Introduction

Humans have transformed primary ecosystems into highly modified agricultural and urban areas [1]. This encroachment of people in natural habitats, driven by population growth and increasing demand for natural resources [2], has led to a growing number of human-wildlife interactions. Negative interactions between people and wildlife pose a serious threat to nature conservation [3] as the species involved are often persecuted and killed by people. This results in the contraction of their ranges or even the extirpation of their populations [4]. Consequently, negative interactions between people and wildlife emerge as major conservation challenges, which that need to be mitigated to ensure long-term species conservation together with the welfare of human societies that share space with them.

Predation of livestock is a major source of conflict between people and large carnivores [5]. A wide variety of prevention measures, such us electric fences or guarding dogs, exist to deal with carnivore attacks on livestock. However, their effectiveness varies greatly depending on the species and context [6, 7]. Furthermore, conflicts with large carnivores may remain even after implementing mitigation measures aimed at increasing tolerance towards species, such as economic compensation [8] or culling [9]. This persistence of the problem despite the application of preventive and mitigation measures points out the importance of human aspects in fuelling conflict [10].

Retaliatory killing of carnivores is a frequently response to livestock predation. Species killing may be responsible for up to 85% of deaths in some populations of large carnivores [11], potentially leading to extirpation of populations [12], or even the species extinction [13]. Such outcomes will have important implications for ecosystem functioning [14, 15], as well as for the ecosystem services that large carnivores provide to humans [16]. Human-carnivore conflicts are heterogeneous in nature, making it difficult to generalise solutions across study cases [17]. In addition, poorly fitted management strategies for a particular case may ultimately lead to an exacerbation of the problem [18]. A multidisciplinary approach that include both ecological and social perspectives of the problem is therefore needed to obtain a comprehensive understanding of human-carnivore conflicts [19], and to facilitate coexistence among people and large carnivores [20].

We studied the ongoing conflict between extensive cattle ranching and the conservation of jaguars (Panthera onca) and pumas (Puma concolor) in the Brazilian Amazon deforestation frontier. In this biodiversity hotspot [21], livestock predation by jaguars and pumas is frequent [22], and preventive measures such as electric fences, or the inclusion of creole cattle and water buffalos has proved to be effective against livestock predation by carnivores, especially in small ranches [23]. However, habitat degradation caused by cattle ranching expansion [24] together with retaliatory killing still remains important factors of large carnivores decline in the area, especially for the jaguar [25, 26]. Recent studies have shown that the current jaguar range is 14% less than the estimate for 2015 [27]. One of the causes of this rapid decline is the expansion of cattle ranching, which is often associated with deforestation and retaliatory killings, the combined effects of which can lead to jaguar extirpation [28, 29]. Conflict mitigation in this region is therefore essential to promote the long-term conservation of jaguars and pumas. The aim of this study was to obtain a comprehensive understanding to the conflict, by looking at the causes of livestock mortality, the factors influencing livestock predation, and the attitudes of livestock owners. Our hypotheses were that (I) livestock predation is not the main cause of cattle mortality, (II) both landscape attributes and livestock management practices are important in livestock predation by large carnivores, and (III) ranchers’ attitudes towards large carnivores are influenced not only by previous predation incidents, but also by social factors.

Material and methods

Study area

The study was carried out in a partially deforested region of approximately 19,000 km² in the transition zone between the Cerrado and the Amazon biomes (Fig 1). The landscape in the study area was dominated by moist broadleaf forests in the Amazon biome, and tropical and subtropical grasslands in the Cerrado biome. The primary economic activities were cattle ranching and agriculture [30]. Amazon and Cerrado are biomes that harbour high biological diversity [21] and may provide essential ecosystem services, such as disease mitigation and local climate regulation [31]. These regions have been vastly affected by deforestation [32] due to agricultural expansion [33, 34]. Cattle ranching expansion [24] was favoured by the construction of Trans-Amazon highway [35] that facilitated access to intact forests, and by the government policies that fostered land tenure and speculation [36]. This, together with the eradication of livestock diseases [37] and technological developments in beef production [38] positioned Brazil as the world’s largest beef exporter [39]. To guarantee preservation of tropical forests, Brazilian legislation (No. 4,771, of September 15, 1965) made it mandatory to preserve a proportion of woodland, known as legal reserves, on privately owned land. In 2001 (No. 2,166–67, of August 24, 2001), an interim measure was implemented, requiring the preservation of 80% of woodland in the Legal Amazon, and 20% of woodland in the Cerrado biome.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 1. Study area map.

