How spices are produced and processed

Ripe or nearly ripe pepper berries on the spike, lightly crushed, put into sacks and soaked in water until the fleshy fruit wall disintegrates and can be rubbed off the seed.

In a 60 h retting of abraded berries followed by amplicon sequencing, the bacterial community shifted from one dominated by Firmicutes to one dominated by Prevotella, which rose from under 1 per cent to about 75 per cent of reads, with Lactococcus, Selenomonas, Streptococcus, Weissella and Enterobacter also prominent; among fungi, Candida rose to about 90 per cent. In separate work based on culture, isolates of Bacillus subtilis, Bacillus licheniformis and other bacteria taken from black pepper removed the fruit wall of fresh berries within five days and secreted pectinase, cellulase and other hydrolytic enzymes. No source read documents a starter culture in commercial use.

A regional handbook records soaking for seven to ten days, preferably in slow-running water, followed by three to four days of sun-drying. The laboratory study that followed the microbes used a much shorter soak of 60 h at room temperature in a buffer at pH 5.5, after mechanical abrasion; the pH rose to 5.92 at 36 h and returned to 5.50, pectin in the fruit wall fell by two-thirds, and the share of berries peeled passed 97 per cent by 48 h.

The compounds behind the faecal, farmyard note of some white pepper, 3-methylindole, 4-methylphenol and butanoic acid, were shown to form during retting in a Thai production plant. In model fermentations, a short fermentation under water with frequent changes of water gave white pepper without substantial amounts of them. The handbook account agrees in its own terms: pepper retted in stagnant water dries grey and musty.

The whole green pod of Vanilla planifolia, harvested nearly odourless with its vanillin bound to glucose as glucovanillin, which made up 10.67 per cent of the dry weight of the fresh bean in one study.

Curing is called a fermentation in the trade, and the pod’s own enzymes do much of the work, but microorganisms are present throughout and have been shown to take part. Sequencing of beans through a Hainan curing found Bacillus at 0.09 per cent of bacterial reads after blanching, 72 per cent after sweating, 3.3 per cent after drying and 87.5 per cent early in conditioning, with Lactococcus alongside it in the finished bean; Aspergillus dominated the fungi from sweating onwards. A study of traditional curing in Indonesia, by plating and DNA profiling, found that fungi and yeasts disappeared after scalding and that heat-tolerant bacilli related to Bacillus subtilis, B. licheniformis and B. smithii developed during the hot period that followed. Bacillus isolates from curing beans produced β-glucosidase and hydrolysed glucovanillin in culture.

The studied Hainan process blanched the beans at 70 °C for 5 min, sunned them for about 6 h a day and sweated them in an oven at 55 °C for six days, dried them to about 30 per cent moisture over 36 sunny days and conditioned them in closed boxes for six months. In the Indonesian study the beans were held at up to 65 °C for over a week after scalding. A regional handbook describes scalding for 30 to 60 s, sweating for 24 to 48 h, several days of alternate sunning and storing, and two to three months of conditioning, and gives five to six months for the whole process in Mexico.

Through curing, glucovanillin fell from 10.67 per cent to 0.17 per cent of dry weight and vanillin rose from 0.03 per cent to between 2.7 and 3.0 per cent, most of it appearing during sweating. Vanillic acid, p-hydroxybenzaldehyde and p-hydroxybenzoic acid were released alongside it. A study of traditional Mexican curing found that the many other glucosides in the bean are hydrolysed at different stages, some only late and some not at all within ninety days, and that the freed compounds are further converted into one another.

Fresh garlic bulbs, held for a period at a controlled high temperature and high humidity until the cloves turn black.

A review gives the conditions under which black garlic is made as 60 to 90 °C at 80 to 90 per cent relative humidity. Both reviews read call the process a fermentation while describing it as a heat treatment; neither names a microorganism or a starter culture. Two studies that followed bulbs through it agree. In one, two Italian varieties held in a household machine for twelve days went from about pH 5.9 to about 3.7 and from 63 to 65 per cent moisture to between 22 and 28, and yeasts counted on the fresh bulbs were no longer found at six days. In the other, a Korean cultivar aged for fifteen days ended at pH 4.17 or 4.41, depending on how fast it was heated.

Black garlic does not give off the strong pungent odour of fresh garlic, which the reviews attribute to its reduced allicin content, and it has a sweet-sour flavour. The change in aroma is attributed to the enzymatic and non-enzymatic browning reactions that run during the long heating. In one laboratory ageing, alliin, the compound from which fresh garlic makes its pungency when cut, fell quickly in the first five days, and fructose, steady for the first five days, rose after that, which is in keeping with the sweetness.

