ASU-SIBR Social Insect Bio Repository, Arizona State University, Tempe, Arizona, USA

CASC California Academy of Sciences, San Francisco, California, USA

UNAM Colleción Nacional de Insectos, Instituto de Biología, UNAM, Ciudad de México, México

UAAAN Departamento de Parasitología, Universidad Autónoma Agraria Antonio Narro, Saltillo, México

USNM National Museum of Natural History, Smithsonian Institution, Washington, DC, USA

Specimens (workers and queens) were manually collected during daylight at three localities in the Mexican states of Nuevo Leon and Coahuila (Table 1), specifically at the eastern and western edges of the northern Gran Sierra Plegada: south and west of the city of Monterrey, Nuevo Leon, and in the municipality of Saltillo, Coahuila. One nest was excavated at Lomas de Lourdes, Saltillo, Coahuila, on September 3, 2014. Approximately 200 workers, one queen, and a substantial amount of fungus garden were collected.

The geographic distribution map (Figure 1) and Table 5 provide detailed information about the localities where T. pakawa was collected. A single queen (dealate female) was collected on the soil surface at La Estanzuela, Nuevo Leon, and its measurements are also reported herein.

Figure 1. 

Geographical distribution of Trachymyrmex pakawa n. sp. and its closest relative, Trachymyrmex smithi, in the southwestern USA and northern Mexico (from Rabeling et al. 2007 and this study). Red stars: collection records of T. pakawa. Blue circles: reported collection points of T. smithi. Yellow area, Chihuahuan desert. Green area, Sierra Madre Oriental mountain range.

Morphological characters from field-collected specimens were examined with a SMZ-168TL Motic stereoscope (Motic Inc., Hong Kong) at 50x magnification and greater. Measurements of Head Length (HL), Head Width (HW), Scape Length (SL) and Mesosomal Length (ML) (all in mm) as defined by Rabeling et al. (2007) were obtained using a Mitutoyo Digimatic 500-195-30 digital caliper with a resolution of 0.01 mm under the stereoscope. Cephalic Index (CI) = HW*100/HL and Scape Index (SI) = SL*100/HW were also calculated (Rabeling et al. 2007). Means were calculated from a total of 22 individual workers.

Photographs were taken with a Leica MZ16 APO dissecting microscope (Leica Microsystems, Wetzlar, Germany) using a ProgRes 3008 digital camera mounted on the microscope. Images were taken with the PictureFrame 2.3 software package (Optronics, Goleta, CA). Sequential images were stacked to final 3D form with Helicon Focus software (Helicon Soft, Ltd., Kharkov, Ukraine). Minor corrections (color, background, etc.) were made using Adobe Photoshop CS5 (Adobe Systems, San Jose, CA USA).

Initially (in 2010) total genomic DNA was extracted from 1 to 3 worker ants of T. pakawa and T. smithi following the lithium chloride method (Huang et al. 2000), and cytochrome oxidase subunit I (COI) was amplified for initial bioinformatic analysis (below). Subsequently, for both COI and Elongation Factor 1-α (EF1-α-F1) analysis, total DNA was extracted from single worker ants of both species (n = 3 each) using the Pure Link^TM Genomic DNA Mini Kit (Invitrogen Carlsbad, CA USA. Identical COI sequences were obtained with both methods (see below). Specimens used for analysis were collected at the localities indicated in Table 1.

