Chemical ecology of tannins and other phenolics: we need a change in approach


Correspondence author. E-mail:


1. Tannins are one of the most studied groups of plant secondary metabolites in research related to chemical ecology. They are traditionally thought to form an important factor of plant defence against herbivorous insects.

2. For a long time, tannins’ anti-herbivore activity was thought to derive from their protein precipitation capacity that rendered plant tissues non-nutritious and unpalatable for herbivores. Recent evidence suggests that tannin activity cannot be explained quite this simply, as tannin oxidation should also be taken into account as a defence mechanism for plants.

3. Tannins show very high variability in their structures with several hundred unique molecules detected in plants. These molecules are unevenly distributed in the plant kingdom and only very seldom—if ever—do two plant species share the same tannin pool. In many cases tannin composition varies even within organs of the same plant species and individual. Still, the overall tannin composition of many plant species is as of yet unknown.

4. Chemical ecology of tannins is challenging due to its multi-disciplinary nature. To facilitate research on tannins, we must provide ecologists and chemists with methodological and collaborative alternatives that enable the true and holistic investigation of all important questions that may arise from the field. So far this has not been possible with the tannin oxidation hypothesis, since simple and widely usable methods have not been available.

5. The aim of this review is to give a clear but detailed view of the chemical ecology of tannins and the methodology used to study them. In addition, we introduce a new method to estimate the oxidative activity of all types of tannins and other phenolics that might cause oxidative stress to herbivores. Hopefully our arguments and method will lead to clear changes in the approaches we take to tannins and their exciting biological activities, and we will witness a new era of flourishing and productive research in the chemical ecology of tannins.


The chemical complexity and heterogeneity of plant tannins means that they do not lend themselves to ready quantitative assessment, and this has produced a confused picture of their real significance – both evolutionary and ecological…” (Haslam 1988)

Plants offer an enormous, but not insuperable challenge for anyone trying to work out in detail the products of their chemical machinery. A typical plant cell synthesizes hundreds or thousands of organic molecules that do not take part in the cell’s primary functions such as growth and reproduction (Macel, van Dam & Keurentjes 2010). Therefore, these molecules are called secondary metabolites, which slightly underestimates their real significance for plants. Whereas, primary metabolites are essential for the growth of single plant cells, many secondary metabolites are essential for the survival of the plant individual, mainly due to their defensive properties against insect herbivores (Haukioja 1991, 2003, 2005). Moreover, most plant species share the same pool of primary metabolites (e.g. carbohydrates and amino acids). The abundance of plant primary metabolites and their relatively simple structures and chemistry has enabled the development of routine analysis methods for their detailed quantification and the production of high-quality commercial standards. The opposite is true for most secondary metabolites and especially for their chemically most complex and challenging subgroup, tannins.

Chemical ecology of tannins is challenging in many ways. Perhaps the biggest challenge is included in the term ‘chemical ecology’ itself. It is anything but simple to couple chemistry and ecology smoothly: most ecologists tend to see chemistry as the discipline of complicated structures and non-understandable or non-approachable methods. By contrast, most chemists see ecology as the discipline that tends to create all explaining theories and is ready to utilize chemistry only when it is simplified to the extreme. Luckily this is not true for all of us, since there is no such discipline as chemical ecology without both of the counterparts. The success of chemical ecology as a field demands that we provide both parties (ecologists and chemists) such methodological and collaborative alternatives that enable the holistic investigation of all intriguing questions that may arise from this challenging field. Having said this, so far the holistic approach has not been applied to tannins, or their biological activities either. More specifically, we have almost fully neglected one group of tannins – ellagitannins (ETs) – and one type of tannin activity – oxidation – although both of these might form a major, but yet non-studied chemical defence for many plant species against herbivores.

Traditionally, it was thought that tannins’ protein precipitation capacity (PPC) was solely responsible for their anti-herbivore activity. High levels of tannins in the diet of an herbivore were supposed to make the diet less nutritive via precipitation of proteins in the digestive tract. This was very logical reasoning and raised the importance of condensed tannins (CTs) and gallotannins (GTs) over ETs, since CTs and GTs have generally much higher PPC than ETs (e.g. Haslam et al. 1992; Kilkowski & Gross 1999). For this reason tannin analysis in ecological studies has mostly concentrated on total CTs (easily analysed by the acid-butanol assay) or on ‘total tannins’ altogether (easily analysed indirectly by measuring the PPC of plant extracts). Unfortunately this approach has severely neglected the less PPC-active ETs (cf. Moilanen & Salminen 2008). In addition, the crucial role of protein precipitation of tannins was strongly challenged by findings that showed tannin–protein interactions can be fully inhibited in many arthropod herbivores by gut surfactants (Martin, Martin & Bernays 1987) and highly alkaline gut conditions, especially in the lepidopteran larvae (midgut pH 9–12; Feeny 1969; Dow 1992; Appel 1993; Gringorten, Crawford & Harvey 1993; Skibbe et al. 1996; Harrison 2001; for a list of pH values, see Gross, Brune & Walenciak 2008). The PPC reactions would require acidic to neutral conditions, whereas the high pH in caterpillars might favour tannin auto-oxidation. For this reason Appel (1993) hypothesized that tannins might derive much of their anti-herbivore activities from being ‘oxidatively activated’ at high pH, and not via hydrogen bond derived protein precipitation at low to neutral pH: the products and byproducts of tannin oxidation could damage nutrients in the gut lumens of insect herbivores or produce cytotoxic effects in their tissues (Felton et al. 1992; Thiboldeaux, Lindroth & Tracy 1998; Hagerman, Dean & Davies 2003). Besides, tannins and other polyphenols may be oxidized at low to neutral pH by plant polyphenol oxidases (PPOs) and peroxidases as well, as already noted in Appel’s seminal paper (Appel 1993). Therefore, if we understood in detail tannin oxidation and its consequences to herbivores that would significantly widen our views of potent anti-herbivore activities of tannins.

