THERYA, 2025, Vol. 16(2):249-258

Delimitation of regional management units for desert
bighorn sheep in Baja California. An application
of the potential species distribution model

Enrique De Jesús Ruiz Mondragón1, Fernando Isaac Gastelum Mendoza1, Guillermo Romero Figueroa1*, Crystian Sadiel Venegas Barrera2,
María Evarista Arellano García1, Israel Guerrero Cardenas3, Eloy Alejandro Lozano Cavazos4, and Raul Valdez5

1Facultad de Ciencias, Universidad Autónoma de Baja California. Carretera Ensenada-Tijuana 3917, CP. 22860, Ensenada. Baja California, México. Email: ruize56@uabc.edu.mx (EJRM); gastelummendozaisaac@gmail.com (FIGM); gromero4@uabc.edu.mx (GRF); evarista.arellano@uabc.edu.mx (MEAG)

2Instituto Tecnológico Ciudad Victoria. Boulevard Emilio Portes Gil 1301, CP. 87010, Ciudad Victoria. Tamaulipas, México. Email: crystianv@gmail.com (CSVB)

3Centro de Investigaciones Biológicas del Noroeste. Instituto Politécnico Nacional 195, CP. 23096, La Paz. Baja California Sur, México. Email: guerrero04@cibnor.mx (IGC)

4Universidad Autónoma Agraria Antonio Narro. Antonio Narro 1923, CP. 25315, Saltillo. Coahuila, México. Email: alejandrolzn@yahoo.com (EALC)

5Department of Fish, Wildlife and Conservation Ecology, New Mexico State University. University Avenue 1780, CP. 88003, Las Cruces. New Mexico, USA. Email: rvaldez@nmsu.edu (RV)

* Corresponding author: https://orcid.org/0000-0002-4191-9828

Models of the potential geographic distribution of species are decision-making tools for wildlife population management, especially for species with broad ranges, such as bighorn sheep. In the present study, a potential geographic distribution model was generated for managing bighorn sheep in Baja California, Mexico. The model was produced with the maximum-entropy algorithm to estimate the geographic range of the species. The variables used as predictors were climate, relief, and vegetation. Meanwhile, known sites where bighorn sheep were recorded were obtained from aerial counts in Sierra Juarez in 2012 and at the regional level in 2021 by Romero-Figueroa et al. (2024). Additional records of terrestrial observations used were reported by Ruiz-Mondragón et al. (2023) for Sierra Juarez in 2016, Sierra La Asamblea and Calamajué in 2021, and Sierra Santa Isabel and Sierra Juarez in 2022, as well as records from the Global Biodiversity Information Facility (GBIF 2021). The geographic distribution model revealed that bighorn sheep in the state of Baja California are distributed along the mountain range of the Gulf of California, covering an approximate area of 317 160 ha. The variables that contributed the most to the construction of the model were roughness, type of vegetation, and precipitation of the coldest quarter. The geographic distribution model was used to define 12 regional management units for the species. Each unit is shared between two or more agrarian communities. In Baja California, bighorn sheep should be managed through community monitoring, habitat protection, and sustainable use programs with the participation of all rural communities that own land within the distribution range of this species.

Los modelos de distribución geográfica potencial de especies son herramientas para tomar decisiones sobre el manejo de las poblaciones de vida silvestre, especialmente de especies que ocupan grandes extensiones de área, como el borrego cimarrón. En el presente estudio se generó un modelo de distribución geográfica potencial que puede ser utilizado para el manejo del borrego cimarrón en el estado de Baja California, México. El modelo se generó con el algoritmo de máxima entropía para estimar el área de distribución geográfica de la especie. Las variables utilizadas como predictoras fueron climáticas, de relieve y de vegetación. Mientras que, los sitios conocidos donde se registró al borrego cimarrón se obtuvieron de conteos aéreos realizados en Sierra Juárez en 2012, y a nivel regional en 2021, por Romero-Figueroa et al. (2024). Asimismo, se incluyeron registros de observaciones terrestres reportados por Ruiz-Mondragón et al. (2023) en Sierra Juárez en 2016; en la Sierra de la Asamblea y en Calamajué en 2021; y en Sierra Santa Isabel y Sierra Juárez en 2022. Así como, del Sistema Global de Información sobre Biodiversidad (GBIF 2021). El modelo de distribución geográfica reveló que la especie en el estado de Baja California se distribuye a lo largo del macizo montañoso del Golfo de California, en una superficie aproximada de 317,160 ha. Las variables que contribuyeron más en la construcción del modelo fueron la rugosidad, el tipo de vegetación y la precipitación del trimestre más frío. El modelo de distribución geográfica se utilizó para definir 12 unidades de manejo regional para la especie. Cada una se comparte entre dos o más comunidades agrarias. En Baja California el manejo del borrego cimarrón se debe realizar a partir de programas de monitoreo comunitario, protección del hábitat y aprovechamiento sostenible en los que, se considere la participación de todas las comunidades rurales que poseen terrenos dentro del área de distribución de esta especie.

Keywords: big game species; ecological niche model; maximum entropy; Ovis canadensis; wild sheep; wildlife management.

Introduction

Management units are geographic areas delimited for the conservation and sustainable management of wildlife species and their habitats (Swihart et al. 2020). However, their definition and delimitation consider only political boundaries without taking into account the biological aspects of the species (Bischof et al. 2016). An example of this issue is the management of bighorn sheep (Ovis canadensis) populations in Mexico. The distribution ranges of this species frequently surpass the boundaries of management units, which are delimited by property lines where there are no barriers restricting the movement of animals (Rubin et al. 2009; Ruiz-Mondragón et al. 2018). Therefore, to achieve sustainable bighorn sheep management in Mexico, it is necessary to design and delimit regional management units (RMUs) whose boundaries match the limits of the distribution area of the populations of the species (Gallina-Tessaro et al. 2009).

