Yeast Interactions with Anthocyanins during Red Wine Fermentation

Transcription

1 104 Medina et al. Yeast Interactions with Anthocyanins during Red Wine Fermentation Karina Medina, 1 Eduardo Boido, 1 Eduardo Dellacassa, 1,2 and Francisco Carrau 1 * Abstract: The capacity of yeast to modify anthocyanin concentration during red wine fermentation was studied with the objective of classifying yeast strains according to their capacity for color removal during vinification. A model red grape juice medium was used to better understand yeast-anthocyanin interactions. An anthocyanin extract of Vitis vinifera cv. Tannat was prepared from grape skins and diluted in a white must in order to obtain a red grape juice without solids. Anthocyanin removal was confirmed to be higher for compounds with higher polarity. Acylated anthocyanins (acetyl and p-coumaryl compounds) were differentially removed, and the percentage of removal for each compound was determined. Results showed no correlation between color intensity and total anthocyanin concentration after fermentation, possibly because of the presence of anthocyanin derivatives formed during the process. HPLC-MS analysis allowed identification of anthocyanin derivatives, while the quantification of several derivatives was performed by HPLC-DAD. The sum of these derivatives showed a direct correlation with the color intensity obtained with each strain, thus explaining the color variability observed. Key words: yeast, anthocyanin, red grape must medium, anthocyanin-derived pigments Anthocyanins are phenolic plant metabolites belonging to the flavonoid family, which are responsible for most red and blue colors in fruits, berries, and flowers. Differences among the aglycones are a consequence of the number of hydroxyl groups present in the molecule and whether the hydroxyls are methylated. Anthocyanin composition is an important quality parameter for red grapes because of the significance of these compounds in determining color of the resulting wines. Anthocyanin profiles of grapes, determined by the relative proportions of the different anthocyanins, are characteristic for each grape variety. Moreover, concentrations of the different compounds can vary significantly within grape cultivars according to environmental conditions, including climate, soil, and vineyard management practices (Boulton et al. 1996). Wine color depends on grape pigments and how those pigments are modified during vinification, storage, and aging (Ribereau-Gayon 1982, Boulton 2001, Vivar-Quintana et al. 2002, Schwarz et al. 2003). More recently it has been 1 Sección Enología, Cátedra Ciencia y Tecnología de los Alimentos, 2 Cátedra de Farmacognosia y Productos Naturales, Facultad de Química, Universidad de la República, Montevideo, Uruguay. *Corresponding author [ fcarrau@fq.edu.uy] Acknowledgments: We would like to thank the financial support of the Comisión Sectorial de Investigación Científica (grant to K. Medina), Instituto Nacional de Investigaciones Agropecuarias (INIA, project INIA- LIA 055), and the Programa de Desarrollo Tecnológico (PDT, project S/ C/OP/07/02). We also thank the critical comments of Prof. Patrick Moyna on this work. Manuscript submitted October 2004; revised December 2004 Copyright 2005 by the American Society for Enology and Viticulture. All rights reserved. demonstrated that, during fermentation, yeast releases secondary metabolic products into the medium, such as pyruvic acid and acetaldehyde, some of which react with anthocyanins to produce derivatives such as vitisin A, vitisin B, and ethyl-linked anthocyanin-flavanol pigments (Morata et al. 2003b, Asenstorfer et al. 2003, Lee et al. 2004, Eglinton et al. 2004). Wine color stability during aging can also be affected by anthocyanin interactions with other polyphenolic compounds, proteins, and polysaccharides (Glories 1984, Escot et al. 2001). Because of the importance of color in red wines and the amount of pigment removed by yeast, in the present work we report the effect of different