalism promotes and sustains conflicts because they elicit strong emotional reactions based on psychological, biological and cultural dif- Fig. 1. Space-time cube. ferences (Alesina & La Ferrara, 2000). Civil conflicts are stimulated and Source: Li et al., 2010. entrenched by colonial favouritism and differences in political ideology (Manotas-Hidalgo et al., 2021). Growing numbers of small arms provide tools for mercenaries, rebels, and jihadist to cause chaos across border regions. Table 1 Centrality of conflicts in Africa per year, by subregion and conflict event. Year Town Latitude Longitude Country 1997 Yaounde 3.867 11.517 Cameroon 1998 Odeama Creek 4.347 6.435 Nigeria 1999 Bandundu − 3.317 17.367 Democratic Republic of Congo 2000 Opala − 0.508 24.229 Democratic Republic of Congo 2001 Shabunda − 2.694 27.346 Democratic Republic of Congo 2002 Colline Rukoko − 1.65 29.267 Rwanda 2003 Kanyasi 1.391 30.441 Democratic Republic of Congo 2004 Ndrele 2.306 30.493 Democratic Republic of Congo 2005 Omee 2.189 31.367 Uganda 2006 Anaka 2.594 31.963 Uganda 2007 Iten 0.67 35.508 Kenya 2008 Kapsabet 0.204 35.105 Kenya 2009 Bavi 1.431 30.298 Democratic Republic of Congo 2010 Omoro 2.75 32.5 Uganda 2011 Jabal Kurgul 13.2 26.967 Sudan 2012 Berunda 2.336 30.299 Democratic Republic of Congo 2013 Mirmir 8.335 30.016 South Sudan 2014 Aweil 8.767 27.397 South Sudan 2015 Gumbolo 6.933 27.95 South Sudan 2016 Faraksika 5.02 29.71 South Sudan 2017 Bangusa 4.802 28.764 South Sudan 2018 Sodi 5.288 26.094 Central African Republic 2019 Sarh 9.15 18.383 Chad 2020 Goumoun 9.932 15.552 Chad 2021 Kabo 7.698 18.63 Central African Republic Conflict by Subregion North Africa As Saddadah 31.468 14.631 Libya Southern Africa Olievenhoutbosch − 25.917 28.105 South Africa West Africa Kopargo 9.841 1.542 Benin East Africa Khorof Harar 2.202 40.754 Kenya Central Africa Abongisia 0.204 25.595 Democratic Republic of Congo Conflict type Battles Tore 4.502 30.157 South Sudan Explosions/Remote violence Nyakma 11.417 30.533 Sudan Protests Fotokol 12.373 14.228 Cameroon Riots Kisangani − 1.167 24.4 Democratic Republic of Congo Strategic developments Irabanda 5.949 22.07 Central African Republic Violence against civilians Mabanga 2.183 27.933 Democratic Republic of Congo 3 A. Osman et al. S o c ia l S c i e n c e s & H u m a n i t i e s O p en 8 (2023) 100557 Fig. 2. Distribution of centrality of conflicts in Africa. 3. Materials and method feature, (mean feature and median feature) was used to assess the cen- trality of the conflict locations in Africa based on the Euclidean distance 3.1. Data sources and processing function. The study further employed the standard deviation ellipsoid equation (Equation (1)). The data used for this study were obtained from a variety of re- ⎛ ∑n n ⎞ ∑ positories. The data on conflict statistics for Africa was obtained from x 2 ⎜ i xiyi ⎟ the Armed Conflict Location and Event Data Project, which spanned C ⎜ i=1 i=1= ⎟⎝∑n ∑n ⎠ (Equation 1) from 1997 to 2021. The Greenberg Diversity Index 2021 and the Pew x 2iyi yi Research Centre provided data on ethnic and religious diversity indexes, i=1 i=1 respectively. Data on governance performance came from the Mo Ibra- − −Where x and y are the coordinates of the features (i) and {x and y} as him Governance Index, climate change data came from Eckstein et al. (2021) and data on guns from the Small Arms Survey database. The the mean centre for the total features. Furthermore, the standard devi-ation for x and y was based on the function (Equation (2)) to summarise United Nations Sustainable Development Global database provided data on corruption, infrastructure (water, power, waste), schools, internet, the central tendency of conflicts, dispersions, and directional trends. health, economic growth, stunting growth, wasting, level of education, ⎛( ) √̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅(̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅)̅̅̅̅̅⎞∑n ∑n ∑n ∑n ∑n 2 unemployment, poverty, and human rights violations. A total of ⎜ x− 2 + y− 2 ± x− 2 − y− 2 + 4 xly ⎟ ⎜ l l l l l ⎟ thirty-six (36) independent variables were compiled from the various σ1 2 ⎜ i=1 i=1 i=1 i=1, = ⎟⎜ ⎟ sources and merged into a single file. The merged data was spatially ⎝ 2n ⎠ joined with a shapefile of the political boundary of African countries. Data gaps (countries without scores) were filled using the K-nearest (Equation 2) neighbour spatial in ArcPro 2.1 software. K-nearest neighbour uses the average score of four neighbours (nations) to fill in countries without The variances of the conflict points were scaled by an adjustment scores. Furthermore, all merged datasets were normalized [X factor of one (1) generating an ellipse which encompassed about 68% of new = (Xi – Xmin)/(Xmax – Xmin )* 100] with Xi as the value, Xmin minimum the conflict points. For Q2 (Space and time simultaneously have no influ-n = value of the variable, Xmax = maximum value of the variable] on a scale ence on the occurrence of conflicts in Africa) the incremental spatial of 0–100 to ensure consistency and ease of comparison between nations. autocorrelation analysis based on Moran’s I function (Equation (3)) was employed. n n 3.2. Data analysis ∑ ∑wi,j zizn j I i=1 j=1= n (Equation 3) The study employed several analytical techniques because of the So ∑ z2i uniqueness of each research question proposed. For Q1 (Notable central i=1 points of conflicts in Africa), a spatial central tendency, measuring central 4 A. Osman et al. S o c ia l S c i e n c e s & H u m a n i t i e s O p en 8 (2023) 100557 Table 2 Mean conflict fatalities per year, subregion, and conflict event. Latitude Longitude Fatalities Nearest Conflict Town Country 1997 4.104 10.775 8 Pouma Cameroon 1998 3.139 9.275 16 Kribi Cameroon 1999 0.139 14.318 33 Akouaka Gabon 2000 2.159 17.246 6 Imese Democratic Republic of Congo 2001 1.228 19.462 7 Bokakata Democratic Republic of Congo 2002 0.519 25.192 7 Kisangani Democratic Republic of Congo 2003 2.425 23.866 6 Gubu Democratic Republic of Congo 2004 2.986 25.980 6 Mboki Democratic Republic of Congo 2005 2.645 27.421 3 Banda Democratic Republic of Congo 2006 4.453 27.574 3 Dalundue Democratic Republic of Congo 2007 2.847 30.276 3 Biringi Democratic Republic of Congo 2008 0.862 31.028 2 Kyebando Uganda 2009 1.010 27.990 4 Basiri Democratic Republic of Congo 2010 2.967 28.899 3 Tora