However, for over 15 years Appel’s clever oxidation hypothesis lacked solid scientific evidence, in part because we knew little about which types of tannins might be prone to oxidation and which not. This gap has been significantly closed during the past 5 years through collaborations between insect biologists and tannin chemists (Barbehenn et al. 2006a,b, 2009a,b; Barbehenn, Weir & Salminen 2008; Moilanen & Salminen 2008; Roslin & Salminen 2008). First, Barbehenn et al. (2006a) tested the in vitro pro-oxidant activity of several purified and chemically well-characterized tannins. The results of these measurements were in striking contradiction with the traditional assumptions of which tannins were active in plant defence. They found that the most active protein precipitants, GTs and CTs, were the least oxidatively active, while ETs, the weakest protein precipitants were clearly the most oxidatively active tannins (Barbehenn et al. 2006a). Strikingly, certain types of CTs were even found to inhibit the pro-oxidant activity of ETs (Barbehenn et al. 2006b), whereby the CTs functioned as biological antioxidants (cf. Hagerman et al. 1998; but see Barbehenn et al. 2009a). Further evidence showed that foliar phenolics with high ET content oxidize more efficiently in the insect midgut than do foliar phenolics containing high CT content (Barbehenn, Weir & Salminen 2008). Finally, it was shown that even the individual ETs differ sevenfold in their pro-oxidant activity—the most active ones being constructed on acyclic glucose core and least actives on cyclic glucose core—and that the level of activity could be predicted by knowing the exact ET structure (Moilanen & Salminen 2008). However, it should be noted that the true role of ETs or tannin oxidation in ecological studies is yet to be unambiguously determined, since this would require independent results from multiple plant–insect systems. There is a great need for research methods that are relevant for studying the oxidation hypothesis, and easy enough to implement that they can be employed by a wide community of biologists with limited chemistry skills. So far these methods have not been available to most of us.

The aims of this review are twofold. First, we review the chemistry, biosynthesis and ecological importance of tannins. Secondly, we give a detailed overview of the current methodologies employed to understand the chemical ecology of tannins in plant–herbivore research. The ultimate purpose is to raise discussion among researchers of this field about existing methodological pitfalls, and how new methods could raise the bar in understanding the consequences of tannins for interactions between plants and their biotic and abiotic environment. However, since the very best and most reliable methods to quantify individual pro-oxidant ETs with liquid chromatography mass spectrometry are not available to the general ecologist, a new simplified method is introduced for the estimation of all types of pro-oxidant phenolics. This method uses a high-quality commercial standard and is suitable for any laboratory as it requires no special skills or equipment, other than a plate reader or a spectrophotometer. In practise, this method measures the proportion of ETs or other phenolics that is easily oxidized at high pH in any plant extract. We hope that with the help of the new method, all plant–herbivore scientists will have the ability to conduct thorough and rigorous tests of the functioning of Appel’s original tannin oxidation hypothesis in several plant–herbivore systems (Appel 1993). If the true tannin composition of all the studied plants was simultaneously determined in collaboration with analytical phytochemists, the true significance of ETs as plant defence compounds could perhaps be unambiguously revealed. Hopefully the method and especially the forthcoming discussion will change the approaches to tannin quantifications with respect to their exciting biological activities.

About tannin definitions and activities

Before going into the details of tannin methodology, it is important to first discuss what we mean by the term ‘tannins’. In our view, the traditional definition of tannins confuses their true ecological significance. Tannins are typically defined as water-soluble phenolic compounds that are able to bind and precipitate proteins and other macromolecules within aqueous solutions. Additionally, tannins are known to bind metals and to form blue to black complexes with iron(III)salts, and to have molecular masses between 500 and 3000 g mol−1. This traditional view of tannins also describes the principle mechanism by which tannins’ anti-herbivore activity was thought to be derived; the nutritive content of a herbivore’s diet could be severely affected if tannins bound and precipitated all the proteins and micronutrients. However, it is problematic to generalize tannin definitions or activities in this way, because it would exclude many ETs that do not conform to this definition because of their low protein binding capacity at low to neutral pH.

We should be able to see and define tannins in the same light that herbivores do. While browsing a single plant, herbivores do not consume only tannins that bind proteins or trace elements, or only those tannins that cause oxidative stress via auto-oxidation at high pH or via enzymatic oxidation at lower pH. They do not consume only CTs or ETs either. They swallow all the tannin molecules that are present. It is these tens or hundreds of plant-specific individual molecules that subsequently cause effects on herbivores. The magnitude of a tannin’s effects depend in part on its concentration, but especially structure, since all biological activities of any chemical come to fruition due to the specific physical properties of a given chemical.

