AP-Level Biology

A drought on a small island

Daphne Major is a volcanic cone in the Galapagos, small enough that Peter and Rosemary Grant and their students could catch, measure, band and follow essentially every medium ground finch on it. In 1977 the rains failed. The small, soft seeds the finches preferred ran out first, and what remained were large, hard seeds that only birds with deep, powerful beaks could crack. Most of the population died.

The survivors were not a random sample. They had measurably deeper beaks than the population they came from, and when they bred, their offspring inherited that difference. In one generation, on one island, in one dry season, the average beak of a wild population changed. Nothing was trying to change. Birds with shallow beaks simply starved.

That is the whole mechanism. This lesson takes it apart, then asks what evidence there is that the same mechanism, running for four billion years, produced everything alive.

Four conditions, and then it is arithmetic

Natural selection follows necessarily whenever four things are true of a population.

1. Variation. Individuals differ. Beaks differ in depth, bacteria differ in membrane proteins, plants differ in flowering time.
2. Heritability. Some of that variation is passed to offspring. A trait that varies only because of nutrition or damage cannot be selected, because there is nothing to inherit.
3. Overproduction. More offspring are produced than can survive. A single cod releases millions of eggs; on average two survive to breed, or the population would explode.
4. Differential survival and reproduction. Which individuals leave descendants is not random with respect to the variation.

Given all four, the frequency of the favoured heritable variants must rise. There is no additional assumption. Darwin published the argument in 1859, but the logic is close to a mathematical necessity, which is why it was independently reached by Alfred Russel Wallace and why the two men's papers were read together to the Linnean Society in 1858.

Key idea: fitness in biology means reproductive success, not strength, health or speed. An organism that lives fifty years and leaves no offspring has a fitness of zero.

What selection is not

Three claims sound like evolution and are not.

• Individuals do not evolve. A finch's beak does not grow during a drought. The population's average changed because of who died. Evolution is a change in allele frequencies across generations, and it has no meaning for a single organism.
• Selection has no goal. It cannot plan for a drought that has not happened, and it cannot preserve variation for later use. It acts on what is present now.
• Use and disuse do not shape offspring. Jean-Baptiste Lamarck proposed in 1809 that a giraffe stretching for leaves would pass a longer neck to its calves. Bodily changes acquired in a lifetime are not written back into the germ line, and the modern evidence for that is molecular rather than merely observational.

Three shapes selection can take

Type	What it does to the distribution	Example
Directional	shifts the mean toward one extreme	beak depth on Daphne Major after 1977
Stabilising	narrows variation around the existing mean	human birth weight: very small and very large newborns have historically had higher mortality
Disruptive	favours both extremes over the middle	bill size in some seed-eating birds where only large and small seeds are abundant

Sexual selection is a fourth pattern with its own logic. A peacock's train is a burden in every respect except mating, and it persists because reproductive success, not survival, is what fitness counts. Selection routinely produces traits that would look like poor engineering to anyone measuring only survival.

Evidence one: fossils, including ones found on purpose

The fossil record is incomplete, since fossilisation is rare, but it is not random, and its order is a testable prediction. In 2004 Neil Shubin's team went to Ellesmere Island in the Canadian Arctic specifically because the rocks there are about 375 million years old, the age at which a fish-to-tetrapod transition should appear, and because they were freshwater sediments of the right type. They found Tiktaalik roseae, an animal with fish scales and gills alongside a mobile neck, ribs and fins containing the bones of a wrist. The prediction specified the age, the rock type and the anatomy before anything was dug up.

Evidence two: homology and structures that do nothing

Look at the forelimb of a human, a bat, a whale and a horse. They perform completely different jobs and they are built from the same bones in the same order: one upper bone, two forearm bones, a cluster of wrist bones, then digits. That is homology, similarity from shared ancestry rather than shared function. Contrast it with analogy: the wings of a bird and a fly do the same job and share no structure at all.

Vestigial structures make the same point more sharply. Whales and some snakes carry reduced pelvic bones connected to no limb. Flightless cormorants on the Galapagos have wings too small to lift them. An engineer designing from scratch would not include them; an inherited body plan being modified would.

Evidence three: the molecules agree with the bones

Nearly every organism on Earth uses the same genetic code, in which the triplet GGA specifies glycine in a bacterium, a mushroom and a whale alike. That universality is exactly what common ancestry predicts and what independent origins would not.

Better still, the degree of molecular difference tracks the degree of anatomical difference. Compare the protein cytochrome c across species and humans differ from chimpanzees at essentially no positions, from horses at a dozen or so, and from yeast at many dozens. Nobody arranged that. Two trees built from utterly different data, one from skeletons and one from sequences, keep coming out the same, and agreement between independent methods is the strongest kind of evidence science has.

Evidence four: where things live

Australia has kangaroos, koalas, wombats and Tasmanian devils, and until humans arrived it had almost no placental mammals. Nothing about the Australian climate demands pouches; deserts elsewhere are full of placental rodents. What explains the pattern is history: Australia separated from the other southern landmasses while marsupials were the dominant mammals there, and the lineage diversified in isolation. Island species are typically most closely related to species on the nearest mainland, not to species in the most similar habitat, which is a prediction of descent and not of design.

Evidence five: it happens fast enough to watch

• Antibiotic resistance. Penicillin entered wide clinical use in the 1940s and resistant Staphylococcus aureus was reported within a few years. Methicillin-resistant strains followed methicillin. The pattern repeats for every antibiotic, because a bacterial population contains variants, the drug kills the susceptible ones, and the survivors reproduce in hours.
• The long-term evolution experiment. Richard Lenski has propagated twelve populations of Escherichia coli continuously since 1988, freezing samples so that any generation can be revived and compared with its own ancestors. Around 31,500 generations, one population evolved the ability to use citrate as a carbon source under oxygen, something the species is defined as not doing.
• Pesticide and herbicide resistance in insects and weeds follows the same course, at considerable agricultural cost.

The one humans did on purpose

Darwin opened On the Origin of Species not with wild animals but with pigeon breeders, because artificial selection makes the mechanism impossible to miss. Every breed of dog descends from wolves within roughly the last fifteen thousand years. Cabbage, broccoli, cauliflower, kale, kohlrabi and Brussels sprouts are all the same species, Brassica oleracea, selected by farmers for different organs of the same plant: terminal buds, flower clusters, leaves, stems and lateral buds respectively. If choosing which individuals breed can do that in a few thousand years, differential survival over hundreds of millions is not a modest claim to make.

Common misconceptions

• "Organisms evolve traits because they need them." Variation arises first, without reference to need, and selection acts afterwards. If the useful variant is not present, the population goes extinct instead.
• "Evolution is just a theory." In science a theory is an explanatory framework supported by tested evidence, not a guess. The testable proposal is a hypothesis; a law describes a pattern without explaining it.
• "Humans evolved from chimpanzees." Humans and chimpanzees share a common ancestor that was neither. Cousins, not ancestors.
• "Evolution means progress toward complexity." Tapeworms lost their digestive systems and cave fish lost their eyes, because in those environments the structures cost more than they returned. Simplification is as much an evolutionary outcome as elaboration.
• "Fitness means the healthiest individual." It means the one leaving the most surviving offspring, which is why a peacock's train and a salmon's suicidal spawning run are both fit.

Where this leaves us

Natural selection follows from four conditions, variation, heritability, overproduction and differential reproduction, and once they hold, the change in allele frequencies is arithmetic rather than aspiration. Fitness counts descendants, not strength, and selection has no foresight: the Daphne Major finches did not respond to a drought, they were sorted by one. Its shapes are directional, stabilising and disruptive, with sexual selection producing traits that hurt survival and help mating. The evidence for common descent comes from independent directions that keep agreeing: fossils found where their age predicted, the one bone-two bones-wrist-digits pattern shared by limbs that do different jobs, vestigial pelvises in whales, a genetic code shared across all domains, molecular differences that track anatomical trees, marsupials confined to a landmass that drifted away, and resistance evolving in bacteria and weeds fast enough to be measured in a laboratory notebook.

Sources

1. OpenStax. (2018). 18.1 Understanding evolution. In Biology 2e. openstax.org
2. OpenStax. (2018). 19.3 Adaptive evolution. In Biology 2e. openstax.org
3. National Human Genome Research Institute. (n.d.). Talking glossary of genomic and genetic terms. genome.gov
4. Darwin, C. (1859). On the Origin of Species by Means of Natural Selection. London: John Murray.
5. Grant, P. R., and Grant, B. R. (2014). 40 Years of Evolution: Darwin's Finches on Daphne Major Island. Princeton University Press.

How many carriers, given how many patients?

Cystic fibrosis is a recessive disorder. Among people of Northern European ancestry, roughly one newborn in 2,500 is affected, meaning they carry two copies of a nonfunctional CFTR allele. How many unaffected people carry one copy?

You cannot count them. Carriers are healthy and, before genetic testing, invisible. But you can compute them.

1. Affected individuals are the homozygous recessive class, so q² = 1/2500 = 0.0004.
2. Therefore q = √0.0004 = 0.02.
3. Since there are only two alleles here, p = 1 - q = 0.98.
4. Carriers are the heterozygotes: 2pq = 2(0.98)(0.02) = 0.0392.

About 3.9 percent of the population, or roughly one person in 25, carries a cystic fibrosis allele without knowing it. That figure matches what carrier screening actually finds, which is a good check on a model built from nothing but a square root and a subtraction.

Why this matters: Hardy-Weinberg turns a visible number, the frequency of the rare phenotype, into an invisible one, the frequency of the allele and its hidden carriers.

Where the equations come from

Imagine every allele in a population poured into one pool. A fraction p of them are A, a fraction q are a, and since these are the only options, p + q = 1. Now let gametes combine at random, which is like drawing two alleles from that pool.

	A (p)	a (q)
A (p)	AA: p²	Aa: pq
a (q)	Aa: pq	aa: q²

Adding the cells gives the second equation, p² + 2pq + q² = 1, where p² is the frequency of homozygous dominant individuals, 2pq of heterozygotes, and q² of homozygous recessives. Godfrey Hardy, a mathematician, and Wilhelm Weinberg, a physician, published this independently in 1908, and the point they were making was not the algebra. It was that allele frequencies do not change on their own. Dominant alleles do not become more common merely by being dominant. Something has to act.

Going the other way: from counts to frequencies

When you can distinguish all three genotypes, count alleles directly rather than using the equation. Suppose 500 plants are scored: 320 RR, 160 Rr, 20 rr.

1. Each plant has two alleles, so there are 1000 alleles in total.
2. R alleles: 2(320) + 1(160) = 800. So p = 800/1000 = 0.8.
3. Then q = 0.2.
4. Expected counts if the population is at equilibrium: p²(500) = 0.64(500) = 320; 2pq(500) = 0.32(500) = 160; q²(500) = 0.04(500) = 20.

Observed and expected match exactly. This population is in Hardy-Weinberg equilibrium for this locus.

The five assumptions, which are really a list of the ways evolution happens

Assumption	If it fails	What that mechanism does
No mutation	mutation	creates new alleles; the ultimate source of all variation, but slow per generation
No gene flow	migration	moves alleles between populations, making them more similar
Infinitely large population	genetic drift	random sampling changes frequencies, strongly in small populations
Random mating	non-random mating	changes genotype frequencies without changing allele frequencies
No selection	natural selection	changes allele frequencies non-randomly with respect to phenotype

No real population satisfies all five, and that is exactly why the model is useful. It is a null model: it says what would happen if nothing were happening. When the data depart from it, the departure is evidence, and its pattern points at which assumption failed.

A population that is not at equilibrium, and what that tells you

Now score a different sample of 500 plants: 380 RR, 40 Rr, 80 rr.

1. R alleles: 2(380) + 40 = 800, so p = 0.8 and q = 0.2. Identical allele frequencies to the first population.
2. Expected counts are therefore the same as before: 320, 160, 20.
3. Observed: 380, 40, 80. There are far too many homozygotes of both kinds and far too few heterozygotes.

Run a chi-square test on that, which measures whether observed counts differ from expected by more than sampling noise:

χ² = Σ (observed - expected)² / expected

• (380 - 320)²/320 = 3600/320 = 11.25
• (40 - 160)²/160 = 14400/160 = 90.00
• (80 - 20)²/20 = 3600/20 = 180.00
• Total: χ² = 281.25, with 1 degree of freedom for this test.

The critical value at the 0.05 level with 1 degree of freedom is 3.84. A value of 281 is not close, so the deviation is real. And notice what it means: the allele frequencies are unchanged, so nothing here proves natural selection. A heterozygote shortage of exactly this kind is the signature of non-random mating, typically inbreeding, or of a sample that is really two separate subpopulations pooled together.

Balancing selection: why a harmful allele stays common

The sickle cell allele of the beta-globin gene causes severe disease in homozygotes. Selection against it should be intense, and yet in parts of sub-Saharan Africa its frequency reaches roughly 0.1. Hardy-Weinberg makes the puzzle quantitative: at q = 0.1, the affected fraction is q² = 0.01, one percent, while carriers are 2pq = 2(0.9)(0.1) = 0.18, eighteen percent.

The resolution is that the three genotypes face different environments. Homozygotes for the normal allele are fully susceptible to falciparum malaria; homozygotes for the sickle allele have sickle cell disease; heterozygotes have substantial protection against severe malaria and only mild symptoms. Where malaria is endemic, the heterozygote has the highest fitness, and selection therefore preserves both alleles rather than eliminating either. This is heterozygote advantage, and it is the clearest case in human genetics of selection maintaining variation rather than removing it. The geographic overlap between historical malaria distribution and sickle allele frequency is the evidence.

Drift: evolution with no fitness differences at all

Genetic drift is change in allele frequencies by pure sampling accident. Flip a fair coin ten times and you will often not get five heads; sample twenty alleles from a population and the frequencies in the next generation will wander. The smaller the population, the wilder the wandering.

• Bottleneck. Northern elephant seals were hunted down to a few dozen animals in the 1890s. The population has since recovered to well over a hundred thousand, but it carries almost no genetic variation, because the variation that was not in those few dozen survivors is simply gone.
• Founder effect. The Old Order Amish of Lancaster County descend from a small group of eighteenth-century immigrants. Ellis-van Creveld syndrome, a rare recessive form of dwarfism, occurs among them at a frequency far above its frequency elsewhere, because one founder happened to carry the allele.

Drift is not selection. It has no direction and favours nothing. It can drive a beneficial allele to extinction and fix a mildly harmful one, and in small populations it routinely overpowers weak selection.

What equilibrium does and does not prove

A population matching Hardy-Weinberg proportions is not thereby proved to be free of evolution, for a reason worth understanding: one generation of random mating restores the expected genotype proportions no matter how distorted they were before. So genotype frequencies can look perfectly ordinary in a population whose allele frequencies are shifting steadily under selection.

The real evidence for evolution is a change in allele frequency across generations. Genotype proportions in a single generation test something narrower: whether mating is random and the population is unified.

Common misconceptions

• "Dominant alleles become more common over time." Dominance is about how alleles are expressed in heterozygotes, not about their frequency. This is the point Hardy and Weinberg were making in 1908, against exactly this claim.
• "2pq is the frequency of Aa, so it should be pq." There are two ways to make a heterozygote, A from the mother and a from the father or the reverse, so the two pq cells add.
• "q is the frequency of the recessive phenotype." q² is the phenotype frequency; q is the allele frequency, its square root.
• "Genetic drift is a kind of selection." It is random sampling. It has no relationship to fitness and no direction.
• "A population at Hardy-Weinberg equilibrium is not evolving." A single generation of random mating restores the proportions regardless of what selection is doing to the allele frequencies underneath.

Putting it together

Hardy-Weinberg is two equations, p + q = 1 and p² + 2pq + q² = 1, and one idea: allele frequencies stay put unless something moves them. Working from a phenotype frequency, take the square root to get q and you can compute the carriers you cannot see, which is how one affected newborn in 2,500 implies one carrier in 25 for cystic fibrosis. Working from genotype counts, count alleles directly. The five assumptions are a catalogue of the mechanisms of evolution, so a departure from the expected proportions is evidence, and its pattern identifies the culprit: a heterozygote deficit with unchanged allele frequencies points at non-random mating rather than selection, as a chi-square value of 281 against a critical value of 3.84 confirms. Selection does not always remove variation, as sickle cell heterozygote advantage in malarial regions shows, and drift changes frequencies with no reference to fitness at all, which is why bottlenecked elephant seals and founder populations look the way they do.

Sources

1. OpenStax. (2018). 19.1 Population evolution. In Biology 2e. openstax.org
2. OpenStax. (2018). 19.2 Population genetics. In Biology 2e. openstax.org
3. Nature Education. (2008). The Hardy-Weinberg principle. Scitable. nature.com
4. Wikipedia. (n.d.). Hardy-Weinberg principle. en.wikipedia.org

A salamander that is one species at one end of the ring and two at the other

The salamander Ensatina eschscholtzii lives in the forested hills that ring California's Central Valley. It cannot live in the dry valley floor, so its range is a horseshoe: one arm runs down the coastal mountains, the other down the Sierra Nevada, and the two arms meet again in the south.

Follow the ring in either direction and each population interbreeds with its neighbours. The colour changes gradually, the genetics changes gradually, and nowhere along the way is there a line to draw. But where the two ends come back together in Southern California, the coastal and inland forms live alongside one another and largely do not interbreed. By the standard test they are two species. By the same test applied one population at a time, they are one.

Nothing is wrong with the salamanders. The awkwardness belongs to the word "species", which was built to label the endpoints of a process and is being asked to label the middle of it.

What a species is, and the four ways of saying it

The biological species concept, associated with Ernst Mayr, defines a species as a group of populations whose members actually or potentially interbreed in nature and are reproductively isolated from other such groups. It is the most useful definition for thinking about how species form, and it fails in four situations.

• Asexual organisms. Bacteria do not interbreed at all, so the criterion says nothing.
• Fossils. You cannot test whether two extinct populations could have interbred.
• Ring species and hybrid zones. As above, reproductive isolation can be partial and geographically graded.
• Species that do hybridise. Many oak and canid species produce fertile hybrids and remain distinct anyway.

Practical alternatives exist: the morphological concept, based on measurable form, which is what museum work and palaeontology use; the phylogenetic concept, the smallest group sharing a common ancestor and diagnosable by shared characters; and the ecological concept, defined by the niche a population occupies. Each is a tool with a domain of use. None of them is a discovery of what species truly are, because the underlying reality is a continuum.

The point: speciation is the gradual acquisition of reproductive isolation, so a sharp definition is asking for something the process does not supply.

The barriers, before and after fertilisation

Type	Barrier	Example
Prezygotic	Habitat isolation	two garter snake species in the same region, one mainly aquatic and one terrestrial
Prezygotic	Temporal isolation	species that breed in different seasons, or flower weeks apart
Prezygotic	Behavioural isolation	firefly species with distinct flash patterns; birdsong that only conspecific females answer
Prezygotic	Mechanical isolation	flowers shaped for one pollinator, so pollen is never delivered elsewhere
Prezygotic	Gametic isolation	sea urchins releasing gametes into open water, where surface recognition proteins allow fertilisation only within a species
Postzygotic	Reduced hybrid viability	hybrid embryos that develop poorly and rarely survive
Postzygotic	Reduced hybrid fertility	the mule: a horse has 64 chromosomes and a donkey 62, so the mule's 63 cannot pair evenly in meiosis
Postzygotic	Hybrid breakdown	first-generation hybrids are fine, but their offspring are weak or sterile

Prezygotic barriers are cheaper, because no gametes are wasted, and selection tends to strengthen them where two diverging populations meet. That strengthening is called reinforcement.

Allopatric speciation: split the range first

Allopatric speciation happens when a geographic barrier divides a population, gene flow stops, and the two halves diverge under different selection pressures and independent drift until they are isolated even if reunited.

The clearest natural experiment is the Isthmus of Panama, which closed roughly three million years ago and turned one ocean into two. Snapping shrimp of the genus Alpheus now exist as pairs of closely related species, one on the Pacific side and one on the Caribbean side, with each pair more closely related to its partner across the isthmus than to other shrimp on its own side. When the pairs are brought together in the laboratory, the more anciently separated pairs are the less willing to mate, which is what a gradual accumulation of isolation predicts.

Sympatric speciation: no barrier required

Sympatric speciation happens without geographic separation, and it is rarer but not exotic.

• Polyploidy in plants is the common route. An error in cell division doubles the chromosome number, and the tetraploid individual cannot produce viable offspring with its diploid parents, since a triploid hybrid cannot pair chromosomes evenly. It can, however, self-pollinate or breed with other tetraploids. Reproductive isolation appears in a single generation. Bread wheat is hexaploid, carrying three ancestral genomes, and a large share of flowering plant species have polyploidy somewhere in their history.
• Host race formation. The apple maggot fly, Rhagoletis pomonella, fed on native hawthorn fruit in North America until apples were introduced. In the mid-1800s a population began using apples, which ripen earlier. Since these flies mate on their host fruit, and the two hosts ripen at different times, the apple and hawthorn populations now mate largely apart while living in the same orchards. Gene flow between them is reduced but not zero, which is what speciation in progress looks like.

Slow and steady, or long stasis and sudden change

Darwin pictured gradual, more or less continuous change. In 1972 Niles Eldredge and Stephen Jay Gould pointed out that the fossil record more often shows long periods of little change punctuated by rapid transitions, and proposed punctuated equilibrium as the typical pattern. The debate is about tempo, not about mechanism: both descriptions are natural selection acting on populations, and both patterns occur. Rapid, in geological terms, can still mean tens of thousands of years.

