Introduction

You are, in a fairly literal sense, made of things that used to be somewhere else. The carbon in your muscle was in the air a few years ago. The calcium in your bones was in rock, then in soil, then in a plant, then in a cow, then in a glass. The iron carrying oxygen around your body right now was in the ground before it was in a lentil. Every atom of you crossed a boundary to get there, and almost all of it crossed through your mouth.

That boundary is the subject of this book. What crosses it, where it came from, what it does once it is inside, and what you actually need to know to make decent decisions about it.

Three kinds of thing cross that boundary

Food, which is mostly plants and animals, arrives as an enormously complicated mixture. An apple is not "sugar and fibre and vitamin C." It is several thousand distinct compounds, most of them present in tiny amounts, most of them never studied, some of them made by the tree specifically to discourage things from eating it. Your body takes that mixture apart, uses some of it, ignores a great deal of it, and hands a large fraction of the rest to thirty-odd trillion bacteria living in your colon who have their own opinions about what to do with it.

Medicine arrives as the opposite: one molecule, purified to within a fraction of a percent, at a dose worked out in clinical trials, in a tablet engineered to dissolve at a particular speed in a particular part of your gut. Medicine is precise in a way food never is. That precision is exactly why it works and exactly why it can hurt you.

Everything else arrives without asking. Caffeine and alcohol and nicotine, which are drugs we have agreed not to call drugs. The additives that keep bread soft and sausage pink. The compounds that did not exist in the raw ingredients and got created by the pan. Lead in old pipes, arsenic in rice paddies, mercury in big fish, cadmium in cocoa. Plastic and non-stick coating and whatever is in the air of a kitchen with the gas on and the window shut.

Most books handle one of these three and pretend the other two are somebody else's department. That is a mistake, because your liver does not know the difference. The same enzyme system that breaks down a painkiller also breaks down a compound in grapefruit, which is why the two together can be dangerous. The same transporter that absorbs iron from spinach also absorbs lead from a water pipe. The body has one set of doors, and everything comes through them.

What this book actually contains

The farming. This book starts in the soil, and it does not rush. If you want to understand why a supermarket apple is picked hard in September and sold to you in July, why the entire world's banana trade runs on a single sterile clone that cannot breed, why some crops need a cold winter and others die in one, or why "organic" does not mean "no pesticides," you need to understand how a farm works. So there are nine chapters on that first: what a plant is doing when it makes fruit, what soil is made of, how climate and water and chill hours decide what will grow where, what happens across a farming year, how varieties are bred and grafted, what gets sprayed and why, what the labels mean, and how produce survives the trip to you.

The produce, crop by crop. Then the aisle itself. Apples get a full chapter, because the reader who prompted this book asked for one and because the apple turns out to be a remarkably good teacher: it is a clone, a cold-storage marvel, a fibre delivery system, and a case study in why "an apple a day" has both more and less behind it than people think. Then bananas, citrus, berries, grapes, stone fruit, melons, mango and papaya, pineapple and kiwi, avocado, tomato, pear, pomegranate, figs and dates, the tropicals most people have never bought, olives and coconut, and the marketing category called superfruit. Then vegetables: the greens, the brassicas, the onion family, the roots, potatoes, the cassava and sweet potato staples that feed a billion people, squashes, chillies, legumes, mushrooms, seaweed, and the herbs and spices that are, chemically, the most concentrated plant material you will ever eat. Then grains, nuts, oils, dairy, meat, fish, fermented food, salt and sugar, and the honest chapter about ultra-processed food.

The digestion. Between the growing and the eating sits a section on what actually happens after you swallow. The journey of one bite from mouth to toilet with timings. Blood sugar and why the same gram of sugar behaves differently depending on what it arrived with. Protein, fat, fibre, and the microbiome. Every vitamin and every mineral, what each one does, what happens when you run short, and which ones people actually run short of. The plant chemicals with health claims attached, sorted into what is established and what is a press release. The natural toxins genuinely present in ordinary food. And whether the time of day you eat matters, which has a more interesting answer than either camp admits.

The medicine cabinet. Then drugs, on the same fixed template. What a drug even is: absorption, distribution, metabolism, excretion, half-life, receptors, dose-response. How a drug gets discovered, tested, approved, and physically manufactured, including what is actually in a tablet besides the drug. How to read a package insert. Then the drugs themselves: paracetamol, ibuprofen and the NSAID family, aspirin, opioids, cold and flu combinations, antihistamines, stomach drugs, antibiotics, antivirals, vaccines, blood pressure and cholesterol drugs, diabetes and weight drugs, antidepressants and the rest of the psychiatric cabinet, inhalers, hormones, and supplements. Then the situations that change all the rules: children, pregnancy, old age, interactions, and how to tell a real side effect from a coincidence.

The chemicals. Finally the things that are neither food nor medicine. Caffeine, which is the world's most used psychoactive drug by an enormous margin. Alcohol. Nicotine. Food additives and what the E numbers mean. What cooking creates that was not there before. Heavy metals and where they actually come from. Plastics, PFAS, packaging, and cookware. The air indoors. And a chapter on toxicology from scratch, because "the dose makes the poison" is the single most useful sentence in this entire subject and almost nobody applies it consistently.

What this book will not do

It will not tell you that any single food is magic. No fruit cures anything. The strongest, most repeatedly confirmed findings in nutrition are boring: eat more plants, eat more fibre, eat less ultra-processed food, drink less alcohol, do not smoke. Every exciting claim beyond that is weaker than its headline.

It will not pretend the evidence is stronger than it is. Nutrition science is hard. You cannot randomise ten thousand people to eat blueberries for thirty years. Most of what we know comes from observational studies, which can tell you that blueberry eaters are healthier and cannot tell you that blueberries made them so. When a claim rests on a mouse study or a test tube, this book says so.

It will not give you medical advice. Doses appear in these pages because you cannot explain a drug without them, and they are illustrative. Real dosing depends on your weight, your kidneys, your liver, your other medicines, your pregnancy status, and your age. Use this book to understand what your pharmacist and doctor are doing, and to ask better questions. Do not use it to overrule them.

It will not moralise about food. There is no clean and unclean here, no detox, no superfood, no foods that "flush toxins." Your liver and kidneys do that, continuously, and they do not need help from celery juice.

Who this is for

Anyone who has stood in a supermarket wondering whether the expensive tomatoes are different, whether frozen peas are worse than fresh, whether that apple has been in storage since last autumn, and what exactly the wax on the skin is. Anyone who has looked at a pill and wondered what the other 90 percent of the tablet is. Anyone who has been told to take ibuprofen with food and wanted to know why, or whether it is even true. Anyone who read a headline about a chemical in their pan and could not tell whether to panic.

No background is assumed. Not chemistry, not biology, not farming. Every term is defined the first time it appears, every mechanism is explained rather than named, and where a number matters it is given rather than gestured at.

Start with How to read this book, which explains the twelve-part template that most chapters follow and, more importantly, the safety rules.

👉 On to how to read this book.

How to Read This Book

A short chapter of ground rules, so the rest of the book reads faster.

The safety rules, first

This book explains drugs and doses in enough detail that it would be easy to mistake it for clinical guidance. It is not.

Never act on this book alone. Doses, thresholds, and drug choices in these pages are illustrative, chosen to explain a mechanism. Real prescribing depends on your kidney function, your liver, your weight, your other medicines, your pregnancy status, your age, and a dozen things a book cannot know. Use it to understand what your clinician is doing and to ask better questions. Do not use it to start, stop, or change a medicine.

Three things override everything else in these pages.

Emergencies are not reading material. Difficulty breathing, swelling of the lips or tongue, chest pain, one-sided weakness, sudden severe headache, confusion, a seizure, a child who is floppy or will not wake, or any suspected overdose means emergency services now. Chapter 101 exists to be read before you need it, not during.

Paracetamol overdose is the specific danger to know. It is the one common household medicine where the gap between a normal dose and a liver-destroying dose is small, where you can feel completely fine for a day or two while the damage happens, and where the antidote works brilliantly if given early and poorly if given late. It gets its own chapter (Chapter 65) partly for that reason. If you remember one medical fact from this book, make it that one.

Allergy is not dose-dependent in the way the rest of this book is. Most of what follows rests on "the dose makes the poison." A true allergy breaks that rule: a trace can kill someone who is sensitised. Chapter 98 handles it separately.

Most chapters follow a fixed twelve-part template

Two templates, actually. Both have twelve sections in a fixed order, so you can skip straight to the part you want.

Every produce chapter (fruits, vegetables, and the rest of the plate)

1. What it is: botanically and practically
2. Where it comes from: origin, domestication, and who grows it now
3. How it is grown: soil, climate, water, the plant's life cycle, harvest
4. What is inside it: composition and a nutrition table per 100 g
5. What it does in your body: the mechanism, not the marketing
6. What the evidence actually shows: with the strength of that evidence stated
7. Who should eat more, who should be careful: ages, conditions, allergies, drug interactions
8. When and how to eat it: time of day, pairing, peel or no peel, raw or cooked
9. Choosing, storing, and ripening
10. The varieties worth knowing
11. Myths and confusions
12. The bottom line

Every medicine chapter

1. What it is and what it treats
2. How it works inside you: the mechanism
3. How it is made: from starting chemical to the tablet in your hand
4. Dose, timing, and how to take it
5. What it actually does well: the evidence, honestly
6. Side effects: with real numbers where they exist
7. Who should be careful, and who should avoid it
8. Interactions: with food, drink, and other drugs
9. Overdose and warning signs
10. Brand names and formulations: so you can recognise the same drug in six boxes
11. Myths and confusions
12. The bottom line

The foundation chapters (farming, digestion, toxicology) and the closing practical chapters do not follow either template, because they are explaining machinery rather than cataloguing an item.

Every chapter ends the same way

After the bottom line, every chapter carries a "Sources and notes" section naming the specific trials, agencies, databases, and papers behind that chapter's claims, followed by an "Open questions" paragraph flagging what is genuinely unsettled, disputed, or likely to change.

That second part is deliberate. A book like this is most misleading where it sounds most confident, so each chapter is required to say out loud what it does not know. If you only read one thing in a chapter you are sceptical about, read its open questions.

The consolidated list of authorities, and where to go to check any of it yourself, is in sources and further reading.

Signs and boxes

In short: lines sit under most section headings and compress that section into one sentence, so you can skim a chapter and stop only where it matters.

Don't be confused: ... boxes untangle two things people routinely mix up. Paracetamol and ibuprofen. A food allergy and a food intolerance. Organic and pesticide-free. Antibiotics and antivirals. Mixing these up is not a small error, it changes what you think the thing is.

Bold marks a technical term the first time it is defined. Tables compare things that are genuinely comparable. A "👉" at the end of a chapter points to the next one.

How to read the evidence labels

Nutrition and medicine differ enormously in how well anything is known, and the book says which is which using four labels.

Where a chapter says something is thin, that is not an insult to the food. Blueberries are excellent. The claim that blueberries specifically improve memory is thin. Both things are true at once.

About the numbers

Nutrition figures are per 100 g of the raw edible portion unless stated, drawn from standard national food composition databases. They vary by variety, ripeness, soil, and season, sometimes by a factor of two. Treat them as the right order of magnitude, not as a measurement of the specific apple in your hand.

Production figures are approximate, mostly reflecting the early-2020s picture from the UN Food and Agriculture Organization. Rankings between adjacent countries shift year to year; the shape of the picture (China and India dominating most crops by volume, tropical exports concentrated in a handful of countries) is stable.

Drug doses are typical adult figures for the country where that product is sold over the counter, and the same drug is legally sold at different maximum doses in different countries. The book flags the big divergences.

Two unit systems. This book uses metric, with US customary in parentheses where it helps. Temperatures are Celsius with Fahrenheit alongside for anything you might do in a kitchen or a fridge.

Reading paths

If you came for the produce, read Chapter 1 through Chapter 9 for the farming background, then dip into whichever crop you buy. The crop chapters stand alone.

If you came for the medicine cabinet, read Chapter 62 and Chapter 64 first, then go straight to the drug you care about. Everything after that assumes those two.

If you came because a headline frightened you, read Chapter 87 and Chapter 97. They are the two chapters most likely to change how you read everything else.

If you want the short version of the whole book, read Chapter 100. It is the practical distillation, and it is deliberately boring.

👉 Now the ground floor: what a plant is, and why it bothers making fruit.

What a Plant Is, and Why It Makes Fruit

TL;DR. A plant is a machine for turning sunlight, air, and water into sugar, and then turning some of that sugar into everything else it needs. It does not make fruit for you. Fruit is a bribe: a package of sugar wrapped around seeds, evolved to make an animal eat it and carry the seeds somewhere else. Almost everything you like about fruit (sweetness, colour, smell, softness) is advertising aimed at an animal. Almost everything you dislike about vegetables (bitterness, astringency, pungency) is the plant's chemical defence aimed at an insect or a fungus. Understanding those two sentences explains an enormous amount about what is in your food and why.

Key takeaways

• Photosynthesis makes sugar from carbon dioxide and water using light. Everything else in a plant, and therefore almost everything else in your diet, is built from that sugar plus about fourteen mineral elements pulled from soil.
• A fruit is a seed dispersal device. Sweetness, colour, and softness appear only when the seeds are ready, which is why unripe fruit is sour, green, hard, and often mildly toxic.
• "Fruit" and "vegetable" are two different classifications that people use interchangeably. Botanically, tomatoes, cucumbers, peppers, aubergines, pumpkins, avocados, and green beans are fruits. Culinarily they are vegetables. Neither definition is wrong; they answer different questions.
• Plants cannot run away, so they defend themselves chemically. A large share of the interesting compounds in your diet, from the heat in chilli to the sulphur bite of garlic, are defensive.
• Roughly half of a plant's dry weight is carbon that came out of the air. Soil supplies minerals and water, not bulk.

The one reaction everything else rests on

In short: Plants combine carbon dioxide from the air with water from the soil, powered by sunlight, to make sugar and release oxygen.

The whole edifice of agriculture, and of your diet, sits on one chemical reaction:

$$6,\mathrm{CO_2} + 6,\mathrm{H_2O} + \text{light} ;\longrightarrow; \mathrm{C_6H_{12}O_6} + 6,\mathrm{O_2}$$

Six molecules of carbon dioxide, six of water, some light energy, and you get one molecule of glucose plus six of oxygen. That is photosynthesis, and it happens inside chloroplasts, small green compartments packed into leaf cells. The green comes from chlorophyll, a molecule that absorbs red and blue light strongly and reflects green, which is the only reason plants look the colour they do.

Two things about this reaction routinely surprise people.

The carbon comes from the air, not the soil. A 100-kilogram tree is not 100 kilograms of soil rearranged. Take a mature apple tree, dry it out, and roughly 45 percent of what remains is carbon, and essentially all of that carbon arrived as an invisible gas through microscopic pores in the leaves. Soil contributes water and about fourteen mineral elements, which together account for only a few percent of dry weight. The seventeenth-century experiment that showed this (growing a willow in a weighed pot of soil for five years and finding the soil had barely lost any mass) is still the cleanest demonstration.

The oxygen you are breathing is a waste product. Plants split water to get electrons, and oxygen is what is left over. For the first two billion years of life on Earth there was essentially no free oxygen in the atmosphere; photosynthetic organisms put it there, as pollution.

Don't be confused: plants respire too. Photosynthesis and respiration are not opposites performed by different kingdoms. Plants also burn sugar with oxygen to release energy, exactly as you do, all day and all night. In daylight, photosynthesis runs far faster than respiration, so the net flow is carbon in and oxygen out. At night only respiration runs. This matters practically: a picked fruit is still alive and still respiring, burning through its own sugar, which is why cooling produce extends its life. That is the entire basis of the cold chain in Chapter 8.

What a plant is built from

In short: Sugar becomes starch for storage, cellulose for structure, and, with added nitrogen and sulphur, protein.

Glucose is the raw feedstock. Everything else is made from it.

The dividing line between starch and cellulose deserves a moment, because it is the difference between a calorie and a non-calorie. Both are chains of the identical glucose molecule. In starch the links are α-1,4 bonds; in cellulose they are β-1,4 bonds. Your digestive enzymes can cut the first and cannot cut the second. That single geometric difference is why a potato feeds you and paper does not, and why "fibre" exists as a category at all. Cows can handle cellulose only because they outsource the job to microbes in their rumen. You do the same thing on a smaller scale in your colon, which is the subject of Chapter 14.

The parts of a plant, and which parts you eat

In short: Every plant part is eaten by someone somewhere, and which part it is predicts the nutrition surprisingly well.

That last row is why nuts and beans are calorically dense and lettuce is not. A seed is a survival capsule that has to fund an entire seedling until it can photosynthesise, so it is packed with concentrated reserves. A leaf is a working solar panel that needs to be thin, so it is 90 percent water. Nothing about this is a coincidence, and it means you can predict a food's rough nutritional shape from what part of the plant it is before you look anything up.

Why plants make fruit at all

In short: Fruit is a payment to animals for transporting seeds, and the plant only pays once the seeds are viable.

A plant that drops all its seeds directly beneath itself has a problem: its offspring compete with it for light, water, and minerals, and a disease or pest that finds the parent finds the whole family. Getting seeds away from the parent is worth paying for.

Some plants pay the wind (dandelion, maple, grass pollen). Some pay water (coconut). Some pay with hooks that catch on fur. And some pay animals directly, in sugar. That is fruit.

The economics are precise, and they explain every property fruit has:

Sugar is the fee. Sugar is metabolically expensive for a plant to make and give away, so it is not given away casually. It is loaded into the fruit late, in the final weeks of ripening, and often not even manufactured there: in apples and many other fruits, starch is stockpiled in the flesh over the summer and then converted to sugar in a rush at the end. That is why a starch test with iodine tells a grower when an apple is ready to pick.

Colour is the advertisement. Unripe fruit is green, which is camouflage against leaves. Ripe fruit turns red, orange, yellow, purple, or black, which are the colours that stand out against green foliage to an animal with colour vision. Primates and birds have colour vision. Most mammals do not, and, revealingly, fruits dispersed by bats tend to be dull-coloured and strongly scented instead.

Smell is the second advertisement. A ripening fruit releases dozens of volatile compounds, esters and aldehydes and terpenes, in a specific blend. The banana smell is largely isoamyl acetate; the fresh apple note is largely hexyl acetate. These are signals, broadcast on purpose, and they are the first thing lost when fruit is picked unripe and ripened artificially. It is the main reason a supermarket tomato smells like nothing.

Softness is the delivery mechanism. Ripening enzymes dissolve the pectin gluing cell walls together, so the flesh softens and the animal can eat it. This is a controlled, programmed process, not decay, though decay follows close behind.

And the seeds are protected. Small seeds are meant to be swallowed and survive the gut intact, sometimes germinating better for the trip. Large seeds are meant to be spat out or dropped. Many seeds are actively defended with bitter or toxic compounds: apple pips, cherry stones, apricot kernels, and peach pits all contain amygdalin, a compound that releases cyanide when crushed and digested. In normal eating this is irrelevant, and there are people who have poisoned themselves with apricot kernels sold as a health food. That is Chapter 18.

Unripe fruit is deliberately unpleasant. Before the seeds are viable, being eaten is a total loss for the plant. So unripe fruit is hard, sour with organic acids, low in sugar, green, and often astringent with tannins that bind to the proteins in your saliva and make your mouth feel dry. An unripe persimmon is the extreme case, and it is genuinely unpleasant in a way that feels almost hostile, because it is.

Don't be confused: ripening and rotting are different processes. Ripening is programmed, driven by the plant's own hormones and enzymes, and it improves the fruit. Rotting is external: fungi and bacteria digesting the fruit for themselves. Ripening makes rotting easier by softening tissue and raising sugar, so they run in sequence, but they are not the same thing and they are controlled by different levers. Cold slows both. Ethylene gas accelerates only the first.

Ethylene: the ripening hormone that runs the fruit trade

In short: A simple gas made by fruit itself triggers ripening, spreads between fruits, and is the single most important chemical in global produce logistics.

Ethylene ($\mathrm{C_2H_4}$) is about as simple as an organic molecule gets: two carbons, four hydrogens. Plants make it, and it acts as a hormone that triggers ripening, leaf drop, and flower senescence. Crucially it is a gas, so it diffuses from one fruit to the fruit next to it.

This splits all fruit into two categories, and the split governs how they are shipped, stored, and sold.

This is why bananas can cross an ocean. They are picked deep green, shipped at about 13 °C (55 °F) in an atmosphere carefully kept low in ethylene, and then, days before sale, moved into sealed ripening rooms and dosed with ethylene gas to trigger the change on a schedule. The banana in your kitchen was ripened by a person pressing a button. Avocados, tomatoes, mangoes, and pears go through variants of the same process.

It is also why the old advice works. Put an unripe avocado in a paper bag with a banana and it ripens faster, because the bag traps the banana's ethylene. Keep apples away from potatoes, because apple ethylene makes potatoes sprout. Keep bananas away from everything, because they emit a lot. And it is why a single rotting fruit really does spoil the bowl: damaged tissue floods out ethylene.

Non-climacteric fruit has no such trick. A strawberry picked white stays sour forever. This is why strawberries are expensive, why they arrive already fragile, and why the worst supermarket produce tends to be non-climacteric fruit shipped a long way.

Plants defend themselves chemically, and you eat the defences

In short: Bitter, pungent, and irritating flavours are usually anti-pest chemistry, and in small doses many of them are good for you.

A plant cannot flee, hide, or fight. Its options are physical armour (thorns, tough skins, silica in grass blades) and chemistry. Chemistry it does extremely well.

Two consequences follow, and they pull in opposite directions.

First, the defences are often good for you at food doses. A compound designed to be mildly toxic to an insect can, at the far lower relative dose a human gets, act as a mild stressor that triggers your own protective responses. That idea has a name, hormesis, and it is one of the better explanations for why vegetables are healthy beyond their vitamin content. The isothiocyanates from broccoli, for instance, activate a cellular pathway (Nrf2) that switches on your own antioxidant and detoxification enzymes. You are not getting antioxidants from broccoli so much as being provoked into making your own.

Second, the defences are still toxins, and dose still decides. Green potatoes contain solanine, and enough of it causes real poisoning. Bitter cassava kills people who process it carelessly, and has caused a permanent paralytic disease called konzo in famine conditions. Rhubarb leaves have enough oxalate to be dangerous. Nutmeg in large amounts is a genuine hallucinogen and a genuine poison. "Natural" is not a safety category, and Chapter 87 is dedicated to that point.

Third, and this one is uncomfortable: breeding has systematically stripped defences out, because bitter does not sell. Modern cucumbers have had cucurbitacins bred down, modern brussels sprouts had a specific bitter glucosinolate reduced in the 1990s (which is genuinely why they taste better than they did in the 1980s), and modern lettuce is far less bitter than wild lettuce. Some of what got removed was probably beneficial. This is a real trade-off in plant breeding, not a conspiracy, and it is discussed in Chapter 5.

Fruit or vegetable? Two different questions

In short: Botany classifies by plant part; cooking classifies by sweetness and where the food appears in a meal. Both are consistent internally, and they disagree constantly.

Botanically, a fruit is the mature ovary of a flower, containing the seeds. That is a structural definition and it does not care how the thing tastes.

By that definition, all of these are fruits: tomato, cucumber, courgette, pumpkin, aubergine, sweet and hot peppers, green beans, peas in the pod, okra, avocado, olive, and every grain of wheat and rice (a grain is a specialised dry fruit called a caryopsis). Meanwhile rhubarb is a stem, broccoli is a flower bud, and a strawberry is not a berry at all: the red flesh is a swollen flower base, and the actual fruits are the little pips on the outside. A banana, a tomato, a grape, an avocado, and a chilli, on the other hand, are berries in the strict botanical sense.

Culinarily, a fruit is sweet and eaten as such, and a vegetable is savoury and eaten as part of a main course. That is a use definition and it does not care about ovaries.

Neither is wrong. The US Supreme Court settled the practical version in 1893 in Nix v. Hedden, ruling that tomatoes were vegetables for tariff purposes because that is how people used them, while explicitly acknowledging the botany. That is the correct resolution: use the definition that answers your question.

This book organises produce the culinary way, because that is how you shop, and flags the botanical facts where they matter (mostly because they explain the plant's behaviour in the field).

What a plant needs from the outside world

In short: Light, carbon dioxide, water, a workable temperature, and about fourteen mineral elements. Everything in farming is about supplying those reliably.

Strip agriculture down and it is the business of supplying six things:

1. Light, which sets the ceiling on how much sugar can be made. This is why row spacing, pruning, and orchard shape are obsessions, and why greenhouses in northern winters run supplemental lighting.
2. Carbon dioxide, free in the air at roughly 420 parts per million and rising. Commercial greenhouses often enrich to 800 to 1000 ppm because it measurably raises yield.
3. Water, in large quantities. A single maize plant transpires a couple of hundred litres over a season. Most of the water a plant takes up is not consumed, it evaporates out of the leaves as a side effect of opening pores to let carbon dioxide in.
4. A temperature range the plant's enzymes work in, plus, for many temperate species, a cold period they cannot do without.
5. Mineral nutrients, dissolved in soil water: nitrogen, phosphorus, potassium in bulk, then calcium, magnesium, sulphur, and trace amounts of iron, manganese, zinc, copper, boron, molybdenum, chlorine, and nickel.
6. Freedom from being eaten, which is where pests, disease, and the entire pesticide question come in.

Chapters 2 through 4 take these in turn. The next one starts with the least glamorous and most important: what soil actually is.

The bottom line

• A plant is a sugar factory powered by light. Almost all of the mass of your food came out of the air.
• Fruit exists to be eaten. Its sweetness, colour, smell, and softness are signals aimed at animals, released only once the seeds are ready, which is why unripe fruit is actively unpleasant.
• Ethylene, a simple gas, controls ripening in about half of all fruits and is consequently the most commercially important molecule in the produce trade.
• Bitterness and pungency are chemical defences. At food doses many of them are good for you; at high doses several are frankly poisonous. Natural is not a safety category.
• Which part of the plant you are eating (root, leaf, seed, fruit) predicts its nutritional shape better than almost any other single fact.

Sources and notes

Photosynthesis, plant anatomy, and the starch/cellulose distinction follow standard plant physiology texts; Taiz and Zeiger, Plant Physiology and Development, is the usual reference. The seventeenth-century willow experiment is Jan Baptist van Helmont's, described posthumously in Ortus medicinae, 1648. Ethylene as a ripening hormone and the climacteric/non-climacteric split follow standard post-harvest physiology; Wills and Golding, Postharvest, covers it. Volatile compounds of banana and apple (isoamyl acetate, hexyl acetate) are from food chemistry literature summarised in McGee, On Food and Cooking. Amygdalin in Rosaceae seeds and the plant defence compound classes are covered in the EFSA opinion on cyanogenic glycosides (2019) and in standard plant secondary metabolite reviews. Hormesis and the Nrf2 pathway follow Mattson's and Talalay's work, discussed further in Chapter 17. Nix v. Hedden, 149 U.S. 304 (1893), is the tomato tariff case. Bitterness reduction in modern brussels sprouts is documented in Dutch breeding literature from the 1990s.

Open questions. How much of the health benefit of vegetables is attributable to hormetic stress responses rather than to fibre, micronutrients, or displacement of other foods is not settled, and probably cannot be resolved with current study designs. Whether the flavour and defence compounds bred out of modern cultivars mattered to health, and by how much, is an open and largely untested question.

👉 Next: what soil actually is, and why "dirt" is the wrong word for the most complicated ecosystem in agriculture.

Soil, From Rock to Root

TL;DR. Soil is not dirt. It is a four-part mixture of ground-up rock, decayed organic matter, water, and air, colonised by more living organisms per handful than there are humans on Earth. Its texture (the sand, silt, and clay ratio) is fixed and decides how it holds water. Its structure (how those particles clump) is fragile and decides whether roots and air can get in. Its pH decides which minerals dissolve, and therefore which nutrients a plant can actually reach even when they are present. Getting a crop is mostly a matter of getting those three right, because a plant with poor roots cannot be rescued by anything you spray on its leaves.

Key takeaways

• A teaspoon of healthy soil holds on the order of a billion bacteria plus metres of fungal filament. Soil fertility is largely a biological property, not a chemical one.
• Texture is permanent, structure is not. You cannot change sand into clay, but you can destroy good structure in one wet ploughing season and it takes years to rebuild.
• pH is the master variable. Iron is abundant in almost all soils and unavailable in alkaline ones. Most crops want pH 6.0 to 7.0.
• Nitrogen, phosphorus, and potassium are needed in bulk, which is why fertiliser bags carry three numbers, and why the invention of synthetic nitrogen fixation roughly doubled how many people the planet can feed.
• It takes centuries to build a few centimetres of topsoil and a single bad season to lose it. Erosion, not nutrient depletion, is the dominant long-term threat.

What soil is made of

In short: By volume, a good agricultural soil is about 45 percent mineral, 5 percent organic matter, and 50 percent pore space, half filled with water and half with air.

Take a spade of good topsoil and it separates into four components.

That last row is the one people forget. Roots need oxygen. A root is not a passive straw, it is living tissue burning sugar to actively pump minerals in against a concentration gradient, and that costs oxygen. Fill every pore with water and roots suffocate within days. This is why overwatering a houseplant kills it and looks identical to underwatering: in both cases the roots stop working. It is why flooded fields are a disaster for most crops, and why rice, which grows in standing water, needs a special anatomical trick (aerenchyma, internal air channels running from shoot to root) to survive it.

Texture: sand, silt, and clay

In short: Particle size determines everything about water and nutrient holding, and it is set by geology, not by the farmer.

Soil mineral particles are sorted by size into three classes. The sizes look arbitrary and are not: they mark the points where physical behaviour changes.

The difference is surface area, and it is not subtle. One gram of sand has a surface area of a few square centimetres. One gram of clay has a surface area of hundreds of square metres, because clay particles are not grains but stacked plates. Nutrients stick to surfaces, so clay holds nutrients and sand does not.

A loam is a balanced mixture, roughly 40 percent sand, 40 percent silt, 20 percent clay, and it is the ideal for most crops: enough clay to hold water and nutrients, enough sand to drain and let air in. When a wine region or an orchard district is described as having "good soil," loam or a sandy loam is usually what is meant.

You can determine texture yourself in about a minute. Take a walnut-sized lump of moist soil and work it in your palm:

• Falls apart, feels gritty, will not form a ball → sand or sandy loam
• Forms a ball, feels smooth and floury, ribbon breaks quickly → silt loam
• Forms a ball, feels sticky, can be squeezed into a ribbon several centimetres long before breaking, takes a polish when rubbed → clay

Texture cannot practically be changed. To turn a hectare of clay into a loam you would need to haul in thousands of tonnes of sand. Gardeners who add a bag of sand to clay usually make it worse, because a little sand in a lot of clay produces something closer to concrete. What you actually do with a difficult texture is work with it: choose crops that suit it, add organic matter, and manage drainage.

Structure: the part you can ruin

In short: Structure is how particles clump into crumbs, and it decides whether roots, water, and air can move. Good structure is built by biology and destroyed by machinery.

Texture is the ingredient list. Structure is how those ingredients are assembled, and it is where soil biology does its most visible work.

In a well-structured soil, particles are glued into aggregates: crumbs a few millimetres across, held together by fungal threads, bacterial secretions, root exudates, and the sticky organic compounds collectively called humus. Between the crumbs run large pores that carry air and drain excess water; within the crumbs, small pores hold water against gravity for roots to drink later. That dual pore system is the entire trick.

Good structure looks like a well-made crumble topping. Bad structure looks like either loose dust or a solid block.

How structure is destroyed:

• Compaction. Driving heavy machinery over wet soil crushes the pore space. A tractor pass on saturated ground can create a plough pan, a compacted layer 20 to 30 cm down that roots cannot penetrate. It can take years and deep-rooting cover crops to break.
• Excessive tillage. Ploughing breaks aggregates apart, exposes protected organic matter to rapid microbial burning, and leaves bare soil exposed to rain. This is the central argument for no-till farming.
• Bare soil and rain impact. A raindrop hits bare ground at roughly 9 metres per second. It shatters aggregates and the fine particles seal the surface into a cap, which then sheds water instead of absorbing it, which then causes erosion.
• Sodium. Sodium ions make clay particles repel each other and disperse, collapsing structure. This is the mechanism behind salinisation ruining irrigated land.

How structure is built: organic matter, living roots in the ground as much of the year as possible, fungal networks left undisturbed, earthworms, and staying off wet fields. All of these are slow.

Soil life: the part that is actually doing the work

In short: Soil is an ecosystem, and most of the nutrient supply to plants is mediated by organisms rather than by chemistry alone.

A single teaspoon of healthy topsoil contains on the order of a billion bacteria, several metres of fungal hyphae, thousands of protozoa, and dozens of nematodes. A hectare of good pasture soil can hold several tonnes of living organisms. This community does several jobs no chemical can replace.

Decomposition. Everything that dies is dismantled by soil organisms, releasing the nitrogen, phosphorus, and sulphur locked in it back into plant-available form. This is called mineralisation and it is temperature-dependent, which is one reason cold soils in early spring are effectively nutrient-poor even when they test rich.

Nitrogen fixation. Nitrogen gas makes up 78 percent of the atmosphere and no plant can use it, because the triple bond holding N₂ together is one of the strongest in chemistry. Certain bacteria can break it. The most important are rhizobia, which form nodules on the roots of legumes (peas, beans, lentils, clover, alfalfa, soy) in a genuine trade: the plant supplies sugar and an oxygen-free compartment, the bacteria supply ammonia. A good clover or alfalfa stand can fix on the order of 100 to 250 kg of nitrogen per hectare per year, free. This is the single most important fact in the history of crop rotation and it explains why legumes appear in every traditional rotation on Earth. It is also why Chapter 47 matters to farming as much as to dinner.

Mycorrhizal fungi. Roughly 80 percent of land plant species form partnerships with fungi that colonise their roots and extend a network of hyphae far beyond the root zone. Hyphae are much thinner than roots and reach pores roots cannot. They pass phosphorus, zinc, and water to the plant in exchange for sugar. Phosphorus in particular barely moves in soil (it diffuses a millimetre or two), so a plant relying on roots alone rapidly exhausts the phosphorus in its immediate neighbourhood. Mycorrhizal networks are the main reason plants get enough of it. Heavy tillage and high phosphate fertiliser both suppress these fungi, which is a genuine argument against over-fertilising.

Earthworms. Not universal (large parts of North America had no native earthworms after the last glaciation, and introduced ones are a mixed blessing in forests), but where present they are transformative. They drag organic matter down, create vertical channels that drain water and let roots follow, and their castings are richer in available nutrients than the surrounding soil. Worm counts are a reasonable field proxy for soil health: a spadeful of good soil in spring should hold ten or more.

Don't be confused: "feeding the plant" and "feeding the soil" are different strategies. Conventional fertiliser practice supplies plant-available nutrients directly, which works fast and reliably. Organic and regenerative practice supplies organic matter that soil organisms convert into plant-available nutrients, which works slowly, buffers better, and builds structure. Neither is magic. The first can produce excellent yields on degrading soil; the second can produce poor yields on excellent soil if the timing is wrong. Chapter 7 takes this apart properly.

The nutrients a plant needs, and where they come from

In short: Seventeen elements are essential. Three are needed in bulk, and those three are what fertiliser bags contain.

Plus carbon, hydrogen, and oxygen from air and water, which supply about 95 percent of dry weight.

Two diagnostic patterns are worth memorising because they generalise. Mobile nutrients (nitrogen, phosphorus, potassium, magnesium) can be pulled out of old leaves and moved to new growth, so deficiency shows in the old leaves first. Immobile nutrients (calcium, iron, boron, sulphur) cannot be moved once deposited, so deficiency shows in the new growth first. Which end of the plant is yellow tells you which half of the table to look in.

The three numbers on a fertiliser bag

A bag marked 10-10-10 is 10 percent nitrogen, 10 percent phosphate (P₂O₅), and 10 percent potash (K₂O) by weight. The last two are expressed as oxides for historical reasons and not because those compounds are in the bag. So a 20 kg bag of 10-10-10 carries 2 kg of actual nitrogen.

Nitrogen is the one that runs out fastest, because it is not held by soil particles (nitrate carries a negative charge, as do clay surfaces, so they repel) and it leaches away with rain. This is why nitrogen is applied every year and why nitrate pollution of groundwater and rivers is the characteristic environmental cost of intensive agriculture.

The industrial fix for nitrogen deserves a paragraph of its own. Before 1909, all the nitrogen entering agriculture came from legumes, lightning, manure, and mined Chilean nitrate. The Haber-Bosch process, which combines atmospheric nitrogen and hydrogen under roughly 200 atmospheres of pressure and 400 to 500 °C over an iron catalyst to make ammonia, removed that ceiling. It is now estimated that a substantial fraction of the nitrogen atoms in your body passed through a Haber-Bosch reactor, and that without synthetic nitrogen the planet could feed perhaps half its current population. It also consumes on the order of 1 to 2 percent of global energy and is a significant source of greenhouse emissions. It is simultaneously one of the most important inventions in human history and one of the largest environmental problems in agriculture, and any honest account of farming has to hold both.

pH: the master variable

In short: Soil pH decides which nutrients dissolve. A soil can be rich in iron and the plants can still starve of it.

pH measures acidity on a scale where 7 is neutral, lower is acidic, higher is alkaline, and each step is a tenfold change. Soils range from about 3.5 (peat bogs) to about 9.5 (arid alkaline soils).

pH matters because nutrient availability, not nutrient presence, is what a plant experiences. Each element dissolves best in a particular pH window:

This is why blueberries, which want pH 4.5 to 5.5, cannot simply be planted in ordinary garden soil and fed harder. At pH 7 they cannot take up iron no matter how much iron is present. It is why tea, rhododendrons, and potatoes prefer acid ground, and why brassicas prefer it limed (partly for nutrition and partly because the clubroot pathogen dislikes alkalinity).

Changing pH: adding ground limestone (calcium carbonate) raises pH, slowly, over months to years; the amount needed depends heavily on texture, because clay resists change far more than sand does. Elemental sulphur lowers pH, via bacteria that oxidise it to sulphuric acid, which means it works only when the soil is warm and biologically active. Both are slow. This is why matching crop to existing soil is usually smarter than fighting the soil.

Cation exchange capacity, in plain terms

In short: CEC is the soil's ability to hold positively charged nutrients against being washed away. Clay and organic matter provide it; sand does not.

Clay particles and humus carry negative charges on their surfaces. Positively charged nutrients (calcium²⁺, magnesium²⁺, potassium⁺, ammonium⁺, and also hydrogen⁺ and aluminium³⁺) stick to those surfaces, held loosely enough that roots can trade for them but tightly enough that rain does not wash them straight through. The size of that reservoir is the cation exchange capacity.

A pure sand has a CEC near 1 to 3. A clay loam rich in organic matter might be 20 to 30. Practically:

• Low CEC (sandy soil): feed little and often, because nutrients wash through. Irrigate little and often for the same reason.
• High CEC (clay, high organic matter): buffers well, holds a reserve, forgives irregular feeding, but is slow to correct if it is wrong.

Adding organic matter is the only realistic way to raise CEC on a sandy soil, which is another reason compost and cover cropping keep coming up.

Where soil comes from, and how fast it goes

In short: Soil forms at roughly a centimetre per century and erodes far faster than that under bad management. This is the real sustainability problem in agriculture.

Soil forms from parent rock through weathering: freeze-thaw cracking, chemical dissolution by slightly acidic rain, root pressure, and biological acids. The rate is glacial, on the order of 0.5 to 2 cm of topsoil per century in temperate conditions, and slower in cold or dry climates.

Erosion rates on tilled, bare, sloping ground routinely exceed that by an order of magnitude. The 1930s American Dust Bowl is the famous case (deep ploughing of prairie that had never been ploughed, followed by drought, followed by wind that removed the topsoil of an area the size of a small country), and it produced the modern soil conservation movement. Similar losses are ongoing and less visible in many places.

The practices that hold soil in place are the same short list every time: keep it covered, keep living roots in it, disturb it as little as possible, and diversify what grows in it. Those four principles are the core of what gets marketed as regenerative agriculture, and unlike most agricultural marketing, they rest on solid soil physics.

A soil test, decoded

If you ever have soil tested, here is what the report is telling you.

Two practical notes. Sampling dominates accuracy: fifteen to twenty cores mixed together across a field beats one careful sample from one spot, because soil varies more over ten metres than most people expect. And more is not better for phosphorus: surplus P runs off into water bodies and causes algal blooms, which is a leading cause of freshwater degradation worldwide.

The bottom line

• Soil is a living, four-part system: minerals, organic matter, water, and air, in roughly equal halves of solid and pore space. Roots need the air as much as the water.
• Texture (sand, silt, clay) is fixed by geology and decides water and nutrient holding. Structure (how particles clump) is biological, fragile, and the thing management actually controls.
• pH governs whether nutrients that are present are usable. Below 5.5 aluminium turns toxic; above 7.5 iron and zinc lock away. Most crops want 6.0 to 7.0.
• Nitrogen, phosphorus, and potassium are needed in bulk. Nitrogen is the one that leaches, needs replacing yearly, and comes overwhelmingly from the Haber-Bosch process or from legumes.
• Soil builds at about a centimetre a century and can be lost in a season. Cover it, keep roots in it, disturb it less, and diversify it.

Sources and notes

Soil texture classes, structure, cation exchange capacity, and nutrient availability by pH follow standard soil science texts; Brady and Weil, The Nature and Properties of Soils, is the standard reference. Rhizobial nitrogen fixation rates and mycorrhizal phosphorus transfer are from agronomy literature and FAO soils publications. Haber-Bosch history, its share of global energy use, and estimates that roughly half the nitrogen in human tissue passes through it follow Smil, Enriching the Earth, 2001, and subsequent nitrogen-cycle analyses. Soil formation and erosion rates are from FAO and USDA Natural Resources Conservation Service assessments; the Dust Bowl account follows Worster, Dust Bowl, 1979. Phosphorus runoff and freshwater eutrophication follow standard limnology and the FAO Status of the World's Soil Resources report.

Open questions. Estimates of soil carbon sequestration potential under changed management vary widely between studies and are genuinely contested, partly because measuring change at field scale is hard and partly because gains saturate and reverse. How much of soil fertility is biological rather than chemical, and therefore how much can be restored by management alone, remains argued.

👉 Next: climate, water, and the growing season, and why some fruit trees genuinely need a cold winter to work.

Climate, Water, and the Growing Season

TL;DR. Where a crop can grow is decided by four numbers: how cold it gets in winter, how much heat accumulates in summer, how long the frost-free window is, and how much water arrives and when. Temperate fruit trees have an additional and counterintuitive requirement: they need a specific amount of cold each winter, and without it they will not flower properly, which is the single biggest reason apples and cherries do not grow in the tropics. Almost every fact about where your food comes from traces back to these constraints, and climate change is currently moving all of them.

