Ant species records were retrieved from Dáttilo et al. (2020) and combined with the newly-recorded species in this study. A geometric interpolation-based approach was used to review the species richness patterns. We used the package “sf” (Pebesma 2018) in R 4.0.3 to generate a gridded (0.5 degrees) species richness map of Mexico. To adjust the species richness weighted by distances, we use a tuning value of p = 0.5, which combines high weights for low distances and low weights for high distances. The weighted species richness was also adjusted by a sampling effort considering a factor based on the ratio of the number of species recorded in each quadrat and the maximum species number reported for each centre of species richness in the grid. Hence, the maximum relative sampling effort in each quadrat will be closer to 1 (Raedig et al. 2010). We consider a narrow endemic species sensu Gentry (1986) when the species interpolated distribution range is equal to or less than five quadrats (5000 km² approximately) inside the grid. This limit of maximum distribution has been previously applied to delimit insect endemism (Torres-Cambas et al. 2016, Grill et al. 2002, Mosconi et al. 2014, Bazelet et al. 2016). To estimate the distribution of the narrow endemics, we propose a factor of 0.5 as the limit of the upper number of occurrences per grid (Morawetz and Raedig 2007). To perform these estimations and map the distribution patterns, we used the R package “sperich” (Lange et al. 2012). We associated the vegetation type and land cover type per record as in Dáttilo et al. (2020) to caclculate a correlation of state-based species richness over habitat richness in R 4.0.3. We used the habitat categories from INEGI (2007) which cover all the territory of Mexico at a scale of 1:250 000 and is the result of a standardised and field verified process of the categorisation of the vegetation and land use coverage, frequently used for ecological analyses of Mexico.

Field collections were performed from March 2016 to November 2017 in 14 States of Mexico. Collection sites (Table 1) represented ecologically diverse habitats in the Nearctic and Neotropical Regions of Mexico (Morrone 2010). We aimed to retrieve abundant and/or ecologically-dominant ant species in each habitat, trying to represent different ecological preferences, life strategies and phylogenetic lineages. All samples from this study were hand-collected. The average area of the sampled localities was 12 km² ± 27 km². Sampling was performed for at least five hours per locality and sites were selected as far as possible from human settlements. Whenever possible, ants were collected from the same nest or at the colony foraging pathway. Specimens were preserved in molecular grade ethanol (99%). Each colony location was geopositioned by using a Garmin GPSmap 62s or by the app GPS Essentials v. 4.4.25. Habitat type associated with each sample was recorded in the field and corroborated by the vectoral metadata of vegetation and land use series IV chart (INEGI 2007). The annual mean temperature (amt) and precipitation (amp) from the climate of each locality was retrieved by the WorldClim V.2.0 database (Gotelli et al. 2011, Fick and Hijmans 2017). All environmental variables were related to the geopositioned colonies on QGIS 3.10.1.

We report 19 new distributional records for 14 species from central and north Mexico. The record for Stenamma schmitii Wheeler, 1903 is the first for Mexico, while 13 of them are new state-level records. The species belong to 11 genera and four subfamilies (Dolichoderinae, Dorylinae, Formicinae and Myrmicinae).

New records were found in seven States in central Mexico and one in Nuevo Leon in the north part of the country. Most of these records (68%) were found in undisturbed habitats, such as natural grasslands, oak forests and shrublands associated with the oak forests. We also found four records in induced grasslands and one in a human settlement, A Quinta, at the ridge of the Cumbres mountain range. The localities where we found these ants were of high elevation (14 were at 2050 m a.s.l. or more) and low annual mean temperature (17.86 ±2.63⁰C) with precipitation that ranged from 172.25 to 45.5 mm/cm².

The analysis of 21,741 records of 888 species distributed in 856 quadrats of 0.5° revealed a maximum species richness of 251 with an average of 14.82 ± 27.8 species per quadrat. One third of the quadrats had no species recorded (Fig. 1).

The highest ant species richness was found in the southeast region of Mexico, stretching along the southeast part of Sierra Madre Oriental and Sierra de Chiapas mountain systems. There are also two separated quadrats with high richness in central Jalisco and eastern Quintana-Roo (75 and 102 species per quadrat, respectively, Fig. 1). The quadrat with the highest species richness (251 species) is located at the southern border of the country between Tabasco and Chiapas States. However, the weighted species richness adjusted by sampling effort shows only one large area (1.14 x 105 km²) of high diversity with a centroid in Tabasco, Veracruz de Ignacio de la Llave and Chiapas States and a single quadrat in Quintana-Roo (Fig. 2).

