Animal Evolution:
Revision Notes

Compiled as a final-year zoology student at the University of Edinburgh, based on the information given in lectures.

Adaptation

Adaptations are traits favoured by evolution because they serve a specific purpose.
How to test if a trait is an adaptation: Is it complex? Is it the optimal solution to a problem? Analyse on first principles. Determine fate of deviants (thought experiments, correlation studies, genetic manipulation, historical analysis, comparison of species).
Why a trait might not be an adaptation: chance, consequence of laws of chemistry/physics, by-product of an adaptation, result of trade-offs, evolutionary time lag, result of adaptation by others, historical constraints, environmental variation.
Criticism of adaptationism: it is untested storytelling, and forgets that organisms are integrated wholes.
Is malarial fever an adaptation? Good for host? (It kills parasites.) Good for parasites? (It immobilises host. It prevents parasites killing host - unlikely, this is group selectionist.) Incidental consequence of stimulating immune system? Drug trials suggest recovery from malaria is slower if fever is too high, but result is uncertain.
Parasitic adaptations? Use of intermediate hosts; huge egg output; making intermediate host more susceptible to predation. These could, however, be accidental or 'unintentional' consequences of parasitism.

Natural selection

(Ridley 2003, Evolution; Dawkins, The Blind Watchmaker)

Selection acts if there is: variation, fitness differences, inheritance.
Selection coefficient for a genotype, s = 1 - w (where w is fitness).
Selection differential is difference in mean phenotype between selected (breeders) and unselected.
Response to selection is change in trait value after selection: R = s × h² (where h² is heritability).
If population is not at equilibrium, trait frequency distribution of offspring will differ predictably from parents'. Trait frequency distribution will also differ among age classes or life-history stages.
Experiments show that guppies are selected for crypsis where there is predation, and sexually selected for conspicuous colour spots in the absence of predators (Endler 1980).

Experimental evolution

(Elena & Lenski 2003)

Experimental evolution uses natural selection rather than artificial selection - organisms "select themselves".
How rugged are fitness landscapes? Lenski et al (1991) bred test tubes of E. coli over thousands of generations. Fitness (compared with ancestral population) increased over time, although rate of increased slowed. All populations followed a similar trajectory of fitness increase, but there was some variation. An increase in cell size occurred, paralleling the increase in fitness. More extreme divergence between populations was seen when novel environments (different sugars) were used - underlying genetics of adaptation here may be different.
Contingency - how important is evolutionary history? Populations of E. coli transferred from glucose to maltose medium evolve to a similar fitness irrespective of starting point, but this masks variation in other traits such as cell size (Travisano et al 1995). Populations of Phi6 bacteriophage may evolve to different endpoints. Conclusion: populations starting from different points may evolve to different peaks, which may or may not differ in fitness.
Selection in different environments - there may be a 'cost of adaptation' due to antagonistic pleiotropy (genes good in one environment are bad in another) or mutation accumulation (genes neutral in selected environment deteriorate). Chlamydomonas evolved in light conditions fare worse in dark conditions and vice versa, but it is possible to evolve generalists that thrive in light or dark - mutation accumulation is important (Bell & Reboud 1997). In E. coli, deterioration of ability to survive in one environment mirrors improvement in another, and 'mutator lines' (with an unnaturally high mutation rate) deteriorate at a similar rate to non-mutator lines - antagonistic pleiotropy is important.
What limits adaptation? The supply of new mutations, or the efficiency of selection? Mutator lines of E. coli adapt faster with small populations where the supply of mutations is limiting, but not with larger ones where the rate of spread of mutations is limiting (De Visser et al 1999). Larger populations of asexual Chlamydomonas adapt faster, but there are diminishing returns as population size is increased. Sex increases efficiency of selection in larger Chlamydomonas populations, but has less effect on small populations (Colegrave 2002).