The seeds of the African locust bean, Parkia biglobosa, cleaned and sorted, boiled, dehulled and then left to ferment. The same kind of seasoning is made in West Africa from other protein-rich seeds, and the names differ by country: iru or dawadawa in Nigeria and Ghana, soumbala or soumbara elsewhere in the region.

In iru from producers and markets across Nigeria, DNA profiling of 16 samples found bacteria related to Bacillus subtilis as the one consistent presence, alongside relatives of Staphylococcus vitulinus, Staphylococcus saprophyticus, Morganella morganii, Bacillus thuringiensis, Tetragenococcus halophilus and others. Culturing gave counts of 10⁶ to 10⁷ CFU/ml and 280 isolates, dominated by B. subtilis and B. amyloliquefaciens, with members of the Bacillus cereus group among them. Sequencing of dawadawa from six communities in Ghana likewise found Bacillus, Staphylococcus, Streptococcus and Lactobacillus as the main genera and B. subtilis the most abundant species, with home-made samples more diverse than commercial ones. All of this describes spontaneous fermentation; the Nigerian study isolated strains as candidate starter cultures but does not report one in use.

The fermentation is described as spontaneous, on the solid boiled seed, at ambient temperature and under uncontrolled conditions. It is alkaline: the pH rises as proteins are broken down extensively to peptides, amino acids and ammonia, and that rise favours Bacillus. Market soumbala in northern Côte d’Ivoire had a pH of about 6 and a moisture content of 20 to 24.7 per cent as paste, 7.3 to 9.3 per cent as powder and 8.6 to 10.7 per cent as granules.

The fermented seeds are used as a flavouring condiment across West African cooking, and are sold as whole fermented seeds, paste, granules and powder.

Members of the Bacillus cereus group were among the bacteria cultured from Nigerian iru, and the DNA profiles included relatives of Morganella morganii and staphylococci; the authors argue from this for controlled fermentation and good manufacturing practice. In 54 market samples of soumbala from Korhogo, aerobic mesophilic counts of 6.17 to 8.38 log cfu/g exceeded the standard the authors applied, while coliforms, moulds and yeasts were below it; the powder was the most contaminated form. Those authors note that the powder is often used without cooking and advise that soumbala be added during cooking.

Soybeans, black-seeded for Liuyang and Yangjiang douchi and ordinary for Pingjiang douchi and the bacterial douchi of Gansu. The beans are steeped, cooked and cooled before anything grows on them.

Douchi is sorted into four types by the organism that leads the first stage: Aspergillus, Mucor, Rhizopus and bacterial. They change as a batch goes on: in one Yangjiang batch followed for 35 days, Aspergillus led the first stage and the yeast Millerozyma and the bacteria Staphylococcus and Bacillus the second, though Staphylococcus was abundant from the first day. In seven commercial douchi from Sichuan and Chongqing, sequencing found Bacillus, Tetragenococcus and Weissella as the main bacteria and Aspergillus, Mucor and Penicillium as the main fungi; across fifteen bought around China, Bacillus and Candida led. In home-made bacterial douchi from Gansu, where nothing is inoculated, Bacillus and Ignatzschineria were found in both places sampled, with the yeast Pichia. Traditional douchi is described as fermented by organisms from the surroundings and the raw material.

Two stages. Mould or bacteria first grow on the cooked beans; in the Aspergillus-type process the mould is then washed off, to keep the douchi from tasting bitter, and the beans are salted and left to ferment again, in one factory in sealed tanks. Salt and time differ widely by type. Liuyang douchi has under 3 per cent salt and a short ageing; Yangjiang douchi over 10 per cent; Pingjiang douchi is heavily salted and aged for a year. Traditional Mucor-type douchi can be started only between November and February, because Mucor grows only at low temperatures, and is aged for more than ten months.

Rebuilding the aroma of each of three types and leaving out one compound at a time showed that guaiacol mattered to the smoky Liuyang douchi; phenylacetaldehyde, dimethyl trisulfide and 2-acetylpyrrole to Yangjiang douchi; and phenylacetaldehyde, dimethyl trisulfide and a pyrazine to Pingjiang douchi. The same test on a Yongchuan douchi picked out 2,3-butanedione, dimethyl trisulfide, acetic acid, acetylpyrazine and guaiacol most clearly. In a Yangjiang batch followed through its making, the main volatile compounds of the first stage were the alcohols and aldehydes of the bean; guaiacol appeared in the second stage, and 3-octanone, present in a small amount from the start, rose quickly there. The sauce-like note was the one the rebuilt mixtures matched least well. In seven douchi from Sichuan and Chongqing glutamic acid was the most abundant free amino acid.