How are the activities of specific tannin structures triggered in the herbivore? Appel (1993) thoroughly described most of the possibilities. To summarize, it is just as important to know the pH conditions prevailing in different parts (mouth, foregut, midgut, hindgut etc.) of the herbivore species as it is to know the exact tannin structures of the plant species. Why? The very same tannin structure may show high activity, or no activity, in different herbivore species or herbivore organs, depending on the environment. As a rule of thumb, mammals have neutral to acidic digestive organs. By contrast, the pH of the midgut in most insects tends to be neutral to alkaline (even pH 12), and other gut regions in insects are typically neutral to acidic (Harrison 2001). Thus, the oxidatively active tannins are likely to function in the alkaline midguts of many insect species, where the PPC-active tannins are inactive. Likewise, the PPC-active tannins are likely to function in the more acidic environment, where the pro-oxidant tannins are inactive. It is interesting that the very same plant sample may contain both pro-oxidant and PPC-active tannins and that these may function in series in the herbivore, depending on the successive chemical environment. In addition, plant tannins may be partially oxidized by plant PPOs during the very act of plant browsing by the herbivore. This, in turn, must have direct effects on the capacity of the tannins to subsequently function via PPC or pro-oxidant activities. Unfortunately, high-quality tannin studies have not yet unambiguously revealed the tannins most prone to PPO oxidation. Nevertheless, it appears that the structural features that determine tannin oxidation at high pH are not the same that determine their oxidation by PPOs at lower pH’s (Moilanen & Salminen 2008).

Finally, one neglected but important tannin activity relates to the hydrolysis of hydrolysable tannins (HTs). It is known that galloyl glucoses and ETs produce at least gallic and ellagic acids in the midgut, hindgut and faeces of Lepidopteran insects and that the level of hydrolysis varies among insect individuals (Salminen & Lempa 2002; Barbehenn et al. 2009b). So far there have not been any insect studies truly investigating the effects of the hydrolysis products as they are released from ETs in an insect. Nevertheless, we would not be highly surprised even if many of the yet non-identified effects of tannins on herbivores would be based rather on tannin metabolites (resulting from oxidation or hydrolysis) than on intact tannins themselves. A similar view has recently emerged to explain the positive effects of tannins and other polyphenols on human health (cf. Del Rio et al. 2010).

The most unambiguous tannin definition is based on their chemical structures

It is commonly accepted that tannins are divided into three main groups: HTs, proanthocyanidins (CTs) and phlorotannins. Phlorotannins are found mainly in marine organisms such as brown algae (Arnold & Targett 2002) and they are structurally perhaps the most simple tannin group. Individual phlorotannins are composed of two or more phloroglucinol (Fig. 1a) units that are attached to each other via C-C or C-O-C bonds, thus yielding oligomers such as the tetrameric phlorotannin (Fig. 1b). Further structural variations may include additional OH-groups in the molecules or additional bonds between the monomers. This structure-based definition very clearly differentiates phlorotannins from other types of phenolic compounds.

Figure 1.

 Structures of phloroglucinol (a), the phlorotannin building unit, and a tetrameric phlorotannin (b) consisting of four phloroglusinol units.

Condensed tannins are the most common group of tannins and they typically consist of two or more monomeric (+)-catechin or (−)-epicatechin units (Fig. 2a–b). These types of CTs are called procyanidins (PC), while the other common CT group, prodelphinidins (PD), consist of (+)-gallocatechin or (−)-epigallocatechin units (Fig. 2c–d). Other types of common (Fig. 2e–f) or rarer (Fig. 2g–l) monomer units are shown in Figure 2. In plants, CTs are found as oligomers (two to ten monomer units) or polymers (>10 monomer units). Interestingly, Barbehenn et al. (2006b) showed that especially PD-rich polymers are able to inhibit ET oxidation at high pH in vitro, while PC oligomers and polymers were less effective. As can be anticipated even from the multiple monomeric units, the true structures of all individual CT oligomers and polymers in a plant sample are very difficult to determine. However, at least the average nature of the monomeric building blocks (PC/PD-ratio) and the mean degree of polymerization may be readily determined (Cheynier & Fulcrand 2003; Karonen et al. 2007). The number of monomer units in a polymer may be as high as 83, as shown by Cheynier et al. (1999); Fig. 2m), resulting in a molecular mass of over 23900 g mol−1.

Figure 2.

 Structures of the common (a–f) and more rare (g–l) monomeric building units of condensed tannins, and an example of the polymeric condensed tannins (m).

Structurally perhaps the most complex group of tannins are the HTs. This is the group that is little studied in plant–herbivore research and perhaps for this reason—in addition to the structural complexity—this group suffers from most of the errors in detection and/or quantification (see later). HTs are first divided into three subclasses such as simple gallic acid derivatives, GTs and ETs. Simple gallic acid derivatives contain five or less galloyl groups (Fig. 3) that are most commonly esterified to either glucose (monogalloyl and pentagalloyl glucoses in Fig. 3b, d) or quinic acid (monogalloyl quinic acid in Fig. 3c). Gallic acid derivatives that contain six or more galloyl groups are defined as GTs and are further characterized by having one or more digalloyl groups (heptagalloyl glucose in Fig. 3e; Gross 1999). The number of galloyl groups in the molecules increases PPC, but simultaneously oxidative capacity is lowered (Kilkowski & Gross 1999; Barbehenn et al. 2006a). Although GTs are rarer in nature than ETs, they are the main components (hexa to decagalloyl glucoses) of commercially available tannic acid and thus responsible for the high PPC (and the low pro-oxidant activity) of this commonly used hydrolysable tannin preparation. Gallic acid and several simple gallic acid derivatives, such as in Fig. 3b–c, are found in most tannic acids as impurities and they may cause tannic acid to show some pro-oxidant activity. Fig. S1 in Supporting Information shows that tannic acid is a mixture of tens of polyphenols and the polyphenol content differs significantly within the tannic acids. Figs S1 a–e represent HPLC chromatograms of proper tannic acids with high GT content (GTs surrounded by the broken line) and low content of impurities (peaks not surrounded by the broken line). Figs S1 f–l show that lower quality tannic acids may contain more impurities than actual GTs. Most strikingly, chromatograms S1a and S1l are from two different lots of J.T. Baker’s tannic acid sold under the very same product code, but during separate years. It is clear that commercial chemistry firms cannot be fully trusted on this issue and it is highly recommended that researchers use at least HPLC-DAD (Fig S1, methods used described in Salminen et al. 1999) to check the composition of different tannic acid batches or lots before use. Additionally, mass spectrometric analysis will specifically reveal if the tannic acid consists mostly of quinic acid based impurities (see Fig. S2a) or glucose based GTs (see Fig. S2b; J.-P. Salminen & S.H. McArt, unpublished data).