Democratic Republic of Congo 2011 14.082 24.571 3 Amar Allah Sudan 2012 3.953 26.349 2 Samongo Democratic Republic of Congo 2013 8.924 27.470 2 Akuem South Sudan 2014 8.948 25.557 2 Khor Shamam South Sudan 2015 7.078 25.137 2 Abd El Lait Sudan 2016 6.687 26.205 2 Abd El Lait Sudan 2017 7.390 24.768 2 Khor Dulu South Sudan 2018 7.123 22.934 1 Bani Central African Republic 2019 10.655 19.156 1 Boum-Kabir Chad 2020 9.920 17.635 1 Kouno Chad 2021 8.739 18.490 1 South Danamadji Chad Subregion North Africa 28.407 17.148 2 Zillah Libya Southern Africa − 25.537 27.307 0 Thekwane South Africa West Africa 10.203 0.146 3 Nazawni Ghana East Africa 1.652 38.960 3 Hadado Kenya Central Africa 0.568 22.505 6 Djolu Democratic Republic of Congo Conflict type Battles 7.302 26.667 7 Abulu Sudan Explosions/Remote violence 14.066 28.336 3 Umm Badr Sudan Protests 11.642 15.972 0 Dourbali Chad Riots 0.195 20.490 1 Befale Democratic Republic of Congo Strategic developments 7.384 19.273 0 Kouanga Central African Republic Violence against civilians 3.863 24.276 3 Yangili Democratic Republic of Congo − Zi as the deviation of the conflict attributes i from the mean (x − X) hotspots/coldspots produced from the analysis were new, consecutive, i with wi,j as weight between i and j spatially, S0 as the aggregate of the intensifying, persistent, diminishing, sporadic, and oscillating. spatial weights and n as the total number of conflicts. Aggregated spatial Q3 (Factors influencing the various conflict space-time hotspots/coldspots ∑ ∑ weight was defined as S n n w . The z-score (Zi) function is in Africa) analysis was possible by interpolating the independent vari-0 = i=1 j=1 i,j given as ables based on inverse distance weighted interpolation technique to generate a surface for the thirty-six (36) individual variables. Per the Ii − E[I] (Equation 4) interpolated variables, a surface extraction tool was used to generate the Zi = √̅̅̅̅̅̅̅̅ V[I] values for each specific space-time hotspot/cold spot. A multi- collinearity test was used to assess the variance of inflation (VIF) factor − 1 for the thirty-six (36) variables with VIF below 3.5 accepted for further with E[Ii] = (Equation 5) (n − 1) analysis. A VIF below 3.5 is a good measure of ensuring independent variables are not related to one another (Prather & Kaspari, 2019). [ ] and V I E I2 E I 2[ ] = − [ ] (Equation 6) Fourteen variables (access to internet, access to pipe water, access to universal health, climate change, corruption index, ethnic diversity, With I is the Moran’s I value and n = number of conflicts. municipal waste management, poverty, restrictions on press freedom, Incremental autocorrelation was employed to measure the intensity primary education, religious diversity, subjective well-being, unem- of clustering based on the spatial distance between conflict points by ployment and unsentenced detained) met the 3.5 VIF criteria. A multi- assessing a series of distances and their statistically significant peak z- nomial logistic regression was performed to assess the likelihood of the scores. The modelling parameters for the incremental spatial autocor- fourteen variables determining the various types of space-time hotspot relation were the number of bands as twenty and the distance function and coldspots. as Euclidean and incremental distance determined by the incremental spatial autocorrelation algorithm automatically. With significant peak z- 4. Results scores and distances identified, the study modelled the locations of conflicts and their corresponding time/day of occurrence by structuring 4.1. Notable central points of conflicts in Africa them into space-time bins based on the Mann-Kendell trend test (Ken- dall, 1955; Mann, 1945). The study adopted the space-time cube algo- Per spatial central tendency analysis, the study identified one central rithm from ArcPro 2.1 to generate emerging hotspots/coldspots by point for each year. In 1997, the central conflict point was in Yaoude aggregating conflict points to netCDF data cube space-time bins. Major (Cameroon) whereas in 2021, it was found in Kabo (Central Republic of 5 A. Osman et al. S o c ia l S c i e n c e s & H u m a n i t i e s O p en 8 (2023) 100557 Fig. 3. Mean distribution of conflicts in Africa. Africa). Between 1997 and 2021 countries with more central conflict but East Africa had the highest number of fatalities (144,251). Although, points were the Democratic Republic of Congo and South Sudan with 7 Southern Africa had an ellipsoid of 2,087,609.44 km2 it recorded fewer and 6 points respectively (Table 1; Fig. 2). fatalities (22,786) when compared to other subregions (Table 3: Fig. 4). Per subregions in Africa, North Africa had As Saddadah (Libya), West Conflicts in East Africa and Southern Africa had trends of conflicts Africa had Kopargo (Benin), Khorof Harar (Kenya) for East Africa, with rotations less than 500. Southern Africa had zero (0) fatality while Abongisia (Democratic Republic of Congo) for Central Africa and Olie- East Africa had a mean of 3 fatalities. The ellipsoid for the year 1998, venhoutbosch for Southern Africa as the central points. In terms of con- riots type of conflict, and West Africa had the most countries and fa- flict types, the Democratic Republic of Congo had central points for riots talities (Table 3). and violence against civilians in Kisangani and Mabanga, respectively. The concentration of mean distribution of conflicts from 1997 to 2021 was within Central Africa with no point in West, North and 4.2. Space and time simultaneously influence conflicts in Africa Southern Africa. Democratic Republic of Congo had the highest counts of mean conflicts from 1997 to 2021 with about eleven different periods Initial spatial analysis identified significant spatial distribution of (Table 2; Fig. 3). conflicts based on years, type, and region. With conflict events per Democratic Republic of Congo had eleven mean points from 2000 to specific year, the average number of peaks at which conflict points 2012. Akouaka (Gabon) had the highest mean fatality with about 33 cluster was two (Table 4). It was observed that 2006 had the most in- deaths in 1999. Per conflict types, Abulu (Sudan) had the highest mean cremental distance of 52.53 km with a peaking distance of 1267.18 km fatality with 7 deaths while Dourbali (Chad) had no fatality. Mean while 2020 had the least distance