Described as widely used as a condiment in Chinese dishes, with fried fish, meat, vegetables and canned foods given as examples, and as eaten in its own right as well as cooked with. None of the studies names a dish.

In fifteen douchi bought in fifteen regions of China, six biogenic amines were measured. Total amines ranged from 5.09 to 679.46 mg/kg, below the 900 mg/kg the authors take from other work as the level of concern for the total. Taken singly, histamine was found in five samples, at 1.43 to 213.13 mg/kg, and was above the 50 mg/kg the authors cite in two of them; β-phenethylamine ranged from 1.47 to 191.24 mg/kg and was above the 30 mg/kg they cite in six. No sample exceeded the level they cite for tyramine. The authors attribute the amines to microbes that break down amino acids, and tested strains that form none.

Melon seeds, boiled to loosen the thin seed coat, dehulled, boiled again and wrapped in leaves to ferment; in southern Nigeria they are sometimes boiled with particular leaves, which producers say give a desired flavour.

From ogiri made in the laboratory, the study isolated Bacillus subtilis, Corynebacterium species, Lactobacillus acidophilus and Staphylococcus aureus by culture. From a sample bought in a market it isolated Bacillus subtilis, Staphylococcus aureus, Enterobacter species, Escherichia coli and Candida albicans. No starter culture was used.

In the laboratory process reported, the seeds were boiled for 30 min, dehulled, boiled for a further 30 min, wrapped in leaves of Thaumatococcus daniellii (or in foil or a closed container for comparison) and held at about 38 °C for 120 h, after which fermentation was ended by drying. The finished condiment had a pH above 7 in every laboratory sample; the market sample stood at 6.36.

Used in small amounts to flavour soups and stews; in the study’s tasting it was stirred into a stew and boiled for about two minutes.

Total bacterial counts ranged from 1.0 × 10⁵ CFU/ml in the sample fermented in a closed container to 8.4 × 10⁵ in the market sample, whose bacterial and fungal counts were significantly higher than those of the laboratory samples. Staphylococcus aureus was isolated from laboratory and market ogiri alike, and Escherichia coli from the market sample. The authors cite earlier reports that fermenting and packing the condiment in leaves encourages contamination by unwanted microorganisms and shortens its keeping time.

Small or planktonic shrimp and similar crustaceans, mixed with salt and left to ferment spontaneously into a paste.

The organisms found are salt-tolerant bacteria, and they differ between products. In ten commercial Indonesian terasi, sequencing and plate counts found traditional products dominated by Tetragenococcus, Bacillus, Weissella and Halanaerobium, while industrial products held no detectable microorganisms, which the authors put down to sterilisation. In seven Thai kapi, nanopore sequencing found Lentibacillus species dominant in three and Staphylococcus sciuri in three others, with Alkalibacterium kapii dominating one. A low-salt terasi was dominated by Tetragenococcus, Aloicoccus, Alkalibacillus, Atopostipes and Alkalibacterium, and a Chinese shrimp paste by Tetragenococcus throughout its fermentation.

One of the Thai products studied was traditionally fermented with 25 per cent salt by weight. The low-salt terasi was followed through 28 days of fermentation.

Forty-eight volatile compounds were identified across the Indonesian products, among them sulphur compounds, short-chain fatty acids and trimethylamine. Tetragenococcus, Bacillus and Halanaerobium were statistically associated with trimethylamine, dimethyl disulfide and short-chain fatty acids, associations the authors themselves call exploratory. Traditional products held a greater variety of aroma compounds than industrial ones.

Described in the Indonesian study as a staple condiment of Indonesian cooking.

Histamine was detected in all seven Thai kapi, at 9.5 to about 39 mg/kg, below the 50 mg/kg limit the authors cite. Two products, a wet-market paste and a community-enterprise paste, carried high microbial loads including Staphylococcus aureus, Bacillus cereus and Clostridium perfringens, while a certified export product carried no detected pathogens; the authors conclude that a 25 per cent salt level alone does not ensure safety without hygienic processing. In a low-salt terasi made in the laboratory, histamine and cadaverine were highest at the start of the fermentation and fell as it went on. In a Chinese shrimp paste, six bacterial genera were positively correlated with biogenic amine levels.

Small sea fish and salt, and nothing else to begin with. The Chinese sauce studied was made from whole anchovies; Malaysian budu is described as made from fish of the genera Stolephorus, Sardinella or Decapterus. The liquid that the salted fish gives up over months is the sauce. Colatura di alici di Cetara, a protected Italian one, is made from the European anchovy, Engraulis encrasicolus, and medium or coarse sea salt alone.