Figure 3.

 Examples of the structures of simple gallic acid derivatives (b–d) and gallotannin (e).

Gallic acid derivatives and GTs are much rarer in plants than are ETs. The diverse ETs structures described to date (>500) may be divided into six subgroups: hexahydroxydiphenoyl (HHDP) esters (Fig. 4b), dehydro-HHDP esters (Fig. 4c) and their modifications (Fig. 4d), nonahydroxytriphenoyl (NHTP) esters (Fig. 4e), flavonoellagitannins (Fig. 4f), and oligomers with both varying degree of oligomerization and types of bonds between the monomers (Fig. 4g-h). For simplicity, all the ETs in Fig. 4 contain ET-typical HHDP groups as parts of their structures. One of the simplest ET definitions shows them as compounds yielding ellagic acid upon hydrolysis of the HHDP group (see Fig. 4a–b). However, ET biosynthesis yields also further metabolites where HHDP groups are modified via biosynthetic oxidation. These do not produce any ellagic acid upon hydrolysis and they can be recognized as ETs only by knowing their biosynthetic origin in the ET pathway. Still, on the basis of their structures, ETs are distinct from all other tannin groups.

Figure 4.

 Examples of the structures of the ellagitannin subgroups.

About tannin biosynthesis

It was shown above that tannin structures can be very complex and to unravel all of them is a tedious, but an exciting task for any phytochemist. However, even though most ecologists may not be interested in the esoteric details of variation in tannin structure, knowledge of tannin biosynthesis can provide insight into the differences in tannin content among species and their tissues, and ultimately their ecological and evolutionary consequences. For example, a typical plant cell does not produce both GTs and ETs simultaneously, and the efficient production of ETs must have direct negative effects on CTs and flavonoid biosynthesis. For this reason both ETs and CTs do not accumulate to high concentrations in the same tissue: they typically show opposite seasonal patterns in their contents, such that ETs peak in young tissues whereas CTs are most abundant in mature leaves (Ossipov et al. 2001; Salminen et al. 2001, 2004; Riipi et al. 2002).

Plant cells (chloroplasts) fix carbon dioxide via the Calvin cycle into glyceraldehyde-3-phosphate, which can be further transformed and accumulated into storage carbohydrates such as sucrose and starch. These are then degraded when necessary either by glycolysis (main products: glyceraldehyde-3-phosphate, phosphoenolpyruvate and pyruvate) or through the oxidative pentose phosphate pathway (main products: erythrose-4-phosphate and glyceraldehyde-3-phosphate) into more simple molecules (Fig. 5). Glyceraldehyde-3-phosphate has a special role, since it is readily available from the Calvin cycle, and can be partitioned between the two carbohydrate-degrading pathways as well, depending on demand. The first branch of the pathway occurs between the acetate/malonate and shikimate pathways. Both of these pathways are essential for the biosynthesis of CTs, while ETs rely solely on the shikimate pathway (Fig. 5). If significant levels of glycolytic phosphoenolpyruvate are directed for the shikimate pathway (together with erythrose-4-phosphate), the production of pyruvate for the needs of acetate/malonate pathway is significantly reduced. This has direct negative effects on CT biosynthesis, since they would need malonyl-CoA as one of their building blocks. However, all plant tannins rely on the efficient function of especially the shikimate pathway, while the phlorotannin-producing brown algae only need the acetate/malonate pathway for their tannin synthesis.

Figure 5.

 A biosynthetic scheme showing the various branching points in the biosynthesis of different types of tannin groups, and other phenolics as well.

The second major branch point occurs at 3-dehydroshikimic acid (Ossipov et al. 2003). This is the precursor for the synthesis of gallic acid, the primary building block of all HTs. Efficient production of gallic acid negatively affects the synthesis of shikimic acid and its products: CTs, flavonoids, and caffeic and coumaric acid derivatives (Fig. 5). The hydrolysable tannin pathway contains a third major branch point at pentagalloyl glucose, the precursor for both GTs and ETs. To date, there are no reports of the simultaneous production of significant amounts of both of these hydrolysable tannin classes in the same plant tissue. Moreover, coniferous plants are only reported to synthesize CTs and never HTs (see also Kubitzki & Gottlieb 1984).

