Species delimitation in the Stenocereus griseus (Cactaceae) species complex reveals a new species, S. huastecorum

The Stenocereus griseus species complex (SGSC) has long been considered taxonomically challenging because the number of taxa belonging to the complex and their geographical boundaries remain poorly understood. Bayesian clustering and genetic distance-based methods were used based on nine microsatellite loci in 377 individuals of three main putative species of the complex. The resulting genetic clusters were assessed for ecological niche divergence and areolar morphology, particularly spination patterns. We based our species boundaries on concordance between genetic, ecological, and morphological data, and were able to resolve four species, three of them corresponding to S. pruinosus from central Mexico, S. laevigatus from southern Mexico, and S. griseus from northern South America. A fourth species, previously considered to be S. griseus and commonly misidentified as S. pruinosus in northern Mexico showed significant genetic, ecological, and morphological differentiation suggesting that it should be considered a new species, S. huastecorum, which we describe here. We show that population genetic analyses, ecological niche modeling, and morphological studies are complementary approaches for delimiting species in taxonomically challenging plant groups such as the SGSC.


Introduction
Morphological characteristics of many cactus species are highly prone to convergent and parallel evolution, as well as losses and reversals [1]. Consequently, there are few synapomorphies for supporting phylogenetic relationships among taxa [2,3]. Molecular data are also limited toimprove the understanding ofevolutionary relations among species because of the limited availability of nuclear markers for this group [4],low plastid sequence divergence [5]  relatively recent origin and diversification of the Cactaceae [6], and high rates of homoplasy due to incomplete lineage sorting [7]. Therefore, interspecific and lower taxonomic group divergence times are expected to be relatively short as in the Pilosocereus aurisetus complex, a relatively young group of species 0.33-0.87 Myr old [8], and in the genus Harrisia which includes species approximately 0.20-0.37 Myr old [9]. Frequently, the taxonomic clusters within these closely related groups of species lack morphological and genetic characters for clearly defining their limits [10]. The Stenocereus griseus species complex (SGSC) has long been considered a taxonomically puzzling group of taxa. High species similarity in the complex is evident based on stem and flower morphology, triterpene composition, and other characters [11]. Gibson [12] considered a broad distribution for the complex: S. griseus from northern Mexico to coastal Venezuela, S. deficiens in coastal Venezuela, S. pruinosus in southern Mexico, S. longispinus in southern Mexico, S. laevigatus distributed in southernmost Mexico and northern Guatemala, and S. hystrix in the Greater Antilles. Gibson [12] warned about the extensive overlap of morphological features and geographic distribution of these taxa and pointed out a possible anthropogenic influence on their distribution. In fact, Bravo-Hollis [13] suggested that S. griseus populations from northern Mexico could have been introduced from Venezuela by humans.
