🌿 Choosing Varieties

In field and market language, sweet cassava means a variety low enough in cyanogenic glucosides to be eaten after simple cooking, such as boiling or roasting, without extended detoxification. Bitter cassava means a variety that requires extended processing, meaning grating, pressing, fermenting and drying, before it is safe to eat.

Read those two definitions again and notice what they are about. Neither one is about flavour on the plate. Both are about what must happen to the root before a person swallows it. The words were borrowed from taste because taste was the tool farmers had. The meaning is safety.

This matters because the word sweet does damage when it is heard as a preference. A trader who tells you a variety is sweet is not describing dessert. They are making a claim that this root can go from soil to pot without the long processing chain. If that claim is wrong, someone eats a lightly cooked root carrying more cyanogenic potential than their body can handle. Konzo, the paralytic disease caused by chronic cyanide exposure from badly processed cassava combined with a protein-poor diet, is real, documented, and has affected tens of thousands of people, overwhelmingly children and women of childbearing age, in central and southern Africa. That history is why this course refuses to soften the point.

Now the honest limits of the taste test. The sweet and bitter split is a genuinely useful working distinction for farmers, and this course does not tell you to abandon it. But it is not a fixed laboratory cutoff. A Ugandan study of local and improved varieties found that bitter taste correlated with measured cyanogenic glucoside level but was not identical to it. Correlated but not identical is the exact phrase to hold onto. Bitterness points in the right direction and it does not give you a number.

Worse, the number itself moves. Cyanogen content in the same variety varies with soil, with drought stress, with plant age, and even with which part of the root is tested. So a variety that tasted mild on your neighbour's field last season is not guaranteed to behave the same way on your field this season after a dry spell.

There is a historical grading scale worth knowing because you will meet it. A classification from Bolhuis in 1954, cited in the FAO review, grades fresh peeled root as innocuous below 50 milligrams of HCN per kilogram, moderately poisonous at 50 to 100 milligrams per kilogram, and dangerously poisonous above 100 milligrams per kilogram.

Teach yourself to hold that scale correctly. It is one historical reference scale for raw peeled root. It is not today's regulatory line for a food you sell. The practical target repeatedly cited for a finished, ready-to-eat cassava food product is 10 milligrams of HCN equivalent per kilogram, which is far stricter than the innocuous band on the Bolhuis scale.

Why the difference? Because finished-product safety and raw-root safety are two different measurements. Processing removes only part of the total cyanogenic potential. A root that would sit in the innocuous band raw is not automatically a finished product that meets a 10 milligram per kilogram target, and a root far above the innocuous band can be brought down by a full processing chain. The raw number and the sold number are not the same number, and they answer different questions.

One more point of honesty about that 10 milligram figure. It is real and it appears repeatedly across the literature, but the reference this course is built on could not trace it to one single clause of one single primary document in this pass. Treat it as the number every source agrees on rather than a clause you can quote chapter and verse. The actual enforcement authority is your national standards body, and residual-cyanide testing before you sell a processed product is set by that authority, not by any course.

So what do you do today? Use the taste check as a real, useful first filter that tells you which varieties your community has learned to trust for direct cooking. Then treat it as a filter and not a certificate. Any variety intended for sale as a fermented or dried product goes through the full process regardless of how it tasted, and any product you sell is subject to whatever testing your national food-safety authority requires. Ask them what that is.

There is one sourced field comparison in the reference for this course that does more work than any general statement about sweet and bitter, and it is worth learning in detail.

One Ugandan study measured cyanogenic potential across six local and six improved varieties at thirteen months of maturity. Here is what it reported.

Among the six local varieties, the lowest was Nyar-udota at roughly 88.5 milligrams of HCN equivalent per kilogram and the highest was Nyar-papoga at roughly 175 milligrams per kilogram. The reported mean across all local varieties is given as 201.65 milligrams per kilogram.