Adaptive radiation

When a lineage reaches unoccupied ecological space, it can diversify quickly into many species with different roles. The Galapagos finches are the textbook case; the Hawaiian silversword alliance, a group of plants ranging from mat-forming rosettes to trees and vines all descended from a single Californian tarweed ancestor, is a more dramatic one; and the cichlid fishes of the African Great Lakes have produced hundreds of species in the geological blink of an eye. The common ingredients are opportunity, isolation and time.

How to read a tree without misreading it

A phylogenetic tree is a hypothesis about ancestry. The tips are the taxa being compared, usually living ones. Each internal node is a common ancestor, and each split is a divergence event. A clade is a node plus everything descended from it, which is what "monophyletic" means.

Three rules stop most misreadings:

1. Relatedness is measured by the most recent common ancestor, not by distance along the page. Two tips drawn side by side can be less closely related than tips at opposite ends.
2. Branches rotate freely at a node, like a mobile hanging from the ceiling. Swapping the two sides of a fork produces the same tree.
3. Living taxa are not ancestral to each other. A tree showing humans next to chimpanzees says they share an ancestor, not that one descends from the other.

Trees are built from shared derived characters, features that arose in a common ancestor and are inherited by all its descendants. An outgroup, a taxon known to have branched off earlier, tells you which state of a character is ancestral. Where the data conflict, the usual criterion is parsimony: prefer the tree requiring the fewest evolutionary changes. Conflicts arise because of homoplasy, the independent evolution of the same feature in separate lineages, which is why wings alone would group bats with birds.

Building one, from a table

Score five animals for four characters, using the lamprey as the outgroup.

Taxon	Vertebral column	Jaws	Amniotic egg	Hair and mammary glands
Lamprey	yes	no	no	no
Tuna	yes	yes	no	no
Frog	yes	yes	no	no
Lizard	yes	yes	yes	no
Mouse	yes	yes	yes	yes

Read the columns from left to right and the nesting builds itself. All five have a vertebral column, so that character defines the whole group and cannot resolve anything inside it. Jaws separate the lamprey from the other four. The amniotic egg separates lizard and mouse from tuna and frog. Hair separates the mouse. The resulting tree branches in that order, and every clade is defined by the character that appeared at its base. Distinguishing frog from tuna would need another character, such as limbs with digits, which is exactly why Tiktaalik mattered in the previous lesson.

Common misconceptions

• "Species are natural, sharply bounded kinds." The boundaries are as fuzzy as the process that makes them, which is why four working definitions coexist.
• "Speciation needs a mountain range or an ocean." Polyploidy achieves it in one generation with no geography at all.
• "Taxa next to each other on a tree are the most closely related." Only the most recent common ancestor decides that; branches rotate freely.
• "A tree shows which living species came from which." All the tips are contemporary. Ancestors are the internal nodes, and they are usually extinct.
• "Hybrids are always sterile." Many are not, which is one of the reasons the biological species concept has exceptions. The mule is memorable, not universal.

The short version

A species is easiest to define as a reproductively isolated group, and that definition breaks down exactly where speciation is happening, as the Ensatina ring around the Central Valley shows. Isolation comes from prezygotic barriers, which prevent mating or fertilisation, and postzygotic ones, which reduce hybrid viability or fertility as in the mule's 63 unpairable chromosomes. Allopatric speciation follows a geographic split, with the snapping shrimp of the Isthmus of Panama as a dated natural experiment; sympatric speciation happens without one, most commonly through polyploidy in plants and visibly in progress in the apple and hawthorn races of Rhagoletis. Tempo varies between gradual change and long stasis broken by rapid transitions. On a phylogeny, relatedness is set by the most recent common ancestor rather than by position on the page, clades are defined by shared derived characters, an outgroup roots the tree, and parsimony chooses between competing hypotheses when convergent evolution muddies the data.

Sources

1. OpenStax. (2018). 18.2 Formation of new species. In Biology 2e. openstax.org
2. OpenStax. (2018). 18.3 Reconnection and speciation rates. In Biology 2e. openstax.org
3. OpenStax. (2018). 20.2 Determining evolutionary relationships. In Biology 2e. openstax.org
4. OpenStax. (2018). 20.1 Organizing life on Earth. In Biology 2e. openstax.org
5. Mayr, E. (1942). Systematics and the Origin of Species from the Viewpoint of a Zoologist. Columbia University Press.

A hundred metres of water, lifted with no pump

The tallest known coast redwood stands about 116 metres. Water enters at the roots and leaves through pores in the topmost needles, and there is no pump anywhere in the tree. A suction pump could not do it in any case: atmospheric pressure can raise a column of water only about 10 metres before it breaks.

What lifts the column is evaporation at the top and the fact that water molecules cling to one another hard enough to be pulled as a thread. Each molecule leaving a leaf tugs on the one behind it, and the tug is transmitted down a continuous chain of water through microscopic xylem tubes to the roots. The chain is under tension, like a rope, and it does not snap.

That chain is the subject of this lesson. Almost every property of water that biology depends on comes from one structural fact about a molecule with three atoms.

Two hydrogens, an oxygen, and an uneven share

Water is H₂O: an oxygen atom covalently bonded to two hydrogens. Covalent bonds share electrons, but they do not share them fairly. Oxygen has an electronegativity of 3.44 on the Pauling scale against hydrogen's 2.20, so the shared electrons spend most of their time near the oxygen.

Two consequences follow. The oxygen carries a partial negative charge and each hydrogen a partial positive one, making the bond polar covalent. And the molecule is bent rather than straight, with an H-O-H angle of about 104.5 degrees, so the two positive ends sit on the same side. If water were linear, the charges would cancel and the molecule would be nonpolar. It is not, so water is a small molecule with a distinct positive end and a distinct negative end.

Put many of them together and the positive hydrogen of one is attracted to the negative oxygen of another. That attraction is a hydrogen bond. It is weak, roughly a twentieth of the strength of the covalent bonds inside the molecule, and it lasts only picoseconds before breaking and reforming elsewhere. But each water molecule can make up to four of them, so at any instant liquid water is a shifting network rather than a collection of separate particles.

The core of it: polarity plus hydrogen bonding explains every property below. There is really only one idea in this lesson.

Cohesion, adhesion, and the redwood

Cohesion is water sticking to water; adhesion is water sticking to other polar surfaces. Together they explain the tree. Adhesion to the walls of the xylem stops the column slipping back, cohesion holds it together as a thread, and evaporation from the leaves, called transpiration, supplies the pull.

The same cohesion produces surface tension, the resistance of a water surface to being broken. Molecules at the surface have no neighbours above them, so their hydrogen bonding is all sideways and downward, pulling the surface taut. A water strider stands on it without wetting its feet, and a paper clip laid flat will float on water it is eight times denser than.

Specific heat: why coasts have mild summers

Raising the temperature of one gram of water by one degree Celsius takes 4.184 joules, several times what most solids require, and about five times what dry rock or sand takes. Heating water means breaking hydrogen bonds, and the energy that goes into breaking them does not go into making molecules move faster, which is what temperature measures.

The biological payoff is thermal stability. Oceans and lakes warm and cool slowly, so aquatic habitats are among the most temperature-stable places on Earth, and coastal regions have narrower seasonal swings than inland ones at the same latitude. Inside an organism, a body that is largely water resists sudden temperature change from a metabolic burst or a cold morning.

Heat of vaporisation: why sweating works

Turning one gram of liquid water at 100 degrees Celsius into vapour takes about 2,260 joules, and evaporation at body temperature costs even more per gram. Every molecule that leaves must break all its hydrogen bonds at once, and the molecules with the most energy leave first.

The ones left behind are therefore cooler. That is evaporative cooling, and it is why sweating cools a mammal, why panting cools a dog, and why transpiration keeps a leaf in full sun from cooking. It is also why humid air is so much harder to tolerate: when the surrounding air is already saturated, sweat does not evaporate, and unevaporated sweat cools nothing.

Ice floats, and lakes do not freeze solid

Almost every substance is denser as a solid than as a liquid, because cooling packs particles closer. Water reaches its maximum density at about 3.98 degrees Celsius and then becomes less dense as it cools further and freezes. Ice is about nine percent less dense than liquid water, so it floats.

The reason is geometric. In ice, each molecule is locked into four hydrogen bonds in a rigid open lattice, and holding those bonds at fixed angles takes more room than the jostling, constantly rearranging liquid does.

Think about what would happen otherwise. Ice forming at the surface would sink, exposing more water to freeze and sink in turn, until a lake was solid from the bottom up and everything in it was dead. Instead a floating sheet insulates the water below, which stays liquid at around 4 degrees. Fish overwinter. Polar ecosystems exist at all.

The solvent, and the hydrophobic effect

Drop table salt into water. The partial negative oxygens surround each Na⁺ ion and the partial positive hydrogens surround each Cl^- ion, forming hydration shells that pull the crystal apart ion by ion. Anything charged or polar dissolves the same way, and such substances are called hydrophilic. Sugars, amino acids and nucleic acids all qualify.

Nonpolar substances such as oils cannot form hydrogen bonds, so water molecules next to them lose bonding partners. The system settles by minimising the contact area, which forces oil molecules together into droplets. This is the hydrophobic effect, and note carefully what it is: not an attraction between oil molecules, but the exclusion of them by water reorganising itself. It is the reason a lipid bilayer assembles spontaneously and the main force folding a protein so that its greasy residues end up buried in the middle. Two of the largest structures in cell biology are built by water refusing to associate with something.

pH, and what one unit means

Water ionises slightly: H2O yields H+ and OH-. In pure water at 25 degrees the hydrogen ion concentration is 10^-7 moles per litre, and pH is defined as the negative logarithm of that concentration, hence a pH of 7.

Because the scale is logarithmic, each unit is a factor of ten. A solution at pH 3 has 10^-3 moles per litre of hydrogen ions, which is one hundred times the concentration at pH 5, not two units' worth. Some real values:

Fluid	Approximate pH
Stomach contents	1.5 to 3.5
Lemon juice	about 2
Human blood	7.35 to 7.45
Seawater (surface)	about 8.1
Household bleach	about 12.5

Blood's range is narrow because enzymes are proteins, and a protein's shape depends on charged side chains whose ionisation state changes with pH. Move blood pH by a few tenths and enzymes begin to lose their shapes; below about 6.8 or above about 7.8 the consequences are life-threatening.

Buffers, and what happens when one is overwhelmed

A buffer is a pair of molecules that can donate or accept hydrogen ions, holding pH nearly constant as acid or base is added. Blood uses the bicarbonate system:

CO2 + H2O is in equilibrium with H2CO3, which is in equilibrium with H+ + HCO3-

Add acid and bicarbonate mops up hydrogen ions by shifting the equilibrium left. Add base and carbonic acid releases them by shifting it right. Because carbon dioxide is exhaled, breathing rate adjusts the whole system within seconds, which is why hyperventilation raises blood pH and holding your breath lowers it.

The ocean runs the same chemistry, and it is being pushed. Carbon dioxide dissolving into seawater produces hydrogen ions, and surface ocean pH has fallen from roughly 8.2 to about 8.1 since the beginning of the industrial era. That sounds small until you remember the logarithm: it is roughly a 30 percent increase in hydrogen ion concentration. The extra hydrogen ions react with carbonate ions, reducing the carbonate available to organisms building shells and skeletons from calcium carbonate, which is the mechanism behind concern for corals, pteropods and shellfish larvae.

Common misconceptions

• "Hydrogen bonds are a kind of covalent bond." They are attractions between molecules, about a twentieth as strong, and they break and reform constantly at room temperature. That weakness is what makes them useful.
• "Water is the universal solvent, so it dissolves everything." It dissolves polar and charged substances. Nonpolar substances it actively excludes, which is a feature rather than a failure.
• "Oil molecules attract each other, which is why they clump." Water reorganising to preserve its own hydrogen bonding pushes them together. The hydrophobic effect is about water, not about oil.
• "pH 6 is only slightly more acidic than pH 7." It is ten times more concentrated in hydrogen ions, and pH 5 is a hundred times.
• "Ice floats because it has air in it." Pure ice floats. Its open hydrogen-bonded lattice takes up more space than the liquid arrangement.

What to carry forward

Water is bent and its bonds are polar, so each molecule has a positive and a negative end and can hydrogen bond to as many as four neighbours. Everything else follows. Cohesion and adhesion draw a continuous thread of water 116 metres up a redwood and hold a strider on a pond surface. A specific heat of 4.184 joules per gram per degree buffers organisms and coastlines against temperature swings, and a heat of vaporisation near 2,260 joules per gram makes sweating and transpiration effective coolants. An open lattice makes ice about nine percent less dense than liquid water, so lakes freeze from the top and life continues underneath. Polar and charged substances dissolve in hydration shells, while nonpolar ones are squeezed together by the hydrophobic effect, which is what assembles membranes and folds proteins. And because water ionises, pH matters: the scale is logarithmic, blood is held between 7.35 and 7.45 by the bicarbonate buffer, and the same chemistry running in the ocean has already raised surface hydrogen ion concentrations by about 30 percent.

Sources

1. OpenStax. (2018). 2.2 Water. In Biology 2e. openstax.org
2. OpenStax. (2018). 2.1 Atoms, isotopes, ions, and molecules. In Biology 2e. openstax.org
3. National Oceanic and Atmospheric Administration. (n.d.). What is ocean acidification? National Ocean Service. oceanservice.noaa.gov
4. Wikipedia. (n.d.). Properties of water. en.wikipedia.org

One amino acid out of 146

The beta chain of human haemoglobin is 146 amino acids long. In the sickle cell allele, position 6 carries valine instead of glutamate. That is the entire difference: one residue in one of two chains, in a protein of several hundred residues, in a genome of three billion base pairs.

Glutamate is charged and sits comfortably on the wet outside of a folded protein. Valine is greasy and does not. The substitution creates a small hydrophobic patch on the surface of every haemoglobin molecule, and when oxygen levels fall, those patches stick to matching pockets on neighbouring molecules. The haemoglobin polymerises into stiff fibres, the red cell deforms into a crescent, and the cell jams capillaries.

Vernon Ingram showed in 1957 that this was the difference, and it was the first time a human disease was traced to a specific change in a specific protein. It is also the cleanest demonstration in biology that shape determines function and that chemistry determines shape. This lesson is about the four families of molecules that cells build, and the catalysts that build them.

Two reactions, four families

Cells assemble large molecules from small repeating units. A monomer is the unit; a polymer is the chain. Two reactions do all the work.

• Dehydration synthesis joins two monomers and releases a water molecule. One partner gives up a hydroxyl group, the other a hydrogen.
• Hydrolysis is the reverse: a water molecule is added across the bond and the chain splits. This is what digestion is.

Both reactions require enzymes in practice, and the same pair of reactions handles carbohydrates, proteins and nucleic acids alike.

Remember: synthesis removes water, breakdown adds it. Every polymer in the cell obeys this.

Carbohydrates: energy, and also bricks

The monomer is a monosaccharide, typically with the formula C6H12O6. Glucose, fructose and galactose all share that formula and differ only in how the atoms are arranged, which is what makes them isomers and why fructose tastes sweeter than glucose despite being chemically equivalent by mass.

Two monosaccharides joined by dehydration synthesis form a disaccharide linked by a glycosidic bond: glucose plus fructose gives sucrose, glucose plus galactose gives lactose, glucose plus glucose gives maltose. Long chains are polysaccharides, and here the arrangement of a single bond has enormous consequences.

Polysaccharide	Structure	Role
Starch	glucose in alpha linkages, straight or lightly branched	energy storage in plants
Glycogen	glucose in alpha linkages, heavily branched	energy storage in animals, in liver and muscle
Cellulose	glucose in beta linkages, straight, hydrogen-bonded into fibres	plant cell walls; the most abundant organic polymer on Earth
Chitin	modified glucose in beta linkages	arthropod exoskeletons and fungal cell walls

Starch and cellulose are both pure glucose. The difference is the orientation of the linkage, alpha in one and beta in the other, and that orientation is why humans digest bread and cannot digest paper. Our amylases fit alpha linkages only. Cattle survive on grass because microorganisms in their rumen make the enzyme that we do not, and termites do the same trick with gut symbionts.

Lipids: the family defined by what they do not do

Lipids are grouped by a property rather than a shared monomer: they are nonpolar and do not dissolve in water. Three types matter here.

• Triglycerides: one glycerol joined by ester bonds to three fatty acids. Saturated fatty acids have no double bonds, pack tightly and are solid at room temperature, like butter. Unsaturated ones carry one or more double bonds, and a double bond in the natural cis configuration puts a permanent kink in the chain, preventing tight packing, which is why olive oil pours.
• Phospholipids: glycerol with two fatty acid tails and a charged phosphate head. The head is hydrophilic and the tails are hydrophobic, making the molecule amphipathic. Drop enough of them into water and they arrange themselves into a bilayer, tails inward, with no assistance. The membrane in every cell you contain assembled itself for the reason described in the previous lesson.
• Steroids: four fused carbon rings. Cholesterol wedges between phospholipid tails and moderates membrane fluidity, and it is also the precursor from which testosterone, oestrogen and cortisol are built.

Per gram, fats store roughly twice the energy of carbohydrates, because their carbons are more reduced, carrying more hydrogens and therefore more electrons to be harvested.

Proteins: twenty letters, four levels

An amino acid has a central carbon bonded to an amino group, a carboxyl group, a hydrogen and a variable R group. There are 20 R groups in the standard set, and they range from charged to polar to hydrophobic to structurally peculiar, which is the whole reason proteins can do so many jobs. Dehydration synthesis joins the carboxyl of one to the amino group of the next, forming a peptide bond.

Level	What it is	Held together by
Primary	the sequence of amino acids	peptide bonds
Secondary	local alpha helices and beta pleated sheets	hydrogen bonds along the backbone
Tertiary	the overall three-dimensional fold of one chain	R group interactions: hydrophobic clustering, hydrogen bonds, ionic bonds, disulfide bridges
Quaternary	two or more folded chains assembled together	the same R group interactions, between subunits

Haemoglobin has all four: a specific sequence, helices, a folded globin subunit, and four subunits assembled around their haem groups. Now return to sickle cell. The change is at the primary level, one letter. It alters the tertiary surface, adding a hydrophobic patch. That changes the quaternary behaviour, letting molecules stack. And the consequence is a cell shape you can see down a microscope. Four levels, one substitution.

Denaturation is the loss of shape without breaking peptide bonds. Heat shakes apart the hydrogen and ionic bonds holding the fold; a pH change alters which R groups are charged, so ionic bonds that were attracting begin repelling. An egg white turning opaque and solid in a pan is albumin denaturing, and it does not go back.

Nucleic acids: the instructions

The monomer is a nucleotide: a five-carbon sugar, a phosphate group, and a nitrogenous base. In DNA the sugar is deoxyribose and the bases are adenine, thymine, guanine and cytosine; in RNA the sugar is ribose and uracil replaces thymine. Nucleotides join through their sugars and phosphates into a backbone with a direction, conventionally described as running from the 5 prime end to the 3 prime end.

DNA is double-stranded and the two strands run in opposite directions, which is what antiparallel means. The bases pair by hydrogen bonding, A with T through two bonds and G with C through three, which is why GC-rich DNA takes more heat to separate. RNA is usually single-stranded and folds back on itself.

Enzymes: what a catalyst does, and what it cannot do

Every reaction has an energy barrier, the activation energy, that must be paid before it can proceed. Enzymes lower that barrier, typically by orders of magnitude, by binding the substrate in an orientation that strains the bonds to be broken and stabilises the awkward transition state.

What an enzyme does not do is change the free energy difference between reactants and products, or shift the equilibrium position. It changes only how fast equilibrium is reached. An enzyme cannot make an energetically unfavourable reaction favourable; the cell achieves that by coupling it to ATP hydrolysis, which is the next module's business.

The scale is worth appreciating. Carbonic anhydrase, which converts carbon dioxide and water into bicarbonate, can process on the order of a million molecules per second. Without it, the reaction in blood would be far too slow to clear the carbon dioxide your tissues produce.

The substrate binds at the active site, a pocket whose shape and charge distribution fit that substrate. The old lock-and-key picture is close but static; the accepted description is induced fit, in which the enzyme adjusts its shape slightly as the substrate arrives, tightening the grip and positioning catalytic groups.

What speeds an enzyme up and what shuts it down

• Temperature. Rate rises with temperature because collisions become more frequent and energetic, up to an optimum. Beyond it, the protein denatures and the rate collapses. The curve is therefore asymmetric: a gentle rise and a cliff.
• pH. Each enzyme has an optimum set by the charges its active site needs. Pepsin works in the stomach at about pH 2; trypsin works in the small intestine at about pH 8. Neither functions in the other's environment.
• Substrate concentration. Rate rises with substrate until every active site is occupied, then levels off. That plateau is saturation, and past it, adding substrate does nothing; only adding enzyme helps.
• Competitive inhibitors resemble the substrate and occupy the active site. More substrate can outcompete them. Statins work this way, resembling the substrate of HMG-CoA reductase and blocking the cholesterol synthesis pathway. Penicillin acts on a bacterial enzyme that cross-links cell walls by mimicking its natural substrate.
• Noncompetitive and allosteric inhibitors bind elsewhere on the enzyme and change the active site's shape from a distance. Adding substrate does not help, because the problem is not occupancy.
• Feedback inhibition is the cell's standard control loop: the end product of a pathway binds allosterically to an early enzyme in that pathway and switches it off, so production stops when there is enough.
• Cofactors and coenzymes. Many enzymes need a non-protein helper: a metal ion such as the zinc at the heart of carbonic anhydrase, or an organic coenzyme, many of which are made from vitamins. This is a large part of why vitamin deficiencies cause such varied damage.