The actual tannin pathways are in most parts hypothetical, since only the galloyl glucose and GT pathways and two steps of the ET pathway are characterized by enzyme studies (Xie & Dixon 2005; Gross 2009). Otherwise the details of the pathways are deduced from the careful inspections between tannin structures and their seasonal occurrence in the plant cell (e.g. Okuda et al. 1987; Salminen et al. 2001, 2004; Yarnes, Boecklen & Salminen 2008) or by the chemical synthesis of the one tannin from the other (Okuda et al. 2009). The galloyl glucose and GT pathways are rather simple, since larger tannins are synthesized by adding new galloyl groups into the smaller tannins. According to the current view, CTs are thought to be formed by condensation of two or more monomeric units (see Fig. 2). Although the ET biosynthesis is not fully elucidated, its main and most probable steps are shown in Figure 6. It can be seen that the ET pathway contains branching points that could restrict a plant species to contain only specific ET classes as the main metabolites. For example, in the species of Geraniaceae and Punicaceae, dehydro-HHDP esters are the main ETs accompanied by lower levels of HHDP esters, while the Combretaceae and Euphorbiaceae accumulate modified dehydro-HHDP esters (Okuda et al. 2009; Yoshida et al. 2009). Monomeric HHDP esters typically accumulate in the Rosaceae, and monomeric NHTP esters in Casuarinaceae, Lythraceae and Myrtaceae, among others (Okuda et al. 2009; Yoshida et al. 2009). Moreover, the distribution of the over 500 reported ET structures within the seven ET groups (Fig. 6) is such that each plant species contains a rather unique ET composition. However, the oligomeric HHDP esters may have some chemotaxonomic significance. For example, oenothein B (Fig. 4g) is typical for Oenothera and Epilobium and gemin A (Fig. 4h) for Geum. Still, it must be noted that new chemical methods are constantly being developed to detect and characterize ETs and if these methods are efficiently utilized, then clearly new ETs will be constantly revealed from the study of additional plant species and from the application of new methods. For instance, until 2010 no one had found ETs larger than pentamers in nature. However, recent mass spectrometry evidence has shown that ETs up to heptamers (seven monomeric units in the same molecule) along with traces of octamers (eight units) and nonamers (nine units) can be found in roots and leaves of Oenothera biennis (evening primrose; Karonen et al. 2010). These very large ETs, with molecular masses up to 7056 g mol−1 have never been tested for activity.

Figure 6.

 A simplified view of the current view of the ellagitannin pathway, showing the major ellagitannin subclasses and their biosynthetic connections. It is not yet fully known if the dehydro hexahydroxydiphenoyl (HHDP) esters derive only from pentagalloyl glucose or also from HHDP esters.

Methods used for tannin quantifications

Given the structural variability among the tannins, each group and compound should be quantified separately. In principle all HTs can be quantified using liquid chromatography coupled with mass spectrometry (e.g. Salminen et al. 2001; Yarnes et al. 2006). Nevertheless, this still requires a priori identification of single metabolites. On the other hand, the polymeric CTs of many species are so complex, with many isomeric forms, that they simply cannot be separated from each other even by modern chromatographic techniques (e.g. Karonen et al. 2004, 2006; Kelm et al. 2006). For these reasons, and for having to analyse hundreds or even thousands of samples, most ecological studies still use simplified tannin quantifications instead of accurate analysis of individual molecules.

The methods for simple tannin quantifications differ in their sensitivity, but especially in specificity. In this respect, CTs are the easiest and safest tannin group to be quantified, since the widely employed spectrophotometric HCl–butanol assay (Swain & Hillis 1959; but see Schofield, Mbugua & Pell 2001) is highly CT-specific. It depolymerises plant CTs to colourful monomeric units that are then detected and quantified at approximately 550 nm. In practice, only plant pigments such as blue to violet anthocyanins or anthocyanidins cause false positives with this method, but even these can be subtracted from the final result if the absorbance is read before and after the acid-catalyzed depolymerisation reaction of the CTs. The most severe problem associated with the HCl–butanol assay is the lack of proper quantification standards; this may cause significant overestimation of the CT contents in the analysed sample (see Rautio et al. 2007). The problem is especially pronounced if commercial quebracho tannin is used, since its CTs are the relatively rare 5-deoxy-CTs (Fig. 2) that give 30-fold lower absorbance with the HCl–butanol assay than the more common PC- or PD-rich CTs (Schofield, Mbugua & Pell 2001; Vivas et al. 2004). Thus, the CT content may be reported as 300 mg g−1 if quebracho is used as the standard, when in fact the true CT content is 10 mg g−1. Nevertheless, the HCl–butanol assay is a very reliable method in estimating the CT levels (low, middle or high) between samples of the same plant species even though absolute values are impossible to obtain.

Hydrolysable tannins are more problematic than CTs to quantify, because they produce multiple hydrolysis products of different chemical nature (Mueller-Harvey 2001). Two of the most common hydrolysis products, gallic acid (from a galloyl group) and ellagic acid (from an HHDP group), are shown in Figs 3a and 4a. There are at least three methods appropriate for quantification of both of these after HT hydrolysis: the rhodanine assay (Inoue & Hagerman 1988), together with the modified potassium iodate technique (Hartzfeld et al. 2002), are currently the best available methods for an approximately estimation of galloyl-containing HTs, while the sodium nitrite method (Bate-Smith 1972; Wilson & Hagerman 1990) is the only simple method to estimate HHDP-containing ETs. In our opinion, none of these methods work optimally in isolation, but the best coverage of the plant HTs is obtained when both galloyl- and HHDP-liberating HTs are quantified separately by two methods and results of these quantifications are then combined. The logic behind this argument is simple: galloyl is the sole group that is attached to the carbohydrate core of simple gallic acid derivatives and GTs (see Fig. 3), but in ETs its presence varies considerably and it may even be absent (structures in Figs 4b–h have zero to three galloyl groups). Thus ETs are clearly underestimated, or even neglected—in comparison to gallic acid derivatives and GTs—by the rhodanine and potassium iodate assays (see also Salminen 2003). However, the sodium nitrite method is not fully optimal for ETs either, since HHDP’s typically make only a portion of functional groups attached to the glucose moieties of ETs (structures in Figs 4b–h have zero to three HHDP groups and six other types of functional groups). Therefore, this method tends to underestimate the ET content as well. However, since galloyl groups do make one of the most common functional groups in ETs (in addition to the HHDP group), the overall underestimation of the total HT content of a plant sample would be significantly smaller, if both galloyls and HHDPs would be separately quantified by the modified potassium iodate technique (Hartzfeld et al. 2002) and the sodium nitrite assay (Wilson & Hagerman 1990), and then combined. Unfortunately, care must be taken with the use of pyridine in the sodium nitrite method, since its incorrect use may pose a toxicity risk. Finally, given the examples seen in the literature, it seems appropriate to note that neither of the galloyl-quantifying methods (Inoue & Hagerman 1988; Hartzfeld et al. 2002) is able to specifically detect and quantify GTs on their own. Therefore GTs cannot be reported to be found in plants unless they are truly identified by more specific methods like liquid chromatography connected to at least diode array or mass spectrometry detectors (see Figs S1 and S2).

