Factors Affecting the Adoption of Anti-Predation Measures by Livestock Farmers: The Case of Northern Chile

Abstract

Livestock farming has been a practice of great importance for the evolution of civilization, not only influencing social, economic, and cultural aspects at a global level, but also food, the economy, and sustainability, especially in developing countries, where it generates significant pressure on natural resources and biodiversity. In this context, conflict arises between wildlife, mainly top predators, and livestock farmers. Despite the efforts of different communities to implement measures against predation, the conflict continues to increase. In Latin America, the livestock sector is growing at a much higher rate than in the rest of the world, particularly in Chile, where around a third of agricultural production units use livestock as their main source of livelihood. To understand the factors influencing the behavior of goat farmers when adopting measures, we applied a hurdle model with social, spatial, economic, and productive information to assess the decision to adopt measures and the intensity of the adoption of such practices. To perform this, we used data from a survey, administered in 2014 to 476 farmers located in the three provinces of the Coquimbo Region. Our dependent variable was defined by six measures: a protection dog, night confinement of the herd, supervised grazing, anti-carnivore corral, the death or capture of the predator, and repelling the predator. The adoption decision, as well as the intensity of adoption, were influenced by the location, household size, the type of livestock, the income generated by the livestock, health management, and access to technical advice. The decision to adopt measures was influenced by the production system and whether it was self-sustaining, while the intensity of adoption was influenced by herd size and the number of losses due to predation. The results showed the importance of developing and adjusting livestock support initiatives in the study area, including those that could be created, based on differentiated measures according to the profiles of farmers in the territory.

1. Introduction

Livestock farming is one of the oldest and most fundamental practices of humanity, serving as a cornerstone in the evolution of civilization. It has influenced social, economic and cultural aspects throughout the world [1,2], mainly due to the contribution of four key factors—milk and meat production, fertilizers, the labor force, and transportation [3,4]—which facilitated demographic expansion [5]. It continues to be of great importance today due to its contribution to food, economics, and sustainability, especially for developing countries, where food insecurity is a permanent concern [6].

The livestock sector uses around 3.9 billion hectares worldwide, of which approximately 52% correspond to extensive low-yield pastures [7]. In developing countries, livestock farming takes place on the least productive lands, not suitable for cultivation, with low levels of productivity and profitability [8], thus playing an essential role in the agricultural economy. This area has focused on increasing production to feed 9.8 billion people, according to projections for 2050 [9,10]. It has been estimated that 90% of the world’s milk and 70% of ruminant meat, as well as a third of pork, poultry and eggs, are produced in small-scale farming systems, thus contributing notably to the food and nutritional security of the poor rural population [11,12]. For small farmers, livestock is the main source of income, supports agricultural diversification and on-farm investment [13], provides employment to the producer and family members [14], acts as a store of wealth [15], and serves as a type of insurance [16], in addition to having cultural implications [17].

In Latin America and the Caribbean, the growth rate of the livestock sector is twice the global average, putting high pressure on natural resources and biodiversity [18,19]. Therefore, the sector needs to optimize the efficiency, utilization, and sustainability of resources. Along with great challenges, the future offers immense opportunities for this field, which can play a key role in improving the lives of millions of people, as it can contribute to the efficient use of natural resources, increasing access to clean and renewable energy, generating large incomes, creating opportunities for value addition and industrialization, stimulating the entrepreneurial spirit of small farmers, reducing inequality gaps, promoting sustainable consumption and production patterns, and increasing the resilience of households to face climate crises [6]. In Chile, commercial cattle farming, both dairy and meat, is primarily located in the southern part of the country, in the regions of La Araucanía, Los Ríos, and Los Lagos. Pork and poultry are produced in the center of the country, mainly in the O’Higgins and Metropolitan regions. Sheep farming takes place almost completely in the Magallanes region, which contains around 75% of the sheep in the country. Meanwhile, the Coquimbo Region holds around 60% of the goat stocks, being recognized for this type of livestock. Finally, the population of domestic South American camelids is concentrated in the three regions in the extreme north of the country, according to the Chilean Agricultural Census [20,21]. At the national level, raising livestock is one of the most traditional activities of peasant family farming, with 28% of agricultural production units having livestock as their main source of livelihood [22].

Human-wildlife conflict is a challenging issue [23,24] and is defined as the process that arises when the behavior of wild animals directly and recurrently threatens the way of life of a person or community which, in response, persecutes, or even eliminates, the species [25]. There has been conflict between wildlife and anthropogenic activities, such as agriculture and livestock, since the beginning of humanity and this continues to this day, being a significant problem in many parts of the world [26,27]. This conflict has increased in recent decades, mainly due to the increase in the human population, and its most direct consequences have been the loss and alteration of habitats [28,29], causing a reduction in wildlife, especially in the populations of top predators, which are in overwhelming decline. Persecution by humans is one of the main causes of this [30,31]. Responses to events vary significantly, and several factors need to be considered, including how accurately the level of damage is assessed, and how severe such damage is, both of which affect the perceived level of conflict. We should also consider the intensity of an individual or group’s response to that conflict and whether the management of that species is directly linked to the reported conflict [32].