Potential geographic distribution models (PGDM) of species are useful tools for predicting the area of occurrence of species, identifying the environmental characteristics of the sites where they thrive, and projecting their presence over wider areas (Guisan and Thuiller 2005; Melo et al. 2020). These models are useful for designing efficient strategies for in situ management of populations, since they allow focusing monitoring and protection actions on the specific areas of the habitat used (Guisan and Thuiller 2005; Villero et al. 2017). One of the main applications of PGDMs in wildlife population management is RMU delimitation (Rodríguez-Soto et al. 2011; Maciel-Mata et al. 2015).

PGDMs are used primarily in the formulation of conservation and management plans and programs for at-risk species of hunting importance or with wide ranges (Guisan and Thuiller 2005; Refoyo et al. 2014; Eyre et al. 2022), such as wild ungulates (Ortíz-García et al. 2012; ENETWILD consortium et al. 2022). However, the reliability and functionality of the PGDM results depend on several factors: scale of the study area, accuracy and randomness of the geographic records of the species, and relevance, autocorrelation, and resolution of the environmental variables used as predictors (Austin 2007; Mateo et al. 2011).

Bighorn sheep are found in extensive mountain ranges with steep slopes, deep cliffs, and large canyons, where vegetation cover is scarce and temperatures are extreme, making it difficult to carry out regular and representative monitoring of population size and dispersal (Hansen 1980; Álvarez-Cárdenas et al. 2009; Ruiz-Mondragón et al. 2018). In the state of Baja California, Mexico, bighorn sheep populations inhabit 13 mountain ranges that cover an area of approximately 967 910 ha from La Rumorosa, on the border with the United States of America, to the Agua de Soda mountain range, located 50 km north of the border with Baja California Sur, Mexico (Romero-Figueroa et al. 2024). In this range, sheep populations are concentrated in areas where conditions are suitable for their persistence (Simmons and Hansen 1980), such as rugged terrain, open vegetation cover, presence of medium-sized shrubs (less than 1.5 m high), and water availability (Hansen 1980; Ruiz-Mondragón et al. 2018; Jones et al. 2022).

This study aimed to develop a PGDM for bighorn sheep, aiming to determine the RMUs for the species in Baja California. The following research questions were addressed: a) Where are the areas with the largest bighorn sheep populations? b) What is the extent of its potential range? c) What is the relative importance of the environmental variables that limit their distribution? d) How many bighorn sheep RMUs can be defined in the state?

Materials and methods

Description of the study area. The study was carried out in the Mexican state of Baja California, which covers an area of 71 446 km². This region extends from 32°43’07” to 28° N, and from 112°17’48” to -118°21’54” W (Figure 1). The main relief forms include mountain ranges, hills, plateaus, descents, plains, valleys, and dunes (INEGI 2001). The predominant climates in the region are temperate dry, very warm very dry, semi-warm very dry, temperate very dry, semi-cold sub-humid, and temperate sub-humid (García and CONABIO 1998). The dominant vegetation types are chaparral, microphyllous scrub, rosetophyllous scrub, and sarcocaulescent shrub (INEGI 2021).

Baja California has 13 mountain ranges where wild bighorn sheep populations thrive (Romero-Figueroa et al. 2024; Figure 1). Land tenure at these sites is mainly ejido (RAN 2016). Similarly, the bighorn sheep distribution area is partially located within natural protected areas (NPAs): the Sierra de San Pedro Mártir National Park (PNSSPM, in Spanish) and the Valle de los Cirios Flora and Fauna Protection Area (APFFVC, in Spanish; CONANP 2024).

Generation of the database of bighorn sheep presence records. The bighorn sheep presence database was built from records of the species obtained from sampling campaigns carried out in different years. This study used data from aerial monitoring conducted by the San Diego Zoo in Sierra Juárez in 2012 (Ruiz-Mondragón et al. 2018) and from the surveillance flight by the Autonomous University of Baja California in 2021 that covered the 13 distribution areas recognized for bighorn sheep in Baja California (Romero-Figueroa et al. 2024). Additional records used regard direct sightings and indirect evidence of the presence of the species (fecal groups and footprints) obtained in terrestrial monitoring as part of a study of the distribution of bighorn sheep in the Sierra Juárez mountain range in 2016 (Ruiz-Mondragón et al. 2018); in terrestrial monitoring carried out within the framework of the Cimarrón Sanctuary project in the Sierra de La Asamblea and Calamajué in 2021; and in surveillance and camera-trap installation tours in Sierra Santa Isabel and Sierra Juárez in 2022 as part of a participatory community monitoring program (Ruiz-Mondragón et al. 2023).

The database was supplemented with records available in the Global Biodiversity Information System (GBIF; 2021), which were refined to exclude occurrence points in the sea or outside the distribution area reported in the literature. Records within 2 km of one other were excluded from our database to reduce spatial bias in the data (Merow 2013).

Predictors of environmental variables. Climate, relief, and vegetation were the environmental variables used to generate the PGDM for bighorn sheep (Rubin et al. 2009; Ruiz-Mondragón et al. 2018; Salas et al. 2018). Of these, 19 variables were bioclimatic, two were topographic (orientation and terrain roughness index), and two were vegetation variables (vegetation type and enhanced vegetation index). Geospatial information was handled in raster format and was processed in the QGIS 3.22.10 geographic information system (QGIS Development Team 2022).

The bioclimatic variables were generated by Cuervo-Robayo et al. (2013) for Mexico with a 90 m spatial resolution, available at the Idrisi Resource Center of the Autonomous University of the State of Mexico (UAEM 2021). These variables were rescaled to a 30 m spatial resolution to match the resolution of the relief and vegetation variables. Orientation and terrain roughness index (TRI) data were extracted from a digital elevation model with a 30 m resolution (INEGI 2013). The vegetation type was obtained from the Series VII of the Land Use and Vegetation Layer of Mexico (INEGI 2021) rasterized with a 30 m pixel resolution. The Enhanced Vegetation Index (EVI) was calculated from Landsat 8 OLI/TIRS satellite images captured between October and November 2022 (USGS 2022), which correspond to the months when the local vegetation greened after the passage of Hurricane Kay, thus favoring the discrimination power of the vegetation index. A mosaic was constructed with the EVI images that covered the entire study area. The generated geospatial information layers were adapted to match the extent of the study area.