Key takeaways

• Chill hours are a hard requirement, not a preference. Most apple and cherry varieties need 800 to 1,200 hours below about 7 °C (45 °F) each winter or they break dormancy erratically and crop poorly.
• Growing degree days predict when a crop will be ready far better than the calendar does, and they are how commercial growers actually schedule harvest and spraying.
• Frost during flowering is the single most destructive weather event in fruit growing, and it lasts one night.
• About 70 percent of all freshwater withdrawal on Earth goes to agriculture. Irrigation is why California, Spain, Egypt, and the Indian Punjab produce what they do, and it is why several aquifers are being emptied.
• Latitude, altitude, and proximity to water are the three levers that let one crop grow in surprisingly different places. A high plateau near the equator can behave like a temperate climate.

Temperature: four different numbers

In short: Cold tolerance, heat accumulation, chilling requirement, and frost timing are separate constraints, and a crop must satisfy all four.

1. How cold it gets, at the worst moment

Every plant has a temperature below which it dies. For a lemon tree that is around -3 °C (27 °F). For an olive, around -10 °C (14 °F). For a hardy apple rootstock, below -35 °C (-31 °F). This is what hardiness zones measure: the USDA system divides the world into zones by average annual minimum temperature, each zone 5.6 °C (10 °F) wide. Zone 5 averages a coldest night of about -29 to -23 °C; zone 9 about -7 to -1 °C; zone 11 never freezes.

Hardiness zones are useful and routinely over-interpreted. They tell you about winter lows and nothing about summer heat, humidity, rainfall, or soil, so a plant can be "zone-appropriate" and still fail. Britain and the US Pacific Northwest share zones with places whose summers are ten degrees hotter, and the crops that thrive differ completely.

2. How much heat accumulates over the season

In short: Plants develop according to accumulated warmth, not elapsed days, which is why the same variety ripens six weeks apart in two regions.

A plant's rate of development depends on temperature, so growers count growing degree days (GDD) rather than calendar days. The arithmetic is simple:

$$\mathrm{GDD} = \sum_{\text{days}} \max!\left(0,; \frac{T_{\max} + T_{\min}}{2} - T_{\text{base}}\right)$$

The base temperature is the threshold below which that crop effectively stops developing: about 10 °C for maize, grapes, and most warm-season crops, and about 4 to 5 °C for many cool-season crops and fruit trees. So a day with a high of 24 °C and a low of 12 °C contributes $(24+12)/2 - 10 = 8$ growing degree days for maize.

Crops carry GDD requirements the way recipes carry cooking times:

Two consequences. First, a short-season variety is not a marketing term; it is a genuinely different plant, bred to complete its cycle in fewer accumulated degree days, usually at a yield cost. This is why northern regions grow different maize varieties than southern ones. Second, GDD explains why the same apple variety is picked in early August in one valley and late September a hundred kilometres away.

3. Chilling requirement: the cold that fruit trees cannot do without

In short: Temperate trees measure winter by accumulating hours in a specific cold band, and refuse to wake up until the count is met.

This is the fact that most surprises people, and it is why you cannot grow apples in Singapore.

Temperate deciduous trees enter dormancy in autumn. Dormancy is not simply a response to cold, it is an internally enforced shutdown with a built-in safety catch: the tree will not resume growth until it has accumulated a set number of chill hours, conventionally hours between roughly 0 and 7 °C (32 to 45 °F). The evolutionary logic is sound. A warm spell in December is not spring, and a tree that flowered in response to it would lose everything to the next frost. The chill requirement is a counter that says "winter has genuinely happened."

If chilling is not met, the consequences are specific and bad: bud break is delayed and scattered, flowering is spread over weeks so pollination fails, leaves emerge patchily, and yield collapses. The tree does not die. It just does not produce.

Plant breeding has produced low-chill cultivars of apple, peach, and blueberry specifically so subtropical regions can grow them, and this is a live and commercially important area. It is also why climate change is a direct threat to fruit growing in places like southern Spain, California's Central Valley, and parts of Australia and South Africa: warming winters mean chill accumulation is falling, and some regions are approaching the point where their existing orchards will no longer perform.

Don't be confused: vernalisation is a different thing with a similar flavour. Chilling requirement releases dormancy in woody perennials. Vernalisation is the cold requirement that lets certain annuals and biennials flower at all: winter wheat sown in autumn will not produce grain unless it experiences winter, and carrots, beets, and cabbages will not bolt to seed in their first year. Same theme, different mechanism, and it is why winter wheat and spring wheat are different products.

4. Frost, and the one night that ruins a year

In short: Open flowers and small fruitlets die at about -2 °C, so a single late spring frost can eliminate a crop that took a year to build.

Cold hardiness is not a fixed property of a tree; it changes dramatically through the year. A dormant apple tree in January shrugs off -25 °C. The same tree in full bloom in April is destroyed by -2 °C. The critical temperatures for apple are roughly:

The vulnerability window is short, a few weeks, and the loss is total: no flowers, no fruit, and no second chance until next year. This is why fruit-growing regions cluster near large bodies of water. Lake Michigan, Lake Ontario, the Great Lakes generally, the Rhine, the sea around Kent and Normandy: large water masses warm slowly in spring, which holds back bud break until the frost risk has passed, and they release heat in autumn, extending the season. Michigan's entire cherry industry exists because of the lake. So does Ontario's Niagara fruit belt.

Growers fight frost with a small arsenal of physics:

• Wind machines: large fans that mix the warmer air sitting 10 to 15 metres up back down to tree level. Works only against radiation frost on still, clear nights, which is the common kind. Useless against a cold air mass moving in.
• Overhead irrigation: counterintuitive but sound. Water releases 334 joules per gram as it freezes (latent heat of fusion), which holds the ice-coated bud at exactly 0 °C as long as you keep spraying. Stop early and you do more damage than you prevented.
• Orchard heaters and burning: expensive, polluting, still used in high-value crops.
• Site selection: cold air is dense and flows downhill, pooling in hollows. A slope is warmer at night than the flat ground below it. This is why vineyards and orchards sit on hillsides, and why the bottom of the valley grows pasture.
• Delayed pruning and evaporative cooling to push bloom later, which buys days.

Light and day length

In short: Some plants flower according to night length, which decides where they can be grown and when they can be forced.

Beyond raw energy for photosynthesis, day length acts as a calendar signal. Plants are classified by photoperiod response:

• Short-day plants (really long-night plants) flower when nights exceed a critical length: chrysanthemum, poinsettia, soybean, rice, some strawberries. Greenhouse growers force poinsettias for Christmas by covering them with blackout cloth.
• Long-day plants flower when nights are short: spinach, lettuce, onion, barley, potato tuberisation in some varieties. This is why spinach and lettuce bolt (rush to flower and turn bitter) in early summer, and why onion varieties are sold explicitly as short-day, intermediate-day, or long-day types, matched to latitude. Planting a long-day onion in Texas produces greens and no bulb.
• Day-neutral plants flower on age or size instead: tomato, cucumber, maize, and the "everbearing" strawberry varieties that made year-round supply possible.

At the equator day length barely varies, which is one reason equatorial agriculture runs on day-neutral and short-day crops and on continuous rather than seasonal cropping.

Water: the largest input by mass

In short: Agriculture uses about 70 percent of humanity's freshwater withdrawals, and the timing of water matters as much as the amount.

Plants use enormous quantities of water, and mostly not for chemistry. Photosynthesis consumes a trivial amount. The rest evaporates from the leaves through open stomata, the microscopic pores a plant must open to let carbon dioxide in. That is the fundamental trade-off of terrestrial plant life: to eat, you must open the door, and when the door is open you lose water.

The efficiency of that trade is measured as water use efficiency, and it varies enormously by photosynthetic type:

CAM plants solve the problem by opening their stomata at night, when it is cool and humid, storing the carbon dioxide as an acid, and using it the next day with the pores shut. That is why pineapple and dragon fruit thrive in places that would kill a lettuce, and it is a good illustration that "drought tolerant" is not vagueness, it is a specific biochemistry.

How much water crops actually need

Approximate seasonal requirements, expressed as depth of water over the field:

Those per-kilogram figures deserve care. A litres-per-kilogram number depends on whether you count rainfall that would have fallen anyway (green water) or only irrigation drawn from rivers and aquifers (blue water), and headlines almost always use the larger, less meaningful figure. Almonds are the famous example: the number is real, and the reason it is a political issue is not the number itself but that Californian almonds are irrigated from a stressed system, and that almond trees, unlike annual crops, cannot be fallowed in a drought year without killing a twenty-year investment.

Timing matters as much as total

Every crop has a critical period where water stress does disproportionate damage:

• Maize: silking and pollination. Drought during that two-week window can halve yield even if rain returns immediately after.
• Wheat: grain filling.
• Potato: tuber initiation and bulking.
• Grapes: deliberately stressed after fruit set, because mild water stress concentrates sugars and phenolics. This is regulated deficit irrigation and it is one of the few cases where growers withhold water on purpose.
• Apple: cell division in the first 40 days after bloom, which sets final fruit size.

And too much water at the wrong time is equally destructive: rain just before a cherry harvest causes the fruit to absorb water through the skin and split, which can destroy a crop in an afternoon. Cherry growers in some regions fly helicopters over orchards to blow the water off.

How irrigation is delivered

Drip irrigation delivers water slowly to the root zone, cutting evaporation and weed growth, and lets fertiliser be dissolved into the water (fertigation) for precise feeding. It is expensive to install and is the main reason high-value fruit in arid regions is economically possible at all.

The dark side is groundwater. The Ogallala Aquifer under the US High Plains, the North China Plain aquifer, and the aquifers under the Indian Punjab are all being drawn down faster than they recharge, in some places by metres a year. Substantial parts of current global food production are running on stored water that will not be replaced on any human timescale. And irrigating with even slightly salty water in a hot climate concentrates salt in the soil as the water evaporates, which is salinisation, the mechanism that ended agriculture in parts of ancient Mesopotamia and is currently degrading millions of hectares.

Humidity, wind, and the things nobody mentions

Humidity decides disease pressure more than almost anything else. Fungal spores need leaf wetness to germinate, so a humid climate means constant fungicide use or disease-resistant varieties. This is why grapes for wine are grown in dry-summer Mediterranean climates, why the Pacific Northwest is excellent for apples (dry summers, low disease pressure) and difficult for stone fruit, and why table grape production concentrates in arid regions with irrigation.

Wind desiccates, breaks branches, scars fruit skin against twigs (which is cosmetic damage that downgrades the crop), and disrupts bee flight, which cuts pollination. Windbreaks are standard in exposed orchards and can raise yields by double-digit percentages.

Hail is the other one-night catastrophe. High-value orchards in Europe and South America increasingly grow under permanent hail netting, which is expensive, changes the light spectrum reaching the fruit, and reduces sunburn as a side benefit.

Sunburn on fruit is a real and growing problem: apples and grapes on the exposed side of the canopy can be damaged above about 40 °C fruit surface temperature. Growers respond with overhead misting, shade netting, and kaolin clay sprays that leave a white reflective film.

The three levers: latitude, altitude, and water

In short: You can approximate a temperate climate near the equator by going up a mountain, which is why highland tropics grow coffee, apples, and strawberries.

Temperature falls with altitude at roughly 6.5 °C per 1,000 metres. That means 1,000 metres of elevation is worth something like 1,000 kilometres of latitude in temperature terms, though not in day length.

This is why:

• Coffee grows on tropical highlands (Ethiopian highlands, Colombian Andes, Kenyan slopes at 1,200 to 2,000 m). Arabica coffee wants 18 to 22 °C, which at those latitudes exists only up high. Lowland tropics grow robusta instead, which tolerates heat and tastes harsher.
• Kenya and Ethiopia grow apples and strawberries commercially at altitude.
• Ecuador and Colombia dominate cut flowers: equatorial day length is stable year-round and altitude supplies cool nights, so roses grow slowly with long, strong stems.
• Mexican and Peruvian highland valleys supply northern-hemisphere winter vegetables.

Proximity to large water bodies is the third lever, moderating both extremes. The Mediterranean basin, coastal California, coastal Chile, the Cape region of South Africa, and southwestern Australia share a Mediterranean climate: wet mild winters, hot dry summers. Those five regions, on five different continents, grow the same crops for the same reason, which is why olives, grapes, citrus, almonds, and stone fruit come overwhelmingly from that short list of places.

What climate change is doing to all of this

This is not a speculative section; the effects are already in the trade data.

• Chill hours are declining in warm fruit regions, threatening orchards that were planted for a climate that is going away. Low-chill breeding is racing this.
• Bud break is advancing faster than the last frost date is retreating in many temperate regions, so the frost risk window is getting worse, not better. Several recent European fruit crop failures follow exactly this pattern: a warm March, early bloom, an April frost.
• Growing zones are moving poleward and uphill, which sounds like a wash and is not, because soils, water, and infrastructure do not move with them.
• Water is being redistributed, with more precipitation falling in heavy events and less in gentle ones, which is worse for soil and for aquifer recharge.
• Crop nutritional quality is measurably shifting: elevated CO₂ raises yields of C3 crops somewhat while reducing their protein, zinc, and iron concentrations by a few percent, an effect confirmed in field experiments. More food, slightly thinner food.

The bottom line

• Four temperature constraints, not one: winter minimum survival, accumulated summer heat, winter chilling requirement, and frost timing at bloom. A crop must satisfy all four.
• Chill hours are why apples, cherries, and pears are temperate crops and cannot simply be moved to warm climates, and warming winters threaten existing orchards.
• One frost night at full bloom destroys a year's crop. That is why fruit regions sit beside lakes and on slopes.
• Water use is dominated by transpiration, not chemistry, and the timing of water stress matters as much as the total. Roughly 70 percent of human freshwater withdrawal goes to agriculture, some of it from aquifers being permanently drawn down.
• Altitude substitutes for latitude at about 1,000 m per 1,000 km, which is why the highland tropics grow temperate crops.

Sources and notes

Growing degree day models, chilling requirement figures, and critical frost temperatures by phenological stage follow extension service publications, particularly Washington State University and Michigan State University tree fruit programmes, which publish the bud-hardiness tables used here. Water use efficiency by photosynthetic type and crop water requirements follow FAO Irrigation and Drainage Paper 56 (Allen et al., 1998). Green and blue water accounting follows Hoekstra and Mekonnen's water footprint work. Aquifer depletion figures for the Ogallala, North China Plain, and Indian Punjab are from USGS and published groundwater assessments. Chill-hour decline under warming and its effect on fruit regions is documented in Luedeling's work on chill models. Photoperiod classes follow standard horticultural references. Elevated CO2 reducing crop protein, zinc, and iron is from Myers et al., Nature, 2014, and subsequent FACE experiments.

Open questions. Which chill model best predicts real bud break is unsettled, and different models disagree about how quickly warming will make particular regions unviable. Projections of where growing zones will move are more confident about temperature than about whether soil, water, and infrastructure will follow.

👉 Next: what a farm actually does across a year, from planting through pollination to harvest.

What a Farm Actually Does, Across a Year

TL;DR. A farm is a business that converts land, water, seed, fertiliser, fuel, and a great deal of timing into a perishable product with a fixed harvest date and a volatile price. An annual crop farm plants, feeds, protects, and harvests inside one season. An orchard is a twenty-year capital investment that yields nothing for the first three to five years and then depends absolutely on bees for a two-week window each spring. Understanding the calendar, the labour, and the economics explains most of what you see in a supermarket, including why some produce is cheap, why some is imported from the other hemisphere, and why perfectly good food is left in the field.

Key takeaways

• An orchard yields nothing for three to five years and is expected to run for 15 to 30. Every decision at planting is locked in for decades.
• Pollination is the tightest bottleneck in fruit growing. Many fruit trees cannot pollinate themselves and need a compatible second variety flowering at the same time, plus insects to carry the pollen.
• Most fruit trees set far more fruit than they can ripen and are deliberately thinned, by hand or chemically, to get saleable size and to prevent biennial bearing.
• Harvest labour is the largest single cost in most fruit and vegetable production, often 40 to 60 percent, and it is the constraint driving mechanisation and crop choice.
• Cosmetic grading standards, not spoilage, cause a large share of on-farm food loss.

Two completely different businesses

In short: Annual crops are a one-year bet, perennial crops are a twenty-year mortgage, and they behave nothing alike.

That last row is why almond growers in a Californian drought behave differently from tomato growers. A tomato grower plants less. An almond grower keeps buying water at any price, because letting the trees die means writing off twenty years.

The annual crop year

In short: Prepare, plant, feed, protect, harvest, and get the field ready again, with each step on a narrow window.

Take a field vegetable crop, say lettuce or carrots, in a temperate climate.

Winter: planning and soil preparation. Decide the crop and variety, based on last year's prices, contracts already signed with buyers, disease history of the field, and rotation requirements. Order seed. Take soil tests. Apply lime if pH needs correcting, because it takes months to act. On heavier soils, primary tillage may happen in autumn so frost can break up clods over winter.

Early spring: seedbed preparation. The soil must be worked to a fine, level tilth, because a small seed needs contact with moist soil particles to germinate. Too coarse and the seed sits in an air pocket and dies; too fine and the surface caps after rain. This is where compaction damage is done, because the temptation to get on the field when it is still wet is enormous and the cost is invisible until later.

Planting. Either direct seeding (a precision drill placing seeds at exact spacing and depth) or transplanting modules raised in a glasshouse. Transplants cost more and buy three to five weeks, which for a crop sold by earliness is often decisive.

Establishment. The most vulnerable period. Slugs, birds, damping-off fungi, and crusting can all take out a stand. Irrigation to get even germination.

Growth: feeding and protecting. Nitrogen usually goes on in split applications rather than all at once, because a single big dose leaches away and can burn roots. Weeds are controlled by herbicide, mechanical hoeing, hand weeding, or plastic mulch. Pests and diseases are monitored, ideally by scouting and thresholds rather than calendar spraying (see Chapter 6).

Harvest. For most vegetables this is a moving target rather than a date: lettuce is cut when heads reach size, and a field is walked repeatedly. For a crop like processing tomatoes or peas, harvest is one machine pass on one day, scheduled by the factory, and the grower's contract specifies quality thresholds that determine payment.

After harvest. Residue is incorporated or left as cover. A cover crop may be drilled immediately, because bare soil over winter loses nitrogen and topsoil.

Rotation, and why nobody plants the same thing twice

In short: Repeating a crop in the same field concentrates its specific pests and diseases until yields collapse.

Every crop has enemies specialised to it, and those enemies persist in soil. Grow brassicas repeatedly and clubroot builds up, and it survives in soil for up to twenty years. Grow potatoes repeatedly and potato cyst nematode accumulates. Grow wheat repeatedly and take-all fungus does the same.

A rotation breaks these cycles by leaving a gap during which the specialist pest has nothing to eat. It also balances nutrient demand, and, in the classic form, includes a legume that leaves nitrogen behind for the next crop.

The four-course Norfolk rotation of the eighteenth century (wheat, turnips, barley, clover) is the textbook example, and it was genuinely revolutionary: it eliminated the fallow year that medieval three-field systems required, so a third more land was in production, and the turnips and clover fed livestock through winter, which meant more manure, which meant higher yields again. Modern rotations are more varied but follow the same logic: alternate deep-rooted and shallow-rooted crops, alternate families, and put a legume in.

The orchard year

In short: A perennial fruit crop is built over five years, and each year runs through dormancy, bloom, fruit set, thinning, growth, harvest, and post-harvest recovery.

An apple orchard is the fullest example, and most other tree fruit is a variation on it.

Establishing the orchard (years 0 to 5)

Site selection comes first and cannot be undone: slope for cold air drainage, aspect for sun, soil depth and drainage, and water rights. Then soil preparation, because this is the only chance to fix subsoil problems before roots occupy the ground for twenty years.

Trees are planted as grafted whips: a chosen variety grafted onto a chosen rootstock (see Chapter 5). Modern high-density orchards plant 2,000 to 4,000 trees per hectare on dwarfing rootstocks, trained flat against wires in a tall spindle or fruiting wall system. This is a completely different object from the mental image of an orchard: it looks like a vineyard, is about 3 metres tall, and is designed so every fruit is within reach from the ground or a low platform and every leaf gets light.

Why the change? Dwarf trees crop in year two or three instead of year seven, produce far more fruit per hectare, ripen more evenly because light penetrates the whole canopy, and cut the cost and danger of ladder work. The trade-off is that dwarfing rootstocks have weak root systems and the trees must be permanently supported on a trellis, plus irrigated, because shallow roots cannot chase water.

Years 1 to 3 are structural: training branches to the wire, removing fruit so the tree invests in wood rather than crop, irrigating, and protecting from deer and voles. The orchard costs money every year and returns nothing. Establishment costs of a modern high-density apple orchard commonly run to tens of thousands of dollars or euros per hectare before the first commercial crop.

The bearing year, month by month

Dormancy (winter): pruning. The single biggest skilled labour job of the winter. The pruner decides, cut by cut, how much crop the tree will carry, how light will reach the fruit, and how big the fruit will be. Pruning removes 20 to 30 percent of the previous year's wood, opens the centre to light, and renews fruiting wood. It is genuinely a craft, it takes years to learn, and mechanical hedging is a compromise that trades quality for speed.

Bud break and bloom (spring): the most dangerous four weeks. Frost risk, as covered in Chapter 3. And pollination.

Pollination, the tightest bottleneck in fruit

In short: Most apple, pear, cherry, and almond varieties cannot fertilise themselves and need a different compatible variety flowering at the same moment, with insects moving pollen between them.

Many fruit trees are self-incompatible: a genetic system recognises pollen carrying the same S-alleles as the tree itself and shuts down the pollen tube before it reaches the ovule. This is an anti-inbreeding device and it is absolute in apples, most sweet cherries, most pears, most plums, and almonds.

Practically, this means:

• An orchard must contain at least two compatible varieties, usually planted in alternating rows, or dedicated crabapple pollinisers placed every few trees.
• Their bloom periods must overlap. Varieties are grouped into bloom-time classes, and a grower matching an early bloomer with a late one gets nothing.
• Some varieties are triploid (three sets of chromosomes instead of two) and produce sterile pollen, so they cannot pollinate anything and need two other varieties in the block to pollinate each other and it. Bramley's Seedling, Jonagold, and Mutsu are all triploid, which is a real orchard design constraint.
• Something must carry the pollen. Wind does not do it for apples; the pollen is heavy and sticky.

That last point is why commercial pollination is an industry. Growers rent honeybee hives at roughly one to three hives per hectare for apples, delivered by truck for the bloom period and removed afterwards. Californian almonds are the extreme case: around a million hectares of almonds bloom in February over a few weeks, requiring on the order of two million hives, which is a large share of all managed honeybee colonies in the United States, trucked across the country in the largest managed pollination event on Earth.

Honeybees are not even the best pollinators for apples. Mason bees (Osmia) and bumblebees work at lower temperatures, in poorer weather, and visit more flowers per minute, and several growers deliberately manage them. Bumblebees are essential in glasshouse tomatoes, which need buzz pollination: the flower only releases pollen when vibrated at a particular frequency, which honeybees cannot do and bumblebees can. Before commercial bumblebee rearing in the late 1980s, glasshouse tomatoes were pollinated by workers with vibrating wands, one flower at a time.

Weather during bloom decides the crop. Bees do not fly below about 12 to 13 °C, do not fly in rain, and do not fly in strong wind. A cold, wet bloom week means poor fruit set no matter how many hives are present. A grower can do everything right for eleven months and lose the year to five bad days.

Fruit set and the June drop. After fertilisation, the tree assesses what it can support and aborts the rest. The natural shedding of excess fruitlets in early summer is called the June drop, and it is not a problem.

Thinning: deliberately throwing fruit away. Even after the June drop, most apple trees carry far more fruit than they should. If left alone, the result is a large number of small, poorly coloured, low-sugar apples, and a tree so exhausted it produces almost nothing the following year. That alternating pattern is biennial bearing, and it is one of the oldest problems in fruit growing.

So growers thin, aiming for roughly one fruit per cluster and a specific spacing along the branch. It is done chemically (with growth regulators or lime sulphur applied during bloom, which is cheap and imprecise) and then corrected by hand (expensive and precise). Removing 60 to 80 percent of the fruitlets to get a good crop feels wrong and is correct.

Summer: growth, irrigation, protection. Cell division in an apple finishes about 40 days after bloom, and everything after that is cell expansion, which is why early-season water stress permanently limits fruit size. Summer pruning opens the canopy so light reaches the fruit surface, because red colour in apples develops only in sunlight and is a major grading criterion. Netting goes up against hail and birds. Sprays continue against scab, mildew, and codling moth.

Harvest. Picked by hand, one fruit at a time, into a canvas bag, emptied into a bin holding 300 to 400 kg. A skilled picker moves one to two tonnes a day. Maturity is assessed by starch-iodine test (staining a cut fruit with iodine, which turns starch black and shows how far the starch-to-sugar conversion has progressed), by firmness with a penetrometer, by sugar with a refractometer, and by seed colour. Different destinations get picked at different maturities: fruit for immediate sale is picked riper, fruit for nine months of controlled-atmosphere storage is picked distinctly firmer.

After harvest. Post-harvest nitrogen, so the tree builds reserves for next spring's flowering, since flower buds for next year are already forming this summer. Leaf fall, sanitation to reduce overwintering disease, and back to pruning.

Labour: the number that decides everything

In short: Hand harvest is the dominant cost in fruit and vegetables, and its availability decides what gets grown and whether it gets picked.

For grain, labour is a rounding error: one operator with a combine harvests a hectare in under an hour. For fruit and vegetables it dominates.

Two consequences run through the entire produce economy.

Mechanisation is bred, not just engineered. Processing tomatoes are machine harvested because in the 1960s breeders at UC Davis produced a tomato with uniform ripening, a tough skin, and a compact determinate plant, developed jointly with the harvester. The machine and the plant were designed for each other. The same has happened with dwarf orchard architecture, with once-over harvested peas and beans, and increasingly with robotic apple picking, which is still not solved at commercial scale because identifying, reaching, and detaching a fruit without bruising it is genuinely hard.

Labour supply decides what is planted. Fruit and vegetable production worldwide leans heavily on seasonal and migrant labour. When that supply is restricted, growers switch to mechanisable crops, move production abroad, or leave fruit unpicked. Crops left in the field for lack of pickers are a regular news story in the UK, US, and Australia, and it is an economic decision, not carelessness: if the price does not cover the picking cost, picking loses money.

The economics, which explain the supermarket

In short: Growers are price takers with fixed harvest dates and perishable goods, which is the weakest possible bargaining position.

A grower faces:

• Fixed costs regardless of yield: land, trees, irrigation infrastructure, machinery.
• A harvest date they do not control, set by the crop.
• A product that loses value by the day.
• Buyers who are far larger than they are. In many countries a handful of supermarket chains control most of the retail market.
• Prices set by aggregate supply. A perfect growing season everywhere means a glut and a price crash. A grower can have their best-ever yield and their worst-ever income in the same year.

Which produces behaviours that look irrational from outside:

Cosmetic grading. Buyers specify size, colour, and shape tolerances. Fruit outside spec is downgraded to juice or processing at a fraction of the price, or not picked at all. A large share of on-farm loss is fruit that is perfectly edible and the wrong shape. The "wonky veg" retail lines of recent years are a partial correction to this.

Contracts and dumping. Processing crops are grown on contract at an agreed price. If the factory rejects a load on a quality specification, the grower may have nowhere else to send it that day.

Counter-seasonal trade. Northern-hemisphere supermarkets want apples, grapes, blueberries, and asparagus year-round. Southern-hemisphere suppliers (Chile, New Zealand, South Africa, Peru, Argentina) supply the northern winter. This is why an apple in a British supermarket in June may be from New Zealand and an apple in December may be from British controlled-atmosphere storage since September. Both are perfectly good; they got there differently.

Vertical integration in bananas and a few other crops, where the same company owns plantations, ships, and ripening rooms, because the logistics are so tight that coordination beats markets.

Greenhouses and the controlled extreme

In short: Protected cultivation trades capital and energy for control, and produces most of the tomatoes, cucumbers, and peppers sold in northern Europe.

At the far end of intensity is protected cultivation. A modern Dutch glasshouse is closer to a factory than a field: hydroponic growing in rockwool or coir rather than soil, computer-controlled nutrient solution, CO₂ enrichment to 800 to 1,000 ppm (often captured from a combined heat and power unit that also supplies heat and electricity), supplementary LED lighting, bumblebee hives for pollination, and biological pest control using predatory mites and parasitic wasps rather than sprays.

The results are startling. Field tomatoes yield perhaps 50 to 100 tonnes per hectare per year. Dutch glasshouse tomatoes exceed 500 tonnes per hectare, with a fraction of the water per kilogram, because water is recirculated rather than lost to soil and evaporation. The Netherlands, a small and cold country, is by value one of the largest agricultural exporters on Earth almost entirely because of this.

The costs are energy and capital. A heated glasshouse in a northern winter is energy-intensive, and the carbon footprint of a locally grown heated winter tomato can exceed that of one trucked or shipped from Spain or Morocco. "Local" and "low carbon" are not the same variable, which is a recurring theme in Chapter 9.

Scale, from smallholding to agribusiness

Worldwide, most farms are small. Estimates put roughly 500 million farms at under two hectares, and those farms produce a large share of the food eaten in low- and middle-income countries. At the same time, a small number of very large operations produce most of what is internationally traded.

Both realities matter for this book. The mango in a European supermarket probably came from an export-oriented operation with cold chain and certification. The mango eaten in the country where it grew probably came from a few trees. Different systems, different economics, same fruit.

The bottom line

• Annual crops are a one-year bet; perennial crops are a multi-decade commitment where the variety and rootstock choice at planting is locked in for twenty years.
• Pollination is the tightest bottleneck in tree fruit: incompatible varieties, mismatched bloom times, or five days of cold rain during bloom can each cost the whole crop.
• Trees set far more fruit than they can finish, so growers deliberately remove most of it to get size, colour, sugar, and a crop again next year.
• Hand labour, not land or inputs, is the dominant cost in fruit and vegetables, and it decides which crops are viable and whether a crop is picked at all.
• A large share of on-farm loss is cosmetic rejection, not spoilage.

Sources and notes

Orchard systems, rootstock-driven planting densities, and establishment costs follow extension publications from Cornell, Washington State, and East Malling. Self-incompatibility and triploidy in apples, and the bloom-overlap requirement, follow standard pomology references. Californian almond pollination hive numbers are from USDA and Project Apis m. reporting. Buzz pollination and the commercial bumblebee industry follow Velthuis and van Doorn's history of bumblebee rearing, 2006. Labour requirements per hectare are drawn from national farm business surveys and extension enterprise budgets. The processing tomato mechanisation history follows the UC Davis breeding and harvester development record. Rotation principles and clubroot persistence follow standard agronomy. Dutch glasshouse yields and the Netherlands' agricultural export position are from Wageningen University reporting and FAO trade data. Smallholder share of global farms is from FAO and Lowder et al.'s farm size analyses.

Open questions. How much on-farm loss is genuinely attributable to cosmetic grading rather than to logistics and price is poorly measured and estimates vary widely. Whether robotic harvesting will reach commercial viability for soft fruit within the next decade is genuinely uncertain.

👉 Next: why your apple is a clone, and how plant breeding actually works.

Why Your Apple Is a Clone

TL;DR. Plant an apple pip and you get a tree that bears nothing like the apple it came from, because apples are wildly variable from seed. So every Gala apple on Earth is a cutting from one 1930s New Zealand seedling, grafted onto a rootstock that controls the tree's size, grown as a genetically identical clone. The same is true of every banana of the Cavendish type, every Hass avocado, and most named fruit varieties. Breeding a new one takes fifteen to twenty-five years. Understanding grafting and breeding explains why fruit varieties are trademarked, why the banana trade is one disease away from disaster, and what "GMO" does and does not mean.

Key takeaways

• Apple seedlings are extremely heterozygous: seeds from a Golden Delicious produce trees that are all different and mostly bad. Every named variety is therefore a clone.
• Grafting joins two plants: a scion that decides the fruit, and a rootstock that decides tree size, precocity, disease tolerance, and soil tolerance.
• Dwarfing rootstocks changed fruit growing more than any variety did, allowing dense orchards that crop in year two instead of year seven.
• Monoculture clonal crops are genetically identical and therefore share every vulnerability, which is exactly how Panama disease destroyed the Gros Michel banana and is currently threatening the Cavendish.
• A genetically modified crop and a conventionally bred crop differ in how the change was made, not in whether the plant was changed. Conventional breeding includes deliberate radiation mutagenesis, which is unregulated.

Why you cannot grow an apple from a pip

In short: Apples carry huge genetic variation and are self-incompatible, so every seed is a genuinely new and usually disappointing plant.

Plant an apple seed and you will get an apple tree. You will not get the apple you ate.

Two mechanisms combine to produce this. First, apples are self-incompatible, so every seed already has two different parents. Second, apples are unusually heterozygous: each individual carries two quite different versions of most genes, so its offspring reshuffle into an enormous range of outcomes. The traits that make a good dessert apple (sweetness balanced with acid, crisp texture, good size, keeping quality, attractive colour) are controlled by many genes at once, and the odds of reassembling a good combination by chance are poor.

Historically, orchardists planted thousands of seedlings and found perhaps one worth keeping. American folk history has John Chapman ("Johnny Appleseed") planting seedling orchards across the Midwest in the early 1800s, and the honest version of the story is that most of that fruit was small, sour, and destined for cider, which is what people actually wanted at the time.

So essentially every apple you have eaten descends by cutting, not by seed, from a single original tree:

Every Granny Smith in the world is a piece of one tree that grew in New South Wales in the 1860s. It is a plant that has been kept alive by continuous cutting for over 150 years.

Grafting: two plants, one tree

In short: A cutting of the desired variety is joined to the root system of a different plant, and the join heals into a single functioning tree with two genomes.

Grafting takes a piece of the variety you want (the scion, a dormant twig with a few buds) and joins it to a rooted plant (the rootstock). Where the cut surfaces meet, the thin layer of dividing cells just under the bark (the cambium) on each side must line up. If it does, the two cambiums produce callus, the callus differentiates into connecting vascular tissue, and water and sugar begin flowing across the join. Within a season it is one tree.

The scion determines the fruit entirely. The rootstock determines almost everything else.

The common techniques:

• Whip-and-tongue graft: matching sloping cuts with interlocking tongues, bound and sealed. The standard for dormant bench grafting of fruit trees.
• Cleft graft: split a cut stump, insert wedge-shaped scions at the edges. Used for topworking, converting an established tree to a new variety, which lets a grower change variety in two or three years instead of replanting and waiting eight.
• Budding (chip or T-bud): insert a single bud under the bark of the rootstock in late summer. Fast, high success rate, and the standard in commercial nurseries and in citrus and roses.
• Approach graft and inarching: joining two plants while both remain on their own roots, used for repairs.

Grafting only works between related plants. Within a species, always. Within a genus, usually (any apple onto any apple, any citrus onto most citrus). Between genera in the same family, sometimes (pear onto quince, which is standard practice and gives dwarfing; almond onto peach). Between families, essentially never. There is no grafting an apple onto an oak.

Don't be confused: grafting is not genetic modification and does not mix the two plants' genes. The scion's cells stay the scion's cells. Fruit from a Bramley scion on an M9 rootstock is entirely Bramley. The rootstock changes the supply of water, minerals, and hormones, which changes vigour and timing, not the genetics of the fruit.

What rootstocks actually control

This is the part that changed fruit growing more than any variety ever did.

The Malling series of apple rootstocks, developed at East Malling Research Station in Kent from 1912 onwards, sorted existing rootstocks into a numbered series by vigour and made tree size a design choice.

Beyond size, rootstocks confer:

• Precocity: how young the tree starts fruiting. Dwarfing stocks crop years earlier, which transforms the cash flow of an orchard.
• Disease resistance: the Geneva series (G.41, G.935 and relatives) from Cornell provides resistance to fire blight and to apple replant disease, which is why they are displacing M9 in new plantings.
• Soil tolerance: heavy soil, drought, high pH, waterlogging.
• Cold hardiness of the root system, which is often the limiting factor further north than the top of the tree is.

Citrus tells the same story: nearly all commercial citrus is budded onto rootstocks chosen for tolerance of Phytophthora root rot, of tristeza virus, of salinity, and of calcareous soil. The famous case is that in the 1800s sour orange was the standard rootstock, tristeza virus arrived, and millions of trees died worldwide because they all shared a rootstock vulnerability.

Grapes tell the most dramatic version. In the 1860s the root-feeding aphid phylloxera arrived in Europe from North America and destroyed most of the vineyards of France within about two decades. The solution, arrived at after enormous argument, was to graft European Vitis vinifera scions onto American Vitis rootstocks, which had co-evolved with the pest and tolerate it. Essentially every wine grape in the world today grows on American roots, and has done since about 1900.

How a new variety is actually bred

In short: Cross two parents, grow tens of thousands of seedlings, and spend fifteen to twenty-five years discarding almost all of them.

Classical fruit breeding is a long game with brutal arithmetic.

Step 1: choose parents and cross them. Emasculate the flower (remove anthers before they shed pollen), apply pollen from the chosen father by hand, and bag the flower to exclude other pollen. A breeding programme might make hundreds of crosses in a season.

Step 2: grow the seedlings. Perhaps 10,000 to 100,000 seedlings from a season of crosses. Each is genetically unique.

Step 3: wait. Apple seedlings on their own roots take four to eight years to fruit. Some programmes graft seedlings onto dwarfing rootstocks to force earlier fruiting and save years.

Step 4: cull ruthlessly. First selection is on fruit appearance and taste, and typically 95 to 99 percent are eliminated immediately for being small, dull, mealy, diseased, or simply unremarkable.

Step 5: propagate and trial the survivors. The few hundred remaining are cloned and planted in replicated trials across several sites and several years to test yield, storage, disease resistance, and consistency.

Step 6: commercial trial and release. Maybe one to five varieties out of the original tens of thousands reach market.

Total elapsed time: 15 to 25 years for apples, similar for most tree fruit, faster for strawberries and annuals. Honeycrisp was crossed in 1960 at the University of Minnesota, nearly discarded, and released in 1991. Cosmic Crisp was crossed in 1997 at Washington State University and released commercially in 2019, after an investment reported in the tens of millions of dollars.

That investment explains club varieties. Older varieties are freely propagatable; a modern variety is often patented and trademarked, and the right to plant it is licensed to a limited group of growers who must follow quality rules and pay royalties. Pink Lady is the trademark under which fruit of the Cripps Pink variety meeting a quality standard is sold. Jazz, Envy, and SweeTango work similarly. The commercial argument is that restricting supply protects quality and price and funds the next variety; the criticism is that it concentrates control of the crop. Both are true.

Faster tools

Marker-assisted selection tests a seedling's DNA for markers linked to traits (scab resistance, fruit acidity, flesh firmness) in the first weeks of life rather than waiting years for fruit, letting a breeder discard 90 percent of the field before planting it out. This is now standard in serious fruit-breeding programmes and it is not genetic modification; it is just testing before planting.

Mutation breeding, which surprises people, involves exposing seeds or buds to gamma radiation or chemical mutagens to generate random mutations, then screening for useful ones. It has been used since the 1950s, thousands of released crop varieties trace to it, and it is entirely unregulated as a breeding method in essentially every jurisdiction. Red grapefruit varieties, some barleys used in brewing, and various rice and wheat lines came this way.

Sports. A sport is a spontaneous mutation in a single branch of an existing tree, producing, say, redder fruit or earlier ripening. An observant grower spots it, cuts that branch, grafts it, and has a new variety in three years instead of twenty. A large share of commercial apple "varieties" are sports of a smaller number of genetic originals: there are dozens of redder, striped, or earlier Gala sports, all genetically Gala.

Hybrids, F1 seed, and why you cannot save the seed

In short: F1 hybrid seed is produced by crossing two inbred parent lines each generation, gives uniform vigorous plants, and does not breed true.

For annual vegetables, the dominant technology is different.

Breeders create two inbred lines by self-pollinating for many generations until each is genetically uniform. Inbred lines are typically weak. Cross them and the offspring show hybrid vigour (heterosis): larger, more uniform, higher yielding than either parent. That first-generation cross is F1 hybrid seed.

F1 seed has real advantages: uniformity (essential for machine harvest and for supermarkets that want identical produce), vigour, and the ability to combine disease resistances from both parents. It also has a commercial property that matters: save seed from an F1 plant and the next generation segregates into a mess of different types. So growers buy new seed every year, which funds breeding and also locks in dependence.

Open-pollinated and heirloom varieties breed true from saved seed, are more variable, and are usually lower yielding, but they can be maintained by growers independently and preserve genetic diversity. Both models have a place, and the argument between them is more political than biological.

Genetic modification and gene editing

In short: GM inserts specific genes, gene editing alters existing ones, and the meaningful questions are about each individual trait, not about the category.

A genetically modified organism in the regulatory sense is one where DNA has been introduced using laboratory techniques rather than sexual crossing, usually via Agrobacterium (a soil bacterium that naturally transfers DNA into plants, hijacked for the purpose) or a gene gun.

The commercially significant GM traits are few, and only a handful are in fresh produce:

Most GM acreage is in four commodity crops, which means most people's GM exposure is via animal feed, oils, and sugar rather than via fresh fruit and vegetables.

Gene editing (CRISPR and relatives) is a different tool: rather than inserting foreign DNA, it makes a targeted change to a gene the plant already has, often producing a result indistinguishable from a natural mutation. Regulators disagree on whether this counts as GM. The United States, Japan, Argentina, and several others generally treat edited crops without foreign DNA as conventional; the European Union has treated them as GM, though the rules are under active revision. A high-GABA tomato has been sold in Japan; non-browning mushrooms and higher-yield varieties are in the pipeline.

On safety, the scientific position is settled at the level of the technique: major reviews by the US National Academies, the European Commission's own multi-decade research programme, and the World Health Organization have found no evidence that approved GM foods pose greater risk to human health than their conventional counterparts. That is a claim about the approved products, not a blanket claim that any conceivable modification is safe, which is why each trait is assessed individually.

The serious arguments about GM are not toxicological. They are about herbicide resistance in weeds driven by over-reliance on one herbicide, about corporate concentration and seed patents, about gene flow to wild relatives and organic farms, and about whether the technology has been aimed at farmer-facing rather than consumer-facing problems. Those are real arguments and this book will not pretend they are settled. What it will not do is treat "contains GM ingredients" as a health warning, because the evidence does not support that.

Monoculture and the clone problem

In short: Clonal crops are genetically identical, so a pathogen that defeats one plant defeats every plant on Earth.

Grafting and cloning give uniformity, which the market demands. They also mean the crop has no genetic variation to fall back on.

The banana is the canonical disaster. Until the 1950s the export banana was the Gros Michel variety: bigger, sweeter, and tougher-skinned than what you buy now. A soil fungus, Fusarium oxysporum f. sp. cubense race 1, causing Panama disease, spread through Latin American plantations and destroyed the trade, because every plant was the same clone and none had resistance. The industry switched to Cavendish, which resisted race 1.