Salt-tolerant bacteria, working alongside the fish’s own enzymes. In a traditional anchovy sauce followed through a year of fermentation in a Guangdong factory, 94% of the proteins recovered from the liquid came from bacteria, and at genus level chiefly from Halanaerobium, Psychrobacter, Photobacterium and Tetragenococcus. The study identified which organisms were active by the proteins they left; it did not count them. Where the process is shortened in a factory a mould starter may be added: one industrial sauce was made with Aspergillus oryzae.

The traditional Chinese sauce was fermented with 20% salt at 25 °C for a year, stirred twice a day; its authors give 10 to 18 months and 25 to 30% salt as the usual traditional range. Budu is described as fermented in closed tanks at 30 to 40 °C for 6 to 12 months, after which palm sugar, tamarind and flavourings are added and the sauce is boiled and filtered. Four commercial budu measured 11.8 to 22.5% salt and pH 4.50 to 4.92, and one fell below the 15% salt that Malaysian food regulations require of it. The Italian colatura must show at least 20 g of salt per 100 g and a pH of 5 to 7 when its salting ends; its rule gives no time or temperature.

Fermentation breaks the fish protein down to free amino acids and small peptides, which is where the savoury depth comes from; the Chinese study traces the routes by which its bacteria could make glutamic acid and the savoury nucleotides, as predictions from the genes and proteins found. The smell is a separate matter. Forty-five volatile compounds were tentatively identified across four commercial budu, among them alcohols, aldehydes, ketones, furans and short-chain acids such as acetic and 3-methylbutanoic acid, whose odours are described as vinegary and as rancid, sweaty and cheesy.

Used across South-East and East Asia as a condiment and a cooking seasoning for its savoury depth, and known by a different name in each country: yu-lu in China, nam pla in Thailand, patis in the Philippines, bakasang in Indonesia and budu on the east coast of Peninsular Malaysia. On the Amalfi coast of Italy, colatura di alici is used as a seasoning and as a sauce for pasta, and its registration says that at Cetara it is served with spaghetti or linguine on Christmas Eve.

Histamine is the documented chemical concern; the authors of one study call it one of the most watched safety indicators in fish sauce. The factory-made anchovy sauce they worked on measured 81.56 mg per 100 mL, against limits its authors cite of 50 mg/kg from the United States regulator for seafood and 400 mg/kg in European Union rules. The high salt is what holds back spoilage and pathogenic bacteria, which is the reason Malaysian regulations set a minimum for it.

Soybeans, water and whatever microbes survive the boiling, and nothing else: it is made without salt. The beans are soaked, boiled and left to ferment. Fresh, wet thua nao keeps for about two days, so it is steamed or roasted, or mashed, pressed into flat discs and dried in the sun, and the dried discs keep for months.

Bacillus leads. In 65 samples from villages in six northern provinces, sequenced for their bacteria, it was on average 67 per cent of the community: 86 per cent in the 35 dried samples and 47 per cent in the 30 wet ones, which were more varied and also carried Lactobacillus, Enterococcus and Globicatella. In five village batches followed by culture, Bacillus subtilis was there from start to finish and a Lactobacillus only on the first day. Nothing is inoculated in the village product. In three laboratory batches left to ferment unaided, Bacillus and its relatives and enteric bacteria of the order Enterobacterales together made up over 99 per cent of what was found; one batch was almost wholly the former, and the other two held both throughout. Moulds were grown from dried samples in four of seven districts surveyed.

The village process is described as two to four days at ambient temperature, which one paper gives as 28 to 34 °C; makers say it takes longer in the cool season. The two laboratory studies of its chemistry held their beans at 37 °C for three days, one with a commercial Bacillus subtilis culture. The beans turn alkaline: a review puts the finished product as high as pH 8, and a naturally fermented laboratory product measured 8.23 to 8.69. Dried and market samples are nearer neutral: 6.92 to 7.23 in discs dried four ways in one laboratory, 6.29 to 6.70 in three samples from Mae Hong Son. Earlier analyses, as one paper quotes them, put the water content at 57 to 65 per cent.

Ammonia and pH rose through the three days in five village batches. In laboratory batches acetate rose sharply in the first day, and the free amino acids leucine, alanine, valine and glutamine rose and stayed up, while citrate fell away. One study, read in abstract, put total volatile compounds at 35 micrograms per kilogram in the cooked beans and 3,500 after 72 hours. Wet village samples held more acetate, propionate, butyrate and isovalerate than dried ones. In dried discs made with a starter culture, 65 volatile compounds were identified, among them 2,5-dimethylpyrazine and trimethylpyrazine. How the discs were dried changed the smell: sun-drying gave more of the green, fatty products of fat oxidation such as hexanal, and sun and hot-air drying more of the roasted and smoky compounds of browning than microwave or vacuum drying.