Reports suggest that small livestock farmers are the actors most affected by predatory attacks, which are often associated with poor management practices. Livestock losses on small farms have a significant impact on rural livelihoods, as the predation of a single animal can cause a significant loss of income for a family [33]. The wild carnivores commonly implicated in livestock losses in Chile are the puma (Puma concolor), foxes (Lycalopex griseus, L. culpaeus and L. fulvipes), wildcats (Leopardus spp.) and, to a lesser extent, the quique (Galictis cuja) [34,35,36]. Birds of prey, especially diurnal ones, can also feed on poultry, such as chickens, turkeys, and ducks [37,38]. It is also important to mention the case of domestic dogs that cause significant attacks on flocks, a generally undervalued threat [28,39,40,41,42].

Anti-predator measures can be categorized as either lethal or non-lethal. Lethal measures involve the killing of the predator, while non-lethal measures include preventive strategies such as the use of deterrent elements (visual, sonic, or chemical), the improvement of enclosures (anti-carnivore pens, meshes for raptors, electric fences), and the use of livestock practices. These practices may include night confinement, supervised grazing, the use of protective dogs, the use of larger animals (llamas, donkeys), the maintenance of the health of the herd (healthy animals are more difficult to prey on), and mitigation measures such as incentives, compensation, or livestock insurance [30,43]. It has been shown that the management practices of local producers significantly influence the incidence of livestock predation [44,45].

Although there are various studies related to ecological aspects of carnivore populations and their behavior [30,31,36], as well as measures against predators that factor in costs, effectiveness, and predation risk models [38,46], there continue to be countless cases showing low or no success in communities that adopt measures, even if these are subsidized by the government. On the other hand, a few studies focus on qualitative aspects of the conflict, focusing on understanding the perceptions and cultural significance of ranchers [44,47,48,49,50].

From a simplistic point of view, farmers generally do not adopt new practices for two reasons. They are either unwilling, or they lack the capacity to do so [51]. However, agricultural producers are continually engaged in processes of technological change within their productive units. This allows them to reformulate their strategies to adapt them in a more adjusted way to the permanent environmental, economic, social, and political changes that occur, both within their systems and the environment in which they carry out their productive activity [52]. The adoption of technologies or measures by low-income farmers must meet a variety of their needs and be acceptable from a sociocultural perspective, in addition to addressing technical and economic considerations [53]. Evaluating the factors that influence the adoption of potential technologies can play a decisive role in developing feasible and sustainable programs according to Nederlof and Dangbégnon [54]. Previous studies have shown that some socio-productive factors influence the adoption or innovation of certain practices. Variables such as the age of the farmers, gender, the size of the farm, and the number of heads of cattle [53,55,56], as well as the location or geography of the environment, have been documented [57].

To the best of our knowledge, there is no research focused specifically on the reasons that lead farmers to implement measures against attacks. Despite the existence of preventive measures and numerous approaches, the conflict between humans and carnivores continues to increase in various parts of the world [58]. This may be related to the fact that the complex social factors that cause this problem are not being targeted [59], which is of key importance in addressing and solving this problem. Along these lines, Currey et al. [60] highlight the importance of considering the human dimensions of human–wildlife conflict for mitigation strategies. This study aims to contribute to the understanding of the conflict by establishing factors that positively or negatively influence the implementation of measures. It seeks to support future research and inform the approaches governmental and non-governmental institutions use to address the livestock and predators’ conflicts. Since the budgets for these issues are usually scarce in developing countries like ours, it is it even more necessary to invest them in the best possible way.

2. Background on Livestock and Conflict in the Coquimbo Region

According to the National Agricultural Censuses [20,21], around 60% of the country’s goat livestock is in the Coquimbo Region, as well as around 50% of mules and donkeys [21,22]. Goat production in Chile has been traditionally kept in the hands of small- and medium-sized farmers. This is an activity carried out through extensive or semi-intensive systems, mainly because they are low-cost production alternatives. For this reason, the areas intended for goat farming correspond to marginal lands, which, added to the above, minimizes the development potential [61]. In this sector, there is evidence of stagnation regarding income level, promoting lifestyles that lead to poverty and an increase in the gap with respect to other productive sectors [62].

