A New Assay for Quantifying Brown Algal Phlorotannins and Comparisons to Previous Methods

Journal of Chemical Ecology, Vol. 22, No. 7, 1996

A NEW ASSAY FOR QUANTIFYING BROWN ALGAL PHLOROTANNINS AND COMPARISONS TO PREVIOUS METHODS

J. LEWIS STERN, I'* ANN E. HAGERMAN, 2 PETER D. STEINBERG, I FRANK C. WINTER, 3 and JAMES A. ESTES 4

~School of Biological Science, University of New South Wales Sydney, N.S. W. 2052 Australia ZDepartment of Chemistry, Miami University Oxford, Ohio 45056 ~University of Auckland, l_x,igh Marine Laboratories RD5 Warkworth, New Zealand 4National Biological Survey, University of California Santa Cruz, California 95064

(Received May 30, 1995; accepted March 4, 1996)

Abstract--Quantitative measurement of phlorotannins (polyphenolics) in brown algae (Phaeophyta) by colorimetric assays can be confounded because: (1) most such assays also react to nonphlorotannin substances (interferences) and (2) the appropriate reference compound for such assays is not always clear, although phloroglucinol is typically used. We developed a new assay in which 2,4-dimethoxybenzaldehyde (DMBA) reacts specifically with 1,3- and 1,3,5-substituted phenols (e.g., phlorotannins) to form a colored product, This new assay, as well as eliminating the problem of measuring interferences, is inexpensive, rapid, and can be used with small sample volumes. We recommend it for all assays of phlorotannins from one or a set of closely related species where the structural types of phlorotannins present are likely to be similar among samples. It is also appropriate for broader surveys of phlorotannin levels across many species, but in this case a reference must be chosen with care. We also compared the DMBA assay to existing assays, including the Folin-Denis Iboth before and after the samples were mixed with polyvinylpolypyrrolidone (PVPP)] and the Prussian blue assays. PVPP was not 100% efficient (and often much less) at removing phlorotannins from solution, and its effectiveness varied among different phlorotannins. Thus, in contrast to previous studies, measuring phenolic levels in extracts before and after

*To whom correspondence should be addressed.

1273

(g}98-0331/96/0700 1273509.50/0 rc~ 1996 Plenum Publishing Coz3~oration 1274 J.L. STERN ET AL.

treatment with PVPP will not necessarilyresult in an interference-freemeasure of phlorotannins. Based on an analysis of reactive substances in red and green algae (which do not contain phlorotannins) in the Folin-Denis and Prussian blue assays, we estimate that the average level of interferences (nonphlorotannins) in brown algae measured in these two assays is on the order of 0.5 % by dry weight.

Key Words--Phlomtannins, DMBA, Folin-Denis assay, brown algae, seaweeds, phenolics, PVPP.

INTRODUCTION

Polyphenolics are one of the most common classes of secondary metabolites in terrestrial plants (Harborne, 1991; Haslam, 1989; Swain, 1979), marine angio- sperms (McMillan, 1984), and macroalgae (Ragan and Glombitza, 1986; Stein- berg and van Altena, 1992). They are ecologically important in both terrestrial and marine systems, deterring or inhibiting herbivores, epiphytes, pathogens, or competitors (e.g., allelopathy) (Beress et al., 1993; Bernays et al., 1989; Hay and Fenical, 1988). In particular, there is a substantial empirical and the- oretical literature on the effects of polyphenolics against herbivores (Coley et al., 1985; Feeny, 1976; Steinberg, 1992; Swain, 1979). Polyphenolics can also significantly affect broad-scale ecosystem properties in both terrestrial (Schles- inger, 1991) and marine (Carlson and Mayer, 1983) systems. Although terrestrial and marine plant polyphenolics are similar in some respects, there are some fundamental differences among the compounds. Ter- restrial polyphenolics, or tannins, are polymers based on flavonoids or gallic acid (Haslam, 1989). Algal polyphenolics, or "phlorotannins" (Ragan and Glombitza 1986), which are only known from the brown algae (Phaeophyta), are restricted to polymers of phloroglucinol (1,3,5-trihydroxybenzene). Phlo- rotannins and terrestrial tannins vary in their linkages between monomeric units, in their size (dimers to 600,000 amu), and to some extent in the types of substitutions present (i.e., halogenation, sulfated groups) (Ragan and Glombitza, 1986). While the abundance and ecological importance of polyphenolics in natural systems has long been recognized, the ability to quantitatively measure these compounds in plant tissue remains a complex issue (Mole and Waterman, 1987a,b; Ragan and Glombitza, 1986; Ragan and Jensen, 1977; Van Alstyne, 1995). Because the compounds are typically large, polar, and water soluble, they are difficult to measure by the standard analytical techniques used to analyze smaller nonpolar metabolites such as terpenes and acetogenins. Analysis of algal polyphenolics is further complicated by the difficulty of separating the individual compounds from the polymeric mixtures which occur naturally in vivo. Consequently, polyphenolics have typically been analyzed in one of two QUANTIFYING BROWN ALGAL PHLOROTANNINS 1275 ways. The first is via various spectrophotometric procedures based on the reaction of a colored reagent with the easily oxidized phenolic functional group. The best known of these procedures is the Folin-Denis test, or variations thereof, such as the micro-Folin Denis (Arnold et al., 1995; Targett et al., 1995; Van Alstyne, 1995) or Folin-Ciocalteau (1927) tests. This assay and its variations have been used extensively in the analysis of phenolics from both marine and terrestrial plants (Ragan and Jensen, 1977; Mole and Waterman, 1987a,b; Stein- berg, 1989). The second set of methods uses the ability of polyphenolics to bind to macromolecules such as proteins, thus generating a measure of "relative tanning ability" (Hagerman and Butler, 1989). Such techniques have been very useful, but they are subject to a number of sources of error, of which two are perhaps the most significant for phlorotannins (Ragan and Glombitza, 1986; Steinberg, 1989; Van AIstyne, 1995). Firstly, because specific molecules are not analyzed, but rather total phenols or tanning ability, variation in the reactivity of structurally different polyphenolics (due to variation in size, structure, etc.) is not separable from variation in reactivity due to different amounts of compounds. Secondly, nonpolyphenolic interferences can react with the reagents used in some colorimetric assays, resulting in an overestimate of phenolic levels. In this paper we compared three spectrophotometric assays in an attempt to assess the variation in the three assays resulting from different types of phlorotannins from marine brown algae and to determine the effect of interferences in these assays. We compared the Folin-Denis technique, used with and without polyvinylpolypyrrolidone (PVPP), a resin which binds to, and removes polyphenols in solution; the Prussian blue assay, which has been suggested to be less sensitive than the Folin-Denis assay to interfering compounds; and a new assay, which is based on the ability of 2,4-dimethoxybenzaldehyde (DMBA) to react specifically with 1,3- and 1,3,5-substituted phenols (i.e., phlorotannins), and which does not react with tannic acid containing only ortho- and parahy- droxyl-substituted phenolics). We compared the color yield of these three techniques for both purified phlorotannin fractions and crude extracts from a number of brown algae. The effect of interferences on the measurement of phlorotannins was examined by measuring phenolic levels in our extracts before and after the addition of PVPP and by measuring the reactivity of extracts from red and green algae, which lack phlorotannins.

