The Andean highlands hold two records of plants too bitter to eat as they are, turned into a staple or a preserved food. One is chuño and tunta, preserved foods finished from native bitter potatoes (papa amarga) by freezing and river soaking. The other is lupin beans (tarwi, Lupinus mutabilis), which contain bitter alkaloids and are finished by water treatment. In the wording of the material itself, the purpose of both processes is to "remove the bitterness." This note traces their standard definitions and process chains using only what is recorded by Peru's national standards, a producers' field guide from the Puno region, and a review paper on seed composition.
The process for making tunta and chuño from native bitter potatoes is recorded step by step in Peru's national standard and in a producers' field guide.
The explanation that prolonged river soaking washes out the bitter compounds is recorded as a narrative passed down among producers and technicians, but verification by chemical analysis has not yet been confirmed.
One source reports that water treatment and boiling remove 89-97% of the alkaloids from bitter lupin (tarwi) seeds.
A path to reducing bitterness through breeding, such as the low-alkaloid cultivar 'Inti,' is recorded, but a combined cultivar that resists insect damage while keeping only the seed's bitterness low remains at the proposal stage.
Bitter Potatoes Have a Name in the Standard
Peru's national standard defines Andean potatoes by dividing them into three categories: native bitter varieties, native sweet varieties, and improved varieties. The bitter varieties are called papa amarga, a category in the standard that groups together several native cultivar groups.
Tunta is derived from Native bitter varieties (papa amarga: S. juzepczukii, S. curtilobum, commonly known as Piñaza, Lucki, Locka, Choquepito, Parina, etc.), native non-bitter varieties (papa dulce: S. tuberosum spp. andigena, commonly known as Imilla negra, Imilla blanca, etc.), and improved varieties (S. tuberosum spp. andigena, commonly known as Canchán, Ch'aska, Perricholi, etc.).contextual
Scope: The Peru altiplano
Note: Definitions 4.8/4.9 of the NTP standard and Table 2 of the CIP practical guide agree.
Defines papa amarga as a group of native varieties belonging to the 3 species S. juzepczukii, S. courtilobum, and S. ajahujuiri, containing glycoalkaloids and bitter, and papa dulce as S. andígena, S. stenotonum, and others, containing no glycoalkaloids and having good flavour.
Limitations: A definition under the standard, not a report of quantitative compositional analysis.
Tabulates native bitter varieties (S. Juzepczukii, S. Curtilobum; commonly known as Piñaza, Lucki, Locka, Choquepito, Parina), native sweet varieties (S. tuberosum spp andigena; commonly known as Imilla negra, Imilla blanca, Sani imilla, Peruanita, Palita), and improved varieties (commonly known as Canchán, Ch'aska, Perricholi, etc.).
Limitations: A list of common names based on interviews conducted in the growing region, not a report of genetic identification.
The preserved food itself, made from this bitter potato as its raw material, is also defined by the standard together with its process. Tunta is recorded as a food made through a sequence of dehydration steps: multiple rounds of freezing, soaking in flowing water, and sun-drying.
Scope: The Peru-Bolivia altiplano (above 3800m elevation)
Note: Based on definition 4.1 of NTP 011.400.
Defines tunta as "a food product made from tubers by a dehydration process of repeated freezing (protected from sunlight during the day), soaking in running water (a river), and sun-drying. The surface is rough and white, with depressions originating from the eyes of the source potato."
Limitations: A definition from a standards document, not a report of chemical analysis or sensory testing.
A field guide for producers in the Ilave river basin, Puno region, records this dehydration process broken down further into 10 steps: receiving, sorting and classifying, first freezing, soaking in water, second freezing, peeling and washing, drying, rubbing and winnowing, sorting and packing, and storage.
Tunta and chuño processing comprises: 10 steps: receiving (recepción), sorting and classifying (selección y clasificación), first freezing (1er congelado), soaking in water (inmersión en agua), second freezing (2do congelado), peeling and washing (descascarado y lavado), drying (secado), rubbing and winnowing (rozado y venteo), sorting and packing (selección y envasado), and storage (almacenamiento).contextual
Scope: The Ilave river basin, Puno region (Peru)
Note: A description from an official guide that systematized the practices of producers in the Ilave river basin, Puno region.