The variance inflation factor (FIV = 1*[1-r2]-1) was used as a criterion to exclude redundant variation between variables (Akinwande et al. 2015). The index was obtained from multiple regressions used to estimate the correlation of the variables considered for the potential distribution model; the variables whose information was contained in any other variable were excluded (FIV > 5; Alvarado-Avilés et al. 2020). From this analysis, eight environmental variables were selected for use as predictors to estimate the distribution area of bighorn sheep in Baja California (Table 1).

Prediction of potential geographic distribution. The Maxent 3.4.4 program (Phillips et al. 2020) was used to predict the geographic distribution area of bighorn sheep in Baja California. The algorithm was implemented based on the criteria by Phillips et al. (2006) for basic niche modeling with a logistic output format, and the result indicates the probability of occurrence of the species of interest in a geographical space. The model was generated with a mean of 50 replicates with 1,000 iterations each. We used 80% of the localities of occurrence to construct the model and 20% to validate it. The predictive accuracy of the model was determined by calculating the area under the curve (AUC) of the receiving operating characteristic (ROC), and the fraction of sites classified erroneously as absences (omission errors) was determined by calculating the omission rate and the mean predicted area.

The number of replicates used for the model construction was determined based on the normality of the distribution of AUC values of the replicates (Plasencia-Vázquez et al. 2014) using the Shapiro-Wilk test; it was found that the 50 replicates fit a normal distribution (W = ٠.٩٦; p = 0.11). In the Maxent program interface, the Do Jacknife to measure variable importance and Create response curves options were activated. The Jacknife analysis was performed to evaluate the percentage of contribution of variables to the model generation. Response curves were created to determine the range of values for each variable within which the species is likely to occur (Phillips et al. 2006).

The continuous predictive model was transformed into a binary model (presence-absence). The cut-off threshold was determined from the mean of the minimum presence of training of the 50 replicates generated (Alvarado-Avilés et al. 2020). The binary model was projected over the Baja California mountain ranges (INEGI 2001), whose contours were used as the basis to define bighorn sheep RMUs, since these relief forms include the suitable habitat for the populations of this species (Hansen 1980; Álvarez-Cárdenas et al. 2009). The boundaries of RMUs were established at the points where the decrease in the concentration of binary-model pixels coincided with the boundaries of the mountain ranges, which was interpreted as an indication of the presence of barriers that limit the dispersal of organisms and, therefore, as the natural limit for a given bighorn sheep population (Epps et al. 2007). Similarly, the binary model was projected onto NPA polygons (CONANP 2024) to determine the fraction of the potential geographic distribution area found within an NPA in Baja California.

Results

A total of 509 records of bighorn sheep were obtained for Baja California: 183 from population monitoring and 326 from the GBIF. After excluding records less than 2 km from each other, the database was reduced to 201 locations: 102 from population monitoring and 99 from the GBIF (Figure 2). The potential bighorn sheep distribution area in Baja California was calculated at 317 160 ha (4 % of the state area), stretching along the mountain massif of the Gulf of California coast from the US border to the Baja California Sur border. The AUC of the potential distribution model was 0.93, with a standard deviation of 0.02, and the omission rate was 0.40 (Figure 3).

The most important variables for the PGDM were roughness, vegetation type, and precipitation of the coldest quarter, which together contributed 78 % to the model construction. Each of the remaining variables contributed less than 10% (Table 2). According to the maximum-entropy algorithm, bighorn sheep in Baja California thrive in places where the roughness index varies between 35 m and 165 m covered by microphyllous scrub, sarcocaulescent scrub, riverbank vegetation or natural palm grove, and the precipitation of the coldest quarter ranges between 30 mm and 55 mm.

We defined 12 Regional Management Units (RMUs) for bighorn sheep in Baja California (Table 3; Figure 4). Within these RMUs, 85 % (271 044 ha) of the bighorn sheep distribution area is located on ejido land shared by 22 agrarian communities in the state. Furthermore, 23.6 % (71 520 ha) of the potential geographic distribution area of bighorn sheep is within a natural protected area: 22.3 % (70 626 ha) in Valle de los Cirios and 0.3 % (889 ha) in San Pedro Mártir.

Discussion

The most recent proposals regarding bighorn sheep distribution in Baja California were those of Lee et al. (2012) and Gutiérrez-Granados et al. (2020). Lee et al. (2012) indicate that bighorn sheep are distributed throughout the Gulf of California mountain range and interrupted in the Agua de Soda mountain range, approximately 50 km north of the border with Baja California Sur (Figure 1). For their part, Gutiérrez-Granados et al. (2020) point out that the species is distributed continuously throughout the Gulf of California mountain range. Both proposals are inconsistent with the PGDM generated in this study, which indicates that in Baja California, bighorn sheep are distributed in patches throughout the Gulf of California mountain range (Figure 2).

Lee et al. (2012) defined the bighorn sheep distribution area in Baja California as the mountain ranges in which sightings were recorded in the aerial monitoring performed in the state. Therefore, the delimitation of the distribution area was conditioned by the geographical scope of the aerial surveillance. This explains why the authors point to the Agua de Soda mountain range as the distribution limit of bighorn sheep, since there were no flights south of this mountain range (Romero-Figueroa et al. 2024). However, despite its limitations, the distribution proposed by Lee et al. (2012) was hugely relevant because it represented the first effort to define the boundaries of the bighorn sheep range in Baja California, and is, therefore, one of the main references used to define the study area of the research on bighorn sheep in the state (Escobar-Flores et al. 2015; Ruiz-Mondragón et al. 2018; Ruiz-Mondragón et al. 2023; Romero-Figueroa et al. 2024).