Common misconceptions

• "Enzymes are used up in the reaction." They are regenerated unchanged and can run millions of cycles. That is what makes them catalysts.
• "Enzymes make reactions happen that otherwise could not." They lower the activation energy of reactions that are already energetically favourable. An unfavourable reaction needs energy coupling, not a better catalyst.
• "Denatured proteins have broken peptide bonds." The sequence survives; the folding does not. A denatured protein has full primary structure and no function.
• "Starch and cellulose are different sugars." Both are glucose polymers. Only the linkage orientation differs, which is why one is food and the other is fibre.
• "Saturated and unsaturated fats differ in how many carbons they have." They differ in double bonds. A cis double bond kinks the chain and stops tight packing, which is why unsaturated fats are liquid at room temperature.

Pulling it together

Cells build polymers by removing water and break them by adding it, and this one pair of reactions serves carbohydrates, proteins and nucleic acids alike. Carbohydrate function turns on linkage geometry, so starch feeds us and chemically identical cellulose does not. Lipids are defined by insolubility: triglycerides store energy densely, phospholipids are amphipathic and assemble into bilayers unaided, and steroids tune membranes and become hormones. Proteins are built from twenty amino acids whose R groups drive folding through four structural levels, and the sickle cell substitution shows how a single residue propagates from primary sequence to a visible cell shape; denaturation destroys the fold without touching the sequence. Nucleic acids store information in a directional, antiparallel, base-paired form. And enzymes speed all of it up by lowering activation energy without changing the equilibrium, subject to temperature, pH, saturation, and inhibitors that either compete for the active site or reshape it from elsewhere, with feedback inhibition as the cell's standard off switch.

Sources

1. OpenStax. (2018). 3.1 Synthesis of biological macromolecules. In Biology 2e. openstax.org
2. OpenStax. (2018). 3.4 Proteins. In Biology 2e. openstax.org
3. OpenStax. (2018). 6.5 Enzymes. In Biology 2e. openstax.org
4. OpenStax. (2018). 3.3 Lipids. In Biology 2e. openstax.org
5. Wikipedia. (n.d.). Enzyme. en.wikipedia.org

Cork, 1665

Robert Hooke put a shaving of cork under a microscope of his own construction and drew what he saw for Micrographia: a honeycomb of tiny empty boxes. He called them cells, after the small rooms monks slept in. What he was actually looking at were the walls of long-dead plant tissue, with nothing living left inside, which makes the founding observation of cell biology an image of empty containers.

It took two centuries to assemble the rest. Matthias Schleiden concluded in 1838 that plants are made of cells, Theodor Schwann extended it to animals in 1839, and in 1855 Rudolf Virchow added the part that connects cell biology to evolution: every cell comes from a pre-existing cell. There is no spontaneous generation of cells, which means every cell in your body is part of an unbroken line of divisions reaching back billions of years.

Why cells are small, in numbers

Everything a cell needs enters through its surface, and everything it must expel leaves the same way. Meanwhile the demand for supplies is set by the volume of cytoplasm to be served. Those two quantities do not scale together.

Cube side	Surface area	Volume	Surface area to volume
1 unit	6	1	6 : 1
2 units	24	8	3 : 1
4 units	96	64	1.5 : 1

Double the size and you quadruple the surface but multiply the volume by eight, so the ratio halves. A cell that grows too large cannot supply its interior fast enough, which is why most cells sit between roughly 1 and 100 micrometres across and why growth beyond a threshold triggers division rather than further expansion.

Cells that need more exchange cheat by changing shape rather than size. Intestinal lining cells carry microvilli, thousands of finger-like projections that multiply absorptive surface many times over without adding volume. A neuron is thin and enormously long for the same reason.

Why this matters: the surface area to volume argument explains cell size, microvilli, the folding of mitochondrial cristae and the stacking of thylakoids. It is one idea reused four times.

Two ways to build a cell

Feature	Prokaryotic	Eukaryotic
Nucleus	none; DNA in a nucleoid region	membrane-bound nucleus
DNA	usually one circular chromosome, plus plasmids	multiple linear chromosomes with histones
Membrane organelles	none	many
Ribosomes	70S	80S in the cytosol, 70S inside mitochondria and chloroplasts
Typical size	1 to 5 micrometres	10 to 100 micrometres
Cell wall	peptidoglycan in bacteria	cellulose in plants, chitin in fungi, none in animals

Prokaryotes are not primitive failures. They are the most abundant and metabolically diverse organisms on the planet, and their simplicity is efficient: with no internal membranes to cross, a bacterium can divide in twenty minutes.

Following one protein out of a cell

The best way to see the endomembrane system is to follow something through it. Take insulin, made by beta cells in the pancreas and exported into the bloodstream.

1. Nucleus. The insulin gene is transcribed into messenger RNA, which leaves through a nuclear pore.
2. Ribosome and rough ER. Translation begins on a free ribosome, but the first stretch of the new chain is a signal sequence that acts as an address label. It directs the ribosome to the rough endoplasmic reticulum, and the growing chain is threaded into the ER lumen. The rough ER is rough because it is studded with ribosomes doing exactly this.
3. Folding and first modifications. Inside the ER the chain folds, disulfide bridges form, and sugars may be attached. Misfolded proteins are detected here and destroyed rather than shipped.
4. Transport vesicle. A piece of ER membrane buds off enclosing the cargo and travels to the Golgi.
5. Golgi apparatus. Vesicles fuse with the receiving face, and the contents are modified in stages as they progress toward the shipping face: sugars trimmed and added, and in insulin's case a segment cut out to produce the mature hormone. The Golgi also sorts, tagging each product for its destination.
6. Secretory vesicle and exocytosis. A vesicle buds from the shipping face, waits near the plasma membrane, and when blood glucose rises, fuses with it, releasing insulin outside.

That is a production line with quality control, staged modification, addressing and dispatch, and it is why the compartments exist at all.

What compartments buy you

• Incompatible conditions in one cell. Lysosomes hold digestive enzymes that work best near pH 5 and pump hydrogen ions inward to maintain it, while the surrounding cytosol sits near pH 7.2. Both exist a membrane apart. It also means a leaking lysosome does limited damage, because its enzymes are far from their optimum outside.
• Concentration. Confining enzymes and substrates to a small volume raises their concentrations and therefore their reaction rates.
• Membrane as workbench. Electron transport chains are embedded in membranes, so more membrane means more capacity. This is why mitochondria fold their inner membrane into cristae and chloroplasts stack thylakoid discs.
• Gradients. A gradient is only useful if something contains it. The proton gradient that makes ATP exists because a membrane holds it.

The parts list

Structure	Function
Nucleus	houses DNA; double membrane with pores controlling traffic
Nucleolus	site of ribosome subunit assembly
Ribosome	protein synthesis; free in cytosol or bound to rough ER
Rough ER	synthesis and folding of proteins destined for membranes, organelles or export
Smooth ER	lipid and steroid synthesis; detoxification in liver cells; calcium storage in muscle
Golgi apparatus	modification, sorting and dispatch of proteins and lipids
Lysosome	hydrolytic digestion of macromolecules, damaged organelles and engulfed material
Peroxisome	breaks down fatty acids and detoxifies hydrogen peroxide using catalase
Mitochondrion	cellular respiration and most ATP production; double membrane with cristae
Chloroplast	photosynthesis; double membrane plus internal thylakoid membranes
Central vacuole	in plants, stores water and solutes and provides turgor pressure
Cytoskeleton	microfilaments of actin for shape and movement, intermediate filaments for tension, microtubules for transport tracks and for the spindle
Cilia and flagella	motility, built on a 9 plus 2 arrangement of microtubules

Two organelles that used to be free-living

Mitochondria and chloroplasts are odd in ways that a purely internal origin does not explain. Lynn Margulis argued in 1967 that they descend from free-living prokaryotes engulfed by an ancestral cell and retained rather than digested. The evidence has held up:

• Both have two membranes, as engulfment by a membrane-bound vesicle would produce.
• Both carry their own circular DNA, resembling a bacterial chromosome rather than the linear, histone-wrapped chromosomes of the nucleus.
• Both have their own 70S ribosomes, the prokaryotic size, not the 80S ribosomes of the surrounding cytosol.
• Both divide by splitting in a manner resembling binary fission, and cannot be built from scratch by the cell.
• Antibiotics that target bacterial ribosomes also affect mitochondrial ones, which is a real clinical side effect and a hard fact to explain any other way.

Endosymbiosis is also why plant cells have both organelles and animal cells only mitochondria: the photosynthetic partner was acquired later, in the lineage leading to plants and algae.

Plant against animal

Plant cells add a rigid cellulose cell wall, chloroplasts, a large central vacuole and plasmodesmata, the cytoplasmic channels that connect neighbouring plant cells through their walls. Animal cells have centrioles and lysosomes prominently, use a flexible extracellular matrix rather than a wall, and connect through tight junctions, desmosomes and gap junctions. The cell wall is why a plant cell placed in pure water swells and stops rather than bursting, which the next lesson makes quantitative.

Common misconceptions

• "Prokaryotes are just simpler, less evolved cells." They have had exactly as long to evolve and dominate the biosphere by number and by metabolic range. Simplicity here is a strategy.
• "The cell membrane and the cell wall are the same thing." The membrane is a selective phospholipid bilayer present in every cell. The wall is a rigid external structure present only in some, and it is not selective.
• "Ribosomes are organelles like the others." They are not membrane-bound. They are ribonucleoprotein machines, and prokaryotes have them too.
• "Mitochondria make energy." They convert it. Energy in the bonds of glucose is transferred to ATP, with a large fraction lost as heat.
• "Bigger cells are more advanced." Cell size is limited by geometry. An organism grows by making more cells, not larger ones.

Where this leaves us

Cell theory says all organisms are made of cells and every cell comes from a pre-existing one, which makes the cell an unbroken lineage as well as a unit of structure. Cells stay small because surface area rises as the square of length while volume rises as the cube, and the same principle reappears as microvilli, cristae and thylakoid stacks. Prokaryotes hold their DNA in a nucleoid and manage without internal membranes; eukaryotes partition the job among organelles, and following insulin from nucleus to rough ER to Golgi to secretory vesicle shows why: staged modification, quality control, addressing and dispatch all need separate rooms. Compartments also allow incompatible conditions, concentrate reactants, provide membrane surface for embedded machinery and contain the gradients that store energy. Mitochondria and chloroplasts stand apart because they used to be free-living bacteria, and their double membranes, circular DNA, 70S ribosomes and independent division are the reasons we believe it.

Sources

1. OpenStax. (2018). 4.3 Eukaryotic cells. In Biology 2e. openstax.org
2. OpenStax. (2018). 4.4 The endomembrane system and proteins. In Biology 2e. openstax.org
3. OpenStax. (2018). 4.2 Prokaryotic cells. In Biology 2e. openstax.org
4. Cooper, G. M. (2000). The Golgi apparatus. In The Cell: A Molecular Approach, 2nd ed. NCBI Bookshelf. ncbi.nlm.nih.gov
5. Hooke, R. (1665). Micrographia. London: Royal Society.

Why the drip bag says 0.9 percent

Intravenous saline is 0.9 percent sodium chloride by mass, and the figure is not arbitrary. It is the concentration at which the solution has the same water-attracting power as the inside of a red blood cell, so no net water movement occurs.

Infuse pure water instead and the cells swell. Water floods in down its concentration gradient, the membrane stretches, and red cells, which have no wall, burst. Infuse a much more concentrated salt solution and the reverse happens: water leaves and the cells shrivel into spiky husks.

Everything about that follows from one structure, a phospholipid bilayer, and one physical tendency, the movement of water toward regions where it is more tightly held. This lesson does both, and does the arithmetic.

The fluid mosaic model

S. J. Singer and Garth Nicolson proposed the current picture in 1972. A membrane is a phospholipid bilayer, tails inward and heads out, in which proteins are embedded like objects floating in a two-dimensional fluid. Both words matter. Mosaic: the proteins are numerous and varied. Fluid: lipids and many proteins drift laterally, so the membrane is a liquid film rather than a wall.

Its components:

• Phospholipids, forming the bilayer and providing the hydrophobic barrier.
• Integral proteins, spanning the membrane, working as channels, carriers, receptors and enzymes.
• Peripheral proteins, attached to one face, often linking to the cytoskeleton.
• Cholesterol, wedged between the tails.
• Glycoproteins and glycolipids, carrying sugar chains on the outer face, used for recognition. The ABO blood groups are differences in these sugars.

Fluidity depends on temperature and on the tails: unsaturated fatty acids, with their kinks, prevent tight packing and keep membranes fluid in the cold. Cholesterol works in both directions, which is the interesting part. At high temperature it restrains phospholipid movement and reduces fluidity; at low temperature it wedges between tails and stops them packing into a solid. It is a buffer, not a stiffener.

What gets through, and what does not

The hydrophobic core is the filter.

Molecule type	Crosses the bilayer directly?	Examples
Small nonpolar	readily	O2, CO2, steroid hormones
Small uncharged polar	slowly	water, urea
Large polar	no	glucose, amino acids
Ions	no, regardless of size	Na+, K+, Cl-, H+

Note the surprise in row four: a sodium ion is tiny and still cannot cross, because it drags a shell of water molecules and the hydrophobic core will not accept the charge. Everything on the lower two rows needs a protein.

Key idea: the bilayer stops charged and large polar substances, so almost everything a cell must import or export needs a dedicated protein, and the cell decides what it makes.

Passive transport: downhill, no ATP

• Simple diffusion. Molecules move from higher to lower concentration by random motion alone. Oxygen entering a cell and carbon dioxide leaving both work this way, and both gradients are maintained because respiration constantly consumes one and produces the other.
• Facilitated diffusion. Still downhill and still free, but through a protein. Channels form a hydrophilic pore; many are gated by voltage or by a bound signal. Carriers bind the solute and change shape to release it on the other side. Aquaporins, discovered by Peter Agre and colleagues, are water channels that speed osmosis enormously where it is needed, in kidney tubules and root cells.
• Osmosis is the diffusion of water across a selectively permeable membrane, from where water is more concentrated to where it is less.

Tonicity describes the solution, not the cell

Solution	Animal cell	Plant cell
Hypotonic (fewer solutes outside)	swells, may lyse	becomes turgid; the wall stops it bursting
Isotonic	no net change	flaccid; plants wilt in isotonic surroundings
Hypertonic (more solutes outside)	shrivels	plasmolyses, the membrane pulling away from the wall

Two things worth noticing. The healthy state is opposite in the two cases: an animal cell wants isotonic surroundings, a plant wants hypotonic ones. And the terms are comparative, so "hypertonic" always needs a "compared to what".

Water potential, and how to compute it

Concentration alone is not enough once physical pressure is involved, as it is inside a walled plant cell. Water potential, written with the Greek letter psi, combines both. Water moves from higher water potential to lower water potential, always.

Ψ = Ψs + Ψp

where Ψs is the solute potential and Ψp the pressure potential. Pure water in an open container has both terms zero, so its water potential is zero, and that is the maximum: adding solute always makes the value negative.

Ψs = -iCRT

• i is the ionisation constant: 1 for sucrose, which does not dissociate, and 2 for NaCl, which splits into two ions.
• C is molar concentration.
• R is the pressure constant, 0.0831 litre bars per mole per kelvin.
• T is temperature in kelvin, which is degrees Celsius plus 273.

Worked example one. Find the solute potential of 1.0 M sucrose at 25 degrees Celsius.

1. i = 1, C = 1.0, R = 0.0831, T = 298.
2. Ψs = -(1)(1.0)(0.0831)(298) = -24.8 bars.
3. In an open beaker there is no added pressure, so Ψp = 0 and Ψ = -24.8 bars.

Worked example two. Find the water potential of 0.15 M NaCl at 22 degrees Celsius.

1. NaCl dissociates into two ions, so i = 2. T = 295.
2. Ψs = -(2)(0.15)(0.0831)(295) = -7.35 bars.

Notice that a 0.15 M salt solution is more concentrated in particles than a 0.15 M sugar solution, and its water potential is correspondingly twice as negative. Particles are what count, not molecules.

Predicting the direction, three scenarios

Take a flaccid plant cell whose solute potential is -2.0 bars and whose pressure potential is 0, so its water potential is -2.0 bars.

Placed in	Solution water potential	Water moves	Result
Pure water	0	into the cell, from higher to lower psi	turgid; entry stops when pressure potential reaches +2.0 and the cell's psi equals 0
0.1 M sucrose at 22 C, psi = −2.45	−2.45	out of the cell	slight shrinkage
0.15 M NaCl at 22 C, psi = −7.35	−7.35	strongly out of the cell	plasmolysis

The first row is the one worth dwelling on, because it shows what pressure potential is for. Water enters, the cell presses against its wall, the wall pushes back, and that back pressure raises the cell's water potential toward zero. Equilibrium arrives not when the concentrations match but when the potentials do. A plant with no turgor pressure is a wilted plant, which is why watering works.

Active transport: uphill, and it costs

Primary active transport uses ATP directly. The sodium-potassium pump is the standard example and one of the busiest proteins in your body: for each ATP hydrolysed, it moves three sodium ions out of the cell and two potassium ions in, both against their gradients. A large share of a resting animal's energy budget goes into running it.

The gradient it builds is then a battery. Secondary active transport spends that stored gradient instead of ATP: in the small intestine, a cotransporter carries glucose into a cell against the glucose gradient by letting sodium flow in down its own steep gradient at the same time. No ATP is used at that protein, but the gradient it consumes was made with ATP elsewhere. Oral rehydration therapy works because of exactly this coupling, which is why the solution contains both salt and sugar.

Plants and fungi use proton pumps in the same way, exporting hydrogen ions to power the uptake of nutrients.

Bulk transport, for anything too large

• Phagocytosis: the membrane engulfs a large particle, such as a macrophage swallowing a bacterium, and the vesicle fuses with a lysosome.
• Pinocytosis: non-specific sampling of extracellular fluid in small vesicles.
• Receptor-mediated endocytosis: specific molecules bind receptors that cluster in a coated pit before it pinches inward. Cholesterol is imported this way as LDL particles, and inherited defects in the LDL receptor cause familial hypercholesterolaemia, in which cholesterol accumulates in the blood because cells cannot take it up.
• Exocytosis: a vesicle fuses with the plasma membrane and releases its contents, as in the insulin secretion traced in the previous lesson.

Common misconceptions

• "Facilitated diffusion requires energy because it uses a protein." It does not. The protein provides a route; the gradient provides the direction and the drive.
• "Osmosis is water moving toward high concentration." Toward the high concentration of solute, which is the low concentration of water. Saying it in terms of water potential removes the ambiguity: water moves to the more negative potential.
• "A hypertonic cell." Tonicity compares two solutions. A cell can be in a hypertonic solution; the cell itself is not hypertonic in isolation.
• "Water potential is always negative." Pure water at atmospheric pressure is zero, and a turgid cell's pressure potential is positive, so the sum can approach zero from below and inside a xylem vessel under tension it becomes strongly negative.
• "Small ions cross membranes easily because they are small." Charge, not size, is the obstacle. Ions need channels.

Summing up

A membrane is a fluid phospholipid bilayer with proteins drifting in it, and cholesterol buffers its fluidity in both directions. The hydrophobic core admits small nonpolar molecules easily, small uncharged polar ones slowly, and charged or large polar ones not at all, so most traffic requires a protein and the cell controls the traffic by controlling the proteins. Passive routes, simple diffusion, facilitated diffusion through channels or carriers, and osmosis, all run down gradients at no cost. Tonicity describes the surroundings, and the healthy state differs between walled and unwalled cells. Water potential makes the prediction quantitative: Ψ = Ψs + Ψp with Ψs = -iCRT, so 1.0 M sucrose at 25 degrees comes to −24.8 bars while 0.15 M NaCl at 22 degrees comes to −7.35 bars, and water always moves toward the more negative value until the potentials match. Active transport spends ATP to build gradients such as the three-sodium-for-two-potassium exchange, secondary transport spends those gradients instead, and anything too large for a protein moves by endocytosis or exocytosis.

Sources

1. OpenStax. (2018). 5.1 Components and structure. In Biology 2e. openstax.org
2. OpenStax. (2018). 5.2 Passive transport. In Biology 2e. openstax.org
3. OpenStax. (2018). 5.3 Active transport. In Biology 2e. openstax.org
4. OpenStax. (2018). 30.5 Transport of water and solutes in plants. In Biology 2e. openstax.org
5. Wikipedia. (n.d.). Water potential. en.wikipedia.org

The diet pill that cooked people

In 1933 Maurice Tainter at Stanford reported that a chemical called 2,4-dinitrophenol caused rapid weight loss. Within a few years it was being sold to hundreds of thousands of Americans. It worked. It also killed people, by raising body temperature until proteins denatured, and in 1938 the newly empowered Food and Drug Administration removed it from the market.

What DNP does is precise and instructive. It slips through the inner mitochondrial membrane carrying hydrogen ions, so a gradient the cell has carefully built up simply leaks away. Fuel is still burned, oxygen is still consumed, and no ATP is made. The energy comes out as heat instead.

That tells you something about how respiration works before you learn any of it. The energy from glucose is not handed directly to ATP. It is used to build a gradient of hydrogen ions across a membrane, and the gradient is what makes the ATP. Puncture the membrane's ability to hold that gradient and the whole system runs hot and produces nothing.

ATP, the cell's small change

Adenosine triphosphate is adenine plus ribose plus three phosphate groups in a row. Those three phosphates all carry negative charge and are held close together, which is unstable in the way a compressed spring is unstable. Break the bond to the terminal phosphate and about 30.5 kilojoules per mole are released under standard conditions.