The first comprehensive taxonomic review of the SGSC was accomplished by Arreola-Nava [14] as part of a survey of the whole genus Stenocereus in which extensive synonymy was found: S. deficiens and S. longispinus were considered synonyms of S. griseus and S. laevigatus, respectively, and Antillean S. hystrix or S. fimbriatus were considered as illegitimate names for S. peruvianus. Conversely, S. pruinosus has remained unchanged. Therefore, four species are currently considered members of the SGSC (Table 1), and it can be included into the larger S. griseus group that also includes S. fricii and S. chacalapensis., Phylogenetic analysis based on the plastid rpl16 intron and the whole trnL-trnF region, along with morphological characters including rib number and stem color [12,13], did not resolve the species in this group [15].
The lack of agreement may be due to the intricate distributional pattern of the complex: herbarium specimens records describe assorted accessions of S. griseus and S. pruinosus in northern and southern Mexico (Chiapas and Yucatán) where a third species, S. laevigatus ( Fig  1) is well represented. Records of co-occurrence can be explained either by misidentification or sympatry.
Other than these taxonomic issues, introgression may be occurring among members of the complex. Parra et al. [16], reported measurable gene flow between S. laevigatus and S. griseus in the southern and northern limits of the range of S. pruinosus. If gene flow is occurring within the SGSC complex, understanding genetic barriers between species will likely shed light on species boundaries throughout their ranges. The distribution of the SGSC includes the Chihuahua and Puebla-Oaxaca diversity centers of cacti [17] interrupted by the Trans-Mexican Volcanic Belt (TMVB), a biogeographic barrier dating back to the Middle Miocene (19.5 to 16 Myr ago) [18]. The Puebla-Oaxaca center includes the Tehuacan Valley region, an early quaternary (2 Myr) sedimentary basin [19] where 64% of Mexican cactus species occur [20] and the Oaxaca Central Valleys. Both regions are isolated by highlands of the Sierra Madre del Sur including the Mixteca Alta and northern and southern ranges in Oaxaca [21], where most SGSC records have originated, mainly those of S. pruinosus (Fig 1). Tothe east of the Sierra Madre del Sur, the distribution of SGSC continues through the Pacific Coast and the Isthmus of Tehuantepec, a well-known biotic barrier [22] for temperate organisms [23]. For the Cactaceae and specifically the tribe Cacteae, this area is part of a larger Central America [24] or Mexican Pacific Coast [25] region The estimated divergence times for plant taxa across the Isthmus of Tehuantepec ranges from 25 to 4 Myr [26 and references therein], the most recent corresponding to the splitting of Rhipsalis baccifera populations [23]. In southern Mexico and Central America, divergence times are associated with the Yucatán Peninsula Karst (3.6 Myr to 18,000 years old) [26], the Central Depression of Chiapas, a 3 Myr old valley formed by the uplift of Chiapas volcanic ranges [27], and the Motagua-Polochic canyons, an active fault at least 15 Myr old [28]. S. peruvianus is restricted to the Greater Antilles that are 35 Myr old [29]. Finally, a distributional gap is found through the Lesser Antilles, except for the Leeward Islands [30]. The southernmost records of the genus Stenocereus are found in Colombia and Venezuela [14], specifically in the Caribbean Coast of North Colombia and Venezuela and the Inter-Andean Valleys [31][32][33].
The aim of this study is to delimit the SGSC by implementing population genetics clustering methods (Bayesian clustering and distance-based methods) and testing the congruency of these groupings with ecological and morphological analyses. We tested the hypothesis that S. griseus is a homonym comprising two different taxa, one from Mexico and the other from northern South America. We complemented the species delimitation with spatial references of taxa of the SGSC and its genetic barriers, and the description of the new species S. huastecorum.

