In the past decade, the number of people living within 100 km of active volcanoes surpassed the astonishing mark of 1 billion people, which means that around 14.3% of the entire world population is somehow exposed to volcanic hazards (Freire et al., 2019). Although there is a consensus in the scientific community that global volcanic activity is not increasing, the increasing population living in the vicinities of volcanoes, whose growth rate has exceeded the average growth rate of the global population since 1975 (Freire et al., 2019), requires a better understanding of volcanic processes, and greater effectiveness of volcanic risk assessment, communication and management.

In natural disaster research, risk can be defined as the result of the interaction of a given hazard (H) and the exposed elements (E), which are characterized by various dimensions of vulnerability (V) and levels of resilience (R) or, in other words, the combination of all losses potentially resulting from a particular natural disaster (UNDRO, 1991; UNISDR, 2004). The hazard factor concerns the characteristics of the natural phenomenon which, regarding volcanic events, may include a variety of products (e.g. tephra fallout, pyroclastic density currents - PDCs, lava flows, gas, lahars). The hazard assessment of a particular system must consider the occurrence frequency, spatial extension, and intensity of all potential hazards (e.g. Smith, 2013). The exposure factor is related to the number and distribution of people and assets located in a dangerous area, while their propensity to damage in case of a hazardous event is expressed by the vulnerability factor. The vulnerability is a dynamic concept that relies on the set of physical, social, economic and environmental characteristics of a system or a community that increases the associated susceptibility to the impact of a natural hazard (UNISDR 2004; UNISDR, 2009). While these three factors contribute to the risk, the resilience acts to reduce it, since it corresponds to the system’s ability to adapt to changes, overcome disturbances and maintain its functionality either by preserving pre-disaster conditions or tolerating and adapting to larger changes (e.g., Gunderson and Holling, 2002; Gaillard, 2007). Thus, the assessment of volcanic risk is complex as it depends on the analysis of multiple primary (e.g., tephra fallout, PDCs, lava flows) and secondary (e.g., lahars, tsunamis) hazards acting over various temporal and spatial scales (e.g., Connor et al., 2015), of the vulnerability of the exposed elements as well as of the resilience of each system to each of these hazards. Hence, a complete analysis of each factor is fundamental for an overall risk assessment (Aspinall and Blong, 2015). In regions characterized by a large number of active volcanoes such as Latin America and Indonesia, risk assessment of all systems might be impractical. In such a case, a risk ranking strategy that includes all main risk factors helps prioritize the volcanic systems that most require a formal risk assessment and/or the implementation of other risk reduction strategies (e.g., mitigation measures, early warning systems, emergency plans).

Ranking strategies of volcanic risk have been developed worldwide during the last decades with the purpose of identifying the active volcanic systems that represent the greatest threat either globally or locally and, therefore, prioritizing and targeting risk reduction strategies (e.g., Bailey et al., 1983; Lowenstein and Talai, 1984; Uhira, 2003; Magill and Blong, 2005a, Magill and Blong, 2005b; Ewert et al., 2005, Ewert., 2007, Ewert et al., 2018; Lara et al., 2006; Palma et al., 2009; Pereña et al., 2015; Scandone et al., 2016; Macedo et al., 2016; Santamaría and Bernard, 2018; Elissondo et al., 2016; Elissondo et al., 2017); a detailed review is presented in the companion paper (Nieto-Torres et al., 2021). Such rankings are especially useful in regions characterized by a large number of active volcanoes but limited resources (e.g., Latin America), enhancing the need of hierarchization and prioritization of investments. Additionally, the possibility of rapid update of the ranking (when new detailed information is made available) ensures a dynamic assessment of the risk in relation to changes of hazard, exposure and vulnerability and an adaptation of mitigation measures, allowing the reallocation of investments if and when needed.

The existing ranking strategies have already allowed significant advances in the knowledge and hierarchization of volcanic threats (e.g., updates in the American ranking–Ewert et al., 2005; Ewert et al., 2018; Ewert 2007–and in the Chilean ranking–Lara et al., 2006; SERNAGEOMIN 2020). Nonetheless, these rankings do not integrate information on vulnerability and resilience, which are key to a comprehensive characterization of the overall risk posed by volcanic systems and the identification of effective mitigation and response measures. Considering that risk reduction is the ultimate goal when working with natural hazards and their threats, it must be taken into account that although hazard impact can be mitigated, the volcanic event itself cannot be easily controlled or modified, and the exposure cannot be easily reduced by relocating infrastructure and established communities, which could be associated with large economic costs and social distress (e.g., Bowman and Henquinet, 2015; Barclay et al., 2019; De Natale et al., 2020; Iuchi and Matter, 2020). As a consequence, changes in hazard and exposure factors are hardly feasible. On the other hand, reducing vulnerability and increasing resilience are expected to be more effective and should be the first targets in risk reduction strategies. As a result, any risk ranking that aims at prioritizing systems that require the implementation of risk reduction strategies should also include vulnerability and resilience parameters.