METHODS AND MATERIALS

Chemical Assays Phenolic levels in crude extracts of brown algae and in "purified" phlorotannin fractions (see Preparation of Pure Phlorotannin Fractions, below) were 1276 J.L. STERN ET AL. quantified by three methods. The first, the Folin-Denis procedure for measuring total phenolic content (Ragan and Jensen, 1977; Swain and Hillis, 1959), has been used extensively for measuring phlorotannins in brown algae (Ragan and Glombitza, 1986; Steinberg, 1989; Targett et al., 1992; Van Alstyne, 1995). The second, the Prussian blue assay (Price and Butler, 1977), is also well described in the literature but has not been used as extensively as the Folin- Denis procedure. Our new technique, the DMBA assay, is based on the vanillin- H2SO4 reaction (Butler, 1982; Butler et al., 1982; Putnam and Butler, 1985). Folin-Denis Assay. Folin-Denis reagent was prepared by dissolving 25 g of sodium tungstate (Na2WO4"2HeO) and 5 g of dodecamolybdophosphoric acid (12MoO3.H3PO4-H20) in 175 ml distilled water, adding 12.5 ml phosphoric acid to the solution, boiling under reflux for 2 hr, and then making up to 250.0 ml. The sample (1.0 ml) was mixed with 4.0 ml of this reagent and 8.0 ml of a saturated sodium carbonate solution and then made up to a reaction volume of 25.0 ml with distilled water. Color was allowed to develop for 5-6 hr, which allowed the precipitate that formed to settle before the absorbance was read at 725 nm. Prussian Blue Assay. Samples were comprised of 100 /xl of extract or diluted extract in 50.0 ml of distilled water in a 125 ml flask. Three milliliters of ferric ammonium sulfate (0.1 M FeNH4(SO4)2 in 0.1 M HC1) was then added to successive samples at intervals of 1.0 minute. Exactly 20 min after the addition of ferric ammonium sulfate to each sample, 3.0 ml of potassium ferricyanide [0.008 M K3Fe(CN)6] was added. Exactly 20 min after the addition of potassium ferricyanide, the absorbance at 720 nm was read. Solvent-only blanks were included and subtracted from sample absorbances. Like Folin-Denis, Prussian blue reacts with reducing compounds such as phenols, but the phenolic group is not directly incorporated into the colored product. DMBA Assay. Stock solutions of 2,4-dimethoxybenzaldehyde (2 g/100 ml solution) (Sigma D-3269) and of hydrochloric acid (16.0 ml concentrated hydrochloric acid per 100.0 ml solution) were prepared in glacial acetic acid. The working reagent, prepared by mixing equal volumes of these two solutions just prior to use, was kept at room temperature before starting the assay. The working reagent is slightly yellow when mixed, but becomes purple (h .... = 565 nm) upon standing, so it must be prepared fresh each day. Blanks of the same volume as the sample were used in all assays because some color formation occurs in the absence of phlorotannins. To ensure that the response was linear for each type of purified tannin, standard curves were prepared. Aliquots of each phlorotannin solution (up to 40/~g from 10 p.g/gl solution) were mixed with 10/zl of DMF (N,N-dimethylformamide) and 2.5 ml of the working reagent (DMF is necessary for use in protein precipitation assays as described in Stern et al. (in press). The reaction mixtures varied in volume by 4/zl (of 2.5 ml) when testing different concentra- QUANTIFYING BROWN ALGAL PHLOROTANNINS 1277 tions; this was done to minimize the volume of added methanol, at the lowest level of pipetting possible (1 tzl), within the range of sensitivity of the assay (10 p.g). The absorbance was determined at 510 nm after exactly 60 minutes at 30~ For fresh extracts of brown algae, 10-100 ~1 were used in the DMBA assay, depending on the concentration of phlorotannins. Preparation of Pure Phlorotannin Fractions. In order to assess the variation in response of different phlorotannins in the different assays, phlorotannin fractions from six species of brown algae were prepared following the methods of Ragan and Glombitza (1986) as modified by Steinberg and van Altena (1992) and van Altena and Steinberg (1992). In brief, several kilograms of fresh seaweed were extracted in 80% methanol under nitrogen at 4~ for three days with regular mixing. The filtered extract was rotary evaporated to a small aqueous volume at 40~ Any precipitates were then filtered and the filtrate was lyoph- ilized, triturated with methanol several times (to remove salt), and then adsorbed onto microcrystalline cellulose. This column was eluted with toluene (to remove pigments) and then with = 3 liters of 2 : 1 acetone-water. This was rotary evaporated, redissolved in water, and then freeze dried. These phlorotannin fractions contained a mixture of phlorotannin polymers and oligomers but were pure in the sense of containing 95 + % phlorotannins, based on the absence of peaks in ~3C NMR spectra except those in the three regions typical of phlorotannins (96- 106, 124-131, and 141-164 ppm) (van Altena and Steinberg, 1992). Phlorotannins from the kelps (Order Laminariales) Agarum cribrosum, Eck- lonia radiata, and Dictyoneurum californicum, and from the fucoids (Order Fucales) Sargassum vestitum and Carpophyllum maschalocarpum were used. High-molecular-weight phlorotannins from the fucoid Fucus vesiculosus were obtained from M. A. Ragan. Catechin (C1251; Sigma, St. Louis, Missouri) and phloroglucinol (l,3,5-trihydroxybenzene) (P3502; Sigma) were also tested in the assays. Assays of Phenolic Levels in Fresh Extracts of Brown Algae. Phenolic levels in fresh methanol extracts of a variety of brown algae also were measured. Extracts were prepared by grinding fresh algal tissue (0.5-1 g) in a Sorvall Omni Mixer in 85 % MeOH and storing at 4~ overnight prior to assaying for phenolic levels. Extracts from 11 species of Australian brown algae (Phaeophyta) from four orders were measured using the DMBA and Folin-Denis assays. These included representatives of the Order Laminariales (Ecklonia radiata), Dictyotales (Di- lophus marginatus, Dic~. opteris acrostichoides, Dictyota dichotoma, Hornoeo- strichus sinclarii, Zonaria diesingiana), Ectocarpales (Endarachne binghamiae), and Fucales (Sargassum vestitum, S. linearifolium, Phyllospora comosa, Cystophora rnoniliformis). For each species, six individual algal plants were collected in summer (February and March) in New South Wales, Australia. In New Zealand, nine species of brown algae were assayed by the Prussian 1278 J.L. STERN El" AL. blue and Folin-Denis techniques for comparison of relative sensitivity to interferences. These included Carpophyllum flexuosum, C. maschalocarpum, C. angustifolia, C. plumosum, Ecklonia radiata, Sargassum sinclarii, Xiphophora chondrophylla, Lessonia variegata, and Cystophora torulosa. Five individuals from each species were collected and analyzed in summer and autumn at the University of Auckland, Leigh Marine Laboratories, New Zealand.