Ethics statement
The permit for collecting plant material in Mexico for studies was provided by national or federal authorities of the Mexican Ministry of Environment and Natural Resources (SEMAR-NAT) and the National Commission for the Natural Protected Areas (CONANP); in Colombia collection was made under permission of the Ministry of Environment and Sustainable Development (MINAMBIENTE). In addition, we obtained permission from the local authorities and communitarian assemblies of the villages whose territories contained the Stenocereus populations we studied. None of the studied taxa are specially protected or endangered species.

Sampling and study sites
The distribution of the SGSC was compiled from records of biodiversity databases including Tropicos, GBIF, REMIB, and CONABIO, together with information retrieved from herbarium specimens from CHAP, CHAPA, COL, ENCB, IEB, MEXU, UTMC, and XAL (codes following Index Herbariorum [34]). We collected 10-15 cm rib strips from 377 individuals over 35 Mexican populations and four from Colombia (Fig 1); samples were preserved in silica gel for transporting to the laboratory where the tissue was frozen, and then lyophilized in a Christ Alpha 2-4 LD Freeze Dryer (Martin Christ Freeze Dryers, Osterode, Germany).

Molecular methods
DNA was isolated from either frozen or lyophilized chlorenchyma using the CTAB-based DNA isolation procedure [35]. We tested 15 microsatellite loci previously developed for Polaskia chichipe [36,37], Stenocereus stellatus [38], and Stenocereus gummosus [39]. PCRs with loci suitable for genotyping (see genotyping and markers suitability in the Results section) were carried out in a MultiGene OptiMax (Labnet International, Inc., Edison, NJ, USA) or in a 2700 thermal cycler (Applied Biosystems, Foster City, CA, USA). We pooled three primers from P. chichipe and one from S. stellatus in one multiplex reaction, as well as four primers from S. gummosus in another one. In addition, we separately amplified primer JCS73 isolated from S. stellatus (see Table A in S1 Appendix for primers and multiplex reactions details). Every reaction was driven to a 5 μL final volume containing 2.5 μL Platinum Multiplex PCR Master Mix (Applied Biosystems, Foster City, CA, USA), 2 μL PCR grade H 2 O, 0.5 μL DNA template (50-200 ng/μL), and a negligible volume of primer mix reaching 70 nM. JCS73 required replacing 0.5 μL of H 2 O by the same G/C enhancer volume in order to assure amplification success. Both multiplex reactions required an annealing temperature of 56˚C, while we used 50˚C for JCS73; 40 cycles were used in every PCR reaction. Additional cycling conditions were implemented following manufacturer directions.

Genotyping and marker suitability
Capillary electrophoresis was performed in a 3130xl Genetic Analyzer (Applied Biosystems, Foster City, CA, USA). Genotyping was achieved by using the Peakscanner software v1.0 (Applied Biosystems, Foster City, CA, USA), while scoring errors and null alleles were searched through MICRO-CHECKER [40], and linkage disequilibrium tests were performed by using Genepop [41]. Once the suitable marker set was determined, we performed two types of grouping analysis: Bayesian clustering and genetic distance based methods.

Bayesian clustering
Three Bayesian methods were used: STRUCTURE [42], which performed with one million Markov chain Monte Carlo discarding the first 100 000 as burn-in and testing up to 15 groups,10 iterations each. The group number (K) was determined via the Evanno method [43] through the online service STRUCTURE HARVESTER [44]. The Geneland R Package [45] was performed using 100,000 iterations with a 10,000 thinning, testing 20 groups with 10 repetitions each. We considered as group number the highest likely value of the resulting distribution. TESS [46] was run testing from 2 to 20 Kmax, 100 000 sweeps were performed discarding the first 10%; then the top 20% DIC values runs were filtered and averaged by every Kmax in order to use the criterion recommended by François and Durand [47] for choosing the most likely number of groups. Both STRUCTURE and TESS most likely group number (K and Kmax, respectively) iterations were analyzed with CLUMPP [48] in order to summarize individual assignment values.

Distance-based approaches
A Nei's standard genetic distance [49] matrix was calculated per population with MSA software [50] bootstrapping it 1000 times. Then, the Barrier [51] program was run for testing three barriers (considering four taxa). Finally, a UPGMA phenogram [52] was built in MEGA 6 [53] with the same distance matrix used for Barrier.

Geostatistics and Bayesian consensus
Kriging interpolation was performed from each Bayesian clustering individual Q-matrix with the ArcMap 10.1 (Redlands, CA, USA) Geostatistical Analyst extension. We defined the genetic groups as high probability areas (! 60% of belonging probability) and their equivalence when these comprised the same populations across the three clustering methods (Fig 2). We summed the kriging raster files of equivalent genetic groups (EGGs) in order to obtain a consensus value for each pixel. Finally, each EGG consensus raster was standardized and we projected the top 75% belonging probability as areas (Bayesian consensus, hereafter "BC"), each named after the most common species records included. Distance-based approaches were superimposed over the BC map (Fig 3) in order to spot biogeographic barriers, thus summarizing every genetic method employed.