In this context, this manuscript, together with the companion paper (Nieto-Torres et al., 2021), aim at developing an inclusive Volcanic Risk Ranking (VRR) method that accounts for key parameters and elements of all the four main risk factors (hazard, exposure, vulnerability and resilience). The first paper (Nieto-Torres et al., 2021) presents the rationale of the proposed method in detail, while this work discusses its application to the volcanic systems in Latin America with activity reported for the last 1,000 years.

The Latin America region represents a great study opportunity; hosting an extensive volcanic arc on its western edge, the continent is characterized by a large number of active volcanoes (almost 300 systems with Holocene activity) coupled with large populations exposed to the associated hazards (Figure 1). Around 10% of the South American population and 70% of the population of México and Central America live within 100 km of an active volcano (Brown et al., 2015; Freire et al., 2019). The volcanism in this region is marked by compositional and morphological diversity (from stratovolcanoes to monogenetic volcanic fields and lava domes), as well as by the variation and alternation of eruptive styles. A dozen large events (VEI ≥5) have occurred in the region in the past 1,000 years (Figure 2) which, together with more recent events of lesser size, demonstrate the potentially devastating consequences of volcanic activity on modern societies (e.g., Chaitén [2008–2010], Arenal [2010], Calbuco [2015], Colima [2017], Sabancaya [2016–2018], Masaya [2018], Fuego [2018]). In particular, the 1985 eruption of Nevado del Ruiz (Colombia), El Chichón in 1982 (México), Santa María (Guatemala) and Mt Pelée (Martinique–here mentioned due to its location and consequent genetic relation to the studied systems) in 1902 are noteworthy since they account for about 3/4 of the deaths related to volcanic events in the 20th century and are responsible for Latin America’s second place in the ranking of regions with the highest volcanic lethality (Tilling, 2009; Delgado-Granados et al., 2015).

Regardless of the shared official language, the socio-cultural affinity, the increased knowledge of local volcanic systems and the increasing number of volcano observatories and of volcanoes monitored in the region, the large number of volcanoes on borders, the scarcity of economic resources, the different social vulnerabilities faced by different countries and the lack of systematized and integrated information still prevent the risk associated with the volcanism in the Latin American region to be fully assessed (Delgado-Granados et al., 2015). We expect the outcomes of this work to contribute to the optimization of available resources and to the development of strategic lines of action for Disaster Risk Reduction in the region.

The volcanism in Latin America is related to the Ring of Fire, also called the circum-Pacific Belt, a 40,000 km long horseshoe-shaped active belt defined by the tectonic plates’ boundaries that fringes the Pacific Ocean. Around 75% of the Earth’s volcanism and 90% of Earth’s earthquakes occur along its path.

In México, most of the active volcanoes are located along the Transmexican Volcanic Belt, a Miocene-Holocene continental volcanic arc nearly E-W trended, around 1,000 km in length, extending from the Mexican Pacific coast to the Gulf of México, and of irregular width varying between 80 and 230 km (Gómez-Tuena et al., 2007). This volcanic arc transversely overlaps NW-SE-trended mountain ranges such as the Sierra de California, Sierra Madre Occidental, Sierra Madre de Chiapas and the Chiapaneco Volcanic Arc, which also encompass some volcanic systems with Holocene activity (either confirmed or estimated) (Ortega-Gutiérrez et al., 1992; Macías, 2005; Gómez-Tuena et al., 2007; Mora et al., 2007). The Transmexican Volcanic Belt results from the interaction of the Pacific and Cocos plates with the North American and Caribbean plates since the Cretaceous. The migration and consequent overlap of the magmatic arcs represents the response to variations in the interaction (direction and angle of the subduction slab) between these plates (McDowell and Clabaugh, 1979; Damon et al., 1981; Gómez-Tuena et al., 2007; Cadoux et al., 2011; Mora et al., 2012).