Goat farming in this territory started with the Spanish invasion and the introduction of animals, brought from dry regions such as Andalucía and Castilla, which proliferated to the detriment of llamas due to the generation of more immediate profits and the lower level of care necessary for their maintenance [63]. Along with the introduction of goats, the Spaniards applied a land tenure system, marking the beginning of agricultural communities (an agricultural community consists of a land tenure system that combines community rights over large undivided properties with private rights within them—178 agricultural communities are in this territory [64,65]; various studies allow us to affirm that, apparently, more than half of the region’s rural poverty is concentrated in agricultural communities) and the practice of transhumance in dryland sectors, in a way that had been known in Castilla since in the 16th century [62]. Goats and agricultural communities are two elements that can be used to identify the rural sector of the region. Goat breeding benefits from the extraordinary adaptability of these animals to difficult habitats. Despite the clearly deficient production and management systems in the area, they can still provide a wide range of products for sale or self-consumption, such as meat, milk, hides, guano, and, more importantly, cheese.

Most farmers suffer attacks by wild carnivores on their herds; however, very few of them report these attacks to the authorities. Goats suffer the greatest number of attacks, followed by poultry [66]. The increase in these attacks produces a decrease in the livestock mass and, therefore, in the economic income of livestock farmers, which generates conflict between livestock farming and the conservation of carnivore species. According to Sepúlveda et al. [46], among the carnivore species that cause the most deaths to livestock are the culpeo fox, the puma, and the dog. The puma causes on average the most deaths in goats, followed by sheep, horses, and cattle, but culpeo foxes and dogs are the ones affecting the greatest number of farmers. This conflict has intensified in part due to climate change, since Chile is a country with great risk in this regard [67]. In this context, the Coquimbo Region has experienced the greatest drought on record [68], which has produced a decline in natural prey, leading predators to interact more often with humans. In addition, it has contributed to A decrease in livestock, which is why the damage received has a greater impact on livestock farmers.

3. Materials and Methods

3.1. The Study Area

The Coquimbo Region is in the northern area of Chile, and it has an area of 40,579.90 square kilometers, equivalent to 5.37% of the national territory. According to Chilean Population Census of 2017, it has a population of 757,586 inhabitants. Its regional capital is the city of La Serena. It has three provinces, Elqui, Limarí, and Choapa, which are distributed transversally from north to south, covering all mountain range, valley, and coastal sectors (Figure 1).

The climate is in a transition between Mediterranean desert and semi-desert climates, each displaying different nuances: it humid and cloudy on the coast, and there is warm steppe terrain in the interior. The coastal area is characterized by the presence of humidity and cloudiness, with moderate temperatures. The inland area is characterized by the absence of cloudiness, with high temperatures and thermal oscillations when compared to the coast. Precipitation tends to decrease, except in the Andes Mountains where it increases again [69].

Orographically, the region is fundamentally characterized by the presence of four large groups of morphological units: the Andean Mountain chain, with considerable heights that exceed 6000 m above sea level, which joins the Coastal Mountain Range through transverse ranges; and the coastal plain and the intermediate depression, with little development in the region and longitudinal corridors in a north–south direction. The Andean–Coastal Mountain system reaches heights between 600 and 1000 m above sea level in a transverse arrangement of E-W, generated by the erosion of streams and rivers. In this way, a complex internal belt of mountainous relief, which is about 50 km in average width and difficult to access, has been formed topographically [70].

In relation to its biodiversity, it is one of the most interesting regions in the country [62], since it is part of Mediterranean Chile, considered one of the 25 “World Biodiversity Hotspots for Conservation Priority” [71]. In this area, many endemic flora and fauna species are recorded due to the high climatic and topographic heterogeneity, recurrent climatic changes in the past, and the nature of Chile as a biogeographic island [72]. It brings together various biogeographic elements, which have experienced significant evolutionary radiations, as well as relict communities typical of forests from the south of the country. On the other hand, there are various species with northern and southern limits of distribution. These characteristics indicate a greater susceptibility to habitat reductions [69]. At the national level, the Coquimbo Region is in second place among the regions in terms of the lowest percentage of protected area under protection categories, reaching only 0.37% [73].

3.2. Survey and Sampling

The data used in this study include social, economic, and productive information, as well as livestock losses due to predatory attacks and an assessment of wildlife (the predators included puma, culpeo fox, chilla fox, domestic dog, and birds of prey) obtained from a survey carried out during 2014 among 503 livestock producers in the Coquimbo Region. These results were part of the study “Diagnosis of the population status of the puma (Puma concolor) and evaluation of the interaction with livestock farming in the Coquimbo Region”, carried out by an environmental consulting firm, whose technical counterpart corresponded to the Agricultural and Livestock Service (SAG) and was financially supported by the Regional Government of Coquimbo [46]. The relevance of surveys is related to the amount of information captured and their level of coverage, which is not common in this type of study.

Sampling was carried out randomly on lists of livestock producers, provided by the municipalities of each province and the INDAP (INDAP: National Institute for Agricultural Development (Instituto de Desarrollo Agropecuario) of the Chilean Ministry of Agriculture). The questionnaires included socioeconomic information, information about the farm and from the productive systems, and information about losses, adopted measures, and perceptions of the conflict. Information related to perception contained values between 1 and 4, where 1 indicates that there is no damage; 2 indicates light damage; 3 indicates important damage; and 4 indicates very important damage to each predator.

