**1. Introduction**

Mexico holds an exceptional species richness and endemicity of amphibians and reptiles,ranking in the top three countries worldwide [1]. It has over 376 species of amphibians and 864 species of reptiles, and approximately 65% and 57% of these species, respectively, are endemic [2,3]. However, it has been argued that both amphibians and reptiles are the two most vulnerable groups of terrestrial vertebrates, being at significantly higher risk than mammals and birds [4–6] for threats such as habitat loss and fragmentation. One third of the species of amphibians worldwide are threatened with extinction according to IUCN [7], and only 35% of Neotropical and Nearctic species are in the IUCN Least Concern Category [5,6]. Species of reptiles have been less studied, but they are also highly vulnerable to these threats [8], and it is likely that reptiles are at high risk of extinction as well [4]. In general, both groups are highly dependent on the environmental conditions and are very vulnerable to pathogens, invasive species, ultraviolet-B exposure, and pollution [9–12]. Several factors, such as habitat loss and fragmentation, water pollution, climate change, and mining have been identified to negatively affect their breeding activities, reproduction, and survival performance [13,14], leading to the reduction of species range size and local population extirpation.

Mexico has lost over 13.5 million ha of natural vegetation in the last 50 years [1] due to annual deforestation rates greater than 1% nationwide [15]. This habitat loss and fragmentation a ffects most ecosystems and species of flora and fauna, increasing the vulnerability and extinction risk of species [16]. Since species of amphibians and reptiles have dispersal limitations [17], their movements between habitat fragments are limited and going from unfavorable to favorable habitats is unlikely [13,14]. As a result, the Global Amphibian Assessment suggests that habitat loss impacts 89% of the threatened amphibian species in the Americas, which is three times the impact of any other threat [18]. Recent studies have proposed that human-induced habitat loss is important, affecting the diversity and abundance of amphibian and reptile species [19] in Neotropical habitats [20], xeric habitats [21], and dry plains [22]. Furthermore, small-range species are more likely to show population declines [23]. For example, 70% of Mexican species of amphibians and 80% of species of reptiles have restricted distributional ranges and high environmental specialization, increasing their vulnerability [24]. In fact, the distributions of endemic amphibian and reptile species have declined 80% and 70%, respectively [25].

In addition to habitat loss, mining activities are suspected to impose a high risk to species of amphibians and reptiles, but this has been poorly studied. Mexico ranks second in silver production worldwide and is one of the countries with the largest production of gold, zinc, copper, and other minerals [26]. There are currently 1531 mining projects (884 more than in 2010), of which 1113 projects are in the exploration stage (where perforations are made to determine the available minerals), 63 are under the construction of the mine, 274 are in the production stage where the minerals are being extracted, and 81 have been postponed [27]. After exploitation, environmental regulations recommend closing and restoring the a ffected areas. However, the companies are not forced to elaborate an integrated plan of mining closure and restoration [27]. Most mining projects are open pit mining, which is conducted at large scale, generating pollution of rivers and aquifers with heavy metals, large quantities of polluting debris, acid drainage, continuous emissions of gases and dust into the atmosphere, and the local removal of all plant and animal species [28,29]. Mining activities are considered of public utility. They are prioritized over any other use or activity in the territory and can be conducted regardless of the regime of land tenure, such as territories of indigenous people, urban areas, and private and social property [29]. There are currently more than 24,000 terrestrial active mining concessions covering over 20 million ha, and 14 marine concessions covering approximately 740 marine ha [30–32]. A total of 85% of mining activities are located in areas with vegetation cover holding ecological integrity [33,34]. Furthermore, the legislation does not restrict the possibility of establishing mining activities in most categories of protected areas [29], which has resulted in 73 mining projects covering more than 2 million ha inside protected areas and Ramsar sites; while 60,000 ha are located inside the core zones of protected areas [31,35,36].

Academic and NGO organizations assessing species extinction risk at a global level (IUCN) and at a national level (Mexican governmental ecological regulations) consider habitat loss and habitat fragmentation as the main variables to assign species extinction risk; the higher the proportion of habitat loss in their distributions, the higher the category assigned for species extinction risk [37]. However, other potential factors a ffecting the conservation status of species, such as mining activities, are largely underestimated. In this study, we (1) analyzed the combined impact of habitat loss and mining activities on potential species distributions of Mexican endemic amphibian and reptile species, and (2) determined the modifications and area needed to conserve a minimum proportion (20%) of the extant distribution of these species.

