Ontologies and zoonotic risk-related knowledge, attitudes and practices: a cross-sectional survey in Bolivia, Brazil, Chile and Guatemala Authors: Katja Radon, Caroline Kuhn, Carlos Fernando Pinto Navia, Carlos Roberto Vásquez-Almazán, Denise Siqueira de Carvalho, Luciana Salini Abrahão Pires, Magda Clara Vieira da Costa-Ribeiro, Marcia Adler, María Soledad Burrone, Maria Teresa Solís-Soto, on behalf of KAP Wildlife project team Published: 2026 ABSTRACT Abstract AbstractObjectivesTo examine whether Indigenous Peoples’ and Local Communities’ (IPLC) ontologies are associated with knowledge, attitudes and practices (KAP) related to wildlife cohabitation and zoonotic disease transmission in biodiversity-rich areas of Latin America.DesignCross-sectional household survey using a standardised KAP questionnaire. Ontologies were classified using latent class analysis. Associations between ontology classes and outcomes were assessed using multivariable logistic regression models.SettingUrban, rural and protected areas in biodiversity-rich regions of Bolivia, Brazil, Chile and Guatemala.ParticipantsA total of 2903 individuals aged ≥10 years were recruited through random household sampling (response rate 85%).Primary and secondary outcome measuresPrimary outcomes were defined according to the KAP framework. Knowledge outcomes comprised combined knowledge of zoonotic disease transmission from wildlife to humans and knowledge of zoonotic risks associated with wildlife trade. Perceived training needs related to zoonotic disease prevention were analysed as a secondary knowledge outcome measure. Attitudes were measured through risk perception, operationalised as concern about zoonotic disease transmission. Practices included self-reported hunting and slaughtering of wildlife.ResultsThe analysis identified three distinct ontology classes: Relational environmentalism (52% of the population), characterised by strong spiritual connections to animals and a tendency to protect wildlife; Dualistic environmentalism (28%), with a weaker spiritual connection to animals but a commitment to wildlife conservation; and Neutral (20%), demonstrating little spiritual connection to animals and a neutral attitude towards wildlife conservation. In the logistic regression analyses, both environmentalism groups exhibited greater knowledge of zoonotic transmission and concern about outbreaks, with members of the Relational class demonstrating higher levels of these attributes. Furthermore, members of the Dualistic environmentalism class were less likely to have close contact with animals.ConclusionsIn Latin America’s biodiversity-rich regions, individuals whose ontology aligns with environmentalism appear to demonstrate a heightened awareness of zoonoses, particularly those who adhere to a Relational environmentalism perspective. Consequently, the integration of IPLC cultural knowledge holds potential to enhance wildlife conservation measures and contribute to the mitigation of disease transmission. Further research is needed to explore causal pathways and the integration of culturally grounded approaches into public health interventions.Keywords: Climate Change, Cross-Sectional Studies, INFECTIOUS DISEASES STRENGTHS AND LIMITATIONS OF THIS STUDY The study used random household sampling across urban, rural and protected areas in four countries, enhancing the diversity of sociocultural contexts represented. The study used random household sampling across urban, rural and protected areas in four countries, enhancing the diversity of sociocultural contexts represented. Ontologies were derived using latent class analysis, allowing data-driven classification rather than predefined categories. Ontologies were derived using latent class analysis, allowing data-driven classification rather than predefined categories. The cross-sectional design limits causal interpretation regarding the directionality of associations. The cross-sectional design limits causal interpretation regarding the directionality of associations. Self-reported practices (eg, hunting and slaughtering) and self-identification of indigeneity may be subject to reporting and social desirability bias. Self-reported practices (eg, hunting and slaughtering) and self-identification of indigeneity may be subject to reporting and social desirability bias. Introduction Latin America is one of the most biodiverse regions globally, with over 40% of the world’s species endemic to the region.1 2 However, biodiversity is under significant threat from human activities such as deforestation, mining, monoculture farming and unregulated wildlife exploitation.3,5 Over the past five decades, Latin America has experienced the highest wildlife population declines worldwide, with a 95% reduction, largely driven by land-use changes that increase the risk of zoonotic spillover.6 7 Indigenous Peoples and Local Communities (IPLCs) play a crucial role in conservation efforts, possessing deep ecological knowledge and fostering spiritual ties to ecosystems. Many IPLC knowledge systems have been described as reflecting relational human–nature ontologies, rooted in animism, that emphasise interconnectedness rather than the nature-culture divide introduced by European colonial perspectives.8,10 Many Indigenous traditions across Latin America, including Andean, Amazonian and Mayan cosmologies, perceive humans, animals and natural entities as inherently alive and sentient.11,14 This perspective informs conservation practices and has influenced legal frameworks such as the ‘rights of nature’ in some Latin American countries.15 Despite their stewardship, IPLCs face increasing challenges from poverty, extractivism, climate change and conservation approaches that often exclude local participation.16 Meanwhile environmental activism centred on animal protection is gaining traction in Latin American countries, particularly among younger, urban populations, though it often emphasises market-driven lifestyle changes (eg, veganism) rather than systemic ecological integration.17 In contrast, many IPLCs depend on wildlife for food and income, leading to human-wildlife interactions and potential conflicts, such as livestock predation and crop destruction.18 These interactions, exacerbated by deforestation