Among the six improved varieties, the lowest was NASE 19 at roughly 101.8 milligrams per kilogram and the highest was NASE 14 at roughly 116.5 milligrams per kilogram. The reported mean across all improved varieties is given as 108.75 milligrams per kilogram.

Notice something before going further. The reported mean for the local varieties, 201.65, sits above the highest named local variety in the same table, 175. This course reports both figures as the source gives them and does not quietly reconcile them. Treat the named-variety figures as the named-variety figures and the means as the study's own summary values, and if you need to build on either one, go to the study itself. Manufacturing a tidier number would be exactly the habit this course exists to break.

Now the two findings that matter for your decisions.

The first. Improved varieties in this Ugandan sample were, on average, markedly lower in cyanogenic potential than the local varieties tested, 108.75 against 201.65 milligrams per kilogram, which works out at roughly 46 percent lower. That is a large gap and it is one of the strongest practical arguments for improved material that this reference contains. Note carefully what it is not: it is not a claim that improved always means safer everywhere. It is six improved varieties against six local varieties in Uganda at thirteen months.

The second finding is the one every learner should carry for life. Every one of these twelve varieties exceeded the cited 10 parts per million reference value in the raw, unprocessed state. All twelve. Including the ones people would call the sweeter ones. Including the lowest, at roughly 88.5, which is still nearly nine times the cited finished-product reference value.

Sit with that. It is the single clearest sourced illustration in this whole course that no cassava variety should be treated as safe to eat raw. Not the local favourite, not the improved release, not the one your grandmother always boiled. The improved varieties were lower, and lower is genuinely better, and lower was still far above the reference value in the raw state.

This does not mean everyone eating boiled cassava in Africa is being poisoned. Cooking does work, processing does work, and the reference value applies to the finished product a person actually eats, not to the raw root. What it means is that the safety comes from the processing step, not from the variety being inherently harmless. Remove the processing and the variety will not save you.

The same study also found that cyanogenic content peaked at 8 to 10 months of plant maturity before declining by month 13. So the twelve figures above are figures at thirteen months, which is towards the lower end of that variety's own curve. Harvest at nine months and the same variety may be carrying more. Harvest age is a cyanide-relevant variable, not just a yield variable, and this is under-quantified across varieties generally.

How should you use a study like this? Not by looking for NASE 14 in your own market, because a variety name from a Ugandan study is not a recommendation for your district. Use it three ways instead. First, as proof that improved breeding programmes can and do shift cyanogenic potential downward in a measurable way, which is a reason to take improved material seriously. Second, as proof that no variety earns an exemption from processing. Third, as a model of what a real variety comparison looks like: named varieties, a stated maturity age, a stated country, and a measured range rather than a slogan. When somebody offers you variety advice, ask for those four things. If they cannot give them, you have an opinion, not evidence.

Improved is a word that gets sold to farmers hard, so let us be precise about what it means and what it does not.

An improved cassava variety is one released by a breeding programme, typically IITA in Ibadan or a national root-crop research institute such as NRCRI in Nigeria, NARO in Uganda, CSIR in Ghana or ZARI in Zambia, after selection for particular traits. The traits selected for in the cassava programmes covered by this course's reference are cassava mosaic disease resistance, more recently cassava brown streak disease tolerance, higher root yield and higher dry-matter content.

So improved means improved for those specific things, in the environment where the selection was done, against the disease strains present at that time. It does not mean improved in every direction at once. It does not mean immune. And as Lesson 2 showed, it does not mean the roots are safe raw.

Now the yield question, which is where the argument for improved material is usually made, and where the numbers need careful reading. The reference gives yield figures from many contexts and they spread enormously. Traditional smallholder methods are cited at 5 to 12 tonnes per hectare in one synthesis and 5 to 8 in another. Improved varieties, well managed, are cited at 40 to 60 tonnes per hectare, with another source giving above 30 tonnes per hectare as a stated attainable ceiling with improved management.