PCR amplification of the COI gene fragment (marker) was carried out (Jacobson et al. 2006) in a volume of 25 ml consisting of 2.5 ml of 10× PCR buffer, 0.75 ml of 50 mM MgCl₂, 1.25 ml of 2.5 mM dNTPs, 1.0 ml of genomic ant DNA, 1.3 ml of 13.6 mM CI-J2195 primer, 1.0 ml of 15.2 mM Jerry Garcia-CI primer (Jacobson et al. 2006), 0.3 ml of TaqDNA polymerase, 5U/ml (all reagents from Invitrogen, Carlsbad CA USA), and 16.9 ml of MilliQ water. PCR amplifications were performed at 95 °C (1 min), 35 cycles of 94 °C (1 min each), 43 °C (1 min), 70 °C (2 min), and 70 °C (2 min). PCR amplification of the nuclear gene fragment of the first intron of the EF1-α-F1 was carried out using the primer pairs U52.1 and L53 as described by Rabeling et al. (2007), in a volume of 25 ml consisting of 2.5 ml of 10X PCR buffer, 0.75 ml of 50 mM MgCl₂, 0.25 ml of 20 mM dNTP’s, 1.0 ml of genomic ant DNA, 0.5 mM of U52.1 and L53 respectively, 0.3 ml of Taq DNA polymerase (5U/ml) (all reagents from Invitrogen, Carlsbad CA USA), and 19.2 ml of MilliQ water. PCR amplifications were performed at 94 °C for 2 min, 35 cycles of denaturation at 94 °C for 30 sec, annealing at 51 °C for 45 sec, extension at 72 °C for 90 sec, and a final extension of 72 °C for 10 min. Amplification products were visualized on 1.2% agarose gel stained with ethidium bromide.

Nucleotide sequencing and purification of PCR fragments were performed at Macrogen Corporation USA (Rockville, MD) using CI-J2195, Jerry Garcia-CI, U52.1, and L53 primers for sequencing.

The raw sequences were edited in CodonCode aligner (CodonCode Corporation). We included in our dataset previously published sequences; see Table 1 for a complete list of accession numbers of sequences used and the novel ones obtained in this study. Sequences were aligned using CLUSTALW implemented in Mega 6.0 (Tamura et al. 2013). Our final concatenated alignment comprises the partial sequences of the COI marker (313 bp) and EF1-α (715 bp) marker for a total of 1028 aligned sites and 131 parsimony-informative sites.

We submitted four data blocks to PartitionFinder v1.1.1 (Lanfear et al. 2012) as follows: first, second, and third positions of COI and the intron region of EF1-a-F1. Based on the PartitionFinder analysis under the Bayesian information criterion (BIC) and the “all” search algorithm, our dataset was divided into two partitions: 1) First and second positions of COI and EF1-α and 2) third positions of COI. For the first partition the best model was General Time Reversible with invariant sites and a gamma distribution (GTR+I+G) and for the second partition the best model was GTR+G. We conducted a Bayesian phylogenetic analysis employing MrBayes v3.2.5 (Ronquist et al. 2012), with nucmodel= 4by4, nruns = 2, nchains = 4, and sampled freq = 100. Nodes that had posterior probabilities greater than 0.95, were considered well supported. Additionally, a parsimony method to produce a median joining network was used to visualize genetic distance and the geographic associations among T. pakawa and T. smithi haplotypes using PopArt v.4.6 (http://popart.otago.ac.nz).

Divergence dates between T. pakawa and T. smithi were estimated using the mitochondrial data with the Bayesian program BEAST v1.4.8 (Drummond and Rambaut, 2007) following the procedure implemented by Seal et al. (2015) for timing the evolutionary events in T. septentrionalis. We applied a lognormal relaxed clock, a Yule process tree prior, a UPGMA starting tree, and a TN93 substitution model (Topaliv2.5). The clock model was estimated from a normal distribution with a mean of 0.075 substitutions/sites/lineages/myr. Three independent experiments were performed for 20 million generations and sampled every 2000 generations, with 10% of the trees discarded (burn-in). The output file was evaluated with TRACER v1.4.8 and the effective sample sizes exceeded 200 (Drummond and Rambaut 2007).

The evolutionary distances were computed using the p-distance method and represent the proportion of nucleotide changes calculated in MEGA v6 (Tamura et al. 2013). Standard errors of mean p-distances among taxa were calculated using 1000 bootstrap replicates.