The initial database of 503 surveys, distributed in the three provinces of the region, was reduced to 476 observations (

Table 1. Surveys applied in the three provinces of Coquimbo Region.

Province	Observations
	n	%
Choapa	165	34.7
Limarí	176	37.0
Elqui	135	28.3
Full sample	476	100.0

), mainly due to the incomplete information obtained for certain variables relevant to this study. These correspond to 4.9% of the number of total UPAs (UPA: agricultural productive unit), according to the 2007 agricultural census, and 8.8% of goat farmers in the region, according to 2018 figures from the Regional Government of Coquimbo (estimated at 5391 according to data from the Regional Government of Coquimbo (2018) from the following presentation: “Relevance of goat activity and farmers in the Limarí Province”).

3.3. The Model

To analyze the factors that influence the adoption of measures against livestock predation, a two-hurdle model proposed by Cragg [74] was used. This model assumes that the decision to adopt can precede a decision on the intensity of use, and that both decisions can be explained by different factors or by the same factors, but with different effects. This model was used in a similar study by Jara Rojas et al. [57] on the adoption of agroforestry practices in silvopastoral systems in Colombia, and in the work of Roco et al. [75], evaluating the adoption of practices against climate change.

The model can be expressed as follows:

$$\begin{array}{l}d*={z_i}^{\prime }\alpha +{\epsilon }_i\\ d_i=1 \mathrm{if} d_i*>0, \mathrm{and} 0 \mathrm{if} d_i*\le 0\end{array}\}\mathrm{First}\mathrm{hurdle}\left(\mathrm{adoption}\mathrm{decision}\right)$$

(1)

$$\begin{array}{l}y*={x_i}^{\prime }\beta +{\mu }_i\\ y*=y_i* \mathrm{if} y_i*>0, \mathrm{and} d_i*>0\\ y_i=0 \mathrm{otherwise}\end{array}\}\mathrm{Second}\mathrm{hurdle}\left(\mathrm{intensity}\mathrm{of}\mathrm{adoption}\right)$$

(2)

where d_i* is a latent variable that describes the producer’s decision to adopt measures and d_i is the adoption decision observed, taking the value of 1 if the farmer adopts at least one practice and 0 if they do not take any measure. y* is the latent variable that describes the intensity of adoption and y_i is the observed response of adoption intensity, measuring the number of anti-predation practices carried out. It is important to mention that our dependent variable is equal to the sum of adopted practices for each farmer. z, y, and x are vectors of variables explaining the adoption and intensity. α and β are vectors of parameters. εi and μi are error terms.

To determine whether decisions concerning adoption and adoption intensity are made jointly or separately, a likelihood ratio (LR) test was performed. The test compares the maximum likelihood values of two models: a count data regression model (Poisson regression) and a zero truncated regression. The test allows us to evaluate whether the models are significantly different from each other and can be defined as follows:

λ = −2 (L_P − L_L − L_ZT)

(3)

where L_P − L_L − L_ZT are the log-likelihood values for the Poisson regression model, Logit regression model, and truncated regression at zero, respectively. LR has a chi-square distribution and degrees of freedom equal to the number of explanatory variables included in the model.

3.4. Measures Adopted and Variables

The response variable represents whether the farmers adopted any measure against predation and how many of these they adopted. The measures considered in our analysis are described in

Table 2. Possible anti-predation measures and its adoption by livestock farmers.

Measure	Description	n (%)
Herd protection dog	A dog that may or may not be of a breed specific to this work, but that generally has a large size and is trained with the livestock to protect.	41 (8.6)
Night confinement	This is a traditional practice in the area, in which the herd is enclosed in a pen with restricted space to spend the night. This may or may not be covered, and is generally located close to the home or a rustic building named “ruco”.	336 (70.5)
Grazing	This is the action of accompanying, guiding, and supervising livestock, establishing sectors where they graze or browse.	87 (18.2)
Pen against depredators	This is a corral that usually has higher fences than is common in the area, with a roof and sometimes even a mesh between the roof and the fence.	33 (6.9)
Capture or death of the predator	This is action by farmers to kill, by different means (shots, poison, hanging, dogs trained in hunting), the animal accused of the attacks, or to capture it through cages or snares (huaches).	18 (3.7)
Drive away the predator	This involves scaring away the animal causing predation through different techniques, such as lights, loud sounds, or chasing.	16 (3.3)
Non-adoption of measures		85 (17.8)

.

The independent variables considered in this study are detailed in

Table 3. Description and statistical summary of the variables used in the study.