and habitat fragmentation, also heighten the risk of zoonotic disease transmission, with approximately 60% of emerging infectious diseases being zoonotic, the majority originating from wildlife.19 Zoonotic transmission is rarely linear; instead, multiple species, including domestic and wild animals, often serve as hosts.20 The emergence of SARS-CoV-2 has underscored the need for a deeper understanding of human-wildlife interactions to mitigate future risks. Ontologies shape human–wildlife interactions and influence zoonotic risk.21 22 However, the role of relational ontologies, such as animism, in contrast to dualistic naturalistic perspectives, remains largely unexplored. It is hypothesised that these ontologies may function as both barriers and enablers for zoonosis prevention.21 22 Studies suggest that IPLC-managed areas experience lower deforestation rates and greater forest regeneration, contributing to biodiversity conservation and reduced zoonotic risk.23 Despite these insights, however, there has been a paucity of research examining whether different ontologies are actually associated with distinct knowledge, attitudes and practices (KAP) towards wildlife in Latin America. This study therefore investigates this relationship, using data from the cross-sectional epidemiologic component of the ‘Knowledge, Attitudes and Practices towards the Risk of Zoonotic Diseases, Wildlife Trade and Wildlife Consumption in Latin America’ study. Materials and methods A detailed description of the overall project has been published previously.24 The cross-sectional design, including study areas, participant selection, survey instruments, variables and data analysis procedures, is summarised below. Study areas and population Four countries’ areas were selected for their diverse climate, biodiversity, wildlife and culture (figure 1). In Brazil, the project was conducted in the Atlantic Forest, a region with 5% of the world’s vertebrate species. It involved three rural communities representing the population of the region: Caiçaras, Quilombolas and Castelhanos-colony. In Chile, the project focused on the O'Higgins region, centred on the Paredones river basin. It encompassed the communities of Navidad (coast), Doñihue (valley) and Machalí (mountain). In Bolivia, the project was conducted in the Department of Chuquisaca, in the Municipality of Presto, within the El Palmar Integrated Management Natural Area. Quechua communities inhabit this region. In Guatemala, the project was implemented in the Ecoregion Lachua, a vast tropical wetland area in 49 indigenous communities of the Maya-Q'eqchi’ people. Study areas (from North to South): (A) Ecoregion Lachua, Guatemala (photo: Carlos Roberto Vásquez-Almazán); (B) Presto, Province of Jaime Zudañez, Bolivia (photo: Carlos Fernando Pinto Navia); (C) Castelhanos colony, Paraná Brazil (photo: Luciana Salini Abrahão Pires); (D) Lo Miranda, O’Higgins Region, Chile (photo: Pilar Macarena Mansilla Vivar). The map was created with openstreetmap.org. The photos were taken by the KAP wildlife study group and reused with permission. KAP, knowledge, attitudes and practices. A two-stage cluster sampling procedure was conducted in each area to include 900 participants, with a random sample of Enumeration Areas achieving 300 in rural, urban and protected areas. We invited all household members aged 10 years or older into the study, with 100 individuals in three age groups (children, adults and elderly) in each area. The sample size considered the balance between feasibility of implementing fieldwork and statistical robustness, so the minimum sample size for developing a multivariable prediction model with binary outcomes was considered, considering at least 100 people per group.25 Written informed consent was obtained from all participants. For participants under 16 years of age, written informed consent was obtained from a parent or legal guardian, and verbal or written assent was obtained from the minor participant in accordance with local ethical requirements. Community members participated in the study through the survey, which was designed to capture their experiences, priorities and practices related to zoonotic risks and wildlife use. Study instruments The standard questionnaire on KAP was adapted from previous studies.26 27 The English version of the questionnaire is provided in the online supplemental file S2 and includes questions on demographics, education, occupation and medical history, and zoonotic disease-related KAP.24 The instrument was translated into local languages and then piloted in each country (≈20) to assess clarity, with feedback used to revise it in the final questionnaire. The interviews, conducted between September 2022 and September 2024 by trained interviewers who were fluent in the local languages, lasted approximately 30 min, and responses were standardised and coded for entry into Epi Info or Open Data Kit. Participant data were pseudonymised by assigning unique identification codes and stored on a secure platform compliant with applicable data protection regulations. Patient and public involvement Community representatives and members of Indigenous and local communities were engaged during the preparatory phase of the study to introduce the research aims and ensure the cultural appropriateness of the survey. The questionnaire was piloted and refined based on feedback regarding clarity, language, contextual relevance and the time required for completion. Translations into local languages were reviewed by bilingual experts to ensure comprehensibility. Prior to fieldwork, the study was presented to local authorities and through local media channels. Community representatives facilitated access to households and supported recruitment by informing residents about the study. Participants were not involved in determining the outcome measures, which were defined by the research team. Dissemination includes sharing results with participating communities through locally appropriate formats, such as community meetings, radio broadcasts and the distribution of culturally appropriate educational materials. Variable definition For the analyses, we defined the variables as follows: Sampling variables: Country: Brazil (reference), Bolivia, Chile and