Animal Diversity: Systematics, Classification and Phylogenetic Relationships
2. Classical characters: organisation, symmetry, germ layers and coelom
Classical classification sorts animals by a few deep features of the body plan, read from adult anatomy and embryology. They are informative because they are complex, slow to change and tied to development; their weakness, as Section 6 shows, is that some evolved or were lost more than once.
| Character | States | Examples |
|---|---|---|
| Level of organisation | Cellular (no true tissues); tissue; organ-system | Porifera; Cnidaria and Ctenophora; Bilateria |
| Symmetry | Asymmetric; radial; biradial; bilateral; pentaradial in adults only | Most sponges; cnidarians; ctenophores; Bilateria; echinoderms (larvae bilateral) |
| Germ layers | Diploblastic (ectoderm and endoderm, jelly-like mesoglea between); triploblastic (mesoderm added) | Cnidaria, Ctenophora; Bilateria |
| Body cavity | Acoelomate; pseudocoelomate (cavity from the blastocoel, mesoderm only on the body-wall side); eucoelomate (cavity enclosed by mesoderm) | Platyhelminthes; Nematoda, Rotifera; annelids, molluscs, arthropods, echinoderms, chordates |
| Origin of coelom | Schizocoely (split within mesodermal bands); enterocoely (pouches budded from the archenteron) | Annelids, molluscs; echinoderms, hemichordates, amphioxus |
| Segmentation | Metamerism: serial repetition of body units | Annelida, Arthropoda, Chordata (somites) |
Embryology split the triploblasts into two lines. In protostomes the blastopore typically becomes the mouth, cleavage is spiral and determinate (each early blastomere has a fixed fate) and the coelom arises by schizocoely; many have a trochophore larva. In deuterostomes the blastopore becomes the anus and the mouth forms anew, cleavage is radial and indeterminate (an isolated early blastomere can still form a whole larva) and the coelom arises by enterocoely.
3. The major phyla and the chordate plan
| Phylum | Grade | Diagnostic features |
|---|---|---|
| Porifera | Cellular; no tissues; mostly asymmetric | Choanocytes drive water in through ostia and out through the osculum; skeleton of spicules or spongin |
| Cnidaria | Tissue; diploblastic; radial | Cnidocytes with nematocysts; gastrovascular cavity with one opening; polyp and medusa forms |
| Ctenophora | Tissue; diploblastic; biradial | Eight rows of ciliary comb plates; adhesive colloblasts |
| Platyhelminthes | Triploblastic; acoelomate | Dorsoventrally flattened; flame cells (protonephridia); gut blind or absent |
| Nematoda | Pseudocoelomate | Round in section; cuticle moulted; only longitudinal body-wall muscles; complete gut |
| Annelida | Coelomate; metameric | Segments separated by septa; setae or parapodia; nephridia; closed circulation in earthworms |
| Mollusca | Coelomate; coelom reduced to the pericardial cavity | Mantle, muscular foot, radula (absent in bivalves); open circulation except in cephalopods |
| Arthropoda | Coelomate; metameric | Jointed appendages; chitinous exoskeleton that is moulted; haemocoel and open circulation |
| Echinodermata | Coelomate deuterostome | Water vascular system with tube feet; calcareous ossicles of mesodermal origin; adults pentaradial |
| Hemichordata | Coelomate deuterostome | Proboscis, collar and trunk; pharyngeal slits; a stomochord, not a notochord |
| Chordata | Coelomate deuterostome | Notochord, dorsal hollow nerve cord, pharyngeal slits, post-anal tail, endostyle or thyroid |
The chordate characters need appear only in the embryo. In Urochordata (tunicates) the notochord lies in the tail of the tadpole larva and is lost at metamorphosis; in Cephalochordata (amphioxus) it runs from head to tail throughout life; in Vertebrata it is largely replaced by the vertebral column. Vertebrates comprise the jawless lampreys and hagfishes and the jawed gnathostomes: cartilaginous fishes, bony fishes and tetrapods. Among tetrapods the amniotes (reptiles, birds, mammals) share an egg with amnion, chorion and allantois, a synapomorphy that freed reproduction from water.
4. Reading characters: homology, polarity and monophyly
A phylogeny is inferred from homologous characters, similarities inherited from a common ancestor (the forelimb bones of a bat, a whale and a human). Homoplasy is similarity not due to common descent: convergence (wings of birds and insects, camera eyes of octopus and vertebrates), parallelism, or reversal (limb loss in snakes). Character polarity, which state is ancestral, is set by outgroup comparison: the state shared with a related group outside the study group is taken as ancestral.
