Cochemiea. Plastid relationships and the young radiation of Baja California

From red flowers to a broader genus

Classical Cochemiea comprised a small group of cacti with cylindrical stems and red tubular flowers. The 2020 protologue of Cochemiea thomasii used this scope and provided a key to six species. Today's Cochemiea in POWO includes 36 accepted species, including many former Mammillaria. Changes to genus boundaries resulted from comparing the evolutionary history of a broader set of plants. [García Morales et al. 2020, pp. 1–5] [2-WWW]

For some current species, the historical name appears in a paper's title, material description or tree label. C. halei, C. poselgeri, C. pondii, C. maritima and C. setispina bear Mammillaria names in the 2021 tree of Breslin and coauthors. These are the same epithet combinations now linked to accepted names. The scope of the analysed sample is retained separately from the number of species in the current list. [Breslin et al. 2021, pp. 313, 317–323]

What was read from plastids

The 2021 analysis covered 88 specimens representing 78 taxa. It used the large single-copy region of the plastid genome, abbreviated LSC. The matrix contained 93808 aligned DNA positions. Three approaches to tree reconstruction, maximum likelihood, maximum parsimony and Bayesian inference, produced consistent major groups. Traditional broad Mammillaria contained plants from several separate lineages. [Breslin et al. 2021, pp. 310–314; Fig. 2]

Plastids are cellular organelles that include chloroplasts. The regions analysed belong to the same linked genome. A large number of positions strengthens the reading of its history, but does not create as many independent genealogies as numerous unlinked nuclear loci would. The resulting arrangement is a strong hypothesis of plastid relationships among the specimens studied. Delimiting species or reconstructing gene flow also requires information suited to those questions. [Breslin et al. 2021, pp. 310–315]

One broad lineage, different boundaries between names

Breslin and coauthors proposed a narrower Mammillaria and a broader Cochemiea. In the latter, they also included material of the current Neolloydia conoidea and Ortegocactus macdougallii. In 2023, Chincoya and coauthors recovered a similar broad lineage but favoured separate genera, citing morphological, ecological and biogeographic differences among its smaller groups. The discussion therefore also concerned the choice of rank for the identified subgroups. [Breslin et al. 2021, pp. 313–316] [Chincoya et al. 2023, pp. 15–16]

In October 2026, POWO accepts Cochemiea with 36 species and retains separate Neolloydia and Ortegocactus. The atlas follows this current list and describes published proposals within their own scope. The 2021 and 2023 analyses lacked Mammillaria mammillaris, the type species of Mammillaria. The authors related the name to earlier data and relatives studied, while the 2023 paper identified the need to include the type directly in the sample. [2-WWW] [2-WWW] [2-WWW] [Breslin et al. 2021, pp. 314–316] [Chincoya et al. 2023, pp. 15–16]

Spines and flowers within the lineage

Hooked central spines were common in the expanded Cochemiea lineage. Reconstruction of character states on the tree allowed assessment of where this shape might have appeared or disappeared. C. halei has straight spines and grouped with C. poselgeri on the tree. Hooked spines also occurred in Mammillaria senilis, retained in narrower Mammillaria. A single striking character does not coincide with the full boundaries of a genus. [Breslin et al. 2021, pp. 312–315; Fig. 2–3]

The elongated red flower of the former Cochemiea group has characters of a syndrome usually associated with hummingbird pollination. In the description of C. thomasii, the association was presented as presumed. Phylogenetic analyses compare the distribution of such characters on a tree without observing all pollinators in the field. In C. thomasii, the described similarity to C. halei has a morphological basis, because the new species was not included in the DNA samples of the studies discussed. [García Morales et al. 2020, pp. 1–5]

Recent splits in regional lineages

In 2022, the analysis was expanded to 89 specimens, including 39 Cochemiea taxa under the authors' treatment. Time trees were estimated in BEAST and RevBayes using secondary calibrations from earlier studies. The lack of direct fossil calibrations for this group meant that timing also depended on adopted estimates. The broad Cochemiea lineage began approximately 5,5 million years ago, while many of its regional splits were much younger. [Breslin et al. 2022, pp. 1475–1481; Fig. 1–3]

Of the 35 regional taxa considered, the analysis placed the origin of 28 lineages between about one million and 200 thousand years ago. Such dates indicate intense recent differentiation. For the youngest nodes, uncertainty intervals are broad enough to make precise assignment of species splits to a single event, such as the isolation of a particular island, difficult. Similarity between lineage ages and geographic changes provides a hypothesis to assess. [Breslin et al. 2022, pp. 1478–1483]

The diversification rate depends on the model

The BAMM model estimated the mean diversification rate of broadly defined Cochemiea at 1,15 lineages per million years, with a reported 95% interval of 0,895–1,454. TurboMEDUSA gave a median of 1,99 lineages per million years. Both approaches indicated an elevated rate in this part of the tree, but differed in value and in the details of changes along branches. These are parameters of a modelled history of lineage accumulation, dependent on dating, sampling and analytical assumptions. [Breslin et al. 2022, pp. 1476–1477, 1480–1482; Table 1]

In the simpler BiSSE model, hooked spines were associated with faster diversification. Applying HiSSE, which allows effects of hidden, unmeasured character states, greatly weakened this association. The result allows a possible role for the hook to be considered, but does not establish it as the cause of the radiation. The comparison shows how a correlation with one character can coexist with other changes in the environment and plant biology. [Breslin et al. 2022, pp. 1477, 1479–1480, 1483–1484]

Dispersal, geography and comparison with 2023

The 2022 reconstructions considered nine geographic areas and indicated repeated movements of lineages between the peninsula, islands and mainland. The authors linked regional differentiation to tectonic processes, climate changes and dispersal opportunities. Bird transport of seeds remains one proposed explanation. A model of historical ranges does not track an animal or an individual seed-transport event. [Breslin et al. 2022, pp. 1476–1479, 1482–1484; Fig. 1, 3]

In 2023, Chincoya and coauthors analysed 52 plastid loci from 142 genomes representing 103 taxa. Some raw data came from the 2021 study. Their biogeography included thirteen regions, and the origin of the entire mammilloid clade was estimated at 7,37 million years, with a 95% interval of 4,86–10,02 million years. For smaller lineages close to Cochemiea, an association with northwestern Mexico was reconstructed. The conclusion that Baja California was colonized through dispersal was based on dates younger than the opening of the Gulf of California. [Chincoya et al. 2023, pp. 3–8, 12–15]

In this sample, the classical red-flowered group was represented by C. halei, C. pondii and C. poselgeri. Different sample numbers, calibrations, choices of plastid regions and divisions of regions affect comparison of the two studies' dates and routes. Agreement in some results strengthens the picture of recent differentiation in northwestern lineages, while disagreement about genus scope remains an explicit part of the interpretation. [Chincoya et al. 2023, pp. 7–8, 12–16]