Ninety-five Principles for Evolutionary Biology

I
When (1) individuals within populations vary, (2) part of this variation is heritable, (3) some individuals are more successful at surviving and reproducing than others, and (4) some of the variation in fitness is caused by the variation in traits - then there can be a change in the frequencies of the traits from generation to generation. This type of change is ‘evolution’. Microevolution is change in allele frequencies across generations.

II
• Selection sorts among individuals, but the population is what evolves.
• Selection acts on phenotypes, but evolution is a change in gene frequencies.
• The adaptive process is not forward looking; every step is a response to immediate conditions.
• New states can evolve beyond what was previously present.
• Correlated response to selection complicates optimization.
• Selection is non-random (and mutation is undirected and does not respond to specific 'needs').
• Evolution can change direction; it's not uni-directional progress.
• Fitness is not tautological: the realities of physics etc. cause relationships between traits and fitness.
• Selection acts among individuals, not usually for the good of the species.

III
• Point mutations are caused by mistakes in DNA replication and damages (as from radiation) that are not correctly repaired. Some point mutations change the amino acid (replacement substitutions), while others do not (synonymous substitutions). The latter are presumably nearly neutral.
• An insertion or deletion of a base pair often causes a frameshift with loss of function.
• New genes can come from gene duplications, which can happen through unequal crossing over during meiosis.
• An inversion is a kind of chromosomal alteration in which the order of genes is flipped. Inversions generally prevent crossing over so alleles are locked together as a unit. They then tend to adapt to local conditions, forming clines along environmental gradients.
• Polyploidy is when the whole genome is doubled (2n->4n). It doesn’t seem to work much in animals, but is very common in some groups of plants. It instantly makes new “biological species” that later can diverge in morphology, ecology, etc.

IV
Mutation at some rate is inevitable. Every gene has some probability of mutating (~1 phenotypically obvious mutation per 100,000 gene copies), there are lots of genes in eukaryotic genomes (e.g. 30,000 in humans), so many individuals are born with a new mutation. And the number is surely much higher if you included all neutral point mutations.

V
Genetic variation in natural populations was classically expected to be low (one allele should be better than all others and selection should keep deleterious ones at very low frequencies), but enzyme electrophoresis found lots of variation. This could be because of lots of neutral variation, or it could be selection favors different alleles at different times and/or places. Now even more variation is evident at the DNA level, again suggesting much of it is neutral.

VI
Hardy-Weinberg equilibrium:  when there is no selection, no linkage to selected loci, no mutation, no gene flow, an infinite population, and random mating, then (1) sex by itself does not change allele frequencies and (2) the diploid genotypic frequencies are freq(AA) = p², freq(Aa) = 2pq, and freq(aa) = q², where p is the frequency of A, and q is the frequency of  a in the gene pool.

VII
Single-locus model of genetical evolution (as represented by program AlleleA1) demonstrates
• Advantageous new recessive alleles take a long time to rise from 1% to modest frequency.
• Advantageous new dominant alleles take off quicker but then drag later on near fixation.
• Maximally additive fitnesses have the quickest evolution.
• All of them evolve a lot in 1000 generations.
• Heterozygote superiority results in a balanced polymorphism.
• Homozygote superiority means that the initial frequency matters.
• Mutation can cause very slow (non-adaptive) evolution by itself.
• With selection mutation can be the rate limiting step.
• Very small populations drift faster, but even big populations drift eventually.

VIII
There is a mutation-selection balance by which the equilibrium frequency of a deleterious allele is the sqrt of the mutation rate / selection .coefficient Selection has a hard time eliminating the last copies of a deleterious recessive allele (if q=1/100, then q²=1/10000), which is why all populations of diploid eukaryotes have a bunch of rare genetic diseases, or in other words, carry a genetic load.

IX
Negative frequency dependent selection is when the rarer allele has an advantage, and results in an intermediate allele frequency (e.g. ~1:1 sex ratio).

X
Gene flow between subpopulations that have different allele frequencies can make any one subpopulation seem like there is homozygote excess. A common effect of migration is to homogenize subpopulations that would otherwise differentiate under drift or local adaptation.

XI
In finite populations, there is genetic drift (changes in allele frequency due to random sampling of which gametes make it into the next generation). Every pop follows a unique path. Drift goes faster in small populations, but given enough time, p can change a lot even with N_e=400. The probability an allele goes to fixation is equal to its initial frequency. Founder effect and other bottlenecks are forms of drift. Drift is non-adaptive evolution, but it may later have fitness consequences (e.g. populations may suffer from inbreeding depression).

XII
Mutations happen. (1) deleterious alleles appear and are eliminated by selection, (2) neutral mutations appear and are lost or fixed by drift, and (3) advantageous alleles appear and are swept to fixation by selection. Advocates of the neutral theory of molecular evolution hold that there are many mutations in category 2 that account for vast numbers of differences between lineages.

XIII
For neutral alleles, the substitute rate = neutral mutation rate. In small populations there are few mutations but those that occur are more likely to be fixed; in large populations there are more mutations but they’re more likely to be lost. So, allopatric pop’s inevitably diverge through the accumulation of neutral substitutions (as well as, of course, selective sweeps).

XIV
Neutral theory is the null model for understanding molecular evolution. One contrasts divergence in pseudogenes, non-coding regions, synonymous substitutions, and amino acid replacement substitution. (Actually, synonymous substitutions are not totally neutral. Codon bias can cause weak selection because some tRNAs are more available than others, thereby affecting the efficiency of translation.) Certain types of loci seem to have been under strong positive selection: recently duplicated genes that have taken on a new function; genes involved in sperm/egg recognition; genes involved in disease resistance; etc. But there is so much variation that there is plenty of room for neutral and nearly neutral evolution at the molecular level.