Variable	Description	Mean	Min–Max	Std Dev
Dependent variables
Measure adoption	Dichotomous variable takes value 1 if the farmer takes at least one measure, or 0 otherwise	0.821
Intensity	Number of measures adopted ranging from 1 to 4	1.115		0.80
Independent variables
Sociodemographic
Age	Age of the farmer in years	54.9	18–94	13.8
Gender	Male	0.586
	Female	0.413
Family size	Number of family members	3.16	1–12	1.58
Location
Elqui	Dichotomous variable takes value 1 if the farm is located in Elqui Province (north), or 0 otherwise (omitted)	0.28
Limarí	Dichotomous variable takes value 1 if the farm is located in Limarí Province (center), or 0 otherwise	0.37
Choapa	Dichotomous variable takes value 1 if the farm is located in Choapa Province (south), or 0 otherwise	0.34
Productive
Sanitary management	Dichotomous variable takes value 1 if the farmer implemented sanitary management of herd regularly (deworming and vaccination), or 0 otherwise	0.56
Intensive system	Dichotomous variable takes value 1 if the production system is intensive (confined animals), or 0 otherwise	0.082
Semi-intensive system	Dichotomous variable takes value 1 if the production system is semi-intensive (animals are released during a period of the day or by season), or 0 otherwise	0.563
Extensive system	Dichotomous variable takes value 1 if the production system is extensive (animals graze freely on extensive lands), or 0 otherwise (omitted)	0.355
Herd size	Number of standardized animal units (AUE) in the farm	20.1	0.25–336.3	22.6
Losses	Number of standardized animal units (AUE) loosed due to predators	2.218	0–32.0	2.98
Meat	Dichotomous variable takes value 1 if the production system is oriented to meat production, or 0 otherwise (omitted)	0.096
Milk	Dichotomous variable takes value 1 if the production system is oriented to milk production, or 0 otherwise	0.252
Double purpose	Dichotomous variable takes value 1 if the production system is oriented to double purpose production (meat and milk), or 0 otherwise	0.546
Self-subsistence	Dichotomous variable takes value 1 if the production system is oriented to self-subsistence, or 0 otherwise	0.098
Agricultural income	Agricultural income in dollars according to year of survey completion	1403	0–17,532.1	1701
Associativity	Dichotomous variable takes value 1 if the farmer declares to belong to some organization, or 0 otherwise	0.64
Technical assistance	Dichotomous variable takes value 1 if the farmer declares to receive technical assistance, or 0 otherwise	0.37

. We considered sociodemographic variables (age, gender, size of the family group, whether one belongs to an indigenous community); location, classified into provinces (Elqui, Limarí, Choapa); productive data (performance of sanitary management, type and orientation of productive system, total number of animals, total losses due to predation); membership in an organization; and technical assistance through advice.

To carry out this study, both livestock ownership and animal losses were adjusted in animal equivalence units (AUE) according to standardized tables (we used tables from the Ministry of Agriculture, Fisheries and Food of Spain and from the United States Department of Agriculture USDA, Natural Resources Conservation Service (NCS)). For example, one mature goat is equivalent to 0.15 AUE, one sheep is equal to 0.20 AUE, an adult cow is equivalent to 1 AUE, and a mature bull is equal to 1.35 AUE.

4. Results and Discussion

The age of the producers surveyed varied between 18 and 96 years, with an average of 54.9 years; the family size fluctuated between 1 and 12 members, with an average of 3.16 members. Only 9 people declared themselves to be indigenous people. In terms of gender, the survey included a higher percentage of male producers, accounting for 59%, compared to 41% for women (Table 3).

Regarding production, as shown in Table 3, 56% of respondents declared themselves to be implementing some form of sanitary management, while 15% utilized it only in summer. The predominant type of production management was semi-intensive (56.3%), followed by extensive (35.5%) and intensive methods (8.2%), respectively. Furthermore, the primary productive orientation was dual-purpose (54.6%), followed by dairy specialization and, finally, meat production. The income, considered in dollars, ranged between USD 0 and 17,532, with an average annual income of USD 1701. Of those surveyed, 64% were affiliated with an association such as a breeding committee or agricultural community, while 37% received productive advice, either directly from INDAP or through one of its specific programs, like PRODESAL or PADIS (PRODESAL is the local development program of INDAP that aims to build technical and productive capacity among low-income, subsistence, and family farmers and their families, with the goal of increasing their share of revenues along the value chain [76]; PADIS is the program for the development of small peasant producers of the rainfed area and its objective is to support rural families who have conditions of drought, reinforce the origin of property (communities), and to strengthen their forestry and agricultural activities [77]).

The livestock ownership of the respondents is presented in

Table 4. Total livestock and losses in terms of animal species in the sample.