Assessment of Interferences Phenolic levels in plant tissue may be overestimated if nonphenolic compounds in algal extracts react with assay reagents, e.g., ascorbic acid or peptides in the Folin-Denis assay (Ragan and Glombitza, 1986). While the DMBA assay was developed in order to avoid such interferences, this assay may not be suitable for all studies. Thus, the problem of interferences was further addressed in two ways. PVPP Treatments. Polyvinylpolypyrrolidone (PVPP) is a polymer that can remove tannins and phlorotannins from acidic aqueous solution (Loomis and Bataille, 1966; Yates and Peckol, 1993). Thus, one way of assessing the levels of interferences in an algal extract is to measure total phenolics in the extract with the Folin-Denis or Prussian blue procedure before and after treatment with PVPP (e.g., Yates and Peckol, 1993). Any color yield in a Folin-Denis assay after treatment with PVPP should be due to nonphlorotannins, such as proteins or ascorbic acid (or free phloroglucinol, which adheres poorly or not at all to PVPP), and in theory an interference-free measure of phlorotannins could then be obtained by subtraction. Extracts treated with PVPP should not yield any color in the DMBA assay, since DMBA reacts specifically with phlorotannins. We compared phenolic levels in untreated samples and PVPP-treated samples using both the DMBA and Folin-Denis assays. These experiments give us information about both the efficacy of PVPP in removing phlorotannins from solution and the amounts of nonphlorotannin interferences in our fresh algal extracts. Because both Yates and Peckol (1993) and Targett et al. (1995) note that PVPP is effective in a slightly acidic environment, the pH of the methanol extracts of algae was adjusted if necessary by the addition of 10 /zl of acetic acid to a pH of =4. Both fresh extracts and pure phlorotannins were treated with 20 mg PVPP/ml solution (Kellem et al., 1992; Yates and Peckol, 1993). The mixtures were shaken overnight at 4~ centrifuged, and the supernatants assayed for phenolic level. "Phenolic" Analyses of Red and Green Algae. A second way of assessing likely levels of interfering compounds in brown algae is to assay species of red and green algae, which lack phlorotannins (Ragan and Glombitza, 1986). Color obtained in such assays should be due only to interferences such as ascorbic acid, polypeptides, or, in the case of some red algae, simple phenols (Phillips QUANTIFYING BROWN ALGAL PHLOROTANNINS 1279 and Towers, 1982). This method assumes that the levels of these interferences, such as ascorbic acid or proteins, are not substantially different in red and green versus brown algae. In our view this assumption would be strengthened by looking at a taxonomically wide range of species. Thus we measured total phenolics in extracts of 12 species of red algae (seven from California, five from New Zealand) and five species of green algae (three from California, two from New Zealand) in the Folin-Denis and Prussian blue assays. The California algae tested were Prionotus lanceolata, Endocladia muricata (Order Crypto- nemiales), Halosaccion glandiforme (Rhodymeniales), Iridaea flaccida, Mas- tocarpus papillata (Gigartinales), Porphyra sp. (Bangiales), Rhodomela larix (Ceramiales), Chaetomorpha sp., Cladophora sp. (Cladophorales), and Ulva sp. (Ulvales); New Zealand algae tested were Vidalia colensoi (Ceramiales), Gigartina circiuta, Melanthalia abscissa (Gigartinales), Prilonia sp. (Nema- lionales), Pterocladia lucida (Gelidiales): Ulva sp. (Ulvales), and Codiumfrag- ile (Siphonales). Sample size was five individuals from each species in all cases. Extraction procedures followed those described above for brown algae. Although the chemistry of the DMBA reaction should preclude any color formation with red and green algae (unless phlorotannins are present), we assayed a selection of red and green algae (which are known to contain secondary metabolites) with DMBA. We extracted and assayed the following seaweeds: Lau- rencia rigida, Delisea pulchra, Caulerpa scalpeliformis, and Codium fragile. Four species that lack known secondary chemistry were also tested (the red algae Soliera sp., Amphiroa sp., Corallina sp., and the green alga Ulva lactuca).

RESULTS

Reaction between Phlorotannins and 2,4-Dimethoxybenzaldehyde (DMBA) Preliminary characterization of the reaction of Ecklonia radiata and Car- pophyllum maschalocarpum phlorotannins with DMBA in glacial acetic acid showed that formation of the colored adduct was both time- and temperature- dependent (Figure 1). A reaction time of 1 hr was selected since at that time absorbance was relatively stable; slight variations in reaction time did not cause large changes in absorbance. Adequate sensitivity was obtained even at room temperature, but we chose to use slightly elevated temperatures (30~ to ensure good temperature control despite seasonal fluctuations in room temperature. Higher temperatures gave increased sensitivity, but also increased problems with degradation of DMBA and with acetic acid fumes during the reaction. Based on these results, phlorotannins were routinely allowed to react with the DMBA for exactly 60 min at 30~ before determinations were made. The Xm,x for the colored product of the reaction is 510 nm so spectrophotometric measurements were made at that wavelength. 1280 J.L. STERN ET AL.

22~ C ~- 3.3C E C 0.8" ---- 46 ~ C

0.6' 0 U C W 0.4" 0 W

0.2"

0.0~ 0 50 100 150

Time (minutes)

Fie. 1. Kinetics of the reaction between 20 tzg of Carpophyttum maschatocarpum phlorotannins and 2,4-dimethoxybenzldehyde at three temperatures.