The Central American volcanic front (CAVF), a result of the subduction of the Cocos plate beneath the Caribbean plate since the Early Miocene, is around 1,100 km in length. It extends through the whole isthmus, from the México-Guatemala border to Panamá, with a volcanic gap around 200 km long from the SE Costal Rica to the W Panamá. The volcanic range is composed of 50 major volcanic centers and hundreds of other small vents aligned along 7 main segments more or less parallel to the coast (Stoiber and Carr, 1973; Alvarado et al., 2012). The segments’ boundaries are marked by the offset of the chain of active volcanoes; minor alignments with small active volcanic vents may occur transversely to the main chain (Stoiber and Carr, 1973). Local tectonic trends control the volcanic systems’ dynamics and the systems with larger magnitude events or more continuous activity are preferentially located near the segments’ boundaries (Stoiber and Carr, 1973; Alvarado et al., 2012). At least 26 volcanic systems are considered active, i.e., have confirmed eruptive events in the last 10,000 years.

In South America, the Cordillera de Los Andes represents one of the highest and largest (ca. 8,000 km length) subaerial mountain ranges in the planet, whose formation began during the late-Paleozoic (Tilling, 2009). Since the late-Triassic the region has also hosted the largest, although discontinuous, continental volcanic arc whose magmatism is related to the subduction of the Nazca and Antarctica Plates beneath the South America Plate. The volcanic arc can be divided into four segments: the Northern Volcanic Zone (NVZ) from Colombia to Ecuador, the Central Volcanic Zone (CVZ) along southern Peru and Northern Chile, the Southern Volcanic Zone (SVZ) extending from central do southern Chile and finally the Austral Volcanic Zone (AVZ), along the southernmost region of the continent. These segments are separated from each other by volcanically inactive gaps that may be a result of changes in the slabs dip (Barazangi and Isacks, 1976; Pilger, 1984).

The Latin America volcanic arc host around 21% of the volcanoes classified as active during the Holocene by the Global Volcanism Program (GVP) from the Smithsonian Institution (GVP Holocene Volcano List, 2013), and 8 of the 20 top countries with the highest overall volcanic threat defined by Brown et al. (2015) based on the number of active volcanoes, the exposed populations and the average hazard index. The volcanism comprises a wide range of compositions, from basalts to rhyolites, with prevailing intermediate compositions and is mainly characterized by high, steep-sided volcanoes, commonly capped by ice or snow.

According to the GVP database, 123 volcanoes have eruptions reported during the last 1,000 years in Latin America, of which 13 are located in México (1 on the border with Guatemala), 7 in Guatemala, 5 in El Salvador, 10 in Nicaragua, 6 in Costa Rica, 1 in Panamá, 10 in Colombia (1 on the border with Ecuador), 20 in Ecuador, 8 in Peru, 41 in Chile (2 and 5 are located on the borders with Bolivia and Argentina, respectively) and 2 in Argentina (Figure 3); the majority of them (∼67%) are stratovolcanoes. The shifts between eruptive styles, common in the volcanoes of the region, may bring, however, some complexity to the hazard assessment.

The VRR method described in the companion paper by Nieto-Torres et al. (2021) and applied here to the Latin America volcanic systems with activity in the last 1,000 years builds on the parameters proposed by Ewert et al. (2005), Ewert (2007), Ewert et al. (2018), but also includes vulnerability and resilience factors. We consider two ranking strategies in order to assess the risk:

1) Assessment of the contribution of the factors that build the risk (a 3-factor VRR from now on referred to as VRR (1)):

It is important to stress that a risk ranking is not a hazard or a risk assessment. As a result, the hazard parameters and elements considered in the proposed method relate exclusively to volcanological features and to the potential phenomena associated with a certain system; however, they do not include probabilistic assessments. Systems that present multiple hazards are assigned higher scores, meaning higher hazard. In total, 9 hazard parameters (Table 1) are considered, of which 4 correspond to volcanic products or unrest phenomena, whose score varies between 0 (non-occurrence) and 1 (occurrence). The parameters and their score ranges are (between square brackets): volcano type [0–1], maximum VEI [0–4], recurrence rate [0–4], type of volcanic hazard [0–5 depending on the occurrence of Holocene PDCs, lava flows, lahars, volcano-related tsunami and the hydrothermal explosion potential], sector collapse potential [0–1], existence of primary lahar source [0–1], seismic unrest [0–1], ground deformation [0–1], active fumarole or magmatic degassing [0–1]. The assessment of the various parameters considered here relies primarily on the GVP volcano database, being modified or updated when necessary with local research institution data (Table 2). It is important to highlight that, although the present work evaluates only volcanic systems with confirmed eruptions in the last 1,000 years, the analysis considers the entire Holocene activity record.