Saccharomyces cerevisiae strains on anthocyanins extracted from Tannat wines. The presence of grape skins and other solids during red wine fermentation makes it difficult to study yeast-anthocyanin interactions during winemaking. Previous studies reported complications because of the presence of solids, the natural flora present in grapes, and significant variations in the maceration technology (Cuinier 1997, Morata et al. 2003a, Caridi et al. 2004). A model red grape juice (RGJ medium) was developed using an anthocyanin extract from grape skins of Vitis vinifera cv. Tannat. This cultivar, widely grown in Uruguay, is one of the richest varieties in polyphenolic compounds (Carrau et al. 2001, Gonzalez Neves et al. 2001). Its name is derived from its unusually high tannin content compared with other red grapes. The RGJ medium allowed us to standardize fermentation conditions and achieve reproducible yeast-anthocyanin interactions during vinification. This strategy also reduced anthocyanin losses from other causes, such as skin adsorption (Ribereau- 104

2 Yeast Interactions with Anthocyanins during Fermentation 105 Gayon et al. 2002). The method represents an improved approach for specifically studying the effect of yeast-anthocyanin interactions. Materials and Methods Yeast strains. The yeast strains examined were isolated from fermentations conducted in different Uruguayan wineries. They were identified as Saccharomyces cerevisiae (strains CP881, CP882, CP863, CP874, CP873, and KU1) through DNA microsatellite markers (González Techera et al. 2001). Two commercial yeasts were used as controls: Montrachet 522 (University of California, Davis) and Lalvin 254 D (Lallemand, Montreal, Canada). Yeasts were maintained on YEPD medium slants (1% yeast extract, 2% peptone, 2% glucose, 2% agar, containing 0.1 M citrate-phosphate buffer, ph 4.5) and incubated for 12 hr at 20 C on fresh medium before use in fermentation experiments. Red grape juice medium. A simulated red grape juice (RGJ medium) was designed for fermentations. It consisted of a Chardonnay white must (ph 3.6; total reducing sugars, 200 g/l; total acidity as H 2 SO 4, 3.25 g/l; free amino nitrogen, 213 mg/l) chemically analyzed according to Zoecklein et al. (1995) and then sterilized by filtration through a 0.45 µm cellulose acetate membrane. This white grape must was stored at -20 C until used. The must was then supplemented with an anthocyanin extract prepared as follows. Tannat grape skins were macerated for 72 hr with 5 g/l potassium metabisulfite under agitation. The liquid was separated by filtration, supplemented with 2 g/ L tartaric acid, and evaporated under vacuum at 40 C to one-third of the initial volume. The extract was then lyophilized and stored under nitrogen at -20 C until used. The anthocyanin extract (2 g) was dissolved in one liter of white grape must and then sterilized by filtration (0.2 µm). Total anthocyanin concentration in the final RGJ medium was ~750 mg/l. Fermentation conditions. Fermentations were conducted using sterilized 125 ml Erlenmeyer flasks filled to 50 ml with RGJ medium and stopped with Müller valves containing concentrated sulfuric acid. Duplicate flasks were inoculated with 10 5 cfu/ml (Carrau 2003) and incubated at 20 C without shaking. Carbon dioxide was measured by weight loss and the number of dead cells determined by the methylene blue-staining technique. Total cell population was determined in an improved Neubauer chamber. Color intensity and hue. Wines were diluted 1:10 (v/v) in a ph 3.2 buffer (0.05 M Na 2 HPO 4, 0.75 M citric acid) in order to avoid effects because of ph differences and copigmentation (Boulton 2001). Absorbance at 420 nm, 520 nm, and 620 nm was determined using a Spectronic Genesys 2 spectrophotometer (Spectronic Instruments, Rochester, NY), with a 1 cm path-length quartz cell. Hue (H = A420/A520) and color intensity (IC = A420 + A520 + A620) values were calculated from these measures. Quantitative analysis of total anthocyanins. Upon removal of yeast cells by