Thai food scientists describe it as serving chiefly as a condiment, or mixed into chilli pastes, and not eaten plain as Japanese natto is. Another group describes a savoury taste and a strong meaty smell, and notes that it is free of salt, unlike shrimp paste. A Thai study of dish names calls it the main seasoning of Tai Yai cooking. Across the border in north-eastern Myanmar, pe poke is pressed into wafers and sun-dried in the same way; in Myanmar the fresh beans are also eaten as a side dish and as fritters.

One survey cultured 24 market samples from Chiang Rai and Phayao, nine fresh and fifteen dried. Escherichia coli was grown from 13, Salmonella from three and Bacillus cereus from two; Clostridium perfringens and Staphylococcus aureus from none. A survey of dried thua nao from seven districts grew Aspergillus flavus and Aspergillus niger from the samples of two, and measured no toxin. The authors of the village-batch study note that the product is cooked again, by steaming or roasting, before it is eaten.

Raw freshwater fish, salt, and rice bran or roasted rice flour, sealed in a container and left for at least six months to a year. The rice is described as the source of carbon for the fermenting organisms. In some cases a little finished pla-ra or a bacterial starter culture is added.

Salt-tolerant and lactic acid bacteria together, and not the same ones in every jar. In five market samples from three north-eastern provinces, sequencing found Tetragenococcus, Halanaerobium and Lactobacillus among the leading genera: Tetragenococcus muriaticus in two samples, Halanaerobium fermentans the most abundant in a third and jointly so in a fourth, and Lactobacillus rennini leading the sample from Roi Et. A later study that followed the fermentation reports that the variety of bacteria fell and that of fungi rose as it went on, that Halanaerobium increased, and that Penicillium was the main fungus.

The five market samples, each fermented for more than a year by its seller’s account, measured 7 to 10 per cent salt, pH 4.83 to 7.15 and 90 to 450 mg of lactic acid per litre. It is kept at ambient temperature and is said to last a year or longer.

A very strong smell, and a saltiness and sourness that depend on how much salt went in and how much lactic acid the fermentation made. Glutamic acid is reported as its most abundant amino acid, which is where the savouriness comes from. The taste differs from one community to the next, and buyers are said to hold strong preferences for particular sources.

Both a condiment and a dish in Thailand and Laos. The paper read names it as the seasoning of north-eastern Thai cooking, in namprik pla-ra, kang lao, namya and somtum pla-ra, the green papaya salad eaten all over the country.

The seeds of roselle, Hibiscus sabdariffa, the plant whose calyx is dried for drinks. The studies read call bikalga an alkaline fermented food used as a condiment in Burkina Faso, and one says it is widely eaten there and in neighbouring countries.

Seventy bacteria isolated by culture from bikalga made in different regions of Burkina Faso were identified by DNA typing and sequencing. Bacillus subtilis was the commonest, at 42 of the 70, followed by Bacillus licheniformis at 16; the others were B. cereus, B. pumilus, B. badius, Brevibacillus bortelensis, B. sphaericus and B. fusiformis, as the abstract abbreviates them. A later study describes the fermentation as dominated by species of the Bacillus subtilis group. Lactic acid bacteria are present too: Enterococcus faecium, Enterococcus hirae and Pediococcus acidilactici were isolated from bikalga and soumbala. The authors propose selecting starter cultures; none of the abstracts reports one in use.

A study of how it is made at production sites in Burkina Faso lists the main steps as cleaning the seeds, cooking them with an alkaline liquid drawn from ash, fermenting, steaming and drying. The pH rose with cooking and fermentation, to 8 or 9, and then fell with steaming and drying, in some cases to 6. The seeds studied held 28.69 per cent crude protein and 26.39 per cent carbohydrate before and 26.47 and 13.7 per cent after, with moisture, ash and titratable acidity higher in the fermented seed.

Used as a condiment. The abstracts read say no more than that about how it is cooked with.

Forty-one Bacillus isolates from soumbala and bikalga were screened for toxins. All nine Bacillus cereus among them produced enterotoxin when grown in fermenting African locust bean, and no isolate of another species did; the gene fragment specific to producers of the emetic toxin was not detected in any isolate. The test fermentation was of locust bean, not of roselle seed. In a separate laboratory study, liquid from cultures of three Bacillus subtilis strains isolated from bikalga inhibited Listeria monocytogenes, Staphylococcus aureus and Bacillus cereus; that was shown in culture and not in the condiment.

A paste of red pepper with ginger, red onion and garlic, to which smaller quantities of other spices and salt are added; the paper lists cardamom, fenugreek, cumin, basil, black cumin, rue, coriander, rosemary and thyme among them.

