Endemism refers to a spatiotemporal character shown by each taxon or biotic group with a restricted geographic distribution (Anderson 1994). Although this term has received several different meanings (see Anderson 1994; Peterson and Watson 1998; Noguera-Urbano et al. 2017), two different levels have been usually distinguished: a spatial level, in which an area is categorised as an endemism area based on the occurrence of more than one species with rather restricted and largely congruent ranges (Haffer 1981; Anderson 1994); and a temporal level, in which endemic taxa can be classified according to their inferred evolutionary age (Stebbins 1942, 1974; Stebbins and Major 1965; Major 1988).

Spatial endemism responds to ecological, evolutionary, geographical, and climatic factors, all of which influence processes promoting the evolution of endemism areas (Stebbins and Major 1965; Anderson 1994; Linder 2008; Harrison and Noss 2017). Endemism areas have been generally attributed to historical processes, such as Pleistocene refugia (Hewitt 1996; Tzedakis et al. 2002) or major geological events (Jetz et al. 2004). Thus, endemism areas may result from either long-term climatic stability, which potentially reduces extinction events, or geographic complexity, which may promote the development of heterogeneous habitats that may enhance or limit biotic dispersal (Fine 2015; Harrison and Noss 2017). Regarding temporal endemism, two categories may be defined according to the time of origin: paleo-endemics and neo-endemics (Stebbins and Major 1965; Prentice 1976; Major 1988). Paleo-endemics refer to ancient and geographically restricted taxa for which current ranges represent remnants of formerly widespread distributions (Malik 2016). In contrast, neo-endemics refer to taxa of relatively recent origin that have not dispersed beyond their ancestral distributional range (Prentice 1976). The boundary between the recognition of paleoendemics and neo-endemics has not been clearly established. However, some authors locate the boundary between the Miocene/Pliocene transition (5–6 million years ago [Mya]; Fjeldså and Lovett 1997). Da Silva and Bates (2002) situate the boundary on the Pliocene/Pleistocene transition; thus, paleo-endemics are ancient lineages (> 2.6 Mya), while neo-endemics are recent lineages mostly originated during the Pleistocene. Areas with significant concentrations of paleo- or neo-endemic taxa are referred to as centres of paleo-endemism or neo-endemism, respectively (Mishler et al. 2014). Centres of paleo-endemism acted as places for survival (‘species museums’), where taxa persisted over time. In contrast, centres of neo-endemism (‘species cradles’) may have played a major role in relatively recent speciation and evolutionary events (Stenseth 1984; Jablonski 1993; Gaston and Blackburn 1996).

Early studies on assemblages of paleo-endemics and neo-endemics were carried out in regions previously recognised as refugia, such as California (Stebbins and Major 1965; Kraft et al. 2010), the tropical Andes (Fjeldså 1995), tropical Africa (Fjeldså and Lovett 1997), the Mediterranean Basin (Verlaque et al. 1997), the South African Cape region (Verboom et al. 2009), and southern China (López-Pujol et al. 2011). Recent studies have applied a phylogenetic approach and novel phylogenetic tools, including new metrics, such as relative phylogenetic diversity and relative phylogenetic endemism, and new methods, such as categorical analysis of paleo- and neo-endemism (CANAPE: Rosauer et al. 2009; Mishler et al. 2014). These analyses have located potential centres of endemism and classified them based on branch lengths in the phylogenetic tree of inhabiting taxa, allowing for a quantitative distinction among centres of neo- and paleo-endemism (Mishler et al. 2014). These novel methods have been used in analyses including community assembly, evolutionary biogeography, bioregionalism, and conservation studies in different geographic regions, such as Australia (Mishler et al. 2014; Schmidt‐Lebuhn et al. 2015), California (Thornhill et al. 2017), New Zealand (Heenan et al. 2017), Chile (Scherson et al. 2017), Mexico (Sosa et al. 2018), and North America (Mishler et al. 2020). In Mexico, endemism areas of vascular plants were recently assessed by Sosa et al. (2018) using a phylogenetic perspective on the distributional range of both paleo- and neo-endemic species. Their findings identified paleo-endemism areas (e.g. in Baja California, the Sonoran Desert, the northern Chihuahuan Desert, the Sierra Madre Oriental, the western Neovolcanic Belt, the Tehuacán-Cuicatlán Valley (TCV), and the Balsas Basin), neo-endemism areas (e.g. in the Sonoran Desert), and super-endemism areas concentrating both paleo- and neo-endemic taxa (e.g. in the northern Mexican Plateau and Sierra Madre de Chiapas).