centrifugation and filtration, total anthocyanin concentration in each wine was measured spectrophotometrically at 540 nm, after dilution in waterethanol-hydrochloric acid (70:30:1), according Di Stefano et al. (1989). Analysis of anthocyanins by HPLC-DAD. Anthocyanins were analyzed by high-performance liquid chromatography (HPLC) using a Hewlett-Packard Series 1050 (Palo Alto, CA) equipped with an HP Chem Station and photodiode-array detector (DAD). Gradients of solvent A (waterformic acid, 95:5, v/v) and solvent B (acetonitrile) were applied over a C 18 reversed-phase 150 mm x 4.6 mm column (Phenomenex, Torrance, CA) as follows: 10 to 30% B linear (1.0 ml/min) from 0 to 40 min, then 30 to 35% B (1.0 ml/ min) from 40 to 48 min, and 35 to 10% B (1.0 ml/min) from 48 to 49 min, then 10% from 49 to 55 min to reequilibrate the column to initial conditions (Rivas-Gonzalo et al. 1995). Anthocyanins were detected by scanning from 280 to 600 nm. Quantification was performed against an external standard at 530 nm and expressed as a function of malvidin-3- glucoside concentration. Samples (20 µl) of previously filtered wines (200 µl) were injected into the HPLC. Determinations were made in duplicate. The following anthocyanins were identified in all wines: delphinidin, cyanidin, petunidin, peonidin, and malvidin, as well as their 3-glucosides, 6-acetylglucosides, 6-p-coumarylglucosides, and 6- caffeoylglucosides. Identification of anthocyanin-derived pigments by HPLC-MS. HPLC-mass spectrometry (MS) analyses were performed using a Finnigan LCQ MS detector (Thermoquest, San Jose, CA) equipped with an atmospheric pressure ionization source, using electrospray ionization (ESI) interface. Both the sheath gas and the auxiliary gas were a mixture of nitrogen and helium. Sheath gas flow was 1.2 L/ min and auxiliary gas flow was 6 L/min. Capillary voltage was 4V and capillary temperature was 195 C. Spectra were recorded in positive ion mode between m/z 120 and The mass spectrometer was programmed for a series of three consecutive scans: a full mass, a MS 2 scan of the most abundant ion in the full mass, and a MS 3 of the most abundant ion in the MS 2. The normalized energy of collision was 45%. The anthocyanin-derived pigments were identified by similar M + and fragmentation patterns (Alcalde-Eon et al. 2004). Statistical analysis. Significant differences among replicates and treatments (control and each yeast strain) were assessed using a two-way ANOVA (2 replicates x 9 treatments) for the anthocyanin compounds. Statistical analysis was performed using GenStat software (VSN, Herts, UK). Results Anthocyanin profile of Tannat skin extracts. The yield of anthocyanins extracted from Tannat grape skins was 2.0 g/kg. Similar HPLC anthocyanin profiles were

3 106 Medina et al. obtained for both the RGJ medium and Tannat grapes (Figure 1). Commercial preparations of grape anthocyanins showed significant differences from these profiles (data not shown). Yeast growth and viability. All fermentations were finished at day 7. Similar final total populations of ~10 7 cells/ ml and 2.6% dead cells were obtained with no significant differences between strains. Yeast effect on total anthocyanin and color parameters of wines. Total anthocyanin concentration immediately after alcoholic fermentation for the wines obtained with different yeast strains is shown in Figure 2. An average decrease of 17% in anthocyanin concentration was found for all yeast strains studied when compared with unfermented RGJ medium, with no significant differences among strains (main difference 16 mg/l, 3.3%). However, color intensity was affected by yeast strain, with a significant decrease of color between some fermentations and the control (Figure 3), but without significant differences for alcohol content and final ph (data not shown). The smallest decrease in color intensity, compared to the unfermented control, was obtained for strain CP882 (1.5% decrease) while the largest loss was found for strain M522 (14% decrease). The values