ATP is not a store; it is a currency. Cells hold very little of it and recycle it furiously, phosphorylating ADP back to ATP as fast as it is spent. An adult human turns over something close to their own body mass in ATP over a day, while carrying only a few grams at any moment.

Its use is energy coupling: an unfavourable reaction is run alongside ATP hydrolysis, usually by transferring the phosphate group to a substrate or an enzyme, so that the pair together is favourable. This is the answer to the question left open in Module 2, of how a cell drives reactions that are energetically uphill.

Bottom line: respiration is the business of regenerating ATP, and everything below is the machinery for doing so efficiently.

Redox, in the only sense you need

Oxidation is loss of electrons and reduction is gain of electrons. In organic chemistry electrons usually travel with hydrogen atoms, so a molecule losing hydrogens is being oxidised.

Glucose is a highly reduced fuel: its carbons carry many hydrogens, and therefore many high-energy electrons. Respiration strips those electrons away in controlled steps and passes them, eventually, to oxygen. Burning glucose in a flame does the same chemistry in one uncontrolled step and gives you only heat.

The electrons are carried by two shuttles. NAD⁺ accepts two electrons and one proton to become NADH; FAD accepts two of each to become FADH₂. Every time you see NADH being made, electrons are being moved out of fuel and into a carrier.

The overall reaction is

C6H12O6 + 6 O2 → 6 CO2 + 6 H2O

releasing about 2,870 kilojoules per mole of glucose. The cell captures roughly a third of that in ATP and releases the rest as heat, which is a better efficiency than a petrol engine achieves.

Stage one: glycolysis, in the cytosol

Ten enzyme-catalysed steps split a six-carbon glucose into two three-carbon pyruvate molecules. It happens in the cytosol, needs no oxygen and no organelle, and occurs in essentially every living thing, which is why it is thought to be the oldest part of the pathway.

• Investment phase. Two ATP are spent to phosphorylate the sugar. This looks perverse and is not: adding charged phosphates traps the sugar inside the cell and destabilises it for splitting.
• Payoff phase. The six-carbon molecule is cleaved into two three-carbon pieces, each of which is oxidised and phosphorylated. Four ATP are produced by substrate-level phosphorylation, where a phosphate is transferred directly from a substrate to ADP, and two NAD⁺ are reduced to NADH.

Net per glucose: 2 ATP, 2 NADH, 2 pyruvate. Four made minus two spent.

Stage two: pyruvate oxidation

If oxygen is available, pyruvate is transported into the mitochondrial matrix. There a large enzyme complex removes one carbon as carbon dioxide, oxidises the remaining two-carbon fragment, reducing NAD⁺ to NADH, and attaches the fragment to coenzyme A to give acetyl-CoA.

This happens twice per glucose, since glycolysis produced two pyruvates. Per glucose: 2 CO₂, 2 NADH, 2 acetyl-CoA. That is the first carbon dioxide you exhale, and it is worth noticing that the carbon in your breath came from the food you ate, not from the air.

Stage three: the citric acid cycle

Acetyl-CoA hands its two-carbon acetyl group to a four-carbon molecule, oxaloacetate, forming six-carbon citrate. The cycle then oxidises the molecule step by step, releasing two carbons as carbon dioxide and regenerating oxaloacetate to accept the next acetyl group.

Per turn: 3 NADH, 1 FADH₂, 1 ATP by substrate-level phosphorylation, 2 CO₂. Two turns per glucose, so 6 NADH, 2 FADH₂, 2 ATP and 4 CO₂.

Account for the carbon now. Glucose had six. Two left in pyruvate oxidation and four in the cycle, so by this point every carbon atom from the glucose has departed as carbon dioxide. Notice what has not happened: barely any ATP has been made. Four so far, out of thirty-odd. Almost the entire yield is sitting in the reduced carriers, and cashing them in is the next stage.

Stage four: the electron transport chain and chemiosmosis

NADH and FADH₂ deliver their electrons to a series of protein complexes embedded in the inner mitochondrial membrane. The electrons pass from complex to complex, each with a greater affinity for electrons than the last, releasing energy at each handover. Three of the complexes use that energy to pump hydrogen ions from the matrix into the intermembrane space.

At the end of the chain sits oxygen, which accepts the spent electrons along with hydrogen ions to form water. This is the only place oxygen is used in the whole of respiration, and it explains why an oxygen supply is required: without a final acceptor, the chain backs up, no more electrons can enter, the carriers stay reduced, and the citric acid cycle stops for want of NAD⁺. Cyanide kills so quickly because it blocks the last complex, with exactly that effect.

The pumping builds a hydrogen ion gradient, higher concentration outside the inner membrane than in, which is both a concentration gradient and a voltage. Peter Mitchell proposed in 1961 that this gradient, and not any direct chemical intermediate, is how the energy is stored, an idea so unpopular at the time that he spent years defending it and eventually received a Nobel Prize for it in 1978.

The hydrogen ions flow back into the matrix through ATP synthase, a rotary molecular motor. The flow spins part of the enzyme, and the rotation forces conformational changes that assemble ATP from ADP and phosphate. It is a turbine. This process is called oxidative phosphorylation, and the coupling of a gradient to ATP synthesis is chemiosmosis.

Now DNP makes sense. It ferries hydrogen ions back across the membrane without passing through the turbine, so the gradient collapses, no ATP is made, and the energy appears as heat. Brown adipose tissue in mammals does the same thing deliberately, with a protein called thermogenin, to generate heat in newborns and hibernators.

The tally, and why the number is not exact

Stage	Direct ATP	NADH	FADH2
Glycolysis	2	2	0
Pyruvate oxidation	0	2	0
Citric acid cycle	2	6	2
Total	4	10	2

Each NADH delivering electrons at the first complex yields about 2.5 ATP, and each FADH₂, which enters further along and so drives less pumping, yields about 1.5.

• 10 × 2.5 = 25
• 2 × 1.5 = 3
• Plus the 4 direct ATP
• Total about 32 ATP per glucose, and about 30 in tissues where the shuttle that moves cytosolic NADH into the mitochondrion costs something.

Older textbooks give 36 or 38, using whole-number ratios of 3 and 2. The modern figures are not whole numbers because ATP synthase requires a non-integer number of protons per ATP, and because the gradient also powers transport work. Any exact figure quoted without a range is overstating the case; the honest statement is roughly 30 to 32.

Without oxygen

Glycolysis needs a supply of NAD⁺ to keep going, and normally the electron transport chain regenerates it. With no oxygen, that route closes, and unless NAD⁺ is regenerated some other way, glycolysis halts within seconds.

Fermentation is that other way. It does not make ATP itself. Its entire purpose is to oxidise NADH back to NAD⁺ so glycolysis can continue producing its 2 ATP per glucose.

• Lactic acid fermentation. Pyruvate accepts the electrons and becomes lactate. This runs in vertebrate muscle during intense exercise and in the bacteria that make yoghurt and sauerkraut.
• Alcoholic fermentation. Pyruvate loses a carbon as carbon dioxide and the resulting acetaldehyde accepts the electrons, becoming ethanol. Yeast does this, which is why bread rises and beer exists.

One common belief needs correcting here. Lactate is not the cause of the muscle soreness that arrives a day or two after hard exercise; blood lactate returns to baseline within about an hour. Delayed soreness comes from microscopic damage to muscle fibres and the inflammation that follows.

Other fuels, and the throttle

Glucose is not the only input. Fats are broken into fatty acids, which are chopped two carbons at a time into acetyl-CoA and fed into the citric acid cycle, which is why fats yield so much more energy per gram. Proteins are deaminated and their carbon skeletons enter at whichever point matches their structure. Respiration is a hub, not a single line.

Its rate is controlled by feedback inhibition, the mechanism from Module 2. Phosphofructokinase, an early enzyme of glycolysis, is inhibited allosterically by ATP and by citrate and stimulated by AMP. When the cell has plenty of ATP, the pathway throttles back; when ATP has been spent and AMP accumulates, it opens up. The cell regulates its power output at the intake.

Common misconceptions

• "Mitochondria make energy." They transfer it, from the bonds of fuel molecules to ATP, losing about two thirds as heat. Energy is never created.
• "Oxygen is used throughout respiration." It is used at exactly one point, as the final electron acceptor. Everything upstream stops without it only because the chain backs up.
• "Fermentation is a way of making ATP without oxygen." Glycolysis makes the ATP. Fermentation exists to regenerate NAD⁺ so glycolysis can keep running.
• "Lactic acid causes next-day muscle soreness." Lactate clears within about an hour. Delayed soreness is tissue damage and inflammation.
• "Respiration produces exactly 38 ATP." Roughly 30 to 32 is the current estimate, and the value genuinely varies with tissue and conditions.

Putting it together

ATP is the cell's spendable currency, recycled constantly rather than stored, and respiration exists to regenerate it. Glucose is a reduced fuel, and the pathway strips its electrons onto NAD⁺ and FAD in stages: glycolysis in the cytosol yields 2 ATP, 2 NADH and 2 pyruvate; pyruvate oxidation in the matrix yields 2 NADH, 2 acetyl-CoA and the first two carbon dioxides; the citric acid cycle turns twice for 6 NADH, 2 FADH₂, 2 ATP and the remaining four carbon dioxides. At that point every carbon from the glucose has gone and only 4 ATP exist, because the value is in the carriers. Oxidative phosphorylation cashes them in: electrons descend the chain, protons are pumped across the inner membrane, oxygen accepts the spent electrons to make water, and the returning protons drive ATP synthase as a turbine, for a total near 30 to 32 ATP. Chemiosmosis is why an uncoupler such as DNP burns fuel and produces only heat. Without oxygen, fermentation regenerates NAD⁺ so that glycolysis alone can supply 2 ATP per glucose.

Sources

1. OpenStax. (2018). 7.2 Glycolysis. In Biology 2e. openstax.org
2. OpenStax. (2018). 7.3 Oxidation of pyruvate and the citric acid cycle. In Biology 2e. openstax.org
3. OpenStax. (2018). 7.4 Oxidative phosphorylation. In Biology 2e. openstax.org
4. OpenStax. (2018). 7.5 Metabolism without oxygen. In Biology 2e. openstax.org
5. OpenStax. (2018). 6.4 ATP: adenosine triphosphate. In Biology 2e. openstax.org

A willow tree, 200 pounds of soil, and five years

In the 1640s Jan Baptist van Helmont planted a willow sapling weighing about five pounds in a pot holding two hundred pounds of dried soil. He covered the pot, watered it with rainwater, and waited five years. The tree then weighed around 169 pounds. The soil, dried and weighed again, had lost about two ounces.

Van Helmont concluded that the extra 164 pounds of wood had come from water. He was mostly wrong, and his experiment was excellent. He had proved that plants do not eat soil, which was the prevailing belief, and he had shown that plant mass has to come from somewhere invisible. It took another century and a half of work by Priestley, Ingenhousz and Senebier to identify the missing ingredient: carbon dioxide from the air.

A tree is mostly air. That is the fact worth carrying: the carbon in a hundred-tonne redwood was extracted, molecule by molecule, from a gas present at roughly four hundredths of one percent of the atmosphere.

The equation, and one thing it hides

6 CO2 + 6 H2O + light energy → C6H12O6 + 6 O2

That is respiration's equation reversed, and it is tempting to say photosynthesis is respiration run backwards. It is not. The pathways are different, the enzymes are different, the compartments are different, and the intermediates are different. What is true is that the two processes together form a cycle at the level of the planet: photosynthesis reduces carbon and releases oxygen, respiration oxidises carbon and consumes oxygen.

The equation also hides where the oxygen comes from. Written this way it could plausibly come from the carbon dioxide. It does not. In 1941 Samuel Ruben and Martin Kamen supplied algae with water containing the heavy isotope oxygen-18 and found the heavy label in the oxygen gas released; supplying labelled carbon dioxide instead did not. The oxygen you are breathing came from split water.

What matters here: photosynthesis has two stages. The light reactions convert light energy into ATP and NADPH; the Calvin cycle spends those to build sugar from carbon dioxide. Only the first needs light directly.

Inside a chloroplast

A chloroplast has two outer membranes and, inside, a fluid called the stroma and a third membrane system of flattened discs called thylakoids, stacked into grana. The stacking is the surface area principle again: the light reactions happen in thylakoid membranes, so more membrane means more capacity, and the thylakoid interior, the lumen, is the compartment that holds the proton gradient.

The division of labour is clean. Light reactions in the thylakoid membrane; Calvin cycle in the stroma.

Pigments, and why leaves are green

A pigment absorbs some wavelengths and reflects others, and the colour you see is what it rejected. Chlorophyll a, the pigment that actually performs the chemistry, absorbs strongly in the blue region near 430 nanometres and in the red near 660, and poorly in the green in between, which is why plants look green. Chlorophyll b and the carotenoids are accessory pigments absorbing at somewhat different wavelengths, broadening the range of usable light and passing the energy on to chlorophyll a. Carotenoids also absorb excess energy that would otherwise damage the chloroplast, and they are what you see in autumn leaves once chlorophyll breaks down.

Theodor Engelmann demonstrated the connection in 1882 with an elegant experiment. He laid a filament of green alga under a spectrum of light produced by a prism, and added bacteria that swim toward oxygen. The bacteria gathered along the parts of the filament lit by blue and red light, and thinned out where the light was green. The action spectrum of photosynthesis, measured by where oxygen was being produced, matched the absorption spectrum of chlorophyll.

The light-dependent reactions

1. Photosystem II absorbs light. Energy funnels to a special pair of chlorophyll molecules called P680, and an excited electron leaves.
2. Water is split to replace it. Photosystem II is the strongest biological oxidant known, and it strips electrons from water. Splitting two water molecules yields four electrons, four hydrogen ions released into the thylakoid lumen, and one molecule of O₂, which diffuses away as a by-product.
3. Electron transport. The excited electron passes along a chain of carriers in the thylakoid membrane, and the energy released is used to pump more hydrogen ions from the stroma into the lumen.
4. Photosystem I re-energises it. The electron, now depleted, arrives at photosystem I, where another photon boosts it again, from a pair called P700.
5. NADPH is made. The re-energised electron is passed to NADP⁺, which with a hydrogen ion becomes NADPH, the reducing power for the next stage.
6. ATP is made by chemiosmosis. Hydrogen ions accumulated in the lumen flow back into the stroma through ATP synthase, exactly as in a mitochondrion, driving ATP production. Here it is called photophosphorylation.

Notice the electron's journey: it starts in water, gets pushed uphill twice by light, and ends in NADPH. Water is the source and NADPH is the destination, with a proton gradient built along the way. Two photosystems are needed because one photon does not carry enough energy to lift an electron all the way from water to NADP⁺.

The Calvin cycle, with the arithmetic

In the stroma, ATP and NADPH are spent to fix carbon dioxide into sugar. Three phases repeat.

1. Carbon fixation. The enzyme rubisco attaches one CO₂ to a five-carbon sugar, ribulose bisphosphate or RuBP. The resulting six-carbon molecule immediately splits into two three-carbon molecules of 3-phosphoglycerate. Rubisco is thought to be the most abundant protein on Earth, which is a comment on how slow it is: plants compensate for a sluggish enzyme by making enormous quantities of it.
2. Reduction. Each 3-phosphoglycerate is phosphorylated with ATP and reduced with NADPH to give glyceraldehyde-3-phosphate, G3P.
3. Regeneration. Most of the G3P is rearranged, using more ATP, back into RuBP so the cycle can continue.

Now count. Take three turns of the cycle, fixing three carbon dioxides.

• Three CO₂ plus three RuBP give six molecules of 3-phosphoglycerate, which become six G3P.
• Five of those six are consumed to regenerate three RuBP. Only one G3P leaves the cycle.
• The cost of those three turns is 9 ATP and 6 NADPH.
• A glucose is six carbons, so it takes two G3P, meaning six turns of the cycle, 18 ATP and 12 NADPH.

The product of photosynthesis is really G3P rather than glucose. From G3P the plant builds glucose, sucrose for transport, starch for storage, cellulose for walls, and the carbon skeletons for amino acids and lipids.

Rubisco's expensive mistake

Rubisco cannot reliably tell carbon dioxide from oxygen. When it grabs O₂ instead, the result is photorespiration: no sugar is made, previously fixed carbon is released, and ATP is consumed cleaning up. On a hot dry day the problem compounds, because the plant closes its stomata to conserve water, carbon dioxide inside the leaf is depleted by the Calvin cycle while oxygen from the light reactions accumulates, and the ratio shifts further in oxygen's favour.

The flaw is probably an inheritance. Rubisco evolved when the atmosphere held far more carbon dioxide and very little oxygen, and discriminating between them was not a problem worth solving.

Two solutions plants actually evolved

	C3 (most plants)	C4	CAM
Where CO2 is first fixed	directly by rubisco in mesophyll cells	by PEP carboxylase in mesophyll cells, into a four-carbon acid	by PEP carboxylase at night, stored as malate in vacuoles
Separation strategy	none	spatial: the four-carbon acid is shuttled to bundle sheath cells where rubisco works in a high-CO2 pocket	temporal: stomata open at night, and the Calvin cycle runs by day behind closed stomata
Best suited to	cool, moist, moderate light	hot, bright, with adequate water	arid environments
Examples	wheat, rice, soybean, most trees	maize, sugarcane, sorghum, crabgrass	cacti, pineapple, agave, many succulents

PEP carboxylase, the enzyme both alternatives use for the first capture, does not bind oxygen at all, which is precisely why it solves the problem. Both strategies cost extra ATP, so they only pay in the conditions that make photorespiration expensive. That is why a lawn of cool-season grass browns in August while the crabgrass in it thrives.

What limits the rate

• Light intensity. The rate rises with light and then plateaus when the photosystems are saturated and some other factor takes over.
• Carbon dioxide concentration. Frequently the limiting factor in bright light, which is why commercial greenhouses sometimes enrich the air.
• Temperature. The Calvin cycle is enzymatic, so its rate rises with temperature to an optimum and then falls as proteins denature and photorespiration increases.

Where a graph of rate against one factor levels off, that factor is no longer the limiting one, and identifying which factor is limiting is the standard question this topic asks.

Common misconceptions

• "Plants take in carbon dioxide and give out oxygen; they do not respire." Plants respire constantly, day and night, in their own mitochondria. In daylight photosynthesis simply outpaces it.
• "The oxygen released comes from carbon dioxide." It comes from water, as isotope labelling showed in 1941.
• "The light-independent reactions happen at night." They happen whenever ATP and NADPH are available, which in most plants is during the day. They are called light-independent because they do not use photons directly, not because they prefer darkness.
• "Plants get their mass from the soil." Van Helmont disproved that in the 1640s. Soil supplies water and mineral ions; the bulk of the dry mass is carbon from the air.
• "Green light is useless to plants." It is absorbed poorly, not not at all, and it penetrates deeper into a leaf and into a canopy than red or blue, which is why it is far from wasted.

Looking back

Plant mass comes from air, a conclusion van Helmont's willow set up and later work completed. Photosynthesis runs in two stages inside the chloroplast: light reactions in the thylakoid membranes and the Calvin cycle in the stroma. Chlorophyll a absorbs blue and red and reflects green, with accessory pigments broadening the range, which is why the action spectrum matches the absorption spectrum in Engelmann's 1882 experiment. In the light reactions, photosystem II splits water to replace lost electrons, releasing the oxygen you breathe; electrons travel a chain that pumps protons into the thylakoid lumen; photosystem I re-energises them onto NADP⁺; and returning protons drive ATP synthase. The Calvin cycle then spends 9 ATP and 6 NADPH per three turns to release a single G3P, meaning 18 ATP and 12 NADPH per glucose. Rubisco's inability to exclude oxygen causes photorespiration, worst when heat closes stomata, and C4 and CAM plants avoid it by separating the first capture from rubisco in space or in time.

Sources

1. OpenStax. (2018). 8.1 Overview of photosynthesis. In Biology 2e. openstax.org
2. OpenStax. (2018). 8.2 The light-dependent reactions of photosynthesis. In Biology 2e. openstax.org
3. OpenStax. (2018). 8.3 Using light energy to make organic molecules. In Biology 2e. openstax.org
4. OpenStax. (2018). 6.1 Energy and metabolism. In Biology 2e. openstax.org

A hormone that never gets in

In the late 1950s Earl Sutherland and Theodore Rall were trying to work out how adrenaline makes a liver cell dump glucose into the blood. They broke the cells open, expecting to find the hormone bound to the enzyme that breaks down glycogen. It was not there. What they found instead was a small, heat-stable molecule that appeared when the hormone was applied to the membrane fraction, and that could trigger glycogen breakdown all by itself in a cell-free extract. They had discovered cyclic AMP, and Sutherland received the 1971 Nobel Prize in Physiology or Medicine for the work.

The finding was strange at the time and is the foundation of the whole field now. Adrenaline does not enter the liver cell. It docks on the outside, and something else carries the message inward. Every hormone you will meet in this course works on one of two plans: it either crosses the membrane itself, or it hands the message to a relay at the door.

The point: a signal has three parts, and they are worth naming before anything else. Reception, a ligand binding a receptor. Transduction, the chain of molecular events that carries and converts the message. Response, the thing the cell actually does.

How far the message has to travel

Cells signal over distances that differ by six orders of magnitude, and the mode is chosen by the distance.

Mode	Range	Example
Direct contact	touching cells	gap junctions in heart muscle; plasmodesmata between plant cells
Paracrine	neighbouring cells	growth factors released into the extracellular fluid during wound healing
Synaptic	across a gap of about 20 nanometres	acetylcholine released by a motor neuron onto a muscle fibre
Endocrine	the whole body	insulin from the pancreas reaching every tissue through the blood
Autocrine	the cell itself	an immune cell secreting a signal that binds its own receptors

Notice that the same molecule can appear in more than one row. Adrenaline is a hormone in the blood and a neurotransmitter in the nervous system. What makes a signal specific is not the molecule but the receptor, and only cells carrying the matching receptor can respond at all. Every one of your cells is bathed in insulin after a meal. A liver cell takes up glucose in response; a mature red blood cell, which lacks the machinery, does not.