The paper describes lactic acid bacteria and yeasts as the organisms of the fermentation. In its own storage trial, lactic acid bacteria rose from 1.63 log cfu/g in the fresh paste to a peak of about 4.3 log cfu/g after 120 days at room temperature and then fell below 3 log cfu/g by 300 days. Counts were by plating; no organism was identified to species, and no starter culture was used.

Traditional awaze is described as fermenting for about 30 days at ambient temperature. Rural households make it in the season after the red pepper harvest in November and keep it through the year in traditional pots and gourds.

Over 300 days of storage in six kinds of container, the highest yeast and mould count, 4.06 log cfu/g, was in paste kept in plastic bags, and the highest total bacterial count, 4.12 log cfu/g, in metal cans. Paste in glass bottles at 4 °C held 3.2 log cfu/g of yeasts and moulds and 2.97 log cfu/g of bacteria at 300 days, which the authors judged acceptable against the standards they applied. Counts were higher at room temperature than under refrigeration in every container, and colour faded faster.

A spicy chilli paste eaten mainly in southern Ethiopia, made from green chillies, fresh sweet basil, garlic, and ginger or rue seeds or both. The study read took this description from earlier work and analysed a single sample collected in the Wolaita zone.

By 16S rRNA sequencing, the one sample had a bacterial community unlike those of the other eight Ethiopian fermented products analysed with it: an uncultured lineage of the family Marinilabiliaceae made up 43.8 per cent of it and Weissella 38.4 per cent. Culturing recovered four strains, of the genera Weissella, Secundilactobacillus and Kocuria. The authors describe the fermentation as spontaneous and note that earlier, culture-based studies had found Lactobacillus and yeasts dominant instead. Fungi were not examined here.

The sample was only mildly acid, within the range of pH 4.6 to 5.6 that the authors cite from earlier studies, and held very little sugar, ethanol or organic acid; citric acid was the main acid, with glutamic, lactic and acetic acids also present. The paper does not say how long or at what temperature datta is fermented.

Dried chillies, rehydrated or boiled, crushed, mixed with salt and sealed to ferment. One study added 6 per cent salt and 5 per cent sugar, with or without 5 per cent Sichuan pepper; another mixed chilli, salt, fresh garlic and fresh ginger at 100:8:5:3.

In a spontaneous fermentation followed by 16S rRNA sequencing, the bacteria that came to dominate depended on the chilli cultivar: Lactiplantibacillus in paste made from one cultivar, Weissella in another, and a slower, less orderly rise of Weissella in a third. No starter culture was used.

The fermentations studied were spontaneous and sealed, in ceramic or food-grade jars, for 35 days in one study and about a month in the other.

Fermented without Sichuan pepper, the paste was mainly sour in an electronic-tongue profile. Sixty volatile compounds were identified, chiefly terpenes, alcohols and esters, and adding Sichuan pepper raised the terpenes in particular, among them limonene, β-myrcene and α-phellandrene. Across cultivars, one paste gained acids and esters late in fermentation, another kept more aldehydes and formed esters earlier, and a third was richer in alcohols, ketones, furans and pyrazines.

Fresh chillies and a cereal, mainly maize or rice, fermented together. In the study read, fresh red chillies were cut up and mixed with an equal weight of rice flour and 6 per cent salt.

Through 90 days of natural fermentation followed by high-throughput sequencing, Lactobacillus was the dominant bacterial genus and Candida the dominant fungal genus, and both were positively correlated with most of the key flavour compounds.

The fermentation followed was natural and ran for 90 days in Guizhou, in a tank whose mouth was sealed and which was stood upside down in a basin of water. The authors give the usual length of the traditional fermentation as 30 to 90 days at room temperature. The pH fell and total acid rose continuously, while organic acids first rose and then declined; lactic and acetic acids were the main ones.

Six organic acids, 17 amino acids and 21 key volatile compounds, mainly esters and terpenoids, were identified as its flavour components. Total free amino acids fell after fermentation, with bitter and umami amino acids rising and sweet ones falling.

Sour, aromatic and hot. It is fried and eaten as a dish in its own right, or used as a seasoning to improve other dishes.

Chilli mash fermented with salt; the conventional product used as a control in one study held 15 per cent salt. Chopped fermented chilli, which a second study calls duo la jiao, is the same thing made by cutting fresh chillies into pieces and salting them, at 10 per cent in that study.

