Evolutionary history and diversification of Cactaceae

Phylogeny and common ancestors

Evolution does not arrange modern genera in order from simple to advanced. Every living species has an equally long history since the common ancestor, although individual lineages have retained different sets of traits. Leafy Pereskia is not an ancestor of mammillarias or opuntias. It is a modern sister lineage to some other cacti, with its own history of change. [de Vos et al. 2025]

This distinction matters when interpreting older diagrams. A comparison of a shrub, a columnar form and a globose plant can compare shoot geometry, but does not demonstrate that one plant shown arose directly from another. Relationships are reconstructed from many traits and DNA sequences. Growth habit is then mapped onto the resulting tree to test where similar solutions arose independently. [de Vos et al. 2025]

The phylogenomic study by de Vos and co-authors included 170 species representing nearly ninety percent of genera and frequently distinguished groups. The authors used hundreds of nuclear genes. This is broad sampling of genera, but not a complete inventory of about 1850 species. Good resolution of the main branches therefore does not imply resolution of every species and genus boundary. [de Vos et al. 2025]

Family age and limits of dating

Cacti have left almost no fossils that can be directly associated with tree nodes. Lineage age is therefore calculated mainly from rates of sequence change and calibrations drawn from broader angiosperm analyses. CactEcoDB summarizes modern reconstructions as consistent with a family crown age of approximately 27 to 37 million years. The crown comprises the last common ancestor of all living Cactaceae lineages and its descendants. [Arakaki et al. 2011]

This is not the date when the first plant that could be called a cactus appeared. The separation of Cactaceae from their closest relatives must have occurred earlier. Later radiations, in which many branches arose within a geologically short time, occurred in different parts of the tree. One figure therefore does not describe the origin of the family, the development of succulence and the rapid increase in species number. [Arakaki et al. 2011]

Archival explanations linked cactus age mainly to continental breakup and the family's absence from Africa. Current datings indicate a much younger crown than the separation of South America and Africa. The natural occurrence of Rhipsalis baccifera outside the Americas is interpreted as later dispersal, not a remnant of Gondwana's breakup. Age estimates without fossils still have a broad margin of uncertainty. [Arakaki et al. 2011]

Six subfamilies and conflict between genes

The classification of de Vos and co-authors distinguishes six subfamilies, Leuenbergerioideae, Pereskioideae, Opuntioideae, Maihuenioideae, Blossfeldioideae and Cactoideae. Four each contain one accepted genus in this treatment. Opuntioideae is divided into three tribes, and Cactoideae into eight tribes and fourteen subtribes. The authors accepted 155 genera and provided their synonymy. [de Vos et al. 2025]

These counts result from an explicit classification decision, not an unchanging property of nature. A gene represents a fragment of history, while species have the history of entire populations. During rapid radiation, different genes may support different branching arrangements because of incomplete sorting of ancestral variants, hybridization or limited information. De Vos and co-authors demonstrated a particularly complex signal around Cactoideae and pereskioid lineages. [de Vos et al. 2025]

Conflict between gene trees does not invalidate phylogeny. It indicates that separations occurred rapidly or involved gene flow, and that a single sequence is insufficient. In practice, classification may be stable at subfamily level while retaining uncertainty within young radiations. A genus name on a label may therefore change despite the plant's unchanged appearance. [de Vos et al. 2025]

Independent origins of succulence

Water storage in fleshy tissues has arisen many times in angiosperms. Similar selective pressure can produce convergent outcomes in unrelated lineages. A cactus and a succulent euphorbia may have green, ribbed stems and reduced leaves, but this does not make them a single closely related group. [Arakaki et al. 2011]

Within Cactaceae, traits likewise did not change as one package. Leaves, degree of stem succulence, CAM metabolism, root architecture and areole organization have their own histories. Leafy lineages are not a frozen image of the ancestor. De Vos and co-authors note that traditionally linking low succulence with a position at the tree's base can be misleading. The closest relatives of Cactaceae outside the family are mostly small herbaceous plants, complicating a simple image of an ancestor as a woody, pereskia-like tree. [Arakaki et al. 2011]

In cultivation, morphological similarity remains useful if it does not replace provenance data. Surface-to-volume ratio affects water balance, and branching affects shoot mechanics. This does not imply identical growth rhythms. Two globose cacti may belong to distant lineages and come from soils with different water retention and temperature patterns. [Arakaki et al. 2011]