Two kinds of receptor, and one useful rule

The rule is chemical. A signal molecule that is small and lipid-soluble can slip through the phospholipid bilayer; a large or charged one cannot.

Intracellular receptors handle the first group. Steroid hormones, which are built on a cholesterol skeleton, pass straight through the membrane. Cortisol, testosterone and oestrogen bind receptor proteins in the cytoplasm or nucleus, and the hormone-receptor complex then binds DNA and acts as a transcription factor, switching particular genes on. The response is slow, taking minutes to hours, because it requires new protein to be made, and it lasts a long time. Thyroid hormone and vitamin D work the same way. Nitric oxide, a gas, also diffuses straight in, and relaxes the smooth muscle of blood vessel walls.

Cell-surface receptors handle everything else, and come in three families worth knowing.

• G protein-coupled receptors. A membrane protein that threads through the bilayer seven times. Ligand binding changes its shape, it activates an attached G protein by causing it to swap GDP for GTP, and the activated G protein moves along the membrane to switch on an enzyme such as adenylyl cyclase. Alfred Gilman and Martin Rodbell shared the 1994 Nobel Prize for identifying these proteins. Roughly a third of all prescription drugs act at receptors of this family.
• Receptor tyrosine kinases. Ligand binding brings two receptor monomers together; the pair then phosphorylate each other on tyrosine residues, and the resulting cluster of phosphates is a docking site where many different relay proteins attach at once. One ligand can therefore trigger several pathways simultaneously. The insulin receptor is one of these.
• Ligand-gated ion channels. The simplest design: the receptor is the channel. Acetylcholine binding at a neuromuscular junction opens a pore, sodium ions flood in, and the muscle fibre depolarises in under a millisecond. No second messenger is involved, which is why this is the design used where speed matters.

The cascade, and why it multiplies

Follow adrenaline into a liver cell and count the steps.

1. Adrenaline binds a beta-adrenergic receptor on the outside of the membrane.
2. The receptor activates a G protein, which activates adenylyl cyclase on the inner face of the membrane.
3. Adenylyl cyclase converts ATP into cyclic AMP, the second messenger Sutherland found.
4. Cyclic AMP activates protein kinase A.
5. Protein kinase A phosphorylates phosphorylase kinase, which phosphorylates glycogen phosphorylase.
6. Glycogen phosphorylase chops glucose units off stored glycogen, and glucose leaves the cell for the blood.

A kinase adds a phosphate group to a target protein, usually taking it from ATP; the added phosphate carries negative charge, which changes the target's shape and therefore its activity. This is the on-switch used everywhere in the cell, and about two percent of human genes code for kinases.

Now see why the chain has so many links. Each activated molecule at one level activates many molecules at the next, because it is a catalyst and is not consumed. One occupied receptor turns on several G proteins. Each adenylyl cyclase makes many cyclic AMP molecules. Each protein kinase A phosphorylates many substrates. The result is amplification: a few molecules of hormone in the blood produce a response involving millions of molecules of glucose. That is why endocrine signals can work at concentrations too low to measure by ordinary chemistry.

Calcium is the other great second messenger. Cytosolic calcium is kept extremely low, roughly ten thousand times lower than the concentration outside the cell or inside the endoplasmic reticulum. Opening a calcium channel therefore produces a huge proportional spike with no energy spent at the moment of signalling, and that spike is what triggers muscle contraction and neurotransmitter release.

Turning it off is not an afterthought

A switch that cannot be released is not a switch. Every element of a pathway has a matching off mechanism: the ligand dissociates and is degraded, the G protein hydrolyses its own GTP back to GDP and shuts itself down, phosphodiesterase destroys cyclic AMP, and protein phosphatases strip the phosphates that kinases added. Because the off machinery runs continuously, the cell's response tracks the current concentration of signal rather than the total it has ever seen.

What happens when termination fails is the clearest evidence that it matters. Cholera is a failure of exactly one step. Vibrio cholerae secretes a toxin that chemically modifies the G protein in intestinal cells so that it can no longer hydrolyse its GTP. The G protein is stuck on, adenylyl cyclase runs without pause, cyclic AMP stays high, chloride ions are pumped into the gut lumen without stopping, and water follows them osmotically. An untreated patient can lose litres of fluid in hours. The bacterium never enters a single cell; it simply jams one molecular switch in the on position.

Why this matters: caffeine acts on the same system from the other end. It inhibits phosphodiesterase, the enzyme that clears cyclic AMP, so signals that would have faded persist. Many drugs act not on the signal but on the off switch.

What the cell does with the message

Responses fall into three broad kinds. A pathway can activate an existing enzyme in the cytoplasm, which is fast and reversible; that is what adrenaline does to glycogen phosphorylase. It can change gene expression by activating a transcription factor, which is slower and longer-lasting. Or it can trigger a change in the cytoskeleton, moving the cell or changing its shape.

One response deserves its own name. Apoptosis is programmed cell death: a signal, either an external death signal or an internal report of irreparable DNA damage, activates enzymes that dismantle the cell from the inside, package the fragments in membrane, and present them for neighbouring cells to eat. Nothing leaks, and no inflammation follows, which is what distinguishes apoptosis from the messy death of an injured cell. Apoptosis carves the webbing from between the fingers of a developing hand, prunes surplus neurons, and removes cells whose DNA damage cannot be repaired. Cancer cells routinely acquire mutations that disable it.

Bacteria signal too. A free-living Vibrio fischeri cell makes no light, because light from one cell would be wasted. Each cell secretes a small molecule continuously, and when the population is dense enough the concentration of that molecule crosses a threshold and switches on the luminescence genes in every cell at once. This is quorum sensing: a way for single cells to measure how many of them there are, and it also governs when pathogenic bacteria switch on virulence genes and build biofilms.

Common misconceptions

• "Hormones enter the cell and do the work themselves." Most do not. Water-soluble signals bind outside and never cross. Only lipid-soluble signals such as steroids get in.
• "A second messenger is a second hormone." It is made inside the cell, by the cell, from its own materials. Cyclic AMP is manufactured from ATP by adenylyl cyclase.
• "Every cell in the body responds to a hormone in the blood." Only cells with the matching receptor respond at all, and different cell types with the same receptor can respond in different ways because they contain different downstream machinery.
• "A long pathway is wasteful; a direct route would be better." Length buys amplification, points at which the signal can be integrated with other information, and points at which it can be regulated. A one-step pathway would have none of that.
• "Apoptosis is the same as cell damage." Apoptosis is orderly and triggered on purpose. Necrosis is uncontrolled, spills the cell contents, and provokes inflammation.

What to remember

Sutherland found cyclic AMP because adrenaline was not where he expected it to be, and the shape of the whole field followed from that. Signalling has three phases: reception, transduction, response. Lipid-soluble signals such as steroid hormones cross the membrane and bind intracellular receptors that act on DNA, giving slow, durable responses. Everything else binds a surface receptor, of which G protein-coupled receptors, receptor tyrosine kinases and ligand-gated ion channels are the three families to know. Inside, second messengers such as cyclic AMP and calcium spread the message, and kinase cascades amplify it, because each catalytic step activates many molecules at the next. Off switches, from GTP hydrolysis to phosphodiesterase to phosphatases, run continuously, and cholera shows what a jammed one costs. Specificity comes from which receptors and which relay proteins a cell happens to contain, which is why the same hormone can mean two different things in two different tissues.

Sources

1. OpenStax. (2018). 9.1 Signaling molecules and cellular receptors. In Biology 2e. openstax.org
2. OpenStax. (2018). 9.2 Propagation of the signal. In Biology 2e. openstax.org
3. OpenStax. (2018). 9.4 Signaling in single-celled organisms. In Biology 2e. openstax.org
4. The Nobel Foundation. (1994). The Nobel Prize in Physiology or Medicine 1994: Alfred G. Gilman and Martin Rodbell. nobelprize.org
5. The Nobel Foundation. (1971). The Nobel Prize in Physiology or Medicine 1971: Earl W. Sutherland Jr. nobelprize.org

The cells that would not stop

On 8 February 1951 a surgeon at Johns Hopkins took a small sample from a cervical tumour in a thirty-one-year-old woman named Henrietta Lacks. She was not asked, and neither she nor her family was told. She died that October. The sample went to George Gey's laboratory, where every previous attempt to keep human cells alive in glass had failed within days. These cells did not fail. They doubled roughly every twenty-four hours, and they have been dividing ever since, in laboratories on every continent, under the name HeLa.

Two facts sit side by side here. The first is an ethical one about consent, which the Lacks family fought over for decades and which changed how tissue samples are handled. The second is biological, and it is the subject of this lesson: an ordinary cell divides a limited number of times and then stops, and a cancer cell has lost the machinery that makes it stop. HeLa carries DNA from human papillomavirus 18 integrated into its genome, and the viral proteins disable two of the cell's most important brakes.

To see what was lost, you have to see what a normal cell does.

The cycle, and the arithmetic inside it

A dividing cell spends most of its life not dividing. For a typical human cell in culture with a cycle of about twenty-four hours, the actual division takes under an hour. The rest is interphase, which has three parts.

Phase	What happens	Chromosomes	Chromatids
G1	growth, protein synthesis, organelles duplicated	46	46
S	DNA replication	46	46 becoming 92
G2	more growth, checking the copy, building spindle components	46	92
M	mitosis and cytokinesis	46 in the parent, 46 in each daughter	92 separating to 46 each

Key idea: S phase does not double the chromosome number. It doubles the DNA. A chromosome that has been replicated still counts as one chromosome, because its two identical copies remain joined at the centromere as sister chromatids. The count of chromosomes changes only at anaphase, when that join is cut and each chromatid becomes a chromosome in its own right. If you can hold that distinction you have removed the single largest source of confusion in this whole topic.

Cells that are not on their way to divide sit in G₀, a quiescent state. Your liver cells wait there and can be recalled if the liver is damaged. Your neurons and mature muscle fibres enter G₀ and, in general, never come out. That is why a severed spinal cord does not heal the way a cut in the skin does.

Mitosis, one stage at a time

1. Prophase. Chromatin condenses until each replicated chromosome is visible as two chromatids joined at a centromere. The nucleolus disappears. The two centrosomes move apart and the mitotic spindle begins to form between them.
2. Prometaphase. The nuclear envelope breaks down. Protein complexes called kinetochores assemble on each centromere, and spindle microtubules attach to them.
3. Metaphase. Chromosomes are pulled to the middle and line up on the metaphase plate, each with its two kinetochores attached to microtubules from opposite poles. That opposition is the whole point: it guarantees one chromatid will go each way.
4. Anaphase. An enzyme cuts the protein rings holding the sisters together, and they separate simultaneously. Kinetochore microtubules shorten and drag the chromatids poleward, while other microtubules lengthen and push the poles apart. This is the shortest stage, often only a few minutes.
5. Telophase. Chromosomes reach the poles and decondense, nuclear envelopes reassemble around each set, and nucleoli reappear. Two nuclei now exist in one cell.

Cytokinesis is a separate event that usually overlaps telophase, and animals and plants do it differently. An animal cell builds a contractile ring of actin and myosin just inside the membrane, which tightens like a drawstring and pinches the cell in two, producing the cleavage furrow you can see under a light microscope. A plant cell cannot pinch, because it has a rigid wall. Instead, vesicles from the Golgi line up at the middle and fuse into a cell plate that grows outward until it joins the existing wall.

Prokaryotes do none of this. A bacterium has a single circular chromosome attached to the membrane; it replicates from one origin, the two copies move to opposite ends, and a protein ring pinches the cell in two. No spindle, no nuclear envelope to dismantle, no chromatids to align. Binary fission is a genuinely different mechanism, not a simplified mitosis.

Timing the cycle from a slide

You cannot watch a root tip cell for a day, but you can count. The trick is that the proportion of cells caught in a given phase is proportional to the time that phase takes, provided the cells are dividing continuously and are not synchronised. Suppose you count 200 cells in an onion root tip squash and find:

Phase	Cells counted	Fraction	Time, in a 24 hour cycle
Interphase	150	0.75	18 hours
Prophase	24	0.12	2.9 hours
Metaphase	10	0.05	1.2 hours
Anaphase	6	0.03	0.7 hours
Telophase	10	0.05	1.2 hours

The arithmetic is one line: fraction times cycle length. Prophase is 24 divided by 200, which is 0.12, and 0.12 of 24 hours is 2.88 hours. The mitotic index, the proportion of cells in mitosis rather than interphase, is 50 out of 200, which is 25 percent. That number is used clinically: a tumour biopsy with a high mitotic index is dividing fast, which is information about how the disease will behave and how it may respond to drugs that attack dividing cells.

What controls it

For decades the cycle looked like a sequence that simply happened. Three lines of work showed it is actively driven and actively checked, and the 2001 Nobel Prize in Physiology or Medicine went to all three. Leland Hartwell, working with budding yeast, isolated mutants that arrested at particular points and named the idea of a checkpoint. Paul Nurse found the gene controlling the same transitions in fission yeast, and then found the human version. Tim Hunt, working on sea urchin eggs, noticed a protein whose concentration rose steadily and then collapsed at every division, and called it cyclin.

The mechanism they uncovered is a pairing. A cyclin-dependent kinase, or Cdk, is present at a constant level and is inactive on its own. Cyclin concentration rises and falls through the cycle, and when enough cyclin accumulates it binds the Cdk and switches it on. The active complex phosphorylates the proteins that carry out the next phase. Then the cyclin is destroyed, the kinase falls silent, and the cycle moves on. Rising and falling protein concentration is what turns a clock into a one-way ratchet.

Three checkpoints ask three questions.

• The G₁ checkpoint asks whether the cell is large enough, has enough nutrients, has received a growth signal, and has undamaged DNA. Passing it commits the cell to divide; failing it usually sends the cell to G₀.
• The G₂ checkpoint asks whether DNA replication finished and whether the new copy is damaged.
• The M checkpoint holds anaphase until every single kinetochore is attached to the spindle. A single unattached chromosome keeps the signal switched on, and the cell waits.

Cancer as a failure of these controls

Cancer is not one disease and it is not caused by one mutation. It is what happens when a lineage of cells accumulates enough mutations to escape the controls above, and the genes involved fall into two categories that behave in opposite ways.

Proto-oncogenes code for the proteins that push division forward: growth factors, their receptors, the relay proteins in the cascades of the previous lesson. A mutation that makes one hyperactive turns it into an oncogene. Because the fault is a gain of function, one mutated copy is enough. Think of a stuck accelerator.

Tumour suppressor genes code for the brakes. p53 is the best known: when DNA is damaged, p53 levels rise, the cycle halts for repair, and if the damage cannot be repaired p53 triggers apoptosis. It is mutated in roughly half of all human cancers. Because a brake fails only when it is gone, both copies must usually be knocked out, which is why tumour suppressor mutations behave recessively at the level of the cell. The retinoblastoma protein Rb is the other classic example, and it was the basis of Alfred Knudson's 1971 argument that inherited cancer predisposition means being born with one copy already broken.

Bottom line: HeLa's viral proteins target p53 and Rb. Both brakes go, the G₁ checkpoint stops functioning, and the cells divide without limit. That is why a tumour taken in 1951 is still growing in 2026.

Common misconceptions

• "Chromosome number doubles during S phase." The DNA quantity doubles and the chromatid count doubles. The chromosome count does not change until anaphase separates the sisters.
• "Mitosis is cell division." Mitosis divides the nucleus. Cytokinesis divides the cell. They are separable, which is why some cells, including many fungal cells and early fly embryos, undergo repeated mitosis without cytokinesis and end up with many nuclei in one cytoplasm.
• "Cancer cells divide much faster than normal cells." Many do not. What defines them is that they divide when they should not, ignore signals to stop, and fail to die when damaged.
• "Mitosis produces genetic variation." Barring mutation, it produces two genetically identical cells. Variation comes from meiosis and fertilisation, which is the next lesson.
• "A single mutation causes cancer." Several are typically needed, in different genes and usually in both categories, which is why incidence climbs steeply with age.

What you now know

The cell cycle runs G₁, S, G₂, M, with interphase taking the great majority of the time and G₀ holding cells that are not on the path to division. S phase doubles the DNA and produces sister chromatids without changing the chromosome count; anaphase is where that count changes. Mitosis proceeds through prophase, prometaphase, metaphase, anaphase and telophase, with the spindle attached to kinetochores from opposite poles so that separation is exact, and cytokinesis then splits the cytoplasm by a contractile ring in animals or a cell plate in plants. Counting cells on a slide gives the length of each phase, because the proportion caught in a phase matches the fraction of the cycle it occupies. Cyclins rise and fall to activate cyclin-dependent kinases, and three checkpoints ask about size and signals, about the completeness of replication, and about spindle attachment. Cancer is the accumulated failure of those controls: oncogenes are accelerators stuck on, tumour suppressors such as p53 are brakes that have been removed.

Sources

1. OpenStax. (2018). 10.2 The cell cycle. In Biology 2e. openstax.org
2. OpenStax. (2018). 10.3 Control of the cell cycle. In Biology 2e. openstax.org
3. OpenStax. (2018). 10.4 Cancer and the cell cycle. In Biology 2e. openstax.org
4. Cooper, G. M. (2000). The eukaryotic cell cycle. In The Cell: A Molecular Approach (2nd ed.). NCBI Bookshelf. ncbi.nlm.nih.gov
5. The Nobel Foundation. (2001). The Nobel Prize in Physiology or Medicine 2001: Leland H. Hartwell, R. Timothy Hunt and Paul M. Nurse. nobelprize.org

Eight million possibilities before anything interesting happens

You have twenty-three pairs of chromosomes. When a cell in your gonads makes gametes, each pair lines up at the midline and its two members are sent to opposite poles, and which member goes which way is decided independently for every pair. Two outcomes per pair, twenty-three pairs: 2²³, which is 8,388,608 distinguishable chromosome combinations. That is before a single crossover has occurred, and before the gamete meets another one carrying its own 8,388,608 possibilities. Multiply the two and you are at 2⁴⁶, about seventy trillion, from shuffling alone.

This is why siblings resemble each other and are not identical, and it is the reason sexual reproduction is worth its considerable cost. An organism that reproduces asexually hands its offspring a copy of itself, which is a fine strategy in a stable environment and a poor one in a changing one. Meiosis is the machinery that generates the alternative.

In short: meiosis is one round of DNA replication followed by two divisions, producing four haploid cells from one diploid cell, and it is not simply mitosis performed twice.

The vocabulary you cannot skip

A human body cell is diploid, written 2n = 46: two copies of each of twenty-three chromosome types, one inherited from each parent. The two members of a pair are homologous chromosomes: the same length, the same centromere position, the same genes in the same order, but not necessarily the same alleles. Chromosome 9 from your mother and chromosome 9 from your father both carry the ABO blood group gene at the same locus; one may carry the A allele and the other the O allele.

A gamete is haploid, n = 23: one member of each pair. Fertilisation restores the diploid number, and the arithmetic only works if the halving happens first. An organism that did not halve would double its chromosome number every generation.

Do not confuse homologous chromosomes with sister chromatids. Sister chromatids are two copies of the same chromosome, made in S phase, identical apart from any crossover. Homologous chromosomes are two different chromosomes from two different parents that happen to carry the same genes.

Meiosis I: the division that halves

Everything unusual about meiosis happens in the first division.

1. Prophase I. Homologous chromosomes find each other and pair along their whole length, a process called synapsis. Each pair, now four chromatids together, is a tetrad. While they are held in register, crossing over occurs: matching segments are cut and exchanged between non-sister chromatids. The points where the exchanges happened remain visible as X-shaped junctions called chiasmata, and they physically hold the homologues together until anaphase. Prophase I is long; in human females it begins before birth and pauses there for years or decades.
2. Metaphase I. The tetrads line up on the metaphase plate as pairs, not as individual chromosomes. Which member of each pair faces which pole is random and independent across pairs. This is independent assortment, and it is a physical event at a specific moment, not an abstraction.
3. Anaphase I. Homologous chromosomes are pulled to opposite poles, each still consisting of two chromatids. The centromeres do not split. This is the reductional division: after it, each pole has one member of each pair, and the chromosome number is halved.
4. Telophase I and cytokinesis. Two cells, each with 23 chromosomes, each chromosome still made of two chromatids. There is no S phase before the next division.

Meiosis II: the division that tidies up

Meiosis II is mechanically mitosis performed on a haploid cell. Chromosomes line up singly at metaphase II, the centromeres split, and sister chromatids separate at anaphase II. The result is four haploid cells, each with 23 single chromosomes. The chromosome number is not reduced again; it was already halved.

Track the numbers through the whole process and the pattern is clear.

Point in the process	Chromosomes per cell	Chromatids per cell
Before S phase	46	46
After S phase, entering prophase I	46	92
After meiosis I	23	46
After meiosis II	23	23

Two processes side by side

	Mitosis	Meiosis
Divisions after one replication	one	two
Daughter cells	two	four
Chromosome number in daughters	same as parent	half the parent
Homologues pair	no	yes, in prophase I
Crossing over	essentially none	routine, in prophase I
Genetic result	identical to parent	all four different
Where in the body	growth, repair, asexual reproduction	gamete production only

The single most useful diagnostic when you look at a diagram and cannot tell which process it shows: look at metaphase. If chromosomes are lined up in pairs, two chromosomes wide, it is metaphase I of meiosis. If they are lined up single file, it is mitosis or metaphase II.