Evolution of development

(Carroll et al 2001, From DNA to Diversity)

Development of an organism depends on selective gene expression.
Genes typically have many enhancers, which can act independently - modular control.
Bristles in Drosophila grow in thousands of precise positions, driven by different enhancers of the achaete/scute gene. Knocking out particular enhancers removes specific bristles.
Types of homologous gene: orthologs (in different species), paralogs (in same species).
There is a limited number of major chemical signalling pathways (around 16-25). They are highly conserved, and used for multiple functions.
"Homologous genes do not necessarily encode homologous structures and homologous structures need not be encoded by homologous genes" (de Beer 1971).
Pax6 gene controls eye development in different phyla that evolved complex eyes independently, but may have had a common ancestor with primitive photoreceptors that expressed this gene (Gehring & Ikeo 1999).
Anatomical structure and expressed genes on the ventral side of protostomes corresponds to those on the dorsal side of chordates. There may have been a dorsal/ventral inversion in the evolution of chordates, or the two body plans may derive from a common ancestor with diffuse dorsoventral organisation (Gerhart 2000).

Cambrian explosion

(Gould 1989, Wonderful Life; Conway Morris 1998, The Crucible of Creation)

What caused the Cambrian explosion? Advances in developmental genes? Environmental changes (e.g. higher atmospheric oxygen)? Ecological innovation (e.g. hard body parts for defence) and adaptive radiation?
Is it an artefact of the fossil record? Unlikely. Knowledge of the Cambrian is based on several fossil deposits. If animals were diverse in the Precambrian their tracks should have fossilised even if their soft bodies didn't.
Cambrian explosion may have been preceded by a 'Precambrian slow-burn' during which body plans diverged.
Was diversity of phyla higher in Cambrian than today? It depends on your interpretation of Cambrian animals.
This has fuelled debate over contingency vs. determinism. Convergent evolution is evidence for determinism, e.g. Anolis lizards have radiated independently into the same set of niches on four different islands (Losos et al 1998).

Hox genes

Hox genes all share a distinctive 'homeobox' region. They code for transcription factors.
Hox genes are side-by-side along a chromosome. Position of Hox genes along a chromosome corresponds to the physical position of expression along the anterior-posterior axis (colinearity). None are expressed in the anterior-most part of an animal. They are turned on early in development and remain switched on.

Hox genes have been found, in some form, in all advanced animal phyla...

Drosophila has 8 Hox genes (other arthropods have additional ones, some of which appear to have evolved into non-Hox genes in Drosophila). ‘Ultrabithorax’ mutation is due to failed expression of a Hox gene, causing third thoracic segment to develop like second one (with proper wings instead of halteres).
Annelids have a similar Hox gene family.
Chordates have an extended 13-gene Hox family, duplicated into four sets.
Echinoderms have Hox genes but their expression in the non-bilaterally symmetrical adults is hard to interpret.
C. elegans has only 4 recognisable Hox genes, and their function is relatively trivial, but this may not be true in other nematodes.
Molluscs have Hox genes, and there is some colinearity in their expression (most evident in nerve ganglia). Squid arms express different combinations of Hox genes, but not colinearly (Hox genes adopting new functions?).

Sexual arms races

Bean weevils: male has spiky penis to damage female, increasing time she takes to mate again; female fights back by trying to throw off male early during copulation (Crudgington & Siva-Jothy 2000).
Drosophila: males inject toxins during mating that kill previous males' sperm, but also selfishly harm females. Exposure to a particular Accessory Gland Protein (AGP) in seminal fluid reduces female life expectancy: mating with castrated males doesn't harm female, but mating with sperm-less males does (Chapman et al 1995). AGP genes appear to be under strong selection - high proportion of non-synonymous mutations (Swanson et al 2001). Males become less 'nasty' if evolved under artificially monogamous conditions (Holland & Rice 1999).