Guatemala. Country: Brazil (reference), Bolivia, Chile and Guatemala. Area: Urban (reference), rural and protected area. Area: Urban (reference), rural and protected area. Age group: Children (10 to 17 years), adults (18 to 60 years, reference) and elderly (61 years and older). Age group: Children (10 to 17 years), adults (18 to 60 years, reference) and elderly (61 years and older). Potential confounders: Gender: Male (reference), female. None of the participants identified as diverse. Gender: Male (reference), female. None of the participants identified as diverse. Level of formal education: None, primary education, secondary education and tertiary education (reference). For children (<18 years), the highest parental level of education was used as a proxy. Level of formal education: None, primary education, secondary education and tertiary education (reference). For children (<18 years), the highest parental level of education was used as a proxy. Self-identification as belonging to an indigenous or ancestry related group: No (reference) and yes. (Residence in a selected study area did not automatically imply classification as IPLC.) Self-identification as belonging to an indigenous or ancestry related group: No (reference) and yes. (Residence in a selected study area did not automatically imply classification as IPLC.) Outcome variables: Knowledge on disease transmission from wildlife to humans: ‘How likely do you think that diseases can be transmitted from wildlife to humans?’ Knowledge on disease transmission from wildlife to humans: ‘How likely do you think that diseases can be transmitted from wildlife to humans?’ Knowledge on zoonosis by wildlife trade: ‘How likely do you think that wildlife trade has an impact on the transmission of diseases?’Both knowledge variables were measured on a 5-point Likert scale and dichotomised to indicate a lack of knowledge (No=‘extremely likely’ or ‘likely’; Yes=any other category). In the sensitivity analysis, respondents who responded ‘likely’ were also designated as lacking knowledge. Knowledge on zoonosis by wildlife trade: ‘How likely do you think that wildlife trade has an impact on the transmission of diseases?’ Both knowledge variables were measured on a 5-point Likert scale and dichotomised to indicate a lack of knowledge (No=‘extremely likely’ or ‘likely’; Yes=any other category). In the sensitivity analysis, respondents who responded ‘likely’ were also designated as lacking knowledge. Training needs: ‘Have you ever had any training related to zoonotic diseases, wildlife trade, or consumption?’ (No/Yes). Training needs: ‘Have you ever had any training related to zoonotic diseases, wildlife trade, or consumption?’ (No/Yes). Risk perception: ‘Are you worried about diseases or disease outbreaks in live animals in your local market?’ (Yes/No). Risk perception: ‘Are you worried about diseases or disease outbreaks in live animals in your local market?’ (Yes/No). Hunting: ‘During the past 12 months, have you hunted an animal?’ (Yes/No). Hunting: ‘During the past 12 months, have you hunted an animal?’ (Yes/No). Slaughtering: ‘During the past 12 months, have you slaughtered an animal?’ (Yes/No). Slaughtering: ‘During the past 12 months, have you slaughtered an animal?’ (Yes/No). Data analysis The analysis was done by IBM SPSS V.29 software. For the latent class analysis (LCA), we used the R package poLCA via the R plug-in. The study population was restricted to those with complete data on all indicator variables used for the LCA. The LCA was then used to classify ontologies based on the patterns of the following six statements regarding the attitude towards wildlife, all scored on a 3-point Likert scale (‘strongly disagree’/’disagree’, ‘neutral’, ‘agree’/’agree’: Humans are part of the animal world. Humans are part of the animal world. (Some) wild animals are connected with my ancestors. (Some) wild animals are connected with my ancestors. Animals see themselves as humans and have their own culture. Animals see themselves as humans and have their own culture. I fear wild animals, we have to protect humans from wild animals. I fear wild animals, we have to protect humans from wild animals. We have to protect wild animals from humans. We have to protect wild animals from humans. Wild animals are beneficial to humans. Wild animals are beneficial to humans. The selection of the model was guided by the principle of parsimony, with simplicity, interpretability, and model fit based on the Bayesian Information Criterion (BIC) being given consideration. Given the robustness of the results, the implementation of multiple-group models (by country) was deemed unnecessary and was thus excluded to enhance simplicity. Additionally, the LCA was repeated, with dichotomised indicator variables (‘strongly disagree’/’disagree’/’neutral’ vs ‘agree’/’strongly agree’). Furthermore, we performed a country-specific LCA to account for heterogeneity between countries. We compared the ontology classes identified via LCA across variables and potential confounders, using absolute and relative frequencies and χ² tests. Subsequently, we examined the associations between class membership and KAP. Frequencies were calculated for each category, followed by crude logistic regression models adjusted only for the sampling variables. Finally, fully adjusted models were developed, incorporating both the sampling variables and potential confounders. As sensitivity analyses, we re-ran the final models using country-specific latent classes, both with and without adjustment for country. To assess potential overadjustment, we repeated the models excluding Indigenous identification from the final models. Finally, we stratified the models for the primary outcomes by country. In addition, we considered the lack of knowledge about transmission from animals to humans for specific zoonotic diseases as outcome (Yellow fever, Chikungunya, Hantavirus, Leptospirosis, Rabies, Brucellosis, Salmonellosis, Bovine Tuberculosis and COVID-19). This study adhered to the Consensus-Based Checklist for Reporting of Survey Studies guideline.28 The completed checklist is provided in the online supplemental file S3. Results A total of 3011 face-to-face interviews were conducted (~85% of the eligible