Although opposite opinions seem to exist, there is no single method available for the determination of the total tannin content of a plant sample. Usually this is attempted by the methods that rely on the specific interactions of tannins with proteins. However, we have tried to show in our arguments presented above that not all tannins are equally precipitated by test proteins used in conventional assays; the most pro-oxidant ETs are especially prone to being poorly estimated. Therefore, if tannins with high PPC are quantified, simultaneously tannins with high oxidative capacity are neglected, and vice versa. Nevertheless, if one knows for sure that protein precipitation is the one mechanism by which tannins affect a certain herbivore species (e.g. mammals or other vertebrates with neutral to acidic gut pH), then it may be fully appropriate to focus only on those tannins that are efficient protein precipitants. There are several methods available and a suitable one should be found among those reported by Hagerman & Butler (1978), Hagerman (1987), Henson et al. (2004) and McArt et al. (2006).

A new method that estimates the oxidative capacity of any plant sample at high pH

One of the main purposes of this article is to show that tannins are diverse compounds that possess oxidative capacity in addition to their most appreciated PPC, and that the former activity may be an important determinant of tannins’ anti-herbivore effect. However, the tannin oxidation hypothesis (Appel 1993) has not been rigorously tested, mainly because of the lack of simple and suitable methods for this purpose. To address this shortcoming, we introduce a new method that could be used in any laboratory with basic facilities capable of measuring total phenolics with the Folin-Ciocalteau assay. Before describing the method, we want to emphasize that tannin or phenolic oxidation is not a simple issue and it will depend on multiple factors such as phenolic structure, exact pH value, concentration of molecular oxygen and metal ions, and the presence or absence of PPOs; no single and simple method is able to take all these into account. Nevertheless, although ETs may have the highest oxidative capacity of all the tannins (Barbehenn et al. 2006a; Moilanen & Salminen 2008), the new method will not be narrowly focused and biased to ETs only, but will measure all phenolics that might cause oxidative stress towards herbivores at high pH.

It is known that the oxidation of o-dihydroxy polyphenols typically yields o-quinones (Quideau, Feldman & Appel 1995; Feldman et al. 1999; Chen & Hagerman 2005), and by doing this the number of the original phenolic OH-groups is decreased in the phenolic structure. Figure 7 shows this process for the HHDP group of ETs: four of the six OH-groups have been oxidized by the loss of four electrons. The more sensitive a phenolic compound is to oxidise at high pH, the more phenolic OH-groups will be oxidized, and more rapidly. If the compound is not sensitive to oxidation, then the initial number of phenolic OH-groups will not change. The logic for the new method is very simple: phenolic hydroxyl groups are quantified by the Folin-Ciocalteau assay before and after oxidation of the phenolic sample by a pH 10 buffer. Altogether the method will yield three types of estimates from the tested sample: (i) the total phenolic content (in mg g−1 dry wt); (ii) the proportion of easily oxidized phenolics (the oxidative capacity as % of total phenolics); and (iii) the total content of oxidative phenolics (the oxidative capacity as mg g−1 dry wt). It should be noted that the method uses a modification of the common Folin-Ciocalteau test and we cannot guarantee functioning of the method if significant changes—that might cause precipitates to occur—are made to chemicals used in the method.

Figure 7.

 A simplified view of the oxidation of the hexahydroxydiphenoyl (HHDP) group of ellagitannins at high pH in the presence of molecular oxygen. The view shows two simultaneous mechanisms by which oxidative damage may be caused: by nucleophilic reactions of the oxidized HHDP group with proteins, and by the formation of highly reactive hydroxyl radicals via the Fenton reaction.

Five types of commercial chemicals are needed in the method: (i) pH 10 carbonate buffer (50 mm: buffer A, J.T. Baker, Deventer, Holland); (ii) 0·6% formic acid (buffer B, J.T. Baker, Deventer, Holland); (iii) a mixture of buffers A and B (buffer C: 9 volumes of buffer A and 5 volumes of buffer B); (iv) 1 N Folin-Ciocalteau reagent; and (v) 20% sodium carbonate (m/v). Gallic acid is used as the quantification standard, since it is readily available in a pure form; alternatively a species-specific phenolic standard could be produced by Sephadex LH-20 chromatography (see below). A stock solution (2·0 mg mL−1) is prepared by first dissolving the gallic acid in a volumetric flask with a few drops of ethanol and adding purified water to the mark. Several dilutions (0·1–2·0 mg mL−1) are then made by diluting aliquots of the stock solution with water.