Species	Total Livestock	Total Losses
Sheep	10,498	1058
Goats	40,851	6645
Equines	1705	130
Cattle	1300	96
Poultry	8880	1839
Others (pigs; rabbits)	422	0

. As determined by the data from the last two agricultural censuses in the region, it presented a high livestock mass of goats, followed by sheep, poultry (chickens, ducks, turkeys, and geese), and then, respectively, horses, cattle, and other animals, mainly pigs and rabbits. The greatest, relatively and proportionally, numbers of losses due to predation were related to goats, followed by poultry, sheep, and to a lesser extent horses and cattle.

As mentioned previously, data collection in the survey also considered the farmers’ perception of the damage caused by each predator. The results of this perception data are represented in Figure 2. The wild predators perceived as most harmful by respondents, as shown in Figure 2 and

Table 5. Frequency distribution of damage perception by livestock farmers according to predator species (n = 476).

Specie	Damage Perception
	Non-Damage (0)	Scarce Damage (1)	Medium Damage (2)	Important Damage (3)	High Damage (4)
Cougar	1	300	43	54	78
Fox (chilla)	2	210	159	280	25
Fox (culpeo)	1	200	90	115	70
Wildcat	5	460	10	1	0
Quique	4	466	6	0	0
Prey bird	11	383	67	11	4

, were culpeo foxes, ranking first, followed by cougars and chillas. Birds of prey ranked third, followed by wildcats and quiques. This may be attributed to the high incidence of fox attacks on young livestock and chickens [35,78,79], which is directly correlated with the species recorded under tenure. The perception of damage is consistent with findings from other studies on recorded attacks [12,46]. However, the place occupied by birds of prey is striking, although it could be explained by the presence of certain livestock types, (primarily goats), and birds.

Table 2 and Table 3 show that 82.1% of farmers incorporated at least one measure against predation. The Province of Choapa presented the highest adoption rate (84.3%), while the Province of Limarí had the highest intensity, that is, the adoption of two (20.5%) and three measures (7.4%), as shown in Figure 3.

The most commonly adopted practice was night confinement, probably because this measure also allows for better management of the herd, since the farmers can count animals, check animal health, and perform other important tasks more efficiently, such as feed supplementation and morning milking, among others. In addition, it implies a degree of protection against predators that attack during the night, dusk or dawn, such as pumas and foxes. A small group of farmers follow the traditional practice of grazing, which does not require investment but requires a high amount of human capital since it requires personnel and time to be dedicated to it. To a lesser extent, there is the use of pens, an efficient measure but one that requires a significant monetary investment for construction. Finally, there is the capture or death of the predator and efforts to ultimately scare it away, which are not carried out often since they require effort and coordination with other members of the community. In addition, community members are aware of the legal repercussions that these practices, punishable by law (Chilean law number 19,473 includes sanctions ranging from fines to minor imprisonment for hunting protected species), may have.

In relation to the model fitted, the value of the LR test yields λ= 205.7, far exceeding the critical table value of 32 at 1% significance, confirming the hypothesis that decisions on the adoption and intensity of use of anti-predation measures are taken separately. Consequently, the hurdle model is the most appropriate for analyzing how the independent variables influence both decisions [80]. Another consideration for model selection is the use of pseudo R² to check the goodness of fit and the percentage that explains or predicts each result, where a double hurdle presents the higher value (see

Table 6. Results of the Poisson model and the two-stage model.