In terms of exposure, the 9 parameters (Table 1) investigated here comprise data on the population, residential buildings, transportation, power, water and telecommunication infrastructure, emergency and critical facilities as well as economic activities exposed to the hazard within four radii: 5, 10, 30 and 100 km. These radii cover the areas most susceptible to the impact of the different types of hazards considered in this study and are in agreement with the limits used by the GVP for population data. In case of volcanic fields, the exposure is analyzed for assets inside the volcanic field and up to 5, 10 and 30 km from the field’s boundary, which is defined by the connection of the outermost volcanic edifices that comprise it. Population data is here evaluated based on density (inhab/km²), i.e. the ratio between the number of inhabitants (according to the GVP database, which is based on the LandScan produced by the Oak Ridge National Laboratory) and the area, considering the whole surface area of each of the radius thresholds previously defined. As a result, volcanic systems without exposed population score 0, while volcanic systems with population density higher than 0 and lower than 1 inhab/km² score 1, with population density between 1 and 10 inhab/km² score 2, between 10 and 100 inhab/km² score 3 and above 100 inhab/km² score 4 (Table 1). This density-based score system was chosen considering that the population density concept is more easily and quickly understood, simplifying the communication with stakeholders. While comparable with the logarithmic scale of previous rankings (e.g., Ewert et al., 2005; Lara et al., 2006; Ewert., 2007; Pereña et al., 2015; Elissondo et al., 2016, 2017), it allows fitting maximum values that facilitate the normalization of the final score and the comparison between different databases. A similar scoring system rationale is used for residential buildings and critical infrastructures, with values ranging from 1 (when the first occurrence appears within 100 km radius from a central volcano or 30 km from the border of a volcanic field) to 4 (first occurrence within the 5 km radius threshold and, in the case of volcanic fields, inside the field). The critical infrastructure considered in this work encompass: 1) transportation, considering either point or linear infrastructure such as airports, air routes, harbors, and primary road and rail networks; 2) power infrastructure, including power generation (power plants), transmission and distribution lines; 3) water supply and wastewater infrastructure; 4) telecommunication; 5) emergency facilities (e.g., hospitals, fire brigades, police offices and military bases) 6) critical facilities (e.g., government offices, educational and scientific institutions, museums, and recreation sites), and 7) economic activities (e.g., agriculture, mining, factories, tourism) (Table 1). The data on the elements at risk were obtained from a combination of sources and databases (Table 3), including the GVP from the Smithsonian Institution, the LandScan produced by the Oak Ridge National Laboratory, local Institutions (Table 2), the Humanitarian OpenStreet Map, the United Nations divisions (e.g., HDX-OCHA and UNESCO), the Food and Agriculture Organization (FAO) and the World Bank database. It must be noted that the assessment of volcanic systems located on geographic borders calculate a unique score that considers the exposed elements of the countries (or administrative units) encompassed by each of the radius thresholds. In addition, a hierarchy of the emergency and critical facilities (e.g., hospitals, schools, fire stations, etc.) was not set, i.e., the existence of at least one asset from any of the considered categories is sufficient for scoring (for example, non-existence or lack of data of government offices but existence of schools results in a score).