for hue showed no significant differences among treatments in our experimental conditions (Figure 3). Effect of yeasts on anthocyanins. ANOVA for anthocyanin values obtained by HPLC-DAD analyses, for the different yeast fermentations, showed significant differences (p < 0.05) among treatments for peonidin, delphinidin, petunidin, malvidin-3-caffeoylglucoside, malvidin-3- (p-coumaroyl)glucoside, and petunidin-3-(p-coumaroyl) glucoside. The concentration of petunidin-3-(p-coumaroyl) glucoside increased for all fermentations. All strains tested showed higher percentages of anthocyanin removal with increasing anthocyanin polarity (Figure 4). Average anthocyanin losses were delphinidin, 45 to 50%; petunidin, 22 to 30%; peonidin, 7 to 27%; malvidin, 10 to 15%; and acylated anthocyanins, 9 to 16%. In this last case, percentage losses for individual components were malvidin-3-acetylglucoside + delphinidin-3-(p-coumaroyl) glucoside, 9.0%; delphinidin-3-acetylglucoside, 2.6%; peonidin-3-(p-coumaroyl)glucoside, 1.9%; petunidin-3- acetylglucoside, 1.2%; malvidin-3-(p-coumaroyl)glucoside, 0.9%; and petunidin-3-acetylglucoside, 0.4%. Cyanidin-3- glucoside concentration was very low in all treatments. Anthocyanin derivatives produced during yeast fermentation. The relationship between total anthocyanin CI Hue *10 Figure 1 Anthocyanin composition of RGJ medium and Tannat grapes. D: delphinidin; C: cyanidin; Pt: petunidin; Pn: peonidin; M: malvidin; Acy: acylated derivatives (acetyl, caffeoyl, and p-coumaryl compounds). Figure 3 Effects of yeast strain on color intensity and hue in wines obtained from RGJ medium fermented by different yeast strains. Control: unfermented RGJ medium. Error bars represent ± standard deviation of the mean (n = 2). Hue *10: for clarity hue values are presented using a factor of 10. Significant differences (p < 0.05) according to LSD are indicated by different letters. Figure 2 Total anthocyanin concentration in wines obtained from RGJ medium fermented by different yeast strains. (Control: unfermented RGJ medium.) Figure 4 Average percentage of variuos anthocyanins removed during fermentations using different yeast strains. Error bars represent ± standard deviation of the mean (n = 2). Acylated derivatives: acetyl and p-coumaryl compounds.

4 Yeast Interactions with Anthocyanins during Fermentation 107 concentration and color intensity values was paradoxical. While all fermentations lost the same quantities of total anthocyanins, some maintained or increased color intensity. This paradox might be explained by the presence of anthocyanin derivatives or the copigmentation effect. Under our experimental conditions, copigmentation was avoided (see Material and Methods), so the presence of anthocyanin derivatives is the most likely explanation. Anthocyanin derivatives were identified by HPLC-MS analysis of the wine samples for all the yeast strains studied (Table 1, Figure 5). Five derivatives were also quantified by HPLC-DAD: petunidin-3-glucoside-8-ethylcatechin (isomers), malvidin-3-glucoside-8-ethylcatechin, malvidin-3- glucoside-4-vinylcatechol, malvidin-3-(p-coumaroyl)glucoside-8-ethylcatechin, and malvidin-3-glucoside-4-vinylguaiacol. A correlation analysis between the concentration of total pigments and color intensity for the different strains was performed to explore the diverse behavior of the strains in relation to color (Figure 6). The correlation obtained was significant at r 2 = 0.92, which suggests an important role for anthocyanin derivatives in contributing to color, and that the formation of these pigments may differ significantly for each yeast strain. Discussion Yeast-anthocyanin interactions have previously been studied with whole grape maceration during fermentation (Cuinier 1997, Morata et al. 2003a, Caridi et al. 2004), must pasteurization (Loiseau et al. 1994, Conterno et al. 1997), or use of yeast lees (Vasserot et al. 1997). However, the presence of solids in the medium and/or the instability of anthocyanins during pasteurization add complexity