From Calyx to Colour: Drying, Extraction and Formulation as the Determinants of Anthocyanin Retention in Hibiscus sabdariffa

1. Introduction

Roselle (Hibiscus sabdariffa L.) is grown across the tropics for its fleshy red calyces, which are harvested two to three weeks after flowering and have long been consumed as infusions, syrups and preserves in West Africa, the Caribbean and Southeast Asia [1,2]. Interest in the crop as an industrial ingredient rests on a single number: calyx anthocyanin content is commonly reported between 150 and 1500 mg per 100 g, roughly three times that of black grapes [1]. As manufacturers retreat from synthetic reds, that pigment density is what makes the plant worth processing.

The difficulty is that the pigment is fragile and the crop is not. Calyces are seasonal, perishable and 80 to 90% water, so some form of stabilisation is unavoidable, and every stabilisation step has a cost in pigment [1]. The literature on this has grown in two directions that rarely meet. One body of work studies the unit operations — drying, milling, powder properties and reconstitution — from a process engineering standpoint [1]. Another studies extraction solvents, pigment profiles and formulation performance from a food chemistry standpoint [3]. A manufacturer has to make decisions across both.

This review follows the material through the whole chain and asks, at each step, what is lost, what the operator controls, and what the published evidence actually supports. It draws principally on a recent process-side review [1] and a recent extraction and formulation study [3], sets their findings against the wider literature, and compares them throughout with butterfly pea (Clitoria ternatea), the other anthocyanin colourant of commercial importance in the region, which is treated in a companion review [4]. The two species are chemically opposite, and the contrast is instructive rather than incidental.

2. The Pigment, and Why Roselle Is the Mirror Image of Butterfly Pea

Roselle colour comes from two principal anthocyanins, delphinidin-3-sambubioside and cyanidin-3-sambubioside, accompanied by minor amounts of the corresponding 3-glucosides [1,3]. In the calyx these have been reported at roughly 71% and 29% of total anthocyanins respectively [1]; in a recent aqueous extract they were quantified at 2755.81 and 1842.89 mg per kilogram, a 60:40 split, with identification confirmed by molecular and product ions at m/z 597 and 303 for the delphinidin derivative and 581 and 287 for the cyanidin derivative, each losing a sambubiose unit of 294 mass units [3].

What matters industrially is what these structures lack. They are glycosylated but not acylated. There are no aromatic side chains to fold back over the chromophore, so roselle anthocyanins have none of the intramolecular copigmentation that stabilises the polyacylated ternatins of butterfly pea [4,5]. The consequence is that roselle behaves the way anthocyanin textbooks say anthocyanins behave: the red flavylium cation dominates at low pH and is the stable species, colour is lost as pH rises through hydration at C2, and the useful formulation window is acidic.

This is not a deficiency so much as a division of labour, and it is the reason the two plants are complementary rather than competing. The calyx is naturally acidic, at pH below 3, because succinic and oxalic acids together account for some 76% of its organic acid content, with malic, citric, tartaric and ascorbic acids making up the remainder [1]. The plant therefore arrives with its own stabilising environment, and its pigments are at home in exactly the acidified beverages and confectionery where butterfly pea is least comfortable. A formulator choosing between them should be choosing on finished-product pH before anything else: roselle below pH 4, butterfly pea between pH 5 and 8 [4].

Table 1. Reported composition of Hibiscus sabdariffa calyces, compiled from M’be et al. [1] and Vargas et al. [3].

Component	Reported range or value	Note
Water (fresh calyx)	80–90 g/100 g	81.35% measured in a recent study; water activity 0.99
Protein	0.9–17.9 g/100 g	Wide range reflects variety and origin
Lipid	0.1–3.9 g/100 g	—
Fibre	2.3–12 g/100 g	Relevant to swelling behaviour on reconstitution
Carbohydrate	3.3–12.3 g/100 g	Approximately 40% glucose
Total anthocyanins	150–1500 mg/100 g	Approximately three times black grapes; principally delphinidin-3-sambubioside and cyanidin-3-sambubioside
Ascorbic acid	Approximately 72 mg/100 g	Higher than orange juice at 49–54 mg/100 g
Organic acids	Succinic and oxalic approximately 76% of total	With malic, citric, tartaric and ascorbic acids
pH	Below 3	Intrinsic acidity stabilises the flavylium cation
Minerals	Potassium, calcium, magnesium, iron, manganese, zinc	Dissolution favoured by the low pH

Table 1. Reported composition of Hibiscus sabdariffa calyces, compiled from M’be et al. [1] and Vargas et al. [3].