That gap is commonly a three to six fold difference. It is enormous, and it is real, and it is also the most misread pair of numbers in cassava.

Here is the misreading. A seller shows you 40 to 60 and your current yield of 8, and the implied promise is that the variety is worth five times your harvest. It is not, on its own. The gap is attributable to a combination of variety choice, planting material quality, spacing, fertiliser, weed control and pest and disease pressure. Not any single lever alone. The 40 to 60 figure describes improved varieties with improved management, and the second half of that phrase is doing at least as much work as the first.

The honest habit to build is this. For any yield figure quoted to you, by an input seller, a cooperative or a government pamphlet, ask which conditions it assumes. Which variety, which country, which year, monocrop or intercrop, fertilised or not, weeded to what standard, national average or on-farm trial. A figure without those conditions attached is advertising.

For comparison, cited national and regional averages sit far below the improved-management ceiling. Nigeria averaged 14.7 tonnes per hectare in 1995 to 1999, Ghana 13.1 and Cote d'Ivoire 10.8 in the same period. The African continental average for 2014 is cited at 8.38 tonnes per hectare, with a range across leading African producer countries of 7.72 to 23.36. The world average cited in the FAO post-harvest compendium is under 10, with Thailand at roughly 15 and South Africa at roughly 50. Those averages include every badly weeded, unfertilised, disease-hit field in the country, which is exactly why they are so far below trial figures.

Now the case for local varieties, which is not sentimental. Local varieties persist because they have been selected, informally and over many seasons, for the thing that matters most in that place. They survive the local disease pressure well enough to still be here. They match the local cooking and processing habits. They match the local soil. Their cycle length fits the local hunger calendar. And their planting material is available from a neighbour, this week, at a price you can pay, which is not a small thing in a crop with no mature certified-seed market.

That last point is the real trade-off. Improved material has to physically reach you as stems, and stems are bulky and perishable. A variety that is genuinely superior and unavailable is worth nothing on your farm.

The sensible position for most learners is not either or. Keep the local varieties that your household and market already trust, and test improved material on a small measured part of your land, as Lesson 6 sets out, before you convert your whole field to something you have not yet seen grow on your own soil.

This is the lesson where a course has to resist the thing courses most want to do, which is print a table of the best current varieties for each country. This course will not print one, and the reason is not caution for its own sake.

Disease and pest pressure in cassava is a moving target, not a fixed fact. Cassava mosaic disease and cassava brown streak disease are actively spreading, mutating into new strains and isolates, and being bred against in real time. Any variety named anywhere as resistant is resistant to the strains known at the time of the study that named it. A variety recommendation is a starting point for a conversation with your nearest extension office and seed-certification scheme, never a permanent fact, and a printed table goes stale within a few years while learners keep reading it.

So learn the structure instead, because the structure lasts.

Start with the two diseases, because they behave differently and they are at very different stages of being solved.

Cassava mosaic disease is caused by a group of whitefly-transmitted geminiviruses. It spreads both by the whitefly vector and, critically, through infected stem cuttings used as planting material. On yield, a 1997 estimate attributed 50 to 70 percent yield reductions to CMD in affected crops. A separate 2016 dataset found CMD-infected plants produced a mean of only 23.6 percent marketable roots, meaning roughly three quarters of the roots from infected plants were not marketable. Those are two different measures, a total yield reduction and a marketable-root proportion, and they should not be averaged together.

CMD resistance breeding has been underway in the region since at least the 1930s and 1940s and is one of the longest-running crop-disease breeding programmes in African agriculture. A definitive dated history of the specific IITA release lineage could not be retrieved for this course, so you will not find one here. What you can take from the length of that programme is that resistant material against CMD exists and has existed for a long time, which is why CMD resistance is the trait most likely to be genuinely present in an improved release available to you.

