Eukaryotes

The relationships of the major lineages of eukaryotes have become much better known during the last decade as more sequence data have been gathered and analyzed (reviewed in Baldauf et al, 2000; Hedges, 2002; Keeling et al., 2005). Because the relationships of single-celled eukaryotes (protists) are intimately tied to the relationships of multicellular eukaryotes (algae, plants, fungi, animals), it is usually more convenient to discuss this subject in terms of overall (higher-level) eukaryote phylogeny, as will be done here. The land plants, fungi, and animals will be discussed in separate sections.

The ease of sequencing ribosomal RNA (rRNA), and especially the small subunit, meant that an initial molecular view and framework of eukaryote phylogeny, from molecules, was based on that gene. For eukaryotes, those trees defined a crown consisting of plants, animals, fungi, and related protists, and a series of lineages along the stem or base of the tree, with the diplomonad Giardia as the earliest branch (Sogin et al., 1989; Schlegel, 1994). Later analyses using complex models and different genes showed that some—but not all—basal branching lineages (e.g., microsporidia, Dictyostelium) actually belong higher in the tree, and their misplacement was the result of long-branch attraction or other biases (e.g., Philippe and Germot, 2000). Subsequently, much recent attention has been placed on building trees with as many genes and taxa as possible, and using different types of analyses, including complex substitution models. This has brought welcomed stability to some aspects of the tree, and remarkable volatility to others.

Some major questions that were once controversial have now been answered to the satisfaction of many in the field. For example, animals and fungi appear to be closest relatives (opisthokonts) to the exclusion of plants (although see Philip et al., 2005), and red algae are on the "plant lineage" and not basal to the divergence of plants and opisthokonts as previously thought. There is growing support that amoebozoans are the closest relatives of opisthokonts (e.g., Amaral Zettler et al, 2001; Baldauf, 2003), and that microsporidia are the closest relatives of fungi.

Recently, there has been an effort to summarize these and other aspects of the eukaryote tree in the form of a five-group arrangement: plants, unikonts, chromalveolates, rhizarians, and excavates (Keeling, 2004; Keeling et al., 2005). Under this scheme, Plantae is defined by the presence of plastids acquired by primary endosymbiosis and includes the land plants, charophytes, chlorophytes, rhodophytes, and glaucophytes. The unikonts are defined by the presence of a single cilium-bearing centriole and include the opisthokonts (animals, fungi, choanoflagellates, ichthyosporeans, and nuclearids) and the amoebozoans. Rhizaria includes the cercozoans, foraminiferans, polycistines, and acanthareans. Chromalveolates include the alveolates (e.g., ciliates, apicomplexans, and dinoflagellates) and stramenopiles (e.g., brown algae, diatoms, haptophytes, and cryptomonads). The excavates include the discicristates (e.g., euglenids and kinetoplastids), oxymonads, and metamonads (e.g., diplomonads, parabasalids, and Carpediemonas); the content of the excavates and relationships among the included taxa are particularly controversial (see below). The above arrangement also agrees with a division of eukaryotes into unikonts and bikonts (Richards and Cavalier-Smith, 2005). The amount of evidence supporting inclusion of different taxa varies considerably, from hundreds of genes in some cases (e.g., animals joining with fungi) to relatively small amounts of morphological or molecular data in other cases (e.g., excavates).

As a point of discussion, this five-group arrangement serves a useful purpose. However, a major problem is that it avoids the question of the root. Technically, without a root there can be no evolutionary polarity or claim of monophyly (i.e., no "five groups"). In fact, analyses of the largest sequence data sets, using complex models of evolution, show with statistical significance that the root lies within one of these five groups, the excavates (Hedges et al., 2001; Bapteste et al., 2002; Hedges et al., 2004), which breaks up the monophyly of the bikonts. This supports the earlier proposal of a basal position for Giardia (a diplomonad), based on rRNA sequences and cytological arguments (Sogin et al, 1989). Because complex models of evolution have been used in these recent studies, there is no clear evidence yet that long-branch attraction or other substitutional biases are responsible for this root position.

