By Scott A. Mori, Chih-Hua Tsou, Arne Anderberg, Ya-Yi Huang & Ghillean T. Prance

Click on phylogeny of Lecythidaceae to retrieve papers related to this subject. A summary of these papers is provided below. All references cited in the essay can be retrieved from the Literature Module.

The use of anatomical, morphological, and molecular data in a phylogenetic context has added a great deal to the understanding of evolutionary relationships within Lecythidaceae and its nearest relatives. Sometimes molecular data has unequivocally supported what we suspected was true; for example, that the central AmazonianAsteranthos brasiliensis did not belong to the Lecythidaceae where it had previously been assigned as the only member of the African subfamily Napoleonaeaceae in the New World (Prance & Mori, 1979). Tsou (1994) was the first to suggest that the closest relatives of A. brasiliensis were the African Syctopetalaceae based on embryology and this was confirmed by Appel (1996) based on morphology, especially the presence of ruminate endosperm in both taxa, as evidence. Molecular data (Morton et al., 1997) also supported these conclusions and A. brasiliensis is now considered as belonging to the Syctopetalaceae (Appel, 1996: Mori et al., 2007). Another example is molecular support for the long-standing hypothesis that actinomorphic flowers preceeded zygomorphic flowers and that Grias and Gustavia are the most basal Neotropical Lecythiaceae (Mori et al., 2007). An additional example is the confirmation that Lecythis is not monophyletic as shown in a recently published two gene tree (Mori et al., 2007). This was not unexpected because Mori (in Mori & Prance, 1990) recognized the extreme androecial variation found in Lecythis as representing separate sections of the genus. The new analysis suggests, however, that the current circumscription of Lecythis is not monophyletic and Mori’s sections should be considered genera.

Finally, our most recent molecular study of Lecythidaceae (Mori et al., 2007) summarizes what is currently known about the classification of the family and this supports previous classifications not based on molecular data. Our study shows that the Napoleonaeaceae, Scytopetalaceae, and Lecythidaceae conform to highly supported clades. Separation of the Lecythidaceae into subfamilies Foetidioideae, Planchonioideae, and Lecythidoideae is consistent with strongly supported clades hypothesized by Mori et al. (2007). Moreover, molecular data demonstrate that the New World Lecythidoideae are sister to the Old World FoetidiaPetersianthus clade, and morphological features suggest that the link is through the actinomorphic-flowered ancestors of Grias and Gustavia that are in turn sister to the zygomorphic-flowered New World taxa.

On the other hand, surprises based on molecular data are making Lecythidaceae specialists more carefully evaluate some aspects of previous systematic accepted relationships of the family. For example, molecular data place Lecythidaceae in the Ericales (Anderberg et al., 2002; Morton et al., 1997; Schönenberger et al., 2005) even though a sister relationship based on morphology had never been hypothesized before. This may be a problem that will never be solved because it is not possible to obtain molecular data for all species of Lecythidaceae and their ancestors over the evolutionary time frame of Lecythidaceae.

Another unexpected result is the suggestion derived from the latest molecular phylogeny that Eschweilera is not monophyletic (Mori et al., 2007). In the tree provided in that paper, Eschweilera appears as two unresolved clades – a large, wide-spread clade with seeds with lateral arils and a smaller clade with a more limited distribution in Central America, the Andean valleys, and western Amazonia having seeds surrounded by a spreading white aril that often completely surrounds the seed (see “Aril spreading” in the Glossary Module). The morphological difference that we have found that separates the two clades is the lateral aril characteristic of the larger clade and the spreading aril of the smaller clade.

Zygomorphy is restricted to the New World Lecythidaceae and appears to have evolved once, but this interpretation requires a reversal back to actinomorphic flowers in theAllantoma lineage (now including the actinomorphic-flowered Cariniana discussed in Mori et al. (2007).

Molecular data applied to New World Lecythidaceae indicate that species in the generaAllantoma, Corythophora, Couratari, Couroupita, Grias, and Gustavia form well supported clades and that under the current circumscriptions Cariniana, Eschweilera, and Lecythisare not monophyletic. Because the position of the monotypic Bertholletia excelsa in relation to the other zygomorphic-flowered genera is not resolved, the final story about its generic relationships has not yet been established. A cladistc analysis of Cariniana, however, clearly indicates that there are two groups in Cariniana and that one of these groups is more closely related to Allantoma than it is to the other species (true Carinianabecause it includes the type). Thus, Huang et al. (in press) have redefined Carinianasuch that half of the erstwhile Cariniana (those with zygomorphic flowers) stay in the genus and the other half (those with actinomorphic flowers) are transferred to Allantoma. This leaves only the question of monophylly in Eschweilera and Lecythis and the relationships ofBertholletia and other zygomorphic-flowered genera to be determined. Thus, the questions – 1) are Eschweilera and Lecythis monophyletic? and 2) what are the relationships among the zygomorpphic-flowered Lecythidaceae? are the most important questions that remain to be solved in the classification of Lecythidaceae.

We propose to tackle these problems in three ways by: 1) sequencing the ITS and MatK genes which have shown enough variation to distinguish closely related species in other studies and adding those data sets to those already obtained from the ndhF and trnL-F genes to generate hypothesized phylogenies, 2) hypothesizing a phylogeny based on anatomical and morphological data, and 3) combining the data sets from molecular sequences and from anatomical and morphological studies.

In addition to questions about classification, we will address hypotheses about character evolution. The combination of studying a family with relatively well-known natural histories of pollination and dispersal and the ability to hypothesize phylogenies based on trees generated from multiple data sets makes it possible to study co-evolution across an entire Neotropical plant family. Thus, my colleagues and I will address questions such as:

  1. Is the presence of actinomorphic flowers in Allantoma a reversal from ancestors with zygmorphic flowers? (see Mori et al. 2007)
  2. Are the different kinds of androecial hoods and two different types of pollen (fodder and fertile) the result of selection by different pollinators?
  3. Do the three bat-pollinated species have a common ancestry? Only one bat-pollinated species, Lecythis poiteaui, is represented in our molecular dataset so we will collect DNA of the other two species known to be bat-pollinated.
  4. Are the winged seeds of species of Allantoma, Cariniana, and Couratari the result of convergent evolution or did they evolve independently? Allantoma lineata, the best-known species of the genus, has water-dispersed seeds and retains only a vestigial seed wing (Tsou & Mori, 1994).
  5. Are the closest relatives of water-dispersed species wind- or animal-dispersed species found in adjacent non-flooded forests as hypothesized by Ducke (1948) as a general Amazonian phenomenon?

In conclusion, Neotropical Lecythidaceae provide an exciting opportunity to learn about the systematics and evolution of a ecological important group of tropical trees. We invite others to join us in this endeavor.

How to cite:

Mori, S. A., Chih-Hua Tsou, Arne Anderberg, Ya-Yi Huang & Ghillean T. Prance. 25 August 2010 onward. The Lecythidaceae Pages ( The New York Botanical Garden, Bronx, New York.