XV
Non-random mating does not by itself change allele frequencies. It does change the equilibrium genotypic freq’s (≠ p²+2pq+q²): e.g., selfing reduces heterozygosity by 1/2 each generation. In combination with selection against homozygotes (inbreeding depression), non-random mating sets the stage for adaptive evolution in mating systems.

XVI
Usually, when you have inbreeding there is a reduction in fitness, called inbreeding depression. It’s worse in harsh environments. It is mostly due to exposing deleterious recessive alleles. Some species self habitually tending to be purged of deleterious recessives.

XVII
Linkage disequlibrium is a correlation in state between the alleles at two loci. It could involve physical linkage on a chromosome, but that’s not in the definition. It can be caused by mating that is non-random with respect to the two loci, multilocus selection, drift, and population admixture. It is diminished by recombination.

XVIII
When two loci are in linkage disequilibrium because of tight physical linkage, selection on one locus can change the allele frequencies at the other locus. This is called hitchhiking.

XIX
Sex to a geneticist is the systematic recombination of genetic elements through meiosis and fertilization. It is easiest to think about in outcrossing organisms like ourselves, but selfing organisms are also sexual.

XX
There are many ancillary costs of sex (you may not find a mate, searching for a mate takes time and energy, you risk predation and sexually transmitted diseases; the mate may demand courtship exertions; it may not work), and there is a more direct “cost of males”, which in organisms where males don’t invest in offspring is an automatic 2-fold disadvantage caused by investing in sons who do nothing demographically. Yet the vast majority of organisms have sex at least occasionally. Why? At the level of population genetics, the only consequence of sex is the reduction of linkage disequilibrium, so any model for the benefit of sex must include (1) a mechanism that eliminates particular multilocus genotypes or produces others, thereby creating linkage disequilibrium; and (2) a reason why recombination is favored. •Muller’s ratchet is one explanation for how sex is maintained. Slowly drift makes the multilocus genotypes function less and less well.  Eventually, the whole lineage goes extinct by the build-up of mutations. •Selection in ever-changing environments may engender short-term benefits of sex. When there’s a trade-off between being adapted to one environment versus being adapted to others, sex recreates favorable multilocus genotypes; the genes for recombination ride to high frequency in the high-fitness genotypes that they help to create.

XXI
When you cross to lines that differ in a polygenic quantitative trait, the range in the F₂s doesn’t extend to the full range of the parentals, but selection with recombination can recover the full range by bringing together all the small or all the large alleles. Natural populations harbor a great deal of hidden variation that can allow for evolution beyond what already exists.

XXII
Quantitative Trait Loci (QTL) studies combine a molecular map of the genome with measurements from an F₂ array. They allow one to see how traits are associated with each marker (linkage group) and the number of major loci responsible for differences between lines.

XXIII
heritability: h²=V_additive/V_phenotype. There are several ways to estimate h², e.g., parent-offspring regression.
selection differential: S = trait_after - trait_before
evolution is response to selection: R = h² S
A small selection differential iterated over 100s of generations can cause an enormous quantitative response to selection.

XXIV
So far, we’ve talked about directional selection, but there is also stabilizing selection (where intermediate dimensions are most fit) and disruptive selection (where extreme dimensions are most fit).  Stabilizing selection is our primary explanation for stasis. Disruptive selection is important in some mechanisms of speciation and the evolution of alternative phenotypes within a species.

XXV
There’s a lot of genetic variance in most natural pop’s. Why? (1) Populations may be on the way to some new and more apt state; (2) Genetic load is caused by deleterious mutations many of which are recessive; (3) There may be heterogeneities in the selective regime.

XXVI
An adaptation is a feature that was formed during history by selection because of a benefit conferred in a specified function. Many biologists do not define adaptations in this way, but instead use the word ahistorically. In hypothesizing a selection scenario, it is helpful to sketch out the prepositions that specify the form of natural selection on traits, via components of fitness, among individuals, and by agents of selection. Then one should explore a series of alternative hypotheses. There are three forms of data relevant to testing adaptive hypotheses. (a) One can study how selection works in similar contemporary systems. (b) One can see how a character is optimal given other characters. (c) One can trace changes in the character on a phylogeny studying correlations between the character of interest and the environment that is thought to favor that character.


XXVII
The comparative method should take into account phylogenetic information. One wants to correlate CHANGES in one character with CHANGES in another character or environment. It is also informative to know the polarity of character change. Finally, an organism has a large number of features that originated at many different times in many different adaptive contexts. Features that were formed for one reason are often co-oped for a different use (exaptations). Also, not all features are adaptations.

XXVIII
Along with positing adaptations, ultimate explanations also often involve recognizing tradeoffs and constraints. A tradeoff is implied when becoming good at one thing makes the organism bad at another thing. It is impossible to become good at all components of fitness. Constraint is a more general idea: the functional features of the organism dictate which evolutionary paths are “easy” or “hard”. The path evolution takes can also be biased by those changes that there is genetic variance for. Lastly, optimal design often does not evolve because each tiny change has to itself be an improvement; gradualism limits reorganization.

XXIX
Proving the details of a historical explanation is not the only reason to think about adaptations.  A second reason is that we further our understanding of evolutionary processes. A third reason is that we further our understanding of the biology surrounding whatever we are studying - in the process of testing hypotheses about ultimate causes, researchers are led to discoveries in proximate biology. 

XXX
Avoid statements of characters arising for the function of allowing groups, species, or higher taxa to survive (unless you really mean it).
bad: Saber toothed cats evolved big teeth for the good of the species.
better: Among the ancestors of saber toothed cats, individuals with bigger teeth were better able to hunt, were more likely to reproduce, and therefore passed their genes for big teeth on to more descendants than did individuals with smaller teeth.
Evolution by selection happens to populations of genes through the differential performance of phenotypes as expressed in individuals.  The fate of taxa is usually irrelevant, and the implication of group selection is misleading.