Cavendish is also a sterile triploid clone, propagated vegetatively, so every Cavendish banana on Earth is genetically the same plant. Tropical Race 4 of the same fungus, which Cavendish does not resist, was identified in Taiwan in the 1990s, has since spread through Southeast Asia, Australia, the Middle East, Africa, and reached Colombia in 2019 and Peru in 2021. It persists in soil for decades and there is no chemical control. The banana trade is running the same experiment a second time and knows it.

Other cases with the same shape: the Irish potato famine (a narrow clonal potato base met late blight), the 1970 US southern corn leaf blight epidemic (a single cytoplasmic male sterility source used in most hybrid maize seed), citrus greening (huanglongbing) now devastating Florida's orange industry, and Xylella killing centuries-old olive trees in Puglia.

The defence is diversity: gene banks such as the Svalbard Global Seed Vault and the many national collections, wild relatives conserved in situ, and breeding programmes that keep introducing resistance. It is slow, unglamorous, and underfunded relative to what it protects.

What breeding has done to flavour and nutrition

In short: Selection for yield, shipping, and appearance has cost flavour in specific documented cases, and the nutrient-dilution story is real but smaller than headlines suggest.

The tomato is the clearest case, and the mechanism is known. A mutation that makes fruit ripen to a uniform light green before turning red (the uniform ripening trait) was selected in the 1930s and 40s because it made harvest scheduling easy and looked better on shelf. In 2012 researchers showed that the gene involved, SlGLK2, also drives chloroplast development, so the same mutation reduces the fruit's photosynthetic capacity and thereby its sugar and carotenoid content. Growers had traded flavour for uniformity without knowing it. That is a real, documented, mechanistically explained loss.

Beyond that specific case, the general claims deserve care:

• Comparisons of old and new food composition tables do show declines in some minerals in some crops over decades. Part of that is a genuine dilution effect: higher-yielding varieties produce more mass per unit of mineral uptake. Part is changes in analytical method and in which varieties were sampled. The effect is real, usually in the range of five to twenty percent for some minerals, and not the collapse that circulates online.
• Selection for shipping has cost texture and aroma in tomatoes, strawberries, and peaches, because fruit picked firm and unripe never develops full volatile profile.
• Selection against bitterness has removed some beneficial compounds along with the unpleasant ones, as covered in Chapter 1.
• Counter-trend: modern breeding programmes increasingly select explicitly for flavour and for nutrient density, because there is now a market that pays for it. Several recent tomato and strawberry releases are flavour-first.

The practical takeaway is not "old varieties good, new varieties bad." It is that ripeness and freshness affect flavour and nutrient content far more than variety does, which is why a supermarket tomato in February and a garden tomato in August seem like different species.

The bottom line

• Named fruit varieties are clones, kept alive by grafting for decades or centuries, because seedlings do not resemble their parents.
• A grafted tree is two plants: the scion decides the fruit, the rootstock decides size, precocity, disease tolerance, and soil tolerance. Dwarfing rootstocks reshaped the entire fruit industry.
• Breeding a new tree fruit variety takes fifteen to twenty-five years and tens of thousands of discarded seedlings, which is why modern varieties are patented and licensed.
• Clonal uniformity is the industry's greatest efficiency and greatest structural risk. The banana has already been destroyed once by it and is being destroyed again.
• Genetic modification and gene editing are breeding techniques. The evidence on approved products does not support treating them as a health risk; the serious debates are about herbicides, patents, and corporate concentration.

Sources and notes

Apple domestication and Malus sieversii ancestry follow Cornille et al., PLoS Genetics, 2012, and Duan et al.'s apple genome work. Malling rootstock history is from East Malling Research Station's published record. The phylloxera epidemic and the grafting solution follow Campbell, Phylloxera, 2004. Breeding timelines and club variety economics are from university breeding programme publications, particularly Minnesota (Honeycrisp) and Washington State (Cosmic Crisp). Mutation breeding numbers are from the joint FAO/IAEA Mutant Variety Database. GM crop traits, acreage, and the Hawaiian papaya case follow ISAAA reporting and Gonsalves' published account. The safety position follows the US National Academies' Genetically Engineered Crops report, 2016, the European Commission's decade of EU-funded GMO research report, 2010, and WHO statements. Banana Panama disease history and Tropical Race 4 spread follow Ploetz's reviews and FAO reporting. Tomato flavour loss traced to SlGLK2 is Powell et al., Science, 2012. Nutrient dilution over decades follows Davis et al., 2004, and its critiques.

Open questions. How much of the measured decline in crop mineral content is genuine dilution from higher yields versus changed analytical methods and cultivar sampling is not resolved. Whether gene-edited crops should be regulated as conventional or as GM remains an active legal question, with the EU position still under revision.

👉 Next: pests, disease, and what actually gets sprayed.

Pests, Disease, and What Actually Gets Sprayed

TL;DR. Roughly 20 to 40 percent of global crop production is lost to pests and disease despite everything thrown at them. Pesticides are the tool that made modern yields possible and they have a genuine cost: some older classes were seriously toxic, resistance evolves relentlessly, and drift and runoff damage things that were not the target. The residues on the food you buy are, in nearly all cases, very small fractions of levels shown to be safe in animal studies, and this is one of the most heavily monitored things in the food supply. The real risks concentrate on the people who apply the chemicals, not the people who eat the produce.

Key takeaways

• Roughly 20 to 40 percent of potential global crop yield is lost to pests, weeds, and disease even with current controls. Without any control, losses in some crops would exceed 80 percent.
• Pesticide toxicity varies enormously. Comparing a modern selective insecticide to 1950s organophosphates or to DDT is like comparing a modern car to one without seat belts.
• Resistance is inevitable and is managed by rotating modes of action, not by spraying harder.
• Organic farming uses pesticides, drawn from an approved list of mostly natural origin compounds. Some of them, like copper, are more environmentally persistent than the synthetics they replace.
• Washing produce removes some surface residue; peeling removes more; neither is the main reason residues are low. Regulatory limits and pre-harvest intervals are.

What is actually attacking the crop

In short: Four categories, each needing a different control strategy: insects, fungi, weeds, and everything else.

The largest single loss category worldwide is weeds, which is unglamorous and true. Weeds do not make headlines but they quietly take a bigger share than insects do.

A few specific enemies are worth knowing because they shape whole industries:

Apple scab (Venturia inaequalis) is why apple growing in humid climates is chemically intensive. Spores overwinter in fallen leaves, release during spring rain, and infect leaves and fruit, producing corky brown lesions that make fruit unsaleable. A wet spring in a susceptible orchard means a spray roughly every seven to ten days. Scab-resistant varieties exist (Liberty, Topaz, and others carrying the Vf resistance gene) and have struggled commercially because consumers buy the varieties they know.

Late blight (Phytophthora infestans), the organism behind the Irish potato famine, is still the most destructive potato disease worldwide. It can destroy a field in under two weeks in the right weather, which is why potato growers spray on forecast-driven schedules tied to humidity and temperature.

Codling moth is the worm in the apple, literally: the larva tunnels to the core. It is the reason apples are sprayed at specific times, and it is also one of the great successes of alternative control, via mating disruption: dispensers flood the orchard with synthetic female pheromone so males cannot locate real females. This works, is target-specific, harms nothing else, and is widely used.

Varroa mite attacks honeybees, not crops, and matters because of pollination (Chapter 4). It is a major contributor to colony losses worldwide.

What a pesticide is, and the classes worth knowing

In short: "Pesticide" is an umbrella covering insecticides, fungicides, herbicides, and more, spanning enormous ranges of toxicity and selectivity.

The word covers any substance used to kill or control a pest. In practice, more than half of global pesticide use by volume is herbicides.

Insecticides, by generation

The direction of travel is genuine: newer insecticides are generally more selective and less acutely toxic to mammals. That progress is real and it is not the end of the story, because selectivity for mammals is not the same as harmlessness to ecosystems. Neonicotinoids are the clearest case: they are strikingly safe for people compared with what they replaced, and they are strikingly bad for pollinators, because they are systemic and turn up in pollen and nectar.

Fungicides

Two broad families. Protectants (copper, sulphur, mancozeb, captan) coat the leaf surface and stop spores germinating; they must be reapplied after rain and resistance to them develops slowly because they hit many targets at once. Systemics (strobilurins, triazoles, SDHIs) are absorbed and move within the plant, work after infection has begun, and are far more prone to resistance because they hit a single molecular target.

Copper deserves a note, because it is the fungicide approved for organic production and it is not benign. Copper does not degrade. It accumulates in soil indefinitely, is toxic to earthworms and soil microbes at accumulated levels, and long-used organic vineyards and orchards in Europe have measurable copper build-up. The EU has been progressively restricting its use, including in organic systems. This is a good illustration that "approved for organic" is a regulatory category, not a toxicological one.

Herbicides

Glyphosate is the most used herbicide on Earth and the most argued about, so it is worth stating what is actually known. It blocks an enzyme (EPSP synthase) in a metabolic pathway that plants and many microbes have and animals do not, which is the source of its low acute mammalian toxicity. On cancer: the WHO's International Agency for Research on Cancer classified it in 2015 as "probably carcinogenic to humans" (Group 2A), a hazard classification meaning the agency judged there is evidence it can cause cancer under some conditions. Regulatory agencies including the EPA, EFSA, and others conducting risk assessments, which ask whether it does so at realistic exposures, have concluded it is unlikely to pose a cancer risk at expected dietary exposures. Both statements can be true because they answer different questions, a distinction covered properly in Chapter 97. The occupational exposure question, for people who mix and spray it for a living, is genuinely less settled than the dietary one.

The larger and less contested problem with glyphosate is agronomic: two decades of heavy reliance on it, especially with herbicide-tolerant crops, has produced glyphosate-resistant weeds across tens of millions of hectares, which has pushed farmers back toward older and more toxic herbicides.

Resistance: the treadmill nobody escapes

In short: Any control applied uniformly and repeatedly selects for the organisms that survive it, and there is no way around this except deliberate variation.

Spray a field and you kill 99 percent of the target. The 1 percent that survives may carry a mutation that made them survive. They breed. Next generation, the survivor fraction is larger. Repeat for a decade and the product no longer works.

This is not a hypothetical: over 500 insect species have documented resistance to at least one insecticide, over 250 weed species to at least one herbicide, and fungicide resistance can appear within two or three seasons for single-target systemics.

Resistance management is therefore a formal discipline with published rules:

• Rotate modes of action, not just brand names. Products carry a FRAC/IRAC/HRAC group number identifying their biochemical target, and the rule is to alternate group numbers between applications.
• Use mixtures of two modes of action, so a resistant individual still meets a lethal compound.
• Leave refuges. In Bt maize and cotton, growers are required to plant a percentage of non-Bt crop nearby so susceptible insects survive there and dilute resistance genes by mating with survivors. This is a genuinely clever piece of applied evolutionary biology and it has measurably slowed resistance where it has been enforced.
• Do not rely on chemistry alone. Rotation, resistant varieties, sanitation, and biological control all reduce selection pressure.

Integrated Pest Management: what good practice actually looks like

In short: Monitor, decide with thresholds, use the least disruptive tool that works, and treat spraying as one option rather than the default.

Integrated Pest Management (IPM) is the mainstream professional standard, and it is much less about ideology than about not wasting money.

1. Prevention. Resistant varieties, rotation, sanitation (removing infected material), healthy soil, and habitat for beneficial insects.
2. Monitoring. Pheromone traps, sticky traps, weekly scouting, degree-day models that predict when a pest will hatch, and disease forecasting models based on leaf wetness and temperature. This is where the modern effort goes.
3. Thresholds. Do not treat because a pest exists; treat when its density crosses the level at which the damage would exceed the cost of treatment. Some aphids in a field are food for the ladybirds you want to keep.
4. Intervention, least disruptive first. Biological control (releasing predatory mites, parasitic wasps, or nematodes), mating disruption, physical barriers and netting, and then selective chemistry, with broad-spectrum products last.
5. Evaluation. Did it work, and what does that mean for next year.

Biological control is underrated by the public and heavily used commercially, especially in glasshouses, where a Dutch pepper grower may release predatory mites (Phytoseiulus against spider mite), parasitic wasps (Encarsia against whitefly), and predatory bugs (Orius against thrips) as standard practice, with almost no insecticide. It works there because the enclosed environment lets you establish predator populations reliably.

What ends up on your food

In short: Residues are regulated by a chain of limits with large built-in safety factors, monitored by testing programmes, and in the overwhelming majority of samples they are far below the legal limit.

Here is the actual chain, because it is more structured than most people realise.

1. Toxicology establishes a no-effect level. Animal studies determine the highest daily dose producing no observed adverse effect (the NOAEL), typically in two species over a lifetime, plus reproductive and developmental studies.

2. Divide by safety factors. The NOAEL is divided by 100 as standard: a factor of 10 for possible differences between animals and humans, and another 10 for variation between humans, with additional factors applied for particular concerns such as effects on developing animals. The result is the Acceptable Daily Intake (ADI), the amount a person could consume every day for life without expected harm.

3. Field trials set the residue limit. Supervised trials applying the product exactly as the label directs measure what residue remains at harvest. The Maximum Residue Level (MRL) is set at or slightly above the highest residue seen in those correct-use trials.

This is the part that is most misunderstood. An MRL is a trading and good-practice standard, not a safety threshold. Exceeding an MRL means the product was probably used incorrectly. It does not mean the food is dangerous, and in most exceedances the residue remains far below the ADI. The two numbers are often separated by a factor of hundreds or thousands.

4. Pre-harvest intervals enforce the gap. Every product label specifies a minimum number of days between last application and harvest, calculated so residues decay below the MRL. This is legally binding and it is the single most effective control on what reaches you.

5. Monitoring programmes test the food. The EU tests tens of thousands of samples a year across member states; the US FDA and USDA run parallel programmes. Published results are consistent year over year: roughly 95 to 99 percent of samples are within legal limits, about half of samples have no detectable residues at all, and dietary risk assessments based on the measured residues put actual consumer intake at a small fraction of the ADI.

Don't be confused: "detected" and "unsafe" are not related. Analytical chemistry can now detect parts per billion, which is roughly one second in thirty years. A headline saying residues were "found on 60 percent of samples" is describing the sensitivity of the instrument, not the safety of the food. The number that matters is the residue as a percentage of the ADI, and it is almost always well under one percent.

The "Dirty Dozen" list, honestly

The annual list of most-residue-bearing produce published by the Environmental Working Group is widely circulated and is genuinely misleading in one specific way: it ranks by frequency and number of detections, not by the toxicological relevance of the amounts found. Analyses that model actual exposure, including work published in the Journal of Toxicology, have found that consumer exposures to the listed pesticides on those foods were typically several orders of magnitude below levels of concern, and that substituting organic for conventional would not meaningfully change that exposure.

The reason to care about the list is nonetheless real, and it is this: if it persuades anyone to eat fewer strawberries or less spinach, it does clear harm, because the established benefit of eating fruit and vegetables vastly outweighs any residue risk that has been demonstrated. If you can afford organic and prefer it, that is a legitimate choice for environmental and farmworker reasons. Choosing to eat less produce because of residues is the one response the evidence does not support.

What washing and peeling actually do

• Rinsing under running water for 30 seconds removes a meaningful share of surface residues, plus dirt and a good part of any surface microbes. Rubbing with your hands helps more than the water alone.
• Baking soda solution (about 1 percent) has been shown in laboratory work to remove more surface residue than plain water for some compounds, because it speeds their breakdown. It is not necessary, but it is not nonsense.
• Vinegar, salt water, and commercial produce washes are not clearly better than water for residues, though vinegar solutions do reduce surface bacteria somewhat.
• Peeling removes more residue than any wash, and also removes fibre and a disproportionate share of some nutrients and polyphenols. For most produce that is a bad trade.
• Systemic pesticides cannot be washed off because they are inside the plant tissue. This is a genuine limitation of washing and the reason pre-harvest intervals matter more than washing does.
• Food safety is the better reason to wash. E. coli, Salmonella, and Listeria on produce cause far more documented illness every year than pesticide residues have ever been shown to. Melons are the notorious case: the rind is not eaten, so people do not wash it, and the knife carries surface bacteria straight into the flesh.

The costs that are real

Nothing above should be read as a claim that pesticides are harmless. The genuine problems are these, and they are not primarily about your dinner.

Applicator and farmworker exposure is where documented human harm concentrates. Acute poisonings number in the hundreds of thousands globally each year, overwhelmingly in low-income countries where protective equipment and enforcement are weak, and paraquat in particular has caused a very large number of deaths. Chronic exposure associations, including some cancers and Parkinson's disease, are studied primarily in agricultural cohorts for this reason.

Pollinators. Neonicotinoid effects on bees are well documented in both laboratory and field studies, which is why the EU banned the main three for outdoor use in 2018. Pollinator decline has multiple causes (habitat loss, varroa, disease, climate) and pesticides are one of them.

Aquatic systems. Pyrethroids are extremely toxic to fish and aquatic invertebrates at very low concentrations. Runoff and spray drift into watercourses is a real and measurable harm.

Ecosystem simplification. Broad-spectrum insecticides kill the predators of the pest as well as the pest, which can cause a secondary pest outbreak where a previously insignificant species explodes because its natural enemies are gone. This is a classic and well-documented failure mode of calendar spraying.

The bottom line

• Pests, weeds, and disease take a fifth to two fifths of world crop production even with modern control. The counterfactual to pesticide use is not the same food with fewer chemicals, it is substantially less food.
• Pesticide classes differ enormously. The trajectory from arsenicals to DDT to organophosphates to modern selective chemistry is real progress in human toxicity, with ecological costs that are still being reckoned with.
• Resistance is inevitable and is managed by rotating modes of action and not relying on chemistry alone.
• Residues on retail produce are monitored heavily and sit far below the levels used to set safety limits. "Detected" says more about instruments than about risk.
• The documented human harm from pesticides falls overwhelmingly on the people who apply them, not on the people who eat the food.

Sources and notes

Global crop loss estimates follow Oerke's analyses in the Journal of Agricultural Science and subsequent FAO assessments. Pesticide class properties, resistance mechanisms, and mode-of-action group numbering follow IRAC, FRAC, and HRAC published classification schemes. The residue regulatory chain (NOAEL, ADI, MRL, pre-harvest interval) follows EFSA and Codex Alimentarius procedures. Monitoring results are from the EFSA annual EU report on pesticide residues in food and the USDA Pesticide Data Program. The Dirty Dozen critique follows Winter and Katz, Journal of Toxicology, 2011. Glyphosate's divergent classifications are IARC Monograph 112 (2015) versus EFSA and EPA risk assessments. Neonicotinoid effects on pollinators follow EFSA's 2018 conclusions and the field studies underlying the EU outdoor ban. Copper accumulation in organic systems follows EFSA's copper review and EU restriction decisions. Occupational poisoning figures are from WHO estimates and the paraquat literature. Washing and residue removal follows Yang et al., Journal of Agricultural and Food Chemistry, 2017 (baking soda).

Open questions. The long-term health effects of chronic low-level pesticide exposure in agricultural workers, particularly associations with Parkinson's disease and some cancers, remain under active study and are not settled. Whether IPM adoption has actually reduced total pesticide load, as opposed to shifting which compounds are used, is poorly measured at national scale.

👉 Next: organic, regenerative, and what the labels actually mean.

Organic, Regenerative, and What the Labels Mean

TL;DR. Organic is a legally defined production standard, not a health claim. It bans synthetic pesticides and fertilisers and permits a list of mostly natural-origin ones, some of which are genuinely toxic. Organic produce carries lower synthetic pesticide residues, has slightly higher levels of some plant compounds and lower cadmium, and has not been shown in any convincing trial to make people healthier. It yields roughly 20 percent less per hectare on average, which means it is better per hectare and often worse per kilogram on environmental measures. Most of the other labels on produce are worth much less than organic, and a few are worth nothing at all.

Key takeaways

• Organic means no synthetic pesticides or fertilisers, no GMOs, and no synthetic additives, plus animal welfare rules. It does not mean pesticide-free, local, or more nutritious.
• Organic yields average about 80 percent of conventional, with the gap much smaller for legumes and perennials and much larger for cereals under high-input conditions.
• Nutritional differences are small. Meta-analyses find higher polyphenols and lower cadmium and pesticide residues in organic crops; no measurable health outcome difference in humans has been demonstrated.
• "Natural," "farm fresh," "free from," and most front-of-pack claims are marketing with no legal definition in most jurisdictions.
• Regenerative agriculture has no single legal definition, which makes it both the most promising and the most abusable term on this list.

What organic certification actually requires

In short: A production process standard, audited annually, covering inputs, soil management, and record keeping, with a multi-year conversion period.

Organic is a process standard. It certifies how the food was produced, not what is in it, and there is no test you can perform on an apple to determine whether it is organic. Certification rests on documentation, farm inspection, and traceability.

The main regimes are the EU organic regulation, the USDA National Organic Program, the UK's certifiers such as the Soil Association, and equivalents in Canada, Japan (JAS), Australia, and elsewhere. They differ in detail and agree on the core.

Prohibited in organic production:

• Synthetic pesticides and herbicides (with narrow exceptions)
• Synthetic nitrogen fertiliser, and most soluble synthetic fertilisers
• Genetically modified organisms
• Sewage sludge as fertiliser
• Irradiation of food
• Most synthetic food additives and processing aids
• Routine prophylactic antibiotics and growth hormones in livestock

Required:

• A documented soil fertility plan built on rotation, cover crops, compost, and manure
• A conversion period, typically 2 years for annual crops and 3 years for perennials, during which the rules apply but the produce cannot be sold as organic
• Buffer zones to limit drift from neighbouring conventional fields
• Separation and traceability records through processing and packing
• Livestock: outdoor access, organic feed, welfare-based space standards, restricted antibiotic use with extended withdrawal periods

Permitted pesticides in organic production is the section that most surprises people. Organic farming uses pesticides, from an approved list of substances that are naturally derived or judged low-risk:

The Bt entry is the sharpest illustration of the philosophical basis of organic standards. Spraying the bacterium's insecticidal protein onto a crop is organic. Having the plant produce the identical protein is prohibited, because the objection is to the technique, not the molecule. Whether you find that coherent depends on whether you think the standard is about substances or about processes. It is explicitly about processes.

Don't be confused: organic is not pesticide-free. If you want pesticide-free food you are looking at "no-spray" claims from individual growers, which are unregulated, or at growing it yourself. Organic certification permits a defined pesticide list, and organic produce in monitoring programmes does sometimes carry residues, both of approved organic substances and of trace drift from neighbours.

Does organic food make you healthier?

In short: Composition differs slightly in measurable ways; no controlled evidence shows a health benefit in people.

Three separate questions get collapsed into one. Take them apart.

1. Is the composition different?

Yes, slightly, and in directions you would predict from the production method. The largest systematic reviews, notably the 2014 meta-analysis of 343 studies led by Barański and colleagues in the British Journal of Nutrition, found in organic crops:

• Higher concentrations of polyphenols and other antioxidant compounds, in the range of roughly 19 to 69 percent higher depending on the compound class. The likely mechanism is that plants receiving less soluble nitrogen and experiencing more pest pressure invest more in defensive chemistry, exactly as Chapter 1 would predict.
• Lower cadmium, roughly half, plausibly because phosphate fertilisers are a cadmium source.
• Four times lower frequency of detectable pesticide residues.
• No meaningful difference in most vitamins and minerals. Vitamin C differences are small and inconsistent.
• Higher dry matter in some crops, which mechanically raises the concentration of everything per gram, and which some analyses do not adjust for.

Organic milk and meat show a modest but reasonably consistent increase in omega-3 fatty acids, which is a grazing effect rather than an organic effect: cattle eating more grass produce more omega-3 whether or not the farm is certified.

2. Does that composition difference matter to health?

There is no good evidence that it does. The polyphenol difference sounds impressive as a percentage and is small in absolute terms relative to your total intake, and polyphenols themselves have weaker health evidence than their reputation suggests (Chapter 17).

3. Do organic eaters have better outcomes?

Large observational studies, notably the French NutriNet-Santé cohort, have reported lower cancer incidence among frequent organic consumers. The study is well conducted and the finding is real as an association. The problem is unavoidable confounding: people who buy organic food in France also, on average, smoke less, drink less, exercise more, are richer, better educated, and eat more vegetables and less processed meat. Statistical adjustment helps and cannot fully remove this. The authors themselves say so.

The honest summary is: organic food is not worse for you, is very slightly different in composition, and has never been shown in a controlled setting to make anyone healthier. If you are choosing organic, choose it for reasons that hold up: lower synthetic pesticide residue if that matters to you, better animal welfare standards, reduced farmworker exposure, and support for a farming system with different environmental trade-offs.

The environmental accounting, which cuts both ways

In short: Organic is generally better per hectare and often worse per kilogram, and which measure matters depends on whether land is the constraint.

This is the genuinely difficult question and it does not have a clean answer.

Where organic performs better:

• Higher soil organic matter and better soil biological activity, consistently measured.
• Substantially higher biodiversity on the farm, typically 30 percent more species, particularly for birds, pollinators, and soil fauna.
• No synthetic nitrogen means no Haber-Bosch energy cost and generally lower nitrate leaching per hectare.
• Lower pesticide load in the surrounding environment, with the copper caveat above.

Where organic performs worse:

• Yields average around 80 percent of conventional, with wide variation. The gap is small (5 to 10 percent) for legumes, perennials, and some fruit, and large (25 to 35 percent) for cereals in high-input systems.
• That yield gap means more land per unit of food. On measures expressed per kilogram of food, organic frequently comes out worse on greenhouse gas emissions and land use.
• Land use is the crux. If producing the same food organically requires 20 to 30 percent more land, and that land comes from somewhere, the carbon and biodiversity cost of the land conversion can exceed the on-farm benefit. A widely discussed 2019 Nature Communications analysis modelled converting England and Wales entirely to organic and found domestic emissions fell while total emissions rose once the overseas land needed to make up the food shortfall was counted.
• Organic livestock generally produce more methane per unit of milk or meat, because slower growth means more lifetime emissions.
• Mechanical weed control instead of herbicide means more tractor passes, more fuel, and in some systems more soil disturbance, which is in tension with the soil goals.

The reasonable conclusion is that this is a genuine trade-off rather than a scoreboard, and that the highest-leverage environmental choices in food are not organic versus conventional at all. They are, in rough order of effect size: how much beef and lamb you eat, how much food you waste, and how much air-freighted produce you buy. Those three dominate. Chapter 9 returns to this.

Regenerative agriculture

In short: A set of soil-first practices with good science behind them and no legal definition, which means the term is being used for everything from serious farming to pure marketing.

Regenerative agriculture is defined by practices rather than prohibitions, and the practices are the ones Chapter 2 endorsed:

1. Minimise soil disturbance (no-till or reduced till)
2. Keep the soil covered (residue, mulch, cover crops)
3. Keep living roots in the ground as much of the year as possible
4. Maximise diversity in crops and rotations
5. Integrate livestock where possible, using managed grazing

Notably, regenerative is not the same as organic and the two sometimes conflict: no-till farming usually depends on herbicide for weed control, which organic prohibits, so organic systems tend to till more. Some of the best soil outcomes come from conventional no-till farms, and some organic farms have poor soil structure from repeated cultivation.

The evidence is strongest for soil structure, water infiltration, erosion reduction, and farm biodiversity, all of which are well supported. The evidence is weakest, and most oversold, for carbon sequestration. Soils do gain carbon under these practices, but the gains saturate after a couple of decades, are reversible the moment practices change, are genuinely hard to measure at field scale, and some early estimates counted carbon moving deeper in the profile rather than being newly added. Treat "regenerative agriculture will reverse climate change" as marketing, and "regenerative agriculture builds better soil and farms" as well supported.

Because there is no certification behind the word in most markets, its value on a package is close to zero. Ask what practices, on what land, verified how.

Every other label, decoded

In short: A few labels are legally defined and audited, most are marketing, and the difference is not visible from the front of the package.

The general rule: a label that names a certifying body and a standard document is worth reading about. A label that is an adjective is worth ignoring.

Hydroponics, vertical farming, and whether soil matters

In short: Plants do not need soil, they need water, oxygen, and dissolved minerals, and the resulting food is nutritionally comparable.

Hydroponics grows plants with roots in nutrient solution or an inert medium such as rockwool or coir. It is standard for glasshouse tomatoes, peppers, cucumbers, and leafy greens, and produces enormous yields with very high water efficiency because water is recirculated.

Two things are commonly asserted and are not supported. "Hydroponic produce lacks nutrients": comparative studies find nutrient content is broadly similar and depends far more on variety, light, and harvest timing than on the medium. Vitamin content can be higher in controlled environments because of consistent light and immediate harvest. "Hydroponic produce is unnatural": the plant experiences the same dissolved ions it would take up from soil water, and cannot tell the difference.

The genuine debates are different. Whether hydroponics should be certifiable as organic is contested: the US permits it, the EU does not, on the reasonable grounds that organic standards are built around soil stewardship and hydroponics has no soil to steward.

Vertical farming, stacking hydroponic layers under LED lighting indoors, is energy-hungry: you are replacing free sunlight with electricity. It makes economic sense for high-value, fast, light-demanding-but-low-light-requiring crops close to expensive urban markets, which is why the sector has converged on leafy greens and herbs, and why several high-profile vertical farming companies have failed trying to grow anything else. Nobody is growing wheat in a tower.

The bottom line

• Organic certifies a production process. It permits a list of approved pesticides, some of which, like copper, are genuinely problematic.
• Organic produce measurably has fewer synthetic pesticide residues, somewhat more polyphenols, and less cadmium. No controlled evidence shows this makes people healthier.
• Organic yields around 20 percent less on average, which makes it better per hectare and often worse per kilogram. Land use is where the argument actually lives.
• Regenerative practices are well supported for soil and biodiversity, oversold for carbon, and legally undefined, so the word alone guarantees nothing.
• Most other front-of-pack claims are marketing. Look for a named certifier and a written standard.
• If you want the highest-leverage food choices for the environment, they are red meat frequency, food waste, and air freight, not the organic aisle.

Sources and notes

Organic standards follow EU Regulation 2018/848, the USDA National Organic Program rule, and Soil Association standards. Permitted input lists are from those regulations. Composition differences follow Baranski et al., British Journal of Nutrition, 2014, and Srednicka-Tober et al. for milk and meat. The absence of demonstrated health outcome differences follows Smith-Spangler et al., Annals of Internal Medicine, 2012, and Dangour et al.'s FSA-commissioned review. The NutriNet-Sante cancer finding is Baudry et al., JAMA Internal Medicine, 2018, with the authors' own confounding caveats. Yield gaps follow Seufert et al., Nature, 2012, and Ponisio et al., 2015. The England and Wales organic conversion modelling is Smith et al., Nature Communications, 2019. Regenerative practice evidence follows Rodale, FAO, and the soil carbon literature discussed in Chapter 2. Hydroponic nutrient comparisons follow controlled-environment agriculture reviews.

Open questions. Whether organic farming scaled globally would be net positive or negative environmentally depends entirely on assumptions about land use change and diet, and reasonable analyses reach opposite conclusions. Soil carbon sequestration claims for regenerative systems are the least settled part of this chapter.

👉 Next: how produce survives the trip from field to shelf, including how an apple can be a year old and still crisp.

From Field to Shelf

TL;DR. A picked fruit is still alive. It is still breathing, still burning its own sugar, and still running the enzymatic programme that ends in mush. Everything in post-harvest handling is an attempt to slow that programme down: cooling within hours, holding at the exact temperature the crop tolerates, controlling the atmosphere around it, and blocking the ripening hormone. Done well, an apple picked in September is crisp the following July. Done badly, a strawberry is worthless in three days. This chapter is why the produce aisle looks the way it does.

Key takeaways

• Cooling within hours of harvest matters more than anything else. Every 10 °C reduction roughly halves to thirds the rate of respiration and therefore of decay.
• Controlled atmosphere storage holds apples for six to twelve months by dropping oxygen to 1 to 3 percent and raising carbon dioxide, which nearly stops respiration.
• Chilling injury is real: tomatoes, bananas, avocados, citrus, and cucumbers are damaged by ordinary refrigerator temperatures, which is why they belong on the counter.
• 1-MCP (SmartFresh) blocks the ethylene receptor and is the single most important post-harvest chemical in the fruit trade.
• Roughly a third of all food produced is lost or wasted. In low-income countries most of it is lost before it reaches a shop; in high-income countries most of it is thrown away by shops and households.

The picked fruit is still alive

In short: Respiration continues after harvest, consuming the fruit's own reserves, and its rate predicts shelf life almost exactly.

When a fruit is picked, it loses its supply line: no more water or sugar from the plant. What it does not lose is metabolism. It keeps respiring, burning stored sugar with oxygen to keep its cells running, giving off carbon dioxide, water, and heat.

$$\mathrm{C_6H_{12}O_6} + 6,\mathrm{O_2} ;\longrightarrow; 6,\mathrm{CO_2} + 6,\mathrm{H_2O} + \text{energy}$$

Respiration rate is the master variable of shelf life. A crop that respires fast burns through its reserves fast, softens fast, and dies fast.

Sweetcorn is the classic demonstration. It respires so fast that at room temperature it converts a meaningful fraction of its sugar to starch within a day, which is why old gardening advice was to have the water boiling before you pick it. Cool it immediately and that loss slows by an order of magnitude. Modern supersweet varieties, bred with mutations that slow sugar conversion, made the advice less urgent, which is an example of breeding solving a logistics problem.

Respiration roughly doubles to triples for every 10 °C rise. This single relationship is the entire economic justification of the cold chain. A crop that lasts 2 days at 25 °C may last 8 to 20 days at 5 °C, and it costs less to refrigerate it than to lose it.

The cold chain

In short: An unbroken sequence of temperature control from field to kitchen, where the first hour matters most and every break is cumulative.

Precooling: the first few hours

Field heat must come out fast. A pallet of lettuce cut at 28 °C and stacked will cook itself through its own respiratory heat. Commercial precooling methods:

Vacuum cooling is the elegant one: reduce the pressure and the water on the lettuce evaporates at low temperature, and because evaporating water absorbs a lot of heat (2,260 joules per gram), the lettuce chills itself from the inside of every crevice. A head of lettuce goes from 25 °C to 2 °C in half an hour, losing about 1 percent of weight per 5 °C of cooling.

Optimum storage temperatures, and the ones that are not near zero

Most produce keeps best just above freezing. A significant minority is damaged there, and this is where domestic refrigerators go wrong.

Chilling injury is a distinct phenomenon from freezing. Cell membranes of tropical and subtropical species lose their fluidity below a threshold, leaks develop, enzymes escape their compartments, and the tissue degrades. Symptoms are pitting, brown staining, failure to ripen, and off-flavours. The cruel part is that damage is often invisible while the item is cold and appears within hours of returning it to room temperature.

The tomato case is the best documented and the most useful. Research at the University of Florida showed that storing tomatoes below about 12 °C reduces the expression of the genes producing the volatile aroma compounds that make a tomato taste like a tomato, and that the loss is not fully reversible on rewarming. A refrigerated tomato genuinely does taste worse, and there is a mechanism for it, not merely folklore. Store them on the counter, stem side down, out of direct sun, and refrigerate only when fully ripe and only for a day or two.

Humidity and water loss

Produce is 80 to 95 percent water and loses it continuously to the air. A loss of 3 to 6 percent of weight makes most produce visibly limp and unsaleable. So storage rooms run at 90 to 98 percent relative humidity for almost everything, with the exceptions of onions, garlic, pumpkins, and nuts, which need dry air to prevent rot.

The practical consequences in a domestic fridge: the crisper drawer exists to raise local humidity, produce in perforated plastic bags keeps far better than produce left bare, and a limp carrot can be revived by soaking in cold water because the loss was water, not structure. Wilting is usually reversible; rot is not.

Controlled atmosphere: how an apple lasts a year

In short: Reduce oxygen to 1 to 3 percent and raise carbon dioxide, and respiration nearly stops without the fruit suffocating.

If respiration needs oxygen, remove most of the oxygen. That is controlled atmosphere (CA) storage, developed in the 1920s at Cambridge by Franklin Kidd and Cyril West, and it is why apples are available year-round.

A CA room is gas-tight. After loading, the fruit's own respiration draws the oxygen down, assisted by nitrogen generators, until the atmosphere is roughly:

• Oxygen: 1 to 3 percent (air is 21 percent)
• Carbon dioxide: 1 to 5 percent (air is 0.04 percent)
• Temperature: 0 to 2 °C
• Humidity: 90 to 95 percent

Under those conditions apple respiration falls to a small fraction of its rate in air, ethylene production is suppressed, softening nearly stops, and the fruit can be held for six to twelve months. The room is not opened until the fruit is needed; personnel entering a CA room without breathing apparatus would lose consciousness, and the doors carry serious warnings for that reason.

Dynamic controlled atmosphere (DCA) pushes further, lowering oxygen to below 1 percent while continuously monitoring the fruit for the first signs of fermentation stress, using either chlorophyll fluorescence sensors or ethanol measurement in the room air, and easing the oxygen back up when the fruit signals distress. It gets more storage life and better texture from the same fruit.

Modified atmosphere packaging (MAP) is the retail-scale version: a sealed bag with a film of specific gas permeability, so the produce's own respiration establishes a beneficial equilibrium inside the pack. This is why bagged salad stays alive for a week and why a bagged salad opened and resealed deteriorates far faster.

Don't be confused: "a year-old apple" is not a scandal. An apple in CA storage since October is not decaying slowly; it is essentially paused. Its vitamin C declines gradually (by perhaps a quarter to a third over long storage), its texture is deliberately protected, and its sugars barely change. The alternative is not a fresher local apple in June, because there is no such thing in the northern hemisphere. The alternative is an apple flown from New Zealand. Both are legitimate; the CA apple usually has the smaller carbon footprint.

Blocking ethylene: 1-MCP

In short: A molecule that occupies the fruit's ethylene receptors without activating them, effectively pausing ripening for months.

Chapter 1 covered ethylene, the gas that triggers ripening. The post-harvest industry has a way to switch that signal off.

1-methylcyclopropene (1-MCP), sold as SmartFresh, is a small gas molecule that binds irreversibly to the ethylene receptors in plant cells. It does not trigger them; it blocks them. Fruit treated with a few parts per billion for 12 to 24 hours after harvest becomes effectively deaf to ethylene, its own and everyone else's, until new receptors are synthesised, which takes weeks to months.

The effect is large. Treated apples stay firm and green-stemmed for months longer, treated avocados and bananas hold, and treated flowers last longer in the vase. It is approved in many countries, leaves negligible residue (it is a gas applied at parts per billion and dissipates), and is not something a consumer will ever see on a label.

The trade-off is quality, not safety. Aggressively treated fruit may never develop full aroma, because the volatile compounds that make ripe fruit smell ripe are themselves downstream of ethylene signalling. A very firm, very bland apple in April is often a 1-MCP artefact. The same is true of the avocado that goes from rock hard to brown inside without passing through a good stage.

Other post-harvest treatments

Waxing. Fruit has a natural waxy cuticle that limits water loss, and washing removes part of it. Commercial packers replace it with a thin food-grade coating: carnauba wax (from a palm), shellac (a resin secreted by the lac insect, so not vegan), or plant-derived and lipid coatings. It reduces water loss and shrivelling, adds shine, and can carry a fungicide. It is edible, present in small amounts, and worth washing off if you dislike it. Shellac on citrus and apples is the reason some fruit is not vegan, which surprises people.

Curing. Potatoes, sweet potatoes, onions, and garlic are deliberately held warm and humid for a week or two after harvest so that wounds heal over with suberin and skins set. Uncured potatoes bruise and rot; cured ones store for months.

Degreening. Citrus grown in warm climates can be fully ripe and still green, because peel colour depends on cool nights breaking down chlorophyll. Since consumers refuse green oranges, packers expose them to low levels of ethylene to destroy the chlorophyll and reveal the orange carotenoids underneath. Nothing about the flesh changes. In some countries oranges are also dyed, which must be declared.

Fumigation and quarantine treatment. Fruit crossing borders often must be treated against fruit fly and other quarantine pests: methyl bromide fumigation (being phased out under the Montreal Protocol), cold treatment (holding at 1 °C for two to three weeks in transit), hot water dipping for mangoes, or irradiation.

Irradiation deserves a note because it is widely misunderstood. Exposing food to ionising radiation kills insects and microbes and inhibits sprouting. It does not make food radioactive, any more than a dental X-ray makes your jaw radioactive. It is approved for specific foods in many countries, must be labelled with the international Radura symbol, causes small nutrient losses comparable to cooking, and is used far less than its safety record would justify, mostly because of consumer resistance.

Sulphur dioxide pads in grape cartons suppress botrytis, which is why table grapes travel well. People with sulphite sensitivity, which overlaps heavily with asthma, should know this.

Shipping and the physical journey

In short: Most produce moves in refrigerated shipping containers by sea; a small, high-value fraction flies, and that fraction dominates the carbon footprint of what it carries.

Reefer containers are refrigerated shipping containers with their own compressor units, plugged into ship power, holding a set temperature to within a degree for weeks. The best ones run controlled atmosphere in transit, so a banana ship is a moving CA store. Roughly all long-distance banana, citrus, grape, apple, kiwi, and avocado trade moves this way.

Sea freight is astonishingly efficient per kilogram. Shipping produce halfway around the world by sea typically adds a small fraction of the food's total carbon footprint, often on the order of 0.1 kg CO₂ equivalent per kilogram of fruit. Production usually dominates.

Air freight is 20 to 100 times worse per tonne-kilometre, and it is used for the small set of goods that cannot survive sea transit: asparagus out of season, fresh berries out of season, green beans and mangetout from East Africa, fresh herbs, some fish, and cut flowers. If you want to reduce the carbon footprint of your fruit and vegetables, the single most effective rule is to avoid air-freighted produce, which is far more targeted than "buy local." Air-freighted goods are usually identifiable by being highly perishable and out of local season.

Ripening rooms are the last stop for climacteric fruit. Sealed rooms with precise temperature control and ethylene dosing, typically 100 to 150 ppm ethylene for 24 to 48 hours at 15 to 20 °C, take bananas through a defined colour scale from 1 (all green) to 7 (yellow with brown flecks). Retailers order by colour stage, so bananas arrive at the shop at exactly the ripeness the buyer specified. This is why bananas in a shop are so uniform, and why a banana ripened at home from truly green never quite matches.

Food loss and waste

In short: Around a third of food produced is never eaten, and the losses happen in different places in rich and poor countries.

The FAO's widely cited estimate is that roughly one third of food produced for human consumption is lost or wasted. The pattern differs sharply by income level:

• Low-income countries: most losses happen early, between field and market. Inadequate cooling, poor roads, poor storage, and pest damage. Post-harvest losses of 20 to 50 percent are common for fruit and vegetables. The fix is infrastructure, and it is one of the highest-return interventions available in food systems.
• High-income countries: most losses happen late, at retail and in homes. Cosmetic rejection, over-ordering, date-label confusion, and household waste. Households are typically the single largest contributor.