## **2. Materials and Methods**

## *2.1. Point Occurrence Data*

The study included Mexican endemic species of amphibians and reptiles. We compiled point occurrence data for 275 species of amphibians and 474 species of reptiles from the Global Biodiversity Information Facility website (GBIF; https://www.gbif.org/; accessed on 25 January 2018). We excluded

(1) all point occurrence data prior to 1970; (2) points that had a resolution lower than 2 decimals of degree or no geographic coordinates (decimal latitud = 0, empty, 99, −99); (3) fossil records; (4) alive specimens (from zoos); (5) data obtained from iNaturalist (www.iNaturalist.com.mx), since those records do not have collected and verifiable specimens, and (6) records that were found within the same pixel of the bioclimatic variables from the WorldClim, used for constructing the models (see below; 1km2). In ArcMap, we eliminated all points that did not coincide with the currently recognized distribution of the species. Once the databases were refined, we included only the species with a minimum of 10 records. The minimum number of 10 records per species was defined based on published information for an adequate species distribution modeling approach in Maxent [38].

## *2.2. Potential Species Distributions*

For each species, the polygons of the Mexican terrestrial ecoregions, including occurrence points, were selected, leaving a 50 km buffer zone surrounding them to be used as the modeling area (M region) [39,40]. The environmental variables used to construct potential species distributions were nineteen bioclimatic variables (~1 km2) from the WorldClim database (https://www.worldclim.org/; accessed on 31 January 2018) [41]. Variables with a correlation r > 0.7 were considered redundant and only one was included [42].

We generated the ecological niche models following the methodology described by Sánchez-Cordero [43]. Using the ENMeval library [44] in the R software [45], 10,000 background points were selected within the modeling area to parameterize the model, the block method was used to divide the presence data into training and testing groups [46], and 5 regularization multipliers and 13 feature classes were established to adjust the models. From a total of 65 models per species, the best model was selected based on the omission rate and the area under the curve (AUC), and projected into a discrete presence/absence map through a maximum sensitivity plus specificity threshold [47]. The area of each potential species distribution was obtained using the Consnet software package [16,48,49] by obtaining the number of cells occupied by each species and multiplying this number by 0.78, the size in km<sup>2</sup> of the used rack cells.

## *2.3. Extant Species Distributions*

From the potential species distribution models, we obtained two scenarios, as follows: (1) Extant species distributions due to habitat loss, and (2) extant species distributions due to habitat loss and mining activities. Habitat loss was estimated based on the land use and vegetation coverage map [33], which contains information on habitat transformation since 1968, and includes transformed areas into agricultural, rural, or urban settlements. For mining activities, we used the official mining concession map, which has information of all mining concessions since 1942 that are still active today [50]. Furthermore, we used the software package ConsNet on potential species distribution, extant species distribution due to habitat loss, and extant species distribution due to habitat loss and mining activities to analyze the area of occupancy of each species under each scenario.

Potential species distributions were compared with extant species distributions under each scenario and the percentage of the reduction in their distributions was obtained. We divided species in four groups according to percentage ranges of their distribution loss, as follows: (1) Species that lost <30% of their distribution; (2) species that lost between 30–50% of their distribution; (3) species that lost between 50–80% of their distribution, and (4) species that lost >80% of their distribution.

## *2.4. Selection of Priority Areas for Conservation*

For each scenario of potential species distribution, extant species distributions due to habitat loss, and extant species distribution due to both habitat loss and mining activities, we obtained the selection of priority areas for conservation using the ConsNet software package, which allows the identification of conservation solutions by defining multiple previously set criteria [16,48,49]. ConsNet allows searching for the best solutions for different objectives according to the required conservation plan. For example, it can search for the minimum selected area and the best surrogate representation without considering any other restrictions, such as shape or connectivity. Similarly, a conservation solution can be searched for by considering, for example, the best representation, minimum area, connectivity, and/or shape configuration. The conservation target for all species was set to 20% of their distribution under the three scenarios, considering that all species are endemic and have limited distributions. We searched for the best representation with the minimum area and shape using the RF4 adjacency algorithm with a basic neighbor selection, and running 200,000 iterations to find the best solution. We obtained the total area (km2) of conservation, perimeter, number of clusters, shape, and total representation of the species on the conservation area network.