XXXI
Complex adaptations usually are thought of as arising by many small steps, not by one saltation in a single individual; each small step, however, is traceable to the substitution of discrete mutations.
bad:  The first protosquid that had an eye was hugely more successful than his blind compatriots.
better: Over many generations, numerous improvements in the visual apparatus of early cephalopods were selected for.
Evolution happens incrementally; biologists debate how incrementally, but major features are not thought to arise like Athena from the head of Zeus.   Gradualism applies both to gradual modification of structures in morphospace and gradual change through an extended period of time.

XXXII
Selection does not anticipate long-range goals and does not compel orthogenesis.
bad: Ever since the Eocene, horses have been selected towards ever increasing size.  From Hyracotherium through Merychippus we finally arrived at the modern Equus.
better: Many lineages of horses have evolved since the Eocene with considerable divergence in size.  The earliest horse, Hyracotherium, was much smaller than the only surviving lineage, Equus.  If we trace the fossil record between the two, we find that greater size evolved in several stages.  The apparent "trend" that could be pointed out is an artifact of ignoring the many forms that are not in the direct line between the first (small) and the modern (large) horse.
Never use the phrase "ladder of life."  Evolution is a bush, not a ladder, and it is misleading to connect a tip to a root ignoring all other branches.  Most trends in macroevolution are not likely to be the result of constant directional microevolutionary selection.

XXXIII
When discussing differential success, avoid absolutism in the form of words like "all", "only", and "every".
bad:  Only the cheetahs that were very fast at running down prey survived.
better: The cheetahs that were very fast tended to have higher rates of survival than those that were just a little bit slower.
The determinism of selection is typically only probabilistic.

XXXIV
Selection does not respond to needs.  It is the result of variants conferring success in the context of intraspecific competition.
bad: Once flat fishes started living on the ocean bottom, they needed to rearrange the position of their eyes to get them both on the "top" side of the fish.
better: Once flat fishes started living on the ocean bottom, those individuals in which the "bottom" eye peeked out from under the fish a bit were better able than their more symmetric compatriots to capture prey.
"Need", "necessary", "had to in order to survive" may apply to extinction but are unlikely to apply to selection or the causal origin of adaptive traits.  True, an adaptation may be shaped primarily during hard times, but not when it is needed outright for the survival of the population.  If they drop a Big one, we’ll need to be tolerant of radiation, and sufficiently far from ground Zero where tolerance is not needed but provides a statistical edge there may be selection favoring more tolerant individuals over less tolerant ones, but where it is really needed, evolution is expected to be too slow to come to the rescue.

XXXV
6. Evolution does not proceed by the inheritance of acquired traits (“the Lamarckian fallacy”).   It is also misleading to imply that adaptations arise through the efforts and moral virtues of the organism (what Darwin disdained Lamarck for).
bad: Monkeys who used their tails to hang from grew very strong tails.  This great tail strength built up over the generations.
better: Monkeys varied genetically in how strong their tails were able to grow when used to hang from.  Genes that allowed the tails to grow stronger increased to fixation over the generations.
The example was chosen because it is one in which the plasticity of growing strong tails might set the context for selection sorting out genes for stronger and stronger tails (“genetic accommodation”).  This might sound like the Lamarckian fallacy but is in essence Darwinian.

XXXVI
Evolution by selection is not evolution by “randomness”.
bad: The flowers of orchids with all their functional elaborations evolved through randomness.
better: The flowers of orchids with all their functional elaborations evolved through non-directed mutation, which generated variation, and deterministic selection, which even when weak can accumulate favorable variants through many generations.
The essence of selection is the non-random relationship between character variation and fitness.  Selection is brainless, algorithmic, and automatic, but it is not random.  Randomness can, however, result in nonadaptive evolution in the form of genetic drift and particular solutions to problems when more than one is possible.

XXXVII
Particular adaptations arose in particular lineages that are associated with particular phylogenetic levels; the origin of an adaptation should not be tied to narrower or broader taxa.
bad: Once upon a time, rattlesnakes didn't have pits.  Mutations arose for sensing heat, and these were selected for such that those rattlesnakes with even rudimentary pits ate more often and left more offspring that those rattlesnakes with no heat sensors.  Through progressive improvements, exquisitely functional pits evolved.
bad: Once upon a time, snakes didn't have pits.  Mutations arose for sensing heat, and these were selected for such that those snakes with even rudimentary pits ate more often and left more offspring that those snakes with no heat sensors.  Through progressive improvements, exquisitely functional pits evolved.
better: In the lineage that led to pit vipers, there were snakes that didn't have pits.  Mutations arose for sensing heat, and these were selected for such that those snakes with even rudimentary pits ate more often and left more offspring that those snakes with no heat sensors.  Through progressive improvements, exquisitely functional pits evolved.
Rattlesnakes belong to a group called pit vipers, which unlike other snakes have pits below their eyes that they can use to sense heat differences as small as 1/300ths of a degree.

XXXVIII
Bateman’s principle states that sperm are more numerous and cheaper than eggs, so a few males can sire more than their share of offspring, implying male reproductive success can be more variable than female reproductive success. Sexual selection acts more on males than on females, generally. The exceptions improve the logic. The gender with the lesser parental investment is the most subject to sexual selection (females in seahorses). When both sexes invest heavily, or in monogamous species, things become more complicated, and sort of more equal, but often still males and females follow different 'reproductive strategies'. In addition, there may be more than one strategy for being of a certain sex (e.g. territorial males and sneaker males).

