The Rhipsalis Riddle - or the day the cacti came down from the trees: Part 3
Dr. Phil Maxwell, Bathgate's Road, Waimate, South Canterbury (Copyright, May 1999, reprinted with persmission from the author. This article originally appeared in the New Zealand Cactus and Succulent Journal.) email: philsue@voyager.co.nz

(b) The place of Rhipsalis in cactus evolution
As can be seen from the quotes at the beginning of Part 1 of this article, opinions on the place of Rhipsalis in cactacean evolution are sharply divided. Chorinsky (1931) regarded Rhipsalis and Pereskia as primitive genera in the family. Berger (1926; 1929: 89) classed Rhipsalidaeae as a tribe at the beginning of the subfamily "Cereoideae" [ie Cactoideae] but nonetheless thought their occurrence in the Old World probably resulted either from dispersal by birds or by rafting (Berger 1929: 18). Backeberg (1950: 183) demoted Rhipsalidinae to a subtribe of Hylocereeae but placed it at the beginning of the cereoideae. Britton and Rose (1923) took the opposite course by placing Rhipsalis right at the end of their monograph, after Mammillaria, and although this implies they regarded it as an advanced genus, they did not explain why they adopted this classification. Troll (1937) and Buxbaum (1950) stated explicitly that Rhipsalis and its relatives are advanced taxa. The latter author first ranked them as a subtribe Rhipsalinae of tribe Pachycereae (Buxbaum 1958: 199) but later (Endler & Buxbaum 1979: 18) as a subtribe of tribe Hylocereae near the beginning of the Cactoideae. Hunt (1967:449) demoted the cactus subfamilies to tribes and classed Rhipsalis with other epiphytic cacti (including Epiphyllum and Hylocereus) in a subtribe Cereinae "Group B". Of course it is difficult to express phylogenetic relationships adequately in what is essentially a linear classification, but it is interesting to note that most of these taxonomists have placed Rhipsalis and its allies near the beginning of the Cactoideae, irrespective of whether or not they regarded them as primitive or advanced. Rauh (1979:205) and Barthlott (1979: endpapers) attempted to clarify relationships in the family by arranging the genera in two dimensions to form a vaguely amoeboid diagram. They show Rhipsalis and related genera as forming a rather isolated group ("Gruppe II" in Rauh's diagram), but possibly having affinities with Hylocereus and its allies.

Anyone needing convincing that cactologists have considerable trouble in coming up with a stable classification for the family need look no further than those proposed by an "ad hoc Working Party" set up by the International Organization for Succulent Plant Study (Hunt & Taylor 1986; 1990). In the first of these the Cactaceae are subdivided into two subfamilies, Cactoideae (split into tribes Pereskieae and Cacteae) and Opuntioideae. Rhipsalis and its allies make up "Group II" of Cacteae, and placed between a group comprising the other epiphytic genera (Hylocereus, Epiphyllum, etc), but also Echinocereus and Peniocereus, and another group which includes Cereus, Pilosocereus, Arrojadoa and Melocactus. Four years later (Hunt & Taylor 1990) the Working Party reverted to the old three subfamily classification proposed a century ago by Karl Schumann, ie Pereskioideae, Opuntioideae and Cactoideae, and had shunted Rhipsalis (and disputed segregate genera) to a position ("Group IV") between two groups of South American ground-dwelling genera (Copiapoa, Eriosyce, Parodia, Frailea, Calymmanthium, Browningia, etc). The implication of the latter classification is clear: Rhipsalis is a derived, "specialised" genus.

The question that must be asked is "What characters does Rhipsalis (specifically R. baccifera) have that indicate it is a relatively primitive member of the Cactaceae?" Some that appear to be primitive - at least to me - include the simple, cylindrical stem without ribs or tubercles, the tiny, unsunken areoles, the absence of spines in the adult plant, and the small, pale flowers. The flowers are unusual in lacking a receptacle tube, in contrast with typical members of the Cactoideae, which have a well-developed flower tube. Buxbaum (1950: 118) claimed that the last character "had been used by many authors in classifying the group as primitive", but discounted this because he had concluded on the basis of vegetative features that "the Rhipsalides is a highly evolved group". Boke (1955) studied the vegetative shoot in R. cassytha [= baccifera], and although he noted that this species had what seemed to be primitive characters, he claimed that "its relatively simple anatomy seems best interpreted as having been derived from more complex forms", a conclusion disputed by Croizat (1961:758-759).