of 16.47 km. Maximum peak distance analysis per subregion indicated that Djolu (Democratic Republic of was 2279.46 km attained in 1997 (Moran’s I = 0.02, Z-score = 30.78, p Congo) had the highest mean of 6 fatalities (Table 2; Fig. 3). = 0.00) reduced significantly to 681 km in 2017 (Moran’s I = 0.02, Z- Based on the standard deviation ellipsoid, the study identified the score = 63.14, p = 0.00). direction and number of conflict points within each ellipsoid. The year Conflict type with the least peak distance (472.30 km) was violence 2011 had the largest ellipsoid of 21,248,408.33 km2 with 560,295 fa- against civilians (Moran’s I = 0.05, Z-score = 303.31, p = 0.00) with an talities while 2014 had the least ellipsoid of 13893876.40 km2 and incremental distance of 12.91 km. At the subregional level conflicts in 417,252 fatalities. The year with the most fatalities was 2012 with East Africa (Moran’s I = 0.05, Z-score = 546.02, p = 0.00) had the least 684,037 deaths and ellipsoid of 21054891.29 km2. peak distance at 276.32 km. Protest exhibited the largest ellipsoid compared with other conflict Results showed that four hotspots were persistent thus new hotspots, types with an area of 26,424,782.63 km2 and 582,808 fatalities. For sporadic, oscillating, and consecutive. Oscillating conflict hotspots were subregions Central Africa had the largest ellipsoid, 3,304,200.23 km2, the most frequent in Africa accounting for the most of riots (70.9%) and strategic developments (41.25%) hotspots (Table 5; Fig. 5). Per region, 6 A. Osman et al. S o c ia l S c i e n c e s & H u m a n i t i e s O p en 8 (2023) 100557 7 Table 3 Standard deviation ellipsoid for conflict fatalities in Africa. Rotation X Standard Deviation (Km) Y Standard Deviation (Km) Area of ellipsoid (Km2) No. of conflict points No. of Fatalities Mean Fatalities No. of countries within ellipsoid 1997 110.31 3153.78 1712.68 16968143.57 123,678 516,130 4 18 1998 106.75 3219.24 1481.74 14984835.21 110,882 322,159 3 19 1999 107.94 2816.31 1646.81 14569727.44 103,291 447,493 4 16 2000 115.06 3161.21 1881.65 18686201.40 128,955 557,376 4 17 2001 127.82 3005.80 1813.80 17126816.70 124,404 535,883 4 17 2002 137.24 1732.07 2853.78 15527882.06 118,947 536,565 5 12 2003 119.51 2907.45 1788.97 16339608.52 143,012 467,107 3 17 2004 132.36 2592.68 1705.87 13893876.40 117,622 417,252 4 8 2005 129.75 2570.27 1814.93 14654349.81 135,026 504,101 4 9 2006 128.83 2711.22 1689.03 14385656.67 132,952 501,869 4 9 2007 132.06 2856.00 1872.07 16796061.19 134,417 514,991 4 10 2008 141.64 1907.34 2560.72 15343237.48 132,239 494,180 4 10 2009 143.43 1959.24 2813.44 17316287.91 142,613 645,871 5 13 2010 117.94 2681.66 2110.78 17781755.55 147,044 554,918 4 15 2011 149.33 2109.79 3205.97 21248408.33 160,220 560,295 3 12 2012 148.39 2194.51 3054.13 21054891.29 155,341 684,037 4 18 2013 158.12 2213.71 2926.41 20350931.20 153,112 560,464 4 13 2014 153.25 2068.12 2899.82 18839736.38 141,931 549,768 4 12 2015 157.57 2011.19 2988.76 18883032.32 129,036 640,409 5 13 2016 153.82 1980.57 2933.68 18252846.13 141,767 538,709 4 11 2017 140.68 1995.32 2985.48 18713504.41 145,207 550,946 4 10 2018 134.72 2947.84 1972.08 18262290.27 134,009 532,592 4 14 2019 142.92 1949.78 3383.67 20725394.41 131,952 448,972 3 14 2020 139.59 1936.49 3336.09 20294642.81 124,779 432,235 3 16 2021 137.01 1880.89 3295.75 19473548.69 123,876 430,776 3 17 Conflict Event Battles 113.22 2636.36 1633.10 13525228.21 136,701 527,107 4 8 Explosions/Remote violence 130.96 2927.00 1588.34 14604772.86 122,620 473,376 4 9 Protests 155.48 2124.15 3960.03 26424782.63 158,894 582,808 4 18 Riots 156.54 2147.87 3697.91 24951172.08 138,763 553,124 4 24 Strategic developments 107.10 3031.34 1777.49 16926621.68 153,870 535,884 3 18 Violence against civilians 120.81 2527.13 1642.90 13042685.58 116,608 414,870 4 10 Subregion North Africa 124.44 2329.43 1112.33 8139731.38 31,095 39,566 1 2 Southern Africa 24.82 563.70 1178.88 2087609.44 15,446 2,278 0 1 Central Africa 122.53 1184.78 887.77 3304200.23 37,901 111,076 3 1 East Africa 42.26 954.59 1100.75 3300913.66 56,078 144,251 3 1 West Africa 93.88 1401.18 567.25 2496875.12 28,313 68,219 2 3 A. Osman et al. S o c ia l S c i e n c e s & H u m a n i t i e s O p en 8 (2023) 100557 Fig. 4. Areas covered by standard deviation ellipsoid for conflict fatalities in Africa. oscillating hotspots dominance was evident in Southern Africa = 0.39, p < 0.05) was the main factor likely to cause NHS for ES. For EV, (46.96%), North Africa (29.91%) and West Africa (10.46%) (Table 5; NHS were determined by accessibility to internet (β = 0.51, p < 0.01), Fig. 6). solid waste management challenges (β = 0.46, p < 0.01), unemployment Hotspots in Africa are sporadic, accounting for less than 20% of (β = 0.46, p < 0.01) and restriction on press freedom (β = 0.36, p < hotspot types when analysed per conflict type and at the subregional 0.01). Consecutive hotspot areas were influenced by solid waste man- level. Emerging new conflict hotspot types were explosions/remote agement challenges for EP (β = 2.65, p < 0.01) and EV (β = 0.40, p < violence (EE) and much confined to East (6.42%) and West Africa 0.01). Religious diversity (β = 0.20, p < 0.01) influenced areas with CHS (6.22%) (Table 5; Figs. 5 and 6). for EB. In North Africa ethnic diversity (β = 3.43, p < 0.01), access to Space and time cumulatively revealed areas of significantly low universal health (β = 2.31, p < 0.01) and restrictions on press freedom number of conflicts (coldspots) in Africa. The coldspots spatial distri- (β = 2.67, p < 0.01) were more likely to influence CHS for EB (Table 7). bution was limited to North Africa which had sporadic (0.65%) and Central Africa had over eight factors likely to influence CHS and three consecutive (0.65%) coldspots (Table 5; Fig. 6). for NHS (ethnic diversity, unsentenced detained, access to pipe water). In general, the conflict types with the most hotspot coverage were In East Africa, access to universal health, pipe water and unemployment explosions/remote violence (EE) and protests (EP) with 6652.94(1000) were more likely to affect NHS with CHS and entrenched by unem- km2 and of 7100.25 km2 area, respectively. Battles (Cube area = 71.75 ployment, restrictions on press freedom and climate change. km2 and total cube area = 15420.03 (000)) had the lowest coverage area. Climate change, religious