On the other hand, Gutiérrez-Granados et al. (2020) constructed a PGDM to delimit the distribution area of bighorn sheep. Their work was also an important contribution to the matter because it defined the limits of the habitat available for the species in Baja California. However, it proposes a potential distribution area with an atypical pattern for any wild sheep species since they are not distributed evenly throughout the habitat but tend to concentrate around patches of habitat that provide the resources required by the species to survive, such as water, food, and escape ground (Bleich et al. 1990; Epps et al. 2007; Rubin et al. 2009; Salas et al. 2018). Another constraint of the PGDM generated by these authors is the size of the calculated distribution area, as it is too large to be used in decision-making on the management and monitoring of the species.

The lack of precision of the PGDM by Gutiérrez-Granados et al. (2020) is attributed to the inclusion of altitude and slope as predictor variables, the use of bioclimatic variables from WorldClim (Hijmans et al. 2005), and the use of a database of species occurrence made up entirely of GBIF records. In this type of analysis, altitude and slope are variables with marginal influence on the distribution of wild ungulates that inhabit mountainous areas (Keya et al. 2016; Khan et al. 2016; Ruiz-Mondragón et al. 2018; Salas et al. 2018). The low spatial resolution of WorldClim bioclimatic surfaces is a source of uncertainty for the model, as they do not reflect climate variations at the local level (Harris et al. 2014; Stewart et al. 2022). GBIF is a website with a particularly pronounced spatial bias due to uneven sampling effort, storage, and mobilization of data between the different areas in the range of a species, in addition to the lack of certainty about the quality of the data uploaded to the platform (Beck et al. 2014).

The PGDM presented in this study was developed from a database in which spatial bias was reduced by incorporating a similar number of records from field monitoring and the GBIF (Beck et al. 2014). Likewise, it was constructed using high-resolution bioclimatic variables developed especially for Mexico (Cuervo-Robayo et al. 2013), in addition to other predictor variables that are highly correlated with the distribution of bighorn sheep in the PGDM: terrain roughness, orientation, vegetation cover, and vegetation type (Rubin et al. 2009; Ruiz-Mondragón et al. 2018; Salas et al. 2018). This resulted in a PGDM showing a clustered distribution along a mountain range, consistent with the distribution pattern reported for the species (Bleich et al. 1990; Epps et al. 2007; Rubin et al. 2009; Ruiz-Mondragón et al. 2018). This suggests that this PGDM provides a more accurate representation of the distribution of bighorn sheep in Baja California than the one developed by Gutiérrez-Granados et al. (2020). In addition, it is a more useful tool for decision-making than the model of Gutiérrez-Granados et al. (2020), since it reduces by 75 % the distribution area proposed by these authors, facilitating the identification of areas of importance for the species.

The PGDM AUC assessment indicates good accuracy in discriminating between suitable and unsuitable sites for the species. However, the calculated distribution area is probably smaller than the actual range, as the omission rate of the training points did not fully match the predicted omission rate (Phillips et al. 2006; Figure 3).

The analysis of the contribution of the variables to the construction of the model indicated that roughness was the most relevant habitat component for bighorn sheep in Baja California (Table 3). This is an important variable for wild sheep species, as it is related to the availability of escape ground (Álvarez-Cárdenas et al. 2009; Salas et al. 2018). Furthermore, the presence of bighorn sheep was associated with sites with roughness values between 35 m and 165 m, typical of medium and high mountain ranges with canyons that provide protection to bighorn sheep in the Baja California peninsula (Álvarez-Cárdenas et al. 2009; Escobar-Flores et al. 2015; Ruiz-Mondragón et al. 2018).

The vegetation type was another variable that contributed to the construction of the PGDM (Table 3) because it indicates the availability of forage and water for the species. In addition, it is related to the predator-avoidance strategy of bighorn sheep, which consists of using patches in which the vegetation foliage does not reduce visibility and, therefore, allows detection of predators from a distance (Wilson et al. 1980; Álvarez-Cárdenas et al. 2009; Escobar-Flores et al. 2015). As in other studies carried out in Baja California, it was determined that bighorn sheep are distributed in microphyllous scrub, sarcocaulescent scrub, riverbank vegetation, and natural palm groves (Escobar-Flores et al. 2015; Ruiz-Mondragón et al. 2018).

Precipitation of the coldest quarter was the most important bioclimatic variable for bighorn sheep in Baja California (Table 3). This is attributed to the fact that the highest percentage of rainfall in the state occurs in winter (García and CONABIO 1998), and, therefore, this variable is related to water recharge in the habitat of the species. In addition, winter rains are related to the growth and flowering of quality forage for wild herbivores in Baja California (Delgadillo-Rodríguez and Macías-Rodríguez 2002).

In the state of Baja California, bighorn sheep management units are currently bounded by the boundaries of private land, ejidos, or the common use of ejidos, which have no relationship with the distribution patterns of bighorn sheep populations or their metapopulation dynamics (SEMARNAT 2022). This research outlines the first proposal for the definition of bighorn sheep management units in Baja California based on information on an ecological aspect of the species: its potential distribution. This proposal differs from the previous one by Lee et al. (2012), who proposed three RMUs: the northern RMU, which extends from Sierra Juárez to Sierra de San Felipe; the central RMU, from Santa Isabel to Sierra de La Asamblea; and the southern RMU, which includes the La Libertad, Las Ánimas, and Agua de Soda mountain ranges. This regional management proposal is based on the assumption that there are three metapopulations of bighorn sheep in Baja California; however, they do not provide evidence to support their existence. In this sense, there are also no studies showing that, in Baja California, a bighorn sheep metapopulation is distributed in two or more mountain ranges; on the contrary, genetic studies suggest that a mountain range can host more than one metapopulation (Buchalski et al. 2015).