Buxbaum (1950:76) pointed out that "the seedlings of Rhipsalis do not have primitive hypocotyls or cotyledons, and that the primordial shoot is cereiform. This means, according to the Law of Recapitulation, that the ancestors of Rhipsalides were already quite succulent, few-ribbed, and of a basitonic branching habit like the Cereae". Seedlings of R. baccifera certainly have a swollen, succulent hypocotyl (the first-formed part of the stem) and small cotyledons (Buxbaum 1950: fig. 59b,c) the condition present in the seedlings of most Cactoideae, but not Pereskia species, which have a slender hypocotyl and large cotyledons (Buxbaum 1950: fig.51). This indicates Rhipsalis is more derived than Pereskia in this respect, which is a non-controversial conclusion; however, the seedlings of Hylocereus and Epiphyllum are similar to those of Pereskia (Buxbaum 1950: figs. 58, 59a), which might be taken as evidence that these epiphytic cacti are more primitive than (and possibly ancestral to) Rhipsalis.

An alternative scenario - one that I prefer - is that Epiphyllum, Hylocereus and their relatives on one hand, and Rhipsalis and its allies on the other, form quite separate groups of epiphytic cacti which were independently derived from Pereskia (or an allied genus). This is consistent with the second IOS classification (Hunt & Taylor 1990: see particularly note 4, p. 87), and implies the Cactoideae are diphyletic (ie have two separate origins). I suggest the Hylocereeae probably split off from Pereskia much later than Rhipsalis. Most of the genera included in the Hylocereeae occur in South America, Central America and the Caribbean which implies they were established prior to the disruption of proto-Central America in the Late Cretaceous (see above); however, their absence from Africa is evidence that they evolved after it had completed its separation from South America (about 90-80 Ma).

Another problem involves the provenance of the plants identified as Rhipsalis baccifera (or R. cassytha) in the botanical literature. The type locality of R. baccifera is assumed to be the Caribbean, possibly Jamaica (Barthlott & Taylor 1995: 63) and it seems likely that most of the cultivated plants are from the same area. This is certainly true for the plants studied by Boke which were collected in Cuba (Boke 1955:1). However, Caribbean (and Central American) populations are tetraploid, whereas those from South America are diploid and are therefore less derived. This is consistent with the idea that the R. baccifera group arose in South America. It would be worth checking to see if seedlings of South American "R. baccifera" (which perhaps should be differentiated taxonomically from the Caribbean population because of the chromosomal differences) have the same succulent hypocotyl and small cotyledons, or if they more closely resemble seedlings of Pereskia.

All that needs to be said about Buxbaum's second objection is that there is no such thing as a "Law" of recapitulation. Also known as the "Biogenetic Law", this claims that "ontogency recapitulates phylogeny", ie the growth stages of an individual organism replay its evolutionary history. It is now considered to be an over-generalization, as there is plenty of evidence from other plants and animals that juvenile characters in one species may become the adult characters in its descendant, ie the opposite of recapitulation. The significance of this phenomenon, known as paedsomorphosis, for cactus evolution is discussed below. (See Gould 1977 for a thorough discussion of the relationship between ontogeny and phylogeny.)
We therefore have a situation where many cactologists acknowledge that Rhipsalis is a relatively simple cactus, but claim that this simplicity must be secondarily derived, because of the prevailing notion that it is a specialised genus that arose relatively late in the history of the family. Now parasites often become very simplified as a result of their habit, and so end up looking primitive, but Rhipsalis is of course an epiphyte, NOT a parasite. If Rhipsalis is regarded as being more "specialised" than a strongly ribbed or tuberculate ground-dwelling cactus with prominent, sunken areoles, long spines, and large, deeply coloured flowers, then I suggest this is because most cactophiles are much more familiar with the latter, and tend to take them for granted as exemplars of the family.