In mash fermented at reduced salt with water activity held between 0.87 and 0.91, lactic acid bacteria and yeasts dominated, and in one treatment Bacillus and the dominant yeasts were correlated with the build-up of esters and pyrazines. The opportunistic pathogen Klebsiella reached about 10 per cent relative abundance under reduced salt. In chopped chilli fermented spontaneously at 10 per cent salt, which bacteria took over depended on the chilli variety: in one of the three varieties compared, Weissella reached about 90 per cent relative abundance late in the fermentation.

The study compared 4 to 12 per cent salt against a 15 per cent control at 25 and 40 °C. The 4 per cent salt treatments reached a total acidity of 130 to 200 g/kg against 24 to 58 g/kg in the control and used up their reducing sugars faster. The chopped chilli of the second study was fermented at 20 °C for 120 days and finished at pH 4.07 to 4.26, with a total acidity of 5.68 to 7.69 g/kg, depending on the variety.

In a separate study of fermented Capsicum frutescens chilli, 19 aroma-active compounds were detected and omission tests identified four as key: methyl salicylate, linalool, 2-methoxy-3-isobutylpyrazine and phenylethyl alcohol. Methyl salicylate masked the floral note, while phenylethyl alcohol reinforced the aromas of the other two.

Chopped fermented chilli is described as spicy and sour and as a foundational seasoning of Hunan cooking.

Lowering the salt raised the relative abundance of Klebsiella, an opportunistic pathogen, to about 10 per cent, which the authors of the low-salt study take as showing that strict hygiene of the raw material is a critical control point when salt is reduced.

A grain-based paste with red pepper, made with meju, the fermented soybean block that the biogenic-amine study names as its main protein-rich ingredient.

High-throughput sequencing before and after fermentation found Bacillus, Aerosakkonema and Enterococcus as the main bacterial genera, with Bacillus replacing Aerosakkonema as fermentation went on. Among fungi, Aspergillus dominated before fermentation and the yeasts Zygosaccharomyces and Millerozyma after it. In 35 cottage-industry products, plate counts ranged from 2.79 to 8.73 log CFU/g for aerobic bacteria and 1.56 to 7.15 log CFU/g for yeasts, and five yeast species were identified, Zygosaccharomyces rouxii among them.

Four farm-made pastes were sampled on the day they were made and again after twelve months. After fermentation, salt concentration, pH and reducing sugar had all decreased, while acidity and amino-type nitrogen had increased; salt fell from about 8.2 to 7.3 per cent on average. The pastes differed from one another, and one of the four showed no rise in amino-type nitrogen.

The abundance of Zygosaccharomyces rouxii after fermentation was related by the authors to the flavour of the paste, and a high level of amino-type nitrogen to the abundance of Bacillus before fermentation. The cottage-industry products contained ethanol as their main alcohol, with volatile alcohols, aldehydes, alkanes, nitrogen-containing compounds and terpenes.

Among 35 cottage-industry products, eight carried Bacillus cereus above 4 log CFU/g. Total biogenic amines in the same products ranged from 52.95 to 176.24 mg/kg, with histamine from not detected to 16.94 mg/kg and tyramine from 2.15 to 52.34 mg/kg, all below the toxicity limits the authors cite; they attribute the low levels to the small share of protein-rich ingredients other than meju.

Red chilli, steamed broad beans, salt and wheat flour. The paste is made in three stages: the broad beans are fermented with wheat flour and a mould starter, the fresh chillies are chopped and fermented in salt, and the two are then combined and aged together. The studies read describe it both as a broad-bean paste and as a fermented red pepper paste. A chilli bean paste made in Guizhou leaves out the flour.

The bean stage is started with a mould, Aspergillus oryzae in the processes described; one licensed producer used two strains of it and one of Aspergillus awamori. Followed at three salt levels, that stage was dominated by Bacillus, Staphylococcus, Aspergillus and Mortierella. The salted stages that follow are spontaneous. In the Pixian mash, Aspergillus niger, Candida zeylanoides and Bacillus licheniformis were identified as secreting peptidase and building up amino acids from the bean proteins. In a Guizhou paste, Latilactobacillus led the bacteria and Starmerella and Pichia the fungi across two samplings, but in the six months between them Latilactobacillus fell from about half of the bacteria to about 6 per cent and Pichia from about 30 per cent of the fungi to about 4. Traditional production is described as relying on the microbes of the raw materials and the surroundings.

Heavy salt throughout: one study gives the commercial paste as 15 to 22 per cent salt under the Chinese national standard and the bean stage as 12 to 13. The combined mash is traditionally aged in the open, in the sun, sheltered from rain, stirred by day and left under the dew at night. At one factory the pH of the mash fell from 5.7 to 4.9 over the 60 days sampled. At another, beans and chillies were each salted to 15 per cent, the beans fermented for 180 days, the chillies for 120 and the mixture for 15 more. An experimental batch ran its joint fermentation for 90 days. One study describes the paste as usually ripened for one to three years. All of this is the Pixian paste.