The relationship between mass of phlorotannin and color yield in the DMBA assay for six purified phlorotannin fractions and phloroglucinol is shown in Table 1. The relationship was linear to at least 40/zg, with r 2 values >0.975 for all phlorotannins and phloroglucinol. The limit of detection was approximately 5 p,g, The spectra of the reaction products of several different algal phlorotannins with DMBA were similar, indicating that the adduct formed between DMBA and various phlorotannins was the same. Phloroglucinol, the monomer unit of phlorotannins, forms a peak at 510 nm after an initial peak (within the first minute) at 565 nm, which later is masked by the peak at 510 nm. This is probably due to slight differences in the chemistry of phloroglucinol from phlorotannins; e.g., phloroglucinol lacks the aryl and ether linkages of the polymeric phlorotannins (K. Barrow, personal communication). The DMBA assay is based on chemistry similar to the vanillin assay (Ribr- reau-Gayon, 1972). In the vanillin assay, the use of acidified glacial acetic acid as a solvent increases color yield, simplifies the reaction kinetics, and stabilizes the colored product of the reaction between the aldehyde and phenolic (Butler et al., 1982). Furthermore we found that glacial acetic acid is an excellent solvent for phlorotannin/protein precipitates (Stem et al., in press). We suggest QUANTIFYING BROWN ALGAL PHLOROTANNINS 1281

o o

.3 c~ c5d~5o d ~ 0 Z =

Z ~4 ~ m LQ o.o.o.o.o.d ~5 ~.

~5 + I I I + + I I m m

[-- z oddddddd II II [I II II II II II E>=~o

Z Z b- O O ~N ~ o

~. "-:. d d d + + + + = + = = o o o

< 2 ~N II II II II II -.~ .~~

g ~o Z <

< e,-•~ r~ i~- t,~~ g~ m ~o U , , Z < c~

n~ ON < II II II II II II II II

d =

O <

N = Z e .1 ~e < ,.2

[-- 1282 J.L. STERNET AL.

the use of a fume cupboard, covered waterbath, and cuvettes with stoppers when using this solvent. It has previously been reported that small amounts of methanol inhibit the reaction between vanillin and phenolics (Butler et al., 1982). Similarly, we found that as little as 4% methanol, water, or dimethylformamide in the reaction reduced the color yield in the DMBA reaction. When using methanolic or aqueous extracts of plant material, controls must be included to account for this variation. Quantification of brown algal phlorotannins in aqueous methanol extracts (e.g., such as those used here) does not suffer significantly from this decrease in color yield as only very small volumes of extract are required for use in the assay (see Methods and Materials).

Comparisons among Assays Using "Purified" Phlorotannin Fractions and Phloroglucinol There was considerable variability in the reactivity of different purified phlorotannins within the three assays performed, as indicated in Table 1 by the variation in slopes of the regressions for absorbance versus the mass of each phlorotannin. In general, variation in the reactivity of different phlorotannins was least in the Prussian blue assay, with slopes varying between 0.0822 and 0.I 11 (including phloroglucinol). Variability in reactivity was greatest in the DMBA assay with slopes varying between 0.005 and 0.02 for the six phlorotannins tested (Table 1, excluding phloroglucinol). The reactivity of phloroglucinol (PG) with DMBA is very high, relative to phlorotannins, with the slope of the standard curve for phloroglucinol in the DMBA assay 3-11 times greater than for the phlorotannin regressions. In comparison, the slope for phloroglucinol in the Folin-Denis assay was 1-3 times higher than slopes for the phlorotannins. For a given polyphenolic (p), the ratio of color yield in a vanillin assay (V) to a Folin-Denis assay (FD), relative to phloroglucinol (PG) (i.e., (mvp/ mvPG)/(mFDp/mFDPG) where m is the slope for a given phlorotannin standard curve), may provide information about phlorotannin molecular weight (Ragan and GIombitza, 1986). This is based on the idea that polymerization of phloroglucinol is generally via ether linkages, decreasing Folin-Denis reactive sites (and thus color yield), while vanillin and DMBA color yield is a function of available positions on a benzene ring between meta-oxygen substitutions (Ribrr- eau-Gayon, 1972). Such positions would not be consumed if polymerization is via ether linkages, and therefore color yield should be strictly proportional to the number of phloroglucinol molecules present. Thus, with increasing molecular weight, vanillin should yield proportionally more color per unit mass than Folin-Denis. Color yield within an assay for a phlorotannin must be taken as relative to color yield for phloroglucinol to account for between-assay differences in extinction coefficients. QUANTIFYING BROWN ALGAL PHLOROTANNINS 1283

Because of the similarity in the chemistry of the vanillin-phlorotannin and DMBA-phlorotannin reaction, we might expect a similar trend for the ratio between DMBA absorbance and vanillin absorbance (vanillin is 3-methoxy-4- hydroxy benzaldehyde). Comparison of these ratios in Table 1 indicates a high ratio for the monomer phloroglucinol, as expected, and the next highest ratios are for the relatively small polymers from Ecklonia radiata (van Altena and Steinberg, 1992). However, the polymers from Fucus vesiculosus, which span a large size range (Ragan 1985; Ragan and Glombitza, 1986), also yield a high value, suggesting that the ratio of DMBA to FD does not consistently provide easily interpretable information about the molecular weight of phlorotannins (see Discussion, below). There was some similarity in the response of pure phlorotannins from taxonomically related algae in these assays (Table 1). For example, in the DMBA assay, slopes of the regression lines for the three kelps (Order Laminariales) varied between 0.01 and 0.02 (Table l). Phlorotannins from the confamilial species Sargassum vestitum and Carpophyllum maschalocarpum (Order Fucales, family Sargassaceae), had lower slopes: 0.0054 and 0.0064, respectively. A similar grouping was observed in the Folin-Denis assay. However, absorbance of phlorotannins from the fucoid Fucus vesiculosus (family Fucaceae) was more similar to that of the kelps in both assays.

Assays of Fresh Algae The effect of the choice of standard for the determinations of phenolic levels in crude extracts from 11 species of Australian brown algae is shown in Table 2. Phlorotannin levels were measured in the Folin-Denis and the DMBA assay based on three standards: (1) phloroglucinol, the typical standard used in such studies; (2) the overall regression for each assay (Table 1); and (3) purified phlorotannin fractions from Sargassum vestitum and Ecklonia radiata. The choice of a standard curve substantially affected the measured estimates of phenolic content for most species. For example, estimated levels of phenolics in the DMBA assay for Sargassum vestitum varied between 1.8% dry wt (phloroglucinol reference) and 24.6% dry wt (S. vestitum phlorotannin reference), depending on the reference used. Similarly, the Folin-Denis assay resulted in estimates varying from 6.1% dry wt (phloroglucinol reference) to 23.1% dry wt (phlorotannin reference) for Sargassum vestitum.