Ecological niche modelling (ENM) and comparisons
The SGSC records were obtained by merging the herbarium records with our own collections, and these data were curated by removing points whose values were clearly rounded up or with incorrect geo-references. ENM was performed by Maxent 3.3.3 [54] using 313 records and the 19 bioclimatic variables from Worldclim [55] at 30 arc-second resolution. The variables BIO2 (Mean Diurnal Range), BIO3 (Isothermality), BIO11 (Mean Temperature of Coldest Quarter), BIO13 (Precipitation of Wettest Month), BIO15 (Precipitation Seasonality), and BIO19(Precipitation of Coldest Quarter) that contributed to 95% of the model and showed low correlations were selected to perform individual (species) niche modeling defined by the occurrences intersecting each BC polygon, which contained 59 records for S. griseus, 111 for S. huastecorum, 78 for S. laevigatus, and 65 for S. pruinosus. For each model, 10,000 random points were used in order to extract bioclimatic data for the environmental background.
In order to evaluate differences among species ecological niches, two methods were used: Contact zone analysis [56] and the multivariate method proposed by McCormack et al. [57]. Comparisons were carried out between spatially contiguous BC polygons (S. pruinosus vs. S. laevigatus, and S. laevigatus vs. S. griseus), those suspected to be sympatric (S. huastecorum vs. S. pruinosus) and homonyms (S. huastecorum vs. S. griseus).

Morphometric analysis
Voucher specimens at MEXU and samples from our collections that contained vegetative portions of stems (subapical areolas may contain additional spines derived from the flower and mature areolas often loss upper peripheral spines) were considered to measure areolar Kriging interpolation of individual Q-matrix by each clustering method. The first three maps columns (left to right) depict the interpolated assignment probability for each method (labeled in the heading) and K is the number of groups detected by each. Groups containing the same populations across methods (EGGs) are placed alongside and keep the same color hue, whereas the color gradient saturation represents higher probability: red = S. griseus-Mexico (here designated S. huastecorum); green = S. pruinosus and its subgroups (green shades in Geneland maps); blue = S. laevigatus and its subgroups (blue shades in Geneland maps), and dark gray = S. griseus. White bullets represent populations. EGGs populations are represented at the rightmost column by bullets which follow the color code before described.

Nomenclature
The electronic version of this article in Portable Document Format (PDF) in a work with an ISSN or ISBN will represent a published work according to the International Code of Nomenclature for algae, fungi, and plants, and hence the new names contained in the electronic publication of a PLOS article are effectively published under that Code from the electronic edition alone, so there is no longer any need to provide printed copies.
In addition, new names contained in this work have been submitted to IPNI, from where they will be made available to the Global Names Index. The IPNI LSIDs can be resolved and the associated information viewed through any standard web browser by appending the LSID contained in this publication to the prefix http://ipni.org/. The online version of this work is archived and available from the digital repositoriesPubMed Central and LOCKSS.

Genotyping and marker suitability
Thirteen out of 15 SSRs markers yielded positive PCR products as revealed by 2% agarose gel electrophoresis. However, we discarded four of these loci because they exhibited non-allelic patterns (JCS1 and JCS51), lack of polymorphism (Sgum12), and extensive occurrence of null alleles (Sgum25). After evaluating linkage disequilibrium, we used the final set of loci including Pchi 20, 50, and 54, JCS 49 and 73, as well as Sgum 06, 29, 36, and 39 (for details see S1 Appendix).

Bayesian clustering
STRUCTURE and TESS detected four EGGs including the same populations ( Figures A and B in S2 Appendix), which were designated as S. huastecorum, S. pruinosus, S. laevigatus in northern, central, and southern Mexico, respectively, and S. griseus from northern Colombia. Geneland, however, showed six groups (Fig C in S2 Appendix), two of them were exactly equivalent to S. huastecorum and S. griseus, respectively; whereas two groups (Tehuacán and southern Oaxaca) are contained into S. pruinosus (Fig 2), finally, S. laevigatus splits into Cintalapa populations and the remaining populations of this species (Fig 2).

Distance-based approaches
The Barrier tests revealed three observable geographic barriers: a central/northern barrier supported by high bootstrap values (85-90%), the Isthmus of Tehuantepec (79%), and the third separated the Colombian populations from Mexican taxa, with a 68% support (the purple bars in Fig 3). The UPGMA phenogram (Fig 3) showed four branches matching the STRUCTURE/ TESS clustering; the greatest genetic divergence corresponds to the Colombian S. griseus (Nei'

Ecological niche modeling comparisons
The Contact Zone Analysis (CZA) [56], revealed that every pairwise comparison showed clear suitability differences; however, it failed to distinguish the contact zones of S. griseus with S. laevigatus and S. huastecorum (Fig 5E-5H). Conversely, the McCormack et al. [57] method showed general differentiation in principal components 1, 2, and 4, which together explained 74.87% of the variance, and partially for PC 3, which failed in the comparisons by the CZA. PC 5 showed no differentiation between S. huastecorum vs. S. pruinosus and S. huastecorum vs. S. griseus (Table 2).