Molecular techniques

(Griffiths et al 1996, An Introduction to Genetic Analysis)

Southern blotting is used to locate a known sequence of DNA on a gel. It involves electrophoresis, blotting of DNA onto filter paper, denaturing to produce single-stranded DNA, and combining with a radioactive probe.
Polymerase Chain Reaction (PCR) makes many copies of a DNA strand. DNA is heated to separate strands, two primers (at either end of the DNA region of interest) are added, solution is annealed by cooling, bases and enzyme (thermostable Taq polymerase from thermophilic bacteria) are added, and cycle is repeated many times.
Dideoxy (Sanger) sequencing uses modified bases that halt replication when they become incorporated into a growing DNA strand. The bases can be linked to fluorescent dyes that show up as coloured bands on gel.
Restriction Fragment Length Polymorphisms (RFLPs), used in DNA fingerprinting, can be identified by cutting with restriction enzymes and Southern blotting. Only detects insertions/deletions or mutations at restriction sites.
DNA fingerprinting analyses variation in multi-locus mini-satellite DNA. Relatedness is measured by the band-sharing coefficient (similarity index); the average BSC of unrelated individuals equals the frequency of each band in population. Satellite DNA consists of highly repeated regions believed to function-less and selectively neutral.
DNA profiling, which uses single-locus micro-satellite DNA regions amplified by PCR, has largely superseded DNA fingerprinting. Probability of random match at one locus is 1 - (1-q)² where q is frequency of alleles (rarer ones are thus more diagnostic). Exclusion probability for multiple loci is product of individual probabilities.
Molecular techniques have discovered that: extra-pair fertilisations but not egg dumping occur in indigo buntings (Westneat 1987), the contribution that a beta male dunnock makes to rearing chicks depends on his chances of paternity (Burke et al 1989), dominant meerkats father most of a group's offspring but subordinate immigrants also father a few (Griffin et al 2003), polar bears have genetically-distinct populations (Paetkau et al 1999).

Phylogenetic trees

(Hillis et al 1996, Molecular Systematics)

Cladograms simply show relationships; phylograms also show distance. Trees may be rooted or unrooted.
Unresolved trees have nodes giving rise to more than two branches (polytomy).
Taxa can be monophyletic, paraphyletic or polyphyletic.
Characteristics of phylogenetic methods: efficiency ("doability" given computer limitations), power (amount of data needed), consistency (will it converge on the "right" solution given sufficient data?), robustness, falsifiability.
Measuring phylogenetic difference: nucleotide substitution models (taking into account different probabilities of different types of substitution - transition vs. transversion, base composition bias), distance correction (taking into account multiple substitutions), aligning sequences (may involve a scoring system).
Clustering methods of constructing trees: neighbour joining (link closest individuals first, then next closest, etc).
Optimality methods: maximum likelihood, maximum parsimony (minimising putative evolutionary changes).
Different sites evolve at different rates, providing a variety of molecular clocks suitable for different purpose.
Search strategies: algorithmic (stepwise addition, star decomposition), exhaustive (not computationally feasible for large trees), heuristic (branch swapping and binary evaluation, "walking in tree space").
Problems: long branches attract (the chance of homoplasy - parallel or convergent substitution - increases with genetic distance), evolutionary noise may obscure phylogenetic signal.
Assessing support for trees: bootstrap (resample dataset to produce another with same number of characters and reassess), jack-knife (sub-sample dataset using fewer taxa - quicker but less meaningful).
Example of the use of trees: discovery that African forest and savannah elephants are distinct groups.