Evolution of growth forms

In 2011, Hernández-Hernández and co-authors analysed 224 species representing about eighty-five percent of the genera then recognized. For Cactoideae, they reconstructed growth-form history on a tree built from five plastid and nuclear markers. Results showed repeated origins of globose, columnar and tree-like forms. Epiphytic habit arose independently at least in Rhipsalideae and hylocereoid lineages. [Hernández-Hernández et al. 2011]

The authors estimated the highest probability for a solitary, globose, ribbed Cactoideae ancestor, but the probability for globose habit alone was only 0.30. This is an important example of reconstruction limits. The highest of several values does not thereby become a certain result. Support for a globose, ribbed ancestor of Cacteae was much greater. Elsewhere in the tree, large columns and trees arose from different starting points. [Hernández-Hernández et al. 2011]

CactEcoDB describes growth habit in nine categories, tree-like, barrel-shaped, columnar, cushion-forming, epiphytic, geophytic, clumping globose, solitary globose and shrubby. These categories organize continuous variation and are not always mutually exclusive. One species may be assessed in more than one category, and a young plant may look different from a mature one. [Hernández-Hernández et al. 2011]

Environment and lineage diversification rate

The expansion of dry habitats created new opportunities, but does not explain all variation in cactus diversification rates. The 2014 analysis by Hernández-Hernández and co-authors combined dating, growth habit and sets of traits associated with pollination. The richest radiations in some genera occurred in the late Miocene and coincided with increasing aridity in North America. Diversification rate was also associated with changes in growth form and pollination biology. [Thompson et al. 2024] [Hernández‐Hernández et al. 2014]

The more recent 2024 analysis by Thompson and co-authors compared 39 variables and the phylogeny of 1063 species. Daily temperature range, soil sand content and plant size had the greatest predictive importance. Isothermality and range size had smaller but detectable contributions. For environmental variables, the highest estimated speciation rates occurred at intermediate values, not under the most extreme conditions. [Thompson et al. 2024] [Hernández‐Hernández et al. 2014]

The authors emphasize that the machine-learning model left about seventy-nine percent of variation unexplained. Results from models accounting for relatedness showed general trends but did not demonstrate a simple causal relationship. A variable important to diversification over millions of years is not automatically an instruction for an individual potted plant. Sand in a macroevolutionary analysis does not justify a universal single-component substrate. [Thompson et al. 2024] [Hernández‐Hernández et al. 2014]

Biogeography, environment and collection records

The family's natural centre of occurrence is in the Americas. Its range includes temperate Canada, dry North America, Central American forests and mountains, the Andes, eastern Brazil and Patagonia. The greatest richness is not evenly distributed. Important diversity centres lie in Mexico, the Andes and eastern Brazil. Cacti inhabit deserts, dry forests, cold high mountains and humid forests occupied by epiphytes. [Thompson et al. 2026]

Modern distribution results from migration, barrier formation, climate change, habitat specialization and extinction. No single present-day region can be identified as the exact birthplace of every principal family trait. Spatial data also vary in quality. CactEcoDB combines expert-defined ranges with cleaned occurrence records, but the authors still identify gaps and sampling biases. [Thompson et al. 2026]

In a collection, a useful label retains the name, field number or seed source, provenance location and identification-change history. The plant therefore remains useful for comparison after another revision. Appearance alone describes growth habit, but without provenance it rarely suffices to reconstruct a population's range or requirements. Phylogeny calls for caution here. A similar outline may represent a convergent environmental response rather than close relationship. [Thompson et al. 2026]

Morphological basis for describing Stenocactus × irregularis

Stenocactus × irregularis was described as a hybrid of S. coptonogonus and S. obvallatus, called S. pentacanthus in the source. In a comparison of 132 adult plants, the hybrid had an intermediate number of complete ribs, 15–28, compared with 12–14 in the first parent and 30–53 in the second. Co-occurrence, partially shared flowering time and the combination of traits formed the basis of the parentage hypothesis. The analysis included neither DNA markers nor controlled crosses.[González-Zamora et al. 2023, pp. 90–101; POWO, checked on 11 October 2026][2-WWW]

In a sample of 85 seeds from three hybrid plants, 78% germinated. The result documents germination of the collected seeds without specifying test conditions or duration or following subsequent seedling fate. Details of shoot structure and the scope of analysis are explained in the account of Stenocactus × irregularis.[González-Zamora et al. 2023, pp. 98–101]