One analysis demonstrated sensitivity of the topology to removal of fast-evolving sites, but those results were inconclusive because of a lack of statistical support for most nodes (Arisue et al, 2005). With the great age of these lineages and the fact that long-branch attraction and other substitutional biases may lead to incorrect groupings (Philippe et al, 2000), it is worth being cautious in interpreting any results, even if statistically significant. Future analyses of large numbers of taxa and genes, and testing of hypotheses concerning substitutional biases, should help better resolve the tree of eukaryotes. However, the weight of the current sequence evidence argues, significantly, against the monophyly of at least one of the five groups, the "excavates," and in favour of a root between the metamonads (or at least the diplomonads) and other eukaryotes (Fig. 2A). Thus, a six-group classification would divide the excavates into the discicristates and metamonads.

Another possible location of the eukaryote root, between opisthokonts and all other eukaryotes, has been proposed based on a gene fusion event, joining dihydrofolate reductase and thymidylate synthase (Philippe et al., 2000; Stechmann and Cavalier-Smith, 2002) in many bikont eukaryotes. Evidence of a fusion of three genes in the pyrimidine biosynthetic pathway in unikonts led those same authors to revise their rooting scheme to include amoebozoans with opisthokonts (i.e., all unikonts) in the root (Stechmann and Cavalier-Smith, 2003). However, the subsequent finding of that triple gene fusion in a red alga (Matsuzaki et al., 2004), which is clearly not related to opisthokonts or amoebozoans, undermined the usefulness of that gene fusion character.

Figure 2. Two alternative hypotheses for the phylogenetic tree of eukaryotes. (A). The metamonad root, reflecting a six-group classification. This tree is favoured by phylogenetic analyses of DNA sequence data. (B). The unikont root, reflecting a five-group classification (metamonads and discicristates are combined into "excavates"). Under unikont rooting, the non-unikont eukaryotes (bikonts) are monophyletic.

Even more recently, evidence from the gene structure of myosin genes has been marshalled to further support a root between unikonts and bikonts (Richards and Cavalier-Smith, 2005) (Fig. 2B). In this case, gene and domain evolution is complex and there is homoplasy among the data. Moreover, most of the characters proposed as support for the root actually

Figure 2. Two alternative hypotheses for the phylogenetic tree of eukaryotes. (A). The metamonad root, reflecting a six-group classification. This tree is favoured by phylogenetic analyses of DNA sequence data. (B). The unikont root, reflecting a five-group classification (metamonads and discicristates are combined into "excavates"). Under unikont rooting, the non-unikont eukaryotes (bikonts) are monophyletic.

origin of eukaryotes origin of eukaryotes support the largely uncontroversial grouping of animals, fungi, and amoebozoans (unikonts), which does not define the root position. Both of the two characters indicated as supporting the alternative branch ("bikonts"), which is critical for the claim of unikont rooting, are problematic. One character involves the two-gene fusion, but this turns out to be absent in species that are critical to defining the root (e.g., metamonads), and the clear case of homoplasy involving the triple-gene fusion shows that gene fusions in general are not necessarily reliable characters. The second character is an insertion of 60 amino acids in bikonts. However, only two of the 13 bikonts examined (Trypanosoma and Phytophthora) had this insertion and neither was a metamonad. Furthermore, a recent study (Hampl et al, 2005) claimed to recover excavate monophyly but close scrutiny shows that those authors fixed the root to unikonts and therefore they did not actually test excavate monophyly with a prokaryote outgroup.

Time estimation of protist evolution has lagged behind that of other groups largely because of the complexity of relationships and slower accumulation of sequence data. Recently, a sequence analysis of the phylogeny and divergence times of eukaryotes, including the major groups of protists, was made using 22-188 proteins per node (Hedges et al, 2004), (Fig. 3). Divergence times were estimated using both global and local (including Bayesian) clock methods, and the genes were analysed separately and as a single "supergene." The diplomonad Giardia was found to be basal to the plant-animal-fungi clade, with significant bootstrap support, in Bayesian, likelihood, and distance analyses of 39 proteins (Hedges et al,

2004). Two other protists lineages, the euglenozoans (105 proteins; 38,492 amino acids) and alveolates (73 proteins; 27,497 amino acids) were also found to be basal to the plant-animal-fungi clade with significant support.