The other line of evidence for the antiquity of Rhipsalis is the great diversity of plant form in the genus (even more if it includes species placed in Lepismium). This is greater than in any other cactus genus and indicates a long period of evolutionary radiation (see Fig. 3).

If this scenario is correct, if Rhipsalis has been around since the Early Cretaceous, then it should be possible to find some fossil evidence. According to McCarten (1981) the oldest bona fide cactus fossils are of Late Pleistocene age (a mere 22,000 years old at the most). This is not surprising as most live in habitats where fossilisation is unlikely. Even if a Rhipsalis stem was somehow fossilised as the result of rapid burial then it would probably be very difficult to recognize it as such unless flowers were preserved. (This is not impossible - a considerable number of fossil flowers, some as old as 110 million years, are known.) Pollen stand a much better chance of surviving as fossils, but I know of no fossil cactus pollen. Again, this is not surprising - cacti do not produce large numbers of pollen grains because they are pollinated by insects, birds or bats; it is the wind-pollinated plants that have to produce prodigious quantities of pollen to ensure at least some find a suitable flower, and are therefore well represented in the fossil record. However, I think it is only a matter of time before fossil cactus pollen are discovered. If Cretaceous Rhipsalis pollen is ever found, it will be a major boost for the hypothesis proposed here: even better if it turns up in Africa, Madagascar or Sri Lanka.
It is one thing to claim that Rhipsalis is an ancient, Gondwanic genus, another to suggest it played a significant part in the evolution of the Cactoideae. It is possible that the Rhipsalideae (Rhipsalis, Hatiora, Lepismium and Schlumbergera) represent an evolutionary cul-de-sac (albeit a successful one) that appeared early in the history of the family but remained almost entirely confined to the trees. An alternative hypothesis, which I prefer, is that Rhipsalis - specifically a member of the R. baccifera group - is ancestral to all other Cactoideae (except Hylocereenae).

At first sight this seems implausible - a nondescript, spineless epiphyte giving rise to a suite of mostly spinose ground-dwelling cacti. The first point to make is that some members of the R. baccifera group, including R. baccifera itself, can have an epilithic (living on rocks) as well as an epiphytic habit. More important for our story is the presence at an early stage in the development of R. baccifera of ribbed, bristly stems which are produced prior to the formation of the slender, spineless cylindrical adult stems (Fig. 4). This is the "cereiform" shoot mentioned by Buxbaum (1950: 76) as evidence for a "cereoid" ancestor for Rhipsalis.

Croizat took a characteristically radical position and declared it represented "an ancestral form in the pre-angiospermous stage of evolution" (1961: 759). Another possibility is that it is an adaptation (along with the swollen hypocotyl and abbreviated cotyledons) to protect the seedling in an epilithic habitat. the next step in the evolutionary scenario is represented by R. mauritiana, which occurs throughout tropical Africa, Madagascar, Sri Lanka and the small islands of the west Indian Ocean. Barthlott & Taylor (1995: fig. 12) illustrate a Madagascan specimen which bore flowers and fruits for several years on a spiny or bristly early growth stage before producing more typical spineless "adult" shoots. [They described the spiny stage as "juvenile", but the fact that it bore flowers means that it is by definition, adult.] The final step is R. horrida (restricted to Madagascar) which has dispensed with the spineless shoots, and has in addition bristles on the pericarpel and fruit (see Barthlott 1979: pl. 93 for a colour photograph).

The series R. baccifera - mauritiana - horrida may be interpreted as an example of paedomorphosis, in which the spineless shoots have become progressively reduced, and the bristly "cereiform" shoots which initially formed the juvenile stage have become the adult, flowering stage. It is not difficult to imagine something like R. horrida as the stem group from which the other ground-dwelling Cactoideae evolved; however, R. horrida itself is restricted to Madagascar and is unlikely to be the actual ancestor, particularly as Cactoideae other than Rhipsalis are unknown from outside the Americas. I suggest instead an origin in South America from another paedomorphic species (now extinct) that resembled R. horrida. Paedomorphosis is far from being a rare phenomenon, and is recorded from many groups of plants (eg the cactus Turbinicarpus valdezianus) and animals (eg the axolotl and Homo sapiens).