diversity and subjective well-being were East Africa had the most hotspot zones followed by North Africa and the more likely to influence CHS in West Africa. Emerging NHS were less least as Southern Africa. likely to be a result of increasing poverty (β = 0.84, p < 0.01), and primary education (β = 0.18, p < 0.01) but not restrictions on press 4.3. Causes of conflicts vary across space-time hotspots/coldspots in freedom (β = 1.34, p < 0.01). Also, unemployment, subjective well- Africa being, primary education, and restrictions on press freedom were more likely to account for OHS in West Africa. Using variables which had VIF below 3.5 the study found fourteen variables to explain the difference in hotspots and coldspots across Af- 5. Discussion rica (Tables 6 and 7). Ethnic diversity was less likely to influence the occurrence of new NHS for EB (β = − 0.25, p < 0.01), EV (β = − 0.26, p < This study reaffirms Lis et al. (2021) and Simanda’s (2019a, 2019b) 0.01), EP (β = − 0.89, p < 0.05) and ES (β = − 1.19, p < 0.05) (Table 6). results about the impact of space and time on conflicts. According to the New hotspots (NHS) for EB were influenced by access to universal health study, the centrality of conflict in Africa is determined by the year of the (β = 0.70, p < 0.01), solid waste management challenges (β = 0.87, p < conflict, the type of conflict, and the subregion. Communities in the 0.01) and unemployment (β = 0.74, p < 0.01). Subjective well-being (β Democratic Republic of Congo were found to be the central conflict points 8 A. Osman et al. S o c ia l S c i e n c e s & H u m a n i t i e s O p en 8 (2023) 100557 9 Table 4 Maximum peak distance of clustering for conflict fatalities in Africa. Year, event, subregion Number of peaks Distance Increment (Km) Peak Distance (Km) Moran’s I Expected Index Z-score p-value 1997 2 46.93 2279.46 0.02 − 0.0001 30.78 0.00 1998 2 37.14 1676.80 0.05 − 0.0002 75.09 0.00 1999 2 42.38 1199.94 0.09 − 0.0002 108.85 0.00 2000 2 43.06 1340.02 0.02 − 0.0002 27.31 0.00 2001 2 41.78 830.16 0.02 − 0.0002 12.80 0.00 2002 2 45.23 1156.21 0.02 − 0.0002 25.88 0.00 2003 2 47.47 1013.43 0.08 − 0.0002 60.02 0.00 2004 2 50.06 1647.25 0.02 − 0.0003 25.47 0.00 2005 1 51.25 1482.38 0.02 − 0.0003 14.95 0.00 2006 2 52.53 1267.18 0.03 − 0.0003 22.27 0.00 2007 1 48.06 870.57 0.06 − 0.0002 41.25 0.00 2008 1 38.76 1500.25 0.02 − 0.0002 26.40 0.00 2009 2 48.50 1313.72 0.01 − 0.0003 10.77 0.00 2010 2 48.79 2023.22 0.02 − 0.0002 29.84 0.00 2011 1 39.22 1796.29 0.00 − 0.0001 9.02 0.00 2012 2 34.20 1332.55 0.01 − 0.0001 19.53 0.00 2013 2 32.92 1314.19 0.02 − 0.0001 91.78 0.00 2014 2 27.09 815.37 0.05 − 0.0001 133.15 0.00 2015 2 25.23 780.33 0.05 − 0.0001 142.70 0.00 2016 2 23.19 872.46 0.02 − 0.0001 63.89 0.00 2017 1 23.68 681.13 0.02 − 0.0001 63.14 0.00 2018 1 23.45 721.24 0.03 − 0.0001 62.84 0.00 2019 2 18.85 1009.64 0.03 0.0000 171.21 0.00 2020 2 16.47 878.39 0.02 0.0000 137.29 0.00 2021 1 18.16 830.04 0.03 0.0000 111.72 0.00 Conflict Event Battles (EB) 2 14.46 609.01 0.04 0.0000 404.41 0.00 Explosions/Remote violence (EE) 2 23.24 886.19 0.04 0.0000 135.09 0.00 Protests (EP) 2 20.09 543.64 0.004 0.0000 21.55 0.00 Riots (ER) Strategic developments (ES) 2 21.45 936.19 0.005 0.0000 10.10 0.00 Violence against civilians (EV) 1 12.91 472.30 0.05 0.0000 303.31 0.00 Subregion North Africa (NA) 2 12.68 593.08 0.03 − 0.0001 364.69 0.00 Southern Africa (SA) 2 15.27 443.34 0.01 0.0001 85.39 0.00 Central Africa (CA) 1 11.38 318.22 0.08 0.0000 535.34 0.00 East Africa (EA) 2 8.90 276.32 0.05 0.0000 546.02 0.00 West Africa (WA) 2 98.90 316.32 0.05 0.0000 552.02 0.00 A. Osman et al. S o c ia l S c i e n c e s & H u m a n i t i e s O p en 8 (2023) 100557 Table 5 Space-time cube analysis of conflict hotspots in Africa from 1997 to 2021. Variables Conflict event Subregion EB EE EP ER ES EV NA SA CA EA WA No. of cubes 2149 937 937 1692 1549 2230 926 988 1297 1699 2122 Unit area of cubes (km2) 71.75 7100.25 7100.25 7283.94 7165.62 7180.36 3798.63 998.94 2204.49 2418.57 1276.85 Total area of cubes (1000, km2) 15420.03 6652.94 6652.94 12324.43 11099.60 16012.20 3517.53 986.95 2859.22 4109.60 2709.47 %No pattern 68.31 59.98 73.53 23.52 46.68 63.05 51.30 45.65 69.01 75.81 37.89 % Hotspots New 7.40 17.29 10.78 6.09 2.13 1.57 1.19 0.40 2.16 6.42 6.22 Sporadic 8.47 3.31 7.15 0.18 4.45 18.92 0.54 0.81 13.57 8.48 10.13 Oscillating – 70.09 41.25 29.91 46.96 – 10.46 Consecutive 15.82 19.42 8.54 0.12 5.36 16.46 15.77 6.17 15.27 9.30 35.30 Persistent – – – – – – – – – – – Intensifying – – – – – – – – – – – Diminishing – – – – – – – – – – – Historical – – – – – – – – – – – % Cold spots New – – – – – – – – – – – Sporadic – – – – 0.13 – 0.65 – – – – Oscillating – – – – – – – – – – – Consecutive – – – – – – 0.65 – – – – Persistent – – – – – – – – – – – Intensifying – – – – – – – – – – – Diminishing – – – – – – – – – – – Historical – – – – – – – – – – – Mean Trend z-score No pattern 1.53 1.37 1.3 4.24 3.09 1.9 2.93 1.9 1.48 1.87 2.54 Hotspots New 2.62 4.67 4.79 4.25 4.35 3.88 4.06 4.49 4.2 3.69 3.77 Sporadic 3.19 4.63 4.88 5.27 4.62 4.38 4.62 4.97 4.83 2.55 3.23 Oscillating – 4.8 4.12 4.36 5.01 – – 4.35 Consecutive 4.86 5.08 5.15 5.41 4.6 4.42 3.80 5.05 4.4 4.6 4.26 Persistent – – – – – – – – – – – Intensifying – – – – – – – – – – – Diminishing – – – – – – – – – – – Historical – – – – – – – – – – – Cold spots New – – – – – – – – – – – Sporadic – – – – − 0.04 – − 3.42 – – – – Oscillating – – – – – – – – – – – Consecutive – – – – – – − 3.59 – – – – Persistent – – – – – – – – – – – Intensifying – – – – – – – – – – – Diminishing – – – – – – – – – – – Historical – – – – – – – – – – – Mean Trend p-value No pattern 0.13 0.21 0.21 0.2 0.11 0.15 0.03 0.31 0.26 0.38 0.12 Hotspot New 0.01 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.03 0.02 Sporadic 0.04 0.00 0.00 0.00 0.00 0.01 0.00 0.00 0.00 0.04 0.00 Oscillating – – – 0.00 0.05 – 0.00 – – – 0.00 Consecutive 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 Persistent – – – – – – – – – – – Intensifying – – – – – – – – – – – Diminishing – – – – – – – – – – – Historical – – – – – – – – – – – Cold spot New – – – – – – – – – – – Sporadic – – – – 0.9 – 0.00 – – – – Oscillating – – – – – – – – – – – Consecutive – – – – – – – – – – – Persistent – – – – – – – – – – – Intensifying – – – – – – – – – – – Diminishing – – – – – – – – – – – Historical – – – – – – – – – – – Battles (EB), Explosions/Remote violence (EE), Protests (EP), Riots (ER), Strategic developments (ES), Violence against civilians (EV), North Africa (NA), West Africa (WA), East Africa (EA), Central Africa (CA) and Southern Africa (SA). 