Molecular phylogenetics defines the ancestor of
apicomplexans, ciliates, and dinoflagellates.

It is now well understood that apicomplexans (such as malaria parasites), ciliates and dinoflagellates share a recent common ancestor and form the super-phylum Alveolata1. What is not clear is what that ancestor might have looked like. Each of these three phyla have rather unique characteristics sharing only a few common features such as a second inner membrane system (alveoli) and micropores. For example, dinoflagellates like Pfiesteria and other neurotoxic species have uniquely condensed chromosomes that once were thought to be ancestral to all eukaryotes2. The Ciliophora exhibit uniquely arranged cortical ciliature and generally have two nuclei. The apicomplexans, so named for their distinctive apical organelles, also exhibit gliding motility not shown by any other group of organisms. Determining the ancestral conditions from among these traits has confounded protozoologists. Regarding the origins of the apicomplexans, for example, Levine3 held that the oyster parasite Perkinsus marinus may be ancestral, but that this "is conjectural and never will be proven. There are no fossils". Here we demonstrate with a molecular phylogenetic analysis including the genus Colpodella, and in the absence of fossils, the morphological identity of the ancestral alveolate.

Perkinsus marinus , which contributed to declines of the Chesapeake Bay oyster, was thought to be representative of the ancestral apicomplexan when it was discovered to posses structures reminiscent of a conoid and rhoptries4. However, molecular phylogenetic analyses based on multiple genes clearly place Perkinsus species with the dinoflagellates5. Although all known species of Perkinsus are parasites of marine invertebrates (especially bivalves), species of Colpodella, except for their free-living habits, are indistinguishable from Perkinsus species6,7,8. Colpodella species (prev. Spiromonas) are voracious predators of other free-living protists. With a rostrum, they penetrate through the cell membrane and slowly but completely consume the cytoplasmic contents of their victims. The rostrum of Colpodella species is identical in structure and function to the "conoid" of Perkinsus zoospores. Although Colpodella species were once thought to be kinetoplastids, similar to Bodo, the close relationship between Colpodella and Perkinsus now seems well established8.

Our analysis, however, paints a very different picture. Preliminary comparisons to Bodo confirmed that Colpodella is not closely related to the kinetoplastids. Analysis of aligned SSU rDNA for 26 other alveolate ingroup taxa with 7 stramenopile and fungal outgroup taxa resulted in 2 equally parsimonious trees (length = 3388, RI = 0.526). The consensus placed Colpodella as sister-taxon to the Ciliophora with a Bremer9 support (b) of 14. This support exceeded that found either for monophyly of Ciliophora (b = 11) or of Dinozoa (b = 10). Perkinsus species grouped as sister taxon to the Dinozoa (b = 10). Twenty-one extra steps were required to constrain Colpodella to group with Perkinsus species (Templeton test 0.025 > P > 0.0199; Wilcoxon signed rank test 0.013 > P > 0.008 on four equally suboptimal trees).

The failure of Colpodella and Perkinsus to group together is surprising because they are almost identical morphologically and developmentally. Motile stages of both have a pair of anterior orthogonal flagella, vesicular mitochondrial cristae, inner alveolar membranes, and micropores. Moreover, both have the open-sided truncated "conoid" and sacculate extrusomes or "rhoptries" that occupy the length of the cell. Following feeding, species of Perkinsus and Colpodella lose their flagella, become spherical, exhibit a large central vacuole, encyst and divide8. The only obvious difference in their life cycles is that species of Colpodella appear to only produce 4 progeny per cyst, whereas Perkinsus species can produce up to 32 progeny10. Otherwise there is visibly little to distinguish these genera.

Finding that these two genera do not have an immediate common ancestor, implies that this morphology and life-history must have been maintained unchanged from a more distant ancestry. Because of their respective placements as sister taxa to dinoflagellates and ciliates respectively, that organism with the same morphology ultimately must have been the ancestor of both of those groups and thus also of the Apicomplexa.

Mark E. Siddall
American Museum of Natural History,
New York, New York 10024, USA

Thomas A. Nerad
American Type Culture Collection,
Manassas, Virginia 20110, USA

Kimberly S. Reece
Eugene M. Burreson

Virginia Institute of Marine Science,
Gloucester Point, Virginia 23062, USA

1. Gajadhar, A. A. et al. Mol. Biochem. Parasitol. 45, 147-154 (1991).
2. Loeblich, A. R. J. Protozool. 23, 13-28 (1976).
3. Levine, N. D. The protozoan phylum Apicomplexa (CRC Press, Boca Raton, 1988).
4. Perkins, F. O. J. Parasitol. 62, 959-974 (1976).
5. Siddall, M. E. et al. Parasitology 115, 165-176 (1997).
6. Brugerolle, G. & Mignot, J. P. Protistologica 15, 183-196.
7. Myl'nikov, A. P. Zoologich. Zhur. 7, 5-15.
8. Simpson, A. G. B., & Patterson, D. J. Syst. Parasitol. 33, 187-198 (1996).
9. Bremer, K. Evolution 42, 795-803 (1988).
10. Perkins, F. O. J. Shelfish Res. 15, 67-87 (1996).
11. Goloboff, P. A. Nona. Program and documentation. New York.

FIGURE CAPTION. Phylogenetic analysis of Alveolata using stramenopiles and fungi to root the tree separates the morphologically indistinguishable species of Colpodella and Perkinsus . This suggests that their morphology is plesiomorphic (dashed line) and that it is representative of the ultimate ancestor of alveolates. Colpodella sp. [ATCC 50594] was isolated from brown woodland soil from Gambrill State Park, Maryland in 1993. The strain feeds on Bodo caudatus and was maintained in vitro at the American Type Culture Collection with its naturally ocurring bodonid prey. DNA isolation, amplification of SSU rDNA for Colpodella followed standard protocols. All other sequences were obtained from GenBank. Multiple alignment was effected with Clustal-W. The most parsimonious solution and branch support values (at nodes) were obtained with Nona11 using the ratchet.