Organizes tunta processing into 10 steps and describes the work involved, the tools used, and points of caution for each step. A systematization of producers' practices.
Limitations: Not an experimental verification, but a field guide that observed and systematized the practices of a producer community.
Context: As background description in a paper whose main topic is an antimicrobial-activity test, describes tunta production in the southern Peruvian highlands (elevation 3800-4500m) organized into 8 steps (selection, initial freeze-drying, fermentative soaking, peeling, moist fermentation, heat drying, quality selection, storage). This overlaps with the CIP field guide's 10-step description in its broad outline (the basic structure of freezing, soaking, and drying), but differs in how the steps are divided and in whether a moist-fermentation step is added.
Limitations: There are differences between sources in the description of days and temperatures (the CIP field guide/nawparisun describe an initial freezing of 3-4 nights and soaking of 21-30 days, while this source describes an initial freezing of 3-8 nights and soaking of 15-30 days, plus a separate moist-fermentation step of 21 days). This is a background description in a paper whose main purpose is an antimicrobial-activity test, not primary data focused on verifying the process itself.
The first freezing is carried out by exposing the potatoes to low nighttime temperatures below 5℃ for 3-4 nights. It is recorded that the period chosen for this is May-July, calledépoca de heladas (frost season), when climate conditions of strong sunlight, low humidity, and little cloud cover align.
Tunta and chuño processing requires Multiple rounds (3-4 nights) of freezing under night-time low temperatures below 5℃. Carried out during the winter months of May-July, called the "época de heladas" (frost season), under climate conditions of strong sunlight, low humidity, and little cloud cover.contextual
Scope: Puno region (Peru) altiplano, above 3800m elevation
Note: The NTP standard, the CIP practical guide, and a 2023 field-demonstration study all describe the condition as below 5℃ for 3-4 nights.
Describes the first freezing as being carried out by exposing the potatoes to night-time temperatures below 5℃ for 3-4 nights. The May-July "época de heladas" (frost season) is characterized by strong sunlight, low cloud cover, and low humidity under 40%, which is described as aiding the natural freezing and drying.
Limitations: A description of climate conditions, not a report of instrument-measured data.
In a field-demonstration study, the procedure for the first freezing - exposing the raw potatoes to an environment below 5℃ for 3-4 nights - was carried out and recorded. Bartlett's freezing theory is invoked as the theoretical basis.
Limitations: A record from a small-scale field-demonstration study (8 modules, 181 producers).
The frozen potatoes are next submerged inside an enclosure built from netting and stakes, and soaked in flowing river water for 21-30 days (some reports give an average of about 25 days). It is also recorded that their position is moved every 5 days.
Tunta and chuño processing requires The frozen potatoes are submerged in flowing river water inside an enclosure (jaula/poza) made of netting and stakes, and soaked for 21-30 days depending on the variety (some reports give an average of around 25 days). Their position is moved every 5 days.contextual
Scope: The Ilave and Huenque river basins, Puno region (Peru)
Describes submerging the frozen potatoes in an enclosure built of netting and stakes in the river, and soaking them for 21-30 days depending on the variety. Instructs choosing clear, gently flowing water and avoiding sediment.
Limitations: A description from a practical guide; it does not include measured data such as water quality or microbiology.
For the finishing peeling and for lifting the potatoes out of the river, it is recorded that dedicated tools with local vocabulary are used, such as the large strainerwiscaña (iron-ring diameter 50cm) and the netted clothtaquiña, used for peeling by treading pressure.
Tunta and chuño processing requires Traditional tools and local vocabulary: fishing nets (mallas); "chuñahuis", the term for the open ground used for freezing and drying; "wiscaña", a large strainer used to lift the potatoes out of the river (iron-ring diameter 50cm, net depth 70cm, handle 1.2m); and "taquiña", a netted cloth used for peeling by treading pressure.contextual
Scope: The Ilave river basin, Puno region (Peru)
Tabulates the material and use of each tool. Explains wiscaña as a "colador gigante" (giant strainer), and taquiña as the "manta de malla" (mesh cloth) used in the peeling process (Step 6, Descascarado y lavado).
Limitations: A description of the tools; it does not include a quantitative evaluation of performance or efficiency.