A problem?
The main objection to the hypothesis that Rhipsalis holds a key part in cactus evolution comes from molecular biology, specifically Wallace's studies on chloroplast DNA in the Cactaceae (1995). The cladogram given in his figure 11 does not show Rhipsalis as being particularly primitive, although it doesn't provide much evidence for it being an advanced genus either. Contrary to the impression that many people have, however, molecular approaches to phylogenetic problems are not quite as cut-and-dried as might be hoped. One problem is what happens if one or more taxa are removed from (or are added to) the analysis. If the phylogenetic reconstruction is robust then this will have little or no effect on the resulting cladogram, but it is sometimes found that the effect is drastic and unpredictable. I certainly have trouble reconciling the cladogram showing Leptocereus quadricostatus as "nested within the Pereskia clade" (Wallace 1995: 9, fig. 10) with the one on the next page (fig. 11) where Leptocereus is shown as being one of the most derived taxa in the cacti studied for this project! There are also occasions where molecular analyses give spectactularly unacceptable results. One example comes from a recent molecular phylogenetic analysis of bivalve molluscs where two gastropods (snails) are nested within the bivalves! It is problems of this sort that lead me to conclude that although molecular biology provides a powerful tool for systematic studies, it does not on its own give all the answers.

Synoptic History of the Cactaceae
To summarize: I regard the evidence for an ancient origin for Rhipsalis to be compelling, and further, I think it likely that it was a key player in the evolution of the Cactaceae, although I admit this suggestion is controversial. Important episodes in the history of the family, based on biogeographic, geological and taxonomic evidence are summarized below.

1. Pereskia or a very similar genus evolves from currently unknown ancestor (Portulacaceae?) somewhere in West Gondwana, probably in South America, and probably in the Early Cretaceous. Retains C3 metabolism. The current diversity in the genus suggests an early radiation, possibly forming shrubs, trees and climbers as in extant species.

2. Opuntioideae separate at an early stage in the history of the family, or have an independent origin (ie they are not cacti in the strict sense). [As Gibson and Nobel (1986:44) point out there is quite a problem facing "scientists who want to demonstrate a link between Pereskia and subfamily Opuntioideae on the one hand and with subfamily Cactoideae on the other."] whatever their origin, they probably evolved CAM at an early stage in their history.

3. One climbing species, possibly already with some degree of succulence, becomes epiphytic, in the process losing its leaves and evolving CAM. This is either a species of Rhipsalis or its immediate progenitor. R. baccifera appears at an early stage in the evolution of the genus and spreads rapidly through the tropical rain forests of West Gondwana but not into East Gondwana, possibly because of the formation of a seaway between the two portions of the supercontinent at about 130 Ma. The American and African/Madagascan populations of R. baccifera eventually become genetically isolated and begin to diverge.

4. Some populations of the Rhipsalis baccifera group forsake an epiphytic habitat for one on the ground. R. baccifera itself can be either epiphytic or epilithic and is therefore "preadapted" for such a change in life-style. NO major changes are required, just eventual suppression of the spineless shoots, and the concomitant transfer of the onset of flowering to the "cereiform" stage of growth (see above). Such a change may require no more than a single mutation to one of the genes controlling stem development. This transformation probably took place independently in South America and in Africa/Madagascar. The South American population forms the stock for other ground-dwelling Cactoideae (except Hylocereeae), but the eastern population (which is represented by the extant R. mauritiana and R. horrida and may well have appeared much later) proves to be a cul-de-sac.

5. Hylocereeae split off from Pereskia. Leptocereus may be the link between Pereskia and the typically epiphytic members of the tribe (eg Hylocereus, Epiphyllum and Selenicereus), but see the discussion on molecular systematics above. [Also note that Leptocereus is placed in a different group from Hylocereus and its allies in the second IOS classification (Hunt & Taylor 1990:89).]