- Plesiomorphy: an ancestral state. Apomorphy: a derived state.
- Synapomorphy: a derived state shared by two or more taxa, the only evidence for a clade (hair and mammary glands for Mammalia).
- Symplesiomorphy: a shared ancestral state, which groups nothing (a backbone shared by fishes and tetrapods does not make “fishes” a clade).
- Autapomorphy: a derived state unique to one terminal taxon; it diagnoses that taxon but says nothing about its relatives.
A monophyletic group (clade) contains an ancestor and all its descendants: Mammalia, or Archosauria (crocodilians plus birds). A paraphyletic group contains an ancestor and only some descendants: “reptiles” without birds, “fishes” without tetrapods, “invertebrates”. A polyphyletic group gathers lineages from different ancestors on a convergent character: “warm-blooded animals” (birds plus mammals). Two taxa are sister groups if they share an immediate common ancestor. A tree is read by its nodes, not by the left-to-right order of tips, so rotating branches about a node changes nothing. In a cladogram branch lengths mean nothing; in a phylogram they are proportional to change; in a chronogram, to time. A rooted tree shows the direction of time; an unrooted tree shows only relationships.
5. Molecular tools and tree-building methods
Molecular characters are numerous, heritable, comparable across distant groups and countable. The marker is matched to the depth of the question. Slowly evolving nuclear 18S rRNA, and now hundreds of genes or whole genomes (phylogenomics), resolve splits among phyla; fast mitochondrial genes (cytochrome b, 12S and 16S rRNA) resolve species and populations. DNA barcoding identifies animal species from a standard region of about 650 bp of mitochondrial cytochrome c oxidase subunit I (COI). Older molecular tools were protein electrophoresis, immunological distance and DNA–DNA hybridisation. Rare genomic changes, such as retroposon insertions or rearrangements of mitochondrial gene order, are nearly free of homoplasy and make strong characters.
| Method | Principle | Strength and limitation |
|---|---|---|
| UPGMA (distance) | Repeatedly joins the closest pair and averages distances | Assumes equal rates in all lineages (a clock) and gives a rooted, ultrametric tree; misleads when rates differ |
| Neighbour-joining (distance) | Stepwise joining that corrects for unequal rates (minimum evolution) | Fast and tolerant of rate variation; unrooted tree; sequence detail lost in the distances |
| Maximum parsimony (characters) | The tree needing the fewest character changes | No explicit model; prone to long-branch attraction |
| Maximum likelihood (characters) | The tree and branch lengths that make the data most probable under a substitution model | Statistically consistent if the model is adequate; computationally heavy |
| Bayesian inference (characters) | Posterior probability of trees given data, model and priors, sampled by MCMC | Gives posterior probabilities of clades; sensitive to model and priors |
The simplest distance is the p-distance, the proportion of aligned sites that differ. Because a site can change more than once (multiple hits), p underestimates the true number of substitutions and saturates as divergence grows. The Jukes–Cantor correction, which assumes all substitutions equally likely, gives d = −(3/4) ln(1 − 4p/3) substitutions per site. Worked example: two 200-site sequences differ at 50 sites, so p = 0.25 and d = −0.75 ln(0.667) = 0.304, about a fifth more than p. Support for a clade is measured by the bootstrap: sites are resampled with replacement to build many pseudo-datasets, and the percentage of the resulting trees that contain the clade is its bootstrap value.
6. Classical versus molecular: the modern animal tree
Molecular phylogenies, first from 18S rRNA and now from genome-scale data, upheld Bilateria, Protostomia and Deuterostomia but redrew the protostomes into two clades. Ecdysozoa are animals that moult a cuticle: arthropods, nematodes, tardigrades and onychophorans. Lophotrochozoa (within Spiralia) include annelids, molluscs, flatworms, brachiopods, bryozoans and rotifers, many with spiral cleavage, a trochophore larva or a lophophore. Deuterostomia comprise Ambulacraria (echinoderms plus hemichordates) and Chordata. Whether sponges or ctenophores are the sister group of all other animals is still debated.