Chuño, a very similar preserved food, differs from tunta in both color and process when the two standards' definitional texts are compared. Chuño is a dark-brown dehydrated potato made by repeating freezing and sun-drying without peeling, and the standard's definitional text contains no mention of the prolonged river-soaking process that is central to tunta.
Note: Based on definitions 4.1/4.4/4.5 of the NTP standard.
Defines chuño blanco as 「tunta及びmorayaを、産地・消費地以外で総称する語」 (in translation: "a term that groups together tunta and moraya, used outside their zones of production and consumption"), and chuño as 「皮を剥いていない色が濃褐色の脱水芋で、凍結の反復と天日への曝露によって処理される」 (in translation: "an unpeeled, dark-brown dehydrated potato, processed by repeated freezing and exposure to the sun"). The reference to river soaking found in tunta's definition (4.1) is absent from chuño's definition.
Limitations: A comparison of definitional text in the standard, not a direct field observation of differences in process.
Washing Out Bitterness: A Narrative, and What Came After
The field guide for producers and technicians also records a narrative about why such a laborious river-soaking process is carried out at all. It holds that tunta was once made only from native bitter varieties (papa amarga), and that prolonged soaking in the river washed out the glicoalcaloides (glycoalkaloids), making its consumption possible.
Tunta — origin narrative as told: A narrative recorded in a field guide for producers and technicians states: 「アンデスの歴史においてtuntaは専ら papa amarga(苦味在来種)から作られていた。河川での長時間の浸漬が『グリコアルカロイド』の洗い流しを助け、その消費を可能にしたため」 (in translation: "In Andean history, tunta was made exclusively from papa amarga (native bitter varieties), because prolonged soaking in the river helped wash out the 'glycoalkaloids,' making its consumption possible.")contextual
Scope: The Peru altiplano
Note: The source's narrative is recorded as-is. It does not include any numerical toxicity or safety claims.
The original states that in antiquity tunta was made exclusively from these varieties (bitter potatoes), because prolonged soaking in the river favors the washing-out of glycoalkaloids, making consumption possible.
Limitations: A record of a narrative shared among producers and technicians, not a quantitative demonstration of the washing effect through chemical analysis.
Show dataingredient-tunta —has_historical_origin_narrative→ (the claim sentence above)
claim: cl-tunta-bitterness-narrative / type: cultural / updated: 2026-09-01
This narrative is preserved not as a quantitative demonstration of the washing effect through chemical analysis, but as a record shared between producers and technicians. Even so, the flavor of good-quality tunta itself is described asagradable, ligeramente insípido (pleasant, slightly bland), and it is recorded as having a quality of absorbing the flavor of whatever ingredients it is cooked together with.
Tunta — sensory attribute: Among the traits of good-quality tunta, alongside color (white to matte white) and smell (a mild aroma like an aquatic plant), the flavor 「agradable、ligeramente insípido(心地よく、わずかに淡白)」 (in translation: "pleasant, slightly bland") is listed, and it is described as absorbing the flavor of whatever ingredients are cooked together with it.contextual
Scope: Puno region (Peru)
Note: A description of quality standards by producers and technicians; not an experimental sensory test.
Among the 7 factors that determine commercial quality (forma, color, tamaño, rehidratación, sabor, textura, olor), sabor (flavor) is described as pleasant and slightly bland, pairing well with strongly flavored dishes.
Limitations: A description of quality standards intended for producers; not a standardized sensory evaluation such as a panel test.
Meanwhile, a processing evaluation carried out across several production modules, including the improved cultivar silver, also reported that aligero amargor bastante insignificante (slight, fairly insignificant bitterness) was detected. Even after going through such a laborious process, the bitterness is not necessarily reduced to zero.
Tunta — sensory attribute: In processing evaluations across several modules including the improved variety silver, evaluating "sabor", a 「ligero amargor bastante insignificante(ごくわずかで取るに足らない程度の苦味)」 (in translation: "a slight, fairly insignificant bitterness") is reported to have been detected.observation
Scope: The Huenque and Ilave river basins (Puno region, Peru)
Note: A record from a single small-scale field-demonstration study; not generalized.
Original text: "Se ha evaluado esta característica y se detectó un ligero amargor bastante insignificante." (in translation: "This trait was evaluated, and a slight, fairly insignificant bitterness was detected.")