6. Evolutionary radiation of ground-dwelling Cactoideae into arid and semi-arid habitats during the Cretaceous. Cacti were probably initially confined to South America, but establishment of an isthmus or archipelago between South and North America allowed some interchange of plants and animals during the Late Cretaceous (about 90-80 Ma). Opuntioideae, Pereskia, Hylocereeae, Melocactus and possibly even Mammillaria became established in proto-Central America at this time, and some of these (particularly Opuntioideae) probably spread into North America.

7. Proto-Central America is disrupted by seafloor spreading starting about 80 Ma, its components moving eastwards and carrying with them the ancestors of the cacti now living on the Greater Antilles. North and South America become isolated again and cactus evolution follows different paths on the two continents during the Tertiary. Geological processes, including the uplift of the Rocky Mountains and the Andes, volcanic activity and the development of the San Andreas Fault system produce a wide range of suitable habitats for ground-dwelling cacti.
8 North and South America joined again in the Late Pliocene (about 3.5-3.1 Ma), allowing an important interchange of animals (particularly mammals) and plants between the continents. Cacti probably began to spread back into Central America at this time, but there seems to be little evidence for significant interchange of North and South American cacti.

Summary
I hope I have shown how biogeography can be combined with evidence from other disciplines to throw some light on the evolution of the cacti, despite the fact that they have no fossil record of any consequence. Little of what I have discussed above is completely new, but I have attempted to meld published information with current ideas on historical biogeography, geology, evolutionary theory and taxonomy to produce what I hope is a coherent account of the history of this fascinating group of plants. It is of course highly speculative and can be described (and perhaps by some dismissed) as "story-telling", but speculation is an important part of the scientific process, particularly if some of the ideas can be tested.

Acknowledgements
I thank our Editor for her patience in waiting for this article to be completed and for supplying a copy of Backeberg's 1950 paper, and Nick Perrin for using his scanner to help get the article into the pipeline.

References Cited
(Because I live a considerable distance from major academic libraries it has not been possible to refer to all of the following publications. Those not seen are marked with an asterisk*.)
Allanby, M. (ed) 1985: "The Oxford Dictionary of Natural History." Oxford; New York: Oxford University Press. 688 p.
*Anthony, H.E. 1948: How Rhipsalis, an American cactus, may have reached Africa. Journal of the New York Botanical Gardens 49: 33-38.

Backeberg, C. 1950: Some results of twenty years of cactus research. Cactus and Succulent Journal (US) 22: 181-190.

Barthlott, W. 1979: "Cacti. Botanical aspects, descriptions and cultivation." Cheltenham; Stanley Thomas. 249 p. [English translation of "Kakteen" (1977)]

Barthlott, W.; Taylor, N.P. 1995: Notes towards a monograph of Rhipsalideae (Cactaceae). Bradleya 13: 43-79.

Benson, L. 1982: "The Cacti of the United States and Canada". Standford; Stanford University. 1044 p.

*Berger, A. 1926: "Die Entwicklungslinien der Kakteen". Jena; G. Fischer. iv + 105p p.
Berger, A. 1929: "Kakteen. Anleitung zur Kultur und Kenntis der wichtigsten eingeführten Arten". Stuttgart; E. Ulmer. vii + 346 p.

Boke, N.H. 1955: Development of the vegetative shoot in Rhipsalis cassytha. American Journal of Botany 42: 1-10.

Britton, N.L.; Rose, J.N. 1923: "The Cactaceae. Descriptions and illustrations of plants of the cactus family. Carnegie Institute Publication 248. Vol. 4. vii + 318 p.

Buxbaum, F. 1950-1955. "Morphology of cacti". Pasadena; Abbey Garden Press. 223 p.
Buxbaum, F. 1958: "Cactus Culture based on Biology". London; Blandford. 224 p.

*Camp, W.H. 1948: Rhipsalis - and plant distributions in the Southern Hemisphere. Journal of the New York Botanical Gardens 49: 88-91.