The isolated TCV represents a complex physiographic mosaic of Cenozoic origin (Dávalos-Álvarez et al. 2007) in which internal minor valleys are separated by mountain chains, therefore promoting a very heterogeneous environment. This small area bears the greatest plant diversity of the Mexican arid regions, harbouring more than 3000 species, representing approximately 13% of the estimated flora of Mexico (Casas et al. 2016; Ulloa-Ulloa et al. 2017; Pérez-Valladares et al. 2019). Most of the flora in the TCV have Neotropical biogeographic affinities. However, some of Mexico’s arid and semi-arid plant communities are of Nearctic origin (Rzedowski 1973; Villaseñor et al. 1990). Additionally, a Mexican element has been recognised and includes 13% of the total plant diversity in the valley (Villaseñor et al. 1990; Méndez-Larios et al. 2005). Late Pleistocene climatic changes may have largely influenced the biotic composition of the present flora of the region, suggesting that local plant communities are of recent origin (Valiente-Banuet et al. 2009). The valley is thus a complex biotic mosaic in which up to 21 plant communities have been identified (Pérez-Valladares et al. 2019): xerophytic communities dominate the north-western part, while the south-eastern portion is dominated by warmer climates, favouring the development of more mesic communities (García 1998; Valiente-Banuet et al. 2009; Pérez-Valladares et al. 2019).

Cactaceae is nearly endemic to the Neotropics, and about 1,847 species have been recognised. The main centre of diversification for this family is located in Mexico, with a total of 670 species, 519 of which are endemic to the country (Ulloa-Ulloa et al. 2017). The highest concentrations of Cactaceae in the country occur in arid and semi-arid regions, tropical dry forests, and scrubland vegetation (Mutke et al. 2015) in the Chihuahuan and the Sonoran Deserts, and the Tehuacán-Cuicatlán Valley (Arias-Montes et al. 2012). Significantly, the relatively small area of the TCV harbours the highest diversity of Cactaceae in Mexico (Valiente-Banuet et al. 2009), in which most plant communities are dominated by endemic species of columnar cacti, highlighting this small area as an important diversity centre for the family (Valiente-Banuet et al. 2000; Mutke et al. 2015; Pérez-Valladares et al. 2019). At least 86 Cactaceae species occur in the TCV (Arias-Montes et al. 1997, 2012), such as the “viejito” (Cephalocereus columna-trajani), the “tetetzos” or “teteches” (Cephalocereus macrocephalus and C. tetetzo), and the “chendes” (Polaskia chende) (Fig. 1).

Despite being a Cactaceae hotspot, including both the highest species richness and high levels of endemism (Méndez-Larios et al. 2004; Arias-Montes et al. 2012), no studies address the causes of this high diversity in the TCV using a phylogenetic approach. The high diversity and endemism of Cactaceae (Valiente-Banuet et al. 2009) in the isolated TCV suggests that diversification may have occurred through several pulses during the Pleistocene, which, in addition to the complex topography, favoured both the persistence and the speciation in this group. We, therefore, expected the TCV to be an area in which both neo- and paleo-endemic species may be found. We also expected to locate mixed-endemism areas either in topographically rugged or ecologically complex zones (e.g. ecotones, isolated habitat patches). We here raised the following questions to approach the study of the endemism of Cactaceae in this region: Is the TCV a place for the persistence of ancient taxa (paleo-endemics) or promoting the speciation of new taxa (neo-endemics)? Are Pleistocene climate changes and isolation drivers of origin and diversification of endemic lineages in the TCV? To answer these questions, we first conducted a qualitative assessment of the paleo- and neo-endemism in Cactaceae in the TCV. Then, we undertook a quantitative assessment to understand if the areas in which paleo- and neo-endemic species are distributed may be considered significant endemism areas. The conjunction of both qualitative analyses on species and quantitative analyses on areas may improve the understanding of the evolution of the endemism in the isolated TCV.

Recent work supports that centres of endemism occur in regions with long-term climatic stability, which likely buffered surrounding unfavourable climatic conditions (Wiens and Donoghue 2004; Jablonski et al. 2006; Fine 2015; Harrison and Noss 2017). In addition, the topographic complexity and the isolation of some of these regions enhance speciation rates, producing high concentrations of endemic species (Cai et al. 2023). Climatic fluctuations during the Pleistocene strongly impacted the diversification of taxa by limiting the distribution of many species to isolated regions of long-term spatiotemporal climate stability, which contributed to the development of refugia (Haffer 1969; Fjeldså 1995; Davis and Shaw 2001; Jetz et al. 2004; Harrison and Noss 2017). These refugia harbour high biodiversity and endemism, enabling the persistence of paleo-endemic taxa but also promoting the speciation of novel taxa (neo-endemics) (Harrison and Noss 2017; Cai et al. 2023).