A certain number of vulnerability parameters are considered in this work, encompassing four dimensions: 1) physical vulnerability, which describes the fragility of residential buildings and infrastructures, 2) systemic vulnerability, which accounts for the possibility of cascading propagation of impact through a system and for the interdependency of infrastructures and systems, 3) social vulnerability, assessed here through the indicators of population age, disability, ethnicity, unemployment rate and education level, and 4) economic vulnerability, based on the existence of multiple economic activity within the analyzed radius thresholds. In total, 10 parameters were taken into account for the vulnerability factor (Table 1). The physical vulnerability analysis is based on the predominant (more than 50%) typology, or the material used of residential buildings. For Mexican and Central American systems, the analysis considered the type of materials used (wood scores 3, masonry scores 2 and reinforced concrete scores 1) obtained from Google Street View observations combined with the Open Street Map database, while for the South American systems, the analysis was based on the latest versions of local census surveys, which provided the typology of residential buildings, according to local terminology. “Choza, covacha, rancho, cabaña, vivienda móvil, vivienda indígena o étnica, local no construído para habitación humana, vivienda impovisada, otros tipos de vivenda” were taken as most fragile buildings, resulting in higher scores [3], followed by progressively less fragile structures as “mediaguas and casillas” [2] and “cuarto, pieza, casa, vivienda en quinta, vivienda en casa de vecinidad, vila, departamento” [1]. The systemic vulnerability analysis encompasses the proximity of the volcanic system to a geographic border, the lack of redundancy and restricted accessibility to the critical infrastructure identified in the exposure analysis. In particular, by considering the proximity to geographic borders as a vulnerability factor parameter, our strategy addresses the challenges faced by the border systems, such as the direct effects of tephra fallout, as in the already mentioned case of Chile and Argentina, as well as the closure of customs posts, roads and airways, etc. Here, 2 points are scored if the volcanic system is located within 10 km from the border between two or more countries and 1 point if within 100 km. Regarding the lack of redundancy and restricted accessibility, since these conditions in larger areas increase the vulnerability of the systems, score 1 is given in case of non-redundancy and restricted accessibility within a 100 km radius from a central volcano or up to 30 km from a volcanic field’s boundary; if the critical infrastructure is redundant and most of them easily accessed, no scores are counted. The social vulnerability, a complex concept that relies on the idea that local social conditions and some characteristics of a person influence their ability to anticipate, manage, resist and recover from the damage caused by a hazard (Wisner et al., 2004), is assessed through various indicators: age (divided into two groups considered most vulnerable, i.e., children <5 years old and elderly ≥60 years old), disability, ethnicity, unemployment rate and education level (based on the proportion of illiterate population) (e.g., Wisner et al., 2004; Aceves-Quesada et al., 2007; Cutter., 2013). The social indicators were counted as the arithmetic average of data for each administrative unit (province, region or department) inside the 100 km radius threshold; with scores ranging from 0 to 10, where 10 corresponds to 100% of the population in the considered condition. The larger the fraction of population in the considered condition, the higher the vulnerability of the system. This analysis was based on the latest versions of population census data from local institutions (Table 2). Finally, the analysis of the economic vulnerability considers the existence of multiple economic activities, applying a regressive scoring ranging from 4 (if single economic activity within a 5 km radius from a central volcano or inside a volcanic field) to 1 (if single economic activity within a 100 km radius from a central volcano or 30 km radius from a volcanic field’s boundary) and was based on multiple sources of data such as local institutions, the Humanitarian OpenStreet Map and the United Nations divisions (e.g., OCHA). As was the case for the exposure parameters, the assessment of the volcanic systems on borders set a unique score considering the vulnerabilities of the countries (or administrative units) encompassed within each of the radius thresholds.

The analysis of the resilience factor considers the existence of a series of mitigation and response measures that anticipates the threat and enables disaster management, reducing the likelihood of injury, death, and property damage (total of 13 parameters analyzed; Table 1). These measures are divided into existing strategies and tools to reduce risk, including hazard maps, impact/risk assessment studies, engineering mitigation structures, hazard-based land use planning, capacity of monitoring system, early warning systems and insurance coverage, and elements for preparedness, such as educational activities for the population, exercises and simulations for operational institutions (involved in emergency management, e.g., civil protection) and exercises and simulations for the population. Each existing element adds 1 to the score of the system, with the exception of the hazard maps that score 1 in case of single hazard map or 2 in case of multiple hazard map (or hazard maps with multiple hazards), the risk assessment studies whose availability also scores 2 and the monitoring system, whose score can vary from 4 (in case of full real time monitoring) to 0 (absence of monitoring). Finally, the response measures consider the existence of actions to be implemented during a crisis; a score 1 is attributed in case of existence of evacuation plan, occurrence of successful evacuation during past eruptions and availability of shelters. Such data were obtained from public databases from or direct request to the responsible institutions (geological surveys, volcanic observatories, and civil defense).

Finally, all the data were integrated and processed using the software ArcGIS 10.1. For more details on all the analyzed elements and their scoring system, as well as the sensitivity analyses that supports the choice of the risk equation, the reader is referred to the companion paper by Nieto-Torres et al. (2021).

Our analysis focused on the Latin American volcanoes that had eruptions in the last 1,000 years and the results are presented together within the regional context for a general analysis of the Latin America VRR. Results and diagrams for comparison with previous rankings (when available), can be found in the supplementary materials (Supplementary Material 2–8) divided by country, with the exception of Argentinean and Chilean volcanoes that are presented together. The exception for those volcanoes in the southern portion of the continent is due to the fact that Argentina and Chile share a large part of the region’s volcanic systems and their hazards. Indeed, the most active volcanoes in South America are located on geographic borders; of the 81 South American volcanoes analyzed in this work, 10 are located on or very close to (<10 km) the Argentina-Chile border. In addition, considering that Chile is a N-S elongated country with volcanic systems located closer to its eastern portion, the maximum radius threshold applied in our analysis (100 km) includes areas of neighboring countries, especially Argentina which, due to the prevailing wind conditions in the region (towards E), is significantly affected by tephra fallout and volcanic gas generated by Chilean eruptions.