10 A. Osman et al. S o c ia l S c i e n c e s & H u m a n i t i e s O p en 8 (2023) 100557 Fig. 5. Space-time analysis of conflict per events from 1997 to 2021. on a year-to-year basis and for riots and violence against civilians. The increased Anglophone prejudice, herder-farmer conflicts, and land Democratic Republic of Congo’s high count of centrality can be linked to grabbing (Feldt et al., 2020; Mbih, 2020; Ngong, 2021). the country’s high natural resource base, rurality, forested areas, and the According to O’Loughlin and Raleigh (2008), central conflict spots large number of armed groups that go through villages rapping, recruit- become areas from which it spreads to other areas was evident in this ing, stealing, and murdering innocent people (Perera, 2017; Stearns, study as most grew outward form conflict zones. The largest conflict 2017). South Sudan took over from the Democratic Republic of Congo as zone was in 2012 which encompassed most nations in East and Central the focal point of conflict in 2013, but things have changed dramatically Africa. The countries engulfed share similar socio-economic character- since then, as Chad and the Central African Republic have taken centre istics, which Zupančič et al. (2018) and, Murdoch and Sandler (2002) stage recently. The emergence of South Sudan can be ascribed to the surge claim facilities conflict spread across neighbouring borders. One feature in conflict between ethnic groups fighting for control of the new country, of the conflict zones was that they were all oriented in a north-west as well as Sudanese succession issues (Kulang & Ogbonna, 2021; direction, implying the merging of conflicts from the West, Central, Mohammed & Baba, 2021). Although more interventions are needed in East and North Africa, this is a major concern for sub-Saharan Africa. South Sudan, international bodies and peacekeeping agencies need to be Using the three-dimensional space-time cube analysis, notable con- commended for their support and interventions. flict hotspots in Africa derived were consecutive, oscillating, sporadic, The Seleka rebels, foreign operations, and resource control have all new hotspots while coldspot type was sporadic coldspot. Battles in contributed to an increase in mass killings and sexual violence in Central Nigeria, Burkina Faso, and Somalia were more common, as were ex- Africa Republic (Ahmat, 2018), while the conflict in Chad can be linked plosions/remote violence conflicts in Somalia, and violence against ci- to the rise of violent extremist religious groups partly due to the fall of vilians in Burkina Faso, Ghana, Nigeria, Somalia, South Sudan, and the Muammar Gadhafi regime in Libya and the subsequent rise of ISIS in the Democratic Republic of Congo. Poverty, ethnic diversity, access to Arab region (Mahdi, 2020). South Africa was undeniably the focal point health, pipe water, education and waste management were the driving of violence in southern Africa, which is not surprising given the coun- forces behind these consecutive hotspots. But at the subregional level, try’s high crime rate, violent protests, and xenophobic attacks (Lan- West Africa had factors such as subjective well-being, climate change, caster, 2018). Due to increasing conflicts since the Arab Spring, towns in and restrictions on press freedom which need to be considered in pre- Libya have become the focal point of conflicts in North Africa (Daw, dicting the occurrence of consecutive hotspots. New conflict hotspots on 2020). Protests in Africa have been centralized in Cameroon because of the continent were mainly because of poverty, religious diversity, 11 A. Osman et al. S o c ia l S c i e n c e s & H u m a n i t i e s O p en 8 (2023) 100557 Fig. 6. Space-time analysis of conflict for subregions from 1997 to 2021. 12 A. Osman et al. S o c ia l S c i e n c e s & H u m a n i t i e s O p en 8 (2023) 100557 restrictions on press freedom and rising unsentenced detainers. For East Africa, the influence of climate change enhanced the formation of new conflict hotspots. Manotas-Hidalgo et al. (2021) explained that economic challenges faced by most countries such as unemployment, falling wages and increasing poverty provided enough incentives for people to fight, protest and even join armed groups. Also, the growing entrenched po- litical divide in Africa is highly built along ethnic lines, colonial legacies and inequalities which produces high conflict-risk countries (Alesina & La Ferrara, 2000; Manotas-Hidalgo et al., 2021). The compounding ef- fect of low rainfall serves as a multiplier effect on all other causes of conflicts because of the high dependency of socioeconomic livelihoods on climate and rain-fed agriculture. Climate change is increasing competition among herdsmen and farming communities, generating low production, and increasing famine, ethnic clashes and migration (Sakaguchi et al., 2017). It is clear from the results of the space-time cube analysis that different hotspot patterns have different causes with different levels of influence; hence all conflicts cannot be lumped into hotspots and coldspots without the time dimensions. 6. Conclusions and recommendations The study found that central points of conflict tend to be fluid as they changed with time. The centrality of conflict points grow outwardly encompassing several communities and countries with an orientation favoured in the north-west, north, and east directions. Growing conflict zones were inversely related to the number of fatalities. Space-time cube revealed that conflict hotspots/coldspots have a time dimension with the dominant being consecutive, oscillating, sporadic, and new hotspots. Different hotspots had different explanatory variables. Consecutive hotspots are influenced by access to health, waste management, pipe water, education, ethnic diversity, and subjective well-being. For new hotspots, factors such as climate change, restrictions on press freedom, unsentenced detained, poverty and religious diversity were the major determinants. According to the study, the fluidity of conflict major points, needs continuous analysis and predictive analysis to aid peace- keepers in monitoring and providing fast response to save lives. Discovering hotspot patterns and causalities is critical for developing targeted actions to help combat Africa’s growing conflict threat. Peacekeepers should also be aware that different types of conflicts coexist and require different approaches at different times of response. 