In Mexico, several regions, including mountain chains and desert areas, have been widely recognised as hotspots for plant and animal richness and endemism and have frequently been suggested as Pleistocene refugia (e.g. Becerra 2005; Weeks et al. 2005; Delgado-Salinas et al. 2006; Flores-Villela and Martínez-Salazar 2009). These refugia likely show a mixture of potential paleo- and neo-endemic taxa, pointing to multiple historical processes involved in the origin and maintenance of biodiversity. Particularly, the isolated TCV is a small area surrounded by mountains likely acting as barriers that promoted the isolation of mainly arid and semi-arid climate-adapted taxa, leading to the evolution of an area with high biodiversity and a high proportion of endemic species (Dávila et al. 2002; Méndez-Larios et al. 2004). The family Cactaceae has 27 endemic species in the TCV. This number does not seem very large compared to the 519 species of Cactaceae endemic to the country, which could be interpreted as a limitation to this analysis, but this pattern of narrow endemism is common to other areas of the country, for example, the Chihuahuan Desert has 229 endemic species. However, this is 50 times more extensive than the TCV (Hernández et al. 2004). Still, if regional floras are analysed, local endemism also records few species, 23 species in Cuatro Ciénegas, Coahuila (Pinkava 1984), six in Mapimí, Durango (García-Arévalo 2002), or ten in El Huizache, San Luis Potosí (Hernández et al. 2001). In TCV, most endemic species are geographically restricted, such as Thelocactus tepelmemensis, recently described by Davis et al. (2018) and whose only known population has been found growing on limestone rock faces in a narrow canyon in northern Oaxaca. Some species, however, are widely distributed, such as Cephalocereus tetetzo, a branched columnar cactus widely distributed in xerophytic shrublands and tropical dry forests across the valley (Arias-Montes et al. 1997).

The distinction between paleo-endemics and neo-endemics has relied on various criteria, including geography, taxonomy, cytology, geology, climate, and phylogeography (Favarger and Contandriopoulos 1961; Bramwell 1972; Major 1988; Cronk 1992); however, these criteria may be ambiguous. A bounded timescale allowed us to discriminate between palaeo- and neo-endemics, suggesting that molecular phylogenies may be one of the most accurate methods to estimate temporal endemism in plant lineages. We, therefore, considered the criterion of Da Silva and Bates (2002) as accurate for our study because it is based on a historical framework from a molecular phylogenetic analysis, which seems to fit well with the local geological history of the region (Dávalos-Álvarez et al. 2007). Thus, according to our dated phylogeny, 76% (16) of the 21 endemic species were classified as neo-endemics, suggesting that the TCV may have played a major role in recent speciation and diversification in Cactaceae. We found that only 24% (5) of the endemic Cactaceae species were classified as paleo-endemic species (> 2.6 Mya), suggesting a relatively minor role of the TCV region in the persistence of ancient taxa. Mammillaria napina and Lemaireocereus hollianus are paleo-endemic species with a divergence estimated at 5.18 and 4.43 Mya, respectively. Divergence times of other probable paleo-endemics, such as Mammillaria huitzilopochtli (3.11 Mya) and M. hernandezii (3.01 Mya), are not clear because these values are near the boundary of 2.6 Mya. Furthermore, the HPD values show that some neo-endemic taxa could also be classified as paleo-endemic, such as Ferocactus latispinus subsp. spiralis with a divergence time estimated at 1.8 Mya (HPD 0.08–4.41 Mya), Opuntia tehuacana 1.79 Mya (HPD 0.75–3.04 Mya), and Polaskia chende 1.7 Mya (HPD 0.36–3.6 Mya).