The PGDM indicates that in Baja California the bighorn sheep shows a clustered distribution pattern, typical of wild sheep (Figure 2); that is, specimens of the species are concentrated in cores of suitable habitat connected by patches that function as biological corridors (Bleich et al. 1990; Epps et al. 2007; Rubin et al. 2009; Salas et al. 2018). The cores of suitable-habitat concentration are delimited by natural barriers that restrict the displacement of animals, which, according to the analysis of the contribution of variables, may be relatively flat areas with no escape terrain for the species (Berger 1991), a vegetation type unsuitable for the species because of a dense vegetation cover that reduces visibility (Bleich et al. 1997), or areas devoid of nutritious forage and water sources due to extremely aridity (Epps et al. 2004). This is why the boundaries of cores of suitable-habitat concentration were used to define the boundaries of RMUs for bighorn sheep, since the natural barriers between them contribute to confining the populations of the species, and, in this way, the abundance in each core of suitable-habitat concentration does not undergo significant fluctuations due to migratory processes (Epps et al. 2007; Creech et al. 2014).

The largest proportion of the bighorn sheep distribution area in Baja California is ejido land, and, in general, the land tenure of RMUs delimited in the present study corresponds to more than one ejido. This implies that each RMU should establish monitoring and protection programs for the bighorn sheep population with the participation of all landowners within the RMU (Dowsley 2009; Mandujano-Rodríguez and González-Zamora 2009). Furthermore, decision-making on the sustainable use of sheep should prioritize ejido management units and be based solely on the results of joint monitoring by all management units within an RMU (Adhikari et al. 2021).

The main benefit of regional management is the prevention of overexploitation of bighorn sheep populations. In Mexico, this problem is common to all game species at sites where the distribution area of local populations is shared between two or more individual management units (Gallina-Tessaro et al. 2009). This is due to the fact that a particular and independent exploitation quota is granted to each management unit based on the results of individual monitoring of a given local population (Mandujano-Rodríguez 2011). For this reason, to ensure the sustainable use of bighorn sheep, exploitation quotas must be established at the regional level rather than at the level of individual management units (SPA 2013; Ruiz-Mondragón 2014; 2017).

However, granting regional exploitation quotas for bighorn sheep poses a serious social challenge: the distribution of the benefits of hunting the species. The alternatives to resolve this issue could be the establishment of regional wildlife conservation management units (UMA, in Spanish) or the distribution of exploitation quotas that correspond to each RMU among the ejido management units involved, based on the fraction of the total habitat that belongs to each. The formation of regional UMAs is not considered the best option because there are considerable differences in the number of members in each ejido (Ruiz-Mondragón et al. 2023) and in the fraction of the bighorn sheep habitat owned by each (Figure 4). In this regard, it is worth noting that two regional UMAs were formed in Baja California whose viability could not be verified because they never started operations: the UMA named Ejidos Asociados de Baja California, comprising the ejidos Cordillera Molina, Hermenegildo Galeana, José Saldaña, and Plan Nacional Agrario; and the UMA Valle de los Cirios, made up of the ejidos Nuevo Rosarito, Revolución, and Tierra y Libertad. The distribution of hunting permits granted to each RMU among the ejido management units based on the fraction of the bighorn sheep habitat owned by each is considered the most viable alternative to solve the problem of the distribution of the economic benefits of sheep hunting, since the larger the area available, the greater the investment required for its management (Ortega-Argueta et al. 2016). This approach ensures that larger management units have sufficient financial resources to invest in the conservation of bighorn sheep populations and their habitat, while smaller units also participate in the economic benefits of bighorn sheep exploitation.

In Baja California, approximately 25 % of the bighorn sheep range is within an NPA, and this fraction of the habitat concentrates about 38 % of the total population of the species in the state (Romero-Figueroa et al. 2024). This situation has important implications for both bighorn sheep conservation and NPAs. On the one hand, Mexico’s National Commission of Natural Protected Areas (CONANP, in Spanish) has the power to participate in the formulation and monitoring of the correct implementation of the work plans developed to manage one-quarter of the habitat available for the species and more than one-third of the bighorn sheep population in the state of Baja California (CONANP 2006; SEMARNAT 2013). On the other hand, bighorn sheep management units within NPAs can potentially become the main promoters of the conservation of these sites, since they can provide working groups for biodiversity surveillance and monitoring, invest in infrastructure and in the implementation of habitat improvement actions, and finance productive diversification projects around bighorn sheep (Brenner and De la Vega 2014; Sandoval et al. 2019).

Conclusions

The calculated distribution area for bighorn sheep in Baja California was 317 160 ha, extending throughout the state through the Gulf of California mountain range. The most influential environmental variables in the construction of the distribution model were roughness, vegetation type, and precipitation of the coldest quarter. The predictor variables were related to the presence of escape terrain, its suitability for the predator-avoidance strategy, and water and food availability. Based on the PGDM, 12 RMUs were delimited for bighorn sheep in Baja California, whose land tenure is ejido. It is recommended that the management of bighorn sheep populations within each RMU be carried out based on monitoring, protection, and sustainable use programs for bighorn sheep populations, involving the participation of all ejidos that own land within it.

Acknowledgments

The authors wish to thank the UABC Foundation for its support in managing the financing of this project and the Autonomous University of Baja California for the support provided through project 400/2975. Finally, the first author thanks the Consejo Nacional de Ciencias, Humanidades y Tecnología (CONAHCYT) for the economic resources granted during postgraduate studies. María Elena Sánchez-Salazar, M. Sc., translated the manuscript into English.

Literature cited

Adhikari, L., et al. 2021. Community-based trophy hunting programs secure biodiversity and livelihoods: Learnings from Asia’s high mountain communities and landscapes. Environmental Challenges ٤:١٠٠١٧٥.

Akinwande, M., H. Dikko, and A. Samson. 2015. Variance inflation factor: as a condition for the inclusion of suppressor variable(s) in regression analysis. Open Journal of Statistics ٥:٧٥٤-767.

Alvarado-Avilés, J., et al. 2020. Potential distribution of Plestiodon copei (Squamata: Scincidae), an endemic and threatened lizard of Mexico. South American Journal of Herpetology 18:56-67.

Álvarez-Cárdenas, C., et al. 2009. Evaluación de elementos estructurales del hábitat del borrego cimarrón en la Sierra del Mechudo, Baja California Sur, México. Tropical Conservation Science 2:189-203.