XXXIX
Characters in males may evolve features used in interfering with other males, such as weapons. Their sperm production and delivery also evolves in the context of competition and depending on the mating system. And, sexual selection can be caused by mate choice. Female choice may evolve by any mixture of several causes: (a) for acquiring good genes, (b) for direct acquisition of resources, (c) as a pre-existing sensory bias, or (d) through runaway coadaptation with a male character. Runaway sexual selection is thought to follow after there is some female choice - then females who choose will have sexy sons who will sire more than their share of granddaughters who will tend to be choosy.

XL
Sexual selection will proceed until it is fully countered by survival selection. The form of sexual selection depends on parental investments and on who controls what resources and mating opportunities.

XLI
the naturalistic fallacy - implying that because something is natural (adaptive) that it is ethically right.
anthropomorphism - making something seem human (that oughtn’t to be), e.g., making an animal seem as if it has conscious intentions when it may well be following fixed action patterns.
teleology - reading purpose, intent, design into the features of the natural world, e.g., making adaptations seem as though they are for a purpose and reveal the intentions of a designer.
typology - treating the essential type of something (a species) as though it were real, more real than the flawed instances.  Actually the characteristics of a taxon should be thought of like averages, ‘unreal’ summarizations.

XLII
Kin selection is one explanation for altruistic behavior. Selection shapes characters in individuals that benefit kin (although from the point of view of genes, selection is still favoring the selfish allele). Hamilton’s rule states an altruistic feature will be selected for if the cost to the actor is less than the benefit to the recipient times the coefficient of relatedness:  C<Br. In other words, selection favors traits that maximize inclusive fitness, which is made up of direct fitness + indirect fitness. The relationship between the character and the indirect component is kin selection. Many examples of cooperative breeding involve resources being few in number but able to support a larger family. Once altruism is established it becomes adaptive to know who is kin and to masquerade as kin if possible.

XLIII
Reciprocal altruism or at least ‘byproduct mutualism’ is another way in which seeming altruism can be favored.  Altruism is favored when each individual repeatedly interacts with the same set of friends, opportunities for altruism are frequent during the course of a friendship, the animals have good memories, positions change frequently, and costs are less than benefits. Most examples, though, are in animals that are already social seemingly for other reasons.

XLIV
The most extreme form of altruism is eusociality, characterized by specialized non-reproductive castes, cooperative brood care, and overlapping generations. It has arisen many times in the Hymenoptera, a few times in other insects, in snapping shrimp, and in mole-rats. In the Hymenoptera, haplodiploidy is thought to pre-dispose the evolution of eusociality because it is better to raise a sister (r=3/4) than a daughter (r=1/2). This increase in inclusive fitness is only realized when there is a female biased sex ratio. Haplodiploidy is probably part of the explanation. The evolution of altruism is also favored when costs are low and benefits high. In the Hymenoptera, fancy nesting arrangements seem to be a precursor to eusociality. In naked mole-rats, factors that may have propelled the evolution of eusociality include resources being patchy, extreme inbreeding, and the already fixed system of overlapping generations and material care as seen in other species of mole rats.

XLV
Parent-offspring conflict (when to wean) is expected because for mothers costs outweigh benefits sooner than for offspring. For mothers, weaning/fledging/etc. is optimal when the risk to the existing offspring equals the benefits to a future offspring, but for the existing offspring itself, it would do best to wait until the costs are 2x the benefits. Then it becomes better to have a full sibling.

XLVI
Life history evolution is about how life table statistics evolve. Life tables emphasize fecundity m_x and survivorship l_x. Due to the constraints in how to deploy time and energy, tradeoffs lead to alternative strategies (reproduce early but die young versus live long but reproduce late in life; disperse versus have high fecundity). These can be seen in different species and sometimes within a species.

XLVII
Senescence is an acceleration of the rate of mortality with age, or a decline in the rate of fecundity. There's a lot of proximate biology here, but the winning evolutionary explanation posits (a) that deleterious mutations that act late in life (after the organisms have usually done most of their reproducing) have very little effect on fitness and so have higher frequencies when it comes to mutation-selection balance, and (b) a mutation that is a little bit beneficial in youth will be selected for even if it is lethal in old age. Ecological deaths that are not the fault of the organism cause declining survivorship, and this sets up the age structure in which selection favors senescence.

XLVIII
One would expect that there is an optimal clutch size. If a bird lays too few eggs, then opportunity is lost. If a bird lays too many eggs, then they can’t all be raised. However, generally, birds lay slightly fewer than they can fledge. They may be saving some energy for the next clutch and/or doing a good job a raising the right number of healthy fledglings rather than fewer fledglings that would be less well fortified.

XLIX
Generally, there may be a tradeoff between the size and number of offspring (seeds, fish or insect eggs, etc.). The fitness of an offspring is presumably a decelerating function of size. So, there is an intermediate optimal offspring size from the mother's point of view. Also, mothers may be plastic, laying larger eggs on poorer hosts.

L
Antibiotic resistance has evolve many times and on a human time scale very quickly. Selection is hard to avoid. Initially resistance comes at a cost, but eventually the resistant bacteria evolve to be just as fit in the absence of antibiotics.

LI
Evolving pathogens have short generation times, so they can adapt faster than their hosts, and they have large population sizes, which provides great variation through mutation. Only under certain conditions do they evolve to be non-virulent. Patchy genetic structure (caused by there being few instances of multiple infection) can make for higher level selection among founders. In contrast, virulence is favored by a high incidence of multiple infections and when transmission does not depend on the host being healthy (vectorborne diseases, waterborne diseases).