7. Areas for further studies To aid in peacekeeping missions, new studies could focus on fore- casting important locations of disputes. Machine learning techniques can be incredibly beneficial in predicting future conflict locations by analysing past conflict trends to predict future scenarios. This will aid security services in preparing and supplying the necessary staff to deal with conflicts. Furthermore, investigations with spatial structural equations on the interaction between conflicts and sustainable devel- opment goals can help understand the domino effects of conflict. CRediT authorship contribution statement Adams Osman: Conceptualization, Methodology, Software, Meth- odology, Software, Data curation, Visualization, Investigation, Formal analysis, Writing – original draft, Writing – review & editing. Alex Barimah Owusu: Conceptualization, Methodology, Software, Investi- gation, Formal analysis, Writing – original draft. Kofi Adu-Boahen: Methodology, Software, Data curation, Visualization, Writing – review & editing. Emmanuel Atamey: Methodology, Software, Data curation, Visualization, Investigation, Formal analysis, Writing – original draft. 13 Table 6 Determinates of space-time hotspots and coldspots per conflict types in Africa. Variables EB EE EP ER ES EV CHS NHS CHS NHS CHS NHS CHS NHS SHS CHS NHS OS SCS CHS NHS β β β β β β β β β β β β β β β Intercept 3.81 − 74.96b 134.97 − 247.63 − 29.88 − 66.11 − 31.38 5.47 5.59 37.63b − 58.98 58.16b 208.32 − 36.41b − 64.21b Ethnic Diversity − 0.09 − 0.25b − 2.75 2.12 − 0.87a 116.28 0.27 0.09 0.07 − 0.08 − 1.19a − 0.14 − 0.54 − 0.26b − 0.20b Corruption Perception Index 0.00 − 0.06 2.15 0.41 0.83 14.27 0.23 − 0.23 0.06 − 0.23 − 2.94a − 0.74b 4.27 − 0.23a − 0.22 Access to Universal Health 0.00 0.70b 4.42 − 3.97 0.25 226.01 0.04 0.09 0.09 0.02 2.49 − 0.29 − 10.56 − 0.45b − 0.38a Access to Pipe Water − 0.01 0.09 0.21 − 0.51 − 1.79b 7.15 − 0.05 − 0.03 − 0.14 − 0.52b 0.37 − 0.75b − 1.53 0.00 − 0.04 Net Primary School enrolment − 0.08 − 0.13 − 2.30 1.45 − 0.13 74.75 0.28 − 0.03 0.04 − 0.14 − 0.32 − 0.01 2.10 0.14b − 0.03 Municipal Solid Waste Management Challenges 0.19a 0.87b − 0.83 − 0.35 2.65b 914.69 0.16 0.16 0.06 0.06 0.72 0.18 0.47 0.40b 0.46b Access to Internet − 0.31a 0.15 − 1.54 3.20 − 1.28 117.83 0.29 − 0.02 0.26 1.06b 1.95 0.89b 1.63 0.43b 0.51b Poverty 0.26a − 0.32a − 0.48 0.73 0.59 − 266.06 − 0.36 − 0.14 − 0.30 − 0.50 − 2.43a − 0.49 1.20 − 0.05 0.17 Unemployment 0.20 0.74b 4.15 2.24 0.74 − 31.35 − 0.37 − 0.35 − 0.31 − 0.08 0.77 − 0.44b − 2.29 0.04 0.46b Unsentenced Detained − 0.22 − 0.24 0.03 − 1.31 0.46 14.63 0.22 0.13 0.13 0.14 − 0.27 0.24b 0.14 0.07a 0.19a Restrictions on Press Freedom − 0.18 − 0.37a − 0.18 1.74 − 0.12 − 5.13 − 0.24 0.09 0.07 − 0.06 0.75 0.51b 0.70 0.43b 0.36b Religious Diversity 0.20a 0.20a − 0.81 − 3.46 − 0.15 − 172.54 − 0.18 − 0.08 − 0.08 − 0.23 0.26 − 0.22 − 1.77 − 0.16b − 0.20b Climate change − 0.06 0.08 − 1.15 − 1.41 0.09 62.22 0.13 0.13 0.13 0.04 0.39a 0.04 − 0.64 0.16b 0.11a Subjective Well-being − 0.02 − 0.34 − 3.13 0.14 − 2.02b − 379.16 0.30 0.15 0.22 0.37b − 0.30 0.15 − 3.89 − 0.43b − 0.68b Model Fit − 2 Log Likelihood (Final) 547.9 29.6 150.3 473.3 352.8 761.5 McFadden 0.61 0.96 0.72 0.40 0.73 0.45 p < 0.01 = a, P < 0.05 = b, Reference category: Sporadic hotspot, CHS=Consecutive Hotspots, NHS=New Hotspots, OS=Oscillating Hotspot, SCS=Sporadic Coldspots, Battles (EB), Explosions/Remote violence (EE), Protests (EP), Riots (ER), Strategic developments (ES), Violence against civilians (EV). A. Osman et al. S o c ia l S c i e n c e s & H u m a n i t i e s O p en 8 (2023) 100557 Table 7 Determinates of space-time hotspots and coldspots per subregions in Africa. Variables North Southern Central East West CHS NHS CHS NHS CHS NHS CHS NHS OHS β β β β β В β β β β Intercept − 224.86b 1000.78 – − 84.62 − 878.78a − 198.23b − 250.23a 170.66b 77.92a 271.66b Ethnic Diversity 3.43b 35.54 – 0.87a 8.09a 0.33 1.52a − 0.58b − 0.13 − 0.25 Corruption Perception Index 0.37 − 19.53 – 1.20 − 9.31a − 2.92b − 2.52a − 1.16b 0.37 − 1.08a Access to Universal Health 2.31b − 98.17 – 1.77 − 8.32 − 0.60 3.59a − 2.35b − 3.11b − 4.17b Access to Pipe Water 1.13 − 31.82 – − 0.30 10.29a 1.52 2.39a 0.10 0.11 − 0.24b Net Primary School enrolment 0.20 − 118.12 – − 1.21b − 4.21a − 2.80b − 2.23b 0.26b 0.18b 0.49b Solid Waste Management Challenges 1.11 113.59 – 1.31b 6.68 0.31 0.91 − 2.20b − 1.28b − 3.35b Access to Internet − 1.83b − 107.24 – − 1.27 9.62 0.37 − 1.72 − 0.20 0.05 − 0.90b Poverty − 0.66 248.82 – − 0.36 − 4.07 − 4.28b − 5.58b 0.17 0.84b 0.64b Unemployment − 3.56b − 7.21 – 0.68 2.52 2.57b 2.74a − 0.84b − 0.67b − 1.88b Unsentenced Detained 0.17 − 53.12 – 1.54b 10.86b 0.93 − 0.06 0.35 − 0.59b 0.10 Restrictions on Press Freedom 2.67b − 104.62 – − 1.38 − 11.11a 2.30b − 0.78 1.07b 1.34b 2.31b Religious Diversity 0.45 12.16 – − 1.09b − 1.30 − 3.88b − 2.31a 0.38 0.10 0.22 Climate change − 0.08 26.67 – − 0.50a − 2.47b 2.06b 2.17b 0.30b − 0.16 0.12 Subjective Well-being − 0.75 24.13 – 1.87b − 6.19 − 0.54 − 1.26 0.97b 0.24 1.33b – Model Fit – 2 Log Likelihood (Final) 134.028 – 189.15 236.27 1985.37 McFadden 0.79 – 0.74 0.74 0.34 p < 0.01 = a, p < 0.05 = b, Reference category = Sporadic hotspot CHS=Consecutive Hotspots, NHS=New Hotspots, OHS=Oscillating Hotspots, North Africa (NA), West Africa (WA), East Africa (EA), Central Africa (CA) and Southern Africa (SA). Declaration of competing interest interests or personal relationships that could have appeared to influence the work reported in this paper. The authors declare that they have no known competing financial Appendix 1. Variables and their level of multicollinearity Variables Normalized Average VIF Climate Change Index 51.6 2.98 Civilians with Firearms 11.6 17 No. of Military Firearms 8.2 3.96 No. of Firearms for Law Enforcers 4.6 7.27 Diversity Ethnicity index 62.7 1.42 Good Governance Score 48.8 52.81 Rule of Law 49.5 33.32 Participation and Inclusion 46.2 46.25 Economic Opportunities 47.8 72.74 Poverty headcount ratio at $1.90/day 54.8 12.34 Poverty headcount ratio at $3.20/day 20.7 7.14 Prevalence of undernourishment 52.2 7.94 Prevalence of stunting in children under 5 years of age 39.1 7.65 Prevalence of wasting in children under 5 years of age 54.0 8.71 Human Trophic Level 62.3 13.67 Life expectancy at birth (years) 31.3 32.8 Universal health coverage (UHC) index of service coverage 17.0 1.11 Subjective well-being 26.5 2.42 Net primary school enrollment 65.4 3.1 Secondary school completion rate 39.1 5.39 Literacy rate (of population aged 15 to 24) 59.8 6.95 Seats held by women in national parliament 43.2 8.9 Population using at least basic drinking water services 47.3 11.2 Population using at least basic sanitation services 33.9 4.76 Scarce water consumption embodied in imports 96.8 9.3 Population with access to electricity 46.5 10.86 Population with access to clean fuels and technology for cooking 25.7 7.23 Adjusted GDP growth 51.1 3.61 Victims of modern slavery 66.6 6.34 Unemployment rate 69.6 1.79 Population using the internet 25.6 2.67 Mobile broadband subscriptions 36.2 1.91 Access to Piped water 69.8 2.21 Satisfaction with public transport 43.2 10.34 Solid waste management challenges 72.3 1.25 CO₂ emissions embodied in fossil fuel exports 99.8 6.87 Unsentenced detainees 52.8 2.55 (continued on next page) 14 A. Osman et al. S o c ia l S c i e n c e s & H u m a n i t i e s O p en 8 (2023) 100557 (continued ) Variables Normalized Average VIF Corruption Perception Index 26.4 2.63 Children involved in child labor 37.2 9.11 Exports of major conventional weapons 99.9 13.4 Restriction on press Freedom 61.7 1.22 Persons held in prison (per 100,000 population) 6.9 8.3 Religious diversity 54.2 2.28 References Kotsadam, A., & Østby, G. (2019). Armed conflict and maternal mortality: A a micro- level analysis of sub-Saharan Africa, 1989–2013. Social Science & Medicine, 239, Article 112526. Adaawen, S., Rademacher-Schulz, C., Schraven, B., & Segadlo, N. (2019). Drought, migration, and conflict in sub-Saharan Africa: What are the links and policy options? Kulang, T. T., & Ogbonna, C. C. (2021). South Sudan: The conundrum of a long-running social conflict. Insights into Economics and Management, 12, 9–16. Current Directions in Water Scarcity Research, 2, 15–31. Aghedo, I., & James Eke, S. (2013). From alms to arms: The Almajiri phenomenon and Kveladze, I., Kraak, M. J., & Van Elzakker, C. P. (2015). The space-time cube as part of a GeoVisual analytics environment to support the understanding of movement data. internal security in Northern Nigeria. Korean Journal of Policy Studies, 28. International Journal of Geographical Information Science, 29(11), 2001–2016. Ahmat, H. A. (2018). The involvement of foreign actors in ethno-political conflict: Case of the republic of Central Africa. Journal of International Trade, Logistics and Law, 4 Kveladze, I., Kraak, M. J., & van Elzakker, C. P. (2019). Cartographic design and the space–time cube. The Cartographic Journal, 56(1), 73–90. (2), 63–74. Alesina, A., & La Ferrara, E. (2000). Participation in heterogeneous communities. Lancaster, L. (2018). Unpacking discontent: Where and why protest happens in South Africa. South African Crime Quarterly, 64, 29–43. Quarterly Journal of Economics, 115(3), 847–904. Allison, S. (2020). Conflict is still Africa s biggest challenge in 2020. Retrieved from Conflict Li, X., Çöltekin, A., & Kraak, M. J. (2010). Visual exploration of eye movement data using ’ is still Africa’s biggest challenge in 2020 - World | ReliefWeb. the space-time-cube. In International conference on geographic information science (pp. 295–309). Berlin, Heidelberg: Springer. Bach, B., Dragicevic, P., Archambault, D., Hurter, C., & Carpendale, S. (2017). A descriptive framework for temporal data visualizations based on generalized Lis, P., Spagat, M., & Lee, U. R. (2021). Civilian targeting in African conflicts: A poor actor’s game that spreads through space. Journal of Peace Research, Article space-time cubes. Computer Graphics Forum, 36(6), 36–61. Barnes, T. J. (2011). Spatial analysis. The Sage handbook of geographical knowledge. Sage. 0022343320961150. Mack, E. A., Bunting, E., Herndon, J., Marcantonio, R. A., Ross, A., & Zimmer, A. (2021). Barnes, T. J. (2022). The discipline that came in from the cold: American human Conflict and its relationship to climate variability in Sub-Saharan Africa. Science of geography becomes a cold war social science. Environment and Planning F. Philosophy, Theory, Models, Methods and Practice, 1(2 4), 145 167. the Total Environment, 775, Article 145646. – – Mahdi, M. (2020). Dialogue with violent extremist groups: Community perspectives in the Lake Bass, C. A. (2017). Emerging hotspot analysis of Florida manatee (Trichechus manatus latirostris) mortality (1974-2012). Nova Southeastern University. Unpublished Chad Basin. African Portal Retrieved from https://www.africaportal.org/publicatio ns/dialogue-violent-extremist-groups-community-perspectives-lake-chad-basin/. Retrieved from https://nsuworks.nova.edu/cgi/viewcontent.cgi?referer=&httpsre Mann, H. B. (1945). Nonparametric tests against trend. Econometrica, 13, 245–259. dir=1&article=1458&context=occ_stuetd. Beetar, M. (2019). A contextualisation of the 2008 and 2015 xenophobic attacks: Tracing Manotas-Hidalgo, B., Pérez-Sebastián, F., & Campo-Bescós, M. A. (2021). The role of ethnic characteristics in the effect of income shocks on African conflict. World South African necropolitics. Current Sociology, 67(1), 122–140. Betty, E. L., Bollard, B., Murphy, S., Ogle, M., Hendriks, H., Orams, M. B., & Stockin, K. A. Development, 137, Article 105153. Mbih, R. A. (2020). The politics of farmer–herder conflicts and alternative conflict (2020). Using emerging hot spot analysis of stranding records to inform conservation management of a data-poor cetacean species. Biodiversity & Conservation, 29(2), management in Northwest Cameroon. African Geographical Review, 39(4), 324–344. Mohammed, A., & Baba, Y. T. (2021). Secession and border disputes in Africa: The case 643–665. of Sudan and South Sudan border. African Journal of Political Science and International Bruch, C., Jensen, D., Nakayama, M., & Unruh, J. (2019). The changing nature of conflict, peacebuilding, and environmental cooperation. Environmental Law Reporter Relations, 15(4), 131–138. Murdoch, J. C., & Sandler, T. (2002). Economic