We identified two T. pakawa haplotypes for each marker (COI and the EF1-α-F1) (Figure 4). A phylogenetic tree was constructed using Bayesian analyses (Figure 4A) of previously published Trachymyrmex sequences and sequences of T. smithi and T. pakawa obtained for this work (Table 1). The T. smithi specimens from Ojinaga, Chihuahua, Mexico, yielded COI and EF1-α-F1 sequences 100% and 99.99% identical to T. smithi Genbank accessions EF539730 and EF539779.1, respectively, from New Mexico, USA; these (New Mexico) and additional T. smithi Genbank sequences for both markers were used in our molecular analysis (Table 1). Our phylogenetic tree (fig. 9A) is consistent with the unrooted tree of Rabeling et al. (2007). From the analyses we inferred that T. pakawa is a member of the septentrionalis species group, and that T. pakawa and T. smithi are sister species. The T. pakawa -T. smithi clade is most closely related to the clade consisting of T. septentrionalis and T. pomonae. These four species are the sister group of a clade consisting of T. arizonensis, T. carinatus, T. desertorum, and T. nogalensis (Figure 4); the monophyletic group including these eight species belongs to the septentrionalis group (Rabeling et al. 2007). The remaining two species, T. jamaiciensis and T. turrifex, are more distantly related to the members of the septentrionalis group and belong to the jamaicensis and urichii groups.

Haplotype network maps indicate the genetic distance between T. smithi and T. pakawa (Figure 4B). For COI, there is a difference of > 40 base pairs (bp) between the two species. For the more slowly evolving nuclear EF1-α marker the nucleotide difference between the two species is 4 bp.

In our pooled analysis of North American Trachymyrmex taxa (Rabeling et al. 2007), the average genetic distance between pairs of species for the COI marker without T. pakawa is 0.139 (SE = 0.012), whereas the average genetic distance for COI between T. pakawa and the remaining taxa is comparable: 0.132 (SE = 0.014). The distance between T. pakawa and T. smithi (0.12) is larger than the distance between T. smithi–T. arizonensis (0.11) and T. smithi–T. carinatus (0.10). These data support the hypothesis that T. pakawa is genetically distinct and reproductively isolated from other Trachymyrmex species.

For the EF1-α marker, the same patterns are present (although overall distances across all ant taxa considered are much smaller), with the distance between T. pakawa and T. smithi (0.002) larger than the distance between T. smithi-T. pomonae (0.000) (Table 4). But besides the genetic distances, unique insertions and deletions for T. pakawa (not shared with the other species) were observed in EF1-a.

The pooled analysis for the concatenated COI-EF1-a sequences indicated that the average genetic distance between pairs of Trachymyrmex species without T. pakawa is 0.082 (SE = 0.006), whereas the average distance between T. pakawa and the remaining taxa is comparable: 0.066 (SE = 0.005). Also the distance between T. pakawa and T. smithi (0.04) is similar to the distance between T. smithi and other species (0.046–0.054), thus supporting the new species status of T. pakawa (Table 4).

In general, the larger genetic distances among these North American Trachymyrmex species and clades are always between [T. turrifex or T. jamaiciensis] and the remaining taxa including T. pakawa.

Both phylogenetic and network analyses showed the same relationships among T. smithi and T. pakawa. No shared haplotypes were found between T. smithi and T. pakawa (Figure 4B). For COI, there is a difference of 34 base pairs (bp), whereas for EF1-α the difference is 3 bp (Figure 4B). Considering the small number of specimens analyzed (five for COI and three for EF1-a) there is possibly significant genetic variation within T. pakawa, since two haplotypes were found for each marker (Figure 4B). Comprehensive analyses are required to elucidate the population structure of T. pakawa.

Bayesian coalescent analyses indicate a divergence time of about 8.00 million years before present (95% confidence interval: 4.8–11.5 mya) between T. pakawa and T. smithi. The divergence of the lineages of T. pakawa and T. smithi could have been driven by the Pliocene-Holocene desertification of southwestern North America.

To include T. pakawa we modified the taxonomic identification key to North American Trachymyrmex workers (Rabeling et al. 2007) as follows:

Specimens of Trachymyrmex pakawa were collected in the Mexican states of Nuevo Leon and Coahuila. The known species range extends between the opposite edges (west and east) of the Gran Sierra Plegada section of the northern Sierra Madre Oriental mountain range. The more geographically distant populations observed and sampled at the edges of this mountain range, are separated by about 80 km on a straight line (see coordinates, La Estanzuela and Cerro de las Letras). In between these known extreme distribution points, several populations have been detected (Table 5).