Neo-endemic species are found in the genera Cephalocereus, Echinocereus, Ferocactus, Mammillaria, Opuntia, and Polaskia. Mammillaria is the most speciose genus in Cactaceae with approximately 180 species, from which more than 90% of the species are distributed in Mexico, and about 85% are endemic to the country (Butterworth and Wallace 2004; Crozier 2005; Hunt 2006; Hernández and Gómez-Hinostrosa 2015). Although the origin and high diversification rates of the genus might be associated with geographic expansion during the aridification of North America in the Miocene 8.62 Mya (HPD 5.83–12.56 Mya; Hernández-Hernández et al. 2014), a high percentage of Mammillaria species could have evolved due to Pleistocene climatic changes (Cervantes et al. 2021). Our results showed the divergence time estimates for TCVMammillaria endemic species ranged from 0.20 Mya (HPD 0.00–1.02 Mya) for Mammillaria haageana subsp. vaupelii to 5.18 Mya (HPD 3.37–7.15 Mya) for M. napina. Thus, six out of ten species included in our phylogenetic analysis are neo-endemics. Remarkably, most of these taxa are geographically restricted; for example, the neo-endemics M. crucigera and M. supertexta are restricted to the south-eastern TCV region, which is geologically recent and mainly comprised of alluvial fans dating back to the Pleistocene-Holocene (Brunet 1967; Dávalos-Álvarez et al. 2007). This region is exposed to humidity from the Gulf slope, which has promoted the development of more mesic vegetation, such as tropical seasonally dry forests. Speciation processes in tropical seasonally dry forests were probably favoured by Pleistocene climatic changes, which might have promoted high rates of diversification in these isolated and climatically stable environments (Pennington et al. 2004).

Other plant groups also show similar patterns to the one found for Mammillaria. Recent studies of ancient lineages, such as cycads, have shown that this gymnosperm group dates to the late Palaeozoic (Norstog and Nicholls 1997); however, extant species such as Dioon (Zamiaceae) are recently differentiated lineages (Gregory and Chemnick 2004; Dorsey et al. 2018). Dorsey et al. (2018) found that Dioon originated at 7.86 Mya (HPD 7.09–8.71 Mya) in the Miocene, and the diversification of extant species occurred during the Pleistocene, suggesting that modern species are not paleo-endemics but rather recently derived neo-endemic species. Notably, Dioon purpusii Rose, D. argenteum De Luca, Sabato & Vázq.Torres, D. califanoi De Luca & Sabato, and D. caputoi T.J.Greg., Chemnick, Salas-Mor. & Vovides are all neo-endemics restricted to the TCV. According to Dorsey et al. (2018), these species have evolved from populations periodically shifting to lower elevations in response to Pleistocene climatic fluctuations, supporting our hypothesis that these climatic changes and isolation might have driven divergence and speciation in endemic lineages of the TCV.

Regarding the spatial level, CANAPE identified two cells of high phylogenetic endemism (one of mixed-endemism and one of super-endemism) in the TCV (Fig. 2). Our results partially agree with Sosa et al. (2018), in which the whole TCV was included within a mixed-endemism centre and no super-endemism areas were detected. Similarly, Mishler et al. (2020) identified centres of mixed-endemism around the TCV, but no centres of super-endemism were reported. Furthermore, a recent study on Cactaceae also detected mixed-endemism and super-endemism areas around the TCV (Amaral et al. 2022). These differences in the categorisation assigned to the endemism areas may be scale-dependent, as has been observed in the estimation of species richness and endemism (Whittaker et al. 2001; Hartley and Kunin 2003; Laffan and Crisp 2003; Chase et al. 2019; Luebert et al. 2022), as well as in spatial phylogenetics (Daru et al. 2020). Recent studies have found that using small scales may recover significantly high values of phylogenetic diversity (Thornhill et al. 2016, 2017; Scherson et al. 2017; Allen et al. 2019; Mishler et al. 2020). We used a small grid-cell size, which allowed us to recover a centre of super-endemism not previously reported, underlying the significance of the TCV in the diversification and maintenance of species diversity in Cactaceae. Therefore, small scales may reveal emergent patterns related to phylogenetic endemism, thus changing the categorisation of endemism centres.

Both the mixed-endemism area (grid-cell 19) and the super-endemism area (grid-cell 20) are dominated by temperate forests, with scattered patches of xerophytic scrub and seasonally dry forest. In the TCV, oak, mixed, and pine forests occur as isolated patches in altitudes between 1,630 and 2,200 m (Valiente-Banuet et al. 2000). The super-endemism area included two exclusive species: Thelocactus tepelmemensis in xerophytic scrub and Mammillaria oteroi in Quercus forests, all other species in this area are widely distributed in the TCV. Species in the mixed-endemism area include Echinocereus acanthosetus, M. hernandezii, and M. tepexicensis (the latter endemic to this grid cell). Collected specimens in herbariums suggest that these species are also distributed in temperate forests, such as Pinus, Quercus, and Juniperus forests. These results support our hypothesis that mixed- and super-endemism areas may be found in topographically or environmentally complex regions.

Our results suggest that the TCV region has played a predominant role as a cradle, promoting the recent evolution of endemic plant species, which have been enhanced by landscape heterogeneity and isolation. Isolation has promoted in situ speciation, which led to high neo-endemism. Additionally, environmental drivers such as the long-term stability of climate and habitats on a reduced spatial scale have favoured the evolution of the TCV endemic plant species.