growth, civil wars, and spatial spillovers. & Analysis, 49, Article 10134. Burke, M. B., Miguel, E., Satyanath, S., Dykema, J. A., & Lobell, D. B. (2009). Warming Journal of Conflict Resolution, 46(1), 91–110. Nakaya, T., & Yano, K. (2010). Visualising crime clusters in a space-time cube: An increases the risk of civil war in Africa. Proceedings of the National Academy of exploratory data-analysis approach using space-time kernel density estimation and Sciences, 106(49), 20670–20674. Cabot, C. (2017). Climate change and farmer-herder conflicts in West Africa. In C. Cabot scan statistics. Transactions in GIS, 14(3), 223–239. Namakula, E. B. (2022). Rethinking United Nations peacekeeping responses to resource (Ed.), Climate change, security risks and conflict reduction. Springer. Daw, M. A. (2020). Corona virus infection in Syria, Libya and Yemen; an alarming wars and armed conflicts in Africa: Integrating African indigenous knowledge systems. Journal of Aggression, Conflict and Peace Research, 14(4), 320–333. devastating threat. Travel Medicine and Infectious Disease, 37, 101652. Ebiede, T. M. (2017). Beyond rebellion: Uncaptured dimensions of violent conflicts and Ngong, D. (2021). African political theology and post-nationalist nationalism: The case of the Cameroon Anglophone conflict. Political Theology, 1–17. the implications for peacebuilding in Nigeria’s Niger Delta. African Security, 10(1), O’Loughlin, J., & Raleigh, C. (2008). Spatial analysis of civil war violence. In The sage 25–46. Eckstein, D., Künzel, V., & Schäfer, L. (2021). Global climate risk index 2021. Who suffers handbook of political geography. Thousand Oaks, CA: Sage. Omorede, C. K. (2014). Assessment of the impact of oil and gas resource exploration on most from extreme weather events, 2000-2019. Berlin: Germanwatch e.V. Elischer, S. (2021). Militaries in sahelian politics: The oxford handbook of the African Sahel. the environment of selected communities in Delta State, Nigeria. International Journal of Management, Economics and Social Sciences, 3(2), 79–99. Oxford University Press. Perera, S. (2017). Burning the tent down: Violent political settlements in the Democratic Engels, B. (2015). Different means of protest, same causes: Popular struggles in Burkina Faso. Review of African Political Economy, 42(143), 92 106. Republic of the Congo. Journal of International Development, 29(5), 628–644. – Petrova, K. (2022). Floods, communal conflict and the role of local state institutions in ESRI. (2022). How create space time cube works—ArcGIS Pro. Retrieved from https://pro. arcgis.com/en/pro-app/latest/tool-reference/space-time-pattern-mining/learnmore Sub-Saharan Africa. Political Geography, 92, Article 102511. Prather, R. M., & Kaspari, M. (2019). Plants regulate grassland arthropod communities createcube.htm. Feldt, T., Karg, H., Kadaouré, I., Bessert, L., & Schlecht, E. (2020). Growing struggle over through biomass, quality, and habitat heterogeneity. Ecosphere, 10(10), Article rising demand: How land use change and complex farmer-grazier conflicts impact e02909. Purwanto, P., Utaya, S., Handoyo, B., Bachri, S., Astuti, I. S., Utomo, K. S. B., & grazing management in the Western Highlands of Cameroon. Land Use Policy, 95, Article 104579. Aldianto, Y. E. (2021). Spatiotemporal analysis of COVID-19 spread with emerging hotspot analysis and space–time cube models in East Java, Indonesia. International Griffith, D. A. (2021). Interpreting moran eigenvector mmps with the getis-ord Gi* statistic. The Professional Geographer, 73(3), 447 463. Journal of Geo-Information, 10(3), 133. – Raleigh, C. (2014). Political hierarchies and landscapes of conflict across Africa. Political Hägerstrand, T. (1970). What about people in regional science? Papers in Regional Geography, 42, 92–103. Science, 24(1), 6–21. Harris, N. L., Goldman, E., Gabris, C., Nordling, J., Minnemeyer, S., Ansari, S., Raleigh, C., Witmer, D. W., & O’Loughlin, J. (2018). The spatial analysis of war. Oxford … University Press. Potapov, P. (2017). Using spatial statistics to identify emerging hot spots of forest loss. Environmental Research Letters, 12(2), Article 024012. Ruggeri, A., Dorussen, H., & Gizelis, T. I. (2017). Winning the peace locally: UN peacekeeping and local conflict. International Organization, 71(1), 163–185. Johnson, S. D., & Braithwaite, A. (2017). Spatial and temporal analysis of terrorism and Sakaguchi, K., Varughese, A., & Auld, G. (2017). Climate wars? A systematic review of insurgency. The Handbook of the Criminology of Terrorism, 232–243. Kendall, M. G. (1955). Rank correlation methods. London: Griffin. empirical analyses on the links between climate change and violent conflict. International Studies Review, 19(4), 622–645. Kibris, A. (2021). The geo-temporal evolution of violence in civil conflicts: A micro analysis of conflict diffusion on a new event data set. Journal of Peace Research, Shimazaki, H., & Shinomoto, S. (2007). A method for selecting the bin size of a time histogram. Neural Computation, 19(6), 1503–1527. Article 0022343320978695. Simandan, D. (2016). Proximity, subjectivity, and space: Rethinking distance in human geography. Geoforum, 75, 249–252. 15 A. Osman et al. S o c ia l S c i e n c e s & H u m a n i t i e s O p en 8 (2023) 100557 Simandan, D. (2019a). Competition, delays, and coevolution in markets and politics. Thrift, N. (1996). Spatial formations (Vol. 42). Sage. Geoforum, 98, 15–24. Thrift, N. (2005). Torsten Hägerstrand and social theory. Progress in Human Geography, Simandan, D. (2019b). Iterative lagged asymmetric responses in strategic management 29(3), 337–340. and long-range planning. Time & Society, 28(4), 1363–1381. Van den Bosch, J., & Raubo, J. (2017). Armed conflicts in Africa: Fragile states and rising Simandan, D. (2020). Being surprised and surprising ourselves: A geography of personal instability. Przegląd Strategiczny, 10, 445–480. and social change. Progress in Human Geography, 44(1), 99–118. Zupančič, R., Pejič, N., Grilj, B., & Peen Rodt, A. (2018). The European Union rule of law Stearns, J. K. (2017). The democratic republic of the Congo: An elusive peace. In War and mission in Kosovo: An effective conflict prevention and peace-building mission? peace in Africa’s great lakes region (pp. 33–47). Cham: Palgrave Macmillan. Journal of Balkan and Near Eastern Studies, 20(6), 599, 61. 16