This species inhabits very rocky soils, on moderate to very steep slopes, usually between and under large limestone boulders and rocks with a very thin (10–20 cm) cover of lithosol (limestone derived). This seems the main common aspect to most habitats where T. pakawa has been found. The sites are rather diverse montane habitats of the southern Nearctic: gallery forests, oak and oak-pine forests, and xerophilous Chihuahuan scrub on slopes (“submontane scrub”) (Table 4).

With the possible exception of the Lomas de Lourdes (Saltillo) population, all these localities are within stands of lechuguilla (Agave lechuguilla), or within 300 m or less from rocky outcrops where this plant is present. Even the gallery forest at La Estanzuela is within a few hundred meters from lechuguilla stands. This underscores the xerophilous nature of T. pakawa.

Coahuila state

1. We collected T. pakawa in August of 2012–2015, above the Lomas de Lourdes area of Saltillo, on the slopes of the Sierra de Zapalinamé, a small isolated mountain range separated from the main Sierra Madre range by narrow, arid valleys. 25.35998°N, -100.98060°W. Altitude is 1750 m. The ant was also collected at 25.365181°N, -100.983217°W, at an altitude of 1700. Vegetation at these sites is a xerophilous forest of small oaks (4–5 m tall) (Quercus laeta and the endemic Q. saltillensis), mountain mahogany (Cercocarpus spp.), antelope bush (Purshia plicata), weeping juniper (Juniperus flaccida), and madrone (Arbutus xalapensis). The canopy is rather dense; the ants were observed mainly on exposed slopes. Annual precipitation here is 400 mm, with abundant fog spells. On this western (rain shadow) side of the Sierra Madre Oriental, the soil is granular, and usually covered with a thick, but usually dry, litter layer of oak leaves. The ants live in the more mesic creeks and microhabitats of the range (NW slope) in an area with a few very small, intermittent springs, indicating a shallow water table. As elsewhere for T. pakawa, this population is not abundant; only five colonies and one nest entrance have been observed despite active searching over several years. The ants nest in very shallow lithosol covering a rocky layer, or among large limestone boulders (outcrops); the nests are possibly at least one meter deep in order to reach moisture pockets in spaces underneath the boulder layer. The tridimensional structure of aquifers is complex on these slopes, and nesting location might be influenced by the water table.

2. Cañon de San Lorenzo, Sierra de Zapalinamé. 3 August 2013. 25.330996°N, -100.986779°W; altitude is 1900 m. This site is at about 5 km (south) from the previous location. Foraging workers were collected on a sun-exposed limestone peak (western slope) covered with xerophilous scrub (A. lechuguilla), sotol (Dasylirion sp.), and Yucca filifera; above and at short distance from a mesic creek with Arizona cypress, (Cupressus arizonica). The foragers were walking among boulders. The site is a generally arid landscape with very small, scattered springs.

3. Reforestation edge, Universidad Autónoma Agraria Antonio Narro (UAAAN) fields, located at 5 km from the Cañon de San Lorenzo population. 10 October 2013. 25.351378°N, -101.041464°W. Altitude is 1650 m. The habitat is Chihuahuan desert; the observed nest was under low scrub of Acacia greggii (uña de gato, cat´s claw) at the edge of an area reforested with exotic pines (Pinus halepensis).

4. Cerro de las Letras, west of UAAAN campus. 10 September 2012. 25.35885°N, -101.04916°W; altitude is 1750 m. This site is on the steep eastern slope of a hill; the vegetation consists of Chihuahuan desert scrub with sotol, cacti (Opuntia spp.), Tecoma stans (yellow bells) and Yucca sp. The site is right above the creosote bush (Larrea tridentata) life zone. At this location (and probably others), T. pakawa is sympatric with the cryptic, highly xerophilous fungus-growing ant, Cyphomyrmex wheeleri (SRSP unpublished observations). One nest entrance of Trachymyrmex pakawa was also observed here, on a very small flat area on rocky outcrop; it had a small pile (a 4 cm fan) of excavated, lighter soil particles on one side of the entrance (Figure 3).