Where the variation actually comes from

Four sources, in order of when they act.

1. Mutation is the ultimate source, because it is the only one that creates new alleles rather than rearranging existing ones. Everything below shuffles what mutation has already produced.
2. Crossing over in prophase I produces chromosomes that are mosaics of the two parental versions. With one or more crossovers per chromosome pair, the number of distinct chromosomes that can be produced is effectively unlimited, which is why the 8,388,608 figure is a floor and not a ceiling.
3. Independent assortment at metaphase I gives 2ⁿ combinations, which for humans is 2²³.
4. Random fertilisation multiplies one gamete's possibilities by the other's: 2²³ times 2²³, about seventy trillion, from assortment alone.

Worth holding on to: only mutation adds new information. The other three rearrange it, and rearrangement is what makes a population able to respond to selection quickly, because it can bring favourable alleles that arose separately into the same individual.

When separation fails

Nondisjunction is the failure of chromosomes to separate properly: homologues staying together in anaphase I, or sister chromatids staying together in anaphase II. The gametes that result carry one chromosome too many or too few, and a zygote formed from one has three copies of a chromosome, called trisomy, or one, called monosomy.

Most such zygotes do not survive early development. The ones that do are the conditions visible on a karyotype, a photograph of the chromosomes arranged in pairs by size.

• Trisomy 21, three copies of chromosome 21, causes Down syndrome. Its incidence rises sharply with maternal age, and the leading explanation is the enormous length of the pause in prophase I: an oocyte ovulated at forty has been held in that state for four decades, and the protein complexes holding chromatids together degrade over time.
• XXY produces Klinefelter syndrome; X0, a single X with no second sex chromosome, produces Turner syndrome. Sex chromosome aneuploidies are survivable in a way autosomal ones generally are not, partly because one X is normally inactivated anyway and the Y carries few genes.

Human gamete production is also asymmetric in a way worth noticing. Spermatogenesis yields four functional sperm from each meiosis. Oogenesis divides the cytoplasm unequally, producing one large egg and small polar bodies that are discarded, because the egg must supply everything the zygote needs before implantation. And the human oocyte does not even finish: it arrests in prophase I until ovulation, then arrests again at metaphase II, and completes meiosis II only if a sperm arrives.

Common misconceptions

• "Crossing over happens between sister chromatids." Exchange between identical sisters would change nothing. Crossing over is between non-sister chromatids of homologous chromosomes, which is why it produces new combinations.
• "Meiosis II halves the chromosome number again." The halving happened in anaphase I. Meiosis II separates chromatids, exactly as mitosis does.
• "DNA is replicated before each of the two divisions." Replication happens once. If it happened twice, the number would never be reduced.
• "Homologous chromosomes are identical." They carry the same genes at the same loci, but frequently different alleles. That difference is the entire basis of the next module.
• "Meiosis produces gametes in all organisms." In many plants and fungi, meiosis produces spores, and gametes are made later by mitosis. The generalisation that holds is that meiosis produces haploid cells.

The takeaway

Meiosis takes one diploid cell through a single round of replication and two divisions to make four haploid cells that differ from each other and from the parent. Meiosis I is the unusual one: homologues pair in prophase I, cross over at chiasmata, line up as pairs at metaphase I, and separate at anaphase I, which is where the chromosome number is halved. Meiosis II is mechanically mitosis, splitting centromeres and separating sister chromatids. Variation has four sources: mutation creates alleles, crossing over builds mosaic chromosomes, independent assortment gives 2²³ combinations in humans, and random fertilisation squares that number. When separation fails, nondisjunction produces gametes with an extra or a missing chromosome, and the survivable outcomes are the trisomies and sex chromosome aneuploidies a karyotype reveals. Keep the diagnostic in your pocket: pairs at the metaphase plate mean meiosis I, single file means mitosis or meiosis II.

Sources

1. OpenStax. (2018). 11.1 The process of meiosis. In Biology 2e. openstax.org
2. OpenStax. (2018). 11.2 Sexual reproduction. In Biology 2e. openstax.org
3. National Human Genome Research Institute. (n.d.). Crossing over. In Talking Glossary of Genomic and Genetic Terms. genome.gov
4. National Human Genome Research Institute. (n.d.). Meiosis. In Talking Glossary of Genomic and Genetic Terms. genome.gov

5,474 round seeds and 1,850 wrinkled ones

In February and March of 1865, an Augustinian friar named Gregor Mendel read a paper to the Natural History Society in Brno. He had spent eight years crossing pea plants in the monastery garden, something like twenty-eight thousand of them, tracking seven characteristics that came in two clean versions each. For seed shape, his second generation gave 5,474 round seeds and 1,850 wrinkled, which is a ratio of 2.96 to 1.

The number that matters is how close that is to 3 to 1. Mendel got similar ratios for all seven traits, and he drew from them the conclusion nobody else had: inheritance is particulate. Something discrete is passed on, not blended. His paper was published in 1866, cited a handful of times, and effectively ignored for thirty-four years.

Peas were an excellent choice, and the choice was not luck. They self-pollinate, so true-breeding lines are easy to establish; they can be cross-pollinated by hand, so a controlled cross is possible; they grow fast; and the seven traits Mendel picked are each controlled by a single gene with two alleles, one fully dominant. Had he chosen a trait like plant height in a species where dozens of genes contribute, he would have seen a smear rather than a ratio, and there would have been nothing to count.

The words, used precisely

• A gene is a unit of heredity at a particular locus on a chromosome. An allele is one version of that gene.
• A diploid organism has two alleles per locus. If they match it is homozygous, written RR or rr; if not, heterozygous, Rr.
• The genotype is the allele pair. The phenotype is what you can observe.
• An allele is dominant if the heterozygote shows its phenotype, and recessive if it does not. Capital letters mark dominance by convention, not importance.
• Generations are labelled P for the true-breeding parents, F₁ for their offspring, F₂ for the offspring of an F₁ self-cross.

The core of it: dominant does not mean common, strong or better. Polydactyly, having extra fingers, is caused by a dominant allele and is rare. Being able to taste PTC is dominant and widespread. Frequency in a population is set by selection and history, not by dominance.

Segregation, and where it happens

Cross a true-breeding round-seeded plant (RR) with a true-breeding wrinkled one (rr). Every F₁ is Rr and every seed is round. The wrinkled character has not been destroyed, because self-crossing the F₁ brings it back in a quarter of the F₂.

Rr x Rr	R	r
R	RR round	Rr round
r	Rr round	rr wrinkled

Genotypes come out 1 RR : 2 Rr : 1 rr. Phenotypes come out 3 round : 1 wrinkled. The law of segregation says the two alleles of a gene separate during gamete formation so each gamete carries only one. You met the physical event in the last lesson: anaphase I, where homologous chromosomes go to opposite poles.

A round seed can be RR or Rr, and you cannot tell by looking. The test cross settles it: cross the unknown with a homozygous recessive, rr. If the unknown is RR, all offspring are round. If it is Rr, half are wrinkled. This is still how breeders identify carriers, and it works because the recessive parent contributes nothing that could mask anything.

Two genes at once

Mendel then crossed plants differing in two traits: round yellow seeds (RRYY) with wrinkled green (rryy). The F₁ were all RrYy, round and yellow. Self-crossing them gave four phenotypes in the F₂ in the ratio 9 round yellow : 3 round green : 3 wrinkled yellow : 1 wrinkled green.

That 9:3:3:1 is the signature of two genes behaving independently, and you do not need a sixteen-cell Punnett square to get it. Treat the genes separately and multiply. Each gene alone gives 3 dominant : 1 recessive, so three quarters of the plants are round and three quarters are yellow. Round and yellow together is three quarters times three quarters, which is nine sixteenths. Round and green is three quarters times one quarter, three sixteenths. And so on.

That multiplication is the product rule: the probability of two independent events both happening is the product of their separate probabilities. Its partner is the sum rule: the probability of either of two mutually exclusive events is the sum. Together they replace enormous Punnett squares with a line of arithmetic. For a cross AaBbCc x AaBbCc, the chance of an aabbcc offspring is one quarter times one quarter times one quarter, which is 1 in 64. The chance of an offspring showing all three dominant phenotypes is three quarters cubed, which is 27 in 64. Drawing that as a Punnett square would need sixty-four boxes.

The law of independent assortment says alleles of different genes are distributed to gametes independently of one another. Its physical basis is the random orientation of each homologous pair at metaphase I. It holds only for genes on different chromosomes, or far apart on the same one, which is the subject of the next lesson.

When your counts do not match your prediction

You predict 3:1 and you count 88 and 34. Is that a 3:1 ratio with sampling noise, or is your model wrong? Judging by eye does not work, because the amount of deviation you should tolerate depends on the sample size. The chi-square goodness-of-fit test answers the question quantitatively.

chi-square = sum of (observed - expected)2 / expected

Four rules govern its use. Use raw counts, never percentages or ratios. Compute the expected numbers from your hypothesis applied to your actual total. Degrees of freedom are the number of phenotypic classes minus one. And compare the result with the critical value at the 0.05 significance level.

Degrees of freedom	1	2	3	4
Critical value at p = 0.05	3.84	5.99	7.81	9.49

If your chi-square is below the critical value, the deviation is the size chance alone would routinely produce, and you do not reject the hypothesis. If it is above, a deviation that large would occur less than five percent of the time if the hypothesis were true, and you reject it.

Mendel's seeds, worked through

Hypothesis: seed shape follows a 3:1 ratio. Total counted: 7,324.

Class	Observed	Expected	o - e	(o - e)2	(o - e)2 / e
Round	5,474	5,493	-19	361	0.066
Wrinkled	1,850	1,831	+19	361	0.197
Total	7,324	7,324			0.263

Expected round is 0.75 times 7,324, which is 5,493; expected wrinkled is 0.25 times 7,324, which is 1,831. Two classes, so one degree of freedom, and the critical value is 3.84. Our chi-square is 0.263, far below it.

Conclusion, stated properly: the deviation between the observed and expected counts is small enough to be attributed to chance, so we do not reject the hypothesis that seed shape is inherited in a 3:1 ratio.

Note what that sentence does not say. It does not say the hypothesis is proved. Chi-square can only fail to reject; other models might fit these data equally well. Ronald Fisher made a sharper point about Mendel's numbers in 1936: across all seven traits, Mendel's chi-square values are consistently smaller than chance should allow, as though the data were too good. The debate over what that means, unconscious bias in classifying ambiguous seeds, a helpful gardener, or a statistical artefact of how the counting stopped, has run ever since. It does not touch the conclusions, which have been confirmed a thousand times over.

A second test, where the answer flips

Yellow-coated mice are a famous problem. Cross yellow with yellow and you never get a true-breeding yellow line; you get yellow and agouti offspring in a ratio that does not look like 3:1. Suppose you count 135 offspring: 88 yellow, 47 agouti.

Test the 3:1 hypothesis first. Expected: 101.25 yellow, 33.75 agouti. Deviations are 13.25 either way, and 13.25 squared is 175.6. Divided by the expected values: 175.6 / 101.25 = 1.73, and 175.6 / 33.75 = 5.20. The chi-square is 6.93, above the critical 3.84 with one degree of freedom, so reject. Whatever is happening, it is not a simple 3:1.

Now test 2 yellow : 1 agouti. Expected: 90 and 45. Deviations of 2, squared is 4. 4 / 90 = 0.044 and 4 / 45 = 0.089, giving a chi-square of 0.13, well below 3.84, so do not reject. The explanation is that the yellow allele is dominant for coat colour and lethal when homozygous: the quarter of embryos that would be homozygous yellow die before birth, and the surviving 3 : 1 becomes 2 : 1 among the living.

Why this matters: chi-square does not tell you what is true. It tells you which hypotheses your data are compatible with. Rejecting 3:1 is what sent geneticists looking for the lethal allele in the first place.

Common misconceptions

• "A 3:1 ratio means exactly three out of every four." It is a probability per offspring, not a quota. Four children of two carriers can all be affected; the chance is 1 in 256, and it happens.
• "Chi-square proves the hypothesis." It can only reject or fail to reject. Failing to reject means your data are consistent with the model, which is weaker and more honest.
• "Use percentages in the calculation." Chi-square depends on sample size, which is the whole point of it. Percentages destroy that information, and 60 percent out of 10 individuals means something very different from 60 percent out of 1,000.
• "The expected values come from the ideal ratio, so they must be whole numbers." They are computed from your actual total and are usually fractional. Leave them fractional.
• "Mendel's laws are universal." They describe genes on separate chromosomes with simple dominance. Linkage, incomplete dominance, epistasis and polygenic traits all break the ratios, which is the next lesson.

Recap

Mendel counted 5,474 round and 1,850 wrinkled seeds and read particulate inheritance out of the ratio. Alleles segregate, one to each gamete, because homologues separate at anaphase I; alleles of different genes assort independently because each pair orients independently at metaphase I. A monohybrid cross of two heterozygotes gives 1:2:1 genotypes and 3:1 phenotypes; a dihybrid gives 9:3:3:1; a test cross against a homozygous recessive reveals an unknown genotype. The product and sum rules do the same work as a Punnett square with far less drawing, and they scale to three genes where a square would need sixty-four boxes. When counts and predictions disagree, chi-square decides: sum the squared deviations divided by expectations, take degrees of freedom as classes minus one, and compare with 3.84, 5.99, 7.81 or 9.49. Mendel's seed data give 0.263 against a critical 3.84 and survive; the yellow mice reject 3:1 at 6.93 and fit 2:1 at 0.13, which is how a lethal allele announces itself.

Sources

1. OpenStax. (2018). 12.1 Mendel's experiments and the laws of probability. In Biology 2e. openstax.org
2. OpenStax. (2018). 12.3 Laws of inheritance. In Biology 2e. openstax.org
3. NIST/SEMATECH. (2012). Chi-square goodness-of-fit test. In e-Handbook of Statistical Methods. itl.nist.gov
4. Mendel, G. (1866). Versuche uber Pflanzen-Hybriden. Verhandlungen des naturforschenden Vereines in Brunn, 4, 3-47.
5. Fisher, R. A. (1936). Has Mendel's work been rediscovered? Annals of Science, 1(2), 115-137.

A ratio that was wrong by a factor of four

In 1905 William Bateson and Reginald Punnett crossed sweet peas differing in two characteristics, flower colour and pollen grain shape, and self-crossed the F₁. Mendel's rules predicted 9:3:3:1. Out of 381 F₂ plants they counted 284 purple with long pollen, 21 purple with round pollen, 21 red with long pollen, and 55 red with round pollen.

Run the chi-square from the last lesson. Expected counts on a 9:3:3:1 model are 214.3, 71.4, 71.4 and 23.8. The four terms come to 22.7, 35.6, 35.6 and 40.9, and the total is about 134.7 against a critical value of 7.81 for three degrees of freedom. The hypothesis of independent assortment is rejected, and not marginally.

Look at which classes are wrong. The two parental combinations, purple-long and red-round, are far more common than predicted; the two new combinations are far rarer. Something is holding purple and long together. Bateson and Punnett named the phenomenon coupling and could not explain it. The explanation, when it came from Thomas Hunt Morgan's laboratory a decade later, was that the two genes ride on the same chromosome.

Remember: Mendel's ratios are what you get under specific conditions: one gene per trait, two alleles, complete dominance, no gene interaction, and genes on separate chromosomes. Break any one condition and the ratio changes in a characteristic way. This lesson is a catalogue of those signatures.

Dominance is not the only relationship

Incomplete dominance. Cross a red snapdragon with a white one and every F₁ flower is pink. It looks like the blending inheritance Mendel disproved, until you self-cross the F₁ and get 1 red : 2 pink : 1 white. Red and white have not merged; they were never altered. One functional allele simply makes less pigment than two. The genotypic ratio, 1:2:1, is exactly Mendel's; only the phenotypes fail to collapse into two classes.

Codominance. In codominance both alleles are fully expressed in the heterozygote, side by side rather than averaged. A roan cow is not pink; it has white hairs and red hairs both present. Human MN blood groups behave the same way, with heterozygotes carrying both antigens on their red cells.

Multiple alleles. A gene can have more than two versions in a population, even though any one person carries two. The ABO gene has three common alleles: I^A, I^B and i. I^A and I^B are codominant with each other and both dominant to i.

Blood group	Possible genotypes
A	IAIA or IAi
B	IBIB or IBi
AB	IAIB
O	ii

Work one cross. A group A woman whose father was group O must be I^Ai. A group B man whose mother was group O must be I^Bi. Their children come out one quarter I^AI^B (AB), one quarter I^Ai (A), one quarter I^Bi (B) and one quarter ii (O). Two parents, neither of them group O, with a one in four chance of an O child at every conception.

Genes that interfere with each other

Epistasis is one gene masking another. Coat colour in Labrador retrievers uses two: one gene decides whether the pigment is black (B) or brown (b), and a second decides whether pigment is deposited in the hair at all (E) or not (e). A dog that is ee has no pigment deposited regardless of its B genotype, and is yellow.

Cross two BbEe dogs and the sixteen boxes still exist, but the phenotypes group differently: 9 black (B_E_), 3 brown (bbE_), and 4 yellow (B_ee and bbee combined, because you cannot tell them apart by looking). The ratio is 9:3:4, and any 9:3:4, 12:3:1 or 9:7 in place of 9:3:3:1 is a signature of epistasis. Yellow Labradors carry a hidden black or brown genotype, which is why two yellow parents can produce a chocolate puppy.

Pleiotropy runs the other way: one gene affecting many characteristics. The sickle cell allele changes a single amino acid in beta-globin, and the consequences include distorted red cells, blocked capillaries, episodes of severe pain, damage to the spleen, delayed growth and increased infection risk. Nothing about the gene is about pain or the spleen; those follow downstream from a protein that behaves differently when deoxygenated.

Polygenic inheritance is many genes affecting one characteristic. Human height, skin pigmentation and body mass are governed by many loci each contributing a small amount, and the result is a continuous bell-shaped distribution rather than discrete classes. Mendel would have found nothing countable here, which is the strongest argument that his choice of seven simple traits was inspired rather than lucky.

The environment enters too. A hydrangea with one genotype produces blue flowers in acidic soil and pink in alkaline. A Siamese cat is pale on its warm body and dark at its ears, paws, tail and face, because the enzyme that makes the dark pigment is temperature-sensitive and works only in the cooler extremities. The genotype fixes a range of possible outcomes; conditions choose from within it.

Genes on the sex chromosomes

In 1910 Morgan found a single white-eyed male among his red-eyed fruit flies. Crossing it revealed a pattern Mendel's laws cannot produce: the results of a cross depended on which parent carried the trait. The gene is on the X chromosome, and the Y carries no equivalent copy.

Females are XX and can be homozygous or heterozygous. Males are XY, and for any gene on the X they have exactly one allele, expressed whether it is dominant or recessive. This is called hemizygous, and it explains the asymmetry directly.

Do the population arithmetic, using Hardy-Weinberg from Module 1. If the recessive allele has frequency q, the proportion of affected males is simply q, because one copy is enough. The proportion of affected females is q². For red-green colour blindness in populations where q is about 0.08, that is 8 percent of males and about 0.6 percent of females: a ratio of roughly thirteen to one, arising from nothing but the difference between one copy and two.

Two rules follow, and both appear constantly in pedigree questions. An affected male inherited the allele from his mother, since his single X came from her. And an affected father cannot pass an X-linked condition to his sons, because he gives them a Y, but all his daughters will be carriers.

Linkage, and how it becomes a map

Genes on the same chromosome travel together unless a crossover separates them, which is exactly what Bateson and Punnett had run into. Morgan's laboratory turned the problem into a measuring instrument.

The insight is geometric. Crossovers happen at more or less random positions along a chromosome, so two genes far apart are separated more often than two genes close together. The frequency of recombinant offspring is therefore a measure of distance.

Work an example. A fly heterozygous for two linked genes is testcrossed against a homozygous recessive, and 2,300 offspring are scored:

Class	Count
Parental type 1	965
Parental type 2	944
Recombinant type 1	206
Recombinant type 2	185

Recombinants total 206 + 185 = 391. Recombination frequency is 391 divided by 2,300, which is 0.170, or 17 percent. By the convention Alfred Sturtevant introduced in 1913, while still an undergraduate in Morgan's laboratory, 1 percent recombination is defined as one map unit, also called a centimorgan. These two genes are therefore 17 map units apart. Do this for three or more genes and the distances can be assembled into a linear map, which is how the first chromosome maps were made, decades before anyone could read a base sequence.

The upshot: the maximum recombination frequency is 50 percent, not 100. At 50 percent the two genes behave exactly as if they were on different chromosomes, because a crossover between them is essentially certain. Genes far apart on the same chromosome are therefore indistinguishable from unlinked genes in a single cross, and are linked only through the genes lying between them.

Reading a pedigree

Pedigrees are the human substitute for a controlled cross, and three questions usually settle the mode of inheritance.

• Does the trait skip generations? If two unaffected parents have an affected child, the allele must be recessive and both parents carriers.
• Does every affected individual have an affected parent? That points to a dominant allele.
• Are affected individuals overwhelmingly male, with affected sons of unaffected mothers? That points to X-linked recessive. An affected father with unaffected children of both sexes argues against it.

Common misconceptions

• "Incomplete dominance is blending inheritance." Blending would make the original phenotypes unrecoverable. The F₂ of pink snapdragons is 1 red : 2 pink : 1 white, so the alleles were intact all along.
• "Codominance and incomplete dominance are the same." Incomplete dominance produces an intermediate; codominance produces both phenotypes at once. Pink versus a coat of mixed red and white hairs.
• "A recombination frequency can exceed 50 percent." It cannot. Multiple crossovers between two distant genes restore parental combinations as often as they break them, so the measured value saturates at 50.
• "X-linked means only males are affected." Females can be homozygous recessive and affected. They are simply much rarer, at q² rather than q.
• "Polygenic and pleiotropic mean the same thing." Polygenic is many genes to one trait; pleiotropic is one gene to many traits. The arrows point in opposite directions.