In an aroma study of three pastes aged two to five years, 165 volatile compounds were identified and 13 were aroma-active. Ethyl isovalerate, β-damascenone, 3-isobutyl-2-methoxypyrazine and sotolone had the highest odour activity values; the last three were described as fruity, red-pepper-like and caramel-like. Among non-volatiles in six samples, citric acid was the most abundant organic acid at 4.1 to 6.3 mg/g, followed by malic acid at 2.1 to 3.6 mg/g, and the savoury amino acids were the most abundant amino acids. Time adds to all of it. Beans followed in a tank darkened as they fermented while volatile compounds accumulated, and in a Guizhou paste six months of the second fermentation raised the organic acids, citric acid most, by a factor the abstract gives as 1.51, and left glutamic acid the most abundant free amino acid, at 4.0 mg/g of dry weight.

A cooking base of Sichuan food. In a study of water-boiled beef made with and without it, leaving the paste out made the dish less savoury and less liked; the paste contributed most to taste, and the dried chilli and Sichuan pepper most to aroma.

Salt is what holds the fermentation to the organisms wanted. When the bean stage was run at 4, 8 and 12 per cent salt, less salt gave more amino acid nitrogen, sugars, organic acids and volatile compounds, and it also left more of three genera the authors class as opportunistic pathogens, Enterobacter, Pantoea and Brevundimonas, whose share fell as the fermentation went on and fell further as the salt rose. The authors read this as high salt holding such organisms back, and say that fermenting with less of it would need something else to do that work.

The pulp of durian, mixed with salt and left to ferment in a sealed container.

Lactic acid bacteria. Earlier studies, as one paper summarises them, found them the leading organisms and isolated Lactobacillus and Leuconostoc species and Fructobacillus durionis, with Lactobacillus plantarum and F. durionis dominant. In tempoyak from Kalimantan, Lactobacillus made up about 90 per cent of the bacterial sequences, and the two species cultured from it were Levilactobacillus brevis and Lactiplantibacillus plantarum. From three batches bought in Malaysian markets, a further study kept seven strains, Lactobacillus plantarum the commonest, with L. crispatus, L. reuteri and L. pentosus.

One account gives 2.5 per cent salt and a week of fermentation in a sealed container. Durian flesh is described as holding sugar enough, 15 to 20 per cent, to feed the bacteria. A second paper describes the fermentation as spontaneous and uncontrolled and quotes an earlier study for an acidity of pH 3.8 to 4.6. The tempoyak bought in Kalimantan for a third had been fermented with salt only and kept in closed containers.

Its sourness is made by the fermentation. Acetic, lactic and propionic acids have been detected in it, and the lactic acid, with the salt, is what lets it keep.

Described as a condiment often used with certain fish and vegetable dishes, and in a second paper as an acid-fermented condiment used with foods such as fish and vegetables. It is made and eaten in Malaysia and on Sumatra and Kalimantan.

Vegetables such as Chinese cabbage and radish, salted, mixed with a seasoning paste and fermented by their own lactic acid bacteria. The paste is red pepper powder with garlic, ginger, spring onion, salted fermented seafood and a starch paste. In one laboratory kimchi the cabbage was 84 per cent of the weight and the red pepper powder 3 per cent.

The spices in kimchi shape which bacteria grow. One group bought red kimchi and white kimchi, made without red pepper powder, from the same company and followed one bacterium in each with a species-specific assay. Weissella koreensis gave the stronger signal in the red kimchi, between the first and third weeks at 4 °C and within two days at 15 or 25 °C. The same group cites earlier work in which kimchi with the powder held a larger share of Weissella and less Leuconostoc and Lactobacillus. Garlic acts too: in kimchi made with 0, 1, 2 and 4 per cent garlic, lactic acid bacteria after one week stood at 6.35, 5.67, 6.99 and 7.47 log CFU/mL, and kimchi with 2 per cent or more held more Lactobacillus and Leuconostoc.

The studies read kept their kimchi cold: one at 4 °C, sampled weekly, and another at 4, 15 and 25 °C, where the same change took weeks in the cold and a day or two in the warm. The seasoning paste ferments on its own as well. A commercial paste held at 10 °C for 20 days lost pH and gained acidity as its lactic acid bacteria grew.

Garlic changed the chemistry as well as the bacteria. Kimchi with 4 per cent garlic was significantly more acid after one week than kimchi with none, and the authors conclude that garlic altered the use of sugars and the making of organic acids. They suggest this bears on flavour and did not test it by tasting.