population). Of these, 2970 had complete data on the LCA variables. Most interviews were conducted in Bolivia (N=900), followed by Guatemala (N=841). In Brazil, no approval for participation of children was obtained; in Chile, families were reluctant to let children take part. Therefore, only 622 interviews with adults and elderly were conducted in Brazil and 607 in Chile. The comparison of sociodemographic variables revealed statistically significant differences between countries in all variables (pχ²<0.001). In Guatemala, more males (56%) than females (44%) participated, while the reverse was true in the other countries. The lowest levels of educational attainment were observed in Bolivia and Guatemala, with one in three lacking formal education, compared with around 8% in Chile and 11% in Brazil. Furthermore, in the former countries, a higher percentage of the population with indigenous ancestry was noted (>90%), in comparison to 7% in Chile and 21% in Brazil (online supplemental table S1). The missing data showed no association with other variables, so it is possible to assume it was missing at random (data not shown). Incorporating the indicator variables (illustrated in online supplemental file S1: table S2 according to country) into the LCA, the BIC was lowest for the seven-class solution (online supplemental file S1: table S3 and figures S1-S4). The most substantial decrease was observed between the two-class and three-class solutions, after which BIC gradually declined with diminishing returns until reaching its minimum. Beyond the three-class solution, the additional classes became increasingly challenging to interpret, as they introduced finer but less meaningful subdivisions. Consequently, we selected the three-class solution for subsequent analyses. Latent class 1 (‘neutral’; prevalence 21%) comprised participants who exhibited a neutral response to most items. Latent class 2 (‘dualistic environmentalism’; 29%) was constituted by participants who were inclined to support animal protection without disclosing cultural and spiritual ties to ecosystems. Latent class 3 (‘relational environmentalism’; 52%), included those for whom animals were spiritually significant and morally important, viewed as a source of utility (online supplemental file S1: figure S2). The distribution of the three ontology classes differed according to country, area, age, level of education, and self-identification with indigenous groups (p<0.05; table 1). Table 1Absolute and relative frequencies of general characteristics of the study population by ontology latent classN=2970MissingsLCA class 1: neutral*LCA class 2: dualistic environmentalism*LCA class 3: relational environmentalism*NN%n%n%Pχ†Total60120.383328.0153651.7Country0<0.001 Brazil396.337860.820533.0 Chile7712.713121.639965.7 Bolivia19121.222124.648854.2 Guatemala29435.010312.244452.8Area60.04 Urban21921.126725.755453.3 Rural25721.433928.360350.3 Protected12517.122731.137951.8Age group†6<0.001 Adults22715.844531.176153.1 Children20136.313324.121939.6 Elderly17217.625526.155156.3Gender‡8 Male29220.838427.472551.70.64 Female30719.744728.680751.7Education§20<0.001 None17625.314420.737754.1 Primary25420.642434.455545.0 Secondary11616.419327.339856.3 Tertiary5216.66922.019261.3Indigenous identification¶8 No16613.944537.158749.0<0.001 Yes43324.538822.094353.5This table includes only data from participants with complete responses to the LCA indicator variables.*Indicator variables used in the LCA are listed in table 2.†Age groups were defined as follows: adults (18–59 years), children (10–17 years) and elderly (60 years and older).‡None of the participants reported identifying as ‘diverse’.§Level of education was determined based on responses to the question: ‘What is the highest level of education you have completed?’ Tertiary education included college, university or professional education. For children, the highest parental level of education was used, based on the question: ‘What is the highest level of education that your mother or father completed?’.¶Indigenous identity was based on responses to the question: ‘Do you belong to any nation or peasant native indigenous people?’.LCA, latent class analysis. Absolute and relative frequencies of general characteristics of the study population by ontology latent class This table includes only data from participants with complete responses to the LCA indicator variables. Indicator variables used in the LCA are listed in table 2. Age groups were defined as follows: adults (18–59 years), children (10–17 years) and elderly (60 years and older). None of the participants reported identifying as ‘diverse’. Level of education was determined based on responses to the question: ‘What is the highest level of education you have completed?’ Tertiary education included college, university or professional education. For children, the highest parental level of education was used, based on the question: ‘What is the highest level of education that your mother or father completed?’. Indigenous identity was based on responses to the question: ‘Do you belong to any nation or peasant native indigenous people?’. LCA, latent class analysis. In the logistic regression analyses, participant knowledge regarding zoonotic diseases was generally found to be higher among individuals classified into dualistic or relational environmentalism class in comparison to those in the Neutral class (table 2; online supplemental file S1: tables S4 and S5). Furthermore, an increase in knowledge about zoonotic diseases was observed with higher levels of formal education. Regarding training needs, no association was identified with the ontology latent classes (online supplemental file S1: table S6). A lower level of formal education was associated with a higher OR for lacking training. Table 2Absolute and relative frequencies, adjusted and unadjusted results of the logistic regression models for lack of knowledge on transmission of zoonosis through wildlife trade*N=2920N%OR95% CIOR95% CITotal91831.4Crude†LowerUpperAdjusted‡LowerUpperCountry Brazil18730.111 Chile9515.80.400.320.530.400.300.55 Bolivia36140.72.091.652.672.371.693.34 Guatemala27534.01.531.201.951.661.162.37Area Urban26625.911 Rural46539.42.011.662.441.911.572.32 Protected18726.20.760.610.950.790.621.00Age group§ Adults40428.611 Children17232.00.840.671.070.800.621.02 Elderly34235.21.381.141.661.301.061.60Gender¶ Male44132.011 Female47730.90.970.821.140.940.801.11Education** None28741.51.511.072.141.501.062.12 Primary36429.81.150.841.591.130.821.56 Secondary20229.01.360.981.901.370.981.91 Higher6521.011Indigenous identification†† No32126.911 Yes59734.50.800.591.080.780.571.05Latent class‡‡ Neutral23239.311 Dualistic environmentalism24229.30.760.590.980.770.600.99 Relational environmentalism44429.50.730.590.900.730.590.91*Not answering ‘very likely’ or ‘likely’ to the question ‘How likely do you think that diseases can be transmitted from wildlife to humans?’