Cassava brown streak disease is the newer and harder problem. It is caused by two related viruses, also whitefly-transmitted, and it differs from CMD in the way that matters commercially: it attacks the storage root itself, causing a necrotic brown rot inside the root that is often invisible from outside until harvest. You can grow a crop that looks healthy all season and discover the loss when you dig, or in the pot. In severe cases CBSD causes up to 100 percent loss in susceptible varieties.

And here is the fact that should govern how you treat any CBSD-resistant claim. As of a 2022 evidence review, there were no CBSD-resistant cultivars available for use by farmers in Zambia, with breeding by the national institute and IITA described as still underway. That is one country, and other countries differ, but it establishes that CBSD-resistant is a claim which was flatly untrue in at least one heavily affected country at that date. Do not accept it from a seller without checking the current national list.

Altitude will not save you either. The whitefly vector is capable of surviving above 1,000 metres, and its potential range covers over 23 million square kilometres in one cited assessment.

So how do you actually judge a resistance claim before you pay?

Ask four questions of whoever is selling or giving you stems. Resistant to which disease, named. Resistant according to which programme or trial, and in what year. Is this variety on the current recommended list for my agroecological zone, and who maintains that list. And has this material been inspected or certified by anybody, or is it simply stems from a field.

Then do the physical check that costs nothing. Look at the mother field before you cut. Leaves with the mottling and distortion of mosaic disease, stems with borer damage, mealybug, or any visible disease symptom mean you walk away, whatever the variety name is. An infected mother plant produces infected planting material. That is the subject of the whole of Module 4.

A course on cassava is really teaching two overlapping businesses, and this is the lesson where you decide which one you are in, because the answer changes every other choice you make.

The first business is food-security farming with a surplus. Sweet, low-cyanide varieties, short-cycle harvest, sold and eaten fresh, boiled, roasted, or made into fufu or gari at household or village scale. The customer is your household and your neighbours, and the quality standard is what they will eat.

The second business is industrial raw-material production. Bitter or high-starch varieties grown at scale and sold to processors for starch, high-quality cassava flour, ethanol, animal feed or industrial adhesives. The customer is a processor with a specification, and the quality standard is dry matter, starch and delivery volume on a date.

Many farms do both. But a variety chosen for one and sold into the other is where money is lost, and this is preventable.

Now the four axes. Frame variety choice as a decision across all four, weighing each for your own situation.

Axis one is end use. Fresh table consumption favours sweet, quick-cooking varieties, because the customer wants a root they can boil today. Industrial starch, flour and high-quality cassava flour production favours high dry-matter, high-yielding varieties regardless of sweetness, because they will always be processed anyway. If your buyer is a starch or flour processor, sweetness is not a quality they are paying for and you should not sacrifice yield or dry matter to get it.

Axis two is disease and pest pressure in your specific locality. Ask the extension office which CMD and CBSD strains and which pests are locally active, and select accordingly. This is a district-level question, not a national one, and the answer changes between seasons. Lesson 4 gives you the four questions to ask.

Axis three is cycle length against your household cash and food calendar. Some cultivars are ready in 9 months and others need 18 months or more, with root quality commonly reported as optimal at 12 to 15 months. A 9-month variety and an 18-month variety serve different household needs on the same farm. A 9-month variety puts food and cash in reach inside a year, which matters if you have school fees or a hungry season to cross. An 18-month variety ties up the land for two rainy seasons and asks you to survive that period on something else. Many farms should be planting both, on different plots, for exactly that reason.

Axis four is cyanogenic potential relative to your intended processing route, and this is the axis most farmers never consider explicitly. A variety destined only for full gari-style fermentation and drying can safely be a higher-cyanogen variety than one destined for boiling and eating within hours of harvest. That is not a licence to be careless. It is the recognition that the safety of what you eventually eat is the product of the variety and the process together, not the variety alone.

Read axis four in both directions, because both directions matter. If you are running a full fermentation and drying operation, you can consider a high-yielding bitter variety that a fresh-table farmer should not plant. And if you are selling fresh roots to households who will boil them within hours, a high-cyanogen variety is the wrong crop for your market no matter how well it yields, because your customers are not going to do the processing that variety requires.