In a separate phylogenetic analysis (Bapteste et al, 2002) of a similar amount of sequence data (123 proteins, 25,023 amino acids; albeit with some missing sequences) and with a greater number of taxa, the same higher-level structure of the "protist tree" was found. In both cases, the results contradict the "five-group" classification of protists (Keeling, 2004; Keeling et al,

2005) and opisthokont or unikont rooting of eukaryotes (Stechmann and Cavalier-Smith, 2002, 2003; Richards and Cavalier-Smith, 2005). A solution to this problem with the five-group classification is to separate the discicristates from a restricted excavate group (metamonads, and possibly oxymonads and malawimonads if future studies show them to be related). Therefore, if either the unikonts or the metamonads form the root of eukaryotes (Fig. 2), it does not contradict this six-group classification scheme. It is also possible that the root of eukaryotes is at yet another position, such as (for example) between the discicristates (e.g., euglenids) and all other eukaryotes.

The timetree (phylogeny scaled to evolutionary time) of eukaryotes shows that plants diverged from the animal-fungi clade approximately 1600 Ma and that animals diverged from fungi approximately 1500 Ma (Fig. 3), reflecting a relative consistency in these time estimates found in studies using large numbers of genes (Wang et al, 1999; Hedges et al, 2004; Blair et al, 2005). In this timetree, the divergence of red algae (Rhodophyta) from the land plant lineage was approximately 1400 Ma, which is consistent with the date (1200 Ma) for the first fossils of red algae (Butterfield, 2000). Although plastids were obtained by some clades of protists through secondary endosymbiotic events, they arose initially on the plant lineage through primary endosymbiosis between a protist and a cyanobacterium. The date of that event is constrained to approximately 1500-1600 Ma (Fig. 3). The alveolates and euglenozoans branch more basally (~1900 Ma) in the timetree of eukaryotes, while the most basal branch (diplomonads) is dated to ~2300 Ma. A separate analysis of genes involved in the mitochondrial symbiotic event dated that event as 1840 ± 200 Ma (Hedges et al, 2001), and together with these data (Hedges et al., 2004) suggest a date of ~1800-2300 Ma for the origin of mitochondria.

The time estimates in this timetree (Fig. 3) compare closely with those in an analysis that focused on divergences among algae (Yoon et al, 2004). In that study, DNA and protein sequences of several plastid genes were analyzed with a local clock method (rate smoothing) and the primary plastid endosymbiotic event was found to be "before 1558 Ma." The split of red algae from green algae was found to be 1474 Ma, also comparing closely with that found in the other study, 1428 Ma (Hedges et al, 2004), despite different genes and methods. In some earlier timing studies (Feng et al, 1997; Nei et al., 2001), sequences from different clades of protists were combined and therefore the results are not comparable, although time estimates for their hybrid protist lineages, ~ 1500-1700 Ma, are similar in general to those here (Fig. 3).

However, a recent time estimation study using a relatively large data set (129 proteins and 36 taxa) obtained younger dates, with the most basal branches among eukaryotes (in this case, between opisthokonts and all other eukaryotes) splitting only 950-1259 Ma (Douzery et al., 2004). In particular, the split of red from green algae was dated as 928 (825-1061) Ma, which is only about 60% as old as the date obtained in those two other studies (Hedges et al., 2004; Yoon et al., 2004) and which directly conflicts with the oldest fossil of red algae at 1200 Ma (Butterfield, 2000). Their dates for the splits between green algae and land plants (729 Ma) and stramenopiles and alveolates (872 Ma) were also younger than the earliest fossils of those groups (e.g., green algae and stramenopiles), 1000 Ma (Woods et al., 1998; Kumar, 2001). Douzery et al. (2004) explained the conflict by attributing uncertainty (723-1267 Ma) to the geologic dating of the red algal fossil, and citing that same reference (Butterfield, 2000). However, this is incorrect because the paleontological reference (Butterfield, 2000) instead lists a date of 1198 ± 24 Ma for the fossil and claims that it is a refinement of an earlier interval spanning 723-1267 Ma.