Variable	Poisson		Two-Stage Model
			Logit		Zero Truncated
	Coeff (Rob. Std. Err.)	dy/dx (Std. Err.)	Coeff (Rob. Std. Err.)	dy/dx (Std. Err.)	Coeff (Rob. Std. Err.)	dy/dx (Std. Err.)
Age	0.000	0.000	−0.14	−0.001	0.009	0.004
	(0.002)	(0.002)	(0.011)	(0.001)	(0.006)	(0.003)
Gender	−0.011	−0.012	0.115	0.012	−0.148	−0.078
	(0.064)	(0.069)	(0.305)	(0.033)	(0.176)	(0.091)
Family size	0.039 *	0.042	0.082	0.009	0.097 **	0.051
	(0.020)	(0.022)	(0.090)	(0.010)	(0.046)	(0.026)
Limarí	0.125	0.137	−0.350	−0.040	0.668 **	0.395
	(0.084)	(0.095)	(0.336)	(0.041)	(0.250)	(0.157)
Choapa	0.162 **	0.179	0.571 *	0.060	0.251	0.138
	(0.077)	(0.088)	(0.346)	(0.034)	(0.235)	(0.136)
Sanitary management	−0.181 **	−0.197	−0.645 *	−0.071	−0.301 *	−0.162
	(0.063)	(0.070)	(0.288)	(0.030)	(0.171)	(0.094)
Intensive system	0.386 **	0.485	2.186 **	0.140	−0.032	−0.016
	(0.100)	(0.141)	(0.573)	(0.022)	(0.294)	(0.152)
Semi-intensive system	0.303 **	0.322	1.916 **	0.244	−0.141	−0.076
	(0.084)	(0.085)	(0.304)	(0.039)	(0.198)	(0.110)
Herd size (AUE)	0.000	0.000	0.001	0.000	0.002 *	0.001
	(0.000)	(0.001)	(0.007)	(0.000)	(0.001)	(0.000)
Losses (AUE)	0.015	0.017	−0.011	−0.001	0.035 **	0.019
	(0.013)	(0.014)	(0.048)	(0.005)	(0.018)	(0.010)
Milk	0.007	0.008	−0.055	−0.006	0.002	0.001
	(0.114)	(0.124)	(0.561)	(0.064)	(0.280)	(0.148)
Double purpouse	−0.194 *	−0.211	−0.412	−0.046	−0.459 *	−0.250
	(0.099)	(0.109)	(0.539)	(0.059)	(0.266)	(0.158)
Self-subsistence	−0.255 *	−0.249	−1.157 *	−0.179	−0.167	−0.082
	(0.151)	(0.132)	(0.582)	(0.114)	(0.408)	(0.188)
Agricultural income	0.000 **	0.000	0.000	0.000	0.000 **	0.000
	(0.000)	(0.000)	(0.000)	(0.000)	(0.000)	(0.000)
Associativity	0.073	0.077	0.547	0.065	−0.032	−0.017
	(0.089)	(0.094)	(0.404)	(0.051)	(0.218)	(0.116)
Technical assistance	−0.236 **	−0.247	−0.802 **	−0.098	−0.396 **	−0.199
	(0.0869)	(0.088)	(0.382)	(0.050)	(0.258)	(0.134)
Log pseudolikelihood	−568.9		−183.6		−284.2
Pseudo R2	0.032		0.181		0.075
N	476		476		391

Notes: * p < 0.09; ** p < 0.05. Rob std err are robust standard errors. std err are standard errors. dy/dx are marginal effects. Estimations were performed in STATA 18.

). Choapa exhibited a greater inclination to adopt measures against predation, whereas Limarí displayed a higher intensity of such measures. This could be attributed to the fact that both provinces, as per census data, boast a higher number of livestock farms and a larger headcount of cattle compared to Elqui. However, they differ in terms of livestock types. The southern province of the region harbors approximately twice as many cattle and horses as Limarí, which in turn possesses twice as many goats as Choapa [21]. It is expectable that, given their higher value, the loss of a cow or a horse presents a significant negative impact on the household economy. This factor may contribute to the higher adoption rates seen among ranchers in Choapa. Conversely, the prevalence of goats causes a greater risk of predation, potentially influencing the adoption intensity in Limarí Province. Additionally, the challenges associated with implementing certain measures may be more pronounced in larger livestock operations, which are more prevalent in Choapa Province.

In contrast to findings in other studies, the age variable did not exhibit a significant relationship with the adoption of measures or their intensity. Previous research has documented a negative correlation between age and adoption [55,81,82], suggesting that older farmers are less likely to adopt measures. Conversely, other articles register a positive association, where the older the age, the higher the rate of adoption of technologies [83,84]. The gender variable also did not present a significant association, unlike in other studies in which women showed a greater tendency to adopt new technologies [53,85,86]. The opposite has also been reported, with women presenting a lower rate of adopting new practices [56,87]. It has been recorded that, in contexts where women bear greater responsibility in production, as may be the case in this study, there are no significant differences in this variable [88].

As expected, the income generated by livestock had a positive influence, but only on the number of measures adopted. This may be because, as an important activity for the household economy, there is a direct interest in reducing losses. Additionally, with higher incomes from livestock, households have greater capital to invest in these measures and can observe a greater return on these investments, in line with the work of Liu et al. [24].

The number of household members promoting the intensity of adoption could be explained by the greater availability of family labor for the various tasks required, especially those that demand personnel directly, such as grazing, the second most used measure. This is consistent with other studies on the adoption of agricultural technologies, which documented the positive influence of household size [53,56,89].

The orientation of the productive system towards self-subsistence negatively influenced adoption. This could be explained by the lack of resources and information, as well as the lower quality of life. These systems are generally found in remote or difficult-to-access rural areas that lack of infrastructure and basic services, leading to other immediate priorities. In addition, this condition is also related to inefficient forms of livestock management. In this context, the lack of economic alternatives may predispose these systems to accept a certain level of losses. In this sense, intensive and semi-intensive farms tended to have greater adoption compared to extensive production. This can be explained by the greater ease in implementing measures on this type of farms, particularly night confinement, which is the most used measure. Furthermore, the second most used measure, grazing, is easier to manage or unnecessary. Additionally, these farms have a greater capacity to build or adapt pens to protect against carnivores. Dual-purpose livestock had a negative influence on both adoption and intensity, likely because most of this type of livestock is concentrated on extensive farms.