LII
Carrying out the adaptationist's program on humans requires considering our environment of evolutionary adaptation. For 99% of human history we were hunter-gatherers, got plenty of exercise, ate no grains, no milk as adults, no refined sugar, not much fat, and no alcohol. We did not do too much close visual work. Women were often pregnant or lactating. Etc. Many features of our physiology are adaptations to those conditions not to L.A. living.

LII
Like many animals, humans seem to have a system for discriminating based on relatedness. Perhaps the amazing thing is just how altruistic step-parents are, but there a lot of evidence that they do less for step-children than genetic children, and the difference seems to have lasting effects on the success of the offspring. Most alarming, is that the relative risks of children being killed by a step parent are very high.

LIII
One of Darwin's contributions was to depict the diversification of life using an evolutionary tree, or phylogeny.  Let's start with the example shown in the phylogeny below.
 
Various lineages of organisms are descended from common ancestors.  In the example, birds share a common ancestor with dinosaurs and then those two groups collectively share a more remote common ancestor with crocodiles, and those three collectively share an even more remote common ancestor with lizards plus snakes, which diverged from each other subsequently.  These parts of this tree are well known.  Others parts of the tree are still poorly known.  Some data suggest that mammals diverged before the radiation of reptiles (turtles through dinosaurs on the phylogeny), while other data suggest that turtles or {lizards+snakes} diverged first.  From here on, we'll assume mammals diverged first, then turtles, then the {lizard+snake} branch.

LIV
It is often helpful to map character changes on a phylogeny.  As examples, consider two characters:  1. having continuous skull bone between the eye and the nostril (A) versus having an opening in that place (B); and 2.  cold-bloodedness(C) versus wwarm-bloodedness(W). Character evolution is reconstructed on the phylogeny shown to the right. To do this, one looks for a branch in the tree where a change in character would explain the distribution of character states in all the taxa.
 
For character 1, there is a single place in the tree, marked with a solid bar labeled 1:  a change from no opening to having an opening along this lineage and inheritance of the opening in all descendants above this point would result in crocodiles, dinosaurs and birds having the opening.  The opening is a shared derived character, or synapomorphy, of the group {crocodiles+dinosaurs+birds}.  With some characters it is not possible to explain the character distribution with a single change.  For character 2, two changes are required:  warm-bloodedness evolved in both the lineage leading to mammals and the lineage leading to birds.  Since we do not know whether dinosaurs were warm-blooded or not (just from these data), we can't say whether warm-bloodnessness arose before the dinosaur-bird split or afterwards.  At any rate, it seems there was convergent evolution in warm-bloodedness. The following words are essential vocabulary for speaking of character evolution.
homology – a character that is similar due to common ancestry
    synapomorphy – shared derived character, e.g., being on land for tetrapods
    symplesiomorphy – shared ancestral character, e.g., being in the water for fishes + sharks
homoplasy – a similarity between two kinds of organisms that is not due to common ancestry
    convergence – separate evolution in different lineages of a feature that ends up being similar
    reversal – changing back to a previous state
One can use phylogenies to judge whether a similarity should be interpreted as a homology or a homoplasy, and (less easily) whether a homoplasy should be interpreted as a convergence or a reversal. The most frequently used criterion for making these judgments is parsimony, which in this context means choosing the explanation that requires the fewest changes in character state.

LV
Parsimony is also the most frequently used criterion for judging among various possible trees: one choose the tree that requires the fewest steps, i.e., the smallest number of changes in state for all the characters used to infer the tree. There are many other methods for phylogenetic inference, most notably various maximum likelihood methods that model the processes of molecular evolution. If there were no homoplasy, inferring the tree would be easy (the similarities would tell you directly which trees were possible), but before you have the tree you cannot tell which similarities are homologies and which are homoplastic. The solution is to get a large number of presumably independent characters and go with the preponderance of evidence on the argument that common ancestry is the cause that most coordinates nested patterns of similarity.

LVI
Branching pattern (the sequence of common ancestors) does not necessarily represent overall similarity or relative difference.  Overall similarity can be inflated due to convergent evolution, and when rates of evolution are unequal some lineages may diverge while others remain relatively similar.  In the diagram below, the lineage that led to {C+D} diverged radically, while the lineages leading to A and to B remained static.
 
Notice on the right that one might well group A and B together, although their most recent common ancestor is more remote than the common ancestor of B and {C+D}.  In our previous example, the birds have evolved physiologically and anatomically very far (some would say) from the living reptiles (turtles though crocodiles). A group that contains a common ancestor and all of its descendants is monophyletic; it’s a whole branch of the tree of life. A group that excludes some of the descendants of its common ancestor is paraphyletic; {A+B} is paraphyletic by the exclusion of {C+D}. Traditional taxonomy was not developed based on a rigorous knowledge of phylogeny, so taxa (classes like Reptilia, orders, families, genera, etc.) cannot be presumed to be monophyletic. Cladistics refers to branching pattern. Someone who argues that taxonomy should be revised to only recognize monophyletic groups is a “cladist.” Phenetics refers to overall similarity. Phyletics refers to the length of branches on the phylogeny.

LVII
There is free rotation around any node.

This means you have to be careful when viewing large complicated trees. They may look different but convey the same branching pattern.

LVIII
Phylogenies can be based on morphology, or molecules, or combinations of data sets. Morphology might sometimes be very helpful if complex structures tend to evolve when groups-to-be are formed and then somehow become locked in place and made to be conservative. Phylogenetic conservatism is a prominent feature of biodiversity. Nevertheless, in principle one can get very large amounts of information by using molecular markers, so at present molecular approaches are adding greatly to our resolution of the tree of life.