Austin, M. 2007. Species distribution models and ecological theory: A critical assessment and some possible new approaches. Ecological Modelling 200:1-19.

Beck, J., et al. 2014. Spatial bias in the GBIF database and its effect on modeling species’ geographic distributions. Ecological Informatics 19: 10-15.

Berger, J. 1991. Pregnancy incentives, predation constraints and habitat shifts: experimental and field evidence for wild bighorn sheep. Animal Behaviour 41:61-77.

Bischof, R., H. Brøseth, and O. Gimenez. 2016. Wildlife in a politically divided world: insularism inflates estimates of brown bear abundance. Conservation Letters 9:122-130.

Bleich, V., J. Wehausen, and S. Holl. 1990. Desert-dwelling mountain sheep: conservation implications of a naturally fragmented distribution. Conservation Biology 4:383-390.

Bleich, V., R. Bowyer, and J. Wehausen. 1997. Sexual segregation in mountain sheep: resources or predation? Wildlife Monographs 134:1-50.

Brenner, L., and A. De la Vega. 2014. La gobernanza participativa de áreas naturales protegidas. El caso de la Reserva de la Biosfera El Vizcaíno. Región y Sociedad ٢٦:١٨٣-213.

Buchalski, M., et al. 2015. Genetic population structure of peninsular bighorn sheep (Ovis canadensis nelsoni) indicates substantial gene flow across US-Mexico border. Biological Conservation 184:218-228.

CONANP. 2006. Programa de Conservación y Manejo Parque Nacional Sierra de San Pedro Mártir. CONANP. D.F., México.

CONANP. 2024. Áreas Naturales Protegidas Federales de México. http://geoportal.conabio.gob.mx/metadatos/doc/html/anpenero2024gw.html. Accessed April 17, 2024.

Creech, T., et al. 2014. Using network theory to prioritize management in a desert bighorn sheep metapopulation. Landscape Ecology 29:605-619.

Cuervo-Robayo, A., et al. 2013. An update of high-resolution monthly climate surfaces for Mexico. International Journal of Climatology 34:2427-2437.

Delgadillo-Rodríguez, J., and M. Macías-Rodríguez. 2002. Componente florístico del desierto de San Felipe, Baja California, México. Boletín de la Sociedad Botánica de México 70:45-65.

Dowsley, M. 2009. Community clusters in wildlife and environmental management: using TEK and community involvement to improve co-management in an era of rapid environmental change. Polar Research ٢٨:٤٣-59.

ENETWILD consortium, et al. 2022. New models for wild ungulates occurrence and hunting yield abundance at European scale. EFSA supporting publication 19:EN-7631.

Epps, C., et al. 2004. Effects of climate change on population persistence of desert-dwelling mountain sheep in California. Conservation Biology 18:102-113.

Epps, C., et al. 2007. Optimizing dispersal and corridor models using landscape genetics. Journal of Applied Ecology 44:714-724.

Escobar-Flores, J., et al. 2015. Detección de las preferencias de hábitat del borrego cimarrón (Ovis canadensis) en Baja California, mediante técnicas de teledetección satelital. Therya 6:519-534.

Eyre, A., et al. 2022. Using species distribution models and decision tools to direct surveys and identify potential translocation sites for a critically endangered species. Diversity and Distributions 28:700-711.

Gallina-Tessaro, S., et al. 2009. Unidades para la conservación, manejo y aprovechamiento sustentable de la vida silvestre en México (UMA). Retos para su correcto funcionamiento. Investigación Ambiental 1:143-152.

García, E., and CONABIO. 1998. Climas. http://www.conabio.gob.mx/informacion/gis/. Accessed October 4, 2020.

GBIF. 2021. GBIF Occurrence Download. https://doi.org/10.15468/dl.77vmp8. Accessed May 3, 2021.

Guisan, A., and W. Thuiller. 2005. Predicting species distribution: offering more than simple habitat models. Ecology Letters 8:993-1009.

Gutiérrez-Granados, G., A. Rodríguez-Moreno, and V. Sánchez-Cordero. 2020. Modelado de la distribución potencial de especies de mamíferos reservorios y vectores de tres zoonosis emergentes en México. Proyecto JM040. Instituto de Biología, Universidad Nacional Autónoma de México, México.

Hansen, C. 1980. Habitat. Pp. 64-78 in The desert bighorn: it ́s life history, ecology and management (Monson, G., and L. Sumner, eds.). University of Arizona Press. Tucson, U.S.A.

Harris, I., et al. 2014. Updated high-resolution grids of monthly climatic observations - the CRU TS3.10 Dataset. International Journal of Climatology 34:623-642.

Hijmans, R., et al. 2005. Very high resolution interpolated climate surfaces for global land areas. International Journal of Climatology 25:1965-1978.

INEGI. 2001. Conjunto de datos vectoriales Fisiográficos. Continuo Nacional serie I. Sistema topoformas. https://www.inegi.org.mx/app/biblioteca/ficha.html?upc=702825267582. Accessed October 4, 2020.

INEGI. 2013. Continuo de Elevaciones Mexicano, 3.0. http://www.inegi.org.mx/geo/contenidos/datosrelieve/continental/descarga.aspx. Accessed September 2, 2020.

INEGI. 2021. Conjunto de datos vectoriales de la carta de Uso del suelo y vegetación. Escala 1:250,000. Serie VII (continuo nacional). http://geoportal.conabio.gob.mx/metadatos/doc/html/usv250s7gw.html. Accessed September 28, 2022.

Jones, A., et al. 2022. Desert bighorn sheep habitat selection, group size, and mountain lion predation risk. Journal of Wildlife Management ٨٦:e٢٢١٧٣.

Keya, Z., et al. 2016. Habitat suitability & connectivity of Alborz wild sheep in the east of Tehran, Iran. Open Journal of Ecology 6:325-342.

Khan, B., et al. 2016. Abundance, distribution and conservation status of Siberian ibex, Marco Polo and Blue sheep in Karakoram-Pamir mountain area. Journal of King Saud University-Science 28:216-225.