Divergence Plants From Algae

Figure 3. A timetree of eukaryotes based on several molecular studies. Divergence times of deuterostome animals are from Blair and Hedges (2005b), those of arthropods are from Pisani et al. (2004), the divergence time of chytrid fungi from higher fungi is from Heckman et al. (2001), that of glaucophyte algae from rhodophytes + chlorophytes (and their terrestrial descendants) is from Yoon et al. (2004), and the position of choanoflagellates and amoebozoans is constrained by phylogeny. Other divergence times, including those of algae, fungi, plants, other animals, and other protists, are from Hedges et al. (2004). The time of origin of the mitochondrion, and its debated position, is discussed elsewhere (Hedges et al., 2001; Hedges et al., 2004). The time of origin of the plastid is constrained at the base of the plastid-bearing clade (Hedges et al, 2004). Gray horizontal bars are 95% confidence intervals.

Figure 3. A timetree of eukaryotes based on several molecular studies. Divergence times of deuterostome animals are from Blair and Hedges (2005b), those of arthropods are from Pisani et al. (2004), the divergence time of chytrid fungi from higher fungi is from Heckman et al. (2001), that of glaucophyte algae from rhodophytes + chlorophytes (and their terrestrial descendants) is from Yoon et al. (2004), and the position of choanoflagellates and amoebozoans is constrained by phylogeny. Other divergence times, including those of algae, fungi, plants, other animals, and other protists, are from Hedges et al. (2004). The time of origin of the mitochondrion, and its debated position, is discussed elsewhere (Hedges et al., 2001; Hedges et al., 2004). The time of origin of the plastid is constrained at the base of the plastid-bearing clade (Hedges et al, 2004). Gray horizontal bars are 95% confidence intervals.

One possible reason as to why the red algae time estimate of Douzery et al. (2004) conflicts with the fossil date is because they rooted their tree to a unikont (amoebozoan). Although they considered such a rooting to be correct, and a kinetoplastid rooting to be a "reconstruction artefact," they nonetheless calculated divergence times with the latter rooting for comparison. In doing so, they obtained an older date (899-1191 Ma, 95% credibility interval) for the chlorophyte-rhodophyte split. Nonetheless, even that estimate nearly conflicts with the fossil record (1200 Ma) and is 3040% younger than the dates obtained by others (Hedges et al., 2004; Yoon et al., 2004) for this split.

An additional explanation for the young dates in that analysis (Douzery et al, 2004) is that the calibrations used were applied incorrectly. For calibrations, they used minimum and maximum constraints based on the upper and lower time boundaries of the geologic periods containing the earliest fossils of a lineage. There are at least two problems with that approach. First, the geologic periods used were more inclusive than documented for the fossils. For example, they used the Devonian period (354-417 Ma) for the split of mammals and actinopterygian fishes. However, the fossil data are much better constrained than that, with the earliest fossils defining that split occurring in the very earliest Devonian, or more likely, late Silurian, 425 Ma (Donoghue et al, 2003). 425 Ma is 20% older than 354 Ma, the minimum date used in the study (Douzery et al., 2004). Even if the earliest fossils were in the Devonian, their date can usually be ascertained to a much finer level (e.g., age, stage, epoch) than major geologic period, and therefore this general approach is flawed and will result in an underestimate of divergence time.

The second problem is the assignment of a maximum date (constraint) for the calibration to the maximum age of the geologic period. For the time estimation analyses, assignment of a maximum calibration constraint means that the true divergence did not happen earlier than that time. But evolutionary biologists, including palaeontologists, usually never interpret the fossil record as a literal history of life, and most would agree that the true divergences occurred earlier (in many cases, even in earlier periods) than the first fossil occurrences. This approach of assigning a maximum close to the time of the first fossil occurrence would only be valid if the conclusions of the study were that the resulting times of divergence represented minimum estimates rather than mean estimates (for more discussion of this topic, see Hedges and Kumar, 2004). However, Douzery et al. (2004) interpreted their resulting time estimates as mean (true) times of divergence and drew attention to the conflict (difference) between their time estimates and other published dates that are older. If interpreted as minimum time estimates, their estimates would not be in conflict with older time estimates. It is likely that these two problems with calibration methodology, combined with a forced unikont rooting, explain why those time estimates for protists and other eukaryotes (Douzery et al., 2004) are much younger than other published analyses (e.g., Hedges et al., 2004; Yoon et al, 2004).

Was this article helpful?

0 0

Post a comment