Date labels cause an enormous share of household waste, and the two common labels mean different things:

Most food carries "best before." Most food thrown away for being "out of date" was perfectly safe. Several retailers have removed date labels from fresh produce entirely for exactly this reason.

The bottom line

• Harvested produce is alive and burning its own reserves. Respiration rate predicts shelf life, and cooling is the lever that controls it: every 10 °C drop slows decay by roughly half to two thirds.
• Cool fast, hold at the right temperature, and keep the humidity high. Getting the first hours right matters more than anything downstream.
• Tropical and subtropical produce suffers chilling injury in a domestic fridge. Bananas, tomatoes, cucumbers, avocados, potatoes, and sweet potatoes belong on the counter.
• Controlled atmosphere storage and 1-MCP are why fresh apples exist in July. A stored apple is paused, not stale, though heavy ethylene blocking can cost aroma.
• Sea freight is cheap in carbon; air freight is not. Avoiding air-freighted produce beats "buy local" as a rule of thumb.
• A third of food is never eaten, and in rich countries most of that loss happens in kitchens, largely through misread date labels.

Sources and notes

Respiration rates, optimum storage conditions, and chilling injury thresholds follow the USDA Agriculture Handbook 66, The Commercial Storage of Fruits, Vegetables, and Florist and Nursery Stocks, which is the standard reference for every number in the storage tables. Controlled atmosphere storage history follows Kidd and West's work at Cambridge from the 1920s. Dynamic controlled atmosphere and chlorophyll fluorescence monitoring follow Prange's published work. 1-MCP mechanism and effects follow Watkins' reviews in Biotechnology Advances, 2006. Tomato flavour loss on refrigeration is Zhang et al., PNAS, 2016, from the University of Florida group. Banana ripening room protocols and the colour scale are industry standard, documented in Chiquita and Dole technical literature. Food loss and waste estimates follow the FAO Global Food Losses and Food Waste report, 2011, and subsequent UNEP Food Waste Index reports. Date label confusion as a waste driver follows WRAP research in the UK.

Open questions. FAO's one-third food loss figure is widely quoted and rests on data of uneven quality, and subsequent analyses using different methods produce meaningfully different numbers. How much aroma is permanently lost to aggressive 1-MCP treatment, as opposed to delayed, is not well quantified.

👉 Next: choosing, storing, and not wasting it, which turns all of this into practical rules for your kitchen.

Choosing, Storing, and Not Wasting It

TL;DR. Frozen vegetables are usually as nutritious as fresh and sometimes better, because they are processed within hours of harvest while "fresh" produce may be a week old. Canned is fine, with caveats about salt and sugar. Dried is concentrated in everything including sugar. The two rules that actually reduce waste are storing things at the right temperature, which is not always the fridge, and buying against a plan rather than an aspiration. This chapter is the practical distillation of the previous eight.

Key takeaways

• Frozen produce is frozen within hours of picking, so it often has more vitamin C and folate than fresh produce that has spent a week in transit.
• Vitamin C and folate are the fragile nutrients. Minerals, fibre, and most carotenoids survive storage, cooking, and freezing essentially intact.
• Cooking increases the availability of some nutrients (lycopene in tomato, beta-carotene in carrot) while destroying others (vitamin C, some glucosinolates).
• Ethylene management in your kitchen is free and effective: separate the emitters from the sensitive.
• "Buy local" is a weak carbon rule; "avoid air freight and waste less" is a strong one.

Fresh, frozen, canned, or dried

In short: All four are legitimate. Freshness at the moment of processing matters more than the category.

The category argument is mostly wrong because it compares an idealised fresh product with a maligned processed one. The real comparison is between a specific frozen pea, frozen two hours after picking, and a specific fresh pea that was picked five days ago in another country.

Freezing deserves defending properly. Vegetables destined for freezing are picked at peak maturity, because the buyer schedules the harvest, and blanched and frozen within hours, often at a plant on the farm. Blanching (a brief immersion in boiling water or steam) destroys the enzymes that would otherwise cause off-flavours and colour loss in the freezer, and it costs 10 to 30 percent of the vitamin C up front. After that, losses nearly stop. Fresh produce, by contrast, loses vitamin C continuously: spinach can lose the majority of its vitamin C within a week at refrigerator temperature, and much faster at room temperature.

The published comparisons, including work from the University of California and from Leatherhead Food Research, find that frozen fruit and vegetables are broadly comparable to fresh, and better than fresh produce that has been stored for several days, which describes most of what is in a supermarket.

Canning loses more, and gains something. Heat destroys vitamin C and thiamine, and it breaks down cell walls, which increases the bioavailability of carotenoids. Canned tomatoes deliver more absorbable lycopene than raw tomatoes. Canned pulses are nutritionally close to home-cooked and vastly more convenient; draining and rinsing removes about 40 percent of the added sodium. The things to check on a can are added salt and added sugar, and whether fruit is in juice or syrup.

Drying removes water, which concentrates everything per gram by a factor of roughly four to six. A useful way to think about a portion: 30 g of raisins is about 130 g of grapes. That is fine as fruit and misleading as a snack, because the volume that feels like a small handful represents a large bunch. Dried fruit is also sticky and slowly released onto teeth, which makes it worse for dental health than the equivalent fresh fruit. Many dried fruits are treated with sulphur dioxide to preserve colour (bright orange apricots are sulphured, brown ones are not), which matters for people with sulphite sensitivity.

What actually destroys nutrients

In short: Heat, water, light, oxygen, and time, in different combinations for different nutrients.

The two rules that follow: use less water and less time, and keep the cooking liquid where you can. Steaming, microwaving, stir-frying, and roasting all retain more water-soluble nutrients than boiling, and microwaving is often the best of all despite its reputation, because it uses little water and short times.

Cutting matters too. Cutting a vegetable exposes cell contents to oxygen and to enzymes that were previously kept separate. That is why cut apple browns and why cut vegetables lose vitamin C faster than whole ones. Cut close to when you cook.

Don't be confused: "raw food is more nutritious" is half right at best. Raw retains vitamin C, folate, and the enzyme that converts broccoli's glucosinolates into their active form. Cooking increases the availability of lycopene, beta-carotene, and lutein several-fold, destroys antinutrients such as lectins and much of the oxalate in some greens, kills pathogens, and makes starch digestible at all. A varied diet with both is better than either dogma.

Choosing produce in the shop

In short: Weight for size, firmness, skin condition, and smell tell you more than colour does.

General rules that hold across most produce:

• Heavy for its size means high water content and juiciness. This is the single most useful test for citrus, melons, and cucumbers.
• Firm, taut skin with no soft spots. Softness means either ripeness or the start of breakdown, and the difference is location: uniform give is ripe, a localised soft patch is damage.
• Smell at the stem end. For melons, pineapples, strawberries, and stone fruit, aroma is the most reliable ripeness indicator. No smell usually means no flavour.
• Green stems and calyxes indicate recent harvest, on strawberries, tomatoes, aubergines, and peppers.
• Avoid pre-cut produce where possible: cut surfaces lose vitamin C fast, cost more, and carry more food safety risk.
• Colour is a weaker signal than people assume, because packers manage it. A green orange can be ripe; an orange tomato may be bred for it; a deep red supermarket strawberry may be flavourless.
• Bruises and blemishes are cosmetic. A scabbed or russeted apple skin is fine. A soft brown bruise means the tissue underneath has broken down.

Storing it at home

In short: Three zones, not one. Room temperature, fridge, and cool dark cupboard, plus deliberate separation of ethylene emitters.

Room temperature (do not refrigerate)

Bananas, tomatoes, avocados (until ripe), stone fruit (until ripe), mangoes, pineapple, melons (whole, uncut), citrus (a week is fine on the counter), cucumbers, aubergines, peppers (a few days), basil, winter squash and pumpkin, garlic and onions.

Once ripe, most of these can go in the fridge for a couple of extra days, accepting some quality cost. An uncut melon on the counter, a cut one in the fridge.

Cool, dark, dry, ventilated cupboard

Potatoes (never the fridge, because below about 4 °C starch converts to sugar, which makes them sweet and, more importantly, produces more acrylamide when fried or roasted, see Chapter 91), sweet potatoes, onions, garlic, shallots, winter squash. Keep potatoes in the dark: light turns them green with chlorophyll, which is harmless in itself but signals the accompanying rise in solanine, which is not.

Keep onions and potatoes apart. Onions release moisture, potatoes release ethylene, and together they spoil each other faster.

Fridge

Leafy greens, brassicas, carrots and root vegetables, berries, grapes, apples (after a week or two on the counter, if you like them cold), cherries, ripe stone fruit, mushrooms (in paper, not plastic), herbs other than basil (stems in water, bag over the top), and all cut produce.

Practical fridge notes:

• Humidity matters more than temperature for greens. A perforated bag or a container with a damp cloth doubles the life of salad leaves. Fully sealed plastic causes rot; fully open causes wilting; perforated is the answer.
• Wash before eating, not before storing. Surface water accelerates rot, especially on berries and greens.
• Mushrooms in paper. Plastic traps moisture and they go slimy.
• The crisper drawer is a humidity-controlled compartment. The slider controls a vent: high humidity (vent closed) for leafy things that wilt, lower humidity (vent open) for fruit that rots.

Ethylene separation, the free upgrade

From Chapter 1: climacteric fruit emits ethylene, and other produce responds to it.

Two ordinary kitchen failures come from ignoring this: a bag of apples next to the potatoes makes them sprout, and bananas in a fruit bowl with everything else ripens the entire bowl in a weekend.

You can also exploit it. To ripen an avocado, peach, or mango faster, put it in a paper bag with a banana or apple; the bag traps ethylene. To slow it down, separate and refrigerate once ripe.

Reducing waste, in order of effect

1. Plan before buying. Household waste is overwhelmingly caused by buying more than the week's actual meals require. A rough plan beats good intentions.
2. Understand the date labels. "Use by" is safety, "best before" is quality. Almost all fruit and vegetables carry, at most, a best-before date, and your senses are a better guide.
3. Store correctly, per the three zones above. This alone extends the life of most produce by days.
4. Freeze the surplus. Almost everything freezes: chopped herbs in oil in ice cube trays, over-ripe bananas (peeled) for baking, blanched vegetables, bread, cheese, milk, and cooked leftovers. Freezing bought produce on the day you realise you will not use it is much more effective than freezing it the day before it dies.
5. Use the whole thing. Broccoli stems (peel and slice, they are excellent), carrot tops, leek greens, herb stems, parmesan rinds, and chicken carcasses are all food. Citrus zest freezes.
6. Revive rather than discard. Limp greens, carrots, and celery recover in cold water. Stale bread makes croutons and breadcrumbs. Slightly soft tomatoes are better for sauce than firm ones.
7. Cook the ugly first. A simple ordering rule that quietly removes most waste.

Food safety in the kitchen, briefly

In short: Pathogens on produce cause far more documented illness than pesticide residues, and the controls are cold, separation, washing, and heat.

• Wash produce under running water and rub it, including things you will peel: melons, avocados, and citrus, because the knife carries surface organisms into the flesh. This is the mechanism behind several serious listeria outbreaks traced to cantaloupe.
• Do not wash raw chicken. It aerosolises campylobacter over the sink and surrounding surfaces. Cooking kills it; splashing spreads it.
• Separate raw meat from everything else, with separate boards or a strict order of work: produce first, raw meat last.
• The danger zone is roughly 5 to 60 °C. Get cooked food into the fridge within two hours, or one hour in hot weather.
• Reheat thoroughly, to steaming hot throughout, not just warm.
• The particular risks in produce: pre-washed bagged salad (a genuine and repeated source of outbreaks, because a single contaminated leaf is distributed through a batch), sprouts and beansprouts (the warm humid germination conditions are ideal for bacteria, which is why they are on the avoid list in pregnancy), melons, and unpasteurised juices.
• Rice is the classic leftover hazard: Bacillus cereus spores survive cooking, and germinate if cooked rice sits at room temperature. Cool it fast and refrigerate within an hour.

The carbon question, honestly

In short: What you eat matters much more than how far it travelled, with air freight as the one exception.

The intuitive model, that food miles dominate a food's footprint, is wrong for most foods. Analyses of full life-cycle emissions consistently find that transport is a small share of the total for most products, typically under 10 percent, and that production dominates.

In rough order of leverage:

1. How much beef and lamb you eat. Ruminant meat is an order of magnitude more emitting per gram of protein than plant proteins or poultry, because of methane from rumen fermentation and land use. This is the single largest dietary lever by a wide margin.
2. How much food you waste. Wasted food carries the entire footprint of producing it, plus its disposal.
3. Air-freighted produce. Small in tonnage, enormous per kilogram. Highly perishable out-of-season goods are the tell.
4. Heated glasshouse production out of season. A winter tomato grown under heat locally can exceed the footprint of one shipped from a sunnier country.
5. Everything else, including packaging and road transport, which are usually smaller than people expect. Plastic packaging that meaningfully extends shelf life can reduce total footprint by preventing waste, which is counterintuitive and reasonably well supported.

"Buy local" is a good rule for supporting local growers, for freshness, and for seasonal eating, all of which are legitimate reasons. It is a weak rule for carbon.

The bottom line

• Frozen is not a compromise. It is often nutritionally superior to fresh produce that has travelled, because it was processed hours after picking.
• Vitamin C and folate are the fragile ones. Minerals and fibre survive everything. Cooking destroys some nutrients and unlocks others, so eat both raw and cooked things.
• Three storage zones, not one: counter for tropical and ripening fruit, dark cupboard for potatoes and onions, fridge for greens, berries, and everything cut. Never refrigerate potatoes, bananas, or unripe tomatoes.
• Keep ethylene emitters away from ethylene-sensitive produce, and use a paper bag with a banana when you want the opposite.
• Waste is the biggest single household lever, on both money and carbon. Plan, store properly, read the date labels correctly, and cook the ugly things first.

Sources and notes

Fresh versus frozen nutrient comparisons follow Rickman et al., Journal of the Science of Food and Agriculture, 2007, Bouzari et al., Journal of Agricultural and Food Chemistry, 2015, and Leatherhead Food Research work. Cooking losses by nutrient and method follow USDA nutrient retention factors and standard food science texts. Carotenoid bioavailability increases with cooking and added fat follow Brown et al., American Journal of Clinical Nutrition, 2004, and related avocado and salad dressing studies. Storage temperatures and ethylene sensitivity follow USDA Handbook 66. Food safety guidance follows the UK Food Standards Agency and US FDA. The life-cycle emissions comparisons and the dominance of production over transport follow Poore and Nemecek, Science, 2018, and Ritchie's analyses at Our World in Data. Packaging's role in reducing waste follows WRAP and life-cycle assessment literature.

Open questions. Whether plastic packaging is net positive or negative environmentally depends on how much waste it actually prevents in practice, which is measured inconsistently. Household waste figures rely heavily on self-report and bin audits with known limitations.

👉 That completes the growing and buying half. Next: what actually happens after you swallow.

The Journey of a Bite

TL;DR. Follow an apple from your teeth to the toilet. Chewing and saliva start breaking down starch. The stomach is an acid tank that sterilises and dissolves, and absorbs almost nothing. The small intestine, which is where essentially all absorption happens, is about six metres long with an internal surface area of roughly a tennis court, and it works with pancreatic enzymes and bile that arrive on cue. The colon takes what your enzymes could not digest and hands it to thirty-odd trillion bacteria, who ferment it into compounds that feed your gut lining and talk to your immune system. Total transit: one to three days, of which the colon is most of it.

Key takeaways

• The stomach absorbs almost nothing except alcohol, water, and a few drugs. Its job is acid, mechanical breakdown, and pathogen control.
• The small intestine has an absorptive surface of roughly 200 to 300 square metres, produced by folds, villi, and microvilli stacking three levels of magnification.
• Nothing is absorbed as food. Starch must become glucose, protein must become amino acids, and fat must be emulsified, split, and reassembled, before anything crosses.
• Everything absorbed from the gut goes to the liver first, via the portal vein, which is why the liver is the body's chemical checkpoint and why it dominates drug metabolism.
• Transit time is 24 to 72 hours, and the colon accounts for most of it.

The bite

In short: Chewing multiplies surface area and mixes in the first enzyme; swallowing is a reflex you cannot fully control once it starts.

Take a bite of apple, about 20 grams. Roughly 17 g of that is water. Around 2.5 g is carbohydrate, mostly fructose, glucose, and sucrose, with some starch if the apple is not fully ripe. About 0.5 g is fibre, mostly cellulose, hemicellulose, and pectin. Traces of protein, almost no fat, some potassium, some vitamin C, and several hundred different polyphenols, most concentrated in the skin.

Chewing does two things. It multiplies surface area, so enzymes have somewhere to work. Halve a cube and you double its exposed area, and thorough chewing raises the available surface enormously. And it mixes in saliva, of which you make roughly 0.5 to 1.5 litres a day.

Saliva contains:

• Salivary amylase (ptyalin), which begins cutting starch into shorter chains. This is the reason bread tastes sweeter the longer you chew it: you are producing maltose in your mouth.
• Lingual lipase, a minor fat-splitting enzyme, more important in infants.
• Mucins, which lubricate.
• Lysozyme, lactoferrin, and immunoglobulin A, an antimicrobial set.
• Bicarbonate, which buffers acid and protects teeth.

Taste is happening simultaneously, and it is a chemical assay, not a decoration. Sweetness signals available energy, umami signals protein, salt signals electrolytes, sourness signals acid (unripe or fermented), and bitterness signals possible toxin. There is now good evidence for a fat taste (oleogustus) as well. Aroma matters more than taste: most of what you call flavour is volatile compounds reaching the olfactory receptors through the back of the nose, which is why food is bland with a blocked nose.

Swallowing is voluntary to start and reflexive after. The epiglottis folds over the airway, breathing stops for about a second, and the bolus enters the oesophagus. From there peristalsis, a travelling ring of muscular contraction, moves it down in about 8 seconds. Peristalsis is why you can swallow upside down.

The stomach: an acid tank that absorbs nothing

In short: Hydrochloric acid at pH 1.5 to 3.5 unfolds proteins, kills most microbes, and releases minerals from food; the stomach's main output is a controlled trickle into the small intestine.

The stomach is the most misunderstood organ in digestion. People assume it is where food is digested and absorbed. It is neither, mostly.

What it secretes, roughly 2 to 3 litres a day:

The acidity is genuinely extreme: pH 1.5 is roughly the acidity of battery acid diluted a little, and a hundred thousand times more acidic than blood. The stomach survives it because of a two-millimetre mucus-bicarbonate layer that maintains a near-neutral pH right at the cell surface. When that defence fails, through Helicobacter pylori infection or through NSAIDs suppressing the prostaglandins that maintain it, you get ulcers. That mechanism is why Chapter 66 spends time on stomach damage.

Mechanical work. The stomach's three muscle layers churn, grinding food against a closed pyloric sphincter until particles are under about 2 mm. The result is chyme, a semi-liquid acidic soup.

Gastric emptying is controlled and slow, and its speed determines a great deal:

Fat and protein in the small intestine trigger hormones (CCK and GLP-1) that slow the stomach down, which is why a fatty meal "sits heavy" and why adding fat or protein to carbohydrate flattens the resulting blood sugar rise. It is also the mechanism the GLP-1 weight drugs exploit (Chapter 76).

What the stomach absorbs: water, some minerals, alcohol (which is why drinking on an empty stomach hits fast, and why food genuinely slows intoxication), aspirin and a few other weak acids, and caffeine to a small degree. Everything else waits.

The small intestine: where almost everything happens

In short: About six metres long, folded and villous and microvillous, with an internal surface area of a couple of hundred square metres, and it does essentially all of the absorption.

The small intestine has three sections: the duodenum (about 25 cm, where the chemical work is set up), the jejunum (about 2.5 m, where most absorption happens), and the ileum (about 3.5 m, which mops up, plus the specific absorption of bile salts and vitamin B12).

The surface area trick

The intestine solves the absorption problem with three nested levels of folding:

1. Circular folds (plicae circulares), ridges running round the inside, multiply surface area about threefold.
2. Villi, finger-like projections about 1 mm long, roughly 20 to 40 per square millimetre, multiply it another tenfold.
3. Microvilli, microscopic projections on each absorptive cell forming the brush border, roughly a thousand per cell, multiply it another twentyfold.

The result is an absorptive surface commonly estimated at 200 to 300 square metres, about the area of a tennis court, packed into a tube you could coil into a bucket. That the surface is so large is why partial intestinal removal can be survived, and why diseases that flatten the villi (coeliac disease, most notably) cause malabsorption of everything at once.

Each villus contains a capillary network, which takes up sugars, amino acids, minerals, and water-soluble vitamins, and a lacteal, a lymphatic vessel that takes up fat.

The chemistry arrives on cue

When acidic chyme enters the duodenum, the duodenal lining releases hormones that summon the two support organs.

The pancreas delivers, through a duct into the duodenum:

plus a large volume of bicarbonate, which neutralises stomach acid so that the pancreatic enzymes, which work near pH 7 to 8, can function at all.

The protein-cutting enzymes are secreted as inactive precursors and activated only in the duodenum, by an enzyme anchored in the intestinal wall. This is not an accident: if trypsin activated inside the pancreas, it would digest the pancreas. That is exactly what happens in acute pancreatitis, and it is as bad as it sounds.

The gallbladder delivers bile, made continuously by the liver and stored and concentrated between meals. Bile contains no enzymes. Its active ingredients are bile salts, which are detergents: molecules with a water-loving end and a fat-loving end, so they surround fat droplets and break them into a fine emulsion. Since lipase can only work at the surface of a droplet, emulsifying a large droplet into a thousand small ones raises the working surface enormously. Without bile, fat digestion largely fails and fat-soluble vitamins (A, D, E, K) are not absorbed.

Bile is also an excretion route: bilirubin from broken-down red blood cells, excess cholesterol, and some drug metabolites leave the body this way. Bilirubin is what colours stool brown, which is why bile duct obstruction produces pale stools and yellow skin.

Final digestion happens on the cell surface

Some enzymes are not secreted into the cavity at all; they are embedded in the brush border membrane, so the last cut happens right at the point of absorption:

• Lactase splits lactose into glucose and galactose. Its production declines after weaning in most of humanity, which is lactose intolerance, and its persistence into adulthood is the mutation (Chapter 54).
• Sucrase splits sucrose into glucose and fructose.
• Maltase and isomaltase finish starch digestion.
• Peptidases finish protein digestion.

Absorption, molecule by molecule

Sugars. Glucose and galactose are pumped in actively by the SGLT1 transporter, using sodium as the driving force. Fructose enters passively via GLUT5, which is slower and has limited capacity; that limit is why large fructose loads can cause bloating and diarrhoea in susceptible people, and why fruit juice does this more readily than whole fruit.

A useful aside: the SGLT1 mechanism, glucose absorption coupled to sodium, is the basis of oral rehydration solution. Water follows sodium, sodium follows glucose, so a solution containing both is absorbed even by an inflamed gut that would reject plain water. That insight, published in the 1960s, has saved tens of millions of lives from diarrhoeal disease, and it is arguably the highest-impact single application of gut physiology in medicine.

Amino acids are taken up by a family of transporters, some sodium-coupled, and short di- and tripeptides are absorbed intact by the PepT1 transporter and broken down inside the cell. That peptide transporter is also how several oral antibiotics get in.

Fats take an entirely different route. Fatty acids and monoglycerides, packaged with bile salts into tiny aggregates called micelles, diffuse into the intestinal cell. Inside, they are reassembled into triglycerides, wrapped in a protein-and-phospholipid coat to make a chylomicron, and exported into the lacteal rather than into the blood. From the lymphatic system they eventually enter the bloodstream at the base of the neck.

That routing is important. Fat, and fat-soluble drugs and vitamins that ride with it, bypass the liver's first-pass metabolism, unlike everything else absorbed from the gut.

Water follows osmotic gradients: about 9 litres a day passes through the small intestine (2 litres consumed, 7 litres of secretions), and roughly 8 litres of it is reabsorbed before the colon.

The portal vein: everything goes to the liver first

In short: Blood from the entire gut drains to the liver before reaching the rest of you, which makes the liver the body's chemical checkpoint.

Blood leaving the intestines does not join the general circulation. It collects into the hepatic portal vein and goes straight to the liver.

The consequences are large:

• The liver sees every absorbed molecule first, at the highest concentration it will ever reach, and can act before anything reaches the brain or heart.
• First-pass metabolism: many drugs are substantially destroyed on this first pass through the liver, which is why oral doses are often much larger than injected ones, and why some drugs cannot be given orally at all. This is covered in Chapter 62.
• Glucose is buffered. After a meal the liver takes up a large share of incoming glucose and stores it as glycogen, blunting the rise reaching the rest of the body.
• Ammonia from protein metabolism and from gut bacteria is converted to urea before it can reach the brain. When the liver fails, it does not, and the result is hepatic encephalopathy.
• Fructose is handled almost entirely by the liver, unlike glucose, which is why very high fructose intake specifically stresses that organ (Chapter 11).

The large intestine: the fermentation vat

In short: What your enzymes could not digest arrives here for bacteria to ferment, producing short-chain fatty acids that feed your gut lining and influence your immune system.

By the time the residue reaches the ileocaecal valve, essentially all the digestible carbohydrate, protein, and fat is gone. What remains is fibre, resistant starch, some undigested protein, bile acids, sloughed-off intestinal cells, and roughly 1.5 litres of water.

The colon is about 1.5 metres long. It has three jobs.

1. Reabsorb water and electrolytes. Of the ~1.5 L arriving, about 1.3 L is reabsorbed, leaving roughly 100 to 200 mL in stool. When this fails, you get diarrhoea, and the reason severe diarrhoea kills is that the losses are fast and the volumes large.

2. Host the microbiome. Roughly 38 trillion bacterial cells, of hundreds of species, at densities of about 10¹¹ per gram of contents, in a near-oxygen-free environment. This is covered fully in Chapter 14, but the essential transaction is this: bacteria ferment fibre and resistant starch into short-chain fatty acids, principally acetate, propionate, and butyrate. Butyrate is the preferred fuel of the cells lining the colon, so a fibre-poor diet literally starves the gut lining. Fermentation also produces gas (hydrogen, carbon dioxide, methane), synthesises vitamin K and several B vitamins, and generates signals that shape immune development.

3. Form and store stool. Final composition is roughly 75 percent water and 25 percent solids, and of those solids, about a third is bacteria, a third is undigested fibre, and a third is sloughed cells, fats, and inorganic material.

Timing, end to end

In short: One to three days total, with wide normal variation, and the colon accounting for most of it.

Faster is not better and slower is not worse within that range. Very fast transit means poor absorption; very slow transit is associated with constipation and, in observational data, with higher levels of some bacterial metabolites of concern. Fibre, water, and physical activity all shorten transit; opioids, dehydration, low fibre, and inactivity lengthen it.

You can measure your own transit time with sweetcorn, beetroot, or blue food colouring, and it is a more informative experiment than most people expect.

The gut talks to the brain constantly

In short: The gut has its own nervous system, sends far more signals up than it receives down, and produces most of the body's serotonin.

The enteric nervous system contains something like 500 million neurons, more than the spinal cord, and can run peristalsis and secretion with no input from the brain. It is routinely called the second brain, which oversells it, but not by as much as you would think.

Communication runs mostly upward. Roughly 80 to 90 percent of the fibres in the vagus nerve carry signals from gut to brain, not the other way. The gut reports stretch, nutrient content, osmolarity, and irritation, and the brain responds with hunger, satiety, nausea, and mood.

Gut hormones are the other channel:

About 90 to 95 percent of the body's serotonin is made in the gut, by enterochromaffin cells, where it mostly regulates motility rather than mood. It does not cross into the brain. This is why "most of your serotonin is in your gut, so gut health controls your mood" is a real fact leading to an unsupported conclusion. The gut does influence mood, via the vagus nerve, immune signalling, and microbial metabolites, but not by exporting serotonin to the brain.

The bottom line

• Chewing and saliva start carbohydrate digestion; the stomach sterilises, unfolds proteins, and meters food out; the small intestine does essentially all the absorption; the colon ferments the leftovers.
• Nothing crosses the gut wall as food. Everything must be cut to its monomers first, and fat must additionally be emulsified, split, absorbed, reassembled, and shipped through the lymph.
• The small intestine's absorptive area is a couple of hundred square metres, produced by three nested levels of folding.
• Everything absorbed except fat goes to the liver first, which is why the liver dominates drug metabolism and why oral doses differ so much from injected ones.
• Total transit is one to three days, and the colon, where your bacteria live and work, is most of it.

Sources and notes

Digestive anatomy, secretion volumes, enzyme actions, transit times, and absorption mechanisms follow standard physiology texts; Guyton and Hall, Textbook of Medical Physiology, and Johnson, Gastrointestinal Physiology, are the usual references. Small intestinal surface area estimates have been revised downward from the often-quoted tennis court figure by Helander and Fandriks, Scandinavian Journal of Gastroenterology, 2014, who put it nearer 30 square metres; the larger figure remains in wide use and both are given here as an order of magnitude. SGLT1 glucose-coupled sodium absorption and the oral rehydration solution story follow the Lancet-era work of Hirschhorn, Cash, and colleagues in the 1960s and the WHO ORS programme. Gut hormone actions (ghrelin, CCK, GLP-1, PYY) follow endocrinology reviews. Enteric nervous system neuron counts and vagal afferent proportions follow Furness, The Enteric Nervous System, 2006. Gut serotonin production and its inability to cross the blood-brain barrier follow standard neuropharmacology.

Open questions. Small intestinal surface area is genuinely disputed by an order of magnitude depending on measurement method. How much the gut-brain axis influences mood in humans, as opposed to in animal models, remains an active and immature research area.

👉 Next: carbohydrates and blood sugar, the nutrient that generates the most argument and the most confusion.

Carbohydrates and Blood Sugar

TL;DR. All digestible carbohydrate ends up as glucose in your blood, and your body is intensely concerned with keeping that number in a narrow band. Insulin brings it down, and a handful of hormones bring it up. What determines the size and shape of the rise is not sugar-versus-starch but the physical form of the food: whether the starch is locked inside intact cell walls, how much fibre, fat, and protein arrived with it, and how fast the stomach empties. Fructose is metabolised differently from glucose, almost entirely in the liver, which is why the fructose in a bottle of soft drink and the fructose in an apple behave differently in practice.

Key takeaways

• Your entire bloodstream contains about 4 to 5 grams of glucose, roughly a teaspoon. The regulation is that tight.
• Glycaemic index describes a food eaten alone in a fixed 50 g carbohydrate dose, which is not how anyone eats. Glycaemic load, and the whole meal, matter more.
• The physical structure of food often matters more than its sugar content. Whole grains, intact pulses, and whole fruit produce far smaller glucose responses than the same carbohydrate milled or juiced.
• Fructose is handled by the liver, does not require insulin to be taken up, and at high intakes promotes fat production in the liver. In whole fruit the doses are small and arrive with fibre.
• Insulin resistance is the central metabolic problem behind type 2 diabetes, and muscle is its most important site.

What a carbohydrate is

In short: Chains of sugar units, from one unit to thousands, and the length and linkage determine everything.

Carbohydrates are built from simple sugar units. The three that matter:

Joined in pairs, they make disaccharides:

• Sucrose = glucose + fructose. Table sugar, from cane or beet. Chemically identical whatever the source.
• Lactose = glucose + galactose. Milk sugar.
• Maltose = glucose + glucose. From starch breakdown; the sugar in malt.

Longer chains make polysaccharides:

• Starch, the plant storage form, in two shapes: amylose (long straight chains, packs tightly, digests slowly) and amylopectin (highly branched, more exposed ends, digests fast). A high-amylose rice or potato produces a smaller glucose spike than a high-amylopectin one, which is a real varietal difference.
• Glycogen, the animal storage form, like a more heavily branched amylopectin, stored in liver (~100 g) and muscle (~400 g).
• Fibre, chains your enzymes cannot cut. Chapter 14.

Don't be confused: "sugar" on a label and "sugars" in nutrition science are different categories. A label's "of which sugars" counts all mono- and disaccharides, including the lactose naturally in milk and the fructose naturally in fruit. Free sugars (the WHO's category) means added sugars plus those in honey, syrups, and fruit juice, and excludes sugars inside intact fruit and in milk. The health guidance targets free sugars, which is why plain milk and whole fruit are not the problem the label numbers imply.

What the body does with it

In short: Glucose is absorbed, buffered by the liver, and driven into cells by insulin, with a whole hormonal apparatus preventing it from ever getting low.

Your bloodstream holds about 4 to 5 grams of glucose at any moment, in around 5 litres of blood at roughly 5 mmol/L (90 mg/dL). One teaspoon. A can of soft drink contains about 35 grams. Obviously it does not all arrive at once, and the reason it does not raise blood glucose sevenfold is that disposal is fast and aggressive.

Insulin, and what it actually does

Insulin is released by the beta cells of the pancreas in response to rising glucose, and to a lesser degree to amino acids and to gut hormones released just from food arriving. Its actions:

• Moves glucose into muscle and fat cells by causing GLUT4 transporters to move to the cell surface. Without insulin, muscle and fat are nearly impermeable to glucose. Brain, liver, and red blood cells are not insulin-dependent, which is why the brain keeps working in type 1 diabetes even while muscle starves.
• Tells the liver to store glucose as glycogen and to stop making new glucose.
• Stops fat breakdown. This is the underappreciated one. Insulin is the primary brake on releasing fat from fat cells, which is why it is fair to call it a storage hormone, and why the absence of insulin in untreated type 1 diabetes produces uncontrolled fat breakdown and ketoacidosis.
• Promotes protein synthesis.

The counter-regulatory hormones

The body treats low blood glucose as a far more urgent emergency than high, because the brain runs on it, so there are four hormones raising glucose and only one lowering it:

The asymmetry tells you what evolution optimised for. Chronically high glucose kills you over decades; low glucose kills you in minutes.

After a meal, hour by hour

Fasting glucose above 7.0 mmol/L (126 mg/dL) on two occasions, or HbA1c at or above 48 mmol/mol (6.5 percent), defines diabetes. The intermediate zone is prediabetes.

HbA1c deserves an explanation because it appears on every diabetes test result: glucose in blood spontaneously and irreversibly attaches to haemoglobin, at a rate proportional to concentration. Since red blood cells live about three months, the fraction of glycated haemoglobin is a running average of blood glucose over roughly 8 to 12 weeks, weighted toward recent weeks. You cannot fix it by behaving well for three days before the test.

Glycaemic index, and why it disappoints

In short: GI measures the glucose response to 50 g of available carbohydrate from a single food eaten alone, which is a useful research tool and a poor guide to meals.

Glycaemic index ranks foods by the area under the blood glucose curve over two hours, after eating a portion containing 50 g of available carbohydrate, relative to pure glucose at 100.

The list already shows the problem. Fructose has a very low GI and is not therefore a health food. Peanuts have a GI of 13 because they contain almost no carbohydrate. Ice cream has a lower GI than wholemeal bread, because of the fat.

Glycaemic load corrects for portion size:

$$\mathrm{GL} = \frac{\mathrm{GI} \times \text{grams of available carbohydrate per serving}}{100}$$

Watermelon has a high GI of about 76 and contains so little carbohydrate per slice that its glycaemic load is around 4, which is negligible. This is why "avoid high-GI fruits" is bad advice.

The deeper limitations:

• GI is measured on a single food eaten alone, fasted. Almost nobody eats that way.
• Adding fat, protein, acid, or fibre lowers the response substantially. Vinegar, notably, reduces the glucose response to a starchy meal by a measurable amount, and so does eating vegetables and protein before the starch in the same meal.
• Between-person variation is enormous. A landmark 2015 study by Zeevi and colleagues at the Weizmann Institute continuously monitored glucose in 800 people and found the same food produced wildly different responses in different individuals, with gut microbiome composition predicting a substantial part of the variation. One person spiked on bananas and not cookies; another the reverse.
• Ripeness, cooking, cooling, and processing shift GI hugely within a single food, as below.

The practical residue of GI is still worth having: prefer intact grains to flour, prefer whole fruit to juice, eat carbohydrate as part of a mixed meal, and cook pasta al dente. The number itself is not worth memorising.

Why physical form beats composition

In short: Starch locked inside intact plant cell walls digests far more slowly than the same starch milled into flour.

This is the most useful and least advertised fact in the whole subject.

Particle size. Wheat as intact kernels, as cracked grain, as coarse flour, and as fine white flour has an identical chemical composition per gram of starch and produces progressively larger glucose responses. Milling smashes the cell walls that would otherwise force enzymes to work slowly from the outside. This is why "wholemeal bread" made from finely milled wholemeal flour has a glycaemic index close to white bread: the fibre is present, the structure is gone. Stone-ground, coarse, and genuinely whole-grain products behave differently from fine wholemeal flour, and the label rarely tells you which you have.

Intact cell walls. Pulses (lentils, chickpeas, beans) have unusually tough cell walls that survive cooking, so a large share of their starch is physically shielded. This is why their glycaemic responses are so low, and why blended lentil soup produces a bigger response than whole lentils.

Resistant starch. Some starch escapes digestion entirely and reaches the colon as fibre:

RS3 is the interesting one domestically. Cooking and then cooling potatoes, rice, or pasta converts a portion of the starch into a resistant form, lowering the glycaemic response by roughly 10 to 30 percent depending on the food, and reheating does not fully undo it. Cold potato salad and next-day rice are genuinely metabolically different from the hot original. The effect is real and modest, and worth knowing rather than building a diet around.

Ripeness. A green banana is mostly resistant starch with a GI around 30; a heavily spotted one is mostly free sugars with a GI above 50. Same fruit, ten days apart.

Fructose: the one that is different

In short: Fructose is absorbed by a different transporter, does not need insulin, is metabolised almost entirely in the liver, and in large doses drives fat production there.

Glucose can be used by nearly every cell in the body. Fructose cannot: it goes to the liver, which converts it to glucose, glycogen, lactate, or fat.

The metabolic difference that matters is regulatory. Glucose metabolism has a control step (phosphofructokinase) that slows down when the cell has enough energy. Fructose metabolism enters the pathway below that control point, via fructokinase, which has no such brake. So a large fructose load is processed regardless of need, and the excess carbon is converted to fat in the liver by de novo lipogenesis.

At high habitual intakes, this contributes to:

• Non-alcoholic fatty liver disease, now the most common liver disease in the world.
• Raised blood triglycerides.
• Increased uric acid, because fructose metabolism consumes ATP rapidly and generates uric acid as a by-product, which is why sugary drinks are associated with gout.

But dose and context decide, and this is where the argument goes wrong in both directions. A medium apple contains roughly 10 g of fructose, wrapped in fibre, inside intact cells, absorbed over an hour, alongside water and potassium and polyphenols. A 500 mL bottle of soft drink contains roughly 25 to 30 g of fructose in solution, absorbed in minutes, with nothing else. Cohort studies consistently find whole fruit intake associated with lower risk of type 2 diabetes while fruit juice and sugary drinks are associated with higher risk. The molecule is the same. Everything else differs.

High-fructose corn syrup is worth a specific note because it attracts disproportionate blame. HFCS-55, used in soft drinks, is about 55 percent fructose and 42 percent glucose. Sucrose is 50 percent fructose and 50 percent glucose, and is split into exactly those in the gut. The two are close enough that no convincing metabolic difference between them has been demonstrated in controlled human studies. HFCS is a problem because it made sugar cheap and ubiquitous in the US food supply, not because it is a uniquely toxic molecule.

Insulin resistance and type 2 diabetes

In short: Cells stop responding well to insulin, the pancreas compensates by making more, and eventually cannot keep up.

Insulin resistance means a given amount of insulin produces less glucose uptake than it should. The pancreas responds by making more insulin, so blood glucose stays normal for years while insulin levels climb. That compensated phase can last a decade or more, is invisible on a standard fasting glucose test, and is where most of the damage begins. Eventually beta cells fail to keep up, glucose rises, and type 2 diabetes is diagnosed.

Where resistance happens:

• Muscle is quantitatively the most important, taking up the majority of post-meal glucose. Muscle insulin resistance appears early and is strongly improved by exercise.
• Liver resistance means the liver keeps producing glucose when it should stop, which is why fasting glucose rises.
• Fat tissue resistance means fat keeps being released into the blood, raising free fatty acids, which worsens resistance elsewhere.

What drives it, in rough order of evidence:

1. Excess fat stored in the wrong places, particularly inside liver and muscle cells and around the viscera. Ectopic fat, not fat as such: this is why some people with a normal BMI are insulin resistant and some with obesity are not.
2. Physical inactivity. Muscle contraction moves GLUT4 to the cell surface without insulin, which is why a single walk after a meal measurably lowers the glucose response and why exercise improves insulin sensitivity within days.
3. Chronic energy surplus.
4. Genetics. Substantial, with far higher risk in South Asian, East Asian, African, and Indigenous American ancestries, at lower BMIs.
5. Sleep deprivation. A few nights of short sleep measurably reduces insulin sensitivity in healthy volunteers. This is a genuinely underrated factor.
6. Some drugs, notably corticosteroids and some antipsychotics.

The two most effective interventions are weight loss and exercise, and the evidence for both is unusually strong. The DiRECT trial showed that substantial weight loss through a structured low-calorie programme put a large share of people with type 2 diabetes into remission at one year, with remission strongly related to how much weight was lost. Exercise improves insulin sensitivity even without weight loss. Neither is easy; both work.

How much carbohydrate should you eat?

In short: There is no essential dietary requirement for carbohydrate, and that fact is much less useful than it sounds.

Unlike essential fatty acids and essential amino acids, there is no essential carbohydrate: your liver can make all the glucose your brain needs from amino acids and glycerol. Very low carbohydrate diets are therefore physiologically possible, and ketogenic diets have genuine medical uses, notably in drug-resistant epilepsy, where the evidence is decades old and solid.

For general health, the honest state of the evidence:

• Both very low and very high carbohydrate intakes are associated with higher mortality in large cohort studies, with the lowest risk around 50 to 55 percent of energy from carbohydrate. The ARIC study and its associated meta-analysis is the most cited version of this U-shaped finding. Observational, with all that implies.
• What replaces the carbohydrate matters more than the reduction. In those same analyses, low-carbohydrate diets replacing carbohydrate with animal fat and protein were associated with higher mortality; those replacing it with plant fat and protein were associated with lower.
• Low-carbohydrate diets produce faster early weight loss, largely water (each gram of glycogen holds about 3 g of water), and at 12 months the difference against other diets in controlled trials is small. Adherence predicts outcome better than macronutrient ratio.
• Type of carbohydrate dominates. Whole grains, pulses, fruit, and vegetables are associated with lower disease risk across essentially every large cohort. Refined grains and sugary drinks with higher. This finding is far more consistent than anything about total quantity.

The WHO recommends free sugars below 10 percent of energy, with a conditional suggestion of below 5 percent, which is about 25 g a day for an average adult. One standard can of soft drink exceeds that.