.†Adjusted for sampling criteria: country, age, area.‡Mutually adjusted for all variables listed in the table.§Age groups were defined as follows: adults (18–59 years), children (10–17 years) and elderly (60 years and older).¶None of the participants reported identifying as ‘diverse’.**Level of education was determined based on responses to the question: ‘What is the highest level of education you have completed?’ Tertiary education included college, university or professional education. For children, the highest parental level of education was used, based on the question: ‘What is the highest level of education that your mother or father completed?’.††Indigenous identity was based on responses to the question: ‘Do you belong to any nation or peasant native indigenous people?’.‡‡Indicator variables used in the latent class analysis are listed in online supplemental file S1: table S1. Absolute and relative frequencies, adjusted and unadjusted results of the logistic regression models for lack of knowledge on transmission of zoonosis through wildlife trade* Not answering ‘very likely’ or ‘likely’ to the question ‘How likely do you think that diseases can be transmitted from wildlife to humans?’. Adjusted for sampling criteria: country, age, area. Mutually adjusted for all variables listed in the table. Age groups were defined as follows: adults (18–59 years), children (10–17 years) and elderly (60 years and older). None of the participants reported identifying as ‘diverse’. Level of education was determined based on responses to the question: ‘What is the highest level of education you have completed?’ Tertiary education included college, university or professional education. For children, the highest parental level of education was used, based on the question: ‘What is the highest level of education that your mother or father completed?’. Indigenous identity was based on responses to the question: ‘Do you belong to any nation or peasant native indigenous people?’. Indicator variables used in the latent class analysis are listed in online supplemental file S1: table S1. In terms of attitudes, participants classified as Dualistic or Relational environmentalists were more likely to express concern about a disease outbreak at a local market than those in the neutral class (OR for dualistic environmentalism 1.4; 95% CI 1.0 to 1.8; OR for relational environmentalism 1.8; 95% CI 1.4 to 2.3; table 3). The same was true for participants with higher levels of formal education. Table 3Absolute and relative frequencies, adjusted and unadjusted results of the logistic regression models for being worried about a disease outbreak in live animals at the local market*N=2911N%OR95% CIOR95% CITotal194866.9Crude†LowerUpperAdjusted‡LowerUpperCountry Brazil14623.511 Chile54390.337.1626.3652.3932.0322.3246.00 Bolivia63672.57.285.619.445.954.228.39 Guatemala62376.79.417.2212.257.645.3111.00Area Urban80178.111 Rural63153.70.340.280.430.360.290.45 Protected51672.71.020.801.310.880.671.15Age group§ Adults89463.511 Children42278.91.190.911.571.521.142.02 Elderly63265.20.720.580.890.850.671.07Gender¶ Male89264.811 Female105668.81.211.001.451.231.021.48Education** None42964.30.360.230.550.360.230.56 Primary76362.50.430.290.650.460.300.69 Secondary48168.80.570.370.870.570.370.87 Higher26586.011Indigenous identification†† No68657.811 Yes126273.21.531.122.091.531.112.11Latent class‡‡ Neutral38666.211 Dualistic environmentalism44353.81.401.051.851.381.041.83 Relational environmentalism111974.41.801.422.291.771.392.25*Answering ‘Yes’ to the question ‘Are you worried about diseases or disease outbreaks in live animals in your local market?’.†Adjusted for sampling criteria: country, age, area.‡Mutually adjusted for all variables listed in the table.§Age groups were defined as follows: adults (18–59 years), children (10–17 years) and elderly (60 years and older).¶None of the participants reported identifying as ‘diverse’.**Level of education was determined based on responses to the question: ‘What is the highest level of education you have completed?’ Tertiary education included college, university or professional education. For children, the highest parental level of education was used, based on the question: ‘What is the highest level of education that your mother or father completed?’.††Indigenous identity was based on responses to the question: ‘Do you belong to any nation or peasant native indigenous people?’.‡‡Indicator variables used in the latent class analysis are listed in online supplemental file S1: table S1. Absolute and relative frequencies, adjusted and unadjusted results of the logistic regression models for being worried about a disease outbreak in live animals at the local market* Answering ‘Yes’ to the question ‘Are you worried about diseases or disease outbreaks in live animals in your local market?’. Adjusted for sampling criteria: country, age, area. Mutually adjusted for all variables listed in the table. Age groups were defined as follows: adults (18–59 years), children (10–17 years) and elderly (60 years and older). None of the participants reported identifying as ‘diverse’. Level of education was determined based on responses to the question: ‘What is the highest level of education you have completed?’ Tertiary education included college, university or professional education. For children, the highest parental level of education was used, based on the question: ‘What is the highest level of education that your mother or father completed?’. Indigenous identity was based on responses to the question: ‘Do you belong to any nation or peasant native indigenous people?’