LIX
Fossils can be included in a phylogenetic reconstruction. Since generally DNA cannot be gotten from fossils, fossils are probably best used in conjunction with a phylogeny whose superstructure is based on extant organisms. The value of adding fossils is much like the value of adding additional extant organisms. We could have inferred the relationships of birds, lizards, and mammals without knowing about crocodiles, but adding crocodiles tells us a bit more about the transition. Crocodiles are a link between lizards and birds. Dinosaurs are another link. When one finds a new link, one can add that “missing link”, and it improves one’s inferences about the branching pattern and about character evolution. The one way in which adding fossils is not just like adding an extant missing link, is that they allow one to date the appearance and disappearance of forms in the fossil record.

LX
Methods such as parsimony can be used to infer branching pattern without rooting the tree. Rooting is done by outgroup comparison, by referring to organisms that are thought to be (just) outside one’s focal group, sharing a most recent common ancestor with the focal group that is more remote than the most recent common ancestors of all the species within the group. To know that one’s outgroups are not part of one’s focal group, someone must have already done at least a crude phylogenetic analysis of the larger group. By stepping inward, obvious relationships can be used in conjunction with more character data to resolve closer and closer relationships.

LXI
Although we can infer a great deal of the tree of life, we are very far from having complete resolution.

LXII
Similarities between species may be homologies or homoplasies (analogies). Homologies are the basis for inferring the tree, but some presumed homologies may actually be homoplasies. You'll only know how to interpret them after you have the tree. The solution is to get a large number of independent characters and go with the preponderance of evidence on the argument that common ancestry is the cause that most coordinates nested patterns of similarity. There are several methods for inferring phylogeny. The simplest is parsimony: you choose the tree that requires the fewest changes in character state.

LXIII
Bootstrapping can help you quantify your confidence. The computer resamples your data, making (say) 1000 resampled data sets; each one is used to infer a phylogeny; you then ask how many of the phylogenies have a monophyletic group of interest.

LXIV
What are the phylogenies good for? (a) Establishing polarity. (b) Classifying/organizing diversity (although traditional taxonomy includes many monophyletic groups). (c) Distinguishing homology from analogy. (d) Narrating the history of adaptations and carrying out the comparative method. (e) Using molecular clocks either relatively or with a calibration. (f) In biogeography for saying how lineages radiated as geographic barriers rose and fell, dispersal events happened, etc. (g) For studies of cospeciation versus host shifts.

LXV
The two huge things that evolutionary biology tries to explain are (a) adaptations, which were treated in the 2nd part of the course, and (b) diversity. Much of this part of the course is on diversification and lack thereof.

LXVI
There are many species concepts. These sound like semantic arguments, but they encapsulate much of the thinking about how diversity evolves, i.e., about speciation.

LXVII
A taxonomic (morphological) species is whatever a competent taxonomist proposes and other competent taxonomists see no reason to revise. This sounds like a cop-out, and certainly strips species of deep meaning, but it works surprisingly well. Folk taxonomists from many cultures agree on the species they recognize. Taxonomic species have proven to have great predictive power, even for traits that the taxonomists didn't know about, especially in sexual organisms. Often distinguishing taxonomic species comes down to showing discontinuities in multiple traits dividing up the organisms. Some species are very uniform groups; others contain great (geographic) variation. Some are ancient; others recent. Try to avoid thinking of them as ideal types; there is no typical perfect mountain lion, just lots of real mountain lions each with its own peculiarities.

LXVIII
A biological species is a set of actually or potentially interbreeding populations separated by other such entities by intrinsic reproductive isolating barriers. These barriers can exist at any stage in the life cycle. Prezyogtic barriers include being adapted to different habitats, being cued to reproduce in different seasons, when females don't recognize males, when courtship doesn't work, when a male's genetalia don't fit or are not pleasing to the female, and when there incompatibilities between egg and sperm. Postzygotic barriers include when embryos don't develop well, when hybrids are feeble, when hybrids are sterile, and when progeny beyond the F₁ generation have poor genetic balance that causes them to be unhealthy. The biological species concept works nicely for sympatric organisms, but not as well for divergent allopatric populations. It is futuristic, asking us to know about interbreeding that might be possible under conditions that don't exist. (Whether or not you can breed them in a zoo or garden is not conclusive, because if they would never meet in nature due to being adapted to different habitats, then they would still be different biological species.) Also notice that the biological species concept doesn’t say how complete the barrier to interbreeding must be for one to recognize them as different species. Thinking of diversity in terms of biological species probably means paying attention to a great many cryptic biological species. Also in practice it could mean demoting many taxonomic species to geographic races that differ in only one striking character.

LXIX
A phylogenetic species is the smallest group that is monophyletic and diagnosable. Diagnosable is meant to mean that all members of the group have a synapomorphy, but this depends on how hard you look, and even many individuals would have a new mutation somewhere in their genome. In principle, a genealogical species in one that has been independent long enough for the variation within the species to have coalesced, but it is impractical to want to know about this for even a good sample of the genome. The phylogenetic species concept is historical; it is about how much history has accumulated between two candidate species. Under the phylogenetic species concept, there are probably many geographically separated populations that would be worthy of recognition.

LXX
Isolation can start out being extrinsic, caused by a rare dispersal event or by vicariance (the division of the geographic range by a geological barrier). Instant intrinsic isolation can occur by polyploidization, a chromosomal change, or even a macromutation. Such instant speciation is probably rare in certain groups, like birds and mammals.

LXXI
Following the biological species concept, the word speciation is most commonly meant to mean the origin of reproductive isolating barriers. This can happen in many ways. The classic scenario is (a) allopatry causes extrinsic isolation, (b) then diversifying selection causes divergence in ecology and morphology, (c) and there is a correlated response to selection in the reproductive system. It is possible that the divergence could occur through genetic drift rather than selection; this has been often suggested in the form of founder effects following a new colonization. Speciation could happen in sympatry if there were strong enough disruptive selection, but most examples involve at least some form of micro-allopatry or micro-allotopy (for instance caused by living and mating on different host plants at different times of year).