The bottom line

• All digestible carbohydrate becomes glucose. Your blood holds about a teaspoon of it at a time, defended by one hormone that lowers it and four that raise it.
• Glycaemic index is a laboratory measure of a food eaten alone; glycaemic load and, far more, the composition of the whole meal are what matter in practice. Individual variation is large.
• The physical structure of a food matters more than its sugar content. Milling destroys the structure that slows digestion, which is why fine wholemeal flour behaves much like white.
• Fructose is metabolised in the liver without a regulatory brake, so high loads promote liver fat and raise uric acid. Whole fruit does not deliver those loads; drinks do.
• Insulin resistance is the central process behind type 2 diabetes, muscle is its most important site, and exercise and weight loss are the two interventions with the strongest evidence.

Sources and notes

Carbohydrate chemistry, insulin action, and counter-regulatory hormones follow standard endocrinology texts. Glycaemic index and load values follow the international tables compiled by Atkinson, Foster-Powell, and Brand-Miller, Diabetes Care, 2008 and later editions. Individual variation in glycaemic response is Zeevi et al., Cell, 2015, from the Weizmann Institute. Resistant starch classification follows Englyst's typology. Particle size effects on glycaemic response follow Jenkins' and Wolever's work from the 1980s onward. Fructose metabolism, de novo lipogenesis, and uric acid generation follow Lustig's and Havel's reviews, with the sugar-sweetened beverage epidemiology from Malik and Hu. HFCS versus sucrose equivalence follows controlled feeding comparisons summarised by White, American Journal of Clinical Nutrition, 2008. The U-shaped carbohydrate-mortality relationship is Seidelmann et al., The Lancet Public Health, 2018, from ARIC. Type 2 diabetes remission is the DiRECT trial, Lean et al., The Lancet, 2018. WHO free sugars guidance is the 2015 guideline.

Open questions. Whether fructose is metabolically harmful at intakes achievable from whole food, rather than from beverages, is not settled. The U-shaped mortality curve for carbohydrate intake is observational and its shape depends heavily on what replaced the carbohydrate, which the studies handle inconsistently.

👉 Next: protein, the nutrient with the clearest practical case and the noisiest supplement industry.

Protein

TL;DR. Protein is the only macronutrient you cannot store, so a supply has to keep arriving. Nine of the twenty amino acids are essential, meaning you must eat them, and the practical question is not whether a food is a "complete protein" but whether your whole day's eating covers all nine, which almost any varied diet does. The official minimum of 0.8 g per kilogram of body weight prevents deficiency; the amount that optimises muscle retention, especially in older people and people losing weight, is higher, around 1.2 to 1.6 g/kg. Protein is the most satiating macronutrient by a clear margin, which is the most practically useful thing about it.

Key takeaways

• There is no protein storage depot. Excess is burned or converted; shortfall is met by breaking down muscle.
• Nine essential amino acids must come from food. Leucine specifically triggers muscle protein synthesis.
• The RDA of 0.8 g/kg is a minimum to prevent deficiency, not an optimum. Older adults and people in a calorie deficit do measurably better at 1.2 to 1.6 g/kg.
• Protein is the most satiating macronutrient, and the thermic effect of processing it costs 20 to 30 percent of its own calories.
• "Complete protein" is a per-meal concern that turns out not to matter if the day's total is varied. Combining rice and beans at the same sitting is unnecessary.

What protein is

In short: Chains of amino acids folded into specific three-dimensional shapes; the shape is the function.

A protein is a chain of amino acids linked by peptide bonds and folded into a precise shape. There are twenty amino acids in the human genetic code, and the sequence determines the fold, and the fold determines what the protein does.

Proteins are not primarily a fuel. They are the working machinery:

You contain roughly 10 to 12 kg of protein as an adult, and you turn over about 250 to 300 grams of it every day, recycling most of the amino acids. Only about 40 to 60 g a day is actually lost and needs replacing, which is why the requirement is far lower than the turnover.

Essential and non-essential

Nine amino acids cannot be made by the human body and must be eaten:

Histidine, isoleucine, leucine, lysine, methionine, phenylalanine, threonine, tryptophan, valine.

The mnemonic that works: PVT TIM HaLL (phenylalanine, valine, threonine, tryptophan, isoleucine, methionine, histidine, leucine, lysine).

Several more are conditionally essential: normally made by the body, but required from food during illness, injury, rapid growth, or prematurity. Arginine, cysteine, glutamine, glycine, proline, and tyrosine are in this group.

Leucine deserves separate mention because it is not just a building block. It acts as a signal: leucine activates the mTOR pathway, which switches on muscle protein synthesis. A meal needs a leucine threshold of roughly 2 to 3 g to trigger a robust synthesis response, which is one reason a small snack of protein does less than a proper serving, and why leucine content is the reason whey and animal proteins outperform some plant proteins gram for gram.

Protein quality, and why it matters less than the phrase suggests

In short: Quality scores measure the essential amino acid profile and digestibility; the practical implication is much weaker than the marketing.

Two scoring systems are in use.

PDCAAS (Protein Digestibility Corrected Amino Acid Score) is the older standard, capped at 1.0, which compresses the top of the scale so that several very different proteins all score a perfect 1.0.

DIAAS (Digestible Indispensable Amino Acid Score) is the newer FAO-recommended approach, measured at the end of the small intestine rather than in faeces, and not capped, so it discriminates better.

The pattern is real: animal proteins generally score higher because their amino acid profile more closely matches human requirements and they are more digestible.

The limiting amino acid concept explains most of the plant scores. Cereals (wheat, rice, maize) are low in lysine. Pulses (beans, lentils, peas) are low in methionine and cysteine. Each covers the other's gap, which is the biochemical basis for why nearly every traditional cuisine on Earth independently arrived at a grain-plus-pulse staple: rice and dal, beans and tortillas, hummus and pita, rice and beans, lentils and bread.

The complementary protein myth, corrected. Frances Moore Lappé's influential 1971 book Diet for a Small Planet argued you must combine complementary plant proteins at the same meal. She retracted that specific claim in later editions, and the retraction is correct. The body maintains a free amino acid pool and recycles a large quantity of protein daily, so amino acids eaten at breakfast are available to combine with those eaten at dinner. The American Dietetic Association and its successors have stated for decades that combining at each meal is unnecessary. Eating varied plant protein across the day is sufficient.

What remains true for plant-based eaters: total protein needs to be somewhat higher, perhaps 10 to 20 percent, to account for lower digestibility, and lysine deserves attention, which means including pulses regularly rather than living on grains and nuts.

How much protein do you actually need?

In short: 0.8 g/kg prevents deficiency; 1.2 to 1.6 g/kg is better for muscle preservation, ageing, and weight loss; above about 2 g/kg there is no further benefit for most people.

The RDA of 0.8 g per kg of body weight per day is derived from nitrogen balance studies and is defined as the amount sufficient for 97.5 percent of healthy adults to avoid losing body protein. It is a floor, not a target, and this distinction is routinely lost.

For an 80 kg adult, that spans 64 g/day at the RDA to 128 g/day at 1.6 g/kg. In food terms, 128 g is roughly a chicken breast, a tin of chickpeas, two eggs, a pot of Greek yoghurt, and the protein incidentally present in bread and vegetables. It is achievable without powders.

Anabolic resistance is the reason the recommendation rises with age, and it is one of the more important practical findings in nutrition of the last twenty years. Older muscle needs a larger dose of protein in a single sitting to trigger the same synthesis response: where a young adult may maximise the response at around 20 g of high-quality protein, an older adult may need 30 to 40 g. Combined with the fact that older people often eat least protein at breakfast, this matters for sarcopenia, the age-related loss of muscle that predicts falls, frailty, and loss of independence. Spreading protein across three meals rather than loading it all at dinner is the practical advice, and it has trial support.

Protein and satiety: the most useful practical fact

In short: Gram for gram, protein reduces subsequent eating more than carbohydrate or fat, and it costs more calories to process.

The thermic effect of food is the energy spent digesting and processing it:

So 100 kcal of protein nets about 75 usable kcal, against about 98 for fat. That is real and too small to build a diet on.

The satiety effect is larger. Controlled feeding studies consistently rank protein above carbohydrate above fat for suppressing subsequent intake, mediated by gut hormones (PYY, GLP-1, CCK) and by slower gastric emptying. In practice, high-protein diets tend to produce spontaneous calorie reduction: people eat less without being asked to. That is the main reason protein is useful for weight management, more than any thermic accounting.

The 2011 Australian "protein leverage" work by Simpson and Raubenheimer proposed a stronger version: that humans regulate protein intake in absolute terms and will keep eating until protein needs are met, so a diet diluted in protein (which describes much ultra-processed food) drives overconsumption of everything else. The hypothesis is contested and the experimental support is genuinely interesting.

Protein sources, compared

In short: Animal foods deliver more protein per calorie with better amino acid profiles; plant foods deliver protein with fibre and without saturated fat.

Two things stand out. Nuts and nut butters look protein-rich per 100 g and are not protein-dense per calorie, because they are mostly fat. And bread, pasta, rice, and potatoes contribute a substantial share of most people's protein simply because of how much of them gets eaten, which is why frank protein deficiency is very rare in any country with adequate calories.

Does high protein damage your kidneys?

In short: In healthy kidneys, no. In existing kidney disease, yes, and protein restriction is a standard part of treatment.

This is one of the most persistent myths and it has a real fact at its centre.

In people with normal kidney function, controlled studies of high-protein diets, including intakes above 2 g/kg maintained for months in resistance-trained people, show no adverse effect on glomerular filtration rate, albumin excretion, or other markers. The increase in GFR seen with higher protein is a normal adaptive response, the same way muscles enlarge with use, not a sign of damage. Major reviews and position stands from sports nutrition and nephrology bodies concur.

In people with chronic kidney disease, protein restriction genuinely slows progression and is a mainstream part of management, typically 0.6 to 0.8 g/kg under dietitian supervision. This is where the myth comes from: an appropriate treatment for a diseased organ has been generalised into a warning for healthy people.

Two genuine cautions do apply. Hydration: metabolising more protein produces more urea, which needs water to excrete, so higher intakes need more fluid. And existing kidney stones: high animal protein raises urinary calcium, uric acid, and oxalate, and lowers citrate, all of which favour stone formation in people who already form them.

Timing, distribution, and the supplement industry

In short: Total daily protein matters most; even distribution across meals is a real second-order effect; the "anabolic window" was oversold.

The anabolic window. For years, the standard advice was to consume protein within 30 to 60 minutes of training. Subsequent work, notably Brad Schoenfeld's meta-analyses, found the effect largely disappears once total daily protein is controlled for. If you have eaten protein within a few hours either side of training, the timing is not the limiting factor.

Distribution does appear to matter, modestly. Studies comparing an even split (roughly 30 g at each of three meals) against a skewed pattern (10 g, 15 g, 65 g) with identical totals find better muscle protein synthesis with the even split. The effect is small in young people and larger in older ones.

Before bed. Slow-digesting casein before sleep raises overnight synthesis and has reasonable trial support. It is a marginal gain, not a foundation.

Protein powders are food, conveniently packaged, and there is nothing wrong with them. Whey is a by-product of cheese making, fast-absorbing, and highest in leucine. Casein is slower. Soy, pea, and rice proteins work; blended plant proteins cover amino acid gaps better than single sources. The genuine cautions are that supplements are regulated far more loosely than food (Chapter 81), that independent testing has repeatedly found heavy metal contamination in some plant-based protein powders, and that most people meeting their needs from food gain nothing from adding powder.

BCAA supplements (branched-chain amino acids: leucine, isoleucine, valine) are one of the clearest cases of a supplement whose evidence does not support its sales. Taking three of the nine essential amino acids in isolation does not build muscle, because synthesis requires all of them; studies find BCAAs alone raise muscle protein synthesis far less than a complete protein containing the same leucine. If you are eating enough protein, they add nothing.

Deficiency, and the two forms of it

Frank protein deficiency is rare where calories are adequate, and severe where they are not.

Kwashiorkor presents with oedema (the swollen belly familiar from famine photographs), skin and hair changes, and an enlarged fatty liver, classically attributed to protein deficiency with adequate calories. Marasmus is total energy deficiency: extreme wasting without oedema. The clean protein-versus-energy distinction has been questioned for decades, and current understanding involves infection, micronutrient deficiency, aflatoxin exposure, and the gut microbiome as well.

The subtler and far more common problem in wealthy countries is sarcopenia: the progressive loss of muscle mass and strength with age, beginning in the fourth decade and accelerating after 60, driven by inactivity, anabolic resistance, and inadequate protein. It is a leading cause of falls, fractures, and loss of independence, and it is substantially preventable with resistance exercise plus adequate protein. That combination, not either alone, is what the trials support.

The bottom line

• Protein cannot be stored, so intake must be continuous. Nine amino acids are essential and leucine additionally acts as the switch that turns on muscle synthesis.
• The 0.8 g/kg RDA is a deficiency floor. 1.2 to 1.6 g/kg suits most people who care about body composition, ageing, or weight loss; athletes go higher.
• Protein quality differences between animal and plant sources are real and easily handled by variety. You do not need to combine complementary proteins within a single meal.
• Protein is the most satiating macronutrient and costs the most to process, which is why higher-protein diets tend to reduce total intake without conscious restriction.
• High protein does not damage healthy kidneys. It is restricted in existing kidney disease, which is where the myth comes from.

Sources and notes

Amino acid requirements, nitrogen balance, and the 0.8 g/kg RDA derivation follow the FAO/WHO/UNU expert consultation and the US Institute of Medicine Dietary Reference Intakes. PDCAAS and DIAAS scores follow FAO's 2013 report on dietary protein quality evaluation. The complementary protein correction follows Lappe's own revision in later editions of Diet for a Small Planet and successive American Dietetic Association position papers. Anabolic resistance and per-meal leucine thresholds follow Wolfe's and Phillips' work. Optimal intake for resistance training follows Morton et al., British Journal of Sports Medicine, 2018, which found benefit plateauing near 1.6 g/kg. Thermic effect and satiety ranking follow Halton and Hu's review. Protein leverage is Simpson and Raubenheimer, Obesity Reviews, 2005. The absence of kidney harm in healthy people follows Devries et al., Journal of Nutrition, 2018, and ISSN position stands. Sarcopenia prevention evidence follows the PROT-AGE and ESPEN expert group recommendations.

Open questions. Whether intakes above roughly 1.6 g/kg confer additional benefit in trained individuals is contested, and individual variation may be larger than the group means suggest. Long-term cardiovascular effects of sustained very high protein intake, and whether the source matters more than the amount, are not well characterised.

👉 Next: fats, where forty years of dietary advice went through a genuine and instructive reversal.

Fats

TL;DR. Fat is a family, not a substance, and the members behave very differently. Two fatty acids are essential and you will die without them. Trans fats from industrial hydrogenation are the one dietary component with an unambiguous case for elimination, and they have largely been eliminated. Saturated fat is the long-running argument: replacing it with polyunsaturated fat lowers heart disease risk in the trials, replacing it with refined carbohydrate does not, and that nuance is where most of the public confusion lives. Dietary cholesterol turned out to matter much less than assumed for most people, which is why egg guidance changed.

Key takeaways

• Two essential fatty acids: linoleic acid (omega-6) and alpha-linolenic acid (omega-3). Everything else your body can make.
• Trans fats are uniquely harmful and have been banned or effectively removed from the food supply in most of the world since around 2018 to 2023.
• What you replace saturated fat with determines the outcome. Polyunsaturated fat lowers cardiovascular risk; refined carbohydrate does not.
• Dietary cholesterol raises blood cholesterol far less than saturated fat does in most people, which is why per-day egg limits were dropped.
• Omega-3 from oily fish has good evidence for triglycerides and moderate evidence for cardiovascular outcomes; supplements have performed worse than the fish in trials.

The chemistry, in the minimum necessary detail

In short: A fat is three fatty acids on a glycerol backbone, and the double bonds in those fatty acids determine everything about how the fat behaves.

Dietary fat is mostly triglyceride: a glycerol molecule with three fatty acids attached. A fatty acid is a chain of carbon atoms with a carboxyl group at one end.

Two features of that chain determine everything:

Chain length:

Saturation, meaning how many double bonds the chain contains:

The geometry is the whole story. A double bond in the natural cis configuration puts a kink in the chain, so the molecules cannot stack neatly, so the fat is liquid and its membranes are fluid. Saturated chains are straight, stack tightly, and are solid. Trans double bonds have the hydrogen atoms on opposite sides, which removes the kink, so a trans fat is chemically unsaturated and physically behaves like a saturated one. That is precisely why industry made them.

Omega-3 and omega-6

The "omega" number counts carbons from the far end of the chain to the first double bond. This matters because humans lack the enzymes to put double bonds at the omega-3 or omega-6 positions, which is exactly why those two families are essential.

The conversion of plant ALA into EPA and especially DHA is poor in humans: commonly cited figures are 5 to 10 percent conversion to EPA and under 1 to 5 percent to DHA, with conversion somewhat better in women (an oestrogen effect, presumably related to pregnancy). This is the single most important practical fact for vegan and vegetarian diets, and the answer is algal oil, which is where fish get their DHA in the first place.

What your body does with fat

In short: Fat is the energy store, the membrane material, the hormone precursor, and the carrier for four vitamins.

• Energy storage. 9 kcal per gram, stored nearly water-free, against 4 kcal per gram for glycogen stored with three times its weight in water. A person carrying 15 kg of body fat has around 135,000 kcal of stored energy against roughly 2,000 kcal of glycogen. This density is why we store energy as fat and not as starch.
• Cell membranes. Every cell is bounded by a phospholipid bilayer, and the fatty acid composition of your diet measurably changes membrane fluidity and function.
• Hormone and signal precursors. Cholesterol is the starting material for testosterone, oestrogen, cortisol, aldosterone, and vitamin D. Arachidonic acid and EPA are the precursors of eicosanoids: prostaglandins, thromboxanes, leukotrienes, which govern inflammation, clotting, and pain. This is the pathway that aspirin and ibuprofen block (Chapter 66).
• Absorbing vitamins A, D, E, and K, which require fat in the same meal. A fat-free salad genuinely delivers less of the vitamins in it, and studies adding avocado or oil to salad show several-fold increases in carotenoid absorption.
• Insulation and organ protection.

Trans fats: the settled case

In short: Industrially produced trans fat raises LDL and lowers HDL simultaneously, has no known safe level, and has been removed from most of the world's food supply.

Partial hydrogenation was invented to turn cheap liquid vegetable oil into a solid, shelf-stable fat that behaves like butter or lard. Hydrogen is bubbled through oil over a metal catalyst, and stopping the reaction partway converts some double bonds to the trans configuration. The result was cheap, stable, and excellent for baking and frying, and it went into margarine, shortening, biscuits, pastry, and fried food from the 1950s onward.

The health effects, established through the 1990s, are unusually clear-cut:

• Raises LDL cholesterol (the harmful one).
• Lowers HDL cholesterol (the protective one). Almost nothing else in the diet does both.
• Raises lipoprotein(a), an independent risk factor.
• Promotes inflammation and endothelial dysfunction.
• Meta-analyses associate a 2 percent of energy increase in trans fat with roughly a 20 to 25 percent increase in coronary heart disease risk, which is a large effect for a small intake.

The regulatory response was decisive. Denmark limited industrial trans fat in 2003, the US FDA removed the "generally recognised as safe" status for partially hydrogenated oils in 2015 with compliance by 2018 to 2021, the EU set a 2 percent limit from 2021, and the WHO has driven an elimination programme worldwide. It is one of the most successful public health nutrition interventions ever run, and modelling and observational data attribute substantial reductions in cardiovascular events to it.

Two residual notes. Ruminant trans fats occur naturally in small amounts in dairy and beef, produced by bacteria in the rumen, and at typical intakes they do not appear to carry the same risk. And labelling loopholes persist in some jurisdictions where products under a threshold per serving may declare zero; the tell is "partially hydrogenated oil" in the ingredients list.

Saturated fat: the argument, honestly

In short: Saturated fat raises LDL, LDL causes atherosclerosis, and replacing saturated fat with polyunsaturated fat reduces cardiovascular events. Replacing it with refined carbohydrate does not, which is the source of most of the confusion.

This is the longest-running argument in nutrition and it deserves to be laid out carefully rather than settled by assertion.

What is not in serious dispute:

• Saturated fat intake raises LDL cholesterol relative to unsaturated fat. This is reproducible in controlled feeding studies.
• LDL is causal in atherosclerosis. The evidence from genetics (Mendelian randomisation across many variants), from drug trials across multiple mechanisms, and from familial hypercholesterolaemia is about as strong as causal evidence gets in human biology.
• Randomised trials replacing saturated fat with polyunsaturated fat reduce cardiovascular events. The Cochrane review of reducing saturated fat found about a 17 percent reduction in cardiovascular events, driven by the trials where the replacement was polyunsaturated fat.

What is genuinely contested:

• The replacement determines the result. Substitution analyses consistently show: saturated fat replaced with polyunsaturated fat lowers risk; with monounsaturated fat, probably lowers risk; with whole grains, lowers risk; with refined carbohydrate or sugar, no benefit and possibly harm. Advice to "cut fat" in the 1980s and 90s often produced the last of these, and the resulting low-fat high-sugar products are a fair criticism of how the guidance was implemented.
• Not all saturated fats behave alike. Stearic acid (18 carbons, prominent in cocoa butter) is roughly neutral on LDL because much of it is converted to oleic acid. Lauric, myristic, and palmitic acids raise LDL more.
• The food matters, not just the fatty acid. This is the strongest recent finding. Cheese and yoghurt, despite substantial saturated fat, are not associated with increased cardiovascular risk in cohort studies, and fermented dairy is sometimes associated with lower risk. Processed meat, with similar fatty acids, is consistently associated with higher risk. The food matrix (calcium, fermentation products, protein structure, and what else comes with it) appears to matter, and mechanisms are still being worked out.
• The observational meta-analyses (Siri-Tarino 2010, Chowdhury 2014) that found no association between saturated fat intake and cardiovascular disease are real and were widely reported as an exoneration. The standard critique is that observational studies of a single nutrient cannot identify what it was replaced by, and that measurement error in dietary recall is large. Both the findings and the critique are legitimate.

The current mainstream position, from the WHO, the American Heart Association, and most national bodies: keep saturated fat under about 10 percent of energy, and replace it primarily with unsaturated fat from plants, not with refined carbohydrate. That position is defensible on the trial evidence. Positions substantially more permissive or more restrictive than that require reading past what the trials show.

Coconut oil is worth a paragraph because it is marketed as an exception. It is roughly 80 to 90 percent saturated, mostly lauric acid, and controlled trials show it raises LDL substantially compared with unsaturated oils, more than butter in some studies. The medium-chain triglyceride argument is weak: the MCT research uses purified C8 and C10 fats, while coconut oil's lauric acid (C12) behaves largely like a long-chain fat. It is a fine flavouring ingredient and there is no basis for treating it as a health food.

Dietary cholesterol: the reversal

In short: For most people, dietary cholesterol has a modest effect on blood cholesterol, because the body compensates by making less.

Cholesterol is not a fat; it is a sterol, and it is essential. Your liver makes roughly 1 to 1.5 g a day, several times more than a typical diet supplies. When dietary intake rises, endogenous production falls, which is a homeostatic loop most people's bodies run effectively.

Consequently:

• The US Dietary Guidelines dropped the 300 mg/day limit in 2015, stating that cholesterol is "not a nutrient of concern for overconsumption." Several other national bodies followed.
• Eggs are the food this most affects. Large cohort studies mostly find no association between moderate egg consumption and cardiovascular events in people without diabetes. Findings in people with diabetes are less reassuring and less consistent.
• Roughly a quarter to a third of people are "hyper-responders" whose blood cholesterol does respond noticeably to dietary intake. If your LDL is high and you eat a lot of eggs, it is worth testing the effect of changing that rather than assuming either way.
• Saturated fat still raises LDL more than dietary cholesterol does, which is the reason the two were conflated for so long: the foods often overlap.

The lipid panel, decoded

In short: LDL carries cholesterol out to tissues and deposits it in artery walls; HDL carries it back; triglycerides are circulating fat.

Cholesterol does not dissolve in blood, so it travels in lipoproteins, particles with a fat core and a protein shell.

Two corrections worth making. "Good and bad cholesterol" is a simplification that has aged badly for HDL: raising HDL with drugs (CETP inhibitors, niacin) has repeatedly failed to reduce events in trials, so high HDL appears to be a marker of something healthy rather than a cause of it. Triglycerides respond dramatically to alcohol, sugar, and refined carbohydrate, far more than to dietary fat, which surprises people.

Omega-3 and omega-6, and the ratio argument

In short: Omega-3 from oily fish has solid evidence for triglycerides and reasonable evidence for cardiovascular events; the popular omega-6-is-inflammatory argument is not well supported.

The established benefits of marine omega-3 (EPA and DHA):

• Triglyceride lowering, dose-dependent and substantial at prescription doses (2 to 4 g a day can reduce triglycerides by 25 to 30 percent).
• DHA is structurally essential in retina and brain, and required in pregnancy and infancy for neural development.
• Cardiovascular outcomes: eating oily fish is consistently associated with lower cardiovascular mortality in cohorts. Supplement trials have been mixed: several large ones (VITAL, ASCEND) found no benefit for primary prevention, while REDUCE-IT found a substantial reduction using high-dose purified EPA (icosapent ethyl) in high-risk patients, though its mineral oil placebo has been criticised. The honest summary: eat the fish; supplements are a weaker substitute with an unsettled evidence base.
• Anti-inflammatory effects, modest joint symptom benefit in rheumatoid arthritis.

The omega-6 argument. A widely repeated claim holds that modern diets have an omega-6-to-omega-3 ratio of 15:1 or 20:1 against an ancestral 1:1 to 4:1, that omega-6 is pro-inflammatory because arachidonic acid makes inflammatory eicosanoids, and that seed oils are therefore harmful.

The evidence does not support the strong version:

• Increasing dietary linoleic acid does not measurably raise arachidonic acid in human tissue, because the conversion is tightly regulated.
• Controlled trials of increased linoleic acid do not show increases in inflammatory markers.
• Cohort studies find higher linoleic acid intake, and higher linoleic acid measured in blood or tissue (which avoids dietary recall error), associated with lower cardiovascular risk and lower type 2 diabetes risk.
• Major bodies, including the American Heart Association after a formal advisory review, conclude that omega-6 polyunsaturated fat lowers cardiovascular risk when it replaces saturated fat.

What is fair to say: increasing omega-3 is more clearly beneficial than reducing omega-6, and the two are not equivalent levers. The seed oil panic that circulates online substantially exceeds the evidence. What is genuinely true is that seed oils are the fat in most fried and ultra-processed food, so intake correlates with a dietary pattern that is bad for other reasons, and heavily reused frying oil does accumulate harmful oxidation products (Chapter 91).

Cooking fats and smoke points

In short: Match the fat to the temperature, prefer stable fats for high heat, and never reuse oil that has darkened or smells acrid.

Smoke point matters less than oxidative stability, which is set by how many double bonds the fat has. Polyunsaturated oils oxidise fastest, producing aldehydes and other compounds associated with harm. This is why flaxseed oil must never be cooked, why oils should be stored dark and cool, and why deep-fry oil should be discarded once it darkens, foams, smokes early, or smells acrid.

Extra virgin olive oil deserves the correction: it is frequently described as unsuitable for cooking, and testing shows it is one of the more stable oils under heat, because its high monounsaturated content and high polyphenol and vitamin E content resist oxidation. It is fine for normal cooking, and using it for deep-frying is mainly a question of cost and flavour.

The bottom line

• Fat is a family. Chain length and double-bond geometry decide how each member behaves.
• Trans fats are the one clearly settled case: harmful at any realistic intake, and successfully removed from most of the world's food supply.
• Saturated fat raises LDL and LDL causes atherosclerosis. What you replace saturated fat with decides whether you benefit: unsaturated fat and whole grains yes, refined carbohydrate no.
• Dietary cholesterol matters much less than once believed, because the body compensates. Eggs are fine for most people.
• Eat oily fish for omega-3, or algal oil if you do not eat fish. The seed oil panic is not supported by the trial or biomarker evidence; the omega-3 case is much stronger than the anti-omega-6 case.

Sources and notes

Fatty acid chemistry and metabolism follow standard biochemistry texts. Trans fat effects on lipids and cardiovascular risk follow Mozaffarian et al., New England Journal of Medicine, 2006, and the WHO REPLACE programme documentation. The saturated fat replacement evidence follows the Cochrane review by Hooper et al., 2020, and the substitution analyses of Li, Hu and colleagues. The observational meta-analyses that found no association are Siri-Tarino et al., 2010, and Chowdhury et al., Annals of Internal Medicine, 2014, with their standard critiques. The food matrix argument for dairy follows Astrup and Thorning's work and the 2020 Advances in Nutrition consensus. Coconut oil trials are meta-analysed in Neelakantan et al., Circulation, 2020. Dietary cholesterol guidance follows the 2015 US Dietary Guidelines Advisory Committee report. LDL causality follows the European Atherosclerosis Society consensus statement, Ference et al., European Heart Journal, 2017. Omega-3 trials cited are VITAL, ASCEND, and REDUCE-IT. The omega-6 position follows the American Heart Association advisory, Harris et al., Circulation, 2009, and biomarker analyses by Marklund et al. Oil oxidative stability follows de Alzaa et al., Acta Scientific Nutritional Health, 2018.

Open questions. How much of the apparent neutrality of dairy fat is food matrix and how much is residual confounding is unresolved. The REDUCE-IT result remains contested because of its mineral oil placebo, and whether purified EPA differs meaningfully from mixed omega-3 preparations is not settled.

👉 Next: fibre and the microbiome, the part of your food you cannot digest and cannot do without.

Fibre and the Microbiome

TL;DR. Fibre is the part of plant food your own enzymes cannot digest, and it is the single most consistently under-eaten thing in modern diets: most populations eat about half the recommended amount. It does mechanical work in the gut and, more importantly, feeds the bacteria in your colon, who ferment it into short-chain fatty acids that nourish your gut lining and signal to your immune system. The evidence linking fibre intake to lower rates of heart disease, type 2 diabetes, colorectal cancer, and total mortality is among the strongest in nutrition. The microbiome field around it is genuinely exciting and, so far, has produced far more press releases than treatments.

Key takeaways

• Recommended intake is around 30 g a day; typical intake in the UK and US is 15 to 20 g. This gap is one of the clearest nutritional deficits in wealthy countries.
• Soluble and insoluble is a useful but crude split. Fermentability and viscosity are the properties that actually predict effects.
• Butyrate, produced by bacteria fermenting fibre, is the main fuel for colon cells. A low-fibre diet starves the gut lining directly.
• Diversity of plant foods predicts microbiome diversity better than quantity of any one fibre. The often-quoted "30 plants a week" target comes from the American Gut Project.
• Most probiotic claims outrun the evidence. A few specific strains for a few specific conditions have real support; most products do not.

What fibre actually is

In short: Plant carbohydrates that reach the colon intact, because human enzymes cannot cut their bonds.

From Chapter 1: starch and cellulose are both chains of glucose, and the only difference is the geometry of the link. Your amylase cuts α bonds and not β bonds. Everything with the wrong bond geometry, plus a few other polymers, arrives in the colon undigested. That is fibre.

The soluble/insoluble split, taught everywhere, is a laboratory classification based on whether the fibre dissolves in water. It correlates loosely with effects and misleads in important cases. The two properties that actually predict what a fibre does are:

• Viscosity: does it form a gel? Viscous fibres (beta-glucan, pectin, psyllium, guar) slow gastric emptying, slow glucose absorption, and trap bile acids, which is how they lower LDL cholesterol.
• Fermentability: do the bacteria eat it? Highly fermentable fibres (inulin, pectin, resistant starch, GOS) produce short-chain fatty acids and also produce gas.

Psyllium is the illustrative exception: it is soluble and very viscous but poorly fermentable, which is exactly why it is the fibre supplement of choice for irritable bowel syndrome. It gives the gel benefits without the gas.

What fibre does in the gut

In short: It adds bulk and holds water, slows absorption where it is viscous, traps bile acids, and, most importantly, feeds bacteria.

1. Bulk and transit. Insoluble fibre holds water and increases stool mass, which stretches the colon wall and stimulates the movement that propels it. Wheat bran is the most effective bulking agent by mass, and each gram of it can add several grams to stool weight. This is why insoluble fibre helps constipation and why it can worsen symptoms in an already irritated bowel.

2. Viscosity effects. Gel-forming fibres slow gastric emptying and create a physical barrier in the small intestine that slows the diffusion of glucose and fat to the absorptive surface. This flattens the post-meal glucose curve. It also traps bile acids and carries them out of the body, and since bile acids are made from cholesterol, the liver must draw on circulating cholesterol to replace them. That is the mechanism behind the LDL-lowering effect of oat beta-glucan, which is one of the few health claims allowed on food packaging in both the EU and US: 3 g a day of oat beta-glucan lowers LDL by roughly 5 to 10 percent.

3. Feeding the microbiome, which is where most of the interesting biology is.

The microbiome, without the hype

In short: Roughly 38 trillion bacteria in the colon, dominated by a few phyla, whose main service is fermenting what you cannot digest.

Current estimates put the number of bacterial cells in the body at around 38 trillion, against roughly 30 trillion human cells, so the "ten times more bacteria than cells" statistic that circulated for decades was an overestimate that has been corrected. It is roughly one to one, which is still remarkable.

Almost all of them live in the colon, in near-total absence of oxygen, at about 10¹¹ cells per gram of contents. A typical adult carries several hundred species, dominated by two phyla: Bacillota (formerly Firmicutes) and Bacteroidota (formerly Bacteroidetes), with smaller contributions from Actinomycetota (including Bifidobacterium), Pseudomonadota, and Verrucomicrobiota (including Akkermansia muciniphila).

Two facts to hold onto. Individual microbiomes differ enormously: two healthy people may share only a modest fraction of their species. And function is more conserved than taxonomy: different species doing the same job, so what the community does matters more than exactly who is in it.

What the microbiome actually does for you

Fermentation into short-chain fatty acids. The central service. Bacteria ferment fibre and resistant starch anaerobically, producing mainly:

Butyrate is the one to know. Colonocytes derive most of their energy from butyrate rather than from blood glucose, which is unusual for a human cell. So a diet with almost no fermentable fibre literally underfeeds the lining of your own colon, thinning the mucus layer and weakening the barrier. There is good animal evidence that fibre-deprived microbes begin consuming the intestinal mucus layer instead, bringing bacteria closer to the epithelium and promoting inflammation.

Other functions:

• Vitamin synthesis: vitamin K2 and several B vitamins (biotin, folate, B12, riboflavin), though how much is absorbed in the colon is debated and it should not be relied upon.
• Immune education: gut bacteria induce regulatory T cells, and germ-free animals have visibly underdeveloped immune systems. Much of the "hygiene hypothesis" literature about allergy and autoimmunity runs through this.
• Colonisation resistance: a dense resident community occupies niches and outcompetes incoming pathogens. Broad-spectrum antibiotics remove this, which is how Clostridioides difficile takes over.
• Bile acid transformation, which affects fat digestion and signalling.
• Drug metabolism, which is real and clinically significant: gut bacteria inactivate the Parkinson's drug levodopa, activate the prodrug sulfasalazine, and reactivate metabolites of the chemotherapy drug irinotecan, which causes its characteristic diarrhoea.
• Gut-brain signalling via the vagus nerve, immune mediators, and microbial metabolites. The animal work here is striking. The human work is early.

What shapes your microbiome

The honest state of microbiome science

This is a field where the gap between what is demonstrated and what is claimed is wide, so it is worth stating plainly.

Well established: the microbiome ferments fibre into SCFAs; it educates the immune system; antibiotics disrupt it; faecal microbiota transplantation cures recurrent C. difficile infection at high rates and is an approved therapy; diet changes composition quickly.

Plausible, being actively researched: links to obesity, type 2 diabetes, inflammatory bowel disease, colorectal cancer, allergy, and possibly mood.

Overstated: almost every consumer claim. Microbiome testing kits sold direct to consumers cannot currently tell you anything actionable, because there is no agreed definition of a healthy microbiome, results vary by sampling and sequencing method, and the recommendations generated from them are generic. Most probiotic supplements do not establish residence in the gut and their effects, where measurable, stop when you stop taking them.

The obesity story specifically is a cautionary example. Early work showing that transplanting microbiota from obese mice made lean mice fatter was genuinely exciting. Attempts to translate it to humans have been far less impressive, the widely cited Firmicutes-to-Bacteroidetes ratio has not held up as a marker, and human FMT trials for obesity have produced modest and inconsistent results.

Probiotics, prebiotics, and the rest of the -biotics

In short: Prebiotics are food for your existing bacteria and work reliably; probiotics are live bacteria and work for a short list of specific indications.

Where probiotics have real evidence, and it is strain-specific rather than species-specific, let alone generic:

• Antibiotic-associated diarrhoea: reasonably good evidence for Saccharomyces boulardii and certain Lactobacillus strains, with meaningful reductions in incidence.
• Acute infectious diarrhoea in children: modest reduction in duration, roughly a day, though two large 2018 trials found no benefit, which tempered enthusiasm considerably.
• Necrotising enterocolitis in preterm infants: this is the strongest evidence base in the whole field, with substantial reductions in a devastating condition.
• Irritable bowel syndrome: some strains help some people. Response is unpredictable.
• Pouchitis after colectomy: good evidence for a specific multi-strain formulation.

Where the evidence does not support the claims: general "gut health" in healthy people, immunity, mood, weight loss, and skin. A 2016 systematic review found no consistent effect of probiotics on the faecal microbiota composition of healthy adults, which is awkward for the entire premise of the consumer category.

Practical guidance if you want to try one: choose a product naming the specific strain (e.g. Lacticaseibacillus rhamnosus GG, not "Lactobacillus blend"), matched to a condition with evidence for that strain, at the dose used in the trials, and give it four weeks.

Prebiotics work more predictably because they do not need to survive, colonise, or compete. They just feed what is already there. Inulin, FOS, and GOS reliably increase Bifidobacterium. The dose-limiting problem is gas: 5 g is usually fine, 10 to 20 g causes noticeable bloating in many people. Which is the correct place to mention the exception.

When fibre is the problem: FODMAPs and IBS

In short: In irritable bowel syndrome, the fermentation that benefits most people causes pain and bloating, and a structured elimination can identify which fibres are responsible.

FODMAP stands for Fermentable Oligosaccharides, Disaccharides, Monosaccharides, And Polyols. They are short-chain carbohydrates that are poorly absorbed in the small intestine, draw water in osmotically, and are rapidly fermented in the colon:

In most people this is exactly what you want. In irritable bowel syndrome, where the gut is hypersensitive to distension, the same process produces pain, bloating, and altered bowel habit. The low-FODMAP diet, developed at Monash University, produces symptom improvement in roughly 50 to 75 percent of people with IBS, which is a strong result for a dietary intervention.

Two cautions that are frequently ignored. It is a diagnostic protocol, not a permanent diet: strict elimination for two to six weeks, then systematic reintroduction to identify personal triggers, then the least restrictive long-term diet possible. And staying on it permanently reduces fibre intake and measurably reduces beneficial bacteria, which is the opposite of what you want. It should be done with a dietitian.

How much fibre, and where to get it

In short: Around 30 g a day, from a variety of plant sources, increased gradually with water.

Recommendations: the UK sets 30 g a day for adults, the US sets 25 g for women and 38 g for men (14 g per 1,000 kcal), and the EFSA suggests at least 25 g. Actual intakes in both the UK and US average around 15 to 20 g. Estimates of Hadza hunter-gatherer intake run to 100 g a day, and of Palaeolithic diets similarly high.

The dose-response evidence is unusually good. A large 2019 Lancet meta-analysis commissioned by the WHO, covering 185 observational studies and 58 trials, found that people eating the most fibre had roughly 15 to 30 percent lower all-cause and cardiovascular mortality, coronary heart disease incidence, stroke, type 2 diabetes, and colorectal cancer, with benefit continuing to rise up to at least 25 to 29 g a day and evidence of further benefit beyond. Whole grains showed the same pattern. That is about as consistent as nutritional epidemiology gets.

The pattern is clear: pulses are the densest practical source by a wide margin, and a single cup of beans or lentils delivers half a day's fibre. Adding pulses to your week is the highest-leverage fibre change available, ahead of bran, supplements, and everything else.

The "30 plants a week" idea comes from the American Gut Project, which found that people eating more than 30 different plant species per week had more diverse gut microbiomes than those eating fewer than 10. It is observational and it makes a good target because it pushes toward variety rather than quantity. Herbs, spices, nuts, seeds, grains, and different varieties all count, which makes 30 much easier than it sounds.

Increase gradually. Going from 15 g to 35 g overnight produces gas, bloating, and cramping, because the bacterial community needs weeks to adapt. Add 5 g a week. And increase water alongside, because bulking fibres without adequate fluid can worsen constipation rather than relieve it, which is the most common way fibre advice backfires.

The bottom line

• Fibre is plant carbohydrate your enzymes cannot cut. Its two properties that matter are viscosity (which slows glucose and traps bile acids) and fermentability (which feeds bacteria).
• Colon cells run primarily on butyrate produced by bacteria fermenting fibre, so a low-fibre diet underfeeds your own gut lining.
• The evidence linking higher fibre intake to lower mortality, heart disease, diabetes, and colorectal cancer is among the strongest and most consistent in nutrition, and most people eat about half the recommended amount.
• Pulses are the densest practical source. Variety of plants matters as much as total grams.
• Prebiotics reliably feed your existing bacteria. Probiotics work for a short list of specific strains and indications and are otherwise oversold. Consumer microbiome tests are not yet actionable.

Sources and notes

Fibre classification, viscosity, and fermentability follow standard nutrition texts and the Institute of Medicine definitions. The dose-response mortality and disease evidence is Reynolds et al., The Lancet, 2019, commissioned by WHO. Oat beta-glucan and LDL is the basis of authorised health claims by EFSA and the FDA. Short-chain fatty acid production and butyrate as colonocyte fuel follow Roediger's original work and subsequent reviews. Microbial cell counts revised to roughly 1:1 with human cells follow Sender, Fuchs, and Milo, PLoS Biology, 2016. Fibre deprivation and mucus layer degradation follow Desai et al., Cell, 2016. Human milk oligosaccharides and Bifidobacterium follow Bode's reviews. Probiotic evidence by indication follows Cochrane reviews for antibiotic-associated diarrhoea and the preterm NEC literature; the two large negative paediatric gastroenteritis trials are Freedman et al. and Schnadower et al., New England Journal of Medicine, 2018. The finding that probiotics do not consistently alter healthy adult microbiota is Kristensen et al., Genome Medicine, 2016. Low-FODMAP evidence follows the Monash University programme. The 30-plants finding is from the American Gut Project, McDonald et al., mSystems, 2018.

Open questions. Almost nothing about the microbiome is yet actionable at the individual level: there is no agreed definition of a healthy microbiome, and consumer testing cannot currently generate useful recommendations. Whether the obesity associations seen in mice translate to humans has repeatedly failed to replicate.