. Indicator variables used in the latent class analysis are listed in online supplemental file S1: table S1. Finally, latent class membership in the dualistic environmentalism was weakly associated with lower odds of engaging in practices such as hunting (OR 0.6; 95% CI 0.4-0.8), compared to the neutral latent class (table 4). Lower levels of education (compared to tertiary education), residing in Bolivia or Guatemala (compared with Brazil), being male (compared with female) and residing outside urban areas (compared to in urban areas) were all associated with higher odds of hunting. These associations were less pronounced for slaughtering (online supplemental file S1: table S7). Sensitivity analyses mainly confirmed our results (data not shown). Table 4Absolute and relative frequencies, adjusted and unadjusted results of the logistic regression models for 12 months prevalence of hunting*N=2872N%OR95% CIOR95% CITotal48616.9Crude†LowerUpperAdjusted‡LowerUpperCountry Brazil386.111 Chile345.61.060.651.711.030.621.71 Bolivia22925.85.143.507.535.143.178.32 Guatemala18524.34.903.337.194.142.506.87Area Urban11411.311 Rural22519.22.401.853.112.251.722.95 Protected14721.41.631.232.151.691.262.26Age group§ Adults19714.111 Children17333.91.791.382.321.731.312.29 Elderly11612.10.730.570.950.640.480.86Gender¶ Male32323.811 Female16310.80.370.300.460.360.290.45Education** * None13820.82.051.263.342.341.433.85 Primary22618.61.941.233.072.091.313.33 Secondary9614.01.951.213.142.181.343.55 Higher268.611Indigenous identification†† No1149.611 Yes37222.10.910.631.300.860.591.25Latent class‡‡ Neutral14525.011 Dualistic environmentalism8210.00.570.410.800.560.400.79 Relational environmentalism25917.60.970.751.250.950.731.23*Answering ‘Yes’ to the question ‘During the past 12 months, have you hunted an animal?’.†Adjusted for sampling criteria: country, age, area.‡Mutually adjusted for all variables listed in the table.§Age groups were defined as follows: adults (18–59 years), children (10–17 years) and elderly (60 years and older).¶None of the participants reported identifying as ‘diverse’.**Level of education was determined based on responses to the question: ‘What is the highest level of education you have completed?’ Tertiary education included college, university, or professional education. For children, the highest parental level of education was used, based on the question: ‘What is the highest level of education that your mother or father completed?’.††Indigenous identity was based on responses to the question: ‘Do you belong to any nation or peasant native indigenous people?’.‡‡Indicator variables used in the latent class analysis are listed in online supplemental table S1. Absolute and relative frequencies, adjusted and unadjusted results of the logistic regression models for 12 months prevalence of hunting* Answering ‘Yes’ to the question ‘During the past 12 months, have you hunted an animal?’. Adjusted for sampling criteria: country, age, area. Mutually adjusted for all variables listed in the table. Age groups were defined as follows: adults (18–59 years), children (10–17 years) and elderly (60 years and older). None of the participants reported identifying as ‘diverse’. Level of education was determined based on responses to the question: ‘What is the highest level of education you have completed?’ Tertiary education included college, university, or professional education. For children, the highest parental level of education was used, based on the question: ‘What is the highest level of education that your mother or father completed?’. Indigenous identity was based on responses to the question: ‘Do you belong to any nation or peasant native indigenous people?’. Indicator variables used in the latent class analysis are listed in online supplemental table S1. Discussion We explored people’s knowledge of wildlife and zoonotic diseases in urban, rural and protected areas in Bolivia, Brazil, Chile and Guatemala. Participants who aligned with environmentalist classes exhibited higher levels of knowledge about zoonotic diseases and greater concern about disease outbreaks in comparison to those with a neutral class. Concurrently, individuals categorised within the relational environmentalism class were also more likely to engage in practices related to animals. The link between relational environmentalism and knowledge or attitudes towards zoonotic disease transmission is consistent with previous research highlighting the importance of indigenous knowledge systems in biodiversity conservation and ecosystem management.19 Indigenous communities often possess an intricate understanding of human-wildlife interactions due to their deep cultural and spiritual ties to nature.29 30 This traditional ecological knowledge (TEK) is often transmitted orally through generations.31 Likewise, the dualistic environmentalism class’s prioritisation of animal protection is congruent with prevailing trends in global environmental activism.17 32 The ontology classes identified were not exclusively associated with indigenous ancestry. This is in alignment with a conceptual shift from the utilisation of the term ‘Indigenous Communities’ to the more inclusive and comprehensive use of IPLCs. Prior research suggests that many rural and local communities retain their TEK due to their dependence on natural resources.33 The One Health approach emphasises the integration of IPLCs’ knowledge in the prevention of zoonotic diseases, acknowledging that their health perspectives are shaped by cultural traditions and lived experiences.34 This study supports the argument that KAP towards zoonotic risk and wildlife coexistence are shaped more by ontologies and ecological interactions than by ethnic or political self-identification alone.35 The finding that formal education plays a role in shaping knowledge about zoonotic diseases aligns with broader public health literature.36 Education has been consistently linked to increased health literacy, empowering individuals to recognise risks and adopt preventive measures. Conversely, the observed gaps in knowledge and training among participants with lower education levels highlight persistent inequities in access to information and resources. In