Here is how to run the decision in practice. Write your intended market at the top of a page. Under it, write your processing capacity, honestly, meaning what you can actually grate, press, ferment and dry within a few days of digging. Then score each available variety against the four axes. A variety that wins on yield and loses on axis four for your particular market is not a good variety for you. It may be an excellent variety for the processor down the road.

And remember what Module 1 established. Fresh roots begin to spoil within 48 to 72 hours of harvest. Your variety choice, your cycle length and your processing capacity are one connected decision. A high-yielding 18-month variety that lands 20 tonnes on you in a week you cannot process is a loss with excellent agronomy behind it.

Everything in this module points at one action: get a current recommendation from a living source, then test it on your own land before you bet your farm on it.

Start with the living source. The correct channel for a current variety recommendation is your national root-crop research institute or its extension service. Across the region these include IITA in Ibadan for regional work, and national programmes such as NRCRI in Nigeria, NARO in Uganda, CSIR in Ghana and ZARI in Zambia. Ask for the current recommended list for your agroecological zone, not for the country as a whole, because a recommendation for a savanna zone is not a recommendation for a forest-margin zone.

When you get there, ask these things and write the answers down with the date and the name of the person answering.

1. Which varieties are on the current recommended list for my zone?
2. Which diseases are active in my district this season, and which of those varieties handle them?
3. What is the cycle length of each variety on the list?
4. Where can I obtain inspected or certified planting material, and what does it cost?
5. Has anything on last year's list been dropped, and why?

That last question is the most revealing one you can ask. A list that never changes is a list that is not being maintained.

Now the test, because a recommendation is still somebody else's evidence until it has grown on your soil.

The design is simple and you do not need a research station. Take one plot you can walk to easily. Divide it into strips, one strip per variety, with your current local variety included as one of the strips. That local strip is not sentiment, it is your control: without it you cannot tell whether the new variety did well or whether the season was simply good. Plant every strip on the same day, at the same spacing, with the same planting material handling and the same weeding. If you change two things at once you will not know which one produced the result.

Keep the strips big enough to matter. Twenty plants per variety is enough to see a difference; five plants is not, because one termite mound will swamp your result.

Then record, every single time, the same short list of things:

• Sprouting. How many cuttings out of the number planted actually sprouted, counted at four weeks. This is often the biggest difference between varieties and it is invisible later.
• Canopy closure. The week you can no longer see bare soil between the rows on that strip. Compare it with the reported 20 to 24 weeks.
• Disease. Walk each strip monthly and count plants showing mosaic mottling or any other symptom. Write the count, not an impression.
• Yield. At harvest, dig each strip separately and weigh the roots separately. Same day, same scale.
• Root quality. Cut roots open. Look for the brown necrosis inside that signals brown streak damage, which is invisible from outside. A strip that yields well and is rotten inside has told you something crucial.
• Kitchen and market result. Cook it or process it the way you actually will, and ask the people who will eat or buy it what they think.

That last one decides more sales than any agronomic figure. A variety that yields 20 percent more and that your buyers reject is not a better variety for you.

On cyanogenic potential, be honest about what you can and cannot do. You cannot measure milligrams per kilogram in your kitchen. What you can do is note carefully how each variety behaves under the processing you use, and never treat a new variety with the shorter processing you used for an old one until you have a reason to. If you sell a processed product, testing requirements are set by your national standards body, and a new variety is exactly the moment to go and ask them.

One last piece of discipline. Write the harvest date and age in months on every yield record. Cyanogenic content and starch content both move with age, so a yield figure without an age attached cannot be compared with anything, including your own record from last year.

After two seasons of this you will hold something no pamphlet can give you: the performance of named varieties on your soil, in your rainfall, against your diseases, judged by your buyers.