Red circles indicate ranches surveyed in 2008 in the Brazilian states of Pará (PA) and Tocantins (TO). Satellite image of the study area was obtained from Landsat 7, with forest areas shown in dark grey. Country and state boundaries were obtained from Natural Earth Data.

https://doi.org/10.1371/journal.pone.0312077.g001

Data analysis

To assess the reliability of survey data on the causes of cattle mortality, we compared the mortality perceived by the surveyed ranchers with the detailed mortality recorded in the logbooks for two ranches (S4 File). To identify potential problems with biased responses we compared the data collected on the presence of predated cattle on ranches (0/1) and the number of predated heads with the respondent profile (i.e. age, education level, length of time living in the region, and length of time working on the ranch).

To identify which factors influenced livestock predation by jaguar and puma, we fitted two different generalized linear models (GLM): a binomial GLM on the occurrence of cattle predation on ranches to evaluate disparities in ranch vulnerability, and a beta-binomial GLM on the proportion of cattle predated on those ranches that had predations to identify which factors were associated with a higher impact of carnivores on the herd. In the latter model, herd size was included as prior weights. We included as predictors a total of six landscape variables and seven cattle management variables (see Table 1 for variables description). Due to the high number of predictors included (13) and the limited number of ranches visited, it was not advisable fitting a single model with all predictors (landscape + management). Consequently, we fitted on each response variable (occurrence of predations, proportion of cattle predated) two different models: one based on landscape variables, and another based on cattle management variables (i.e. four models in total). The number of ranches used to fit the four models ranged from 80 to 119, depending on the predictors included (see S5 File for details). Prior to analysis, quantitative predictor variables with skewed distributions were log-transformed and all quantitative predictor variables were scaled (mean = 0, sd = 1). We did not find multicollinearity among predictors according to the variance inflation factor (VIF < 3) [46]. No residual spatial autocorrelation was detected in any of the fitted models. For each of the four models fitted, we selected candidate models with ΔAIC_c < 2 (S5 File) and computed the average model. We did not include the variable Breeding (0/1) as predictor when modelling the proportion of cattle predated in attacked ranches due to the majority of them (97%) had included this phase of production on their operations.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Table 1. Description of landscape and cattle management variables used to fit models on cattle predation by jaguar and puma.

https://doi.org/10.1371/journal.pone.0312077.t001

To identify if ranchers’ attitudes were associated to the impact of large carnivores on their ranches, we compared their attitudes towards livestock predation and carnivore conservation with the presence of cattle predated on ranches (0/1) and the proportion of the herd affected in attacked ranches. In addition, we compared the number of predations that respondents were willing to tolerate with the size of their herd, the presence of cattle predated on the ranch (0/1), and the proportion of animals predated in those ranches that experienced cattle predation by carnivores. We did not include in this comparison ranches with less than 50 animals in the herd (n = 6), due to response options ranged from less than 5 individuals to more than 50 individuals. We finally tested whether there was any correlation between the attitudes showed by ranch mangers and their age, education level, length of time living in the region, and length of time working on the ranch. We used Kruskal-Wallis, Mann-Whitney U, Chi-square and Fisher tests for the analyses. We used R software version 4.1.0 [47].

Results

Ranch size ranged from 24 to 103,000 ha (median = 2,807 ha), with forest cover varying from 0 to 96% (median = 41%). Herd size within ranches ranged from 2 to 17,000 heads (median = 1,058) and the median cattle density in pastures was 0.86 heads / ha. Number of workers per ranch ranged from 1 to 75 (median = 3). All of the ranches engaged in extensive grazing, with 92% including a breeding phase, 67% a growing phase, and 53% a fattening phase in their operation (refer to Table 1 for definitions). Additionally, 66% of the ranches had a maternity paddock for calving cows and their calves.

When asked about cattle mortality, 41% of ranch managers identified the ingestion of toxic plants as the primary cause of death on their ranches (Fig 2A). Jaguars were reported to be present on 75% of ranches (46% by direct sightings, and 54% by signs), and cattle predation by jaguar and puma was reported on 71% of the ranches. Ranch managers who had worked on a ranch for longer periods reported cattle predation problems more frequently (χ² test = 7.38, df = 2, P = 0.028). On ranches where cattle were predated by jaguar and puma, 28% of ranch managers perceived predation as the main cause of cattle mortality. The average number of heads predated per year and per ranch varied from 1 to 120, which translates to 0.01% - 12.4% of the herd size being affected (Fig 2B). The perceived values obtained from the ranchers’ memory about cattle mortality were consistent with those values obtained from the detailed data in the logbooks (S4 File) and were not influenced by the respondents’ profile. Most ranchers (99%) reported that jaguars and pumas frequently attacked young cattle, and blamed jaguars instead of pumas for most predations (86 and 14%, respectively).

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 2. Impact of jaguar and puma on cattle ranching.