Limitations: Methodological details of the sensory test, such as the number of assessors and the rating scale, are not stated explicitly in the paper, and it may not be a standardized panel evaluation.
Alongside the potato, another candidate for an Andean bitter staple is the lupin bean (tarwi,Lupinus mutabilis). The bitter alkaloids contained in this species' seeds are a group of compounds called quinolizidine alkaloids (QA), and a distinction is recorded in which sweet cultivars have total QA <500 mg/kg while bitter cultivars have >10,000 mg/kg. ForL. mutabilis, there is 1 paper reporting a value of total QA 31,000 mg/kg.
Note: The total QA figure of 31,000 mg/kg for L. mutabilis is a single report (EFSA states explicitly: "Only one paper reporting the total amount of QAs was identified").
States the distinction of sweet cultivars <500 mg/kg and bitter cultivars >10,000 mg/kg. Total QA for L. mutabilis is 31,000 mg/kg (only 1 paper reporting the total amount of QAs was reported to have been found).
Limitations: The total QA figure for L. mutabilis is a single report.
Among the seed's QA composition, the compound lupanine is recorded as accounting for the largest share (47–58%), followed by sparteine (7.4–16%) and 13α-OH-lupanine (7–15%).
To remove this bitter alkaloid, soaking in salt water and washing has traditionally been carried out. A process combining water treatment and boiling is reported to remove 89–97% of the QA in the seed; biological, chemical, and aqueous methods all exist, and a process finishing in as little as 1 hour is also recorded. Another source notes that soaking and heating alone also lower the alkaloid content, and that following it with microbial fermentation lowers it further, though results are not consistent across strains and species.
lupin debittering reduces 89–97% of the QAs (quinolizidine alkaloids) in the seed (by water treatment and boiling). Traditionally, salt-water soaking and washing. Biological, chemical, and aqueous methods exist, and a process as short as 1 hour has also been reportedsupported
Scope: Andean traditional processing and food-processing research
The source states that seed QA levels have traditionally been reduced substantially by soaking the beans in salted water and washing them, and that water treatment and boiling remove 89 to 97% of the QAs present in seeds.
Limitations: The original study behind the 89–97% figure (species, conditions) has not been traced.
The source states that bitter lupins can be "detoxified" (translation of the review's "detoxified") by biological, chemical, or aqueous processes, and that the shortest debittering process takes 1 hour.
Limitations: Process details are in the full text (not read).
Limits: Debittering of bitter L. mutabilis proceeds in stages: soaking plus heating alone lowers alkaloid content to about 50-65% of the initial value, and a subsequent solid-state fermentation with Rhizopus oligosporus (tempeh-making) further lowers it, depending on the study, to as little as 9% of the initial value (Jiménez-Martínez 2007, 48 hours of fermentation after lactic-acid soaking) to 43% (Agosin 1989, a sweet variety). However, in a trial with a different bitter variety, no detectable decrease in alkaloids was seen during the fermentation stage with R. oligosporus NRRL 2710, and results are inconsistent depending on strain, species, and the presence or absence of an acidic environment.
Limitations: This is a secondary synthesis by a review, and the statistical tests and sample sizes of each primary study are not fully given within the review itself. The review itself states explicitly that the inconsistency due to differences in strain and process conditions is 『説明できない』 (in translation: "cannot be explained").
Cultivar breeding is also recorded as another path for managing bitterness. The low-alkaloid cultivar 'Inti', bred through reselection in the early 1980s, was recorded in Gross et al. 1988 with a value of alkaloid content under 0.02%.
Andean lupin — cultivar: 'Inti', bred through reselection of a low-alkaloid line. Alkaloid content under 0.02%. Recorded as bred through reselection in the early 1980s (Gross et al. 1988).supported
Scope: Lupinus mutabilis
Note: An example of debittering through breeding, in contrast to the existing process-lupin-debittering (traditional debittering by water treatment). However, whether 'Inti' has become widely commercially adopted cannot be confirmed from this material.
States that the low-alkaloid L. mutabilis cultivar 'Inti' (alkaloid content under 0.02%) was bred through reselection in the early 1980s and recorded in Gross et al. 1988.
Limitations: A secondhand citation via a review; the original Gross et al. 1988 paper was not read this time. The commercial adoption status of 'Inti' cannot be confirmed from this material.