The history of life

Geological strata give an order of events but no absolute timescale; radioisotopes give absolute dates.
A 'molecular clock' dates divergence between extant species. A neutral sequence evolves at the mutation rate (around 10^-8 bases per year). Over very long timescales, 'saturation' will occur (sequences will no longer be recognisably similar at all), but conserved sequences (e.g. rRNA) provide an empirical clock.
Divergence of related genes (e.g. catalytic and non-catalytic subunits of ATPase) allows rooting of the tree of life.
Events in the history of life: fossil stromatolites (~3500 Ma), cyanobacterial markers (~2800 Ma, diversifying ~2100 Ma), increase in oxygen (~2000 Ma), eukaryote fossils (>1200 Mya), 'Cambrian explosion' (~550 Ma).
Major transitions: replicating molecules to populations of molecules in compartments; unlinked replicators to chromosomes; RNA to DNA and proteins; prokaryotes to eukaryotes; asexual to sexual populations; unicellular to multicellular organisms; solitary individuals to social colonies; primate societies to human societies with language. These events changed the way hereditary information is encoded and involved the coming together of components into an interdependent whole. (Maynard Smith & Szathmáry 1995, Major transitions in evolution)
Organisms benefited from an increasing division of labour: DNA and protein as well as RNA, evolution of specialised enzymes following gene duplication, organelles in eukaryotic cells, separate sexes.
Why did selfish selection on individuals not disrupt the whole? Kin selection, and ‘contingent irreversibility’.

The RNA world

The Last Universal Common Ancestor of life on Earth (LUCA) had DNA making RNA making protein, metabolism involving glycolysis and ATP, and lipid-based cell membranes.
Before that, there may have been an RNA world in which RNA acted as both gene and enzyme.
RNA is still an essential part of ribosomes, mRNA processing, DNA replication, etc.
Nucleotides are used as signals (e.g. cAMP), energy carriers (ATP), cofactors (NAD), etc.
Ribozymes - RNA-based enzymes (Cech 1982) - include self-splicing ribosomal RNA in the protist Tetrahymena, the enzyme RNase P (in which the catalytic centre is RNA-based), ribosomes.

Efficient ribozymes can be selected for in vitro...

Lehman & Joyce (1993) evolved a version of the Tetrahymena ribozyme that uses Ca²⁺ rather than Mg²⁺ as a cofactor. This ribozyme becomes attached to a part of its substrate, allowing molecules that work as catalysts to be separated from those that don't. RNA was multiplied by converting it to DNA and using PCR.
Bauer et al (1989) evolved an RNA sequence de novo that replicates efficiently in the presence of Q-beta replicase (a viral enzyme). The spread of these RNA molecules through a gel follows population genetics equations developed by Fisher. Their spread occasionally accelerated when a new beneficial mutation occurred.
Johnson et al (2001) evolved an RNA polymerase ribozyme, which originally added a few bases to itself, into one that could add up to 14 bases to an external RNA template with 97% accuracy.

However, the simple RNA-copying ribozymes known cannot handle large enough sequences, and are not accurate enough, to allow reliable coding for long RNA molecules. Maybe an even simpler system predated the RNA world?

Evolution of the genetic code

In the size of the genetic alphabet, there is a trade-off between simplicity and versatility - the present 4-base alphabet represents an optimum (Szathmáry 2003).
RNA is translated into protein using a fairly arbitrary genetic code. This may have arisen from an ancestral code that underwent "codon shuffling" or that gradually became less ambiguous.
There have been occasional slight modifications to the genetic code (e.g. in mitochondria). How did these occur without messing up an organism? 'Stop' codons can be reassigned to amino acids without badly affecting protein functionality. If there is biased codon usage (e.g. pressure to use G/C rather than A/T because the former is more thermostable), some codons might be rarely used and could then be reassigned.
The present code does have some regularities: the 1st codon position is associated with four 'biosynthetic families' of amino acids; the 2nd with polarity of amino acids; the 3rd is usually redundant (this minimises translation errors since there is a 'wobble' in the translation machinery that can cause the final base in a codon to be misread).
Amino acids may originally have been cofactors for ribozymes in the RNA world. A system might have developed in which amino acids were 'tagged' with short RNA molecules (the ancestors of transfer RNAs) for easy attachment to other RNAs. Enzymes linking amino acids together may then have developed to stabilise ribozyme/amino acid complexes; these enzymes later evolved into protein-manufacturing ribosomes.

More notes and essays

Animal Evolution: Revision Notes