to the understanding of yeast-anthocyanin interactions during fermentation. The RGJ medium provided a simplified system in which to evaluate those interactions. Moreover, RGJ medium showed a similar anthocyanin profile to Tannat grapes (Figure 1). Eglinton et al. (2004) recently reported a chemically defined medium to which they added purified anthocyanins from a commercial grape skin extract. Although this approach has the advantage of a chemically defined medium, which allows for more repeatable results, the RGJ medium described here may be a better simulation Table 1 Retention time, M +, MS 2, and MS 3 and fragment sizes (amu) found during analysis of five anthocyanin-derived pigments. Retention M + MS 2 MS 3 Compound (min) (amu) (amu) (amu) 1: Petunidin-3-glucoside , 505, 633, 481, , 343, 205 ethylcatechin (isomers) 2: Malvidin-3-glucoside , 519, 647, , 341, 296, 327 ethylcatechin 3: Malvidin-3-glucoside , 463, 402, 430 vinylcatechol 4: Malvidin-3-(p-coumaroyl) , 665, 647, , 341, 325, 313 glucoside-8-ethylcatechin 5: Malvidin-3-glucoside , 461, 445, 416, 309 vinylguaiacol 1: R 1 = OH, R 2 = glucose 2: R 1 = OCH 3, R 2 = glucose 4: R 1 = OCH 3, R 2 = p-coumaroylglucose 3: R 3 = OH 5: R 3 = OCH 3 Figure 5 Chemical structures of the anthocyanin-derived pigments quantified by HPLC-DAD. Numbers in bold correspond to compounds listed in Table 1. Figure 6 Correlation between color intensity and the sum of anthocyanin-derived pigments expressed as malvidin-3-glucoside equivalents (r 2 = 0.92).

5 108 Medina et al. of an industrial fermentation. Further research using anthocyanin extracts from specific cultivars in a chemically defined medium would be required to compare yeast anthocyanin interactions in different varieties. From the data presented in Figures 2, 3, and 4, it can be concluded that total anthocyanin content was similarly affected, according to the polarity of the different compounds, during fermentation by all yeast strains studied. However, color intensity was differently affected according to the yeast strain considered. The relationship between the polarity of anthocyanins and their adsorption by yeasts was in agreement with a previous report (Vasserot et al. 1997), where delphinidin and petunidin were the most adsorbed. This loss of polar anthocyanins could be explained by the affinity of these compounds to the cell wall (Fleet 1991, Cuinier 1997, Vasserot et al. 1997). This emphasized the importance of this interaction in red wine production from varieties with high concentrations of delphinidin and petunidin, such as Tannat (Gonzalez Neves et al. 2001). Our finding that acylated compounds were the compounds least removed by yeasts contrasts with the results of Morata et al. (2003a). They reported that the acyl derivatives (acetyl and p-coumaryl) were most strongly adsorbed in fermentations with grape maceration using Saccharomyces strains from a Spanish collection. The absence of solids in our experimental model may allow the more accurate analysis of this effect. Further work would be required to investigate the importance of solids in the context of the present study. In this work, we demonstrated that the anthocyaninderived pigments identified were formed during fermentation. The presence of some of these compounds was recently reported in wines (Asenstorfer et al. 2003, Morata et al. 2003b, Alcalde-Eon et al. 2004, Lee et al. 2004). Interestingly, the color intensity of these pigments at the acidic ph of wine is greater than that of the anthocyanins, which exist mainly as colorless hemiacetal structures in wine (Bakker and Timberlake 1997). This ph effect could also explain the direct correlation found between anthocyanin derivative concentration and color intensity. We also found that the presence of anthocyanin-derived pigments depended on the yeast strain used and resulted in a significant contribution to the preservation of wine color. Finally, these novel results challenge anthocyanin adsorption by yeasts as one of the main factors affecting wine