The short version

Bateson and Punnett's sweet peas failed a 9:3:3:1 test with a chi-square near 135, and the excess of parental classes was the first evidence of linkage. Incomplete dominance gives an intermediate heterozygote and a 1:2:1 phenotypic ratio; codominance shows both alleles at once; multiple alleles such as ABO give more combinations than a two-allele gene can. Epistasis converts 9:3:3:1 into 9:3:4 or similar when one gene masks another, pleiotropy is one gene with many effects, and polygenic inheritance produces continuous variation that no Punnett square can capture. Environment shifts phenotype within the range a genotype allows, as a Siamese cat's temperature-sensitive pigment shows. Genes on the X are hemizygous in males, so an X-linked recessive appears in males at frequency q and in females at q². Linked genes recombine at a frequency proportional to their separation, 1 percent equals one map unit, and 50 percent is the ceiling at which linkage becomes undetectable.

Sources

1. OpenStax. (2018). 12.2 Characteristics and traits. In Biology 2e. openstax.org
2. OpenStax. (2018). 13.1 Chromosomal theory and genetic linkage. In Biology 2e. openstax.org
3. National Heart, Lung, and Blood Institute. (n.d.). Sickle cell disease. nhlbi.nih.gov
4. The Nobel Foundation. (1933). The Nobel Prize in Physiology or Medicine 1933: Thomas H. Morgan. nobelprize.org

One band where there should have been two

In 1957 Matthew Meselson and Franklin Stahl grew Escherichia coli for many generations in a medium whose only nitrogen source contained the heavy isotope nitrogen-15. Every base in every DNA molecule in those cells was built from heavy nitrogen. Then they washed the cells into ordinary nitrogen-14 medium and let them divide exactly once, spun the extracted DNA in a caesium chloride density gradient, and looked at where it settled.

Three models were live at the time. If replication were conservative, the original heavy helix would stay intact and a brand new light one would appear alongside it: two bands, one heavy and one light. If it were semiconservative, every molecule would have one old heavy strand and one new light strand: a single band at intermediate density. If it were dispersive, with old and new material scattered through both strands, that would also give one intermediate band after one generation, but would keep giving a single band after two.

After one generation there was one band, at intermediate density. Conservative replication was dead. After a second generation there were two bands, one intermediate and one light, which is what semiconservative replication predicts and dispersive replication does not. The experiment settled the question completely in a single figure, which is why it is routinely called the most beautiful experiment in biology.

What matters here: every DNA molecule in you contains one strand that is a physical inheritance and one that was assembled to match it. The old strand is the template, and that is what makes copying accurate.

Why the structure dictates the mechanism

The double helix Watson and Crick proposed in 1953, using Rosalind Franklin's X-ray diffraction data and Erwin Chargaff's observation that A always equals T and G always equals C, has three features that determine everything below. Watson, Crick and Maurice Wilkins shared the 1962 Nobel Prize for it; Franklin had died in 1958 and the prize is not awarded posthumously.

• Complementary base pairing. A pairs with T through two hydrogen bonds, G with C through three. Either strand therefore specifies the other exactly.
• Antiparallel strands. One runs 5' to 3', the other 3' to 5'. The numbers refer to carbons on the deoxyribose sugar, and the direction is a physical property of the backbone, not a label.
• A sugar-phosphate backbone outside, bases inside. The bases are protected, and the information is readable by unzipping rather than dismantling.

One consequence controls the whole of replication: DNA polymerase can add a nucleotide only to a free 3' hydroxyl group. Synthesis therefore runs strictly 5' to 3', and the enzyme cannot start from nothing; it needs an existing 3' end to extend.

What happens at a replication fork

1. Helicase unwinds the double helix at an origin of replication, breaking the hydrogen bonds and opening a fork. Bacteria have one origin; a human chromosome has thousands, or copying would take weeks.
2. Single-strand binding proteins coat the separated strands to stop them snapping back together.
3. Topoisomerase works ahead of the fork, nicking and resealing the DNA to release the torsional strain that unwinding creates. Without it the molecule would knot.
4. Primase lays down a short RNA primer, because DNA polymerase needs a 3' end to extend and cannot make one.
5. DNA polymerase III extends from the primer, adding nucleotides complementary to the template at roughly a thousand bases a second in bacteria.
6. DNA polymerase I removes the RNA primers and fills the gaps with DNA, and ligase seals the remaining breaks in the backbone.

Now the asymmetry. The fork opens in one direction, but the two template strands point in opposite directions. On one template, the 5' to 3' rule lets the polymerase follow the fork continuously: this is the leading strand. On the other, synthesis must run away from the fork, so the polymerase makes a short piece, falls off, waits for more template to be exposed, and starts again further back. These short pieces are Okazaki fragments, each needing its own primer, and ligase later stitches them into one strand. The lagging strand is not a design flaw; it is the only arrangement consistent with a polymerase that works in one direction on an antiparallel molecule.

Accuracy comes in three layers. DNA polymerase inserts the wrong base roughly once in a hundred thousand. Its own proofreading activity, which backs up and excises a mismatched nucleotide, improves that by about a hundredfold. Mismatch repair enzymes then scan the finished duplex and correct what remains, bringing the final error rate to something near one per billion bases. Three imperfect mechanisms in series produce a very good result, which is a pattern worth noticing because biology uses it everywhere.

Linear chromosomes have an end problem: the final primer on the lagging strand cannot be replaced by DNA, so a little is lost each round. Telomeres, repeated non-coding sequences at chromosome ends, are the buffer that gets eaten. The enzyme telomerase rebuilds them in germ cells and stem cells but is switched off in most somatic cells, which is one reason ordinary cells stop dividing after a limited number of generations. Reactivating telomerase is one of the changes most cancer cells make, and HeLa cells from the earlier lesson have done exactly that.

Transcription: copying one gene into RNA

RNA differs from DNA in three ways: the sugar is ribose rather than deoxyribose, uracil replaces thymine, and the molecule is single-stranded. Those differences make it a short-lived working copy rather than an archive.

RNA polymerase binds a promoter, a sequence upstream of the gene that marks where to start and which strand to read. Only one strand, the template strand, is read, and it is read 3' to 5' while the RNA is built 5' to 3'. The other strand, the coding strand, has the same sequence as the mRNA except that it carries T where the RNA carries U. Transcription ends at a terminator sequence, and unlike DNA polymerase, RNA polymerase needs no primer.

In a eukaryote the transcript is not ready to use. Three modifications happen in the nucleus.

• A 5' cap, a modified guanine, is added to the front. It helps the ribosome find the message and protects it from degradation.
• A poly-A tail, a run of adenine nucleotides, is added to the end, which also protects it and influences how long the message survives in the cytoplasm.
• Splicing. Non-coding stretches called introns are cut out by a complex of RNA and protein called the spliceosome, and the coding exons are joined. Most human genes have introns, some of them far longer than the coding sequence.

Splicing is not merely tidying. Alternative splicing lets one gene produce several different proteins by including different subsets of exons, and it is a large part of the answer to why about twenty thousand human genes can specify a much larger number of proteins.

The code itself

Three bases specify one amino acid. Two would give only sixteen combinations, not enough for twenty amino acids; three give sixty-four. Sixty-one code for amino acids and three are stop signals, UAA, UAG and UGA. AUG is both the start signal and the codon for methionine.

The code is redundant but not ambiguous: most amino acids have several codons, and no codon specifies more than one amino acid. The redundancy is patterned, with synonymous codons usually differing at the third base, which means many third-position mutations change nothing. It is also close to universal: the same codons mean the same things in a bacterium, a fungus and a whale, which is powerful evidence for common descent and the reason a human gene can be expressed in bacteria to make insulin. There are minor exceptions, notably in mitochondria.

Marshall Nirenberg and Heinrich Matthaei cracked the first codon in 1961 by feeding a cell-free system an artificial RNA of nothing but uracil. The system produced a polypeptide of nothing but phenylalanine, so UUU means phenylalanine, and the rest of the table followed within a few years.

Translation, worked end to end

Take a DNA template strand reading 3'-TAC GGA TTA CCA ATC-5'.

Transcribe it, remembering that A pairs with U in RNA:

mRNA: 5'-AUG CCU AAU GGU UAG-3'

Now translate. AUG is start and codes methionine. CCU is proline. AAU is asparagine. GGU is glycine. UAG is a stop codon and codes for nothing.

peptide: Met - Pro - Asn - Gly - stop

The machinery that does this has three parts. The ribosome is built from ribosomal RNA and protein in two subunits, and it has three sites where tRNA can sit: A for arriving, P for the growing peptide, E for exit. Transfer RNA is a folded molecule with an anticodon at one end that pairs with the mRNA codon and an attachment site at the other carrying the matching amino acid. Aminoacyl-tRNA synthetases, twenty of them, are the enzymes that charge each tRNA with the right amino acid, and they are where the correspondence between codon and amino acid is physically enforced.

Elongation is a three-step cycle repeated for every codon: a charged tRNA pairs into the A site, a peptide bond forms between the growing chain and the new amino acid, and the ribosome translocates one codon along, shifting the tRNAs to P and E. The peptide bond is catalysed by ribosomal RNA, not by protein, which makes the ribosome a ribozyme and is one of the strongest arguments that RNA came before protein. Termination occurs when a stop codon enters the A site: no tRNA matches, a release factor binds instead, and the finished polypeptide is let go.

What mutations do to all this

Type	What changes	Effect
Silent	a base substitution that gives a synonymous codon	none; the protein is unchanged
Missense	a substitution giving a different amino acid	ranges from nothing to severe, depending on the amino acid and its position
Nonsense	a substitution creating a premature stop codon	a truncated and usually non-functional protein
Frameshift	an insertion or deletion of a number of bases not divisible by three	every codon downstream is misread; usually catastrophic

The sickle cell allele from Module 2 is a missense mutation: one base changes, glutamic acid at position six of the beta-globin chain becomes valine, and a hydrophobic patch appears on the surface of a protein that should not have one. One base out of three billion.

Now delete a base from our worked example. Remove the first C of the mRNA and the message reads AUG CUA AUG GUU AG. The peptide becomes Met-Leu-Met-Val and then runs off the end of a message that no longer has a stop codon in frame. Nothing after the deletion means what it did. That is why frameshifts are so much more damaging than substitutions.

Key idea: information flows DNA to RNA to protein, and each arrow is a different mechanism with its own enzymes, its own error rate and its own regulation. Naming the arrows is not the same as knowing them.

Common misconceptions

• "DNA leaves the nucleus to be translated." It does not. A working copy in RNA is made and exported; the archive stays put.
• "The lagging strand is made 3' to 5'." Every fragment is made 5' to 3'. Only the overall direction of the growing strand appears backwards, because the fragments are laid down in sequence away from the fork.
• "One gene makes one protein." Alternative splicing means one gene can specify many related proteins, and some genes make functional RNA and no protein at all.
• "Redundant means the code is ambiguous." The opposite. Several codons can mean one amino acid, but no codon has two meanings, so translation is never uncertain.
• "Mutations are always harmful." Most are silent or nearly so. Harmful ones are removed by selection, neutral ones drift, and the rare beneficial ones are the raw material Module 1 depends on.

Pulling it together

Meselson and Stahl put one band where conservative replication demanded two, and semiconservative replication has been the answer ever since. Replication depends on antiparallel strands and a polymerase that adds only to a 3' end, which forces one strand to be built continuously and the other in Okazaki fragments, each needing its own primer and ligase to join them. Three layers of checking, insertion accuracy, proofreading and mismatch repair, take the error rate to about one in a billion, and telomeres absorb the loss that linear ends make unavoidable. Transcription reads one template strand 3' to 5' and builds RNA 5' to 3' from a promoter, and in eukaryotes the transcript is capped, tailed and spliced, with alternative splicing multiplying the proteins a gene can specify. The code is triplet, redundant, unambiguous and nearly universal, with AUG starting and three codons stopping. Translation charges tRNAs by synthetase, pairs anticodon to codon in the ribosome's A site, forms peptide bonds with catalytic rRNA, and stops when a release factor answers a stop codon. Substitutions can be silent, missense or nonsense; an insertion or deletion that is not a multiple of three shifts the reading frame and destroys everything downstream.

Sources

1. OpenStax. (2018). 14.3 Basics of DNA replication. In Biology 2e. openstax.org
2. OpenStax. (2018). 15.1 The genetic code. In Biology 2e. openstax.org
3. OpenStax. (2018). 15.5 Ribosomes and protein synthesis. In Biology 2e. openstax.org
4. National Human Genome Research Institute. (n.d.). Translation. In Talking Glossary of Genomic and Genetic Terms. genome.gov
5. The Nobel Foundation. (1962). The Nobel Prize in Physiology or Medicine 1962: Francis Crick, James Watson and Maurice Wilkins. nobelprize.org

A growth curve with a step in it

Feed Escherichia coli a broth containing both glucose and lactose and plot the number of cells against time. You do not get a smooth curve. The culture grows, then stalls for a while, then grows again. Jacques Monod described this two-phase pattern, diauxie, in 1941, and it took twenty years to explain: the bacteria consume the glucose first, and only when it is gone do they pause to build the enzymes for digesting lactose.

The pause is the interesting part. It means the enzymes were not there before, and that the cell can decide to make them. In 1961 Francois Jacob and Jacques Monod published the model that explained how, and in 1965 they shared the Nobel Prize with Andre Lwoff for it. An E. coli cell carries roughly four thousand genes and cannot afford to express them all at once. Regulation is not a refinement on top of genetics; it is most of what a genome does.

The lac operon, switch by switch

An operon is a cluster of genes with a shared control region, transcribed together into one mRNA. It exists in prokaryotes because genes with a common job can then be regulated with a single switch. The lac operon has three parts and one outside regulator.

• A promoter, where RNA polymerase binds.
• An operator, a short sequence just downstream of the promoter.
• Three structural genes, coding for the enzyme that splits lactose, the permease that transports it into the cell, and a third enzyme.
• A separate regulatory gene, elsewhere on the chromosome, which is always transcribed and produces the repressor protein.

With no lactose present, the repressor sits on the operator and physically blocks RNA polymerase. The genes are off. When lactose enters, a little of it is converted to allolactose, which binds the repressor and changes its shape so that it can no longer grip the operator. The repressor falls off, polymerase proceeds, and the enzymes are made. This is negative inducible control: the default is off, and the substrate switches it on.

That is only half the system, and the other half explains the diauxic step. Even with the repressor gone, the lac promoter is a poor one and polymerase binds it weakly. A protein called CAP improves the binding, but CAP works only when bound to cyclic AMP, the same second messenger you met in Module 5. Glucose suppresses cyclic AMP levels. So when glucose is plentiful, cyclic AMP is low, CAP is inactive, and the operon runs at a trickle even if lactose is present.

Glucose	Lactose	Repressor on operator?	CAP active?	Transcription
present	absent	yes	no	none
present	present	no	no	very low
absent	absent	yes	yes	none
absent	present	no	yes	high

Why this matters: two independent switches in series implement a genuinely sensible rule. Make the lactose enzymes only if there is lactose to digest and no better food available. Neither switch alone could express that.

The trp operon runs the logic in reverse. Its five genes make the enzymes that synthesise tryptophan, and the default is on, because a cell that cannot make tryptophan cannot make protein. When tryptophan accumulates, it binds the repressor as a corepressor and activates it, and the activated repressor shuts the operon down. Inducible operons control catabolism, switching on when the substrate appears; repressible operons control anabolism, switching off when the product accumulates.

Eukaryotes regulate at every step

A bacterium transcribes and translates in one compartment, almost simultaneously. A eukaryotic cell separates the two, and every stage in between becomes a control point.

Level	Mechanism
Chromatin structure	Tightly packed DNA is unreadable. Acetylation of histone tails loosens packing and permits transcription; methylation of DNA usually silences a region.
Transcription initiation	Transcription factors bind promoters and distant enhancers; a gene may need a specific combination of factors before polymerase can start.
RNA processing	Alternative splicing selects which exons are included, so one gene yields several proteins.
mRNA lifetime	Short-lived messages give short-lived signals. Small regulatory RNAs, including microRNAs, bind messages and mark them for destruction or block their translation.
Translation	Initiation factors can be blocked, holding a stockpiled message untranslated until a signal arrives.
After translation	Proteins are cut, folded, modified by phosphorylation, or tagged with ubiquitin and destroyed.

Some of these changes are heritable through cell division without any alteration to the DNA sequence, which is what epigenetics means. The clearest visible example is a calico cat. Coat colour in cats is X-linked; early in the development of a female embryo, one X chromosome in each cell is inactivated at random and packed away, and every descendant of that cell keeps the same X switched off. The result is patches, each a clone of one early cell. A calico cat is a mosaic, and this is why almost all calico cats are female.

How one cell becomes many kinds of cell

Your neurons and your liver cells have the same genome. What differs is which genes are expressed, and the question development answers is how that difference gets established in the right places.

Three mechanisms do most of the work. Cytoplasmic determinants are molecules unevenly distributed in the egg, so the first divisions hand different daughter cells different instructions. Induction is one group of cells signalling to a neighbouring group and changing its fate. And morphogen gradients let position be read as concentration: a substance diffusing from one end of an embryo is at high concentration at one end and low at the other, and cells switch on different genes at different thresholds.

Then the body plan has to be organised, and this is where the surprise came. In the 1970s and 1980s Edward Lewis, Christiane Nusslein-Volhard and Eric Wieschaus ran systematic screens for mutations that disrupted the layout of the fruit fly embryo, work that won the 1995 Nobel Prize. They found genes whose mutation does not damage a structure but puts the wrong structure in a place: a fly with legs growing where its antennae should be, or with a second pair of wings where the balancing organs belong.

These are homeotic genes, and the family known as Hox genes has two properties worth memorising. They are arranged on the chromosome in the same order as the body regions they control, front to back. And they are conserved across animals as different as flies, mice and humans, closely enough that a mouse gene can substitute for its fly counterpart in some experiments. The toolkit for building an animal body is largely shared, and the differences between animals lie substantially in when and where those genes are switched on rather than in having different genes.

The tools this understanding produced

Every technique below is an application of something already covered in this course.

• Restriction enzymes are bacterial defences that cut DNA at specific short sequences. Many cut in a staggered way, leaving single-stranded sticky ends that pair with any other fragment cut by the same enzyme. That is what makes DNA from two species joinable.
• Plasmids are small circular DNA molecules that bacteria take up and replicate. Cut a plasmid and a human gene with the same enzyme, join them with ligase, and transform bacteria with the result, and the bacteria make the human protein. Human insulin produced this way was approved in 1982, replacing insulin extracted from pigs and cattle.
• Gel electrophoresis separates DNA fragments by size. DNA carries a negative charge on its phosphate backbone, so it migrates toward the positive electrode; the gel is a mesh, and small fragments thread through it faster than large ones. Fragment sizes are read against a ladder of known lengths.
• PCR copies a chosen stretch of DNA in a tube. Each cycle heats the sample to about 95 degrees Celsius to separate the strands, cools it so that short primers can bind at each end of the target, then warms it to about 72 degrees so polymerase can extend them. The polymerase, Taq, comes from a bacterium in hot springs and survives the near-boiling step, which is what made the process automatable. Kary Mullis received the 1993 Nobel Prize in Chemistry for it.
• Sequencing reads the order of bases directly. The Human Genome Project, using the chain-termination method, produced its finished sequence in 2003 after thirteen years; a comparable genome can now be sequenced in about a day.
• CRISPR-Cas9, adapted from a bacterial immune system, uses a guide RNA to direct a nuclease to a chosen sequence and cut it, allowing targeted edits. Emmanuelle Charpentier and Jennifer Doudna received the 2020 Nobel Prize in Chemistry for developing it as a tool.

The PCR arithmetic is worth doing, because it explains why the technique changed forensics. Each cycle doubles the number of copies of the target, so after n cycles one molecule has become 2ⁿ. After 20 cycles that is about a million copies; after 30 cycles it is 2³⁰, roughly 1.07 billion. Starting material invisible by any other method becomes enough to analyse. DNA profiling then compares short tandem repeats, stretches where a short sequence is repeated a variable number of times, and the combination across a dozen or more such loci is effectively individual.

In short: none of these tools was invented from nothing. Restriction enzymes, plasmids, polymerases and CRISPR were all found in bacteria, doing a job for the bacterium, and then borrowed.

What is genuinely contested

The technical questions are largely settled; the decisions are not. Editing somatic cells to treat a disease in one consenting patient raises different questions from editing an embryo, where the change is inherited by everyone descended from it and no one involved can consent. Genetically modified crops have measurably reduced insecticide use in some systems while concentrating seed supply in a small number of firms. Genetic testing produces information that insurers and employers would find valuable. A biology course can tell you accurately what the techniques do and cannot tell you which of these to permit; that is a decision about values, made by societies, and it goes better when the people making it understand the mechanism.

Common misconceptions

• "Lactose switches the operon on by binding the operator." It binds the repressor, not the DNA. The operator is a DNA sequence; the repressor is the protein that sits on it.
• "Different cell types have different genes." They have the same genome and express different parts of it. The exceptions, such as mature red blood cells that have ejected their nucleus and some immune cells that rearrange their DNA, are genuinely unusual.
• "Epigenetic changes alter the DNA sequence." They alter what can be read: methyl groups on DNA, chemical marks on histones, the packing of chromatin. The sequence itself is unchanged, which is why the effects are reversible.
• "PCR reads a DNA sequence." PCR copies a region. Sequencing reads it. They are often used together and are different operations.
• "CRISPR edits the whole organism." It edits the cells it reaches. Editing an adult means delivering it to enough of the right cells, which is the hard part of most gene therapy.