The total number of animals and the losses caused by predation (standardized in AUE) influenced the number of measures implemented, but not their adoption (Table 6). This could be explained by a lack of confidence in the measures among those who have not taken any. On the other hand, those with larger herds or constant losses may have already implemented measures and are likely to notice benefits from adopting additional anti-predation practices.

Membership in any organization (cooperatives, agricultural communities, and breeding committees) did not show any association with taking measures against predators, neither in the decision to adopt nor in the number of practices carried out (Table 6). This contrasts with other studies that found a higher probability of adopting new technologies, likely due to information exchange [55]. However, it aligns with other findings that did not identify a significant relationship [57].

Contrary to expectations and previous studies [90,91], veterinary advice and management negatively influenced both the adoption and the intensity of the measures. This could be explained by the fact that health management itself can be considered as a measure against predator attacks. As documented, animals in better health present a lower risk of predation. Consequently, livestock farmers may experience fewer losses and thus reduce the adoption of other measures. Alternatively, this negative influence might be due to additional costs; the drugs used in veterinary care increase operating expenses, making livestock farmers more reluctant to invest in additional protective measures. There may also be competition for limited resources, such as time and money, which must be allocated to herd management. Producers may prioritize ensuring the health and performance of the livestock over other measures, or they may perceive threats as low. In this regard, Leguesse [53] recorded that, among producers of small ruminant herds in Ethiopia, the adoption of “veterinary packages” that improved production was higher in sectors with higher incomes and/or a greater number of animals. With respect to technical advice, the lack of information on options to deal with predation could be a factor. This may also be associated with the focus of the advice, which often centers on other productive aspects and neglects this issue. González [92] mentioned the negative effect of extension services on the adoption of various practices, attributing this to technicality. Additionally, some studies show a high rejection rate of state subsidies for measures to prevent predation in herds by the organizations involved in consultancies [93], leading to disappointment or a lack of interest among farmers. Along these lines, Garforth [94] claimed that while livestock farmers are rational, their behavior is influenced by differences in values, motivations, social influences and behavioral types, and the level of trust in advice and information sources.

According to Ramírez-Álvarez et al. [95], there is a need to understand human–carnivore coexistence by considering local realities and reductions in implicit risks for humans, both globally and locally, which is likely the only way to secure the long-term conservation of pumas in human-dominated landscapes. Along these lines, Jaime [96] proposed beginning a community program that promotes the creation of protected areas in the study area, so that animals that damage goat livestock and natural ecosystems can recover, promoting work between goatherds and government agencies. Alternatives, such as economic compensation and community livestock insurance, appear to be efficient options when accompanied with legal regulations and animal management plans to prevent damage [97].

Another relevant aspect is education regarding the ecological importance of wildlife [45]. Several activities, such as meetings and seminars, can be implemented to improve the herder’s management of livestock to minimize losses by predation.

One limitation of the study is that the approach is unable to capture individual differences that influence the attitude of the farmers. Qualitative methods can help along these lines, and future research can consider these approaches to develop a more comprehensive picture of the reality of study. Future research must also consider aspects related to coexistence in land use planning, education aspects, and economic mechanisms in order to improve the outcomes of interventions addressing the complex issue of human–wildlife conflict. A sustainable development approach can help in view of multiple demands related to nature and production.

5. Conclusions

The conflict with wildlife and farmers is a complex and widespread issue with many facets. This study highlights the differences among farmers in the Coquimbo Region regarding their adoption of measures to prevent herd predation. Using a two-stage econometric model, we found that the approaches and types of measures differ, for example, for a self-subsistence producer engaged in the extensive management of dual-purpose small livestock in the Province of Elqui, compared to a specialized intensive goat farmer in the Province of Limarí, or a semi-intensive livestock farm in the Province of Choapa. Therefore, the focus of both governmental organizations and NGOs should be on the profile of the affected livestock farmer, considering mainly productive, economic and socio-geographical factors, rather than on a standardized package of measures.

Additionally, the negative impact of the consultancies is notable. Resources already exist through INDAP, with investment support instruments that can include improvements in corrals, electric fences, and herd protection dogs. It seems that they are not being implemented adequately, which is why the work of project formulation by the technical teams of PRODESAL and PADIS should be oriented towards the prioritization of these issues. On the other hand, this corroborates the importance of working on this problem in an inter-institutional manner, with other organizations involved such as the SAG, the Regional Governments, and the corresponding Municipalities, through their emergency or productive development offices. This collaboration could influence the efficient use of existing financing lines (such as IFP (Incentive for productive strengthening), PDI (Investment development program) and FOA (Annual operating fund) of INDAP), or those that could be created in the near future to combat this complex conflict.

The complexity of human–wildlife conflict requires the planning of goals and projects that considers the different realities among producers and territories in view of a sustainable development approach.