LXXII
A lot of ink has been spent on explaining how evolution in survival characters causes evolution in reproductive characters under the assumption that the two have a completely different genetic basis: this is the double variation problem. But speciation can occur more easily if one allows for single variation in which selection causes evolution of one set of genes that also affect reproductive isolating barriers. In the classic scenario, there is pleiotropy (or possibly linkage) between the characters of survival selection and the reproductive system. In the case of sexual selection, it's obvious that the same characters under selection can cause reproductive isolation.

LXXIII
Diversifying selection (favoring different states in two isolated populations) or disruptive selection (favoring extreme states in one population) greatly facilitates divergence. Another way in which two populations can come to be different is by different unique mutations arising in the two populations, even if they have the same phenotypic effect and are similarly selected for. This is a way for randomness to play a role in speciation.

LXXIV
Many studies have followed speciation in the lab, and still more have attempted to cause speciation in the lab. Sometimes it works, often it doesn’t. When you do get the evolution of reproductive isolating barriers it is usually when you set up very strong diversifying selection on several traits simultaneously. Actually, sometimes they even evolve in sympatry.

LXXV
In certain groups of plants, many novel entities (species) may have arisen through hybridization. In the case of polyploidy, this is easy to see. Diploid hybrid speciation is possible but much harder to detect. In a number of studies people have hybridized plants in the garden and after a few generations they have evolved into a new species that was partially inter-sterile with its parents and possessing of novel characters. The best-studied example of this is the case of the sunflower Helianthus anomalous, which is thought to have arisen in nature from H. annuus x H. petiolaris. It was re-created in the lab by hybridization followed by hybrid-by-hybrid breeding for several generations. Very curiously, the resulting genome contained the same linkage groups coming from the two respective parents. This must mean that there is selection to eliminate other combinations of linkage groups.

LXXVI
Aside from being a cause of speciation, hybridization after ordinary allopatric speciation can tell us about speciation. Very occasionally two divergent species will come back together (usually in intermediate or disturbed habitat), and will make abundant hybrids (e.g., in hybrid zones). What keeps the two parent species from becoming homogenized. The reigning hypothesis is that they are kept in their niches by strong selection.
 
LXVII
Why don't species ranges grow bigger and bigger by adaptation at their edges? Because of gene flow from their centers. They do well in the conditions they are adapted to, so this is where the birth of individuals is greatest. There ensues a source/sink dynamic. In other cases, gene flow is too restricted, and then indeed the lineage can grow bigger, but then it forms local ecotypes.
 
LXVIII
Another way species might form is by peripatric speciation. Mayr supposed that most of the time species are under stabilizing selection, so they show stasis. Bonds of mating keep the species homogenized; genes are all working together in a coordinated way. Occasionally colonists break-off from the main population in peripheral populations, in unusual habitats, where founder events and novel selective pressures force evolution in a few genes. This causes directional selection on other loci to be compatible, which causes selection on yet other loci. There’s a genetic revolution and a reorganization of the genome.
 
LXIX
Is speciation caused by few or many genes? The long-time presumption was many genes, but there's plenty of empirical evidence that it could be due to few genes ("could be" is not the same as "is").

LXXX
Following initial divergence in inter-compatibility, secondary contact may cause further evolution. There could be character displacement in, for example, the characteristics of the beak. Character displacement in characters than can cause reproductive isolation is called reinforcement: the species became partially isolated (often assumed to be in postzygotic barriers) and then there is selection for prezygotic barriers that save effort from being wasted on hybrids. In fruit flies, closely related sympatric species pairs tend to have stronger prezygotic barriers than allopatric species pairs.
 
LXXXI
The fossil record seems to display punctuated equilibrium - species appear relatively quickly (1000s of generations) and then do not change much for long intervals. This is nicely documented for bryozoans. The opposite of punctuated equilibrium is phyletic anagenesis, and there are also example of such gradual change without speciation. Characterizing the record as mostly punctuational is in vogue at present. Mayr’s genetic homeostasis and speciation through revolution originally inspired punctuated equilibrium, but the pattern could be due to more external reasons: if niches are fixed and species are tracking the habitat where they do best, then they will sort themselves into places where they experience stabilizing selection.

LXXXII
The really curious thing about stasis is it seems to be not only at the species level: phylogenetic conservatism is overwhelming around the level of species but also implicit in our ability to characterize genera, families, orders, etc. Conservatism involves traits of obvious importance (how DNA is replicated) and traits that one might think of as trivial (the arrangement of stamens in the mustard family). Conservatism could be due to the character being a functional adaptation that is constantly under stabilizing selection directly. Or it might come to be canalized whereby the genetic system comes to make the character despite hidden genetic variation. This could eventually mean that it would remain conservative even with out stabilizing selection on it per se. Either way, many characters have been imburdened. In different lineages, different types of characters are conservative or labile. Stasis is the great unexplained truth of biological diversity.

LXXXIII
Given that many characters seem to get imburdened, this opens up the possibility of hierarchical clade selection in which some clades are more ‘successful’ than other clades because of their characters. There's a grand analogy between microevolutionary processes effecting adaptations, and macroevolutionary processes effecting diversification. Mutations are to microevolution as imburdened adaptations are to macroevolution. There is the potential for innovations that favor radiation, innovations that favor endurance, innovations that disfavor radiation, and innovations that disfavor endurance. Consider the consequences of an insect lineage becoming herbivorous, a plant lineage adopting bird-dispersal, a bird lineage that evolves outbreeding mechanisms that cause lots of gene flow throughout the range, a clam lineage that evolves to brood its larvae.