Morphometric analysis
Areolar width and length showed no differences between the taxa analyzed. The lower central and radial c-homologous spines length failed to differentiate S. huastecorum from S. griseus and S. pruinosus from S. laevigatus. Both radial and central spines counting displayed full dissimilarity between species (t = 1.989, df = 84, P<0.005 and t = 1.988, df = 85, P<0.05). Since S. griseus lacks upper left and right central spines, these comparisons could be performed only for the Mexican taxa, and among these only S. huastecorum differed from both S. pruinosus and S. laevigatus (t = 2.010, df = 49, P<0.05 and t = 2.011, df = 48, P<0.05) (Fig 6).  Additional observed specimens: Table A in S3 Appendix. Candelabraform tree, up to 9 m tall; trunk many times absent, <30 cm tall and <20 cm width; many branches, ascending, about 15 cm width, unconstrained, grayish green to glaucous; first order branching; mucilage cavities not evident in transverse branch sections; ribs 7 to 9, acute in transverse section, straight in longitudinal section, 12-35 mm tall by 11-35 mm wide at the base, without horizontal constriction between areolas within the same rib; areolas 8-28 mm apart each, round to obovate (scutelliform), 6-11.5 mm long and 6-10.6 mm wide, with light-colored trichomes; radial spines 5-7, subulated, divergent, 8-30 mm long, white when young, grayish when mature; 1 central spine, rarely 3, subulated, robust, 9.7-56 mm long, white with reddish base when young, grayish at maturity; subapical flowers, night anthesis remaining opened until the next morning, infundibuliform, 5.6-6 cm long and 4.4-5 cm wide in anthesis; pericarpel cylindrical, green, 11.5-14.5 mm wide, covered with slightly prominent podaria, separated, with wide oblong scales about 1.4-2.3 mm long by 1.5-3.2 mm wide at the base, reddish, areolas presenting light-colored trichomes, spines 7-13 mm long; receptacular tube 30-54 mm long, podaria with decurrent scales, narrow oblong to spatulate, apex obtuse to mucronated, about 4-8 mm wide with few trichomes and spines; outer perianth segments narrow oblong, apex rounded to truncate, 16-20 mm long and 6.5-10 mm wide, green to reddish from the bottom to the top; inner segments oblong to spatulate, entire margin, up to 2.5 cm long and 1 cm wide, white to pinkish-white; stamina included, numerous, arranged in verticillated series; basifix anthers, yellowish, style 23-52 mm long and 9-19 mm wide, white to pinkish white; stigma lobules 8-10, 2-4 mm long, yellowish white; nectar chamber semi-closed by the lower filaments curvature, 11-19 mm long and 5.9-6.7 mm wide, striated; ovary 8-13.3 mm long and 5.6-10.8 mm wide; fruit ovoid, dehiscent when ripe, about 48 mm in polar diameter by 40 mm in equatorial diameter, dark red, covered by areolas with numerous setose spines, about 11-18 mm long, yellowish white, deciduous at maturity, sweet flesh, red; ovoid, black seeds 1.3-2.2 mm long by 0.9-1.2 mm wide.
Phenology: flowers and fruits are produced over the year, but the peak of production lasts from November to April.
Habitat: S. griseus grows in tropical deciduous forest and xerophytic scrub alongside with Cereus spp., Prosopis juliflora, Bulnesia arborea, Ceiba sp. and Haematoxylum sp. From sea level to 1200 meters above sea level.  Discussion: S. griseus is the sole species of the genus Stenocereus to occur in South America. This name was actually a homonym comprising a previously undescribed northern Mexico species (S. huastecorum) and a second one from northern South America, which shall conserve the name S. griseus by priority principle.