All the phenolic extracts to be tested are also prepared in water in a known concentration, so that final results can be expressed as mg g−1 dry wt (gallic acid equivalents). Here are short general instructions for the preparation of the extract. Weigh the dried and ground plant material (10 mg) into a 1·5 mL Eppendorf tube and extract three times (3 × 2 h, 3 × 800 μL) with acetone/water (7/3, v/v) on a vortex mixer. Separate supernatant solutions after each extraction step by centrifugation in an Eppendorf centrifuge (10 min, 14 000 rpm) and decant them carefully into a new 2 mL Eppendorf tube. Concentrate pooled supernatant solutions into aqueous phase (but not to dryness) under reduced pressure by an Eppendorf concentrator at room temperature (start this immediately after second extraction to make room for the supernatant of the third extraction) and freeze-dry the aqueous residue. Add 500 μL (this volume may be adjusted for each sample type depending on the phenolic concentration) water on the top of the lyophilized residue. Shake the tube for 5 min and then centrifuge it (10 min, 14 000 rpm). Filter the supernatant solution (0·45 μm PTFE) or pipette it carefully into a new Eppendorf tube. All extractable plant tannins are found in this purified extract, while chlorophyll, waxes, carotenoids and other lipophilic impurities are left in the PTFE filter.

The assay method itself has four steps:

1. This step measures the initial total phenolic concentration of the extract (in mg mL−1 gallic acid equivalents) and this value can be used to determine the total phenolic content (mg g−1 dry wt) of the original plant sample. Take 20 μL of the solution (gallic acid solution or the phenolic extract) and add 280 μL of the buffer C. Shake well. Move 50 μL of this mixture into a single plate reader well, add 50 μL of 1 N Folin-Ciocalteau reagent and 100 μL 20% sodium carbonate solution. Shake the plate for 10 s, incubate at 25 °C (shake for 10 s every 10 min) for 60 min and read the absorbance at 730 nm. If you do this step 1 with several gallic acid dilutions first, you readily obtain a calibration curve that converts absorbance values of unknown samples to mg mL−1 (in gallic acid equivalents). At the same time the linear absorbance range of the instrument will be revealed. Three replicates of each sample will produce reproducible results (see Fig. 8a–c). This way it is possible to analyse 32 unknown samples at the same time with one plate. Figure 8a–c shows that the 60 min incubation time is more than enough to record the maximal absorbance of the incubation mixture at 730 nm, and that all incubation times <15 min will result in high measurement error.

Figure 8.

 Total phenolic absorbances for vescalagin (a), oenothein B (b) and ellagitannin rich Rubus chamaemorus extract (c) as a function of incubation time, both before and after the oxidation at pH 10, and the accumulation of colourful oxidation products during the pH 10 oxidation of vescalagin (d), oenothein B (e) and ellagitannin rich Rubus chamaemorus extract (f).

The goal of step 1 is to achieve an absorbance reading of 1·0 (±10%) for the unknown sample (see total phenolic absorbances before oxidation in Fig. 8a–c). If this is achieved, then it is possible to proceed directly to step 3. If 1·0 (±10%) absorbance is not achieved, then it is necessary to proceed to step 2 first, and then repeat step 1.

2. This step dilutes (or concentrates) all the phenolic extracts to the same total phenolic level (±10%). This is essential, since otherwise the following oxidation step will not produce comparable results for all of the unknown samples: more dilute samples will oxidize more rapidly than the concentrated ones. All extracts should be diluted to the level that is equal to a 1·0 (±10%) absorbance value after the total phenolic measurement of step 1; with our plate reader this is almost equal to the absorbance obtained with a 1·0 mg mL−1 gallic acid solution at step 1. For instance, if the absorbance at step 1 was 1·7, then the original phenolic extract needs to be diluted to approximately (1·7–1·0)/1·7 = 41%. All dilutions are made with purified water. After dilution/concentration, repeat the step 1.

3. This step oxidizes the easily oxidized phenolics in the extract. All extracts should now be at the same total phenolic concentration (±10%) and this concentration should be exactly known for each of the samples (in mg mL−1 gallic acid equivalents). Apply 20 μL of the diluted/concentrated extract into a single well. Add 180 μL of buffer A and shake the plate for 10 s. Incubate at 25 °C (shake for 10 s every 10 min) and add 100 μL buffer B after exactly 90 min. This will change the pH from 10 to 6 and the oxidation is simultaneously stopped. At least three replicate measurements are required here as well, such that 32 samples can be oxidized at the same time on a 96-well plate reader. Figure 8d–f shows the accumulation of colourful oxidation products during the oxidation of vescalagin (Fig. 8d), oenothein B (Fig. 8e) and ET-rich Rubus chamaemorus extract (Fig. 8f). It can be seen that the more pro-oxidant vescalagin (Moilanen & Salminen 2008) accumulates oxidation products more rapidly than the less pro-oxidant oenothein B and the ET-rich Rubus extract is in between. The 90 min oxidation time was chosen to achieve sufficient oxidation with most sample types to be detected by the total phenolic assay. However, it would be possible to oxidize the samples for only 30 min as well: comparison of the oxidative capacities (% of total phenolics lost) measured after 30 min and 90 min oxidations would reveal if phenolics of some plant samples are well oxidized already during 30 min (‘vescalagin-type’), or if they continue oxidizing between 30 and 90 min as well (‘oenothein B –type’). Of course, rapid oxidation must be more harmful for an insect than slow oxidation: the exact oxidation time may be freely optimized, depending on need.