(A) Sources of cattle mortality perceived by ranch managers as the most significant in their ranches. P: ingestion of toxic plant, S: snake bite, C: predation by jaguar and puma, PS: ingestion of toxic plant and snake bite, CPS: predation by jaguar and puma, ingestion of toxic plant and snake bite, O: other causes. The percentage of respondents appears on the top of each bar. (B) Percentage of cattle stock annually predated by jaguar and puma in ranches with cattle predation problems (n = 91). Values were calculated as the perceived number of animals annually predated divided by the herd size reported by ranch managers. The box plot shows the data distribution with outliers (black dots) and mean (dashed line). Raw data values coloured by herd size was added to the plot.

https://doi.org/10.1371/journal.pone.0312077.g002

The occurrence of cattle predation on ranches was related to cattle management: ranches including a breeding phase, and with higher cattle density, were more likely to report cattle predation by big cats (Table 2). We did not find a relationship between the landscape variables and the presence of cattle predated by jaguar and puma on ranches. The impact of jaguar and puma on the herd was associated with the landscape: ranches with greater proportion of forest cover and close to national parks showed greater proportion of herd predated (Table 2). We did not find a relationship between cattle management variables and the proportion of herd predated.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Table 2. Model results about cattle predation by jaguars and pumas on ranches.

https://doi.org/10.1371/journal.pone.0312077.t002

Asked about the approaches used to reduce predation in the area, ranchers said that changes in cattle management were prevalent (76%), followed by shooting (50%), poisoning (8%), others (8%), and cage traps (1%). Ranchers perceived economic compensation (50%), followed by translocation and culling (35% and 13%, respectively), as the best management options to reduce cattle predation. Ranchers that deal with cattle predated by jaguar and puma on their ranches preferred cattle management (χ² = 9.12, df = 1, P = 0.003) and the payment of damages (χ² = 4.27, df = 1, P = 0.039) as the best responses to cattle predation.

Ranchers showed contradictory attitudes regarding nature conservation and livestock predation (Table 3). For instance, the majority of respondents considered nature and wildlife as a treasure that should be preserved, but 52% of them considered that laws preserving nature were holding back the region’s development. The protection of jaguars was perceived as necessary by 85% of respondents, and yet 42% would be happier if jaguars disappeared from the area. Jaguars and pumas were perceived as a threat to cattle (70%), and to a lesser extent also as a threat to people (37%). Indeed, 12% of respondents reported they were aware of attacks on humans. Most ranchers perceived levels of cattle predation as an inherent, acceptable risk in cattle business (88%), but half of them did not tolerate attacks in their ranches, partly because some ran purebred cattle. In fact, 68% of ranch managers would not accept to lose more than five head of cattle per year. Ranchers with larger herd sizes tolerated a greater number of losses (Kruskal Wallis, H = 9.22, df = 2, P = 0.010). Ranchers considered that their ranches are doing well (84%). However, they would like to receive help to solve cattle predation on their ranches (90%), and considered that cattle predation by jaguar and pumas should be addressed by authorities (83%) rather than individual ranches (24%).

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Table 3. Ranchers’ attitudes towards big cats and conservation.

The number of respondents (n) and the percentage obtained for each response are shown.

https://doi.org/10.1371/journal.pone.0312077.t003

Attitudes varied depending on the level of education, the length of time working on the ranch, the presence of cattle predated in the ranch, and the proportion of the herd affected in attacked ranches (Fig 3). Respondents with higher level of education, particularly those with university studies, showed higher tolerance to the presence of large predators, and perceived less threat to their cattle. Ranchers who had been working on the ranch for over a decade were less willing to receive help for solving cattle predation, and considered that this problem should be address by authorities (Fig 3A). Ranchers who faced cattle predation issues commonly agreed that jaguars posed a threat to their cattle, they expressed concern about the situation on the ranch, and the need for government intervention to address the problem. Among these ranchers, those who experienced a higher proportion of heads predated believed that finding a solution for livestock predation is necessary, and perceived that laws aimed at protecting nature hinder the development of the region (Fig 3B).

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 3. Factors influencing ranchers’ attitudes.

(A) Rancher profile. (B) Predation on the ranch. Each graph shows the attitude statement, the total number of respondents at each level (n), and the p-value for the corresponding statistical test (P). “Do not know” option was removed from the graphs (and the subsequent analysis) for clarity.

https://doi.org/10.1371/journal.pone.0312077.g003

Discussion

Conflicts over wildlife are often very complex and may exhibit different levels of intensity depending on the tolerance towards the species involved. Our results suggest that the conflict between cattle ranching and big cats conservation in the Amazon arc of deforestation could be framed as intermediate, underlying conflict [18], including contradictory and “not in my backyard” attitudes. In other words, ranchers agreed that it was necessary to protect jaguars and pumas, but also showed intolerance to livestock losses in their properties. The conflict in our study area is relevant for the conservation of big cats because they may be killed in response to livestock predation, as occurred in alarming numbers in nearby regions [48]. Indeed, we found that retaliatory killing of jaguars and pumas was the second most frequent response to livestock predation. Killing of carnivores has seriously impacted jaguar populations, translating into one of the primary causes of their decline [25, 26]. Other motivations such as fear, cultural beliefs, or economic interests may significantly contribute to retaliatory killing of carnivores [49, 50].