👉 Next: the vitamins, all thirteen of them, with what each does and which ones people actually run short of.

The Vitamins

TL;DR. Thirteen compounds your body cannot make in adequate amounts and must obtain from food. Four are fat-soluble and can accumulate to toxic levels; nine are water-soluble and mostly cannot. The deficiencies that shaped history (scurvy, beriberi, pellagra, rickets) are largely solved by food supply and fortification, and the ones people genuinely run short of today are a much shorter list: vitamin D in high-latitude winters, B12 in vegans and older adults, folate before conception, and iron, which is a mineral but belongs in the same conversation. For everyone else, supplementation has repeatedly failed to show benefit in trials, and in a few notable cases has caused harm.

Key takeaways

• Fat-soluble (A, D, E, K) accumulate; water-soluble (C and the B group) are mostly excreted, with B6 the important exception, which causes nerve damage in excess.
• Vitamin D is the one supplement with a broad case in high-latitude populations, and even then the trial evidence for outcomes beyond bone health is weaker than the enthusiasm.
• B12 comes only from animal foods or fortification, is absorbed via a two-step mechanism that fails with age, and its deficiency causes irreversible nerve damage if missed.
• Folate before and during early pregnancy prevents neural tube defects. This is one of the strongest findings in nutrition and the basis of mandatory fortification in over 80 countries.
• Beta-carotene supplements increased lung cancer in smokers in two large trials. More is not safer.

The thirteen, at a glance

RDA values vary between countries; these are broadly the US/EU adult figures. "RAE" is retinol activity equivalents, "NE" niacin equivalents, "DFE" dietary folate equivalents: each accounts for different forms having different potency.

Fat-soluble: A, D, E, K

In short: Absorbed with dietary fat, stored in liver and fat tissue, and capable of accumulating to toxic levels.

Because these are stored, deficiency takes longer to develop and excess is a genuine risk. They also all require fat in the meal to be absorbed, which is why bile duct disease, pancreatic insufficiency, cystic fibrosis, and fat malabsorption produce combined deficiencies of all four.

Vitamin A

Comes in two forms that behave differently. Preformed vitamin A (retinol, retinyl esters) from animal foods is directly usable and is the form that causes toxicity. Provitamin A carotenoids (beta-carotene, alpha-carotene, beta-cryptoxanthin) from plants must be converted, and conversion is regulated and inefficient, which is why you cannot poison yourself with carrots. You can, however, turn orange: carotenoderma is a harmless yellow-orange discolouration of the palms and soles from very high carotene intake.

What it does: Retinal is the light-sensitive molecule in your retina; without it, night vision fails first. Retinoic acid regulates gene expression in epithelial tissue and immune cells.

Deficiency is the leading preventable cause of childhood blindness worldwide, and it also raises mortality from infection substantially. Vitamin A supplementation programmes in deficient populations are among the highest-return public health interventions available. Sequence: night blindness → dry eye (xerophthalmia) → Bitot's spots → corneal ulceration → irreversible blindness.

Toxicity is real and specific to the preformed form. Acute: headache, nausea, raised intracranial pressure. Chronic: liver damage, bone loss, hair loss, and, importantly, birth defects. This is why isotretinoin (a vitamin A derivative used for acne) requires strict pregnancy prevention, and why pregnant women are advised to avoid liver and liver products, which can contain enormous amounts.

Vitamin D

The one with the strongest general case for supplementation and the most oversold outcome claims.

It is not really a vitamin. It is a hormone precursor you make in your skin: UVB radiation converts 7-dehydrocholesterol to previtamin D3, which is hydroxylated in the liver to 25(OH)D (the form measured in blood tests) and then in the kidney to the active 1,25(OH)₂D.

What it does: raises calcium and phosphate absorption from the gut, which is essential for bone mineralisation. It also has receptors in most tissues including immune cells, which is the basis for the wider claims.

Why deficiency is so common:

• Latitude. Above roughly 37 degrees north or south, UVB is insufficient for synthesis for several winter months. In the UK, essentially no vitamin D is made from October to March.
• Skin pigmentation. Melanin absorbs UVB, so people with darker skin need substantially longer sun exposure for the same synthesis, which is why deficiency rates are much higher in dark-skinned populations at high latitudes.
• Indoor life, clothing, and sunscreen. SPF 30 correctly applied blocks most vitamin D synthesis, though in practice people apply too little for that to fully hold.
• Age. Skin synthesis declines substantially in older adults.
• Few food sources. Oily fish, egg yolks, liver, and fortified foods. It is genuinely hard to get enough from an unfortified diet.

What the evidence supports:

• Established: prevention and treatment of rickets and osteomalacia. This is not in doubt.
• Reasonable: fracture prevention in older adults, but mainly in combination with calcium and mainly in institutionalised or deficient populations. Trials in vitamin-D-replete community-dwelling adults have been largely negative.
• Weak or negative: the very large VITAL trial (over 25,000 participants, 2,000 IU/day) found no reduction in cancer incidence or cardiovascular events. Similar large trials have been negative for diabetes prevention, depression, and cognitive decline. Observational associations with almost every disease are abundant and are heavily confounded, because ill and inactive people go outside less.
• Respiratory infections: meta-analyses suggest a small protective effect, largest in people who were deficient and given daily rather than bolus dosing. Modest.

Practical position: in a high-latitude country, a daily supplement of 10 µg (400 IU) in winter, as the UK NHS advises for everyone, or 20 µg (800 IU) for older adults and those with darker skin or little sun exposure, is cheap, safe, and sensible. Doses above 100 µg (4,000 IU) daily long term are not advisable, and very high doses cause hypercalcaemia, kidney stones, and, paradoxically, in some trials, more falls and fractures. More is not better here.

Vitamin E

A group of eight compounds, of which alpha-tocopherol is the one humans retain. It is a lipid-soluble antioxidant protecting membrane fats from oxidation.

Deficiency is rare outside fat malabsorption conditions. And this is the vitamin with the clearest cautionary supplement history: high-dose vitamin E supplements have been associated with increased all-cause mortality in meta-analyses, and the SELECT trial found a statistically significant increase in prostate cancer in men taking 400 IU daily. Do not take high-dose vitamin E.

Vitamin K

K1 (phylloquinone) from green leafy vegetables is the main dietary form. K2 (menaquinones) comes from fermented foods (natto is by far the richest) and some animal products, and from gut bacteria.

Its job is to activate specific proteins by carboxylating them: clotting factors II, VII, IX, and X, and also osteocalcin in bone and matrix Gla protein in blood vessels.

Two practical points. Newborns are given vitamin K at birth because they are born with low stores and no gut flora, and haemorrhagic disease of the newborn, including intracranial bleeding, is devastating and entirely preventable. The evidence here is overwhelming and declining this injection is a genuinely bad decision.

And warfarin works by blocking vitamin K recycling, which is why people on warfarin are told to keep their green vegetable intake consistent rather than low. Sudden changes in intake destabilise the INR in either direction (Chapter 75).

Water-soluble: the B group and C

In short: Not stored to any great extent, so intake needs to be regular, and excess is mostly excreted, with a few exceptions.

Vitamin C

Humans, other primates, guinea pigs, and bats lost the enzyme to make it. Most other animals make their own.

What it does: it is an essential cofactor for the enzymes that hydroxylate proline and lysine in collagen, which is why deficiency causes connective tissue to fail. It is also a water-phase antioxidant and it dramatically enhances absorption of non-haem (plant) iron.

Scurvy is worth describing because it is the archetypal deficiency disease and the history is instructive. Collagen synthesis fails, so existing collagen is not replaced: bleeding gums, loose teeth, poor wound healing (old scars reopen), perifollicular haemorrhages, joint pain, and eventually death. It killed more sailors than combat during the age of sail. James Lind's 1747 shipboard experiment comparing six treatments in twelve sailors is a founding example of the controlled clinical trial, and the Royal Navy still took over forty years to adopt citrus routinely. Roughly 10 mg a day prevents it, which is about a sixth of an orange.

Does it prevent colds? This is the most-tested question in supplement history. The Cochrane review of over 29 trials is clear: routine supplementation does not reduce the incidence of colds in the general population. It does shorten duration slightly, by around 8 percent in adults and 14 percent in children, roughly half a day. The exception is people under extreme physical stress (marathon runners, soldiers in subarctic training), where supplementation roughly halved incidence. Linus Pauling, whose advocacy created the entire belief, was a brilliant chemist who was wrong about this.

Upper limit is 2,000 mg, above which osmotic diarrhoea and increased oxalate excretion (hence kidney stone risk in susceptible people) become issues.

Vitamin B12

The one that matters most for anyone eating a plant-based diet.

Only bacteria and archaea make B12. Animals get it by eating bacteria or bacteria-fed food, or from their own gut bacteria in ruminants. Plants contain essentially none. Fortified foods and supplements are made by bacterial fermentation.

Absorption is a two-stage relay and unusually fragile:

1. Stomach acid and pepsin release B12 from food protein.
2. It binds to haptocorrin from saliva.
3. Pancreatic enzymes release it in the duodenum.
4. It binds intrinsic factor, made by stomach parietal cells.
5. The complex is absorbed by a specific receptor in the terminal ileum.

Break any step and absorption fails. Hence deficiency in: pernicious anaemia (autoimmune destruction of parietal cells), gastric surgery, ileal disease or resection (Crohn's), and crucially atrophic gastritis, which affects a substantial proportion of adults over 60 and impairs the release of B12 from food while leaving supplement absorption intact. That last point is why B12 supplementation is often recommended for older adults regardless of diet.

Metformin reduces B12 absorption measurably, and long-term users should have levels checked. Proton pump inhibitors do the same, by removing the acid needed for the first step.

Deficiency produces megaloblastic anaemia and, separately, neurological damage: peripheral neuropathy, subacute combined degeneration of the spinal cord, cognitive impairment. The neurological damage can become irreversible if untreated, which is what makes this the most important vitamin deficiency to catch early.

Don't be confused: folate can mask B12 deficiency. High folate intake corrects the anaemia of B12 deficiency without touching the neurological damage, so the blood count looks normal while nerve damage progresses silently. This is the main argument raised against high-dose folic acid fortification, and it is why B12 status should be assessed rather than assumed when folate is high.

Liver stores are large, typically 2 to 5 mg against a daily need of about 2.4 µg, so deficiency after switching to a vegan diet may take 2 to 5 years to appear. This delay is exactly why it catches people out. Anyone eating a vegan diet needs a reliable B12 source, either a supplement or consistently fortified foods. This is not negotiable and it is the single most important supplementation fact in this book.

Folate (B9)

What it does: carries one-carbon units for DNA synthesis and methylation. Rapidly dividing cells need it most, which is why deficiency shows up in blood cells and in a developing embryo.

Neural tube defects. The neural tube closes by day 28 after conception, which is often before a woman knows she is pregnant. Adequate folate at that moment substantially reduces the risk of spina bifida and anencephaly. The MRC Vitamin Study of 1991 demonstrated a roughly 70 percent reduction in recurrence with supplementation, and it is one of the cleanest findings in nutritional science.

The public health consequence: over 80 countries mandate folic acid fortification of flour or grain, and neural tube defect rates fell sharply where it was implemented, by roughly 25 to 50 percent. The UK announced mandatory fortification of non-wholemeal wheat flour in 2021 after decades of debate. Recommended supplementation is 400 µg daily for anyone who could become pregnant, from before conception through the first trimester, and higher doses (typically 5 mg) for those with previous affected pregnancies, diabetes, obesity, or on certain anti-epileptic drugs.

Don't be confused: folate and folic acid are not the same molecule. Folate is the natural form in food; folic acid is the synthetic, more stable, more bioavailable form used in supplements and fortification. Folic acid must be reduced by the enzyme DHFR before use, and that step is slow in humans, which is why unmetabolised folic acid appears in blood at high intakes. Methylfolate supplements avoid this step and are useful for people with certain MTHFR gene variants, though the clinical significance of common MTHFR variants is more modest than supplement marketing implies.

The rest of the B group, briefly

Thiamine (B1). Deficiency causes beriberi: wet (heart failure) or dry (peripheral neuropathy). Historically caused by polished white rice, which removes the thiamine-rich bran, and the discovery that unpolished rice cured it was a founding moment for the whole vitamin concept. Today the important context is alcohol: chronic heavy drinking impairs absorption and depletes stores, and acute deficiency causes Wernicke's encephalopathy (confusion, eye movement abnormalities, ataxia), which if untreated progresses to Korsakoff's syndrome, a permanent and profound memory disorder. This is why intravenous thiamine is given before glucose to anyone with alcohol dependence presenting confused: giving glucose first consumes the last of the thiamine and can precipitate the disaster.

Riboflavin (B2). Deficiency causes cracks at the corners of the mouth, sore tongue, and eye symptoms. Notable trivia with a practical edge: it is bright yellow and the excess is excreted, which is why high-dose B supplements turn urine fluorescent yellow within hours. That is harmless, and it is not evidence the supplement is "working."

Niacin (B3). Deficiency causes pellagra: the four Ds, dermatitis (a distinctive sun-exposed rash), diarrhoea, dementia, and death. It devastated the American South in the early twentieth century, where a maize-based diet was the cause. The reason is elegant: maize contains niacin in a bound, unavailable form, and traditional Mesoamerican nixtamalisation (soaking maize in alkaline lime water) releases it. Cultures that adopted maize without adopting the processing got pellagra; those that kept the processing did not. Joseph Goldberger established the dietary cause in the 1910s against fierce resistance from those insisting it was infectious.

B6 (pyridoxine) is the water-soluble vitamin that is genuinely toxic in excess. Chronic intakes above roughly 100 to 200 mg daily cause sensory peripheral neuropathy: numbness, tingling, and unsteadiness, which may be only partially reversible. High-dose B6 supplements are sold widely for premenstrual symptoms and morning sickness, and this risk is under-communicated. Several countries have restricted maximum doses in supplements as a result.

Biotin (B7). Deficiency is rare, though it can be induced by eating large amounts of raw egg white, which contains avidin, a protein that binds biotin and blocks absorption (cooking denatures it). The important modern issue is different: high-dose biotin supplements, popular for hair and nails, interfere with many laboratory immunoassays, producing falsely high or low results for thyroid hormones and, dangerously, falsely low troponin, which is the test used to diagnose heart attacks. The FDA has issued specific safety communications about this. Tell your doctor if you take biotin.

Pantothenic acid (B5). Present in almost all foods, which is what its name means. Deficiency essentially does not occur.

Should you take a multivitamin?

In short: For most well-fed people, no measurable benefit has been shown; for specific groups, specific supplements have strong evidence.

The trial record for general multivitamin supplementation in well-nourished populations is consistently unimpressive. The US Preventive Services Task Force concluded in 2022 that there is insufficient evidence of benefit for cardiovascular disease or cancer prevention from multivitamins, and recommended against beta-carotene and vitamin E supplements specifically because of evidence of harm. Physicians' Health Study II found a small reduction in total cancer incidence with a daily multivitamin over 11 years, which is the most positive large result and is modest.

Meanwhile the cautionary results are notable:

• Beta-carotene: the CARET and ATBC trials both found increased lung cancer in smokers taking supplements. CARET was stopped early.
• Vitamin E: increased prostate cancer in SELECT.
• High-dose antioxidants may blunt some of the beneficial adaptations to exercise training, which is a small but well-replicated finding.
• Calcium supplements (not dietary calcium) have been associated with cardiovascular events in some analyses.

Where supplementation genuinely earns its place:

For everyone else, food remains the better delivery system, and not because of anything mystical. Food supplies vitamins in the forms, ratios, and matrices that absorption mechanisms evolved with, alongside fibre and the thousands of other compounds that isolated supplements do not contain.

The bottom line

• Thirteen vitamins: four fat-soluble that accumulate and can be toxic, nine water-soluble that mostly cannot, with B6 the notable exception.
• Vitamin D deserves supplementation in high-latitude winters, and the outcome claims beyond bone health have largely failed in large trials.
• B12 comes only from animal foods or fortification, and its deficiency causes nerve damage that can become permanent. Anyone eating vegan, and most older adults, need a source.
• Folic acid before conception prevents neural tube defects. This finding is as solid as nutrition gets and underpins fortification in over 80 countries.
• Multivitamins have repeatedly failed to show benefit in well-fed populations, and beta-carotene and high-dose vitamin E have shown harm. Target the specific gaps you actually have.

Sources and notes

Requirements, upper limits, and deficiency syndromes follow the US Institute of Medicine Dietary Reference Intakes, EFSA Dietary Reference Values, and the UK Scientific Advisory Committee on Nutrition reports. Vitamin D trials cited are VITAL (Manson et al., NEJM, 2019) and D2d; respiratory infection meta-analysis is Martineau et al., BMJ, 2017. Beta-carotene harm is the ATBC (1994) and CARET (Omenn et al., 1996) trials; vitamin E and prostate cancer is SELECT (Klein et al., JAMA, 2011). The Cochrane antioxidant supplement mortality review is Bjelakovic et al. Folic acid and neural tube defects is the MRC Vitamin Study, The Lancet, 1991, and the subsequent fortification literature. B12 absorption, atrophic gastritis, and metformin-associated deficiency follow standard haematology and the ADA standards of care. Vitamin C and colds is the Cochrane review by Hemila and Chalker. Pellagra history follows Goldberger's work and Kraut's biography. Thiamine, Wernicke, and the glucose-before-thiamine caution follow standard emergency medicine guidance. Biotin assay interference is the subject of an FDA safety communication, 2017 and 2019. The USPSTF multivitamin position is its 2022 recommendation statement.

Open questions. Optimal vitamin D status, as opposed to the level that prevents rickets, remains genuinely unsettled, and large trials have not supported the observational associations. Whether the unmetabolised folic acid seen at high fortification intakes matters clinically is unknown.

👉 Next: the minerals, where iron, calcium, iodine, and sodium each carry a different kind of trouble.

The Minerals

TL;DR. Vitamins are organic molecules; minerals are single elements, and unlike vitamins they cannot be destroyed by cooking, only leached into the water. The deficiencies that matter globally are iron (the most common nutritional deficiency in the world, affecting roughly a quarter of humanity), iodine (whose deficiency was the leading preventable cause of intellectual disability until salt iodisation), and zinc. The excesses that matter are sodium, which raises blood pressure across whole populations, and supplemental iron and calcium, both of which can do harm when taken without need.

Key takeaways

• Iron deficiency affects about 2 billion people and is the world's most common nutritional deficiency. Absorption differs fivefold between animal (haem) and plant (non-haem) sources.
• Vitamin C dramatically increases plant iron absorption; tea, coffee, and calcium dramatically reduce it. The timing of a cup of tea genuinely matters.
• Salt iodisation is one of the highest-impact public health measures ever implemented, and iodine deficiency is quietly returning in some countries as people switch to non-iodised specialty salts and plant milks.
• Sodium reduction lowers blood pressure at population scale, and roughly 70 to 80 percent of intake comes from processed food rather than the salt cellar.
• Potassium is the mineral almost everyone under-consumes, and raising it lowers blood pressure roughly as effectively as cutting sodium.

The minerals that matter

Major minerals are needed in amounts above 100 mg a day; trace minerals below that. The distinction is about quantity, not importance: you need 1 gram of calcium a day and 150 micrograms of iodine, and running out of either is serious.

Iron: the world's commonest deficiency

In short: Two forms with very different absorption, tight regulation of uptake and no route for excretion, and a set of dietary interactions that matter more than the raw amount in the food.

Why the body is so careful with it. Iron is essential (haemoglobin, myoglobin, and dozens of enzymes) and dangerous (free iron catalyses reactions that generate damaging radicals, and bacteria need it to grow). So iron is always bound to a carrier protein, and critically the body has no mechanism for excreting excess iron. Regulation happens entirely at absorption, via the hormone hepcidin, which shuts down intestinal iron uptake when stores are high or when inflammation is present.

The two forms:

That difference is why iron intake and iron status can diverge so much between diets.

What increases non-haem absorption:

• Vitamin C. The single strongest enhancer: 50 mg with a meal can increase non-haem iron absorption by three- to sixfold, by reducing ferric iron to the absorbable ferrous form. Practically: a glass of orange juice, tomatoes, or peppers with a lentil dish.
• The "meat factor." Meat, fish, and poultry enhance absorption of non-haem iron eaten alongside them, by a mechanism still not fully explained.
• Cooking in cast iron, particularly acidic foods, genuinely adds absorbable iron.
• Fermentation, soaking, and sprouting, which degrade phytate.

What decreases it:

• Phytate in whole grains, pulses, nuts, and seeds. The strongest inhibitor. Soaking, sprouting, fermenting, and leavening bread with real sourdough all reduce it substantially.
• Polyphenols in tea, coffee, cocoa, and red wine. A cup of tea with a meal can reduce non-haem iron absorption by 60 percent or more. This is one of the most actionable facts in this chapter: if you are iron deficient, drink tea and coffee between meals, not with them.
• Calcium, which interferes with both forms. Do not take a calcium supplement with your iron supplement.
• Antacids and proton pump inhibitors, because stomach acid is needed to solubilise iron.

Deficiency is the most common nutritional deficiency worldwide. It progresses in stages: depleted stores (low ferritin, no symptoms) → iron-deficient erythropoiesis → iron deficiency anaemia (fatigue, pallor, breathlessness on exertion, poor concentration, cold intolerance, brittle nails, and sometimes pica, a craving for non-food substances such as ice, clay, or starch, which is a genuinely distinctive clinical clue).

Who is at risk: menstruating women (particularly with heavy periods), pregnant women, infants and toddlers, adolescents in growth spurts, endurance athletes (through foot-strike haemolysis and gut losses), blood donors, vegetarians and vegans, and anyone with gastrointestinal blood loss. That last one matters most: iron deficiency anaemia in a man or a postmenopausal woman requires investigation for a gastrointestinal source, because colorectal cancer presents this way. It is not something to self-treat with supplements.

Supplementation notes. Ferrous sulphate is the standard and cheapest form. Absorption is better on an empty stomach and better with vitamin C, and worse with food, tea, and calcium. Constipation and dark stools are extremely common; the dark colour is harmless. Recent evidence supports alternate-day dosing as at least as effective as daily dosing and better tolerated, because a large dose raises hepcidin and blocks absorption the following day.

Excess iron is a real danger. Haemochromatosis, a common genetic condition particularly in people of northern European ancestry (roughly 1 in 200 homozygous for the main HFE variant), causes excessive absorption and progressive damage to liver, pancreas, heart, and joints. Treatment is repeated blood removal, which is elegantly simple. And iron tablets are a leading cause of fatal poisoning in young children, because they look like sweets and the lethal dose is small. Keep them out of reach; this is a genuine and repeated paediatric emergency.

Calcium and bone

In short: 99 percent of your calcium is in bone, which also functions as the buffer that keeps blood calcium constant; dietary calcium matters, and less than the dairy marketing suggests.

Blood calcium is held in an extremely narrow range because it controls nerve and muscle function, and the body will dissolve bone to maintain it. Bone is therefore both a structure and a reservoir.

Absorption is about 25 to 35 percent of intake and requires vitamin D. It is higher when intake is low and during growth and pregnancy, and lower with age.

Non-dairy sources, since this is the common question:

Spinach is the instructive case. It is high in calcium and also high in oxalate, which binds calcium into an insoluble compound. Its calcium is almost entirely unavailable, which is why spinach's reputation as a calcium source is wrong, and why low-oxalate greens (kale, bok choy, broccoli, collards) are far better sources despite lower raw numbers.

Supplements are not the same as dietary calcium. Several analyses have associated calcium supplements, particularly at high doses without vitamin D, with a small increase in cardiovascular events, an association not seen with dietary calcium. The current position of most bodies is to meet calcium needs from food where possible and to supplement only for a documented shortfall. Supplemental calcium also raises kidney stone risk, while dietary calcium lowers it, because calcium in the gut binds oxalate and stops it being absorbed. That inversion catches people out: low-calcium diets are a risk factor for calcium oxalate stones.

Bone health is not mostly about calcium. Weight-bearing and resistance exercise, adequate protein, vitamin D, not smoking, and moderate alcohol all matter substantially. Calcium is necessary and not sufficient.

Iodine: the quiet success and the quiet backslide

In short: Needed only for thyroid hormones, deficiency in pregnancy permanently impairs brain development, and salt iodisation nearly solved it before specialty salts and plant milks started reopening the gap.

Iodine has exactly one job: it is a structural component of thyroid hormones T3 and T4, which regulate metabolic rate throughout the body and, critically, brain development in the fetus and infant.

Deficiency causes goitre (the thyroid enlarging as it strains to capture more iodine), hypothyroidism, and, in pregnancy, cretinism: irreversible intellectual disability, deaf mutism, and stunting. Milder maternal deficiency is associated with measurable reductions in child IQ, and this is the reason iodine deficiency was described by the WHO as the leading preventable cause of intellectual disability worldwide.

Salt iodisation, begun in Switzerland and the United States in the 1920s and expanded globally from the 1990s, is among the cheapest and most effective public health interventions ever undertaken. The number of iodine-deficient countries has fallen dramatically.

The backslide is worth flagging because it is current and largely unnoticed:

• Specialty salts (Himalayan pink, sea salt, kosher salt) are usually not iodised. Switching from table salt to sea salt for perceived health reasons removes an iodine source while delivering the same sodium.
• Plant milks are often not fortified with iodine, whereas dairy milk is a major iodine source in countries like the UK (via iodophor teat disinfectants and fortified cattle feed). People switching from cow's milk to almond or oat milk can lose a substantial share of their iodine intake without realising it.
• Reduced salt intake for cardiovascular reasons reduces iodine intake where salt is the vehicle.

The UK, in particular, has documented mild iodine deficiency in teenage girls and pregnant women, which matters given the pregnancy stakes.

Excess iodine is also harmful, causing thyroid dysfunction in both directions. The main route is seaweed: kelp (kombu) is extraordinarily high in iodine, and regular consumption or kelp supplements can deliver many times the upper limit. Nori is far lower and fine.

Sodium and potassium: the pair

In short: They work as opposing partners, most people eat too much sodium and too little potassium, and the ratio may matter more than either number.

Sodium

Essential in small amounts, and the physiological requirement is roughly 500 mg a day. Actual intakes average 3,000 to 5,000 mg in most countries, against WHO advice of under 2,000 mg (about 5 g of salt).

Sodium and blood pressure is one of the better-evidenced relationships in nutrition. The DASH-Sodium trial, which fed participants controlled diets at three sodium levels, showed a clear dose-response reduction in blood pressure. Meta-analyses of trials show reducing sodium lowers blood pressure, more in people who are hypertensive and salt-sensitive. Population studies of countries that reduced salt supply, notably Finland's decades-long programme, show corresponding falls in stroke and heart disease.

The nuance, since it is often used to dismiss the whole thing: some cohort studies find a J-shaped curve where very low sodium is also associated with higher mortality. Interpretation is contested, with reverse causation (sick people are advised to cut salt) and measurement error from single spot urine samples as the main critiques. The mainstream position, that most populations eat far more sodium than they need and reducing it lowers blood pressure, survives this challenge.

Where the salt actually is: roughly 70 to 80 percent of sodium intake in industrialised countries comes from processed and restaurant food, not from the salt cellar. Bread is frequently the single largest contributor, not because it is very salty but because of how much is eaten. So the effective interventions are reformulation by manufacturers and cooking from raw ingredients, not moving the salt shaker off the table.

Salt substitutes, replacing part of the sodium chloride with potassium chloride, tackle both sides at once. The large SSaSS trial in rural China found a substituted salt reduced stroke, cardiovascular events, and total mortality. That is a strong result for a very cheap intervention. The caution is that people with advanced kidney disease or on potassium-sparing drugs must not use them, because they cannot excrete the potassium load.

Potassium

Almost everyone eats too little. Recommended intakes are 3,500 to 4,700 mg a day; typical intakes are well under 3,000 mg. It is the mineral most reliably associated with lower blood pressure across trials, and raising potassium intake appears roughly as effective as cutting sodium.

Best sources are not bananas, which are famous and only moderate. Per typical serving: potatoes with skin, beans and lentils, dried apricots, spinach and other greens, avocado, tomato and tomato products, yoghurt, and salmon all deliver more.

The sodium-to-potassium ratio predicts cardiovascular outcomes better than either alone in several analyses, which conveniently points at the same advice from two directions: less processed food, more vegetables and pulses.

Zinc, selenium, magnesium, and the rest

Zinc. A cofactor in hundreds of enzymes and in the zinc-finger proteins that regulate gene expression. Deficiency causes impaired immunity, poor wound healing, hair loss, skin lesions, growth failure in children, and loss of taste and smell. It is common in low-income settings and in diets high in phytate, which binds zinc as it does iron. Zinc supplementation reduces the duration and severity of childhood diarrhoea substantially and is a WHO-recommended treatment. Zinc lozenges for colds have modest evidence for shortening duration, at high doses started within 24 hours, with an unpleasant taste and nausea as common costs. Chronic high-dose zinc causes copper deficiency, which can produce anaemia and neurological problems, so long-term supplementation above the upper limit of 40 mg is a bad idea. Intranasal zinc products caused permanent loss of smell in some users and were withdrawn.

Selenium. Part of the glutathione peroxidase enzymes and of the enzymes that activate thyroid hormone. Soil selenium varies enormously by region, so intake follows geography: Keshan disease, a cardiomyopathy, occurred in selenium-poor regions of China. The window between adequate and excessive is narrower than for most minerals. Brazil nuts are extraordinarily selenium-dense, with a single nut potentially supplying an entire day's requirement or more, and eating a handful daily can cause selenosis: hair loss, brittle nails, garlic breath, and neurological symptoms. One or two Brazil nuts a day is a sensible supplement; ten is not.

Magnesium. Involved in over 300 enzyme systems, in ATP function (ATP is biologically active as a magnesium complex), and in muscle and nerve function. Intakes are commonly below recommendations, particularly on diets low in whole grains, nuts, and greens. Deficiency associations with hypertension, type 2 diabetes, and arrhythmia are consistent observationally; supplement trials show small blood pressure reductions. The popular use for leg cramps has weak evidence. Different salts differ in absorption and in laxative effect: magnesium oxide is poorly absorbed and mostly a laxative; citrate, glycinate, and malate are better absorbed.

Fluoride. Incorporates into tooth enamel as fluorapatite, which is more acid-resistant. Water fluoridation and fluoride toothpaste both reduce dental caries substantially, and the topical effect from toothpaste is now understood to be the more important mechanism. Excess during tooth development causes dental fluorosis, usually mild white flecking, which is why children should use a smear rather than a strip of toothpaste. Very high chronic intakes, as occur naturally in parts of India and East Africa, cause skeletal fluorosis. The safety debate about community water fluoridation at standard concentrations has been extensively reviewed; recent attention to possible neurodevelopmental effects concerns concentrations well above those used in fluoridated supplies, and the interpretation is contested.

Copper, manganese, chromium, molybdenum. Deficiencies are rare in ordinary diets. Chromium supplements are widely sold for blood sugar control and the evidence is weak and inconsistent.

The bottom line

• Minerals are elements: indestructible by cooking, only leached into water. Use the cooking liquid where you can.
• Iron is the world's commonest deficiency. Plant iron absorption swings by several-fold depending on what you drink and eat with it, and vitamin C up, tea and coffee down, are the two biggest levers.
• Unexplained iron deficiency in men and postmenopausal women needs investigating rather than supplementing.
• Calcium from low-oxalate greens and fortified foods is as usable as dairy calcium; spinach calcium is not. Supplements are a weaker option than food.
• Iodine is a solved problem quietly reopening, via non-iodised specialty salts and unfortified plant milks. It matters most in pregnancy.
• Most people eat too much sodium and too little potassium, and most of the sodium is already in the food when they buy it.

Sources and notes

Requirements and upper limits follow the same DRI, EFSA, and SACN sources as Chapter 15. Iron absorption enhancers and inhibitors, including the tea and polyphenol effect, follow Hallberg's and Hurrell's work; alternate-day iron dosing follows Stoffel et al., The Lancet Haematology, 2017. Hepcidin regulation follows Ganz's reviews. Calcium absorption fractions by food follow Weaver and Heaney's measurements. Calcium supplements and cardiovascular risk follow Bolland et al., BMJ, 2010, and its contested replies. Iodine deficiency as a preventable cause of intellectual disability follows WHO and Iodine Global Network reporting; UK iodine insufficiency in adolescents follows Vanderpump et al., The Lancet, 2011. Sodium and blood pressure follows DASH-Sodium (Sacks et al., NEJM, 2001) and the Cochrane reviews; the salt substitute trial is SSaSS, Neal et al., NEJM, 2021. The J-curve controversy follows PURE (Mente et al.) and its methodological critiques. Selenium geography and selenosis follow Rayman's reviews. Zinc for colds follows Cochrane reviews and Hemila's analyses. Fluoride follows the Cochrane water fluoridation review and Public Health England assessments.

Open questions. The sodium J-curve is genuinely contested and turns almost entirely on whether spot urine estimation is adequate, which is a methodological argument rather than a biological one. Recent attention to fluoride and neurodevelopment concerns concentrations above those used in fluoridated supplies, and its interpretation is disputed.

👉 Next: the phytochemicals, where the gap between what plants contain and what the marketing claims is at its widest.

The Plant Chemicals

TL;DR. Plants make tens of thousands of compounds that are not nutrients: pigments, defence chemicals, flavour molecules. Whole plant foods are consistently associated with better health, and the attempt to identify which specific compound is responsible has mostly failed. Isolated compounds in supplements have repeatedly underperformed the foods they came from, sometimes badly. The most likely explanation is that the benefit comes from the whole package (fibre, matrix, hundreds of compounds acting weakly and together, plus what those foods displace), and that "antioxidant" was the wrong framing for most of it.

Key takeaways

• The antioxidant hypothesis, as popularly understood, has largely failed in trials. Antioxidant supplements have shown no benefit and occasionally harm.
• The better current explanation is hormesis: many plant compounds act as mild stressors that activate your own protective systems, rather than mopping up radicals directly.
• Bioavailability is the recurring problem. Most polyphenols are poorly absorbed, extensively metabolised, and reach blood at concentrations far below those used in laboratory studies.
• The ORAC scale was withdrawn by the USDA in 2012 because it did not predict anything in humans. It is still used in marketing.
• Sulforaphane from brassicas and lycopene from cooked tomato have the clearest mechanisms and the best, though still not definitive, evidence.

What phytochemicals are

In short: Non-nutrient plant compounds, mostly defensive or structural, numbering in the tens of thousands.

Nutrients (vitamins, minerals, macronutrients) are compounds whose absence causes a defined deficiency disease. Phytochemicals are not nutrients: you will not get a deficiency disease from a diet lacking quercetin. Yet diets rich in the foods that contain them are consistently associated with better outcomes.

Estimates put the number of distinct phytochemicals in the human diet somewhere between five and ten thousand, with perhaps a few hundred studied in any depth and a handful studied properly in humans. A single apple contains several hundred.

The rough rule that colour indicates phytochemical class is genuinely useful for shopping: red usually means lycopene or anthocyanin, orange means carotenoids, deep green means carotenoids masked by chlorophyll plus glucosinolates in brassicas, purple means anthocyanins, white and pale means organosulphur or flavonols. "Eat a variety of colours" is a marketing slogan that happens to be reasonable advice.

The antioxidant story, and why it collapsed

In short: A plausible mechanism, a memorable narrative, an industry built on it, and a run of large trials that did not support it.

The hypothesis

Metabolism generates reactive oxygen species (free radicals and related molecules) as an unavoidable by-product. These damage DNA, proteins, and membrane lipids. Cumulative oxidative damage was proposed as a driver of ageing, cancer, and cardiovascular disease. Plant foods are rich in compounds that neutralise radicals in a test tube. Therefore, the reasoning went, supplementing those compounds should reduce disease.

Every step is plausible. The conclusion failed.

The trials

That is an unusually consistent negative record for a hypothesis with such a strong theoretical basis.

Why it failed

Several explanations, probably all partly true.

Radicals are not simply bad. Reactive oxygen species are signalling molecules. Your immune cells use an oxidative burst to kill pathogens. Exercise adaptation is partly driven by oxidative stress signalling: several studies show that high-dose vitamin C and E supplementation blunts the training adaptations to exercise. Suppressing an important signalling system wholesale is not obviously beneficial.

Concentration. In vitro studies apply micromolar concentrations that dietary intake rarely achieves in blood.

The wrong mechanism. The direct radical-scavenging capacity of dietary polyphenols in the body is small relative to endogenous antioxidants such as glutathione, superoxide dismutase, catalase, and uric acid. Whatever polyphenols do, direct scavenging is probably not the main part.

Isolation removes the context. A tomato is not lycopene; it is lycopene plus fibre plus potassium plus a hundred other compounds in a physical matrix, eaten instead of something else.

What replaced it: hormesis

The better current framework is hormesis: a mild stressor provoking a protective adaptive response that exceeds the harm of the stressor itself.

The best-worked example is the Nrf2 pathway. Compounds including sulforaphane (from broccoli), curcumin, and various polyphenols are mildly reactive and mildly toxic. Cells detect them and activate Nrf2, a transcription factor that switches on dozens of the body's own antioxidant and detoxification enzymes: glutathione S-transferases, NAD(P)H quinone oxidoreductase, heme oxygenase-1. The net effect is a stronger endogenous defence system, lasting far longer than the compound itself.

Under this model, plant compounds are useful because they are mildly toxic, which is exactly what Chapter 1 said they were made to be. It also explains why mega-dosing does not scale: past the hormetic window, the stressor is simply toxic.

Don't be confused: ORAC scores mean nothing for you. The Oxygen Radical Absorbance Capacity scale measured a food's ability to neutralise radicals in a test tube. Foods were ranked by it and marketed on it for years. In 2012 the USDA withdrew its ORAC database entirely, stating that the values had "no relevance to the effects of specific bioactive compounds on human health" and that they were being misused in marketing. The scores are still quoted on packaging.

Bioavailability: the recurring disappointment

In short: Most polyphenols are poorly absorbed, heavily transformed by the liver and gut bacteria, and cleared quickly.

The typical fate of a dietary polyphenol:

1. Poor absorption. Many are bound to sugars or to fibre and pass through largely unabsorbed.
2. Immediate metabolism. What is absorbed is rapidly conjugated in the intestinal wall and liver, so the compound circulating in your blood is not the compound in the food.
3. Low peak concentrations, typically nanomolar to low micromolar, often 10 to 1,000-fold below the concentrations used in cell studies.
4. Fast clearance, often within hours.
5. Microbial transformation. A large share reaches the colon, where bacteria break it into smaller phenolic acids, and those metabolites may be the biologically active agents. This is increasingly seen as the main story rather than a footnote.

The clearest illustration is soy isoflavones and equol. Daidzein, an isoflavone in soy, is converted by certain gut bacteria into equol, which is substantially more oestrogenically active. Only about 25 to 30 percent of Western populations and 50 to 60 percent of East Asian populations harbour the bacteria to do this. So identical soy intake produces different exposure in different people, and trials that ignore equol producer status are averaging over two different populations. This may explain a good deal of the inconsistency in soy research.

Curcumin is the other famous case. It has genuinely interesting activity in cell studies and terrible oral bioavailability: poorly absorbed, rapidly metabolised, and barely detectable in plasma after ordinary doses. The standard workaround is co-administration with piperine from black pepper, which inhibits its metabolism and raises bioavailability substantially, and which is why nearly every curcumin supplement contains it. Curcumin also has a reputation among medicinal chemists as a PAINS compound (pan-assay interference compound): it produces apparent hits in a very wide range of assays through non-specific mechanisms such as membrane disruption and metal chelation, which makes its enormous in-vitro literature much less informative than its volume suggests. The human trial evidence, mostly for osteoarthritis pain, is modest.

The compounds with the best cases

In short: A handful have mechanisms, human data, and plausibility all pointing the same way.

Sulforaphane and the glucosinolates

The mechanism is elegant. Brassicas store glucosinolates (glucoraphanin in broccoli) in one cell compartment and the enzyme myrosinase in another. When an insect chews the leaf, the compartments break and myrosinase converts the harmless storage compound into sulforaphane, a reactive isothiocyanate. It is a chemical weapon that is only armed on damage.

In humans, sulforaphane is one of the most potent known dietary activators of the Nrf2 pathway. Evidence includes human trials showing accelerated excretion of airborne pollutants in a randomised trial in Qidong, China, using broccoli sprout beverage, and consistent observational associations between cruciferous vegetable intake and lower cancer risk.

The practical consequence is about cooking. Myrosinase is a protein and is destroyed by heat. Boiled broccoli produces far less sulforaphane than raw or lightly steamed. Two workarounds have human evidence: chop and wait, cutting the vegetable and leaving it 40 minutes before cooking so the enzyme does its work first, and adding a raw brassica (mustard powder, rocket, radish, wasabi) to cooked brassicas to supply myrosinase. Your gut bacteria can also perform the conversion, less efficiently. Steaming for about three to four minutes appears to be the sweet spot: enough to soften, not enough to destroy the enzyme.

Broccoli sprouts contain 10 to 100 times more glucoraphanin than mature broccoli, which makes them the most concentrated practical source.

Lycopene

The red carotenoid in tomatoes, and unusual in that cooking and processing increase its availability substantially. Heat breaks cell walls and converts lycopene from the trans to the more absorbable cis form. Tomato paste and cooked tomato sauce deliver several times more absorbable lycopene than raw tomato, and oil in the same dish increases absorption further, because lycopene is fat-soluble. Traditional Mediterranean cooking arrived at exactly the right method by taste alone.

The evidence: consistent observational association between tomato product intake and lower prostate cancer risk, plus supporting mechanisms. Lycopene supplements have not replicated this, which is the recurring pattern.

Lutein and zeaxanthin

These two carotenoids selectively accumulate in the macula of the retina, forming the macular pigment, where they filter blue light and act as antioxidants in a tissue exposed to intense light and high oxygen. This is one of the few cases where a phytochemical has a demonstrated, specific, anatomical destination.

The AREDS2 trial is one of the genuinely positive supplement trials in this field: a formulation including lutein and zeaxanthin (replacing beta-carotene, which was removed because of the lung cancer signal in smokers) reduced progression of age-related macular degeneration in people with intermediate disease. It does not prevent AMD in healthy eyes.

Sources: kale, spinach, and other dark greens; maize; and egg yolk, whose lutein is particularly well absorbed because it comes packaged with fat.

Flavanols from cocoa and tea

Cocoa flavanols have reasonably consistent short-term trial evidence for improved endothelial function and modest blood pressure reduction (a few mmHg). The large COSMOS trial of cocoa extract supplements found no reduction in the primary cardiovascular outcome, though a secondary analysis suggested a reduction in cardiovascular death. Modest, real, and not a reason to eat more chocolate: most chocolate contains little flavanol, since the processing that makes cocoa palatable (fermentation, roasting, and especially Dutch alkalising) destroys most of it, and it arrives with sugar and fat.