Nuevo Leon state

5. La Estanzuela creek (a state park of Nuevo Leon), on the southern edge of the city of Monterrey. 8 August 2009. 25.54042°N,-100.27286°W, 600 m altitude. This location is along an intermittent stream that flows down from the Sierra Madre into the Rio La Silla, which is part of the Rio Grande Basin. This site is on the eastern slope (not rain shadow side) of the Sierra. The locality is a warm-temperate gallery forest of sycamore (Platanus mexicana), Monterrey oak, Quercus polymorpha, and other Quercus species. Trachymyrmex pakawa forages at a few meters from the pebbly riverbed, on loam-clay reddish soil of variable depth with interspersed large rocks. The surrounding habitat within a short distance (sometimes 200 m) from the stream is clearly xerophilus: on sun-exposed boulders, there is a thorny leguminous scrub with Agave lechuguilla, Acacia spp., Caesalpinia mexicana, Cordia boissieri, and Cactaceae. Annual precipitation is 500–700 mm; annual temperatures range between 5–40° C. We collected T. pakawa here repeatedly from 2007–2013.

6. Upland and up the stream at La Estanzuela, at the previous location. 15 August 2012. 25.536932°N, -100.276289°W; 700 m altitude. This site is an ecotone of oak forest (Quercus rysophylla, Quercus polymorpha)-pine forest (Pinus pseudostrobus), over a limestone cliff. A very shallow, reddish litosol between boulders covers the bedrock. A nest entrance was observed there on deeper soil on a slope. The entrance was an inconspicuous hole (2–3 mm diameter) on a flat, sloped area on loamy soil. No accumulations of soil or detritus were observed at this nest entrance, probably due to the slope and rain.

7. Pico Apache (a peak at Cerro de las Mitras mountain) Monterrey, Nuevo León. 24 July 2008. 25.704834°N, -100.397027°W; 950 m altitude. This site is on the NE slope of the mountain, in Chihuahuan desert scrub/”submontane” scrub, with presence of Agave lechuguilla and sotol (Dasylirion sp.) scrub among boulders.

8. Pico Apache, Cerro de las Mitras, Monterrey, Nuevo León. 24 July 2008. 25.704166°N, -100.400833°W; 1300 m altitude. The vegetation at this site is scrub with sotol, Agave bracteosa (squid agave), A. lechuguilla, Opuntia stricta, Calia secundiflora (mescalbean or frijolillo), and short oaks, up to 2.5 m tall (Quercus sp.). Both locations at Cerro de las Mitras are moister than the surrounding plains.

Unlike T. smithi, its closest relative, T. pakawa has very inconspicuous nest entrances that are rather hard to find (Figure 3). There is no mound, crater, or turret as in other attines like Trachymyrmex turrifex or Mycetosoritis hartmanni (Wheeler 1907). Trachymyrmex pakawa workers usually forage and arrive at the nest entrance singly; loose trails of four and six ants were observed late in the afternoon on only two occasions. At any time only about four ants can be seen in the general vicinity of the nest. Foragers are very shy and, when disturbed, they retreat inside the nest or feign death and drop to the ground if disturbed at some distance from the nest. This makes them difficult to detect in the field because they are easily scared. Apparently only a very small fraction of the nest´s population leaves the nest at any time to forage, at least during the day. It is not known if nocturnal foraging is significant. Regarding worker numbers in colonies, at least 250 workers emerged from the partially excavated nest by the spring at Lomas de Lourdes. The observed nest chamber was at about 50 cm deep in the spaces between rocks in unusually moist soil. There appeared to be only one queen in this nest.