Ranching practices in our study area did not seem adjusted to the coexistence with large carnivores. The husbandry practices used in the Amazon arc of deforestation, involving large free-ranging herds with calves, were inherently associated with an increased likelihood of predation by jaguars and pumas. These practices allow livestock to enter natural areas, inevitably leading to potential encounters between domestic animals and wildlife, and turning the region into a livestock predation hotspot [22]. In addition, we found that forested areas intermingled in the agricultural matrix were risky regions for livestock, which can be partially explained by the habitat preferences of jaguars and pumas [51, 52]. This was consistent with other studies that have reported proximity to forest as a predictor of livestock predation by jaguars [53]. On the other hand, forested patches are also important for biodiversity conservation [54], thus highlighting a trade-off between conservation and exploitation of natural areas in the region.

Our study reveals the complexities of conservation conflicts, especially when dealing with human attitudes. We found that ranchers did not perceive predation as a significant cause of mortality in their herds. Despite of this, they showed intolerance towards livestock predation, which was occasionally attributed to the breeding of purebred animals. This would ultimately lead ranchers to perceive every loss as significant, even if it is not in numerical terms. In addition, we found that negative attitudes towards big cats were mainly associated with a lower educational background. Our findings are consistent with previous studies elsewhere that identified mortality causes different from predation as the main challenges in extensive cattle ranching [55–57], and highlighted the importance of human factors in shaping attitudes towards carnivores [58–60].

The conflict addressed in this study is predominantly influenced by human factors, rather than solely by livestock predation. Livestock predation would act as a trigger of the conflict, perpetuated by limited knowledge on the importance of large carnivores, the lack of support from the government to mitigate livestock losses, and the perception that conservation laws conditioned the viability of cattle ranching. As a result, focusing solely on reducing attacks will fail to alleviate this conflict, and may even aggravate it [18]. Providing financial compensation could help to alleviate the conflict, especially in such cases where a single loss is highly significant for the ranch. However, compensation schemes need to be applied carefully to avoid an inadequate verification of losses and compensations, which may result in less attention devoted to livestock protection, and even a failure to increase tolerance towards carnivores [8, 61, 62]. Moreover, implementing actions to increase tolerance towards carnivores, such as education on the ecosystem services they provide, could aid in the conservation of big cats in the area. Preserving jaguars and pumas will ensure the maintenance of global biodiversity in tropical forests [63], and together with initiatives to ensure food security and mitigate the adverse effects of agriculture on the environment [64, 65] will encourage the transition towards more sustainable human communities that favour nature conservation.

Our findings include some caveats, related to the nature and age of the dataset. On one hand, data derived from surveys can be noisier and more biased than evidence obtained directly. However, we are confident that our approach provided robust results, due to the high number of ranches included in the survey, together with the ability of ranch managers to identify carnivores’ attacks on livestock through bites or paw prints. Furthermore, data about cattle mortality collected from surveys were compared with data from ranch record books, and statistical tests were applied to identify any bias on the data. On the other hand, several changes have occurred in the study area since the data collection (S6 File). Firstly, some pastures have been replaced by soybean crops, suggesting that cattle ranching may have been reduced, or even ceased, in some properties. Secondly, since 2008, 14% of the remaining forest in the study area has been cleared, mainly for pasture (S6 File), which may have reduced habitat availability for jaguars and pumas, as well as the abundance of wild prey. Consequently, the problem of predation on ranches that currently maintain livestock activity may have been indirectly exacerbated, thus perpetuating the conflict between cattle ranching and carnivore conservation in the study area. The conversion of pastures to croplands in older deforested frontiers [66], such as the study area, is pushing cattle ranching into new intact areas [67], resulting in deforestation there [34, 68] and the consequent appearance of new conflicts. We believe that the drivers of these new conflicts would be very similar to those highlighted in this study, and therefore, our results would allow to develop future mitigation strategies that favour coexistence between cattle ranching and large carnivores along the deforestation frontier.

Supporting information

Acknowledgments

This research was conducted as part of the Vanessa Díaz-Vaquero’s PhD thesis. We would like to express our gratitude to the Spanish Ministry of Science, Innovation and University for their support (FPU18/02598). Additionally, this research was also part of her master thesis for the University Master’s degree in Biodiversity and Conservation Biology (Pablo de Olavide University). We extend our thanks to the academic committee of the Master’s programme for their valuable feedback on an earlier draft. We are also grateful to Naturatins and the Cantão State Park personnel.