Effects of P VPP Polyvinylpolypyrrolidone (PVPP) failed to completely remove phlorotannins from all fresh algal extracts except those of Ecklonia radiata. Extracts of most algae showed significant color formation (up to 40% of original absorbance) in the DMBA assay after treatment with PVPP (Figure 2). 1284 J.L. STERN ET Am

TABLE 2. PHLOROTANNIN CONTENT (PERCENT DRY WEIGHT) OF I 1 SPECIES OF ALGAE USING THREE DIFFERENT STANDARDS FOR FOLIN-DENIS AND DMBA ASSAYS"

Standard 2, Standard 3, Standard 1, overall regression reference phlorotannin reference Algae PG reference 5: SD + SD 5: SD

Folin-Denis Er(L) 5.08 + 1.78 3.53 +_ 1.16 8.9 5:2.94 Sv (F) 6.10 5:0.75 10.5 5:1.3 23.1 + 2.8 SI (F) 1.88 5:0.45 3.41 + 0.26 not done Cm (F) 9.90 5:1.10 16.5 5:1.8 not done Pc (F) 2.08 5:0.64 3.70 + 1.05 not done Dm (D) 0.33 5:0.016 0.58 5:0_03 not done Da (D) 5.67 5:1.20 10.7 5:1.5 not done Dd (D) 1.29 5:0.09 2.15 5:0.14 not done Hs (D) 5.00 5:0.65 8.60 5:1.16 not done Zd (D) 6.79 + 0.75 11.4 + 1.2 not done Eb (E) 0.95 5:0.41 1.93 5:0.92 not done DMBA Er 1.12 5:0.45 7.7 5:3.0 6.01 5:2.31 Sv 1.80 5:0.38 11.2 5:2.0 24.6 5:2.9 SI -0.04 5:0.10 2.50 5:0.85 not done Cm 4.30 5:0.65 24.3 5:3.1 not done Pc 1.97 5:0.33 5.80 5:1.50 not done Dm 0.19 __. 0.13 1.73 5:0.11 not done Da 0.04 _+ 0.22 3.90 5:0.92 not done Dd -0.06 5:0.09 1.36 5:0.04 not done Hs* 9.45 5:1.06 40.4 5:7.5 not done Zd* 7.58 5:1.62 50.8 + 6.1 not done Eb 0.264 + 0.28 0.959 5:0.409 not done

"Er = Ecklonia radiata, Sv = Sargassum vestitum, SI = S. linearifolium, Sg = S. globulariae- folium, Cm = Cystophora moniliformis, Pc = Phyllospora comosa, Dm = Dilophus marginatus, Da = Dico,opteris acrostichoides, Dd = Dictyota dichotoma, Hs = Homoeostrichus sincalarii, Zd = Zonaria diesingiana Eb = Endarachne binghamiae; L = Laminariales, F = Fucales, E = Ectocarpales, D = Dictyotales. Data are means and standard deviations for N = 6 samples for all species except E. radiata (N = 9). Note that negative values reflect the high y intercept for phloroglucino[ relative to absorbances for phlorotannins. An asterisk indicates an alga with a high phloroglucino[ content, suggesting a phtoroglucinol reference is appropriate. Overall regression reference from Table 1.

PVPP also failed to completely remove from solution purified phlorotannins extracted from Carpophyllum maschalocarpum, Fucus vesiculosus, E. radiata, and S, vestitum (Figure 2). The difference in the efficacy of PVPP in removing E. radiata phlorotannins from fresh extracts versus "pure" fractions is unex- plained, although this may be due to qualitative differences in specific polymers present in the two samples. QUANTIFYING BROWN ALGAL PI'-ILOROTANNINS 1285

100 [] Folin Denis assay [] DMBA assay 80

t:~Q. =: Q. ._= > 60 t~Q. E,--

40 t--

Pc Hs Zd Sv Er Cm Er Cm2 Sv Fv I I I I Fresh Algal Extracts Phlorotannin Fractions

Algae Tested

FIG. 2. Percent of initial absorbance remaining after treatment of fresh extracts and purified algal phlorotannins with PVPP. If PVPP is effective at removing phlorotannins, absorbance in the DMBA assay should be zero. (Pc = Phyllospora comosa, Hs = Homoeostrichus sinclarii, Zd = Zonaria diesingiana, Er = Ecklonia radiata, Cm = Cystophora moniliformis, Cm, = Carpophyllum maschalocarpum, Sv = Sargassum vestitum, Fv = Fucus vesiculosus).

Sensitivity of Prussian Blue Assay to Interferences

If the Prussian blue assay is less sensitive to interfering compounds than the Folin-Denis assay, then the slope of the line relating estimates of phenolic levels from the two assays should be significantly less than one. For phenolic levels in nine species of New Zealand brown algae, measured in two seasons, the regression had a slope of 0.983 with 95% confidence limits of 0.825 and 1.14 (Figure 3). Thus the Prussian blue assay does not differ significantly from the Folin-Denis assay in sensitivity to interfering compounds present in algal extracts.

Phenolic Levels in Red and Green Algae

Levels of reactive substances (nonphlorotannins) measured by the Folin- Denis and Prussian blue assays for red and green algae from California were mostly <0.6% by dry wt (Folin-Denis overall mean = 0.50%, Prussian blue 1286 J.L. STERN ET AL.

20 y = 0.268 + 0.983x R^2 A ID =~

lS C m

W L,. L v 10 W .J... ~O

e- Q a. 5- /.

0 5 10 15 20

% Phenolics (Folin-Denis)

FIG. 3. Phenolic content of nine species of New Zealand algae for summer and autumn measured via the Prussian blue versus the Folin-Denis assay. Regression analysis of mean phenolic levels in each assay (N = 5) for each species is highly significant (P < 0.0001) with slope = 0.983. The slope is not significantly different from 1 (see text). Species of algae used were Carpophyllum flexuosum, C. maschalocarpum, C. angustifolia, C. plumosum, Ecklonia radiata, Sargassum sinclarii, Xiphophora chondrc ohylla, Lessonia variegata, and Cystophora torulosa.

overall mean = -0.22%). Higher levels (mean of --- 1%) were observed m one species, Rhodomela larix; however, this species contains halogenated phenolic monomers at up to 1.2-3.8% dry wt (Phillips and Towers, 1982), which should react in these assays. Five of the seven New Zealand red and green species assayed also contained relatively low levels of reactive substances. Two species, Vidalia colensoi and Melanthalia abscissa, had substantially higher levels of reactive substances in both assays (Table 3). The overall average of reactive substances from 17 red and green algal species was 0.50% (Folin-Denis) and 0.09% (Prussian blue). In contrast to the data in Figure 3, this suggests that the Prussian blue assay may be less sensitive to nonphlorotannin interferences than the Folin-Denis assay. However, the large negative values for some species in the Prussian blue QUANTIFYING BROWN ALGAL PHLOROTANNINS 1287