In breeding programs for the closely related species Lupinus angustifolius, there is a more far-reaching proposal. It is a description proposing, as a future breeding goal, a combined 'bitter/sweet' phenotype that keeps a high alkaloid concentration in the leaves and stems to prevent insect damage, while keeping the alkaloid concentration low only in the seed.
narrow-leafed lupin — variation is maintained by (proposed): A proposal, as a future breeding goal, for a combined 'bitter/sweet' phenotype that maintains a high alkaloid concentration in the vegetative parts (leaves and stems) to deter insect damage, while keeping the alkaloid concentration low in the seed.preliminary
Scope: Narrow-leafed lupin breeding programs
Note: 2 independent reviews (Frick 2017 and Vishnyakova 2020) propose the same direction, but in both cases this is a description of a future breeding goal, not an already-realized registered cultivar.
Proposes that 'one direction for Lupinus breeding is to develop plants with a bitter/sweet phenotype -- vegetative QA concentration high enough to deter insect damage, while seed QA concentration is low.'
Limitations: A proposal-stage description; it does not indicate an already-realized registered cultivar.
The authors discuss, as a breeding goal, dual-purpose cultivars with high alkaloids in vegetative parts and low alkaloids in seed while keeping disease resistance. It is a review drawing on earlier proposals and does not show that this has been achieved.
Limitations: A breeding concept, not an evaluation of an actual cultivar.
Context: States that the authors compared the accumulation of QAs with the expression of the biosynthetic gene lysine decarboxylase (LDC) in developing seeds and pods of a bitter (high-QA) variety of NLL, and that breeders have developed cultivars with low QA content (the so-called sweet cultivars) to bypass the debittering process -- a description that treats QAs (quinolizidine alkaloids) as the bitter component.
Limitations: This claim concerns the future combined (bitter/sweet) phenotype breeding goal proposed by Frick 2017 and Vishnyakova 2020, but this material addresses the contrast between existing bitter and sweet cultivars themselves, and does not directly verify the combined-phenotype breeding proposal itself.
Show dataorganism-lupinus-angustifolius —has_variation_maintained_by→ (the claim sentence above)
claim: clm-lupin-dual-purpose-breeding-goal / type: cultural / updated: 2026-09-02
What is not known
The total QA figure of 31,000 mg/kg for L. mutabilis is a single report, and whether it can be generalized to other cultivars or growing regions cannot be determined within the scope of this ledger.
The staged decline in debittering (soaking/heating alone, then followed by fermentation) is a secondary organization by a review, and the statistical tests and sample sizes of each primary study are not fully shown within the review. The review itself also states that it cannot explain the inconsistency arising from strains and process conditions.
The number of steps in the tunta process differs by source, between 8 steps and 10 steps, and the descriptions of days and temperature also vary in range. Which should be regarded as the standard sequence cannot be decided within the scope of this ledger.
The narrative that river soaking washes out glycoalkaloids is a record shared among producers and technicians, and quantitative verification by chemical analysis is not included in this ledger.
How widely the low-alkaloid cultivar 'Inti' is currently adopted cannot be confirmed from this material.
L. angustifolius's 'bitter/sweet' combined phenotype is a description at the proposal stage, and does not indicate an already-realized registered cultivar.
Frequently asked
How do tunta and chuño differ?
Both tunta and chuño are processed foods that turn native bitter potatoes into a preserved food, but their standard definitions differ in both color and process. Chuño is a dark brown preserved potato made by repeating freezing and sun-drying with the skin left on, and the standard's definitional text does not mention a river-soaking process. Tunta is a white food made through multiple rounds of freezing plus soaking in flowing water and sun-drying, and this river-soaking step is what distinguishes it from chuño in the definition. Both are comparisons of definitional text in standards documents, not reports of directly observed processing differences in the field.
How is the bitterness removed from bitter lupin beans (tarwi)?
The traditional method recorded is soaking and washing the seeds in salt water. One review reports that a process combining water treatment and boiling removed 89–97% of the alkaloids contained in the seeds. Soaking and heating alone also lower the alkaloid content, and following that with microbial fermentation lowers it further, though the same review also notes that results are not consistent across strains and bean types. A process taking as little as 1 hour has also been reported, but the details of the conditions could not be fully traced within the scope of this material.