that sense, it is possible that the children have not yet consolidated positions on the subject. Country-specific results revealed significant contextual differences in zoonotic disease awareness. These discrepancies may stem from differences in public health campaigns, levels of wildlife exposure, environmental and cultural factors, or the representation of IPLCs in health education initiatives.37 The intersection of anthropogenic changes and human–wildlife interactions further amplifies zoonotic disease risks, particularly in Brazil, where expanding urbanisation has altered pathogen transmission dynamics.38 Traditional hunting and wildlife use practices play a role in shaping interactions with wildlife. Communities in Guatemala, for example, rely on wildlife for protein and use animals in cultural rituals, practices that foster both knowledge and risks related to zoonotic diseases. In addition, the specific zoonotic diseases considered differ by country in terms of prevalence. This emphasises the need for regionally adapted, culturally sensitive participatory education campaigns that account for local epidemiological risks, cultural perspectives and existing knowledge gaps, balancing tradition with public health priorities.13 16 Limitations of this study include its cross-sectional design, which is useful for identifying associations but precludes causal inference about relationships among ontologies, KAP. For instance, practices such as hunting are shaped by a complex interplay of cultural, economic and structural factors; therefore, attributing them solely to ontological orientations would constitute an oversimplification. Although the study implemented strategies to favour the representativeness of the study population and therefore the generalisation of the results, through random sampling and achieving a high percentage of response (≈ 85%), it is possible that some groups are underrepresented (eg, people with less interest in wildlife protection issues or zoonotic risks); therefore, the association between the Neutral Class in relation to KAP could be underestimated. Furthermore, some regions and populations were underrepresented in this study. For example, no community in a protected area was included in Chile. Similarly, the exclusion of children in Chile and Brazil due to parental refusal might have skewed age-related findings. Furthermore, although the study applied a two-stage cluster sampling design, identifiers for Enumeration Areas and households were not retained in the analytical dataset. Therefore, clustering at these levels could not be accounted for in the models. This may have resulted in an underestimation of SEs. The reliance on self-reported data can introduce bias. Participants may be reluctant to disclose sensitive behaviours, such as illegal hunting, or may exaggerate their knowledge of zoonotic diseases to align with perceived social expectations. In interpreting hunting practices, the legal context should also be considered. For instance, in Brazil, commercial hunting is prohibited under environmental legislation; however, subsistence hunting by traditional communities – such as the Quilombola39 - is permitted. Furthermore, Indigenous identity was assessed through self-identification and may not fully capture the multidimensional and context-dependent nature of Indigeneity in Latin America. Identity is shaped by historical, cultural and political factors; therefore, misclassification or socially influenced responses cannot be excluded. However, self-identification represents the standard and widely accepted approach to assessing Indigenous belonging.40 41 The use of local interviewers may, in some cases, have exacerbated these biases rather than mitigated them. Despite the provision of comprehensive training to interviewers and the extensive piloting of questionnaires, linguistic and cultural variations may still have influenced the interpretation of certain questions. Variations in educational levels and systems between countries may also have influenced participants’ responses. For instance, the lower formal education levels in Bolivia and Guatemala may have contributed to lower reported training on zoonotic diseases, while cultural knowledge passed down through Indigenous traditions may have compensated for this lack of formal education in certain contexts. However, it does not fully explore systemic barriers, such as access to healthcare, education or government-led wildlife management initiatives, which might have shaped the observed disparities between countries and communities. On the other hand, the study did not explore in depth some elements that might better explain the characteristics of the ontology groups, such as media exposure, type of job, which could be explored in future studies. In addition, local events occurring during the study period—such as spectacled bear (Tremarctos ornatus) attacks in Bolivia, avian influenza outbreaks in Chile and the dengue fever epidemic in Brazil—likely influenced participants’ responses. These incidents may have temporarily heightened awareness of specific zoonotic risks or shifted attitudes toward wildlife conservation,42 introducing contextual variability that may limit generalisability. Additionally, potential bias may have been introduced through the way we defined certain variables. To address this, we applied alternative definitions in our sensitivity analyses. Furthermore, given the diversity of cultures, response behaviours and environmental contexts, there is a compelling argument for the use of country-specific LCA or for the stratification of analyses by country. However, the main findings were largely corroborated by our sensitivity analyses. The use of LCA avoided oversimplifying responses, providing nuanced insights into the diversity of perceptions within and across regions. Results were robust to a variety of alternative LCA definitions indicating the validity of the findings. The study may provide valuable insights into the interplay between ontologies and KAP regarding zoonotic diseases in four Latin American regions. We were able to involve remote communities that are difficult to reach from major cities, including protected areas with which local universities have longstanding