LXXXIV
The processes of community ecology cause macroevolution in the same sense that the processes of population ecology cause microevolution.  So when you see an organism is apt to its environment, you don’t know if this is because of mutations + normal individual selection in this environment or because of ancient innovations + species selection. Remember that usually when individual selection is pitted against group selection, individual selection gets its way because the episodes of selection are more frequent. Well, in the assembly of a community through succession, species selection probably happens more quickly than adaptation due to individual selection.

LXXXV
Clade selection can have an additional effect on the microevolutionary process: it can use up the lineages that are genetically capable of evolving through microevolutionary processes. Such lineage selection has been used to explain the maintenance of sex: all the lineages that easily could invent obligate asexuality may have all gone extinct.
 
LXXXVI
Adaptive radiations are groups where there has been great diversification in many different modes of life. A key innovation may move a clade into an adaptive zone - a set of niches that are open to be filled.  But characters are not the only thing that can affect speciation and extinction  rates. Also there are circumstances. A plant gets dispersed to a new volcanic island and radiates into 100 species. Dinosaurs go extinct, and rats take over the world. 

LXXXVII
These macroevolutionary processes act at scales of the dynamics within a genus all the way up to the pageant of the metazoa. Several lines of evidence suggest a very substantial pre-cambrian history to the metazoa, perhaps in the form of what we now think of as larval morphology. Then in 1% of the history of the earth, all the phyla appeared. Did they just become big and armored? Maybe a sudden increase in O₂ allowed it to happen. This is the primal radiation of animals, but there were subsequent great radiations too, e.g., the orders of mammals.
 
LXXXVIII
Considering the history of the metazoa, there have been five really big mass extinctions: end of Ordovician, end of Devonian, end-Permian, end Triassic, and K-T. The K-T event, we’re pretty sure now, was demarcated by an object  from space the size of a mountain that made the Chicxulub crater. During mass extinctions some rules that are followed by background extinctions no longer hold. A few rules work even during mass extinctions: species with  larger ranges were less likely to go extinct crossing the K-T. After a mass extinction, there seems to be an increase in diversification rates.
 
LXXXIX
The hox genes cause (in a developmental sense) both flies and mice to have all their different segments. Oddly, they are arranged on the chromosome in the same order in which they are expressed. And they are found in all bilaterans. They are found in organisms that don't have segments and in organisms whose segments are all similar. So it seems that analogous patterns have been built on the same underlying developmental cascade of gene expression; there’s deep homology but not exactly in the phenotype, rather in the rhythm of development.
 
LXL
In flowers, something vaguely similar arose: homeotic pattern formation genes cascade in their expression determining what will be an inflorescence, what will be a flower, and then the ABC genes determine which flower parts are which. Many of these genes are found in non-flowering plants. So, convergent evolution has occurred in the sense that two completely separate lineage of complex multicellular organisms have 'discovered' the logic of how development can be orchestrated. This is deep analogy. And once a system is established, it shows enormous conservatism. With maybe only a couple exceptions in 250,000 spp. of angiosperms, the order of organs (when present) is carpels, stamens, petals, sepals, even though one can make other kinds of mutants.

LXLI
Heterochrony is an evolutionary change in the timing of development and maturation. Paedomorphosis is when the derived condition looks like the juvenile of the ancestor (there are 3 ways to do this: neoteny is when development is slowed down; progenesis is when maturation is sped up; and post-displacement is when the initiation of development is postponed). Peramorphosis is when the ancestral adults look like juveniles of the descendant developmental series (acceleration is when development is sped up; hypermorphisis is when maturation is slowed down; and predisplacement is when development is given a head start). Heterochrony could be global or could affect just one organ (the flower, the face). Often it seems to be associated with a change in life history strategy, e.g., from r- to K-strategy. Heterochrony means a lineage can change dramatically in morphology/ecology through a tiny genetic change affecting already coordinated development. Actually, almost all changes in morphology are a form of at least local heterochrony. Few other possibilities exist: de-novo evolution (which is probably extraordinary) and heterotropy (transfer of position).
 
LXLII
Short history of us. Bacteria have been abundant for 3,500,000,000 years. Eukaryotes arose 2,000,000,000 years ago. Animals got big and fancy during the Cambrian, starting 543,000,000 years ago. Humans and chimps diverged 5,000,000 years ago, the oldest fossils assignable to our genus 2,500,000 years ago, and people with modern brain sizes 100,000 years ago. Agriculture got serious about 6,000 years ago, and the steam engine about 200 years ago.
 
LXLIII
Who are are closest relatives? {gibbon,{orangatang,{gorilla, {bonobo, common chimp}, human}}}. We can trace our history through the bush of life: human characteristics evolved in many stages and with some now-extinct cousins. Forms included gracile australopithocines, robust australopithocines, Homo habilis, Homo erectus, etc. They often existed simultaneously. Brain size got bigger through many stages. Human characteristics include: -bipedalism -reduction in hair; more sweating; physiology of long-distance walkers -very handy; tools very elaborate -short snouts for eating cooked food -females sexually interested all the time -perky breasts -language -huge heads -long development times -menapause -natural-born naturalists, telling stories of snakes and when the world was young; not easily taught quantum mechanics.

LXLIV
Every once in a while a lineage of organisms stumbles into a new realm. Retrospectively, we notice it had major consequences, but we can attribute this "major transition" to mundane adaptive advantages. I suggest that the lineage that led up to humans discovered semantic language probably for reasons related to social selection, and through language we are working on figuring out such cool things as how evolution works.

LXLV
Many of the major transitions in life are characterized by a lineage finding a new level of selection. For example, when complex multicellular animals arose, selection among colonies came to dominate over selection among cells. With humans, we get cultural evolution via selection of memes.