Habitat: it inhabits tropical deciduous forest, xerophytic scrubland, thorny scrubland and mezquital, alongside with Prosopis spp., Vachellia spp., Larrea sp., Myrtillocactus geometrizans, and Stenocereus dumortieri. From 200 to 1600 meters above sea level. Discussion: S. huastecorum populations were considered introduced populations of South American S. griseus by Bravo-Hollis [13]. Moreover, it was also determined as S. pruinosus; genetic, ecological, and morphological differences demonstrate it deserves its own designation. This species can be distinguished by its spination pattern (7-9 radial spines and 3 central spines), pericarpel color (deep red), and restricted distribution to northern Mexico. Etymology: the name S. huastecorum follows the previous designation of the genetic entity described by Parra et al. [16] as "huasteca group", given that its distribution coarsely matches the Huasteca ethnolinguistic region, but we rather use the plural genitive ending to emphasize its belonging as a resource for those human groups.
Stenocereus Candelabraformtree, 3-8 m tall; defined trunk, 0.5-1 m tall and 15-20 cm wide, dark green with lustrous surface; mucilage cavities not evident in branch transverse sections; ribs 7, rounded in transversal section, straight in longitudinal section, about 3 cm tall by 5 mm wide at the base, without horizontal constriction between areolas within the same rib; areolas 8-23 mm apart, elliptic to obovate, 7.14-13.6 mm long and 5.5-10 mm wide, with light-colored non glandular trichomes; radial spines 7-11, acicular, 5.7-18 (30) mm long, white when young, fading grayish with age; central spines 1-4, deciduous except for the lower one, acicular, larger and more robust than the radial ones, up to 50 mm long, white, turning gray at maturity; lateral or subapical flowers, infundibuliform, 7.8-8 cm long and 4.5-6.5 cm wide when opened, night anthesis; pericarpel globose to ovoid, about 1 cm wide, green, covered with slightly prominent podaria, with short triangular scales, about 2 m long and wide (at the base), greenish with purplish hues, areolas with scarcely dense trichomes, yellowish white, without spines; receptacular tube 2.9-3.2 cm wide, podaria with decurrent scales, oblong, apex acute to mucronate, variable in length, about 0.7 cm wide, with few trichomes; outer perianth segments narrowly oblong to spatulate with acute apex, about 3 cm long and 1 cm wide, green with purplish hues; inner segments oblong to oblanceolate, up to 2.5 cm long and 1 cm wide, entire margin, white to pinkish-white; stamina included, numerous, in verticillate series; yellowish white filaments; basifix anthers, yellowish; style about 6.5 cm long by 2 mm wide; stigma lobules 7, 1.2 cm long, yellowish-white; nectar chamber partially closed by the inner filaments curvature, 2.3 cm long by 0.9 cm wide, striated wall; ovary 1 cm long and 0.8 cm wide; fruit globose to ovoid, dehiscent when ripe, diameter 5 cm, green with reddish hues, covered by areolas with numerous setose spines, about 1.5 cm long, white, deciduous at maturity, sweet flesh, red; ovoid, black seeds, 1.9-2.9 mm long by 1.3-2 mm wide.
Common name: "Pitayo", "Pitayo de octubre". Phenology: flowers during spring (March to May), fruits from April to June with a second reproductive peak between August and October.
Discussion: Pfeiffer described Echinocactus pruinosus from cultivated plants in Berlin Botanical Garden. According to Stafleu and Cowan [62] Pfeiffer's vouchers were deposited in the KASSEL herbarium, whose collection was destroyed during World War II. Lemaireocereus longispinus type, according to Britton & Rose [65] was cultivated in the New York Botanical Garden but now is no longer present.
Distribution: endemic to Mexico, in the states of Guerrero, Oaxaca, and Puebla.