TABLE 3. PERCENT DRY WEIGHT OF PHENOLIC SUBSTANCES IN RED AND GREEN ALGAE FROM CALIFORNIA AND NEW ZEALAND AS MEASURED BY FOLIN-DENIS AND PRUSSIAN BLUE ASSAYS (N = 5 IN ALL CASES)

Folin-Denis Prussian blue (mean % dry weight 5: SD) (mean% dry weight + SD)

California Red algae Endocladia muricata 0.43 + 0.07 0.05 5:0.09 Halosaccion glaudiform 0.49 5:0.3 -1.9 5:0.82 lridaeaflacida 0.34 5:0.123 0.25 + 0.15 Mastocarpus papiUata 0.30 5:0.09 -0.24 + 0.17 Porphyra sp. 0.38 + 0.05 0.41 5:0.25 Prinotus lancelota 0.42 + 0.34 0.61 5:0.62 Rhodomela larix 1.07 5:0.43 0.95 + 0.23 Green algae Chaetomorpha sp. 0.7 + 0.12 -0.84 + 0.21 Cladophora sp. 0.52 -I- 0.1 -0.56 5:0.36 Ulva sp. 0.38 5:0.15 -0.94 + 0.16 California overall mean 0.50 -0.22 New Zealand Red algae Vidalia colensoi 1.33 + 0.26 1.18 5:0.14 Melanthalia abscissa 1.73 + 0.23 1.41 + 0.16 Prilonia sp. 0.67 + 0.5 0.47 + 0.47 Gigartina circumcincta -0.15 5:0.12 0.21 5:0.07 Pterocladia lucida -0.02 5:0.003 0.19 -J- 0.06 Green algae UIvasp. -0.08 + 0.15 0.31 5:0.12 Codiumfragile 0.19 + 0.15 -0.07 5:0.41 New Zealand overall mean 0.53 0.53

(which contribute to the lower overall mean) are not easily explained and cast some doubt on the utility of this assay for algae. A paired t test across all species from California and New Zealand, using species means as individual replicates, indicates that the Prussian blue values are in fact significantly less (P = 0.0396, df= 16). As predicted from the chemistry of the DMBA-phlorotannin reaction (Ribrreau-Gayon, 1972), analysis of aqueous methanolic extracts of eight species of red and green algae in the DMBA assay (five of which contain nonpolar metabolites such as terpenoids or acetogenins) yielded essentially no "phenolics" (all samples <0.05% dry wt "phenolics"). Moreover, the hma x of these extracts was often shifted higher than 510 nm (e.g., 580 nm for Laurenica 1288 J.L. STERNET AL.

rigida). This suggests that although nonpolar compounds may react in the DMBA assay, color production of such reactions will be slight, and will not interfere with estimation of phlorotannin levels.