Lee, R., et al. 2012. Observations on the distribution and abundance of bighorn sheep in Baja California, Mexico. California Fish and Game 98:51-59.

Liu, H., and A. Huete. 1995. A feedback based modification of the NDVI to minimize canopy background and atmospheric noise. IEEE Transactions on Geoscience and Remote Sensing 33:457-465.

Maciel-Mata, C., et al. 2015. El área de distribución de las especies: revisión del concepto. Acta Universitaria 25:22-38.

Mandujano-Rodríguez, S., and A. González-Zamora. 2009. Evaluation of natural conservation areas and wildlife management units to support minimum viable populations of white-tailed deer in Mexico. Tropical Conservation Science 2:237-250.

Mandujano-Rodríguez, M. 2011. Consideraciones ecológicas para el manejo del venado cola blanca en UMA extensivas en bosques tropicales. Pp. 249-275 in Temas sobre conservación de vertebrados silvestres en México (Sánchez, O., et al., eds.). INE-SEMARNAT, D.F., México.

Mateo, R., A. Felicísimo, and J. Muñoz. 2011. Modelos de distribución de especies: Una revisión sintética. Revista Chilena de Historia Natural 84:217-240.

Melo, S., H. Reyes, and A. Lira. 2020. Ecological niche models and species distribution models in marine environments. A literature review and spatial analysis of evidence. Ecological Modelling 415:108837.

Merow C., M. Smirh, and J. Silander. 2013. A practical guide to MaxEnt for modeling species’ distributions: what it does, and why inputs and settings matter. Ecography 36:1058-1069.

Ortega-Argueta, A., A. González-Zamora, and A. Contreras-Hernández. 2016. A framework and indicators for evaluating policies for conservation and development: The case of wildlife management units in Mexico. Environmental Science & Policy 63:91-100.

Ortíz-García, A., et al. 2012. Distribución potencial de ungulados silvestres en la reserva de la biosfera de Tehuacán-Cuicatlán, México. Therya 3:333-348.

Phillips, S., R. Anderson, and R. Schapire. 2006. Maximum entropy modeling of species geographic distributions. Ecological Modelling 190:231-259.

Phillips, J., M. Dudik, and R. Schapire. 2020. Maxent software for modeling species niches and distributions v3.4.4. Program distributed by the author. American Museum of Natural History. New York, USA.

Plasencia-Vázquez, A., G. Escalona-Segura, and L. Esparza-Olguín. 2014. Modelación de la distribución geográfica potencial de dos especies de psitácidos neotropicales utilizando variables climáticas y topográficas. Acta Zoológica Mexicana (N.S.) 30:471-490.

QGIS development team. 2022. QGIS Geographic Information System 3.22.10. Program distributed by the author.

RAN. 2016. Perimetrales de los núcleos agrarios certificados, entidad federativa Baja California. http://datos.ran.gob.mx/filescd/dgcat/ran_da_dgcat_poligonos_nucleos_agrarios_bc.kml. Accessed October 1, 2020.

Refoyo, P., C. Olmedo, and B. Muñoz. 2014. La utilidad de los modelos de distribución de especies en la gestión cinegética de los ungulados silvestres. Animal Biodiversity and Conservation 37:165-176.

Riley, S., S. DeGloria, and R. Elliot. 1999. A terrain ruggedness index that quantifies topographic heterogeneity. Intermountain Journal of Sciences 5:23-27.

Rodríguez-Soto, C., et al. 2011. Predicting potential distribution of the jaguar (Panthera onca) in Mexico: identification of priority areas for conservation. Diversity and Distributions 17:350-361.

Romero-Figueroa, G., et al. 2024. Population and conservation status of bighorn sheep in the state of Baja California, Mexico. Animals 14:504.

Rubin, E., et al. 2009. Assessment of predictive habitat models for bighorn sheep in California’s peninsular ranges. The Journal of Wildlife Management 73:859-869.

Ruiz-Mondragón, E. 2014. Estado actual de la población del borrego cimarrón (Ovis canadensis weemsi) en la UMA Ejido La Purísima, Baja California Sur, México. UNAM. Estado de México, México.

Ruiz-Mondragón, E. 2017. Una propuesta de manejo para el hábitat del borrego cimarrón (Ovis canadensis), en Sierra Juárez, Baja California, México. UABC. Ensenada, México.

Ruiz-Mondragón, E., et al. 2018. Potential distribution model of Ovis canadensis in northern Baja California, Mexico. Therya 9:219-226.

Ruiz-Mondragón, E., et al. 2023. Community-based workshops to involve rural communities in wildlife management case study: bighorn sheep in Baja California, Mexico. Animals 13:3171.

Salas, E., et al. 2018. Habitat assessment of Marco Polo sheep (Ovis ammon polii) in Eastern Tajikistan: Modeling the effects of climate change. Ecology and Evolution ٨:٥١٢٤-5138.

Sandoval, A., R. Valdez, and A. Espinosa-T. 2019. Desert bighorn sheep in Mexico. Pp. 350-365 in Wildlife Ecology and Management in Mexico (Valdez, R., and A. Ortega, eds.). Texas A&M University Press. College Station, U.S.A.

SEMARNAT. 2013. Programa de Manejo Área de Protección de Flora y Fauna Silvestre Valle de los Cirios. SEMARNAT. D.F., México.

SEMARNAT. 2022. Unidades de Manejo para el Aprovechamiento Sustentable de la Vida Silvestre 2020. http://geoportal.conabio.gob.mx/metadatos/doc/html/umas20gw.html. Accessed June 4, 2024.

Simmons, N., and C. Hansen. 1980. Pp. 160-287 in The desert bighorn: it’s life history, ecology and management (Monson, G., and L. Sumner, eds.). University of Arizona Press. Tucson, U.S.A.

SPA. 2013. Estrategia estatal para la conservación y manejo sustentable del borrego cimarrón (Ovis canadensis cremnobates) en Baja California. Periódico Oficial del Estado de Baja California CXX: 3-126.