Nitrate from beetroot and greens

Dietary nitrate from beetroot, rocket, spinach, and lettuce is reduced to nitrite by bacteria on the tongue, then to nitric oxide, which dilates blood vessels. This is a well-characterised pathway with measurable effects: beetroot juice lowers blood pressure by a few mmHg and modestly improves exercise efficiency in trials. It is one of the better-evidenced ergogenic aids in sport.

The wrinkle, worth noting because it seems contradictory: nitrates and nitrites in cured meat are associated with colorectal cancer risk, because in the presence of haem iron and protein they form N-nitroso compounds. In vegetables, accompanied by vitamin C and polyphenols that block nitrosation, the same ions behave differently. Context again.

Resveratrol, and a cautionary tale

Resveratrol from grape skins was the proposed explanation for the "French paradox" and became one of the most heavily marketed supplements of the 2000s, on the back of research on sirtuins and lifespan extension.

Two problems. First, the amount in red wine is tiny: reaching the doses used in mouse studies would require drinking hundreds of litres a day. Second, the field suffered a serious research integrity failure when a leading resveratrol researcher was found by his university to have falsified data in dozens of papers, which were retracted. Human trials have not shown meaningful benefit. It is the clearest example in this chapter of a compound whose reputation was built on marketing rather than evidence.

What this means for eating

In short: Eat a variety of whole plant foods, prepare them in ways that preserve or release what is in them, and be sceptical of extracts.

The consistent finding across nutritional epidemiology is that whole plant foods are associated with lower risk of essentially every major chronic disease. The consistent finding across supplement trials is that isolated compounds from those foods mostly do not reproduce it.

The practical rules that follow:

1. Variety over intensity. Thirty different plants a week beats large amounts of one superfood. Different compound classes, and probably synergy.
2. Colour is a proxy for compound class, so eating across the spectrum is a reasonable heuristic.
3. Preparation matters more than people think: fat with carotenoids, chop-and-wait or light steaming for brassicas, cooking for lycopene, raw for vitamin C and myrosinase.
4. Skins and outer leaves are where polyphenols concentrate. Peeling an apple removes most of its quercetin, and the outer leaves of a lettuce or cabbage are the most nutrient dense.
5. Be sceptical of extracts and "superfoods." The word superfood has no scientific definition and is banned as an unsubstantiated health claim on packaging in the EU. Exotic berries are not better than local ones; they are more profitable.
6. Some interactions are real. Grapefruit inhibits CYP3A4 and interacts dangerously with many drugs (Chapter 84). St John's wort induces the same enzyme and reduces the effectiveness of many drugs including contraceptives. High-dose green tea extract has caused liver injury. Concentrated plant compounds are drugs.

The bottom line

• Plants contain thousands of non-nutrient compounds, mostly defensive, and the foods containing them are consistently associated with better health.
• The simple antioxidant explanation failed in trials, sometimes with harm. Hormesis, mild stress provoking your own defences, is the better current model.
• Bioavailability is the recurring obstacle: what circulates in your blood is usually a metabolite at a fraction of the concentration used in laboratory studies, and gut bacteria do much of the transformation.
• Sulforaphane, lycopene, lutein and zeaxanthin, and dietary nitrate have the clearest mechanisms and best human data. Resveratrol is the cautionary tale.
• Eat varied whole plants, prepare them thoughtfully, and treat concentrated extracts as drugs rather than food.

Sources and notes

Compound classes and dietary distribution follow the Phenol-Explorer database and standard phytochemistry reviews. The antioxidant supplement trial record is the same ATBC, CARET, HOPE, and SELECT set as Chapter 15, plus the Cochrane antioxidant reviews. USDA withdrawal of the ORAC database, with its stated reasoning, was announced in 2012. Nrf2 activation by sulforaphane follows Talalay's and Fahey's work at Johns Hopkins; the Qidong broccoli sprout trial is Egner et al., Cancer Prevention Research, 2014. Polyphenol bioavailability and microbial transformation follow Manach et al., American Journal of Clinical Nutrition, 2005. Equol producer status follows Setchell's work. Curcumin bioavailability and its PAINS characterisation follow Nelson et al., Journal of Medicinal Chemistry, 2017. Lycopene bioavailability and cooking follows Gartner et al. and Rao's reviews. AREDS2 is the JAMA 2013 report. Dietary nitrate and nitric oxide follows Lundberg, Weitzberg, and Gladwin, Nature Reviews Drug Discovery, 2008, and the mouthwash abolition studies. The resveratrol research misconduct case is the University of Connecticut investigation and subsequent retractions.

Open questions. Which polyphenol metabolites are biologically active in humans, and at what concentrations, is largely unknown, which is why almost every mechanistic claim in this area outruns its evidence. Whether hormesis is a useful general framework or a post-hoc explanation is debated.

👉 Next: the natural toxins already in your food, because "natural" was never a safety category.

The Natural Toxins Already in Your Food

TL;DR. Ordinary food contains genuine poisons. Green potatoes contain solanine, kidney beans contain a lectin that causes violent vomiting if undercooked, cassava releases cyanide and has caused permanent paralysis in famine conditions, and apple pips and apricot kernels contain the same cyanide-releasing compound. Almost all of this is managed by ordinary cooking and ordinary quantities, and traditional food preparation techniques exist largely because of it. The separate category of antinutrients (phytate, oxalate, tannins) does not poison you; it blocks the absorption of minerals, which matters at the margins and is greatly reduced by soaking, fermenting, sprouting, and cooking.

Key takeaways

• "Natural" is not a safety category. The most toxic substances known are natural products.
• Raw or undercooked kidney beans cause severe vomiting and diarrhoea within hours, from phytohaemagglutinin. As few as four or five raw beans can do it. Slow cookers can make it worse by holding beans at a temperature that activates rather than destroys the lectin.
• Green or sprouted potatoes should be discarded, not just peeled, once green is extensive. Solanine poisoning is real and has killed.
• Antinutrients are a nuisance, not a threat, in mixed diets, and traditional preparation methods reduce them substantially.
• Nutmeg, liquorice, star fruit, and elderberries are the surprising domestic hazards.

The framing: dose, preparation, and quantity

Every point in this chapter is an application of one principle, covered properly in Chapter 87: toxicity is a property of dose, not of origin. Botulinum toxin, ricin, aflatoxin, tetrodotoxin, and amatoxin are all natural. Water is toxic at a high enough dose.

The second principle is that traditional food preparation is mostly toxicology. Soaking beans, fermenting cassava, nixtamalising maize, leaching acorns, curing olives, and boiling rather than steaming certain greens are all techniques that exist because the alternative made people ill. When cultures adopt a food without its processing, the results are historically grim: maize without nixtamalisation gave the American South pellagra; cassava without proper processing gives konzo.

Glycoalkaloids: green potatoes and tomato leaves

In short: Solanine and chaconine in potatoes are genuine neurotoxins that concentrate in green skin, sprouts, and damaged tissue.

Potatoes belong to the nightshade family and defend themselves with glycoalkaloids, principally solanine and alpha-chaconine. These inhibit acetylcholinesterase, the same enzyme organophosphate insecticides hit, and disrupt cell membranes.

Where they concentrate: the skin and the layer immediately beneath it (containing the large majority of the total), the sprouts and eyes, and any green tissue. Green colour is chlorophyll and is itself harmless; it is a marker for light exposure, and light exposure also drives glycoalkaloid synthesis. So green is a warning sign rather than the problem.

Levels: a normal potato contains roughly 10 to 100 mg/kg, and the accepted safety limit for commercial varieties is 200 mg/kg. Greened or sprouted potatoes can reach 500 to 1,000 mg/kg or more in the affected tissue.

Symptoms begin at roughly 1 to 3 mg per kg of body weight: nausea, vomiting, abdominal pain, and diarrhoea, progressing at higher doses to headache, confusion, hallucinations, fever, and, in severe cases, death. A famous 1979 incident at a London school hospitalised 78 boys after a meal of stored potatoes, several with serious neurological symptoms. Deaths are rare and documented.

Practical rules: store potatoes dark and cool but not refrigerated; discard sprouts and cut generously around them; peel and cut away any greening; and if a potato is extensively green, bitter, or heavily sprouted, throw it away. Cooking does not help much, because glycoalkaloids are heat-stable to about 170 °C. A bitter taste is the practical warning signal and should be respected.

Green tomatoes contain tomatine, a related glycoalkaloid that is considerably less toxic and is bound in the gut, which is why fried green tomatoes are a dish and not a hazard. Tomato leaves and stems are higher and are not eaten. Aubergine and peppers contain small amounts. The claim that "nightshades cause inflammation" in the general population is not supported by evidence.

Lectins: the kidney bean case

In short: Raw and undercooked red kidney beans cause violent gastrointestinal illness, and this is the one lectin story that is unambiguously real.

Lectins are proteins that bind carbohydrates, present in many plants and concentrated in legumes and grains. Most are destroyed by cooking and are of no consequence.

Phytohaemagglutinin in red kidney beans is the exception, and it is serious.

• Raw red kidney beans contain 20,000 to 70,000 haemagglutinating units; properly cooked beans contain 200 to 400.
• As few as four or five raw beans can cause illness.
• Onset is fast: nausea and vomiting within one to three hours, followed by diarrhoea and abdominal pain. Recovery is usually within a day, and it is thoroughly unpleasant.

The dangerous method is the slow cooker. Heating beans to 80 °C actually increases toxicity to around five times the raw level, and a slow cooker on low may never exceed that. Documented outbreaks trace to exactly this.

The correct method, per the US FDA and UK food standards guidance:

1. Soak dried beans for at least 5 hours (or overnight).
2. Discard the soaking water.
3. Boil hard in fresh water for at least 10 minutes, which destroys the lectin.
4. Then simmer until tender, or transfer to a slow cooker after the hard boil.

Canned kidney beans have been through this process industrially and are safe as sold.

White kidney beans (cannellini) contain roughly a third as much, and broad beans about 5 percent. Other pulses are much lower.

On the broader "lectin-free" diet trend, promoted in popular books: the claim that dietary lectins from cooked legumes, grains, and nightshades cause widespread chronic disease is not supported. Legume consumption is consistently associated with lower disease risk in cohort studies, and pulses are a staple in most of the world's longest-lived populations. The kidney bean hazard is real, specific, and solved by boiling.

Cyanogenic glycosides: cassava, kernels, and pips

In short: Several foods contain compounds that release hydrogen cyanide when the tissue is damaged and the compound meets an enzyme.

The mechanism mirrors the brassica one: a stable glycoside stored in one compartment and an enzyme in another. Crush the tissue and they meet, releasing hydrogen cyanide, which blocks the final step of cellular respiration.

Cassava is the serious one. It is a drought-tolerant, high-yielding root crop that feeds several hundred million people, and bitter varieties contain enough linamarin to be lethal without processing. Traditional processing (soaking, grating, fermenting for several days, pressing, and roasting or drying) reduces cyanide by 95 percent or more, and every step matters. Under famine conditions, when processing is shortened, the result is konzo: a sudden, permanent, symmetrical spastic paralysis of the legs, mainly affecting children and women of childbearing age, documented in outbreaks across Central Africa. It is entirely preventable and it is a direct consequence of shortcutting food processing under duress. A simple "wetting method" developed for African communities reduces residual cyanide further and has been shown to prevent konzo in field trials.

Apricot kernels are the modern rich-country version of the problem. They are sold as a health food and as an alternative cancer treatment (as "laetrile" or "vitamin B17," which is neither a vitamin nor effective; controlled trials found no benefit and real cyanide toxicity). EFSA has assessed that as few as three small kernels can exceed the safe level for an adult, and one for a toddler. Poisonings and deaths are documented. This is one of the clearest cases in the book of "natural" marketing causing direct harm.

Apple pips are the question everyone asks. An apple contains perhaps 5 to 8 pips with roughly 1 to 4 mg of amygdalin each, releasing a fraction of a milligram of cyanide, and only if crushed. A lethal dose would require deliberately chewing the pips of dozens to hundreds of apples in a sitting. Swallowing a few whole does nothing at all, as the intact seed coat passes through. Do not worry about it; do not make a habit of chewing them.

Oxalates

In short: They bind calcium and, in susceptible people, form kidney stones. Rhubarb leaves are genuinely dangerous; rhubarb stalks are not.

Oxalic acid binds calcium, magnesium, and iron into insoluble crystals. Two consequences: the mineral becomes unavailable, and in the urinary tract the crystals can form calcium oxalate kidney stones, which account for roughly 80 percent of stones.

Rhubarb leaves are the domestic hazard. Poisonings occurred during the First World War when the leaves were promoted as a vegetable in Britain. The stalks are fine.

For most people, oxalate is not worth managing. For recurrent calcium oxalate stone formers, it is, and the advice is counterintuitive:

• Drink much more water. By far the most effective intervention, aiming for 2.5 litres of urine a day.
• Eat calcium with oxalate-rich meals. Calcium in the gut binds oxalate and prevents its absorption. Restricting dietary calcium increases stone risk, which is the single most common mistake.
• Reduce sodium, which increases urinary calcium.
• Cooking helps: boiling spinach and discarding the water removes 30 to 87 percent of its oxalate. Steaming removes less.
• Do not take high-dose vitamin C, which is metabolised partly to oxalate.

Green smoothies deserve a specific warning. Blending large quantities of raw spinach, beet greens, or chard daily delivers far more oxalate than anyone would eat as a salad, and cases of oxalate nephropathy (acute kidney injury from oxalate crystals) have been reported. Rotate the greens.

Phytate

In short: The main storage form of phosphorus in seeds; it binds iron, zinc, and calcium and reduces their absorption. Soaking, sprouting, and fermenting reduce it substantially.

Phytic acid is present in all whole grains, pulses, nuts, and seeds. It binds positively charged minerals tightly enough to prevent absorption.

The effect is genuine and quantitatively significant: phytate can reduce iron absorption by 50 to 80 percent and zinc by 30 to 50 percent from a given meal. In populations depending on unleavened wholegrain bread with little animal food, this contributes to widespread iron and zinc deficiency. This is a real global nutrition problem.

In a mixed diet, it is a minor issue, and phytate has demonstrated benefits of its own: it is an antioxidant, it may reduce colorectal cancer risk, and it reduces urinary crystal formation.

How to reduce it, all of which are traditional techniques:

Genuine sourdough is the standout: the long acidic fermentation activates phytase and degrades most of the phytate, which is one concrete reason traditional sourdough delivers more available minerals than fast-fermented commercial bread made from the same flour.

Goitrogens

In short: Compounds that interfere with thyroid iodine uptake, mainly in brassicas, soy, and cassava. Relevant if iodine intake is low, otherwise not.

Brassica glucosinolates break down partly into thiocyanates, which compete with iodine for uptake into the thyroid. Soy isoflavones inhibit an enzyme in thyroid hormone synthesis. Cassava's cyanide is detoxified to thiocyanate, with the same effect.

The practical position: in people with adequate iodine, normal consumption of these foods does not cause thyroid problems, and studies in soy-consuming populations find no increase in thyroid disease. Cooking reduces goitrogenic activity in brassicas substantially. The combination that matters is low iodine plus high goitrogen intake, which is exactly the historical picture in cassava-dependent, iodine-deficient regions.

People on levothyroxine should note a different issue: soy and high-fibre foods reduce absorption of the drug, so take it on an empty stomach, four hours apart from soy or calcium or iron (Chapter 79).

The surprising domestic hazards

In short: A short list of ordinary foods that cause genuine problems in ordinary kitchens.

Nutmeg. Contains myristicin. Doses of roughly 5 to 15 grams (one to three whole nutmegs) cause a genuine and thoroughly unpleasant intoxication: hallucinations, agitation, dry mouth, tachycardia, nausea, and a hangover lasting days. Poisoning cases and rare deaths are documented. Culinary quantities are irrelevant.

Liquorice. Real liquorice contains glycyrrhizin, which inhibits the enzyme that inactivates cortisol in the kidney, producing an effect like excess aldosterone: sodium retention, potassium loss, high blood pressure, and in severe cases dangerous hypokalaemia and cardiac arrhythmia. Cases of hospitalisation and at least one death from heavy liquorice consumption are documented, and the EFSA suggests no more than 100 mg of glycyrrhizin a day, roughly 50 g of liquorice sweets. Most "liquorice" confectionery in the US and UK is flavoured with anise and contains none; the traditional black liquorice in Northern Europe does.

Star fruit (carambola). Contains caramboxin, a neurotoxin normally cleared by the kidneys. In people with chronic kidney disease it accumulates and causes hiccups, vomiting, confusion, seizures, and death. Fatalities are well documented. Anyone with significant kidney impairment should avoid star fruit entirely, and it is not widely known.

Elderberries. Raw berries, and especially the leaves, stems, and unripe fruit, contain cyanogenic glycosides and cause nausea, vomiting, and diarrhoea. Cooking destroys them, which is why elderberry is always used cooked. An outbreak in California in 1983 hospitalised people who drank juice made from raw berries with leaves and stems included.

Raw or undercooked red kidney beans, as above.

Raw egg white in quantity binds biotin (avidin); more relevantly it carries a small salmonella risk. In the UK, eggs bearing the British Lion mark are considered safe to eat raw or lightly cooked by the FSA, including for vulnerable groups, following the vaccination programme in laying hens. Elsewhere the advice differs.

Green almonds, bitter almonds, and unripe ackee. Ackee, Jamaica's national fruit, contains hypoglycin A and must be harvested only when the pods open naturally. Unripe ackee causes "Jamaican vomiting sickness": severe hypoglycaemia, vomiting, seizures, and death.

Wild mushrooms. Outside the scope of this chapter and worth one sentence: several species that kill are hard to distinguish from edible ones, Amanita phalloides accounts for most fatal mushroom poisonings worldwide, and the characteristic pattern is symptoms that subside after a day, giving false reassurance while the liver fails. Do not forage on app identification.

Fish and shellfish toxins. Scombroid poisoning from histamine in poorly refrigerated tuna and mackerel; ciguatera from reef fish in tropical regions, which is heat-stable and can last months; and paralytic shellfish poisoning from algal blooms. None of these are destroyed by cooking, which is the important point.

The bottom line

• Real food contains real toxins. "Natural" describes origin, not safety.
• The genuine domestic dangers are green and sprouted potatoes, undercooked red kidney beans (especially in slow cookers), rhubarb leaves, apricot kernels sold as health food, and, for people with kidney disease, star fruit.
• Traditional preparation methods (soaking, boiling, fermenting, nixtamalising) exist because of this chemistry, and skipping them has caused mass illness repeatedly in history.
• Antinutrients (phytate, oxalate, tannins) block mineral absorption rather than poisoning you. They matter in monotonous plant-based diets, and much less in varied ones. Soaking, sprouting, and fermenting cut them substantially.
• For recurrent kidney stone formers, drink far more water and keep dietary calcium up, not down.

Sources and notes

Glycoalkaloid levels, symptoms, and the 200 mg/kg limit follow EFSA's 2020 opinion on glycoalkaloids in potatoes and the FAO/WHO evaluations; the 1979 London school incident is McMillan and Thompson, Quarterly Journal of Medicine, 1979. Phytohaemagglutinin figures and the slow cooker warning follow the US FDA Bad Bug Book and UK FSA guidance. Cassava processing, cyanide reduction, and konzo follow Banea-Mayambu's and Howard Bradbury's field work and WHO reporting. Apricot kernel risk follows EFSA's 2016 opinion on cyanogenic glycosides in apricot kernels; laetrile trial failure is Moertel et al., New England Journal of Medicine, 1982. Oxalate contents follow Holmes and Kennedy's analyses and the Harvard oxalate content list; stone prevention advice follows the American Urological Association guideline, including the counterintuitive dietary calcium finding from Curhan et al. Phytate reduction by soaking, sprouting, and fermentation follows Gibson, Perlas, and Hotz. Goitrogen relevance follows Bajaj et al. and the bok choy myxoedema case report, New England Journal of Medicine, 2010. Nutmeg, liquorice, star fruit (caramboxin, Garcia-Cairasco et al.), ackee, and elderberry hazards follow published case series and the EFSA glycyrrhizin opinion.

Open questions. Whether moderate oxalate intake matters for people who have never formed a stone is unclear, and green smoothie nephropathy is documented only in case reports. The threshold at which brassica goitrogens matter in iodine-replete people has not been established.

👉 Next: when to eat, where chronobiology has more to say than either side of the meal-timing argument usually admits.

When to Eat

TL;DR. Your metabolism is not the same at 8am and 10pm. Insulin sensitivity is highest in the morning and falls through the day, so the same meal produces a larger glucose rise at night. Eating late is associated with worse metabolic outcomes in both observational and controlled work, and this is the best-supported timing finding there is. Intermittent fasting works about as well as ordinary calorie restriction and no better, which makes it a useful tool for people who find it easier to follow, not a metabolic trick. Almost everything else about meal timing (eating fruit on an empty stomach, not eating after 6pm, the anabolic window, six small meals) is folklore.

Key takeaways

• Insulin sensitivity is highest in the morning. An identical meal eaten in the evening produces a substantially higher glucose response in controlled studies.
• Late eating is associated with worse outcomes and, in tightly controlled crossover trials, with increased hunger and altered energy expenditure at identical calorie intake.
• Intermittent fasting produces weight loss equivalent to calorie restriction in head-to-head trials. The evidence for benefits beyond that is much thinner in humans than in mice.
• Meal frequency has little effect on metabolic rate. The "eat six small meals to stoke your metabolism" advice is not supported.
• Eating vegetables and protein before carbohydrate in the same meal measurably lowers the glucose response, which is the most practical timing finding in the chapter.

The body clock is a real thing

In short: Nearly every tissue runs a roughly 24-hour molecular clock, the master clock is set by light, and the peripheral clocks in liver and gut are set largely by food.

Your cells contain molecular oscillators: a feedback loop of clock genes (CLOCK, BMAL1, PER, CRY) that cycles over roughly 24 hours. The master clock sits in the suprachiasmatic nucleus of the hypothalamus and is synchronised by light hitting specialised retinal cells. Peripheral clocks sit in liver, pancreas, muscle, and gut, and they are entrained substantially by feeding time rather than by light.

That split is the key to the whole subject. Light sets the brain clock; food sets the metabolic clocks. Eat at times that conflict with your light-set clock and the two disagree, a state called internal desynchrony. This is what shift workers live in, and shift work is associated with higher rates of obesity, type 2 diabetes, and cardiovascular disease in consistent epidemiology, with the International Agency for Research on Cancer classifying night shift work involving circadian disruption as probably carcinogenic.

Things that follow a daily rhythm and matter here:

Melatonin is the mechanistic link that ties this together. Melatonin receptors are present on pancreatic beta cells, and melatonin suppresses insulin secretion. So eating a large meal close to bedtime, when melatonin is already rising, means a glucose load arriving exactly when insulin release is being damped. Carriers of a common variant in the melatonin receptor gene MTNR1B, which is one of the strongest known type 2 diabetes risk variants, show this effect more strongly, which is a satisfying piece of mechanistic convergence.

Late eating: the best-supported finding

In short: The same food eaten late produces higher glucose, and controlled trials show late eating increases hunger and changes energy expenditure at identical intake.

Controlled evidence. A 2022 crossover study by Vujović, Scheer and colleagues at Brigham and Women's Hospital is the cleanest demonstration. Participants ate identical meals, in identical amounts, with identical composition, either early or shifted four hours later, under strictly controlled laboratory conditions. Late eating:

• Increased hunger and altered the appetite hormones leptin and ghrelin in the direction of greater hunger.
• Reduced energy expenditure by a small but measurable amount.
• Shifted gene expression in fat tissue toward storage rather than breakdown.

All at identical calories. That is a rare and valuable design.

Other consistent findings:

• Identical test meals produce substantially higher glucose and insulin responses in the evening than in the morning, reproduced across many studies.
• Front-loading calories (larger breakfast, smaller dinner) produces greater weight loss than the reverse at the same total intake in several trials, though not all, and the effect size is modest.
• Night-eating patterns are associated in cohort studies with higher BMI, worse glycaemic control, and higher cardiovascular risk.
• Eating close to bedtime worsens reflux and, at large volumes, disrupts sleep, which has its own metabolic consequences.

Practical reading: finishing the main eating window two to three hours before bed is sensible and low cost. "Never eat after 6pm" is arbitrary; what matters is the relationship to your sleep, not the clock on the wall.

Intermittent fasting: what the trials actually show

In short: Roughly equivalent to continuous calorie restriction for weight and metabolic outcomes; the human evidence for unique benefits is much weaker than the mouse evidence.

The main protocols:

The head-to-head trials. When intermittent fasting is compared against continuous calorie restriction with matched calorie intake, results converge: similar weight loss, similar changes in blood pressure, lipids, and glycaemic markers. A well-designed 2022 trial in NEJM by Liu and colleagues found time-restricted eating plus calorie restriction produced weight loss no different from calorie restriction alone over 12 months. Multiple meta-analyses agree.

Where TRE alone (no calorie counting) has been tested, results are modest. The 2020 TREAT trial found a 16:8 window produced about 1 kg of weight loss over 12 weeks, no better than the control condition, and raised a concern about loss of lean mass. Other trials have found more benefit. The overall picture: some people spontaneously eat less within a shorter window, and some do not.

The mechanisms people cite, assessed:

• Autophagy. Cellular self-cleaning, upregulated by fasting. Real, important, and overwhelmingly studied in yeast, worms, and mice. The timing in humans is not well established, and the confident claims that autophagy "kicks in at 16 hours" are not based on human measurement.
• Ketosis. Real after roughly 12 to 16 hours of fasting depending on prior glycogen. Not in itself a health outcome.
• Metabolic switching between glucose and fat as fuel, promoted by Mark Mattson and colleagues as the core mechanism. Plausible, and the human outcome data do not yet distinguish it from simple calorie reduction.
• Circadian alignment. This is the most promising strand: early time-restricted eating (window in the morning, e.g. 8am to 4pm) has outperformed late windows in several controlled studies for insulin sensitivity and blood pressure, independent of weight. This suggests that when the window sits matters as much as how long it is, and that the popular pattern of skipping breakfast and eating late may be the worst version of TRE.

Who should not do it: people with a history of eating disorders (fasting protocols can trigger relapse, and this is not a minor caution), people who are pregnant or breastfeeding, underweight people, children and adolescents, people with type 1 diabetes or on insulin or sulfonylureas without medical supervision because of hypoglycaemia risk, and people taking medicines that must be taken with food.

The honest summary: intermittent fasting is a legitimate way to reduce calorie intake for people who find eating windows easier than counting. It is not metabolically magic. If you do it, an earlier window is better supported than a later one.

Meal frequency

In short: Total intake matters; how many meals you split it into matters very little.

The idea that frequent small meals raise metabolic rate rests on a misunderstanding of the thermic effect of food. The thermic effect is a percentage of what is eaten, so dividing the same food into six meals produces six small thermic bumps instead of three larger ones, and the total is identical. Controlled studies confirm no difference in 24-hour energy expenditure between two and six meals at matched intake.

What frequency does affect:

• Appetite control, individually and unpredictably. Some people find grazing prevents overeating; others find it prevents ever feeling satisfied. Trials show no consistent advantage either way.
• Glycaemic variability in people with diabetes, where more even distribution can help.
• Protein distribution, where three to four servings across the day beats one large one for muscle protein synthesis (Chapter 12).
• Dental health, where frequency genuinely matters: each exposure to fermentable carbohydrate produces an acid attack on enamel lasting 20 to 30 minutes, so six snacks are worse than the same food at three meals.
• Snacking on ultra-processed food, which is where the practical harm of frequent eating usually lives.

Meal order: the most useful practical trick

In short: Eating vegetables and protein before carbohydrate in the same meal substantially lowers the glucose response.

This one is well replicated and easy to apply. Studies in both healthy people and those with type 2 diabetes find that consuming vegetables and protein 10 to 15 minutes before the carbohydrate portion of a meal reduces the post-meal glucose peak by 30 to 40 percent compared with the reverse order, with the identical food.

The mechanism is a combination of slowed gastric emptying, increased GLP-1 release, and a physical barrier to rapid starch access.

Related, similarly supported interventions:

• Vinegar (a tablespoon in a dressing) with a starchy meal reduces the glucose response measurably, apparently by slowing gastric emptying and inhibiting some starch digestion.
• A 10 to 15 minute walk after a meal lowers the glucose peak substantially, because muscle contraction moves glucose transporters to the cell surface without insulin. Trials find short post-meal walks outperform a single longer walk elsewhere in the day for glucose control.
• Adding fat, protein, or fibre to a carbohydrate meal flattens the curve, which is why a slice of bread with butter and cheese behaves very differently from bread alone.

These are small effects individually and they are free, immediate, and cumulative.

Breakfast, and the argument about it

In short: The observational case for breakfast is strong and heavily confounded; the trial evidence is much weaker.

Breakfast eaters have lower BMI, better diet quality, and better cardiometabolic markers in cohort studies. They also smoke less, drink less, exercise more, and are more likely to be employed in regular hours. This is the standard problem.

Randomised trials assigning people to eat or skip breakfast find much smaller effects. The Bath Breakfast Project found breakfast eaters had higher total energy intake and more spontaneous physical activity, with no net weight difference. A 2019 BMJ meta-analysis of trials found that eating breakfast was associated with slightly higher total daily energy intake and slightly higher weight, though the trials were short and mostly small.

So: breakfast is not metabolically obligatory. It is genuinely useful for people who otherwise under-eat protein, for children (where the school-performance evidence is better), and for anyone who overeats later without it. And a large morning meal fits the insulin sensitivity rhythm better than a large evening one, so if you eat three meals, making breakfast or lunch the largest is the better-aligned choice.

Timing around exercise

In short: Total daily intake dominates; a few timing effects are real and small.

• Carbohydrate before endurance exercise improves performance in sessions over about 60 to 90 minutes. For shorter sessions it makes little difference.
• Fasted training increases fat oxidation during the session and does not increase fat loss over weeks when total intake is matched. It may reduce performance in hard sessions.
• Protein timing: as Chapter 12 set out, the anabolic window was oversold. Protein within a few hours either side of training is sufficient.
• Post-exercise carbohydrate matters for glycogen replenishment if you are training again within about eight hours. Otherwise, the day's total is what matters.
• Caffeine 30 to 60 minutes before exercise has good evidence as a performance aid (Chapter 88).

The myths

In short: A short list of timing claims with nothing behind them.

"Eat fruit only on an empty stomach, or it rots in your stomach behind other food." Physiologically incoherent. The stomach is a mixing chamber at pH 2; nothing rots in it. Fruit eaten with a meal is digested perfectly well, and eating it after protein and vegetables produces a lower glucose response, not a problem.

"Never combine protein and carbohydrate." Food combining diets have been tested and produce no advantage beyond the calorie reduction that comes from the restriction itself. Your pancreas secretes the full enzyme set for every meal.

"Carbohydrates after 6pm turn to fat." Fat storage depends on energy balance, not the clock. The genuine finding, that late eating produces higher glucose responses and is associated with worse outcomes, is about the whole meal and about circadian alignment, not about carbohydrates specifically.

"You must eat within 30 minutes of waking." No.

"Fasting puts you in starvation mode and wrecks your metabolism." Metabolic rate does fall with sustained weight loss (adaptive thermogenesis, a real and well-measured effect of roughly 10 to 15 percent below predicted), but this is a response to weight loss and energy deficit over weeks, not to skipping a meal. Short fasts slightly increase resting metabolic rate, via noradrenaline.

"Detox by fasting." Your liver and kidneys detoxify continuously. Fasting does not accelerate it, and prolonged fasting slightly impairs some liver detoxification pathways, which is the reverse of the claim.

The bottom line

• Metabolism follows a daily rhythm. Insulin sensitivity is highest in the morning and lowest at night, and melatonin suppressing insulin release is a large part of why.
• Late eating is the best-supported timing problem, demonstrated in controlled crossover trials at identical calorie intake. Finishing two to three hours before bed is a reasonable, cheap rule.
• Intermittent fasting matches calorie restriction and does not beat it. If you use it, an earlier window is better supported than a later one, and it is unsuitable for several groups including anyone with a history of disordered eating.
• Meal frequency barely affects metabolic rate. It affects appetite unpredictably and dental health predictably.
• The most useful practical timing tricks are meal order (vegetables and protein first), vinegar with starchy meals, and a ten-minute walk afterwards.

Sources and notes

Circadian physiology and peripheral clock entrainment by feeding follow Panda's and Bass's reviews. Diurnal variation in insulin sensitivity and the melatonin-insulin interaction follow Poggiogalle et al. and the MTNR1B diabetes genetics literature. The controlled late-eating crossover is Vujovic et al., Cell Metabolism, 2022, from Brigham and Women's Hospital. Intermittent fasting versus continuous restriction follows Liu et al., New England Journal of Medicine, 2022, and the TREAT trial, Lowe et al., JAMA Internal Medicine, 2020. Early time-restricted eating benefits follow Sutton et al., Cell Metabolism, 2018. Metabolic switching is Mattson's framework. Meal frequency and energy expenditure follows Bellisle et al. Meal order effects follow Shukla et al., BMJ Open Diabetes Research and Care, 2015 and 2017. Vinegar and glycaemic response follows Johnston's work. Post-meal walking follows Reynolds et al., Diabetologia, 2016. Breakfast trials follow the Bath Breakfast Project (Betts et al.) and the BMJ meta-analysis by Sievert et al., 2019. Anabolic window revision follows Schoenfeld and Aragon's meta-analyses.

Open questions. Whether intermittent fasting has benefits independent of calorie reduction in humans is unresolved, and the autophagy claims that circulate rest almost entirely on animal work with no established human timing. Long-term trials of early time-restricted eating are still small.

👉 That is the machinery. Now the food itself, starting with the fruit that has an entire proverb attached to it: apples.

Apples

TL;DR. The apple is the most engineered fruit in the temperate world: a self-sterile clone grafted onto a size-controlling rootstock, thinned by hand, picked hard, and held in a low-oxygen room for up to a year. Nutritionally it is water, sugar, and about 4 g of fibre, with most of its polyphenols in the skin, which is why peeling one throws away a disproportionate share of what makes it interesting. The proverb overstates the case; the cohort data supporting apple and whole-fruit intake against cardiovascular disease and type 2 diabetes is genuinely decent. Eat the skin, do not juice it, and pick a variety by what you are doing with it.

1. What it is

Malus domestica, a member of the rose family (Rosaceae), which makes it a close relative of pears, quinces, cherries, plums, peaches, apricots, almonds, strawberries, and raspberries.

Botanically an apple is a pome, not a berry, and its structure is unusual: the flesh you eat is not derived from the ovary wall but largely from the swollen hypanthium, the base of the flower. The actual ovary is the papery core containing the pips. In other words, most of an apple is a modified flower cup rather than a fruit wall in the strict sense.

There are more than 7,500 named cultivars worldwide, of which perhaps 100 are grown commercially at any scale and perhaps 15 dominate global trade.

2. Where it comes from

Origin: the Tian Shan mountains of Kazakhstan. The wild ancestor is Malus sieversii, which still grows in forests around Almaty (a city whose name derives from the word for apple). Genetic work led by Amandine Cornille and colleagues traced the domestic apple to that species, with substantial later introgression from the European crab apple Malus sylvestris as the fruit travelled west along the Silk Road, which is why modern apples are genetically a hybrid mosaic.

Domestication is unusual because the apple resisted the normal process. Seeds do not breed true (Chapter 5), so nothing could be fixed until grafting was invented, in China and the Near East, several thousand years ago. Every significant apple variety in history is a chance seedling or a deliberate cross that someone then cloned.

The Romans spread grafted apples across Europe. Monastic orchards preserved varieties through the medieval period. Colonists took seeds, not grafts, to North America, producing the cider orchards of folklore. Commercial variety development accelerated in the nineteenth century and industrialised in the twentieth.

Who grows them now (approximate, FAO early-2020s):

The southern hemisphere producers (Chile, New Zealand, South Africa, Argentina, Brazil) matter out of proportion to their volume, because they supply northern markets in the northern spring and summer.

3. How it is grown

Climate. A temperate crop with a firm chilling requirement: most varieties need 800 to 1,200 hours below about 7 °C each winter to break dormancy evenly (Chapter 3). They tolerate winter lows to -30 °C or lower when dormant and are destroyed by -2 °C in bloom. They want a cool climate with warm, sunny, relatively dry summers: heat above about 30 °C reduces red colour development and causes sunburn, and humidity drives scab and mildew.

The great apple regions all share the same profile: Washington State's Columbia basin (dry, irrigated, sunny, low disease pressure), the South Tyrol, Poland's central plain, New Zealand's Hawke's Bay, Chile's central valley, and Shaanxi in China.

Soil. Deep, well-drained loam, pH 6.0 to 7.0. Apples are notably intolerant of waterlogging, which kills roots and invites Phytophthora collar rot. Apple replant disease is a specific and important problem: replanting apples where apples grew before produces stunted trees, from a soil biological complex built up by the previous orchard. Fumigation, long rotations, or replant-tolerant rootstocks (the Geneva series) are the answers.

Water. Roughly 600 to 800 mm across the season. Timing is critical: cell division ends about 40 days after bloom and sets the final fruit size ceiling, so early-season water stress permanently limits size. Late-season water swings cause splitting and bitter pit.

The trees. Grafted, always. A modern commercial orchard is a tall spindle or fruiting wall: dwarfing rootstock (M9, or increasingly Geneva 41 or 935), 2,500 to 4,000 trees per hectare, trained flat on a wire trellis about 3 metres tall, with permanent posts and irrigation lines. First commercial crop in year 2 to 3, full production by year 5 to 6, economic life 15 to 25 years, yields of 50 to 100 tonnes per hectare in good systems.

Pollination. Apples are self-incompatible. An orchard needs a second, compatible variety flowering at the same time, or dedicated crab apple pollinisers, plus bees to move the pollen (one to three hives per hectare). Triploid varieties (Bramley, Jonagold, Mutsu) produce sterile pollen and require two other varieties in the block. Five days of cold rain during bloom can cost the year's crop.

Thinning. After the natural June drop, growers still remove 60 to 80 percent of the fruitlets, chemically during bloom and then by hand, aiming at roughly one fruit per cluster. Without it: small, pale, low-sugar fruit and biennial bearing, where the exhausted tree crops heavily one year and barely at all the next.

Pests and disease. Apple scab is the dominant fungal problem in humid climates and drives most of the spray programme; powdery mildew and fire blight follow. Codling moth is the principal insect and is now widely managed by pheromone mating disruption rather than insecticide. Apple maggot, aphids, and mites round out the list.

Harvest. Hand-picked into bags and bins, one fruit at a time, with a slight upward twist so the stem stays on the fruit (a stemless apple rots from the hole). Maturity is judged by the starch-iodine test: cut the fruit, apply iodine, and the remaining starch stains black. A fruit for immediate sale is picked at a later stage than one destined for nine months of storage. Firmness (penetrometer) and sugar (refractometer) confirm.

Storage. This is where apples are exceptional. Controlled atmosphere storage at 0 to 2 °C with oxygen at 1 to 3 percent holds apples for 6 to 12 months (Chapter 8), often with a 1-MCP treatment to block ethylene. An apple bought in June in the northern hemisphere was very likely picked the previous September. It is paused, not stale.

4. What is inside it

Per 100 g of raw apple with skin (roughly two thirds of a medium apple):

A medium apple (about 180 g) supplies roughly 95 kcal and 4.4 g of fibre with the skin, which is a genuinely useful contribution toward a 30 g daily target.

The skin is where the interest is. It holds most of the fibre and the great majority of the polyphenols: quercetin glycosides (apples are one of the richest dietary sources of quercetin), catechins and procyanidins, chlorogenic acid, and, in red varieties, anthocyanins. Peeling an apple removes roughly a third of the fibre and a substantially larger share of the polyphenols. Red-skinned and older varieties tend to be higher; a study of heritage varieties found several times the polyphenol content of some modern commercial ones.

Pectin deserves its own mention. Apples are one of the richest common sources, at roughly 1 to 1.5 g per 100 g, concentrated in the flesh and core region. It is a soluble, highly fermentable, gel-forming fibre, and it is the reason apples set jam and the reason apple is a traditional remedy for both diarrhoea (it gels) and constipation (it holds water and ferments).

Malic acid provides the tartness, at roughly 0.3 to 1 percent, higher in cooking varieties. Bramley's Seedling is essentially a malic acid delivery system with fruit attached.

5. What it does in your body

In short: A slow-release sugar package delivered inside intact cell walls, with soluble fibre that gels, feeds bacteria, and traps bile acids.

Eaten whole, an apple's sugars are behind cell walls and mixed with pectin, so absorption is gradual. Its glycaemic index is around 36 and a medium apple's glycaemic load is about 6, which is low. The same apple juiced has the cell walls destroyed, the fibre removed, and produces a much larger and faster glucose rise.

The classic demonstration is a 1977 study by Haber and colleagues: the same apples eaten whole, as purée, and as juice produced progressively faster eating, progressively higher insulin responses, and progressively less satiety, with the juice group showing a rebound hypoglycaemia that the whole-apple group did not. Physical form again.

Pectin reaches the colon intact, where it is fermented into short-chain fatty acids, mainly acetate and butyrate. It also binds bile acids, which the liver must then replace using circulating cholesterol, a mechanism that contributes to the modest LDL-lowering effect seen in apple trials.

Polyphenols are absorbed partially, metabolised extensively, and the majority reach the colon where bacteria transform them. Apple polyphenols have shown effects on endothelial function in short-term human trials.

Fructose is about 60 percent of an apple's sugar, which is a higher proportion than most fruit. In whole fruit form, at roughly 10 g per apple with fibre alongside, this is not the problem that high-fructose beverages are (Chapter 11). It is, however, why apples are a common trigger in fructose malabsorption and are high-FODMAP.

6. What the evidence actually shows

Established: apples contribute fibre, potassium, vitamin C, and polyphenols, and whole fruit intake in general is associated with better health outcomes across essentially every large cohort study.

Strong: apple and pear intake specifically has been associated with lower risk of type 2 diabetes in pooled analyses of the large US cohorts (Nurses' Health Study, Health Professionals Follow-up Study), with roughly a 20 percent lower risk at higher intakes, while fruit juice went the other way. Higher intake of white-fleshed fruit and vegetables (apples and pears dominate this category) was associated with lower stroke risk in a Dutch cohort, with roughly a 50 percent difference across the intake range, which is a large effect and observational.

Mixed: cholesterol lowering. Controlled trials of whole apples or apple pectin generally find small LDL reductions, in the range of a few percent. Real, modest.

Thin: essentially every specific claim about apples treating a named condition. Apple cider vinegar deserves its own note below.

"An apple a day keeps the doctor away" dates from a Welsh proverb recorded in the 1860s ("Eat an apple on going to bed, and you'll keep the doctor from earning his bread") and was tested, semi-seriously, in a 2015 JAMA Internal Medicine analysis of national survey data. Apple eaters used marginally fewer prescription medications, and after adjusting for confounders there was no significant difference in doctor visits. The authors' conclusion was that "evidence does not support that an apple a day keeps the doctor away," which is a fine piece of scientific deadpan. The proverb is directionally sensible and specifically false.