The most hazardous volcanoes are mainly concentrated in the northern (Ecuador, Costa Rica, Guatemala, México and El Salvador) and southern (Chile-Argentina) regions and correspond to (in descending order of hazard): Tungurahua (Ecuador), Santiaguito (Guatemala), Cotopaxi (Ecuador), Irazú (Costa Rica), Reventador (Ecuador), Popocatépetl (México), Tacaná (México), Fuego (Guatemala), Santa Ana (El Salvador), Rincón de la Vieja (Costa Rica), Cayambe (Ecuador), Guagua Pichincha (Ecuador), Sangay (Ecuador), Cerro Azul-Quizapu (Chile-Argentina), Cerro Hudson (Chile-Argentina), Llaima (Chile-Argentina) and Villarrica (Chile-Argentina) (Supplementary Material 1, Figure 4). Most of the volcanoes with the highest score are also noted for their higher eruptive recurrence.

In terms of exposure, the volcanic systems from the northern region stand out, especially in Central America (El Salvador, Nicaragua, Guatemala and Costa Rica), in addition to México, Ecuador and Peru, some of which are among the most hazardous already identified (e.g., Santiaguito, Fuego, Irazú and Guagua Pichincha). The highest rated volcanoes are, in decreasing order: San Salvador (El Salvador), Ilopango (El Salvador), Apoyeque (Nicaragua), Almolonga (Guatemala), Andahua-Orcompampa (Peru), Nejapa-Miraflores (Nicaragua), Jocotitlán (México), Santiaguito (Guatemala), Concepción (Nicaragua), Guagua Pichincha (Ecuador), San Miguel (El Salvador), Fuego (Guatemala), Masaya (Nicaragua), Chacana (Ecuador), Irazú (Costa Rica) and Galeras (Colombia) (Supplementary Material 1, Figure 5). It is worth mentioning the high population density of the volcanic systems in these regions (even in proximal areas); such high population densities in proximal areas are characteristic of Central American countries and Colombia, where most of the volcanic systems have high scores (i.e., 3 to 4 points) for the smallest radius thresholds (5 to 10 km).

The vulnerability analysis may be, in some extent, biased by the local lack of data, especially in relation to critical infrastructure (i.e., telecommunication infrastructure for Colombia, Peru and Chile). In any case, the most vulnerable areas are concentrated mainly in the southern region (Chile-Argentina), although some areas in Ecuador and México are also among the most vulnerable. The highest scores are mainly due to the typology of residential buildings (in the case of the Mexican systems), proximity to borders, the restrict accessibility to critical infrastructure, and to social vulnerability (especially related to the disabled, elderly and, subordinately, unemployed population). Thus, the most vulnerable areas are located around volcanic systems, in decreasing order: Tacaná (México-Guatemala), El Chichón (México), Llullaillaco (Chile-Argentina), Tutupaca (Peru), Isluga (Chile-Argentina), Guallatiri (Chile-Argentina), Reclus (Chile-Argentina), Yucamane (Peru), Mocho-Choshuenco (Chile-Argentina), El Misti (Peru), Sabancaya (Peru), Huaynaputina (Peru), Chaiten (Chile-Argentina), Putana (Chile-Argentina), Quetrupillan (Chile-Argentina), Ticsani (Peru) and Ubinas (Peru) (Supplementary Material 1, Figure 6).

In terms of resilience, some similarities between the different countries were identified. In addition to the existence of national plans for disaster risk management, in general the most hazardous volcanic systems in each country are those where mitigation and response measures have been implemented, increasing the resilience of surrounding areas, such as risk maps, real-time monitoring, educational activities, simulation exercises and emergency plans. The most resilient areas are those around Arenal (Costa Rica), Copahue (Chile-Argentina), Ubinas (Peru), Villarrica (Chile-Argentina), Poás (Costa Rica), Irazú (Costa Rica), Cotopaxi (Ecuador), Tungurahua (Ecuador), Llaima (Chile-Argentina), Turrialba (Costa Rica), Rincón de la Vieja (Costa Rica), Nevado del Ruiz (Colombia), Calbuco (Chile-Argentina), and Puyheue-Cordon Caulle (Chile-Argentina), Popocatépetl (México) and Planchon-Peteroa (Chile-Argentina) (Supplementary Material 1, Figure 7).