Homonymy in Stenocereus griseus
The populations from northern Mexico and northern South America (red circles in Fig 1), which were previously considered to be S. griseus consistently are two different entities; these genetic groups (red and gray in Fig 3) are clearly differentiated. Moreover, the South American populations showed high genetic distances (Nei's D !0.253 0.293) compared with the Mexican populations. From an ecological point of view, a net differentiation of the reciprocal niche model suitability was detected (Fig 5G-5H) as well as in the Principal Components 1, 2 and 4 (Table 2). Finally, areolar features show differences in spination patterns, where the populations from northern Mexico have more radial and central spines. Therefore, we consider that there is enough genetic, ecological, and morphological evidence for proposing that populations for the northern Mexico group are a different species. We name this species S. huastecorum sp. nov., whereas the term S. griseus should remain for naming the South American populations, according to the priority principle [66]. We found no evidence supporting the Bravo-Hollis [13] hypothesis that the northern Mexico populations belong to the same taxon than the South American S. griseus. However, it has been recognized that the SGSC are often transported by humans [12,16], and the record of a single event of such type could misguide to such hypothesis. In addition, the incomplete revision in previous works of the South American vouchers and the scarcity of records in Central Colombia and Venezuela certainly favored the acceptance of homonymy.

Stenocereus pruinosus in central Mexico
Genetic delimitation fully supports the statements by Parra et al. [16] about population clusters in northern Mexico (the Huasteca group) and the eastern Tehuantepec Isthmus (the Chiapas group) as species different to S. pruinosus. Moreover, our study confirmed that the latter has a north-south substructure (green shades clusters in the Geneland column in Fig 2) which corresponds to the Tehuacán-Cuicatlán Valley and the Oaxaca Central Valleys. S. pruinosus is separated from S. huastecorum by a genetic barrier, consistent with the TMVB, and is separated from S. laevigatus by a second barrier represented by the Isthmus of Tehuantepec, a well-known biogeographic barrier (Fig 3). We did not observe genetic evidence of populations or individuals of S. pruinosus occurring in northern Mexico. Therefore, we do not consider a sympatric scenario between S. pruinosus and S. huastecorum in northern Mexico, but rather a long record of misidentified specimens.
Ecological evidence also provides clear distinction of S. pruinosus ENMs from those of S. huastecorum and S. laevigatus given that their comparisons (Fig 5, Table 2) suggest that these species have different ecological niches. Even though areolar morphology can easily distinguish S. pruinosus from S. huastecorum (every variable measured in Fig 6) only spine numbers (Fig 6E and 6F) were able to distinguish S. pruinosus from S. laevigatus. Areolar characters, however, may be confusing if developmental features are not taken into account because areolas may lose spines because of flowering events and branch age, or it may be simply deciduous as in central-left and right spines of S. laevigatus. Poor morphological differentiation is clearly related to the fact that this species pair shows the least interspecific genetic distance (Nei's D = 0.156). This suggests a recent divergence event, which involves the Isthmus of Tehuantepec constraining the distributional range of S. pruinosus to the Tehuacán-Cuicatlán and Oaxaca Central Valleys.

Stenocereus laevigatus in southern Mexico
In Chiapas and Yucatán there are up to three SGSC members records (Fig 1). Genetic clustering showed dominance of a single genetic group (Fig 3) that we consider to be S. laevigatus given that its records are more common than those of other taxa. Secondary genetic structure involving the westernmost populations of Umoa and Cintalapa (Chiapas) was detected by Geneland, Barrier, and UPGMA analyses. This break contrasts with the nesting pattern showed by the Yucatán Peninsula populations within the Chiapas group (Fig 3), even when they are separated by a distributional gap of over 500 km.
We consider the greater genetic divergence within the same region in Chiapas as an indicator of either recent demographic processes or occurrence of artificial selection. Human management is common throughout the distributional range of S. laevigatus, and it is particularly strong in Chiapas. Management has been previously recognized as a genetic-landscape modifier in S. stellatus [38], but historic demography and phylogeographic research are still needed to distinguish between biogeographic and human processes.
Even though this species is genetically and ecologically divergent with respect to S. pruinosus, if only morphology is taken into account, they may arise as cryptic species given that areolar features are very similar or actually identical in young or vegetative branches.