Stewart, S., et al. 2022. Predicting plant species distributions using climate-based model ensembles with corresponding measures of congruence and uncertainty. Diversity and Distributions 28:1105-1122.

Swihart, R., et al. 2020. A flexible model-based approach to delineate wildlife management units. Wildlife Society Bulletin 44:77-85.

UAEM. 2021. Centro de Recursos Idrisi-México. http://idrisi.uaemex.mx/. Accessed February 6, 2021.

USGS. 2022. EarthExplorer. http://earthexplorer.usgs.gov/. Accessed February 1, 2023.

Villero, D., et al. 2017. Integrating species distribution modelling into decision-making to inform conservation actions. Biodiversity and Conservation 26:251-271.

Wilson, L., et al. 1980. Desert bighorn habitat requirements and management recommendations. Desert Bighorn Council Transaction 24:1-7.

Associated editor: Rafael Reyna Hurtado

Submitted: September 3, 2024; Reviewed: December 11, 2024

Accepted: April 23, 2025; Published on line: May 16, 2025

DOI:10.12933/therya-25-6149 ISSN 2007-3364

Figure 1. State of Baja California, Mexico, showing the mountain ranges (marked in grey) in which bighorn sheep occur: (1) Cucapá; (2) Sierra Juárez; (3) Las Tinajas; (4) Las Pintas; (5) San Pedro Mártir; (6) San Felipe; (7) Santa Isabel; (8) San Francisquito; (9) Calamajué; (10) La Asamblea; (11) La Libertad; (12) Las Ánimas; (13) Agua de Soda. Also shown (in green) are the NPAs that are within the distribution area of bighorn sheep. P.N. S.S.P.M.: Parque Nacional Sierra de San Pedro Mártir. A.P.F.F. V.C.: Área de Protección de Flora y Fauna Valle de los Cirios.

Table 1. Environmental variables used as predictors to construct the potential geographic distribution model of bighorn sheep in Baja California.

Variable	Description	Units
Bio08	Mean temperature of the rainiest month	°C
Bio14	Precipitation of the dryest month	mm
Bio18	Precipitation of the warmest quarter	mm
Bio19	Precipitation of the coldest quarter	mm
Orientation	Direction of exposure of a slope	°
TRI	Degree of surface irregularity. It is calculated as changes in terrain elevation within a 3 × 3 pixel matrix and summarizes the point change in elevation in each pixel and in the 8 pixels surrounding it (Riley et al. 1999).	m
Vegetation type	According to the Land Use and Vegetation Layer of INEGI (2021)
EVI	Contrast between absorption and radiation of vegetation (Liu and Huete 1995). The index is used as an indicator of vegetation cover; it takes values between 0 and 1. Values close to zero indicate areas devoid of vegetation, while values close to one (1) are typical of areas densely covered by plant species.

Figure 2. Potential distribution of bighorn sheep in Baja California (dark red). The image shows the presence records obtained from population monitoring (green) and the GBIF (yellow) used to generate the model.

Table 2. Relative contribution (in percentage) of the environmental variables used to estimate the potential distribution area of bighorn sheep in Baja California and ranges of occurrence of the species.

Variable	Contribution (%)	Range of occurrence
TRI	38	35–165 m
Vegetation type *	23	MS, SCS, RV, NPG
Bio19	17	30–55 mm
Bio14	6	0.25–2.5 mm
Bio08	5	7–12 °C
Orientation	5	0°–100°
Bio18	4	0–65 mm
EVI	2	0–0.04

*MS = microphillous scrub; SCS = sarcocaulescent scrub; RV = riverbank vegetation; NPG = natural palm grove.

Figure 3. Area under the curve (AUC; left) and mean omission rate (right) of the 50 replicates used to estimate the potential distribution area of bighorn sheep in Baja California.

Table 3. Fraction of the distribution area of bighorn sheep in each RMU and ejidos within its limits.

RMU	Area (ha)	Percentage	Ejidos*
Las Tinajas-Las Pintas	66 019	20.8	CM, MSC, Jam, JS, PNA
Santa Isabel	59 281	18.7	RAI, Mat, Rev
Sierra Juárez	58 736	18.5	EZ, Jac, AV, CM, MSC, Jam, DS, TK
San Pedro Mártir	30 040	9.5	Tep, PNA
Cucapá	29 474	9.3	EZ, HJ, LM
San Felipe	19 388	6.1	PNA, Del, Mat
La Asamblea	18 496	5.8	Juar, TL
La Libertad-Las Ánimas	13 567	4.3	TL, NR, CNC
Agua de Soda	12 362	3.9	TL
San Francisquito	6213	2	Mat
Calamajué	2781	0.9	Mat, Rev
La Sirena	805	0.3	Ind
Total	317 160	100

*CM = Cordillera Molina; MSC = Misión de Santa Catarina; Jam = Jamau; JS = José Saldaña II; PNA = Plan Nacional Agrario; RAI = Reforma Agraria Integral; Mat = Matomí; Rev = Revolución; EZ = Emiliano Zapata; Jac = Jacumé; AV = Aubanel Vallejo; DS = Dieciséis de Septiembre; TK = Tribu Kiliwas; Tep = Tepi; HJ = Heriberto Jara; LM = López Mateos; Del = Delicias; Juar = Juárez; TL = Tierra y Libertad; NR = Nuevo Rosarito; CNC = Confederación Nacional Campesina; Ind = Independencia.

Figure 4. Regional Management Units (RMU) for bighorn sheep in Baja California (gray): a) Cucapá; b) Sierra Juárez; c) Las Tinajas - Las Pintas; d) San Pedro Mártir; e) San Felipe; f) Santa Isabel; g) Calamajué; h) San Francisquito; i) La Asamblea; j) La Libertad - Las Ánimas; k) Agua de Soda; l) La Sirena. The image shows the potential distribution of the species (dark red), NPAs (green), and ejidos that are within the limits of the RMU. P.N. S.S.P.M.: Parque Nacional Sierra de San Pedro Mártir. A.P.F.F. V.C.: Área de Protección de Flora y Fauna Valle de los Cirios.