Although one third of the quadrats have zero species recorded, after adjusting the species richness by sampling effort, the north-eastern border of Mexico is the only zone that lacks any recorded species (Fig. 2). Both species richness and adjusted species richness suggest the same hotspot in south-eastern Mexico. Considering the proportion of quadrats of zero species (Fig. 1), the most undersampled zones are represented by the north region of Mexico, specifically the States of Sinaloa, Coahuila, Nuevo León, Tamaulipas, Zacatecas and San Luis Potosi. There is also an undersampled region on the southern coast of Michoacán de Ocampo, Guerrero and Oaxaca.

The highest number of endemics (34) was found in the same centroid as the highest species richness, which corresponds to the Neotropical zone of Mexico. Other areas with high endemism are located the northern part of the States of Sinaloa and Baja California.

Even though Mexico has more than 3000 islands (INEGI 2015), only 14 of them have been reported to have ant species (Dáttilo et al. 2020) and, in our analysis, three of them appeared to have a positive narrow endemic index (Guadalupe in the Pacific, Angel de la Guarda in the Sea of Cortes and Cozumel in the Caribbean, Fig. 3).

Six of the species recorded here (N. melanocephalus, N. rugulosus, C. coruscus, C. striatus, C. trepidulus and T. andrei) belong to the genera with the most species recorded in the country, such as Camponotus, which represents almost 10% of the species in Mexico (Dáttilo et al. 2020).

Half of the new reported records were collected in Guanajuato, a State which has a rather low habitat richness (number of different habitat types) and which has been a poorly-sampled region of Mexico (Table 2), with one of the lowest number of native species recorded (37). In contrast, the State of Tabasco has comparable habitat richness, but a dramatically higher number of records than Guanajuato (1251 and 176, respectively).

Alongside Guanajuato, States Michoacán de Ocampo, Aguascalientes, Zacatecas and San Luis Potosí have low numbers of ant records (Table 2). However, these States converge in a region where narrow endemic species are distributed and appear to be in an ant endemism hotspot (Fig. 3) that corresponds to the transition zone between the Nearctic and Neotropical areas of Mexico. The Mexican transition zone has been reported as a centre of endemism of other taxa (plants, mammals) and the Trans-Mexican volcanic belt could have a specific taxa with a primary biogeographic homology (cenocron) where vicariance events were initiated by the volcanic events (Halffter and Morrone 2017, Morrone 2010).

Dáttilo et al. (2020) report only 33 species in 166 records in Guanajuato within 11 different habitats (only 56% of the records have the habitat reported). In comparision, Veracruz de Ignacio de la Llave has 36 different habitat types, 4329 records and 454 species documented. Most of the narrow endemics are also found in Veracruz de Ignacio de la Llave (Fig. 3), where habitats seem to be more diverse and species richness is higher, even after adjusting the values by sampling effort (Fig. 2). This State has a unique geographical position and its species richness and diverse landscape types could be a result of the of the altitudinal gradients ranging from sea level to 5700 m a.s.l. at the Pico de Orizaba.

We have to also consider two field biological stations that have been centres of ant research in Mexico, one from the National Autonomous University of Mexico (UNAM) and the other from the Mexican Institute of Ecology (INECOL). They are located in Veracruz de la Llave and are influencing the number of records and, thereby, the endemicity indexes in their area. This can be also a factor that leads to the two separate quadrats of high richness found in Quintana-Roo and Jalisco, where entomologists from the University of Guadalajara and from the South Border College (Ecosur) conducted numerous field expeditions.

Habitat richness plays an important role in ant species distribution and should be considered when proposing new surveys. We found a significant correlation (R²₍₃₀₎ = 0.55 p < 0.0001) for ant species richness and habitat richness in the state-based records of Mexico (Dáttilo et al. 2020). A similar relation (R² = 0.40) was found by Ryder Wilkie et al. (2010) in Amazonian Ecuador using a multiple regression model with environmental variables and species richness.

The species richness, adjusted by sampling effort, reveals a zone of high species richness forming a corridor from central Jalisco across the border between Michoacán de Ocampo and Guanajuato and connecting to the main hotspot in Veracruz de Ignacio de la Llave, passing through the Trans-Mexican Volcanic Belt. (Fig. 2).