Stenocereus huastecorum, a novel species from northern Mexico
This entity stands as the most cohesive species within the SGSC: its genetic clustering pattern (Fig 3) is apparent regardless the method used. Ecological (Fig 5A, 5B, 5G and 5H; Table 1) and morphological evidence (Fig 6) shows its uniqueness. Therefore, we consider this group deserves the status of a distinct species given that there is enough genetic, ecological, and morphological evidence to distinguish it from S. griseus (its homonym so far) and S. pruinosus, a supposed sympatric species.
The distributional range of S. huastecorum sp. nov. comprises southern Tamaulipas, western San Luis Potosí, northern Querétaro and Guanajuato as well as disjunctive populations in Veracruz on the south slopes of the Trans-Mexican Volcanic Belt, displaying a distributional gap of 350 km. The distribution area is mainly contained in the Sierra Madre Oriental and the Llanura Costera Nororiental (Northeastern Coast Plain) physiographic regions of Mexico, coarsely matching the ethnolinguistic "Huasteca" region after which we name this species.

Summary of species limits and distribution
Every clustering method agreed in major genetic breaks associated with biogeographic regions. We were able to link the range of S. huastecorum with the Llanura Costera Nororiental, Sierra Madre Oriental (Huastecan Karst), and the southern Trans-Mexican Volcanic Belt, which seems to be the strongest genetic barrier when considering the whole complex (Fig 3). However, this is arguable for S. huastecorum, given that its populations dwell on both North and South of the oldest section of the TMVB (19.5 to 16 Myr), that predates the age of the Core Pachycereeae (Pachycereinae+Stenocereinae+Echinocereus) at 7 Myr [24]. The distribution of S. pruinosus is apparently contained in geologically recent lowlands, Tehuacán Valley and Oaxaca's Central valleys, surrounded by the Sierra Madre Sur, and it reaches the Pacific Coast through the Oaxaca's Southern Range, a Miocene volcanic sequence which is also older than the age of the tribe [67]. S. laevigatus is separated from S. pruinosus by the biotic barrier of the Isthmus of Tehuantepec [22] (Fig 3) which we identified as a genetic barrier, although not common for the Cactaceae; [24,25,68,69]. The core distribution of S. laevigatus is associated with the Central Depression of Chiapas and the Yucatecan Karst, specifically with the most recent areas of the Yucatán Península which date to barely 18,000 years [26] suggesting recent colonization. S. griseus seems to be widespread in Caribbean Coast of North Colombia and Venezuela well as Inter-Andean Valleys, and it is isolated from other SGSC members by the Caribbean Sea and Central America with a singular distributional pattern given that Stenocereus and its relatives are clearly North American [24]. Finally to explain both the S. griseus and Antillean S. peruvianus distributions, a more complex, biogeographic hypothesis other than human transport [13] needs to be tested. We observed spatial concordance between biogeographic and genetic barriers, well-known barriers (mainly highlands) such as the Trans-Mexican Volcanic Belt or Sierra Madre del Sur are by far older than the genus Stenocereus, thus being temporally discordant, whereas extant distribution areas are more recent (less than 4 Myr). This odd pattern may suggest that these ranges are soft barriers, with present-day distributions reflecting historical population refugial dynamics [70], or strong human effects on the genetic landscape considering that S. pruinosus is a intensively managed resource [71]. Demonstrating that one or a combination of these hypotheses will require a time-calibrated phylogeny not only of the SGSC but of the whole genus Stenocereus, as well as a phylogeographic approach within the major lineages.

Conclusions
We conclude that (1) the SGSC shows clear agreement between genetic and biogeographic regions. These genetic barriers, however, seem to be temporally discordant with geographic barriers, thus making a time-calibrated approach necessary. (2) Stenocereus griseus is a homonym of the South American species, which should keep the name by priority, and the new species from northern Mexico is here named Stenocereus huastecorum. (3) Co-occurrence of species records represent species misidentification rather than sympatry or admixture. Finally, the use of phylogeographic methods in the SGSC, including populations of Antillean S. peruvianus, is still needed to find evidence of the historical, anthropogenic, and biogeographic processes that lead to current species distributions.