Component	Reported range or value	Note
Water (fresh calyx)	80–90 g/100 g	81.35% measured in a recent study; water activity 0.99
Protein	0.9–17.9 g/100 g	Wide range reflects variety and origin
Lipid	0.1–3.9 g/100 g	—
Fibre	2.3–12 g/100 g	Relevant to swelling behaviour on reconstitution
Carbohydrate	3.3–12.3 g/100 g	Approximately 40% glucose
Total anthocyanins	150–1500 mg/100 g	Approximately three times black grapes; principally delphinidin-3-sambubioside and cyanidin-3-sambubioside
Ascorbic acid	Approximately 72 mg/100 g	Higher than orange juice at 49–54 mg/100 g
Organic acids	Succinic and oxalic approximately 76% of total	With malic, citric, tartaric and ascorbic acids
pH	Below 3	Intrinsic acidity stabilises the flavylium cation
Minerals	Potassium, calcium, magnesium, iron, manganese, zinc	Dissolution favoured by the low pH

3. Drying: Where Most of the Loss Occurs, and Why Temperature Alone Is the Wrong Specification

Drying is unavoidable and it is where the largest single pigment loss happens. Solar drying, the default in producing regions, removes roughly 90% of the water over four to eight days but has been reported to cost 15.3% of phenolics and 36.9% of anthocyanins [1]. Oven drying at 40 to 45 °C reaches a comparable water content in about a day; higher oven temperatures reach 8% dry-basis moisture in hours. Dehumidified air drying operates at room temperature with relative humidity held below 20% and achieves the highest drying rate of the three, 0.212 g water per gram dry matter per minute, in about twenty hours (Table 2) [1]. Comparisons of drying strategies report that the choice of method changes not only retention of bioactive constituents but the colour of the dried calyx itself [6].

The finding that should change practice concerns temperature. Comparing oven drying at 60, 80, 100 and 120 °C to a common end point below 8% moisture, phenolic content was highest at 80 °C, not at 60 °C [1,7]. The explanation is that the lower temperature required a much longer hold, and the extra hours of exposure to heat, light and oxygen cost more than the higher temperature did. Separately, anthocyanins were well preserved at 50 °C and 70 °C but lost at 60 °C, and when phenolics, anthocyanins and organic acids were weighed together, 70 °C was proposed as the best single compromise [1,8].

The practical consequence is that a drying specification written as a temperature limit is the wrong instrument. What governs retention is cumulative thermal exposure, the product of temperature and time, and a specification that caps temperature without capping duration can select for worse outcomes. For a manufacturer the implication is to specify an end-point moisture and a maximum total drying time together, and to prefer the process that reaches the end point fastest at a moderate temperature rather than the one that runs coolest.

4. The Dehydration Paradox: Concentration Is Only Part of the Gain

Measured phenolic and anthocyanin contents rise sharply on drying, and the rise is usually attributed to the removal of water. Recalculating recently published figures on a dry-matter basis shows that concentration accounts for only a minority of the effect (Table 3).

In that study calyx moisture fell from 81.35% to 11.61%, so dry matter rose from 18.65% to 88.39% and the concentration factor is 4.74. Over the same step, total phenolic compounds rose from 1.33 to 20.60 mg gallic acid equivalent per gram, a factor of 15.5, and total anthocyanins from 0.21 to 6.10 mg per gram, a factor of 29.0 [3]. Dividing through by the concentration factor leaves an unexplained gain of roughly 3.3-fold for phenolics and 6.1-fold for anthocyanins. This arithmetic is ours rather than the original authors’, but the mechanism they propose accounts for it: fresh calyces contain mucilage that forms a viscous gel on contact with water and physically impedes anthocyanin extraction, and the intact cell wall limits solvent access to intracellular compartments; dehydration disrupts both [3].