The reddish detritus accumulations (exhausted fungal substrate) typical of several higher attines (Weber 1972, Sanchez-Peña 2005) are piled at 10–60 cm from the entrance hole and are conspicuous when present. The detritus accumulations of T. pakawa are flat, about 6 cm across. On slopes the detritus does not accumulate.

Trachymyrmex pakawa is a rather cryptic ant. At all locations it is never abundant and foraging workers seem to avoid open flat spaces and clearings, instead walking inconspicuously between vegetation and rocks. At La Estanzuela, Nuevo Leon, workers were noticeably more common at highest elevations (650 m), in the oak-pine forest, at the most mesic conditions observed for this ant.

Soil type (or lack of) might be a major, common factor regarding the distribution of T. pakawa. Excepting the lower altitude site at La Estanzuela, rocky, very shallow soils prevail at all locations, with large buried rocks, and always on slopes. Precipitation varies from 300 (UAAAN) to 600 mm (Estanzuela). However, drizzle and fog are frequent at Lourdes and UAAAN, and contribute an unmeasured amount of effective moisture.

This ant nests in the more mesic microhabitats (permanent and intermittent springs, creeks, or under and between large rocky outcrops) in a generally arid mountain range. The extant populations could be relicts of a more widespread distribution that contracted due to the post-glacial desertification process reducing the extension of woodlands and originating the Chihuahuan desert about 8000 years ago (Elias et al. 1995).

Although it is spread over an area covering a few hundred square kilometers, T. pakawa appears to have a discontinuous distribution in its range, and it is not abundant or common.

Trachymyrmex pakawa is not found in disturbed habitats. No population seems to have colonized agricultural areas (irrigated or dry land), or urban/landscaped areas in the neighboring cities (Monterrey and Saltillo) unlike other Trachymyrmex species. In this respect, this species differs markedly from geographically close North American species of Trachymyrmex that colonize (sometimes abundantly) gardens, yards, and landscaped areas in towns and cities, like the following: T. smithi, in northern Chihuahua (Ojinaga) and adjacent Presidio, Texas (SRSP unpublished observations); T. septentrionalis, in Austin, Texas (SRSP unpublished observations); and T. turrifex, in gardens in Pesqueria, Nuevo Leon, and in Matamoros, Tamaulipas, Mexico (Lower Rio Grande Valley); in this last case at more than 10 nest entrances/m² (Sanchez-Peña 2010; see Rabeling et al. 2007). These apparently anthropo-tolerant Trachymyrmex populations live in deep, sandy, light-colored soils, unlike the thin, rocky, shallow limestone-based lithosols and steep alluvial soils near Monterrey and Saltillo colonized by T. pakawa. This spatial separation from humans might result from the strong observed preference of T. pakawa for rocky slopes and its avoidance of flat areas with deep soils.

Some of the collected specimens at La Estanzuela were conspicuously lighter than the ants from Saltillo and other sites; these are dark brown and initially they were confused with the even darker-colored T. smithi. Subsequent collections at La Estanzuela were more uniformly dark. It is possible that the lighter workers had emerged recently (callow workers). However, presumably callow workers from an excavated nest at Saltillo, covered with actinomycete growth (Cole 1952, Rabeling et al. 2007) possessed uniformly dark cuticle under the actinomycete layer. Recently emerged (callow) workers of Trachymyrmex spp. are commonly covered with growth of (potentially symbiotic) actinomycetes that give them a “sugary” appearance (Rabeling et al. 2007). The taxonomic relevance of the color differences described above is unknown.

Additional differences between T. smithi and T. pakawa include: in T. smithi, nests are often readily visible on flat soils in clearings, with conspicuous detritus piles nearby; nests can have a few hundred ants/nest; T. smithi is also somewhat aggressive and, when the nest or workers are disturbed, workers display aggressive behavior, and sometimes exit the nest resolutely, in small numbers, in an aggressive way (SRSP unpublished observations) unlike the very timid T. pakawa.

The discovery of T. pakawa indicates that the temperate Madrean ecoregion of the Sierra Madre Oriental might harbor additional interesting, undescribed arthropod taxa. This underscores the relevance of analysis and urgent conservation plans for these regions and biomes.