| Classical grouping | Classical basis | Molecular verdict |
|---|---|---|
| Acoelomate flatworms at the base of Bilateria | No coelom, simple organisation | Flatworms are lophotrochozoans; their simplicity is partly secondary |
| Pseudocoelomata as one group | Cavity derived from the blastocoel | Not a clade: nematodes are ecdysozoans, rotifers lophotrochozoans |
| Articulata (annelids plus arthropods) | Metameric segmentation | Rejected: segmentation arose more than once or was lost in many lines |
| Cephalochordates as sister of vertebrates | Full-length notochord, segmented muscles | Tunicates are the sister group of vertebrates (Olfactores) |
| Protostomia and Deuterostomia | Fate of blastopore, cleavage pattern | Both upheld as clades |
7. Numericals: counting trees and scoring parsimony
How many trees? Each new taxon can be attached to any branch of an existing tree, and an unrooted bifurcating tree of n taxa has 2n − 3 branches. Hence the number of unrooted trees is (2n − 5)!! = 1 × 3 × 5 × … × (2n − 5) and of rooted trees (2n − 3)!!, so the rooted count for n taxa equals the unrooted count for n + 1. For 4 taxa there are 3 unrooted and 15 rooted trees; for 5, 15 and 105; for 10, about 2.0 million and 34.5 million. Exhaustive search is impossible beyond about a dozen taxa, so programs search heuristically.
Scoring parsimony (Fitch). For one site, work from the tips inwards: a node takes the intersection of its two daughters' state sets if it is not empty; otherwise it takes their union and one step is counted. Tree length is the sum of steps over all sites, and the shortest tree is the most parsimonious. A site is parsimony-informative only if at least two states each occur in at least two taxa. Worked example for taxa W, X, Y and Z (bases listed in that order):
| Site (W X Y Z) | W+X | Y+Z | W+Y | X+Z | W+Z | X+Y |
|---|---|---|---|
| 1: AAAA | 0 | 0 | 0 |
| 2: GGCC | 1 | 2 | 2 |
| 3: CTTT | 1 | 1 | 1 |
| 4: TTCC | 1 | 2 | 2 |
| 5: AGGA | 2 | 2 | 1 |
| 6: AAGG | 1 | 2 | 2 |
| Tree length | 6 | 9 | 8 |
Key takeaways
- Systematics reconstructs relationships; taxonomy describes, names and classifies. The ICZN starts from 1758, applies priority, allows tautonyms and fixes the endings -oidea, -idae and -inae; every name hangs on a type (holotype, lectotype, neotype).
- Classical characters are level of organisation, symmetry, germ layers, body cavity and its origin, segmentation and protostome–deuterostome embryology; they are tendencies with known exceptions.
- Chordates share notochord, dorsal hollow nerve cord, pharyngeal slits and post-anal tail at least in the embryo; the hemichordate stomochord is not a notochord, and tunicates are the sister group of vertebrates.
- Only synapomorphies define clades; symplesiomorphies and homoplasies mislead. Reptiles without birds and fishes without tetrapods are paraphyletic; warm-blooded animals are polyphyletic.
- Molecular trees split protostomes into Ecdysozoa and Lophotrochozoa, dissolving Articulata and Pseudocoelomata. UPGMA assumes a clock and neighbour-joining does not; parsimony minimises changes and suffers long-branch attraction; likelihood and Bayesian methods use substitution models.
- Unrooted trees number (2n − 5)!! and rooted trees (2n − 3)!!; tree length is the sum of Fitch steps; the Jukes–Cantor distance d = −(3/4) ln(1 − 4p/3) corrects the p-distance for multiple hits.
Practice questions (17)
Attempt each one before opening the answer. Every explanation names the tempting wrong option as well as the right one, because that is where marks are lost.
A zoologist erects a new taxon at the rank of family, based on the type genus Testudo. Under the International Code of Zoological Nomenclature, which name has the correct form for a family?
Show answer
Answer: C — Testudinidae
Zoological family names end in -idae, so the family is Testudinidae. Testudinoidea carries the superfamily ending -oidea and Testudininae the subfamily ending -inae; both are tempting because they are valid zoological endings, but for other ranks. The ending -aceae marks a family in botanical nomenclature, not in zoology.The only name-bearing specimen of a moth species, designated in the original description, was destroyed in a museum fire, and no other specimen of the original type series survives. A specialist designates a newly collected specimen as the name-bearer. What is this specimen called?