Two consequences follow. The first is methodological: comparisons of fresh against dried material on a wet-weight basis overstate the benefit of drying by roughly fivefold, and even dry-matter comparisons conflate concentration with extractability unless they are separated explicitly. The second is commercial. If most of the apparent gain is extractability rather than pigment, then the limiting factor in a commercial process is not how much anthocyanin the calyx contains but how much of it the solvent can reach. That reframes drying as a pretreatment for extraction rather than merely a preservation step, and it suggests that mild cell-disruption technologies applied before or instead of thermal drying deserve more attention than they have had.

5. Milling and Powder Properties

Grinding increases specific surface area and improves biomolecule extraction, and it is where powder functionality is set [1]. It also introduces risks: local heating, Maillard reactions and further pigment loss, all of which scale with the energy put into the mill [1]. Ambient and cryogenic grinding have both been proposed as ways to raise bioavailability while limiting thermal damage, and particle size has been shown to affect the physicochemical and phytochemical characteristics of roselle powders directly [1,9], and drying and grinding parameters together determine the physicochemical properties of the resulting powder [10].

Particle size then propagates into every downstream property. Finer powders present more surface for extraction but swell more slowly, because the same high specific surface area that favours dissolution also raises the resistance of the particle bed to water ingress [1,11]. Porosity partly compensates: porous particles take up water by capillarity, so large porous particles and fine particles can show similar swelling kinetics [1]. Composition matters as well, with carbohydrate-rich and low-molecular-weight protein-rich powders swelling and dissolving quickly, while fibre-rich powders absorb water well but solubilise slowly [1]. Given that roselle calyx carries 2.3 to 12 g per 100 g fibre, the fibre fraction is a plausible and under-examined determinant of how a commercial roselle powder behaves in the customer’s vessel.

6. Extraction: Water Outperforms Alcohol

The default assumption in polyphenol extraction is that recovery improves with organic solvent. For roselle anthocyanins the published comparison points the other way. Against a hydromethanolic control, a food-grade aqueous extract recovered 30.7% more total anthocyanins, and a hydroalcoholic extract recovered 40.7% less than the aqueous one, so that the aqueous extract yielded 68.7% more than the hydroalcoholic (Table 4) [3]. Total phenolic compounds showed a narrower spread, with the aqueous extract slightly below the hydromethanolic control and above the hydroalcoholic.

The divergence between the phenolic and anthocyanin results is the informative part. Total phenolics track solvent polarity loosely across a broad and chemically heterogeneous pool, while the two sambubiosides are highly glycosylated and therefore strongly hydrophilic, so water is the better match for them specifically. A process optimised on total phenolics as the response variable will therefore not be optimised for colour, and for a colourant the anthocyanins are the product.

For a manufacturer this is an unusually favourable finding, because the best-performing solvent is also the cheapest, the safest, the easiest to permit and the one that requires no solvent-residue specification. It also means that aqueous extraction, which is what most commercial roselle extract already uses, is not a compromise made for regulatory convenience but the technically preferable route.

7. Formulation and Storage: Sucrose, pH and the Limits of Protection

Syrups prepared from the aqueous extract at sucrose levels of 0, 5, 15, 25 and 35% and held refrigerated at 4 °C for 30 days showed their highest anthocyanin concentration at 5% sucrose, 495.12 ± 40.66 mg malvidin-3,5-diglucoside per 100 g, with a loss of 12.71% over the storage period [3]. Higher sucrose levels did not protect better. This matters because sugar is often added to anthocyanin systems on the assumption that it lowers water activity and stabilises the pigment; within this range that assumption did not hold, and the formulator gains nothing in pigment retention for the calories added.