When considering the VRR (1) (3-factor VRR: H×E×V), the highest scored systems are well distributed through the whole Latin America region, with emphasis on the Ecuadorian and Chilean-Argentinian volcanic systems that cover, together, 10 of the 20 volcanic systems of the highest VRR scores, followed by Guatemala, México, El Salvador, Peru and Nicaragua. In descending order of risk, considering the VRR (1), the 20 highest scored volcanic systems are: Santiaguito (Guatemala), Tacaná (México-Guatemala), Fuego (Guatemala), San Salvador (El Salvador), El Chichón (México), Ubinas (Peru), El Misti (Peru), Cotopaxi (Ecuador), Guagua Pichincha (Ecuador), Concepción (Nicaragua), Quilotoa (Ecuador), Reventador (Ecuador), Villarrica (Chile-Argentina), Santa Ana (El Salvador), Tungurahua (Ecuador), Ceboruco (México), Llaima (Chile-Argentina), Chaitén (Chile-Argentina), Lonquimay (Chile-Argentina), and Carrán-Los Venados (Chile-Argentina) (Table 4; Figure 8). The lowest scores, on the other hand, are related to the volcanic systems with no (Bárcena and Pinta) or low exposure (Ecuadorian island volcanoes). It is important to note that most Ecuadorian island volcanoes (Alcedo, Cerro Azul, Darwin, Ecuador, Fernandina, Marchena, Santiago and Wolf) present low exposure due to a few exposed elements within the highest radii thresholds. In addition, Fernandina, Marchena and Santiago are located on different islands and, as it is also the case of the other volcanic systems, their activity is mostly associated with small-extension lava flows, that, therefore, do not represent a significant threat. This is also the case of Sierra Negra, the Ecuadorian island volcano which is the closest to an inhabited area; the geological record associated with this volcano also shows some explosive activity but of low intensity that, due to the prevailing wind directions at the region (towards W-NW), also do not represent a major threat.

VRR is significantly different when resilience measures are also considered. In fact, according to the 4-factor VRR (i.e., VRR (2): H×E×V/(Res+1)), the 20 volcanic systems with the highest score are (in descending order): Atitlán (Guatemala), Almolonga (Guatemala), Tecuamburro (Guatemala), Michocán Guanajuato VF (México), Putana (Chile-Argentina), Izalco (El Salvador), Lluillaillaco (Chile-Argentina), San Martín Tuxtla (México), Santiaguito (Guatemala), Ilopango (El Salvador), Tacaná (México-Guatemala), Acatenango (Guatemala), Xalapa-Naolinco VF (México), Reclus (Chile-Argentina), Ceboruco (México), Puntiagudo-Cordon Cenizos (Chile-Argentina), Chacana (Ecuador), El Chichón (México), Sumaco (Ecuador) and Jocotitlán (México) (Table 5; Figure 8). VRR (2) sheds light on the importance of the resilience factor for risk reduction, with emphasis on the Ubinas (Peru), Cotopaxi (Ecuador), Villarrica (Chile-Argentina), Tungurahua (Ecuador), Llaima (Chile-Argentina), Puyehue-Cordon Caulle (Chile-Argentina) and Copahue (Chile-Argentina). Some of these volcanic systems are among those in top positions of the VRR (1) but rank lower in VRR (2) because of implemented resilience measures. The volcanic systems with the most risk reduction strategies in place (e.g., hazard and risk assessments, monitoring and early warning systems, educational activities, evacuation plan, etc.), or in other words with highest resilience, are Arenal (Costa Rica), Copahue (Chile-Argentina), Ubinas (Peru), Villarrica (Chile-Argentina), Poás (Costa Rica), Irazú (Costa Rica), Cotopaxi (Ecuador), Tungurahua (Ecuador), Llaima (Chile-Argentina), Turrialba (Costa Rica), Rincón de la Vieja (Costa Rica), Nevado del Ruiz (Colombia), Calbuco (Chile-Argentina), and Puyheue-Cordon Caulle (Chile-Argentina) (Figure 7). The case of the Nevado del Ruiz volcano deserves attention; although this volcano displays a moderate risk level at the Latin America scale (VRR 1) = 90.61/32nd position), it is among those characterized by the highest level of resilience measures implemented (e.g., risk assessment, early warning systems, hazard-based land use planning, evacuation plan; VRR (2) = 11.65/83rd position), demonstrating the importance of the lessons learned from the tragic event occurred in 1985. In contrast, the volcanic systems Ceboruco (México), Quilotoa (Ecuador), Santiaguito (Guatemala), Tacaná (México-Guatemala), El Chichón (México) and Santa Ana (El Salvador) stand out as those of high to moderate risk with the fewest risk reduction strategies in place and, consequently, highest score in VRR (2).