Summing up

Monod's stepped growth curve pointed at a cell deciding what to build. In the lac operon, a repressor blocks the operator until allolactose removes it, while CAP with cyclic AMP is required for strong transcription, so full expression needs lactose present and glucose absent. The trp operon inverts the logic, with tryptophan acting as a corepressor to switch off its own synthesis. Eukaryotes regulate at every level from chromatin packing to protein degradation, and marks such as DNA methylation and histone acetylation are heritable through cell division without changing the sequence, which is what a calico cat's patches display. Development works by cytoplasmic determinants, induction and morphogen gradients, and homeotic Hox genes, arranged in the order of the regions they control and conserved across the animal kingdom, assign structures to positions. The tools follow directly: restriction enzymes and plasmids for recombinant DNA, electrophoresis to separate fragments by size, PCR to turn one molecule into 2³⁰ copies in thirty cycles, sequencing to read them and CRISPR to edit them.

Sources

1. OpenStax. (2018). 16.2 Prokaryotic gene regulation. In Biology 2e. openstax.org
2. OpenStax. (2018). 17.1 Biotechnology. In Biology 2e. openstax.org
3. National Human Genome Research Institute. (n.d.). CRISPR. In Talking Glossary of Genomic and Genetic Terms. genome.gov
4. The Nobel Foundation. (1965). The Nobel Prize in Physiology or Medicine 1965: Francois Jacob, Andre Lwoff and Jacques Monod. nobelprize.org
5. The Nobel Foundation. (1995). The Nobel Prize in Physiology or Medicine 1995: Edward B. Lewis, Christiane Nusslein-Volhard and Eric F. Wieschaus. nobelprize.org

Twenty-five reindeer, and then eight

In 1911 twenty-five reindeer were released on St. Paul Island, a patch of tundra in the Bering Sea about forty square miles in area, to provide meat for the resident community. There were no wolves and no bears. The lichen the animals eat had been accumulating undisturbed for a very long time, and lichen grows back slowly.

The herd grew. By 1938 there were roughly two thousand animals. Then the lichen ran out. By 1950 the herd numbered eight.

Nothing about that story requires a predator, a disease or a hard winter. It requires only a population growing faster than its food supply can regenerate, on an island from which nothing can emigrate. This lesson is about the models that describe such curves, the arithmetic that makes them predictive, and what happens when populations are placed alongside each other in a community.

Measuring a population you cannot count

Population density is individuals per unit area or volume. Dispersion describes their arrangement: clumped, which is the commonest pattern because resources are patchy and many animals are social; uniform, which usually signals territoriality or chemical inhibition between plants; and random, which is rare and implies neither attraction nor repulsion.

You cannot count every field mouse in a meadow, so you sample. Mark and recapture assumes that the marked proportion of your second sample equals the marked proportion of the whole population:

N = (number marked initially x total in second sample) / number marked in second sample

Mark 60 mice and release them. A week later trap 80 mice, of which 15 carry marks. Then N = (60 x 80) / 15 = 320. The estimate depends on assumptions you should be able to state: marked animals mixed back in fully, marks did not fall off or make animals easier to catch, and no substantial birth, death or migration happened in between. Every one of those can fail, and each failure biases the estimate in a predictable direction.

A survivorship curve plots the fraction of a cohort still alive against age. Three shapes recur. Type I is flat then falls steeply: low mortality until old age, typical of large mammals with few offspring and long care, including humans. Type II is a straight diagonal on a logarithmic scale: a constant death rate at every age, seen in many birds and small mammals. Type III drops almost vertically at the start: enormous numbers of offspring, most dying young, as in oysters, most fish and most trees.

Exponential growth, and why it is always temporary

The per capita growth rate combines births and deaths:

r = (births - deaths) / N

A population of 500 with 60 births and 10 deaths in a year has r = (60 - 10) / 500 = 0.1 per year. The change in the population over that year is:

dN/dt = rN = 0.1 x 500 = 50 individuals per year

Notice what happens as N rises with r unchanged. At N = 1,000 the same r gives 100 individuals a year; at 10,000 it gives 1,000. The increment grows because the base grows, which is the definition of exponential growth and the reason it feels like nothing much for a long time and then like an emergency. A useful shortcut is the rule of 70: doubling time in years is roughly 70 divided by the percentage growth rate. At r = 0.1, which is 10 percent, the population doubles in about seven years.

Exponential growth appears in real populations, but only in particular circumstances: a species newly arrived in unexploited habitat, a bacterial culture in fresh medium, a population recovering after a crash. It cannot continue, because resources are finite.

The logistic model, worked with numbers

The logistic equation adds a brake. Carrying capacity, K, is the population size the environment can sustain, and the term that multiplies the growth rate shrinks toward zero as N approaches K:

dN/dt = rN (K - N) / K

Take r = 0.1 per year and K = 1,500, and evaluate the growth rate at several population sizes.

N	(K - N) / K	dN/dt
100	0.93	9.3 per year
500	0.67	33.3 per year
750	0.50	37.5 per year
1,200	0.20	24.0 per year
1,400	0.067	9.3 per year
1,500	0	0

The fastest growth is not at the smallest population nor at the largest, but at N = K/2, where there are enough individuals to reproduce and enough resources left to support the offspring. That is why maximum sustainable yield in a fishery is estimated near half the unfished stock size, and why fishing a population down to a tenth of K reduces not only the stock but its ability to recover.

Bottom line: exponential growth asks only how many individuals there are. Logistic growth asks how many there are relative to what the environment can support. The St. Paul reindeer overshot K badly, and because their overgrazing destroyed the lichen, K itself fell. Carrying capacity is not a constant.

What holds a population in check

Density-dependent factors act harder as the population gets denser: competition for food, water, nesting sites and light; predation, which concentrates where prey are common; parasites and disease, which spread faster in crowds; and the accumulation of waste. These are the factors that produce the logistic shape, because their strength is a function of N.

Density-independent factors act with the same force whatever the density: a hard frost, a hurricane, a fire, a volcanic eruption. They can devastate a population without any reference to how crowded it was.

Most real populations are held by both, and the balance shapes the life history a species evolves.

	r-selected	K-selected
Offspring	very many, small	few, large
Parental care	little or none	substantial and prolonged
Maturity	early	late
Lifespan	short	long
Population size	fluctuates far below K	near K, relatively stable
Survivorship	type III	type I
Examples	dandelions, mosquitoes, most fish	elephants, whales, oaks, humans

The human population reached eight billion in November 2022, according to the United Nations. Its annual growth rate peaked in the 1960s and fell below one percent around 2020, which means the absolute number is still rising while the rate of increase is falling. The mechanism behind that fall is the demographic transition: as economies industrialise, death rates drop first and birth rates follow later, so the population surges during the gap and then stabilises. Age structure diagrams make the future visible; a broad-based pyramid means a large cohort about to enter reproductive age and continued growth even if family sizes fall to replacement, while a column-shaped diagram means a stable or shrinking population.

Populations pressed against each other

A community is the set of populations living together, and their interactions can be catalogued by who gains and who loses.

Competition harms both. Georgii Gause grew two Paramecium species in 1934, separately and together. Separately, both thrived. Together, one consistently drove the other to extinction. The competitive exclusion principle that follows says two species cannot occupy exactly the same niche in the same place indefinitely; one will out-reproduce the other.

A niche is not a place but a role: everything a species does and requires, its food, its timing, its temperature range, its nesting requirements. The fundamental niche is the full range it could occupy; the realised niche is the part it actually occupies once competitors are present. Joseph Connell demonstrated the difference on a Scottish shore, showing that one barnacle species could live across a wide vertical range when its competitor was removed but was confined to the upper shore when the competitor was present. Coexistence often works through resource partitioning, in which species divide a resource by feeding at different heights, times or prey sizes.

Predation and herbivory benefit one and harm the other, and both drive obvious adaptations: speed, camouflage, thorns, toxins. Warning coloration advertises a genuine defence, and mimicry exploits it. A harmless species resembling a dangerous one is Batesian mimicry; two genuinely defended species converging on a shared warning pattern, so that predators need to learn only one lesson, is Mullerian mimicry.

Symbioses are close, long-term associations. Mutualism benefits both, as with mycorrhizal fungi trading soil minerals for plant sugars, or the gut bacteria that digest what you cannot. Commensalism benefits one and does nothing measurable to the other. Parasitism benefits one at the other's cost, and differs from predation in that the parasite usually does not kill its host outright.

The point: not all species matter equally. Robert Paine removed the sea star Pisaster from a stretch of rocky shore in 1966 and watched the number of species there collapse, because the sea star had been eating the mussels that otherwise crowd everything else out. A keystone species has an effect on its community far out of proportion to its abundance, and you generally discover which one it is by losing it.

Communities in time

Succession is directional change in community composition after a disturbance. Primary succession starts where there is no soil at all: a new lava flow, a retreating glacier's rubble. Lichens and mosses colonise bare rock, their acids and their dead bodies build the first thin soil, and larger plants follow over decades to centuries. Secondary succession starts where soil remains, after a fire or on an abandoned field, and is much faster because the seed bank and nutrients are already there. The old idea that succession always ends in a fixed climax community has been abandoned; disturbance is frequent enough in most systems that communities are usually a mosaic of stages.

Common misconceptions

• "Carrying capacity is a fixed property of a place." It depends on what the population does to the environment. Overgrazing on St. Paul lowered K, which is why the crash went so far below the peak.
• "Predators control prey numbers and that is the whole story." Food supply, disease, weather and competition act too, and predators are frequently limited by prey rather than the other way round.
• "A niche is where an organism lives." Where it lives is its habitat. Its niche is everything it does and needs, which is why two species can share a habitat without sharing a niche.
• "Exponential growth means fast growth." It means growth proportional to current size. A population growing at 0.5 percent a year is growing exponentially and slowly, and will still double in about 140 years.
• "Succession ends in a stable climax community." Disturbance is normal, and most landscapes are patchworks of different successional stages rather than a single endpoint.

What to carry forward

Twenty-five reindeer became two thousand and then eight, which is what unchecked growth against a slow-renewing resource looks like. Mark and recapture estimates a population as marked times total recaptured over marked recaptured, giving 320 in the worked case, and survivorship curves sort species into the three familiar shapes. Per capita growth rate is births minus deaths over N, and dN/dt = rN gives exponential growth whose increment rises with the base; the rule of 70 converts a percentage rate into a doubling time. The logistic model multiplies by (K - N)/K, which puts the fastest growth at half of carrying capacity, at 37.5 per year in the worked table, and zero at K. Density-dependent factors scale with crowding and produce that shape; density-independent ones do not. Life histories run from r-selected to K-selected, and the human population, at eight billion and growing at under one percent, is moving through the demographic transition. In communities, competitive exclusion, resource partitioning and the gap between fundamental and realised niches determine who lives where, keystone species hold assemblages together, and succession rebuilds communities after disturbance from bare rock or from remaining soil.

Sources

1. OpenStax. (2018). 45.1 Population demography. In Biology 2e. openstax.org
2. OpenStax. (2018). 45.3 Environmental limits to population growth. In Biology 2e. openstax.org
3. OpenStax. (2018). 45.6 Community ecology. In Biology 2e. openstax.org
4. United Nations, Department of Economic and Social Affairs, Population Division. (2024). World Population Prospects. population.un.org
5. Scheffer, V. B. (1951). The rise and fall of a reindeer herd. The Scientific Monthly, 73(6), 356-362.

Twenty thousand kilocalories, and what reached the top

Between 1953 and 1956 Howard Odum measured almost everything that could be measured about Silver Springs, a clear, constant-temperature spring-fed river in Florida. He wanted an energy budget for a whole ecosystem, expressed in the same units at every level. His published figures, in kilocalories per square metre per year, run like this: producers fixed 20,810; the herbivores that ate them held 3,368; the carnivores that ate those held 383; and the top carnivores held 21.

Look at the shape of that. Herbivores captured about 16 percent of what producers fixed. The next level captured about 11 percent of the herbivores. The top level, about 5.5 percent. Out of the twenty thousand kilocalories entering the system, twenty-one reached the animals at the top, which is one part in a thousand.

Everything in this lesson follows from where the rest went.

Production, gross and net

Producers do not keep all the energy they fix. Gross primary productivity is the total chemical energy captured by photosynthesis. Producers then respire, spending part of it to stay alive. What is left is net primary productivity, the energy actually available to build plant tissue and therefore to feed everything else:

NPP = GPP - respiration by producers

A grassland that fixes 12,000 kilojoules per square metre per year and respires 7,500 has an NPP of 4,500. That 4,500 is the figure that matters ecologically, because it is what a herbivore can eat. Productivity varies enormously by ecosystem: tropical rainforests, estuaries and coral reefs are among the most productive per unit area, open ocean and desert among the least, though the open ocean's vast area makes its total contribution large.

Worth holding on to: two ecosystems can fix the same amount of energy and support very different amounts of life, because respiration takes a different share. Gross is what came in; net is what is available.

Why so little gets through

At every transfer between trophic levels, most of the energy fails to arrive. Three things account for it.

• Not everything is eaten. Roots, bark, bones and whole individuals die without being consumed by the next level, and their energy goes to decomposers instead.
• Not everything eaten is absorbed. Cellulose, chitin and hair pass through, and the energy leaves in faeces.
• Most of what is absorbed is respired. A mammal spends the majority of its energy budget on movement, ion pumping and maintaining body temperature, and every one of those ends as heat. Heat cannot be recaptured to do biological work, so it leaves the ecosystem permanently.

The rule of thumb is that about 10 percent of the energy at one level ends up in the next. The Silver Springs numbers show it is a rough average rather than a law: 16 percent, then 11, then 5.5. Ectotherms do better than endotherms, because they spend nothing on heating themselves, which is why fish farming yields more protein per unit of feed than cattle.

Two consequences follow directly. Food chains are short, rarely more than four or five links, because after four transfers at 10 percent there is a ten-thousandth of the original energy left and it cannot support another level. And a given area of land feeds far more people if they eat plants than if they eat animals that ate the plants.

Ecological pyramids display this. A pyramid of energy is always widest at the base, because energy is lost at every step and cannot be otherwise. A pyramid of biomass can be inverted, and in open ocean plankton communities usually is: the phytoplankton at any instant weigh less than the zooplankton eating them, because they reproduce and are consumed so fast that a small standing crop supplies a large flow. The pyramid of energy measured over a year is still upright.

Matter, unlike energy, goes round

Energy enters as sunlight and leaves as heat. Matter has nowhere to go. The carbon atoms in your body were in the atmosphere, in a plant, in an animal, in rock, and will be again. This is the single most important asymmetry in ecosystem science, and it is why decomposers are not a footnote: without bacteria and fungi returning nutrients from dead tissue to the soil, production would stop within a few growing seasons.

The carbon cycle. Photosynthesis pulls carbon dioxide out of air and water; respiration by every organism, including producers, returns it. Carbon is stored for long periods in wood, soil, peat, limestone and the deep ocean. Combustion of fossil carbon transfers, in decades, carbon that took hundreds of millions of years to bury. The measurement record is unambiguous: continuous readings began at Mauna Loa in 1958 at about 315 parts per million and are now above 420, with an annual sawtooth from the growing season of the northern hemisphere's forests superimposed on the rise.

The nitrogen cycle. The atmosphere is 78 percent nitrogen gas, and almost nothing can use it, because the triple bond between the two atoms is extraordinarily strong. Nitrogen fixation breaks it: certain bacteria, some free-living and some in the root nodules of legumes, convert nitrogen gas to ammonia, and lightning fixes a smaller amount. Nitrifying bacteria then convert ammonia to nitrite and nitrate, the form most plants absorb. Decomposers return nitrogen from dead tissue as ammonia, and denitrifying bacteria complete the loop by returning nitrogen gas to the air. Since the Haber process was industrialised in the early twentieth century, humans have fixed nitrogen on a scale comparable to all natural processes combined, which is why nitrogen pollution is a global problem and not a local one.

The phosphorus cycle. Phosphorus has no significant gaseous phase. It weathers slowly out of rock, dissolves in water, is taken up by producers, passes through food webs, and settles into sediment that may not surface again for millions of years. Because the supply is slow, phosphorus is frequently the limiting nutrient in fresh water, and a sudden input has dramatic effects.

What happens when a cycle is overloaded

Nitrogen and phosphorus from fertiliser, sewage and manure run into rivers and lakes and lift the limiting nutrient. Algae bloom. The bloom shades out rooted plants, then dies, and the bacteria decomposing it consume dissolved oxygen. Fish and invertebrates suffocate. This is eutrophication, and its large-scale version is a seasonal dead zone: the northern Gulf of Mexico develops a hypoxic area every summer, fed by nutrients carried down the Mississippi from agricultural land far inland.

The clearest experimental evidence that vegetation controls nutrient loss came from Hubbard Brook in New Hampshire, where researchers gauged the streams draining several small forested watersheds, then cleared one entirely and prevented regrowth. Nitrate in the water leaving the cleared watershed rose sharply, exceeding public drinking water standards, while the untouched watersheds nearby were unchanged. Intact vegetation had been holding the nutrients in place, and the experiment measured how much.

Biomagnification is a different overload. A substance that is fat-soluble and not readily excreted, such as DDT or methylmercury, is retained rather than passed on. At each trophic level the consumer eats many individuals from the level below and keeps the contaminant from all of them, so concentration multiplies up the chain. The eggshell thinning that collapsed North American populations of ospreys, pelicans and bald eagles in the mid twentieth century is the classic case, and mercury advisories on large predatory fish are the current one.

Ocean acidification is chemistry, not weather. Carbon dioxide dissolving in seawater forms carbonic acid, and surface ocean pH has fallen by about 0.1 unit since the industrial era began. Because pH is logarithmic, that is roughly a 30 percent increase in hydrogen ion concentration. The consequence for organisms that build shells and skeletons from calcium carbonate, including corals, molluscs and many plankton, is that construction becomes energetically harder.

Biodiversity, and what actually reduces it

Extinction is normal; the current rate is not, and the causes are well characterised. Habitat loss and fragmentation lead, because a landscape divided into small patches supports smaller populations, more edge and less interior, and blocks the movement that lets populations recolonise. Invasive species follow, arriving without their predators and parasites. Overexploitation, pollution and climate change complete the list.

So what?: the argument for conservation is not only aesthetic. Ecosystems supply pollination, water purification, flood buffering, fisheries and the genetic library that medicine and agriculture draw on, and those services are expensive or impossible to replace.

It is also worth being precise about what has worked. Chlorofluorocarbons were destroying stratospheric ozone; the Montreal Protocol of 1987 phased them out, and the ozone layer has been recovering since. Lead was removed from petrol and blood lead levels fell. The pattern in each case was the same: an unambiguous measurement, an identified mechanism, and a substitute for the offending substance. Where those three exist, environmental problems have proven solvable at reasonable cost.

Common misconceptions

• "Energy is recycled through an ecosystem like matter." It is not. Energy enters as light and leaves as heat, and that one-way flow is why every ecosystem needs a continuous input.
• "Decomposers sit at the top of the food chain." They act on every level at once, receiving the large fraction of energy that is never eaten by the next consumer, and they return the matter that producers need.
• "The 10 percent rule is a law." It is an average. Silver Springs gave 16, 11 and 5.5 percent at successive steps, and ectotherm-dominated systems transfer more than endotherm-dominated ones.
• "A pyramid of biomass can never be inverted." In plankton communities it routinely is, because a fast-reproducing standing crop supports consumers that weigh more at any instant.
• "More carbon dioxide simply means more plant growth." Growth is limited by whatever is scarcest, usually water or nitrogen, and the warming, drought and ocean chemistry changes that accompany rising carbon dioxide act in the other direction.

Where this leaves us

Odum's spring turned 20,810 kilocalories per square metre per year into 21 at the top, and the losses along the way are the subject. Net primary productivity is gross production minus what producers respire, and it is what feeds everything else. Roughly a tenth of the energy at one level reaches the next, with the rest uneaten, unabsorbed or respired as heat, which is why food chains are short and why eating lower on one feeds more people. Energy flows through and leaves; matter cycles, which makes decomposers indispensable. Carbon moves between air, tissue, soil, rock and ocean, and the Mauna Loa record has tracked its atmospheric rise from about 315 parts per million in 1958 to over 420 today. Nitrogen must be fixed from an inert gas before life can use it, and industrial fixation now rivals the natural process. Phosphorus has no gas phase and therefore moves slowly, which is why adding it to fresh water causes eutrophication. Overloaded cycles produce dead zones, biomagnification of fat-soluble toxins and ocean acidification, while habitat loss leads the causes of biodiversity decline. Hubbard Brook showed by experiment how much nutrient retention intact vegetation performs, and the Montreal Protocol showed that a well-characterised problem with an available substitute can be solved.

Sources

1. OpenStax. (2018). 46.2 Energy flow through ecosystems. In Biology 2e. openstax.org
2. OpenStax. (2018). 46.3 Biogeochemical cycles. In Biology 2e. openstax.org
3. NOAA Global Monitoring Laboratory. (n.d.). Trends in atmospheric carbon dioxide. gml.noaa.gov
4. NOAA National Ocean Service. (n.d.). What is ocean acidification? oceanservice.noaa.gov
5. Odum, H. T. (1957). Trophic structure and productivity of Silver Springs, Florida. Ecological Monographs, 27(1), 55-112.