DISCUSSION

The new DMBA assay described here has a number of advantages over previously used methods for the analysis of phlorotannins. First, because it is specific to 1,3- and 1,3,5-phenolics, it is insensitive to interferences, measuring only phlorotannins. Second, it is inexpensive and rapid. Third, the DMBA assay was specifically designed to investigate the interaction of phlorotannins with other macromolecules. Thus, this assay can be used to quantitatively measure phlorotannin precipitation by protein (Stem et al., in press), thereby linking together previously disparate methodologies. The chemical basis for the reaction between aromatic aldehydes such as vanillin (or DMBA), with activated phenolics, such as phloroglucinol, is an electrophilic attack by the aldehyde and must be carried out in a strongly acidic solution (Ribrreau-Gayon, 1972). The product of the reaction is a colored adduct and the intensity of the color is dependent in part on the structure of the aldehyde (Butler et al., 1982). Although vanillin does react with phlorotannins, we took advantage of the increased color yield obtained with DMBA. The DMBA method can also be easily adapted for determining phlorotannins in plant extracts, gut contents, or other biological samples. Preliminary work in vivo has demonstrated that staining with DMBA can indicate the pres- ence of phlorotannins in the digestive tract of herbivorous marine gastropods and echinoderms (Stem, unpublished data). One disadvantage of the DMBA assay is that the variation in reactivity of different phlorotannin fractions is somewhat greater than that for the Folin-Denis assay and Prussian blue assay (although fewer phlorotannin fractions were examined for this assay). This variation in reactivity within a given assay presumably reflects differences in the chemical structures of different phlorotannins (Ragan and Glombitza, 1986). Thus the choice of a reference compound becomes par- ticularly important for the DMBA assay, because comparing samples to different standards can result in substantial variation in the estimation of phenolic levels (Table 2). In an effort to eliminate this problem by finding a commercially available standard, we compared published data on the reactivity of vanillin from a variety of phloroglucinol structural analogs. Information about molar extinction coefficients for these potential reference compounds (e.g., catechin, epicatechin, phloretin, resorcinol, and butein) (Goldstein and Swain, 1963) suggests that they are inappropriate for the DMBA assay. In some instances, knowledge of the phenolic constituents present in a QUANTIFYING BROWN ALGAL PHLOROTANNINS 1289 particular species can also obviously assist in the appropriate choice of a reference compound. For example, H. sinclarii contains substantial amounts of phloroglucinol (determined by thin layer chromatography, unpublished data), as do many species of Zonaria (Amico et al., 1981; Blackman et al., 1988; Get- wick and Fenical, 1982) and therefore phloroglucinol would be an appropriate reference for such species. For most algal species neither phloroglucinol nor catechin were suitable as references for DMBA because DMBA responds too strongly to these substances relative to other phlorotannins and therefore significantly underestimates the mass of the phlorotannin measured. One potential disadvantage of DMBA is that, like vanillin, it can form chromophores with some nonpolar metabolites (under different conditions from this assay and absorbing at different wavelengths than 5 l0 nm) that may interfere with quantitative determinations of phlorotannins in algal extracts (R. de Nys, personal communication). However, all five red and green algae tested with DMBA known to contain nonpolar secondary metabolites yielded negligible values (<0.05 % dry wt) and thus such metabolites are unlikely to pose a problem. Like DMBA, the Prussian blue and Folin-Denis tests respond differently to different phlorotannins, but with less marked variation. This differential variation is due to differences in chromophore formation. In the Folin-Denis assay oxidation of phenols is coupled with the reduction of pbosphomolybdic and phosphotungstic acids to stable chromophores (Ragan and Glombitza, 1986). Thus, although the reaction does not incorporate the phenols into the colored product, it is stoichiometrically dependent on the number of free hydroxyl groups (Ragan and Glombitza, 1986). Aldehydes such as vanillin and DMBA react with the aromatic compound at positions between meta-substituted hydroxyl groups to yield a colored product (Ribrreau-Gayon, 1972). Ragan and Glombitza (1986) stated that the ratio of color formation between vanillin (and, thus, by implication the vanillin analog DMBA) and Folin-Denis, relative to phloroglucinol, is indicative of phlorotannin molecular weight, with higher ratios indicating lower molecular weights. Our data do not support this for the DMBA assay, as the highest values for this ratio (other than for the monomer phloroglucinol; Table l) were observed for the low-molecular-weight polymers (MW = l0 g) from Ecklonia radiata (van Altena and Steinberg, 1992) and the high molecular weight polymers from Fucus vesiculosus. More generally, the usefulness of obtaining molecular weight information by comparing vanillin- or DMBA-type reactions with Folin-Denis may be limited since: (1) vanillin and DMBA may react preferentially with terminal groups on the mol- ecule, as is the case for terrestrial tannins (Butler et al., 1982); (2) variation in ether and aryl linkages alter reactivity in an unpredictable manner if the structure is unknown; (3) phlorotannins extracted from seaweeds probably span a range of molecular weights; and (4) chromophore production in Folin-Denis is due to 1290 J.L. STERN ET AL. both the number of oxidizable groups as well as their redox potential (which depends on the surrounding molecular environment). This lack of a consistent relationship between DMBA-FD ratios and molecular weight is unfortunate because molecular weight may explain some of the variation in responses of marine herbivores to phlorotannins. In particular, Boettcher and Targett (1992) showed that variation in the molecular weight of different phlorotannin fractions from Lobophora variegata and Ascophyllum nodosum significantly affected assimilation efficiencies in the marine herbivorous fish Xiphister mucosa, with high-molecular-weight phlorotannin fractions the most inhibitory. In contrast to these results, Steinberg and van Altena (1992; their Figure 7a) and van Altena and Steinberg (1992) found no clear evidence for a relationship between feeding deterrence and phlorotannin molecular weight. Not surprisingly, the importance of molecular weight in the phlorotannin- herbivore interaction is likely to depend on the herbivore and phlorotannins in question and the kind of response examined (e.g., feeding deterrence versus physiological measures, etc.). The data presented here for the DMBA assay indicate that it is the method of choice for studies of variation in phlorotannin levels in a single species or a group of closely related species. In such a case a purified phlorotannin fraction should be extracted from the algae and used as a standard. This will eliminate difficulties due to differences between the reactivity of the standard and that of the samples. The appropriate method for broad comparisons among many species from different taxa (Steinberg 1989, Targett et al., 1992) is less clear. The problems associated with using different reference compounds were readily evident for the three colorimetric assays examined in this study. Because a single appropriate reference compound is not readily available for the DMBA assay we recommend that an overall regression consisting of many phlorotannin standard curves may be used as a reference (Tables 1, 2). This seems to minimize the error in estimating phenolic levels across many species and taxa (especially when compared to using phloroglucinol as a standard). The major advantage of the DMBA assay is that it does not measure nonphenolic interferences, as do the Folin-Denis or Prussian blue assays. Based on our analyses of red and green algae, and assuming that levels of nonphenolic interferences are similar in red, green, and brown algae, levels of interferences are likely to account for 0-1.5 % by dry weight (mean = 0.5 %) of the phenolic levels measured using these assays. This is somewhat higher than the estimate of nonphenolic materials in Fucus gardneri measured by Van Alstyne (1995) using the Folin-Ciocalteau assay, but is still probably an acceptable level of error for assays of phenolic-rich algae, which can have levels of phenolics exceeding 10% by dry weight. Estimates of phenolic levels in more phenolic- poor species may be significantly affected by such levels of interferences, how- QUANTIFYING BROWN ALGAL PHLOROTANNINS 1291 ever. Although levels of reactive nonphenolic substances in F. gardneri (Van Alstyne, 1995) were low, the Folin-Ciocalteu assay does not resolve the problem of interferences for brown algae in general. It does have an advantage over the Folin-Denis assay in that it does not suffer from precipitate formation and thus can be used with small sample volumes. The DMBA assay also only uses small volumes of samples and reagents. Measurement of phenolic levels using the Folin-Denis assay before and after mixing the extract with PVPP does not solve the problem of interferences. Although PVPP has been used in Folin assays to correct estimates of phenolic levels for interferences (e.g., Yates and Peckol, 1993), we found that PVPP failed to remove all phlorotannins from solution, and thus this method can significantly underestimate the levels of phlorotannins in an extract. Van Alstyne (1995) also questioned the utility of PVPP in these assays, although Targett et al. (1995) demonstrated reliable removal of phlorotannins after determining optimal conditions. However, optimal conditions are likely to vary for each phlorotannin-PVPP interaction. Our data reflect on the accuracy of previously reported algal phenolic levels in one final respect. That is, using phloroglucinol as a reference compound in the Folin-Denis or DMBA assays significantly underestimated the phlorotannin content of most species assayed, relative to using an overall regression standard or purified phlorotannin fractions (Table 1). These differences undoubtedly reflect differences in the chemistry between phloroglucinol and phlorotannins and emphasize the need to continue to improve techniques for specifically quantifying phlorotannins.

Acknowledgments--The research reported here was supported in part by NSF grant OCE- 9000264 to J. Estes and P. Steinberg and by a University of New South Wales SRG grant to P. Steinberg. We are grateful to Dr. R. Creese and the rest of the staff at The University of Auckland's Leigh Marine Laboratory for their hospitality.

REFERENCES

AMICO, V., CURRENTI, R., ORIENTE, G., PIATELLI, M., and TRINGALI, C. 1981. A phloroglucinol derivative lu the brown alga Zonaria tournefortii. Phylochemistry 20:1451-1453. ARNOLD, T., TANNER, C., and HATCH, W. 1995. Phenotypic variation in polyphenol content of the tropical brown alga [xJbophra variegata as a function of nitrogen. Mar. Ecol. Prog. Ser. 123: 177-183. BERESS, A., WASSERMANN, O., BRUHN, T., BERESS, L., KRAISELBURD, E. N., GONZALEZ, L. V., DEMO'rrA, G. E., and CHAVEZ, P. 1. 1993. A new procedure for the isolation o1: antiHIV compounds (polysaccharides and polyphenols) from the marine alga Fucus vesicuh~sus_ J. Nat. Prod. Lloydia 56:478-488. BERNAYS, E. A., COOPER DRIVER, G., and BILGENEr, M. 1989. Herbivores and plant tannins, Adv. Ecol. Res. 19:263-302. 1292 J.L. STERN ET AL.