VRR analyses at national scale have also been performed. Figure 9 presents the example of Nicaragua together with the comparison with the previous ranking analysis (Palma et al., 2009) based on the method of Ewert et al. (2005, Ewert., 2007); all other national analyses are presented in Supplementary Material 2–8, with the exception of Panama that only has one volcanic system, and México that has been analyzed in detail by Nieto-Torres et al. (2021). First, the national map shows a heterogeneous distribution of the 4 risk factors, even though all volcanic systems are associated with some level of resilience measures implemented. Second, it is interesting to notice that our 3-factor VRR (1) (blue histograms) is very different with respect to the threat ranking of Palma et al. (2009) as well as the threat ranking based on the hazard and exposure analyses presented here (pale-pink circles), with Conception being associated with the highest score. Finally, the VRR (2) highlights the volcanic system with the fewest resilience measures (i.e., Cosiguina).

Volcanic risk analyses are complex as they need to consider multiple primary and secondary hazards that act at multiple spatial and temporal scales as well as multiple dimensions of vulnerability and various aspects of exposure and resilience. The VRR strategies proposed in the companion paper by Nieto-Torres et al. (2021) and applied here to the Latin America volcanic systems with eruptive record in the last 1,000 years, are based on two different equations (VRR (1) and VRR (2)) that combine 3 (hazard, exposure, vulnerability) and 4 key risk factors (hazard, exposure, vulnerability and resilience), respectively, aiming at classifying volcanic systems in order to prioritize risk reduction measures. In addition to highlighting the importance of considering vulnerability and resilience factors when ranking risk besides hazard and exposure, the proposed VRR strategies provide a valuable opportunity for identifying weak aspects of densely inhabited areas located close to active volcanic systems, thus helping ensure means for planning coordinated actions to reduce risk. It is also found that both VRR (1) and VRR (2) allow for a more pronounced diversification of ranking scores with respect to the volcanic threat ranking proposed by Ewert et al. (2005) and Ewert (2007) and applied in various countries in Latin America (see Figure 9 and Supplementary Materials 2–8). A more pronounced diversification of score helps better identify the volcanic systems that require intervention. In addition, as hazard and exposure are difficult to modify and reduce, the main actions for risk mitigation rely on the reduction of vulnerability (by creating redundancy and accessibility to vital infrastructure and reducing social inequalities that reflects on physical, social and economic vulnerability aspects) and increasing resilience. In particular, resilience can be increased through risk assessment studies, elaboration of hazard-based land-use planning (which also contributes to reduce or at least control exposure), implementation of early warning systems, emergency plans and protocols as well as educational activities and simulations to improve risk perception and response capacity. Hence the importance of including vulnerability and resilience in a comprehensive VRR.

Based on this method, among the 123 volcanoes analyzed for Latin America, the volcanoes Santiaguito (Guatemala), Tacaná (México-Guatemala), Fuego (Guatemala), San Salvador (El Salvador), El Chichón (México), Ubinas (Peru), El Misti (Peru), Cotopaxi (Ecuador), Guagua Pichincha (Ecuador) and Concepción (Nicaragua) are those with the highest score in VRR (1) (H×E×V); in contrast, the island Ecuadorian volcanoes (Ecuador, Marchena, Santiago, Darwin, Fueguino, Wolf and Alcedo) are among the systems with the lowest score in VRR (1) (in addition to Bárcena and Pinta volcanoes for which exposure is zero). Hazard and exposure are the main factors influencing these results. Exposure is here especially influenced by population densities, which is either high even in proximal areas (as is the case of Santiaguito, Tacaná, Fuego, San Salvador, Concepción and Quilotoa) or zero to low, which is the case of the Mexican and Ecuadorian island systems that correspond to the systems with lowest score in VRR (1). Although vulnerability is not the main factor controlling the results presented here, its contribution through the typology of buildings and restricted accessibility to vital infrastructure deserves attention.

The VRR (2), on the other hand, highlights the influence of resilience on the risk score, with those systems with few mitigation and response measures implemented being in the top positions. The volcanoes Atitlán (Guatemala), Almolonga (Guatemala), Tecuamburro (Guatemala), Putana (Chile-Argentina), Izalco (El Salvador), Lluillaillaco (Chile-Argentina), San Martín Tuxtla (México), Santiaguito (Guatemala) and Ilopango (El Salvador) are those with highest score in VRR (2). It is important to notice that VRR (1) and VRR (2) is zero at Bárcena and Pinta volcanoes as there is no exposure and that the systems that occupy the lowest positions in VRR (1) (Ecuador, Marchena, Santiago, Wolf, Fernandina, Darwin and Alcedo) maintained their bottom positions even with low or no resilience; this is mostly due to low exposure. It is worth also noting that some systems with moderate to high VRR (1) scores (e.g., Santiaguito, Tacaná, El Chichón and Ceboruco) maintained moderate to high positions in the VRR (2), highlighting the need to improve their resilience in order to reduce risk.