*Chorinsky, F. 1931: Vergleichende-anatomische Unter suchung der Haargebilde bei Portulacaceen und Kakteen. Österreich. Bot. Zeitschrift 80: 308-327.

Crane, P.R.; Friis, E.M. Runsgaard Pederson, K. 1995: The origin and early diversification of angiosperms. Nature 374: 27-33.

*Croizat, L. 1952: Manual of phytogeography. The Hague; Junk.

Croizat, L. 1961: "Principia Botanica or Beginnings of Botany". Caracas; privately published. 2 vols. 1821 p.

Cullmann, W; Götz, E; Gröner, G. 1986: "The Encyclopedia of Cacti". Sherborne; Alphabooks. 340 p. [english translation of "Kakteen" 1984].

Endler, J; Buxbaum, F. 1979: "Die Pflanzenfamilie der Kakteen". 3rd ed. Minden; A. Philler. 169 p.

Gibson, A.C.; Nobel, P.S. 1986: "The Cactus Primer". Cambridge, MA; Harvard University Press. vi + 286 p.

Gould, S.J. 1997: "Ontogeny and Phylogeny". Cambridge, MA; Harvard University Press. ix + 501 p.

Hallam, A. 1985: A review of Mesozoic climates. Journal of the Geological Society, London 142: 433-445.

Hull, D.L. 1988: "Science as a process. An evolutionary account of the social and conceptual development of science". Chicago and London; University of Chicago Press. 586 p.

Hunt, D.R. 1967: "Cactaceae" in Hutchinson, J. (ed) "The genera of Flowering Plants". Vol. 2: 427-467. Oxford; Oxford University Press. [Reprinted as "The genera of the Cactaceae" 1979.]

Hunt, D.R.; Taylor, N.P. 1986: The genera of the Cactaceae: towards a new consensus. Bradleya 4: 65-78.

Hunt, D.R.; Taylor, N.P. 1990: The genera of Cactaceae: progress towards consensus. Bradleya 8: 85-107.

*Irmisch, J. 1876: Über die Keimpflanzen von Rhipsalis cassytha und deren Weiterbildung. Bot. Zeit. 34: 193.

Marshall, W.T.; Bock, T.M. 1941: "Cactaceae with illustrated keys of all tribes, subtribes and genera". Pasadena; Abbey Garden Press. 220 p.

McCarten, N.F. 1981: Fossil cacti and other succulents from the Late Pleistocene. Cactus and Succulent Journal (U.S.) 53: 122-123.

Opdayke, N.D.; Jones, D.S. MacFadden, B.J.; Smith, D.L.; Mueller, P.A.; Shuster, R.D. 1987: Florida as an exotic terrane: paleomagnetic and geochronologic investigation of lower Paleozoic rocks from the subsurface of Florida. Geology 15: 900-903.

Ren, D. 1998: Flower-associated Brachycera flies as fossil evidence for Jurassic angiosperm origins. Science 280: 85-88.

*Roland-Gosselin, M. 1912. Les Rhipsalis découverts en Afrique sont-ils indigènes? Bulletin de Societie Botanique de France 59: 97-102. [English translation in Torreya 13: 151-156 (1913) and Desert Plant Life 19: 121-124 (1947)]

Rowley, G.D. 1978: Phytogeography and the study of succulent plants. Cactus and Succulent Journal of Great Britain 40: 3-5.

Schumann, K. 1897-1898. Gesamtbeschreibung der Kakteen. (Monographia Cactacearum). Neudamm: J. Neumann. xi + 832 p.

Shields, O. 1988: Mesozoic history and neontology of Lepidoptera in relation to Trichoptera, Mecoptera, and angiosperms. Journal of Paleontology 62: 251-258.

Storey, B.C. 1995: The role of mantle plumes in continental breakup: case histories from Gondwanaland. Nature 377: 301-308.

Wallace, R.S. 1995. Molecular systematic study of the Cactaceae: using chloroplast DNA variation to elucidate cactus phylogeny. Bradleya 13: 1-12.