Show answer
Answer: B — A neotype
A neotype is designated when the name-bearing type has been lost or destroyed. A lectotype is the tempting answer, but it is chosen from surviving syntypes of the original series, and none survive here. Paratypes and syntypes are specimens of the original type series, so a newly collected specimen cannot be either.A bilaterally symmetrical, triploblastic animal has a complete gut, a body cavity formed from the blastocoel with mesoderm lining only the body wall, only longitudinal muscles in its body wall, and a cuticle that it moults as it grows. To which phylum does it belong?
Show answer
Answer: D — Nematoda
A blastocoel-derived pseudocoelom, longitudinal muscles only and a moulted cuticle together identify Nematoda. Arthropoda is the tempting choice because arthropods also moult (both are ecdysozoans), but arthropods are coelomates with a haemocoel and jointed appendages. Annelids have a true coelom with circular and longitudinal muscles, and flatworms are acoelomate with a blind gut.In the classical embryological scheme, which of the following are characteristic of deuterostomes as opposed to protostomes?
Show answer
Answer: A — The blastopore becomes the anus and the mouth forms secondarily; B — Early cleavage is radial and indeterminate
Deuterostomes (echinoderms, hemichordates, chordates) classically show the blastopore becoming the anus and radial, indeterminate cleavage. Schizocoely and the trochophore larva are protostome features, typical of annelids and molluscs.Blastomeres are separated at the four-cell stage in embryos of two marine invertebrates. In species X each isolated blastomere forms a complete though small larva; in species Y each forms only part of a larva, with specific structures missing. Which interpretation is most consistent with these results?
Show answer
Answer: A — X has radial, indeterminate cleavage as in a sea urchin, and Y has spiral, determinate cleavage as in a marine snail
Indeterminate cleavage means early blastomeres keep their full developmental potential, so each can form a whole larva, as in sea urchins (deuterostomes). The determinate spiral cleavage of protostomes such as snails fixes each blastomere's fate early, so an isolated cell gives only its allotted part. The reversed pairing is the tempting distractor; the mode of coelom formation and a pseudocoelom have nothing to do with the potency of early blastomeres.Which structure is correctly matched with the phylum in which it is characteristically found?
Show answer
Answer: D — Water vascular system — Echinodermata
The water vascular system with tube feet is unique to echinoderms. Flame cells are the protonephridia of flatworms (annelids have metanephridia), colloblasts belong to ctenophores (cnidarians have cnidocytes), and the radula is the feeding organ of molluscs.Which of the following groups is monophyletic?
Show answer
Answer: D — Archosauria: crocodilians, birds and all other descendants of their latest common ancestor
Archosauria contains a common ancestor and all its descendants, so it is a clade. Traditional reptiles and fishes are paraphyletic because they leave out some descendants (birds; tetrapods). Grouping birds with mammals by endothermy is polyphyletic: endothermy evolved independently in the two lineages, so the group rests on a convergent character.A study of relationships among five species of the cat family uses a species of the dog family as outgroup. One character scored is “presence of mammary glands”, which the dog and all five cats show. For resolving relationships among the five cats, this character is best described as —
Show answer
Answer: B — A symplesiomorphy that is uninformative about relationships among the five cats
The state is present in the outgroup and in every ingroup species, so outgroup comparison makes it ancestral (plesiomorphic) for the cats, and a shared ancestral state groups nothing. It is a synapomorphy only at the level of Mammalia as a whole, which is the tempting misreading. It is not unique to one species, so not an autapomorphy, and mammary glands in cats and dogs are homologous, not convergent.Which of the following statements about reading phylogenetic trees are correct?
Show answer
Answer: A — Rotating the two branches that arise from a node does not change the relationships shown; B — In a phylogram, branch lengths are proportional to the amount of character change
Relationships are carried by the nodes, so swapping branches about a node leaves the tree unchanged, and phylogram branch lengths do scale with change. An unrooted tree shows no direction of descent, and living tips are not ancestors of one another anyway. Adjacency of tips on the page means nothing because branches rotate freely; sister groups are defined by an immediate common ancestor.Which marker is most suitable for resolving relationships among animal phyla that diverged more than 500 million years ago?
Show answer
Answer: C — The nuclear 18S rRNA gene, because its slow rate of change avoids saturation
Deep divergences need slowly evolving sequences; otherwise repeated substitutions at the same sites erase the signal (saturation). 18S rRNA was the marker that first defined the modern animal tree. COI is tempting because it is the standard animal barcode, but its fast rate suits species identification, not phyla; microsatellites and the control region evolve faster still and suit populations.A distance matrix is computed for four species whose lineages have evolved at very different rates, one about three times faster than the others. Which method is most likely to recover a wrong topology from this matrix, and why?