4. This step measures the phenolics that are left non-oxidized after the step 3. Take 50 μL of the oxidized mixture of step 3 into a single plate reader well. Add 50 μL of 1 N Folin-Ciocalteau reagent and 100 μL of 20% sodium carbonate solution. Shake the plate for 10 s, incubate at 25 °C (shake for 10 s every 10 min) for 60 min and read the absorbance at 730 nm. Convert the absorbance to mg mL−1 gallic acid equivalents. The difference in the total phenolic concentration of the extract before and after oxidation will reveal the oxidative capacity of the sample as % of total phenolics (see Fig. 8a–c for total phenolic absorbances before and after oxidation). When this percentage value is multiplied by the total phenolic content (mg g−1) of the original plant sample, the oxidative capacity of the plant sample is revealed as mg g−1 dry wt (gallic acid equivalents). Hint: samples from the plate used in step 3 can be very conveniently transferred to the plate of step 4 by a 12-channel pipette.

We have tested this method on several phenolic extracts (e.g. species of Acer, Betula, Epilobium, Filipendula, Fragaria, Geranium, Lonicera, Lythrum, Oenothera, Pinus, Populus, Quercus, Rosa, Rubus, Salix, Sorbus and Vincetoxicum) and no precipitations occurred that could distort absorbance readings. Figure S3 shows the oxidative capacities (% of total phenolics lost during the 90 min oxidation) of extracts of 24 Finnish plant species. Most interestingly, the oxidative capacities of the phenolic extracts differed significantly, from 1·7% to 44·6%, although the total phenolic contents of the tested samples were the same. Therefore it seems clear that the proposed method is able to provide plant–herbivore researchers two new types of data to be correlated with herbivore performance: the oxidative capacity of the plant sample in both mg g−1 and in percentage of total phenolics. It will be interesting to learn whether the oxidative capacity of plant samples will have a negative net effect on herbivores, since positive effects might also arise via the electrophilic action of oxidized phenolics towards herbivore pathogens (Quideau, Feldman & Appel 1995; Feldman et al. 1999).

The estimation of the nature of the pro-oxidant tannins in the plant sample

If and when new types of results are obtained with the proposed new method in relation to herbivore performance, then it remains an interesting task to determine what specific types of phenolic compounds were the active ones in any given case. This way we may gain more understanding of the types of phenolics—ETs or others—that might be most important in plant defence. More specifically, this can be achieved by collaboration with phytochemists, but hints of the active components can also be achieved by ecologists themselves. This is because already today many groups utilize Sephadex LH-20 gel chromatography as the efficient way of producing crude tannin standards for tannin quantifications (e.g. Appel et al. 2001; Forkner, Marquis & Lill 2004). However, these Sephadex LH-20 methods partially reflect the current trends in ecological tannin research: the most pro-oxidant ETs are actually eluted by 95% ethanol (the wash solution)—and treated accidentally as waste—while the most protein-active tannins are eluted by 70% acetone—and treated as the true, purified tannins of the plant species in question. We propose such a fractionation scheme to be used with Sephadex LH-20 (as shown in Barbehenn et al. 2006b) that may enable separate elution of many types of phenolics with high oxidative capacity or high PPC. This method is described online in Supporting Information.

Concluding remarks

With this review we hope to raise discussion within the field of chemical ecology about the need for more holistic approaches for tannin research. In the future, all investigators interested in plant–herbivore interactions should have good possibilities for measuring (i) the oxidative capacity; (ii) the PPC; and (iii) the tannin groups responsible for these two types of activities. Moreover, when results are obtained about the relationship between herbivore performance and the oxidative capacity of plant tissues, then additional collaboration with tannin chemists could help us to understand the structural mechanisms for why tannins of one plant are highly active, while tannins of another plant remain inactive. Finally, it should be remembered that all tannin molecules do not need to be harmful. They may even be beneficial as has been recently shown for ruminants (see Mueller-Harvey 2006, 2009; Mueller-Harvey et al. 2007; Aufrère, Dudilieu & Poncet 2008; Brunet, Jackson & Hoste 2008), as well as the health-promoting effects of ETs towards humans (Adams et al. 2006; Malik & Mukhtar 2006), as most polyphenols are excellent antioxidants at low to neutral pH. Thus, we encourage all of you with acidic gut conditions to enjoy your share of ETs from pomegranate juice or Rubus berries (Törrönen 2009), while watching herbivores ingesting their potentially toxic dose of ETs from our study plants.


Our current knowledge of tannin chemistry and chemical ecology would not be possible without the help of numerous ecologically oriented collaborators: Ray Barbehenn, Tomas Roslin, Erkki Haukioja, Kyösti Lempa, Lauri Kapari, Anurag Agrawal, Marc Johnson, John Parker, Heidi Appel, Teija Ruuhola, Pasi Rautio, Chris Yarnes and Bill Boecklen just to name a few. JPS thanks Ann Hagerman and Vladimir Ossipov for fruitful discussions of tannin chemistry and chemical ecology, and Marc Johnson and three anonymous reviewers who gave fundamental help in improving the earlier versions of the manuscript. The practical help in the chemistry lab by Jari Sinkkonen, Maria Lahtinen, Riitta Koivikko, Johanna Moilanen, Anu Tuominen, Olli Martiskainen, Piia Saarinen, Kati Tuominen, Terhi Sundman, Matti Vihakas, Jaana Liimatainen, Meiju Puljujärvi, Tuuli Luomahaara, Kristiina Lehtonen, Kirsi Levänsuo, Jenny-Maria Brozinski, Victor Turhanen, Jonna Kenttä, Angelica Preetz and Anna Ctvrtnickova has been crucial as well. Funding by the Academy of Finland is acknowledged (grant no 119659).