Ríos-Casanova (2014) indicated a direct relationship between the species richness and the area of each State, but also showed that the collection effort was higher in those States with more species recorded (Table 2Dáttilo et al. 2020). Even though our sampling was not aimed to exhaustively collect at every locality, we still found new records for the sampled States and one new record for Mexico. This suggests that there is still a great possibility to collect ants that will be new to Mexican States and a considerable number of new species may result from further surveys of the country. A survey performed in southern Mexico and Costa Rica demonstrated that the new species proportion can rise up to almost 80% of the records (Longino 2019).

One third of the quadrats of the country have no species records, and the overall data are highly scattered. The representative character of such a database is far from ideal for the size and the geographic diversity of the country. Therefore, we encourage the use of this information as a guidance for new surveys of ant diversity and focus on those areas with no or little records to improve the coverage of ant species distribution data. A particular effort could be taken to sample the northern Pacific coast of Mexico and the Sierra Madre Occidental (mainly all Sinaloa, southwest Chihuahua and northwest Durango States), where elements of the Nearctic, Neotropical and transition regions converge. This undersampled zone has an interesting west to east habitat gradient where agricultural landscapes, deciduous forests, pine and oak forests can be found (INEGI 2007). Another largely undersampled zone occurs at the east region of the northern border of Mexico, from Tamaulipas to Chihuahua and corresponds to the eastern zone of the Mexican Nearctic. This is a large area composed mainly of a large desert plateau with scattered patches of agricultural zones (INEGI 2007).

Despite the relevant ant distribution information that can be retrieved from surveys of the northern border of Mexico, Sinaloa and even the south regions of Guerrero and Oaxaca, entomologists might have avoided those areas because the lack of security for fieldwork. Surveys on those areas should be undertaken with extreme precautions due to continuous dangers for field biologists.

Hand collecting was the only method used to retrieve the specimens in our study and it is also the most frequent method amongst all the ant records for Mexico. The second most frequent methods are pitfall traps and leaf litter sampling (Winkler extraction) (Dáttilo et al. 2020). Hand collecting has to be standardised for quantitative surveys and is biased by the expertise of the field collector (Gotelli et al. 2011). However, records retrieved by this method have been useful to detect many (95%) of the endemic ant species (Salata et al. 2020). We encourage the use of a mixture of collecting methods, such as pitfall traps, Winkler extractors and baits for new surveys on the undersampled areas, but always using hand collecting as a method to improve the species records.

The distribution of cosmopolitan Dorymyrmex insanus could be used as an indicator for undersampled regions. According to Guénard et al. (2017), this species is present in all Mexican States, but Chiapas, Tabasco, Morelos, Colima, Guanajuato and Zacatecas. Here, we report its presence in Guanajuato, supporting the ubiquitous distribution of this ant across Mexico. Sampling efforts could be directed to those areas where this ant is not yet recorded. These ant species, considered as bioindicators, are often widely distributed and may be ecologically dominant in their respective biomes (Del Toro et al. 2012).

One of the most interesting biological phenomena of Mexico is the transition between the Nearctic and Neotropical biogeographical regions. To better understand the history of the biotic patterns this gradient has been generating, it will be interesting to focus on the comparative evolutionary history of species endemic from both areas, as well species distributed across this transition. Here, we documented that A. texana, C. scutulatus, C. crinosa, D. insanus, N. melanocephalus and S. augustae have distributions combining both biogeographic elements and could serve as useful models to study this phenomenon.

The results of the analysis of the distribution of species richness and endemicity of ants in Mexico were made using the more complete and up-to-date database of ant records in Mexico (Dáttilo et al. 2020). However, those records do not come from a standardised sampling system and the information is highly fragmented; therefore, the distribution patterns presented here can be improved. The necessary data to extend these patterns cannot be generated by a single systematic survey on ant biodiversity in the country. The integration of the generated data by surveys of different studies will be more valuable if the sampling is standardised and repeatable.

The general patterns of the preliminary estimation of endemism must be taken with caution, as the lack of information from some areas of the country might cause an overestimation of the levels of endemism in some ant species. Mexico is still a largely undersampled country for ants compared with regions, such as Florida (Ohyama et al. 2020) or Madagascar (Guénard et al. 2017), where the sampling has been more intense and systematic.

This is the first effort to describe the patterns of ants`species richness hotspots in Mexico and the endemicity patterns for a grid of 0.5° map of the country. To describe the processes that drove the distribution of the narrow endemic species of ants in the transition zone between the Nearctic and Neotropical zones in Mexico, more biogeographic studies are needed (Fisher 2009). It is necessary to generate a standardised sampling system to achieve the necessary representation of each quadrat and include sufficent ant species distribution data to explore the impact of Formicidae in the geobiotic scenarios in Mexico (Halffter and Morrone 2017).