BLACKMAN, A. J., ROGERS, G. 1., and VOLKMAN, J. K. 1988. Phloroglucinol derivatives from three Australian marine algae of the genus Zonaria. J. Nat. Prod. 51:158-160. BOETTCHER, A. A., and TARGETT, N. M. 1992. Role of polyphenolic molecular size in reduction of assimilation efficiency in Xiphister mucosa. Ecology 74:891-903. BUTLER, L. G. 1982. Relative degree of polymerization of sorghum tannin during seed development and maturation. J. Agric. Food Chem. 30: 1090-1094. BUTLER, L. G., PRICE, M. L., and BROTHERTON, J. E. 1982. Vanillin assay for proanthocyanidins (condensed tannins): Modification of the solvent for estimation of the degree of polymerization. J. Agric. Food Chem. 30:1087-1089. CARLSON, D. J., and MAYER, L. M. 1983. Relative influences of riverine and macroalgal phenolic materials on UV absorbance in temperate coastal waters. Can. J. Fish. Aquat. 40:1258-1263. COLEY, P. D., BRYANT, J. P., and CHAPm, F. S., I11. 1985. Resource availability and plant antiherbivore defense. Science 230:895-899. FEENY, P. P. 1976. Plant apparency and chemical defense, pp. 1-40, in J. W. Wallace and R. L. Mansell (eds.). Biochemical Interactions between Plants and Insects. Plenum Press, New York. GERWlCK, W., and FENICAL, W. 1982. Phenolic lipids from related marine algae of the order Dictyotales. Phytochemistry 21:633-637. GOLDSTEm, J. L., and SWAIN, T. 1963. Methods for detemaining the degree of polymerization of flavans. Nature 198:587-588, HARBORNE, J. B. 1991. Plant defenses against mammalian herbivory, pp. 45-60, in P. T. Palo and C. T. Robbins (eds.). The Chemical Basis of Plant Defense. CRC Press, Boca Raton, Florida. HASLAM, E. 1989, Plant Polyphenols. Vegetable tannins Revisited. Cambridge University Press. Cambridge. HAY, M. E., and FENICAL, W. 1988. Marine plant-herbivore interactions: the ecology of chemical defense. Annu. Rev. Ecol. Syst. 19:111-145. KELLEM, S. J., TISCH, M. H., and WALKER, J. R. L. 1992. Screening of New Zealand native plants for enzyme inhibitory activities. N.Z.J. Bot. 30: 199-203. LOOMIS, W. D., and BATAiLLE, J. 1966. Plant polyphenolic compounds and the isolation of plant enzymes. Phytochemistr3' 5:423-438. MCMILLAN, C. 1984. The condensed tannins (proanthocyanidins) in sea grasses. Aquat. Bot. 8:351- 357. MOLE, S., and WATf-RMAN, P. G. 1987a. A critical analysis of techniques measuring tannins in ecological studies I. Techniques for chemically defining tannins. Oecologia 72: 137-147. MOLE, S., and WATERMAN, P. G. 1987b. A critical analysis of techniques for measuring tannins in ecological studies 11. Techniques for biochemically defining tannins. Oecologia 72:148- 156. PHmUPS, D. W., and TOWERS, G. H. N. 1982. Chemical ecology of red algal bromophenols: 1. Temporal, interpopulational and within-thallus measurements of lanosol levels in Rhodomela larix (Turner) C. Agardh. J. E-~p. Mar. Biol. Ecol. 58:285-293. PRICE, M. P., and BUTLER, L. G. 1977. Rapid visual estimation and spectrophotometric determi- nation of tannin content of Sorghum groin. J. Agric. Food Chem. 25:1268-1273. PUTNAM, L. J., and BUTLER, L. G. 1985. Fractionation of condensed tannins by counter-current chromatography. J. Chromatogr. 318:85-93. RAGAN, M. A. 1985. The high molecular weight polyphloroglucinols of the marine brown alga Fucus vesiculosus L: Degradative analysis. Can. J. Chem. 63:294-303. RAGAN, M. A., and GLOMmTZA, K.-W. 1986. Phlorotannins, brown algal polyphenols. Prog. Phycol. Res. 4: 129-241. RAGAN, M. A., and JENSEN, A. 1977. Quantitative studies on brown algal phenolics. 1. Estimation of absolute polyphenol content in Ascophyllum nodosum and Fucus vesiculosus. J. Exp. Mar. Biol. Ecol. 34:245-258. QUANTIFYING BROWN ALGAL PHLOROTANNINS 1293

RIBEREAU-GAYON, P. 1972. Plant Phenolics. Oliver & Boyd, Edinburgh. pp. 23-53. SCHLESINGER, W. H. 1991. Biogeochemistry: An Analysis of Global Change. Academic Press, San Diego. STEINBERG, P. D. 1989. Biogeographical variation in brown algal polyphenolics and other secondary metabolitcs: comparison between temperate Australasia and North America. Oecologio 78:374- 383. STEINBERG, P. D. 1992. Geographical variation in the interaction between marine herbivores and brown algal secondary metabotites, pp. 51-72, in V. J. Paul (ed.). Ecoh~gical Roles for Marine Secondary Metabolites. Comell University Press, New York. STEINBERG, P. D., and VAN ALTENA, I. 1992. Tolerance of marine invertebrate herbivores to brown algal phlorotannins in temperate Australasia. Ecol. Monogr. 62:189 222. STERN, J. t., HAGERMAN, A. E., STEINBERG, P. D., and MASON, P. K. 1995. Phlorotannin-protein interactions. In press. SWAnN, T. 1979. Tannins, pp. 657-682, in G. A. Rosenthal and D. H. Janzen (eds.). Herbivores: Their Interaction with Secondary metabolites, Academic Press, New York. SWAIN, T., and HILLIS, W. E. 1959. The phenolic constituents of Prumts &mtestica. I. The quantitative analysis of phenolic constituents. J. Sci. Food. Agric. 10:63-68. TARGE'Fr, N. M., BOETTCHER, A. A., TARGEq-F, T. E., and VROI.IJK, N. H. 1995. Tropical marine herbivore assimilation of phenolic-rich plants. Oecologia 103:170-179. TARC;E1T', N. M., COEN, L. D., BOr-:TrCHER, A. A., and TANNER. C. E. 1992. Biogeographic comparisons of marine algal polyphenolics: evidence against a latitudinal trend. Oecologia 89:464-470. VAN ALSTYNE, K. L. 1995. Comparison of three methods Ior quantifying brown algal polyphenolic compounds. J. Chem. Ecol. 21:45-58. VAN ALTENA, I. A., and S'rEINBERG, P. D. 1992. Arc differences in the responses between North American and Australasian marine herbivores to phlorotannins due to differences in phlorotannin structure'? Biochem. Syst. Ecol. 20:493-499. YA'rES, J. L., and PECKOL, P. 1993. Effect of nutrient availability and herbivory on polyphenolics in the seaweed Fucus vesiculosis. Ecology 74: 1757-1766.