collaborations. Local involvement was prioritised during data collection. Trained local interviewers fluent in Indigenous languages and dialects conducted surveys, reducing potential biases due to language barriers or cultural misunderstandings. This approach also ensured cultural appropriateness in communication, fostering trust with participants and improving the reliability of the data. We might, however, be criticised for using Indigenous identities and knowledge systems and reducing them to ‘romanticised’ concepts which do not meet the real social and political conditions.43 We tried our best to include IPLCs’ perspectives in our study. The findings presented here are enriched by the qualitative component of the project, which largely supports our results (Data not shown). Collaborative efforts with the communities are currently underway to develop intervention approaches that aim to further explore the interplay between ontologies, behaviours and health outcomes. In conclusion, this study demonstrates that ontologies are associated with variation in zoonotic disease-related KAP across diverse Latin American settings. While ontologies do not determine behaviour, they appear to represent one dimension shaping how individuals perceive and engage with wildlife-related risks. Integrating cultural and relational perspectives into One Health approaches may therefore strengthen zoonotic risk communication and community-based prevention strategies. However, given the cross-sectional design, further research is required to better understand the mechanisms linking ontology and practice. Supplementary material Acknowledgements We thank all participants of the different components of the project. We are grateful for the support of the local authorities and representatives of public and private institutions and organizationsorganisations. We especially thank the entire project team for facilitating interdisciplinary work and opening up to collective learning. The following organizationsorganisations facilitated the study: SERNAP, SEDES-Chuquisaca (Bolivia), Presto Municipality, EducationsEducation Authorities (Bolivia); the City Halls and the Municipal Health Departments of Paranaguá, Adrianópolis and São José dos Pinhais, representatives of the Quilombola communities of the Atlantic Forest Region in Paraná and the Association of Residents of the Communities of Maria Luíza and Colônia Castelhanos, respectively, in Paranaguá and São José dos Pinhais. (Brazil); SEREMI Health (Health regional Department), SEREMI Environment, SAG (Agricultural and Livestock Service), Navidad, Doñihue and Machali municipalities (Chile); Romeo Palacios, Lachua National Park, National Forest Institute -INAB-; Victor Ramos, Vector Unit, Ministry of Public Health and Social Assistance -MSPAS-; Sta Lucia Lachua Research Station, School of Biology, USAC (Guatemala). This publication was funded by the German Academic Exchange Service (DAAD) with funds from the Federal Ministry for Economic Cooperation and Development (BMZ). The funding body had no role in the design of the study; the collection, analysis and interpretation of data; or the writing of the manuscript. Footnotes Funding: The study was supported by the Alliance against Health Risks from Wildlife Trade (grant number 81277747) coordinated by GIZ and funded by the German Federal Ministry for Economic Collaboration and Development (BMZ). Prepublication history and additional supplemental material for this paper are available online. To view these files, please visit the journal online (https://doi.org/10.1136/bmjopen-2025-111451). Provenance and peer review: Not commissioned; externally peer reviewed. Patient consent for publication: Not applicable. Ethics approval: Ethical approval was obtained from the Ethics Committees at the Ludwig-Maximilians-University in Munich, Germany (ID: 22-0254); the Universidad Mayor de Simon in Cochabamba, Bolivia (ID: CEI-FM-UMSS); the Universidad O’Higgins, Rancagua, Chile (ID: 014-2022); the Universidade Federal do Paraná, Curitiba, Brazil (ID: 5.253.420) and the Comité Independiente de Ética K’awil, Guatemala (ID: K’awil - 003-2022). Participants gave informed consent to participate in the study before taking part. Collaborators: KAP Wildlife project team: María Teresa Solís-Soto; Caroline Kuhn; Denise Siqueira de Carvalho; Carlos Roberto Vásquez-Almazán; María Soledad Burrone; Carlos Fernando Pinto Navia; Marcia Adler; Dennis Martin Mendez Heredia; Fabiana Marcela Perez Morales; Luciana Salini Abrahão Pires; Magda Clara Vieira da Costa-Ribeiro; Gustavo Araújo de Almeida; Paulo de Tarso Pires; Angela Maron de Melo; Marilene da Cruz Magalhães Buffon; Tatjana Queiroz Ribeiro de Almeida; Pilar Macarena Mansilla Vivar; Veronica Encina Zamorra; Marie Astrid Garrido; Katja Radon. Map disclaimer: The depiction of boundaries on this map does not imply the expression of any opinion whatsoever on the part of BMJ (or any member of its group) concerning the legal status of any country, territory, jurisdiction or area or of its authorities. This map is provided without any warranty of any kind, either express or implied. Patient and public involvement: Patients and/or the public were involved in the design, or conduct, or reporting, or dissemination plans of this research. Refer to the Methods section for further details. Contributor Information on behalf of KAP Wildlife project team: María Teresa Solís-Soto, Caroline Kuhn, Denise Siqueira de Carvalho, Carlos Roberto Vásquez-Almazán, María Soledad Burrone, Carlos Fernando Pinto Navia, Marcia Adler, Dennis Martin Mendez Heredia, Fabiana Marcela Perez Morales, Luciana Salini Abrahão Pires, Magda Clara Vieira da Costa-Ribeiro, Gustavo Araújo de Almeida, Paulo de Tarso Pires, Angela Maron de Melo, Marilene da Cruz Magalhães Buffon, Tatjana Queiroz Ribeiro de Almeida, Pilar Macarena Mansilla Vivar, Veronica Encina Zamorra, Marie Astrid Garrido, and Katja Radon Data availability statement The data that support the findings of this study are available from the corresponding author, [MTSS], upon reasonable request. References Supplementary Materials Data Availability Statement The data that support the findings of this study are available from the corresponding author, [MTSS], upon reasonable request.