Show answer
Answer: A — UPGMA, because it assumes that all lineages evolve at the same rate and builds an ultrametric tree
UPGMA clusters by averaged distances under a strict-clock assumption, so a fast lineage looks distant from its true sister and is placed wrongly. Neighbour-joining is the tempting option, but it was designed precisely to correct for unequal rates and returns an unrooted tree. Maximum likelihood estimates a separate length for every branch, and the bootstrap measures support; it is not a tree-building method.Parsimony groups two distantly related, fast-evolving species together with strong bootstrap support. Maximum likelihood with a well-fitting substitution model, after adding slower-evolving relatives of each species, places them in separate clades. What is the most likely explanation?
Show answer
Answer: B — Long-branch attraction: parallel changes on two long branches were read by parsimony as synapomorphies
On long branches many sites change by chance to the same state; parsimony, which has no model of multiple hits, counts these parallel changes as shared derived states and joins the long branches. A model-based likelihood analysis and extra taxa that break the long branches remove the artefact. The bootstrap measures consistency within the data, not correctness, so it can strongly support an artefact, and parsimony has no correction for multiple hits at all.Which of the following conclusions are supported by molecular phylogenies of animals?
Show answer
Answer: A — Arthropods and nematodes both belong to the moulting clade Ecdysozoa; C — Tunicates, rather than cephalochordates, are the sister group of vertebrates
Ecdysozoa (arthropods, nematodes and other moulting animals) and the tunicate–vertebrate pairing are well supported by molecular data. Articulata is rejected: annelids are lophotrochozoans and arthropods ecdysozoans, so segmentation arose more than once or was lost. Flatworms nest within Lophotrochozoa, so their acoelomate condition is not evidence of a basal position.How many distinct rooted, fully bifurcating trees are possible for 6 taxa?
Numerical answer — type the value.
Show answer
Answer: 945
Rooted trees for n taxa = (2n − 3)!! = 1 × 3 × 5 × … × (2n − 3). For n = 6 this is 9!! = 1 × 3 × 5 × 7 × 9 = 945. Using the unrooted formula (2n − 5)!! gives 105, the usual slip; equivalently, the rooted count for 6 taxa equals the unrooted count for 7.Four taxa W, X, Y and Z have eight aligned sites; each string gives the bases of W, X, Y and Z in that order: site 1 AAAA, site 2 GGAA, site 3 CCTT, site 4 TCTC, site 5 AAAG, site 6 TTGG, site 7 CTTC, site 8 AGCT. Using Fitch parsimony, the length (total number of steps) of the most parsimonious unrooted tree is ____.
Numerical answer — type the value.
Show answer
Answer: 11
Steps per site on W+X | Y+Z: 0, 1, 1, 2, 1, 1, 2, 3 = 11. On W+Y | X+Z: 0, 2, 2, 1, 1, 2, 2, 3 = 13. On W+Z | X+Y: 0, 2, 2, 2, 1, 2, 1, 3 = 13. Sites 2, 3 and 6 favour W+X, site 4 favours W+Y and site 7 favours W+Z; site 5 (singleton) costs 1 and site 8 (four states) costs 3 on every tree. The shortest tree has length 11. Leaving out the uninformative sites gives 7, which is not the tree length.Two aligned mitochondrial sequences of 400 sites differ at 60 sites. Using the Jukes–Cantor correction d = −(3/4) ln(1 − 4p/3), where p is the proportion of differing sites, the estimated number of substitutions per site is ____ (to three decimal places).
Numerical answer — type the value.
Show answer
Answer: 0.167
p = 60/400 = 0.15. 4p/3 = 0.2, so 1 − 4p/3 = 0.8 and ln 0.8 = −0.2231. d = −0.75 × (−0.2231) = 0.1674, i.e. 0.167 substitutions per site. The corrected value exceeds p (0.150) because some sites have changed more than once; reporting p itself is the usual slip.For four taxa W, X, Y and Z, which aligned site (bases listed for W, X, Y and Z in that order) is parsimony-informative?
Show answer
Answer: C — AGAG
A site is informative only if at least two states each occur in at least two taxa. AGAG (W and Y share A, X and Z share G) costs 1 step on the tree pairing W with Y and 2 on the other two, so it discriminates. A constant site costs 0 everywhere, a singleton (AAAG) 1 everywhere and a site with four different bases 3 everywhere, so none of them prefers any tree.