Applied across food matrices of differing pH, colour response was pH-dependent as expected, with hue angle correlating positively and colour saturation negatively against pH, at correlation coefficients of 0.86 and −0.77 respectively [3]. The practical boundary is therefore set by the matrix, not by the extract: a roselle colourant performs in acidic systems and loses both saturation and hue fidelity as the matrix approaches neutrality. Where a product must sit near pH 6 or above, roselle is the wrong pigment and an acylated anthocyanin source such as butterfly pea is the appropriate substitution [4].

8. Reconstitution: The Step Almost Nobody Specifies

For a powdered ingredient the last processing step belongs to the customer, and it is rarely specified. Reconstitution proceeds through wetting, sinking and swelling, dispersion and finally solubilisation, and a mechanism specific to grinding-induced roselle powder has been proposed in which wetting is followed by swelling, a rapid dissociation and fragmentation stage, and then near-instantaneous molecular solubilisation [1,11].

Two controllable variables govern how much pigment the customer actually recovers. The first is granulometry, where the finest fractions favour anthocyanin extractability. The second is the reconstitution medium, where agitation and a low temperature together improve both extractability and stability [1]. Rehydration behaviour is itself a function of how the powder was made, as shown for foam-mat dried roselle [12]. The second point deserves emphasis because it runs against intuition: hot water extracts faster but degrades the pigment while it does so, and for a colourant the net result of a hot reconstitution can be worse than a cold agitated one. An ingredient specification that states anthocyanin content but is silent on particle size distribution and recommended reconstitution conditions leaves a large part of delivered performance undetermined.

9. What a Commercial Extract Delivers

Almost all of the figures above come from laboratory extracts prepared for study. Commercial material behaves differently, and the gap is worth stating plainly. In a single-laboratory comparison of nine commercial botanical extracts under identical conditions, a 5:1 aqueous roselle calyx extract gave a half-maximal inhibitory concentration of approximately 26.4 mg per mL against the 2,2-diphenyl-1-picrylhydrazyl radical and did not reach 50% inhibition against the 2,2′-azino-bis(3-ethylbenzothiazoline-6-sulfonic acid) radical at the 50 mg per mL solubility limit, placing it in a moderate group roughly one twentieth to one fortieth as potent as a green tea extract on a mass basis [13].

The explanation is dilution rather than poor material. A commercial extract is a manufactured article: the pigment is accompanied by carrier, the extraction ratio sets how much calyx is represented per gram of powder, and the declared anthocyanin content of such products sits in the tens of milligrams per 100 g rather than the thousands of milligrams per kilogram reported for concentrated laboratory extracts [13]. Published anthocyanin figures for the calyx therefore cannot be read across to a finished ingredient without accounting for extraction ratio and carrier load. For buyers, the consequence is that anthocyanin content per gram of supplied powder, not per gram of calyx, is the specification that matters.

10. Research Gaps

Four gaps follow, ordered by what would most change practice. First, the decomposition presented in Section 4 needs direct testing rather than recalculation: studies that separate concentration from extractability by design, using matched dry-matter bases and a disruption control, would settle how much of the drying benefit is matrix opening. Second, cumulative thermal exposure should replace temperature as the reported drying variable, so that results from different dryers become comparable. Third, the solvent comparison that favours water has been made on one material from one origin, and needs replication across cultivars before it can be treated as general. Fourth, almost all stability data stop at 30 days and at refrigeration; ambient storage over commercial shelf life, in the acidic matrices where this pigment is actually used, remains largely unreported.

11. Conclusions

Roselle is a dense and inexpensive source of anthocyanins whose pigments are chemically simple and correspondingly dependent on an acidic environment, which the calyx conveniently supplies. The processing chain, not the raw material, decides how much of that pigment reaches a product. Drying should be specified as cumulative thermal exposure rather than temperature, because shorter and hotter can outperform longer and cooler. Dehydration should be understood as a pretreatment that opens the matrix, since most of the apparent gain in measured pigment is extractability rather than concentration. Extraction should be aqueous, which is both the best-performing and the cheapest route. Sucrose above a few percent adds nothing to pigment retention, and finished-product pH sets the hard boundary of application, beyond which an acylated pigment such as that of butterfly pea is the correct substitution rather than a higher dose of roselle.