Apple cider vinegar is worth separating from apples. There is modest, replicated evidence that vinegar (any vinegar, via acetic acid) reduces the post-meal glucose response to a starchy meal by slowing gastric emptying and inhibiting some starch digestion, and a small Japanese trial found a modest weight difference over 12 weeks. The evidence for the rest of the claims (detox, cure-all, substantial weight loss) is absent. The genuine harms are real: undiluted vinegar erodes tooth enamel, has caused oesophageal burns, and ACV tablets have caused throat injury. If you use it, dilute it, drink it through a straw, and expect a small effect.

7. Who should eat more, who should be careful

Good for: almost everyone. Children (a whole apple is a genuinely convenient fibre-and-water package), people managing weight (high satiety per calorie), people with constipation or with diarrhoea (pectin works both ways), and anyone short of fibre, which is most people.

Be careful if:

• Irritable bowel syndrome or fructose malabsorption. Apples are high-FODMAP: excess fructose plus sorbitol. They are one of the most common IBS triggers, and this is a genuine and frequent problem rather than a theoretical one. Low-FODMAP alternatives are citrus, berries (in moderation), kiwi, and grapes.
• Oral allergy syndrome. Very common in people with birch pollen allergy: the Mal d 1 protein in apple is structurally similar to the birch allergen Bet v 1, producing itching and tingling of the mouth and lips within minutes of eating raw apple. It is usually mild. Because the protein is heat-labile, cooked apple is generally tolerated, which is a useful diagnostic clue. Older varieties (Santana, Elstar, and several heritage apples) are measurably lower in Mal d 1 and are marketed for this. A minority of people, particularly in Mediterranean countries, react to a different, heat-stable protein (Mal d 3), and those reactions can be systemic and serious.
• Diabetes. Apples are fine and are associated with lower risk. The apple juice is not.
• Dental erosion. Malic acid plus sugar, and slow eating, means prolonged acid exposure. Do not brush immediately after; rinse with water and wait 30 minutes.
• Warfarin. Apples are low in vitamin K and not a problem, unlike leafy greens.

Ages. From about 6 months as purée or, in baby-led weaning, as soft cooked pieces. Raw apple is a choking hazard for under-4s because a bitten-off chunk is exactly the wrong shape and hardness; grate or cook it. Whole raw apples are excellent for older children and adults. For older adults with dentures or swallowing difficulty, stewed apple retains the pectin and most of the nutrition.

Apple pips. Contain amygdalin, which releases cyanide when crushed. Swallowing a few whole is harmless because the seed coat passes through intact. Do not deliberately chew and eat large quantities (Chapter 18).

8. When and how to eat it

• Eat the skin. This is the single most consequential choice. Peeling removes about a third of the fibre and most of the polyphenols.
• Whole beats sauce beats juice, in that order, for satiety, glucose response, and fibre.
• Any time of day. The claim that fruit must be eaten on an empty stomach is nonsense (Chapter 19). Eating an apple after the savoury part of a meal produces a lower glucose response than before it.
• Before a meal works well for appetite: a 2009 trial found an apple eaten before lunch reduced total meal energy intake more than an equivalent amount of apple juice or apple sauce.
• With something fatty or protein-rich (nut butter, cheese, yoghurt) blunts the glucose rise further and increases satiety substantially. This is a genuinely good snack combination.
• Cooking softens the fibre, destroys most of the vitamin C, and increases the availability of some polyphenols. Stewed apple with the skin on is a good option for anyone who cannot manage raw.
• Browning is enzymatic (polyphenol oxidase acting on the polyphenols once oxygen reaches them). Lemon juice, salt water, or cold water slow it. It affects appearance and a little vitamin C, not safety.
• Wash it, particularly if unpeeled and particularly if waxed. A 30-second rub under running water removes most surface residue and dirt; baking soda solution removes somewhat more (Chapter 6).

9. Choosing, storing, and ripening

Choosing. Firm with no soft spots or wrinkles. Heavy for its size. The skin should feel taut, not slippery-slack. Russeting (a rough brown patch or netting) is cosmetic and often indicates a good variety. Bruises are soft brown patches beneath intact skin and mean the tissue has broken down there.

Storing. Apples are climacteric and emit substantial ethylene.

• Refrigerate them. Apples last several times longer in the fridge than on the counter, four to six weeks against about a week, and lose crispness far more slowly.
• Keep them away from other produce, particularly potatoes (they sprout), leafy greens, and unripe fruit that you do not want ripened.
• A perforated bag in the crisper is ideal, because they also lose water.
• Do not store bruised apples with sound ones, since damaged tissue floods ethylene.

Ripening. A commercial apple is picked at eating maturity or after, so there is no home ripening to do. If an apple is starchy and bland, it was picked immature and will not improve much, though a few days at room temperature helps slightly.

10. The varieties worth knowing

Cooking vs eating is a real distinction and it is about acid and cell structure. A high-acid apple with weakly bonded cells (Bramley) collapses into a smooth purée, which is what you want for a sauce or a crumble. A dense, lower-acid apple (Granny Smith, Braeburn, Golden Delicious) holds its shape, which is what you want in a tart. Using the wrong one produces either soup or crunchy tart filling.

On Red Delicious, because it is the best cautionary tale in fruit breeding: over decades, growers selected sports that were redder, more uniform, and better shipping, and the flavour and texture degraded until the variety became a byword for beautiful, disappointing fruit. It was the dominant US apple for most of the twentieth century and has been overtaken by Gala. A purely cosmetic selection pressure produced exactly what you would predict.

11. Myths and confusions

Don't be confused: apple juice, apple cider, and apple cider vinegar are three different things. Juice is filtered and pasteurised. Cider in North America is unfiltered, unpasteurised juice; in Britain and Ireland the same word means the alcoholic drink. Vinegar is cider fermented twice, to alcohol and then to acetic acid. Only the last has any supporting evidence for a metabolic effect, and only a small one.

• "Apple seeds will poison you." No, unless you deliberately crush and eat a large quantity (Chapter 18).
• "The wax on apples is dangerous." It is food-grade carnauba or shellac replacing the natural wax removed by washing. It is edible. Shellac is insect-derived, so waxed apples are not vegan, which is the genuinely surprising fact.
• "Supermarket apples are a year old and therefore worthless." They may be up to a year from harvest, and controlled atmosphere storage nearly stops the metabolism. Vitamin C declines by roughly a quarter to a third over long storage and everything else is broadly intact. The realistic alternative in June is not a fresh local apple.
• "Organic apples are much better for you." Slightly fewer synthetic residues, slightly more polyphenols, no demonstrated health outcome difference (Chapter 7). Apples appear near the top of residue-count lists, and the measured residues remain far below the levels used to set safety limits.
• "Apple cider vinegar burns fat." No.
• "Apples clean your teeth." They stimulate saliva, which helps, and they deliver acid and sugar, which does not. Net effect is not protective; rinse afterwards.
• "Brown apple flesh has gone bad." Enzymatic browning is oxidation of polyphenols and is cosmetic.

12. The bottom line

• An apple is a clone on a rootstock, thinned by hand, picked by the starch test, and often paused in a low-oxygen room for months. Nothing about it is accidental.
• Nutritionally it is water, about 10 g of sugar, 4 g of fibre in a medium fruit, some potassium and vitamin C, and a large and variable dose of polyphenols concentrated in the skin.
• Eat it whole and unpeeled. Purée is worse, juice is much worse, and the difference is measurable in glucose response and satiety.
• The cohort evidence for apples and whole fruit against type 2 diabetes and stroke is genuinely good. The proverb is not, and apple cider vinegar does much less than advertised.
• Choose the variety for the job: acidic and collapsing for sauce, dense and firm for tarts and eating.
• Fridge, not fruit bowl, and away from the potatoes.

Sources and notes

Composition figures are from USDA FoodData Central; production figures are approximate FAOSTAT values for the early 2020s. Apple domestication and Malus sieversii ancestry follow Cornille et al., PLoS Genetics, 2012. Orchard practice, rootstock series, starch-iodine maturity testing, and controlled atmosphere storage follow extension publications from Cornell, Washington State, and East Malling, and USDA Agriculture Handbook 66. Polyphenol content and skin concentration follow Boyer and Liu, Nutrition Journal, 2004, and the Phenol-Explorer database. The whole-versus-puree-versus-juice insulin and satiety comparison is Haber et al., The Lancet, 1977. Apple and pear intake and type 2 diabetes is Muraki et al., BMJ, 2013; white-fleshed fruit and stroke is Oude Griep et al., Stroke, 2011. The proverb test is Davis et al., JAMA Internal Medicine, 2015. Apple cider vinegar and postprandial glucose follows Johnston et al., Diabetes Care, 2004; the enamel erosion and oesophageal injury reports are from dental and case literature. Mal d 1 and Mal d 3 allergen behaviour follows Fernandez-Rivas' and Vieths' work on birch-related and LTP-mediated food allergy.

Open questions. How much the polyphenol differences between heritage and modern cultivars matter to health has not been tested in humans. The apple cider vinegar literature is small, mostly short-term, and mostly on glycaemic markers rather than outcomes.

👉 Next: bananas and plantains, the fruit whose entire global trade rests on one sterile clone.

Bananas and Plantains

TL;DR. The banana is a giant herb, not a tree, and its fruit is a sterile clone propagated by cutting for centuries. Every export banana on Earth is essentially the same plant, which is why a soil fungus destroyed the entire trade once already and is currently doing it again. Nutritionally a banana is a convenient package of potassium, vitamin B6, and a starch that converts to sugar as it ripens, so a green one and a spotted one are metabolically different foods. Plantains are the same genus eaten as a starchy staple, and they feed far more people than dessert bananas do.

1. What it is

Musa species, mostly hybrids of Musa acuminata (the A genome) and Musa balbisiana (the B genome). The dessert banana is triploid AAA; most plantains are AAB or ABB.

The plant is a giant herbaceous perennial, the largest flowering herb on Earth. What looks like a trunk is a pseudostem made of tightly wrapped leaf bases; there is no wood. The true stem is an underground corm, and new plants ("suckers" or "pups") emerge from it.

Botanically the fruit is a berry. Commercial bananas are parthenocarpic and sterile: they develop without fertilisation and contain no viable seeds, which is why the little black specks are aborted ovules rather than seeds. Wild bananas are full of hard, pea-sized seeds and are barely edible.

Don't be confused: banana and plantain is a culinary distinction, not a botanical one. There is no clean genetic line between them. In most of Latin America and Africa, the split is starchy-cooking versus sweet-dessert. In South and Southeast Asia, where hundreds of cultivars exist, the distinction barely makes sense.

2. Where it comes from

Origin: Southeast Asia and New Guinea, with archaeological evidence of banana cultivation in the Kuk Swamp of Papua New Guinea dating back roughly 7,000 years, making it one of the oldest domesticated crops. It spread westward through the Indian Ocean to Africa (Austronesian seafarers carried it to Madagascar and thence inland), and to the Americas with Portuguese and Spanish traders in the sixteenth century.

The modern export trade was built in the late nineteenth century by companies like United Fruit, whose political influence in Central America gave the world the phrase "banana republic." It is one of the few fruits whose history is genuinely a political history.

Who grows them now (approximate, FAO):

Uganda, Rwanda, Ghana, and the Democratic Republic of Congo are among the largest plantain and cooking-banana producers, where they are a genuine calorie staple: per capita consumption in parts of Uganda exceeds 200 kg a year.

3. How it is grown

Climate. Strictly tropical to subtropical: 26 to 30 °C ideal, and growth stops below about 14 °C. Frost kills. Needs 1,800 to 2,500 mm of rainfall a year or heavy irrigation, and high humidity. No chilling requirement, which is the reason it is available year-round.

Soil. Deep, well-drained, fertile loam, pH 5.5 to 7.0. Bananas are exceptionally hungry for potassium (a hectare removes hundreds of kilograms a year) and nitrogen. Waterlogging is fatal, and Panama disease is a soil-borne fungus, so drainage matters twice.

Wind is the great enemy. The pseudostem is soft and the leaves are enormous; a strong wind shreds the leaves (reducing photosynthesis) or topples the whole plant. Hurricanes routinely destroy entire national crops, which is a structural feature of the Caribbean banana economy.

The plant's life cycle is unusual and worth understanding:

1. A sucker is planted, or tissue-cultured plantlets are used (now standard for disease-free stock).
2. It grows for 9 to 12 months, producing 30 to 40 leaves.
3. The flower stalk emerges from the centre of the pseudostem and bends downward.
4. Female flowers develop into fruit, arranged in "hands" of 10 to 20 fingers, with 5 to 15 hands per bunch. A bunch weighs 30 to 50 kg.
5. The pseudostem fruits once and dies. It is cut down after harvest.
6. A selected sucker from the corm becomes the next plant, and the cycle repeats. A single corm ("mat") can crop for 10 to 30 years.

Management. Bunches are covered with perforated blue polythene bags to protect from insects, wind scarring, and spray, and to raise temperature slightly. Plants are propped with poles or twine because a loaded bunch will topple them. De-suckering keeps one follower per mat.

Disease is the defining issue. Black Sigatoka, a leaf fungus, requires spraying up to 40 to 50 times a year in wet tropical conditions, which is among the most intensive fungicide programmes in agriculture and a serious occupational exposure issue. Panama disease (Fusarium wilt) is the existential one, covered below.

Harvest and shipping. Cut deep green at about 75 percent maturity, washed, cut into clusters, boxed, and moved into refrigerated ships at 13 to 14 °C in a low-ethylene controlled atmosphere. Two to four weeks later, at the destination, they enter sealed ripening rooms and are dosed with ethylene at 15 to 20 °C for 24 to 48 hours, then delivered at whatever point on the seven-stage colour scale the retailer ordered (Chapter 8).

4. What is inside it

Per 100 g of raw dessert banana (a medium banana is about 118 g peeled):

Vitamin B6 is the standout, and it is under-advertised relative to potassium: one banana supplies roughly a quarter to a third of the daily requirement.

Potassium is over-advertised. A banana has about 400 mg, which is good and not exceptional: a baked potato has roughly double, and so do a cup of white beans, a cup of cooked spinach, or a serving of dried apricots.

Ripeness transforms it. As a banana ripens, amylase converts starch to sugars:

So a green banana is a slow-release resistant starch food and a spotted one is a fast sugar food. The calorie count barely changes; the metabolic behaviour changes a great deal.

Plantains, per 100 g raw: about 122 kcal, 32 g carbohydrate (mostly starch), 2.3 g fibre, 499 mg potassium, and notably 1,127 IU of vitamin A activity from carotenoids, which dessert bananas lack. Cooked, they behave like potatoes.

5. What it does in your body

The banana's practical properties come from three things.

Ripeness-dependent carbohydrate. Green banana starch is largely resistant starch type 2, which passes to the colon and is fermented into short-chain fatty acids (Chapter 14). Green banana flour is used clinically in some settings for exactly this. Ripe banana is quick fuel, which is why it is the standard cycling and running snack.

Potassium and blood pressure. Potassium counteracts sodium's effect on blood pressure by promoting sodium excretion and relaxing vessel walls (Chapter 16). A banana contributes about 10 percent of a day's need.

Pectin and resistant starch for bowel function. Bananas are a component of the traditional BRAT diet (bananas, rice, apple sauce, toast) for diarrhoea, and there is genuine evidence: green banana or banana flour has been shown in trials in Bangladesh to reduce the duration of persistent diarrhoea in children, via short-chain fatty acids promoting sodium and water absorption in the colon.

Not a serotonin source for your brain. Bananas contain tryptophan and small amounts of serotonin and dopamine. Serotonin does not cross the blood-brain barrier, and the tryptophan dose is small relative to a mixed meal's other amino acids, which compete for the same transporter. The "bananas make you happy" claim is not supported by any mechanism that works.

6. What the evidence actually shows

Established: bananas are a good source of B6 and a useful source of potassium, fibre, and rapidly available energy. Green banana resistant starch is fermented and produces SCFAs.

Strong: the general association between whole fruit intake and lower cardiovascular disease and type 2 diabetes risk includes bananas. The potassium-blood pressure relationship is well established at the nutrient level.

Mixed: green banana for diarrhoea in children has good trial support in specific settings; generalisation is less clear.

Thin to absent: claims that bananas improve mood, prevent cramps specifically, or aid sleep. The cramp claim is worth its own note: exercise-associated muscle cramps are now generally attributed to neuromuscular fatigue rather than to electrolyte depletion, and trials of electrolyte supplementation for cramps are unimpressive. Eating a banana for cramps is harmless and probably not doing what people think.

7. Who should eat more, who should be careful

Good for: endurance athletes (a banana performs about as well as a sports drink for carbohydrate during exercise, with fibre and potassium included), children, people with low potassium intake, and anyone who needs a portable food with no packaging problem.

Be careful if:

• Chronic kidney disease, stage 3 and above. Damaged kidneys cannot excrete potassium efficiently and hyperkalaemia causes dangerous cardiac arrhythmias. Bananas are among the first foods restricted. This applies also to people on ACE inhibitors, ARBs, or potassium-sparing diuretics (Chapter 75).
• Latex-fruit syndrome. People allergic to natural rubber latex frequently cross-react to banana, avocado, kiwi, and chestnut, because of structurally similar chitinase proteins. Reactions can be serious.
• IBS. Ripe bananas are higher in FODMAPs (fructose and fructans increase with ripening); unripe bananas are lower and are permitted on low-FODMAP protocols. This is the reverse of what most people assume.
• Diabetes. Fine in whole form; prefer a firmer, less ripe banana, and pair with protein or fat.
• Migraine. Bananas are on some trigger lists (tyramine), with weak evidence.

Ages. Excellent early weaning food from around 6 months: soft, sweet, no preparation, no common allergen. Whole raw banana pieces are soft enough not to be a major choking hazard, unlike apple. Good throughout life; particularly useful for older adults with poor appetite or chewing difficulty.

8. When and how to eat it

• Match ripeness to purpose. Green to yellow-green for slow energy and resistant starch; fully yellow for eating; spotted for baking, smoothies, and freezing.
• Before or during endurance exercise is where it shines: portable, rapidly absorbed, with potassium and B6.
• Pair with protein or fat (nut butter, yoghurt) to blunt the glucose rise, especially for a very ripe banana.
• Freeze overripe bananas peeled for smoothies and baking. Do not freeze them in the skin unless you enjoy dealing with the result.
• Plantains must be cooked. Green plantain fries as tostones or chips; yellow-black ripe plantain fries sweet as maduros; both boil and mash like potato.
• The peel is edible and eaten in parts of South Asia in curries, particularly the tender inner peel of young fruit. It is high in fibre and potassium. Wash it thoroughly given the spray regime.

9. Choosing, storing, and ripening

Choosing. Buy at the ripeness you need in the next few days, since they move fast. Avoid fruit with a bruised or split skin, or with a grey, dull cast, which indicates chilling injury. Green tips with a yellow body is normal.

Storing. This is the classic case where the fridge is wrong.

• Never refrigerate unripe bananas. Below about 13 °C they suffer chilling injury: the peel blackens and, more importantly, the ripening enzymes are damaged so the fruit never ripens properly. It stays starchy inside a black skin.
• Once fully ripe, the fridge is fine and useful: the peel will blacken cosmetically while the flesh holds for several more days.
• Bananas emit a lot of ethylene. Keep them away from anything you do not want ripened. This is the single biggest cause of a fruit bowl going over all at once.
• Separating the fingers and wrapping the crown in cling film slows ripening a little, by limiting ethylene release from the cut stem. The effect is real and modest.

Ripening. Put them in a paper bag, with an apple if you are impatient. Warmth speeds it.

10. The varieties worth knowing

The Cavendish problem. Every Cavendish banana is a genetically identical clone, so a pathogen that beats one beats all of them. Tropical Race 4 of Fusarium oxysporum f. sp. cubense does exactly that: identified in Taiwan in the 1990s, it has spread through Southeast Asia, Australia, the Middle East, Mozambique, and reached Colombia in 2019 and Peru in 2021. It lives in soil for decades, spreads on boots, tools, and water, and there is no fungicide for it. This is precisely the same failure mode that destroyed Gros Michel, running a second time with full advance warning.

The responses in progress: quarantine and biosecurity, conventional breeding of resistant hybrids (slow, because the plants are sterile), tissue-culture selection, and a genetically modified Cavendish carrying a resistance gene from a wild banana, which was approved in Australia in 2024. Whether consumers accept a GM banana when the alternative is no banana is about to be tested in public.

11. Myths and confusions

• "Bananas are the best source of potassium." Moderate at best. Potatoes, beans, spinach, dried apricots, and avocado all beat them per serving.
• "Bananas are radioactive and dangerous." They contain potassium-40, as does all potassium, including the potassium inside you. The "banana equivalent dose" is a teaching device for putting radiation figures in context, not a hazard. Your own body's potassium gives you far more.
• "Bananas cause constipation." Unripe ones can, through tannins and resistant starch. Ripe ones generally do not.
• "Bananas are fattening / too sugary." About 105 kcal for a medium fruit, with fibre. Whole fruit intake is associated with lower diabetes risk, not higher.
• "Bananas help you sleep." The melatonin and magnesium doses are small. No good evidence.
• "Green bananas are unripe and inferior." They are a different food with a different and arguably better metabolic profile.

12. The bottom line

• The banana is a giant sterile herb, propagated by cutting, and the entire export trade is a single clone facing a soil fungus with no cure. This is the most fragile major food supply chain on Earth.
• Nutritionally: notable for vitamin B6, useful for potassium and fibre, and metabolically transformed by ripeness. Green is slow-release resistant starch; spotted is quick sugar.
• Never refrigerate an unripe banana; refrigerate a ripe one. Keep them away from everything else in the fruit bowl.
• Plantains are the same genus as a starch staple and feed far more people than dessert bananas do. They need cooking.
• Restricted in advanced kidney disease and a common cross-reactant in latex allergy.

Sources and notes

Composition figures are from USDA FoodData Central; production figures are approximate FAOSTAT values for the early 2020s. Kuk Swamp domestication evidence is Denham et al., Science, 2003. Cultivation practice, bunch management, and Black Sigatoka spray intensity follow FAO and Bioversity International banana documentation. The Gros Michel collapse and Tropical Race 4 spread follow Ploetz's reviews in Phytopathology and Plant Disease and FAO's TR4 reporting; the GM Cavendish approval in Australia was announced by OGTR in 2024. Ripening room protocols and the seven-stage colour scale are industry standard. Green banana starch composition changes with ripening follow Zhang et al.'s resistant starch analyses. Green banana for persistent childhood diarrhoea is Rabbani et al., Gastroenterology, 2001, from Bangladesh. Latex-fruit syndrome follows Blanco's reviews. Exercise-associated cramp aetiology follows Schwellnus' neuromuscular fatigue work, which displaced the electrolyte explanation. Potassium restriction in CKD follows KDIGO guidance.

Open questions. Whether a resistant-starch-rich green banana diet has meaningful metabolic benefits beyond its fibre contribution has not been established in well-powered trials. Whether TR4-resistant GM or conventionally bred cultivars will be commercially accepted in time is genuinely unknown.

👉 Next: citrus, which includes the one fruit that can genuinely kill you through a drug interaction.

Citrus

TL;DR. Almost every citrus fruit you can buy descends from just three or four wild ancestors, hybridised and back-crossed for two thousand years: the orange is a mandarin-pomelo cross, the lemon is a citron-sour orange cross, the grapefruit is a pomelo-orange cross. They are the classic vitamin C source and their peel is where the aromatic oils and most of the polyphenols live. Grapefruit is the one fruit in this book with a genuine, documented, occasionally fatal drug interaction: it inhibits the enzyme that metabolises dozens of common medicines, and the effect lasts a day or more.

1. What it is

Genus Citrus, family Rutaceae. The fruit is a hesperidium: a modified berry with a leathery, oil-glanded rind and interior segments filled with juice vesicles, which are individual hair-like sacs of juice. That structure is unique to citrus.

The astonishing genetics: genomic work published in Nature in 2018 by Wu and colleagues established that essentially all cultivated citrus derives from a small number of wild ancestors:

From which:

• Sweet orange = a complex mandarin × pomelo hybrid, roughly 3 parts mandarin to 1 pomelo
• Sour (Seville) orange = a different mandarin × pomelo hybrid
• Grapefruit = pomelo × sweet orange, arisen in Barbados in the eighteenth century
• Lemon = citron × sour orange
• Key lime = citron × papeda
• Persian lime = a triploid involving key lime and lemon
• Clementine = mandarin × sweet orange
• Tangelo = mandarin × grapefruit or pomelo

Citrus hybridises readily and many types are apomictic (producing seeds that are clones of the mother), which is why chance seedlings occasionally produce stable new types.

2. Where it comes from

Origin: the eastern Himalayan foothills, spanning northeast India, northern Myanmar, and southwest China, from where the genus radiated about 8 million years ago. Domestication began in China and Southeast Asia; the citron reached the Mediterranean first (by around 300 BCE), followed by sour oranges and lemons via Arab trade routes in the tenth century, and sweet oranges via Portuguese traders in the fifteenth.

Citrus shaped naval history. Scurvy killed more sailors than combat until James Lind's 1747 trial and the eventual Royal Navy adoption of lemon, then lime, juice rations, which is the origin of "limey" as a term for a British sailor (Chapter 15).

Who grows them now (approximate):

Brazil and Florida built the frozen concentrated orange juice industry; Florida's has been devastated by citrus greening disease (huanglongbing), covered below. Spain and Morocco supply most of Europe's fresh fruit.

3. How it is grown

Climate. Subtropical evergreen trees with no chilling requirement. They want warm summers, mild winters, and no frost. Cold tolerance varies sharply:

A single hard freeze can kill an orchard. The Florida freezes of the 1980s pushed the state's citrus industry substantially south.

Cool nights are needed for colour. Peel colour comes from chlorophyll breaking down to reveal carotenoids underneath, and that breakdown requires cool nights. Oranges grown in consistently warm tropical climates stay green while being perfectly ripe inside, which is why commercial packers degreen them with ethylene, and in some countries dye them (Chapter 8).

Soil. Well-drained, pH 6.0 to 7.5. Extremely intolerant of waterlogging, which invites Phytophthora root and collar rot. Sensitive to salinity, which is a growing problem in irrigated arid regions.

Rootstocks are everything. Almost all commercial citrus is budded onto a rootstock chosen for disease resistance, cold hardiness, salinity tolerance, and soil adaptation: trifoliate orange and its hybrids (Carrizo, Swingle citrumelo) are common. The historical lesson is tristeza virus, which killed millions of trees on sour orange rootstock worldwide in the twentieth century and forced a global rootstock change.

Harvest. Citrus is non-climacteric: it does not ripen after picking, so it must be harvested ready to eat. Sugar and acid are measured as a Brix-to-acid ratio, and legal minimum maturity standards exist in most producing regions. Fruit can be "stored on the tree" for weeks or months in many varieties, which is a useful commercial flexibility.

The disease that is destroying the industry. Huanglongbing (citrus greening), caused by a bacterium spread by the Asian citrus psyllid, has no cure. Infected trees produce small, lopsided, bitter, green fruit and decline over a few years. It has devastated Florida, where production has fallen by more than 90 percent from its peak, and it is present in Brazil, China, and much of Asia. Responses include psyllid control, antibiotic trunk injection, and transgenic resistance research. It is the citrus equivalent of the banana's Race 4 and substantially further advanced.

4. What is inside it

Per 100 g of edible portion:

A medium orange (about 130 g) supplies roughly 70 mg of vitamin C, comfortably a full day's requirement.

Beyond vitamin C:

• Flavanones: hesperidin (oranges and mandarins), naringin (grapefruit, which is where the bitterness comes from), eriocitrin (lemons). Concentrated in the pith and membranes, which is why juice contains far less than whole fruit.
• Limonoids: limonin and nomilin, bitter triterpenes. Limonin is why juice from some varieties turns bitter hours after squeezing, as a precursor converts in acid.
• Essential oils in the rind: predominantly d-limonene, plus linalool, citral (the lemon note), and dozens more. The rind is a completely different ingredient from the flesh and delivers most of the aroma.
• Carotenoids: beta-cryptoxanthin in oranges and mandarins (a provitamin A carotenoid, and mandarins are one of the best dietary sources), lycopene in pink and red grapefruit, and beta-carotene.
• Pectin, heavily concentrated in the peel and pith: commercial pectin for jam-making is made overwhelmingly from citrus peel and apple pomace.

5. What it does in your body

Vitamin C is the headline and it does more than prevent scurvy: it is the essential cofactor for collagen synthesis, and it converts dietary non-haem iron to the absorbable ferrous form. That second point is the most actionable: a glass of orange juice or a squeeze of lemon over a lentil or spinach dish can multiply the iron you absorb from it several-fold (Chapter 16).

Citric acid binds calcium in urine and raises urinary citrate, which inhibits crystal formation. This is a genuine, clinically used effect: lemonade therapy (typically 120 mL of concentrated lemon juice diluted daily) is a recognised adjunct for recurrent calcium oxalate kidney stones, and potassium citrate is the pharmaceutical version.

Flavanones (hesperidin, naringenin) have shown modest effects on endothelial function and blood pressure in trials, generally small.

Limonene from the peel has been studied for a range of effects with mostly preliminary evidence.

Acidity and teeth. Citric acid at pH around 2.2 to 2.4 (lemon and lime juice) dissolves tooth enamel directly. This is the most under-appreciated harm in this chapter: habitual lemon water, sucking on lemon slices, and sipping citrus drinks through the day produce documented, irreversible dental erosion. Dentists see it constantly.

6. What the evidence actually shows

Established: vitamin C prevents and cures scurvy. Citrate raises urinary citrate and reduces calcium oxalate stone recurrence. Vitamin C substantially enhances non-haem iron absorption.

Strong: citrus intake is part of the general fruit-and-vegetable association with lower cardiovascular disease and stroke. Beta-cryptoxanthin intake is associated with lower inflammatory arthritis risk in cohort studies.

Mixed: flavanone supplementation effects on blood pressure and endothelial function, which are small and inconsistent. Grapefruit for weight loss: one small 2006 trial found a modest weight difference, and it has not been convincingly replicated.

Thin to absent: vitamin C preventing colds (it does not reduce incidence in the general population; it shortens duration by roughly 8 percent, about half a day Chapter 15). Lemon water for detox, alkalising the body, or boosting metabolism: no mechanism and no evidence. Your blood pH is held between 7.35 and 7.45 by mechanisms nothing you eat can override, and the "alkaline diet" rests on a misunderstanding of that.

7. Who should eat more, who should be careful

Good for: almost everyone. Particularly useful for people relying on plant iron sources, for recurrent calcium oxalate stone formers, and as a low-FODMAP fruit option for people with IBS (oranges, mandarins, lemons, and limes are all low-FODMAP, which makes citrus one of the better fruit choices in that context).

Be careful if:

• You take medication. See the grapefruit section immediately below. This is the single most important safety item in the fruit half of this book.
• You have reflux (GORD). Citrus is a common trigger through direct mucosal irritation.
• You have dental erosion or sensitivity. Drink citrus with meals rather than sipping through the day, use a straw, rinse with water afterwards, and do not brush for 30 minutes after, because brushing softened enamel removes it.
• You take a proton pump inhibitor or antacids. No interaction, but reflux patients often need to know the trigger.
• Phototoxicity. Citrus peel oils, especially bergamot, lime, and to a lesser extent lemon, contain furocoumarins that cause phytophotodermatitis: handling limes and then going into the sun produces a burn and long-lasting brown pigmentation in the pattern of the splashes. "Margarita burn" is a real dermatological presentation, and it is often mistaken for a chemical burn or abuse.
• Citrus allergy exists and is uncommon; oral allergy syndrome to citrus is more common.

Ages. Fine from around 6 months, though acidity commonly causes a harmless perioral rash in babies. Whole segments are a choking risk for toddlers if the membranes are tough; cut them small. Excellent through life.

8. Grapefruit and the drug interaction

In short: Grapefruit inhibits an intestinal enzyme that normally destroys much of an oral drug dose before it reaches the blood, so blood levels of dozens of medicines can rise sharply, and one glass can do it for over 24 hours.

This deserves its own section because it is the clearest example in the whole book of a food behaving as a drug.

The mechanism. Many oral drugs are partly destroyed on their first pass through the intestinal wall and liver by the enzyme CYP3A4 (Chapter 62). Grapefruit contains furanocoumarins, principally bergamottin and 6',7'-dihydroxybergamottin, which irreversibly inactivate intestinal CYP3A4. Because the inactivation is irreversible, the effect lasts until the cells make new enzyme, which takes 24 to 72 hours. Spacing the grapefruit and the tablet by a few hours does not avoid it.

Grapefruit also inhibits the OATP transporters that carry some drugs in, which produces the opposite effect: reduced absorption and reduced efficacy for a different set of medicines.

The size of the effect can be extreme. For some drugs, a single glass of grapefruit juice increases blood levels several-fold. Terfenadine, a former antihistamine, was withdrawn partly because interactions of this kind caused fatal cardiac arrhythmias.

Drug classes affected (this is illustrative, not exhaustive; always check the leaflet):

Which fruits are involved. Grapefruit and pomelo are the main ones. Seville (bitter) oranges and their marmalade contain furanocoumarins too, as do tangelos and limes to a lesser degree. Sweet oranges, mandarins, clementines, satsumas, and lemons do not significantly inhibit CYP3A4 and are safe alternatives.

The practical rule: if you take a prescription medicine, check the patient information leaflet for grapefruit, ask a pharmacist, and if the answer is to avoid it, avoid it entirely rather than trying to time it. The list of affected drugs is long and grows.

9. Choosing, storing, and ripening

Choosing. Heavy for size is the single best test, because weight means juice. Thin, fine-textured skin generally indicates more juice than thick, coarse skin. Firm with slight give; avoid soft spots, which indicate rot beginning. Colour is unreliable: a green orange can be fully ripe, and colour may be the result of degreening or dye. Loose, puffy skin on mandarins is normal for some types (satsumas) and indicates over-maturity in others.

Storing. Citrus is non-climacteric and only degrades.

• Room temperature: about a week. Fridge: two to four weeks, in the crisper.
• Refrigeration is better for storage life. Bring to room temperature before eating or juicing, because cold fruit yields less juice and tastes less aromatic.
• Grapefruit is the most chilling-sensitive citrus and can develop pitting below about 10 °C over long periods.
• Cut citrus goes in the fridge, covered, and lasts two to three days.
• Zest and juice freeze well. Zest citrus before juicing, always; the zest is the more valuable part and it is impossible to recover afterwards.

10. The varieties worth knowing

11. Myths and confusions

Don't be confused: the pith is not waste. The white spongy layer holds most of a citrus fruit's flavanones and a large share of its fibre. It is bitter, and the standard advice to remove it entirely throws away much of what is beneficial. Eating an orange with some pith attached is better than juicing it.

• "Lemon water alkalises your body." Your blood pH is defended within 0.1 of a unit by respiration and kidney function. Nothing you drink changes it, and if it did you would be critically ill. Lemon juice does make your urine more alkaline, which is the source of the confusion and is not the same thing.
• "Vitamin C prevents colds." It does not reduce incidence in the general population.
• "Orange juice is as good as an orange." It has the fibre removed, most of the flavanones lost with the pith, and a much faster sugar delivery. Whole fruit is associated with lower diabetes risk; juice with higher.
• "Fresh juice is always better than from concentrate." Nutritionally they are similar once reconstituted. Both are still juice.
• "Citrus is bad for you if you have a cold sore / arthritis / an ulcer." No good evidence for any of these. Acidity may irritate an existing mouth ulcer directly.
• "You should avoid citrus at night." No, unless you get reflux.

12. The bottom line

• Nearly all citrus is a hybrid of mandarin, pomelo, citron, and papeda, cross-bred over two thousand years.
• One orange covers the day's vitamin C. The bigger practical value is that vitamin C multiplies the iron you absorb from plant foods eaten with it.
• Citrus does not ripen after picking, so buy it heavy for its size and eat it. Store in the fridge, zest before juicing.
• Grapefruit, pomelo, and Seville oranges inhibit CYP3A4 for a day or more and interact seriously with dozens of common drugs. Sweet oranges, mandarins, and lemons do not.
• Citric acid erodes tooth enamel. Habitual lemon water is a real dental problem, and the detox and alkalising claims behind it are false.

Sources and notes

Composition figures are from USDA FoodData Central; production figures are approximate FAOSTAT values for the early 2020s. Citrus phylogeny and hybrid origins follow Wu et al., Nature, 2018. Cultivation, rootstock history, tristeza, and huanglongbing follow University of Florida and University of California citrus extension publications and USDA reporting on Florida production decline. Scurvy history follows Lind's A Treatise of the Scurvy, 1753, and Carpenter, The History of Scurvy and Vitamin C, 1986. Citrate and kidney stone prevention follows the American Urological Association guideline and Kang et al.'s lemonade therapy work. Flavanone content and its concentration in pith follow Phenol-Explorer. Dental erosion from citric acid follows Lussi's erosion literature. The grapefruit interaction follows Bailey, Dresser, and Arnold, Canadian Medical Association Journal, 2013, which lists the affected drug classes and the irreversible CYP3A4 mechanism; furanocoumarin identification follows Guo and Yamazoe's work. Vitamin C and colds follows the Cochrane review by Hemila and Chalker. Phytophotodermatitis follows dermatology case literature.

Open questions. The full list of drugs affected by grapefruit continues to grow and is not comprehensively characterised; the practical rule is therefore to check per drug rather than rely on any list. Flavanone health effects in humans remain small and inconsistent across trials.

👉 Next: berries, the most nutrient-dense fruit per calorie and the most fragile in the supply chain.

Berries

TL;DR. Berries are the most nutrient-dense fruit per calorie you can buy: low sugar, high fibre, high vitamin C, and by far the highest concentration of anthocyanins, the purple-red pigments that give them their colour. They are also the most fragile thing in the produce aisle, non-climacteric so they never improve after picking, and consequently expensive and often disappointing. Frozen berries are picked riper and frozen within hours, which makes them nutritionally equal or better and far cheaper. Cranberry for urinary tract infection prevention has real but modest evidence; almost every other specific berry claim is thinner than it sounds.

1. What it is

A commercial category rather than a botanical one. Almost none of these are berries in the strict sense.

Meanwhile bananas, grapes, tomatoes, avocados, and chillies are botanical berries, which tells you how little the word constrains anything.

2. Where it comes from

Strawberry is the youngest major fruit: the modern garden strawberry was created in Brittany in the 1750s by an accidental cross between Fragaria virginiana from eastern North America and Fragaria chiloensis from Chile, brought back by the French spy and engineer Amédée-François Frézier. Wild European strawberries were tiny by comparison.

Blueberry is one of the very few major fruit crops domesticated in North America and in the twentieth century: Elizabeth White and Frederick Coville selected wild highbush plants in New Jersey and released the first cultivars in 1916.

Raspberries and blackberries were gathered across Eurasia and North America for millennia and cultivated from the medieval period.

Cranberry was a staple of Indigenous peoples of northeastern North America, used for food, dye, and medicine, and commercially cultivated from the early 1800s.

Production (approximate):

Peru's blueberry industry went from nothing to global leadership in about a decade, which is one of the fastest agricultural transformations on record.

3. How it is grown

Strawberries are herbaceous perennials grown as annuals in most commercial systems. Planted on raised beds through black plastic mulch (warms the soil, suppresses weeds, keeps the fruit clean), with drip irrigation underneath. They propagate by runners, which are stolons producing daughter plants, so nurseries multiply plants vegetatively. Types are June-bearing (one heavy crop, short-day), everbearing, and day-neutral (fruiting continuously regardless of day length), and it was day-neutral varieties that made year-round supply possible. Increasingly grown in polytunnels or glasshouses on raised tabletop systems at waist height, which is a labour cost decision: picking accounts for the majority of production cost, and picking from a bent-over position is slow.

Blueberries demand something no other common crop does: acid soil, pH 4.5 to 5.5. In neutral soil they cannot take up iron and go chlorotic and die (Chapter 2). Commercial growers acidify with elemental sulphur, grow in pine bark or peat beds, or grow in pots with tailored substrate. They are shallow-rooted, need consistent moisture, and most cultivars need 800 to 1,000 chill hours, though low-chill southern highbush cultivars opened up warm regions and drove the Peru and Morocco booms. Bushes are productive for 15 to 20 years. Bumblebees are better pollinators than honeybees because blueberry flowers respond to buzz pollination.

Raspberries and blackberries grow on canes with a two-year cycle: primocanes grow vegetatively in year one, become floricanes and fruit in year two, then die and are cut out. Primocane-fruiting ("autumn") raspberries fruit on first-year canes, which simplifies management enormously. They are trained on wire trellises and need cane thinning.

Cranberries are the strange one: low creeping vines on sandy, acidic, peat-based bogs. They are not grown underwater. The bogs are flooded at harvest because cranberries have four internal air chambers, so they float: the bed is flooded, a machine beats the vines to dislodge the fruit, and the floating berries are corralled and pumped out. Flooding is also used for winter frost protection and pest control.

Harvest. All berries are hand-picked for the fresh market, into the punnet the consumer will buy, because a second handling destroys them. Machine harvesting exists for processing fruit and blueberries. Strawberries are picked every two to three days across a season.

Post-harvest. Berries are among the most perishable produce there is: high respiration rate, thin skin, no protective cuticle to speak of. They are precooled by forced air within an hour of picking, held at 0 °C at 90 to 95 percent humidity, and often shipped in modified atmosphere with elevated CO₂ (15 to 20 percent) which suppresses Botrytis grey mould. Grapes and some berries travel with sulphur dioxide pads. Even so, shelf life is days.

4. What is inside it

Per 100 g raw:

Points worth noticing:

• Raspberries and blackberries are among the highest-fibre foods in the whole produce aisle, at 5 to 6.5 g per 100 g, more than double most fruit. Much of that fibre is in the seeds.
• Strawberries beat oranges on vitamin C per 100 g, which surprises most people. Blackcurrants beat everything, at roughly three times an orange.
• Berries are the lowest-sugar fruit category. A cup of strawberries has under 8 g of sugar; a banana has 14 g.
• Anthocyanins are the defining compound class: water-soluble pigments that shift colour with pH, red in acid and blue-purple in alkaline. This is why blueberry pancake batter turns green where baking soda touches it.
• Ellagitannins in raspberries, blackberries, and strawberries are converted by gut bacteria into urolithins, and urolithin A specifically has become a research target for mitochondrial function. Only some people's microbiomes produce it, which is the same producer-status problem as soy and equol (Chapter 17).
• Proanthocyanidins (PACs) in cranberries, specifically the A-type linkage, are the compounds behind the urinary tract effect.

5. What it does in your body

Anthocyanins are poorly absorbed (typically under 1 percent as the parent compound), rapidly metabolised, and largely transformed by colonic bacteria into smaller phenolic acids. Whatever they do, they mostly do it as metabolites. Measured effects in human trials include improved endothelial function (blood vessel dilation) within hours of a high-anthocyanin meal, and modest reductions in blood pressure and LDL over weeks.