color and demonstrate that yeast contribute to maintaining or improving wine color through anthocyanin-derived pigment formation. Further studies are required to understand how yeast metabolites interact with the formation of these compounds and how enological practices could be altered to achieve higher concentrations of important anthocyaninderived pigments. Conclusions Yeast strain selection was demonstrated to be important for color preservation during vinification. Differences in loss of total anthocyanin compounds, as a consequence of their removal by yeast, were not significant for the different strains studied with the RGJ medium. However, color indices showed significant differences among strains. Differences among yeast strains in production of anthocyanin-derived pigments that affect color index were also demonstrated. The production of these stable pigments may be a significant contribution by yeasts to red wine color, potentially allowing longer aging. The yeast-mediated changes reported here for wine pigment composition in a Tannat-based model medium will enhance future studies concerning yeast-color interaction for other important red varieties. Literature Cited Alcalde-Eon, C., M.T. Escribano-Bailón, C. Santos-Buelga, and J.C. Rivas-Gonzalo Separation of pyranoanthocyanins from red wine by column chromatography. Anal. Chim. Acta 513: Asenstorfer, R., A. Markides, P. Iland, and G. Jones Formation of vitisin A during red wine fermentation and maturation. Aust. J. Grape Wine Res. 9: Bakker, J., and C.F. Timberlake Isolation, identification and characterization of new color-stable anthocyanins occurring in some red wines. J. Agric. Food Chem. 45: Boulton, R The copigmentation of anthocyanins and its role in the color of red wine: A critical review. Am. J. Enol. Vitic. 52: Boulton, R., V. Singleton, L. Bisson, and R. Kunkee Principles and Practices of Winemaking. Chapman & Hall, New York. Caridi, A., A. Cufari, R. Lovino, R. Palumbo, and I. Tedesco Influence of yeast on polyphenol composition of wine. Food Technol. Biotechnol. 42: Carrau, F Characterization of yeast in relation to the ability to utilize nitrogen. Studies of aroma compound. Ph.D. thesis, Universidad de la República, Uruguay. Carrau, F., E. Boido, E. Dellacassa, A. Lloret, K. Medina, and G. Versini Vitis vinifera Tannat produces the typical red wine of Uruguay. Abstr. American Society for Enology and Viticulture 52 nd Annual Meeting Technical Abstracts, p. 49. ASEV, Davis, CA. Conterno, L., C. Tortia, J. Minati, and G. Trioli La selezione di un ceppo di lievito per il Barolo. Vigne Vini 24: Cuinier, C Ceppi di lievito e composizione fenolica dei vini rossi. Vigne Vini 7/8: Di Stefano, R., M.C. Cravero, and N. Gentilini Metodi per lo studio dei polifenoli dei vini. L Enotecnico, pp Eglinton, J., M. Griesser, P. Henschke, M. Kwiatkowski, M. Parker, and M. Herderich Yeast-mediated formation of pigmented polymers in red wine. In Red Wine Color: Exploring the Mysteries. A.L. Waterhouse and J.A. Kennedy (Eds.), pp American Chemical Society, Washington, DC.

6 Yeast Interactions with Anthocyanins during Fermentation 109 Escot, S., M. Feuillat, L. Dulau, and C. Charpentier Release of polysaccharides by yeasts and the influence of released polysaccharides on colour stability and wine astringency. Aust. J. Grape Wine Res. 7: Fleet, G.H Cell walls. In The Yeast. Yeast Organelles. Vol 4. A.H. Rose and J.S. Harrison (Eds.), pp Academic Press, London. Glories, I La couleur des vins rouges. 1. Les equilibres des anthocyanes et des tanins. Conn. Vigne Vin 18: Gonzalez Neves, G., C. Gomez Cordoves, and L. Barreiro Anthocyanic composition of Tannat, Cabernet Sauvignon and Merlot young red wines from Uruguay. J. Wine Res. 12: González Techera, A., S. Jubany, F. Carrau, and C. Gaggero Differentiation of industrial wine yeast strains using microsatellite markers. Lett. Appl. Microbiol. 33: Lee, D.F., E.E. Swinny, and G.P. Jones NMR identification of ethyl-linked anthocyanin flavanol pigments formed in model wine ferments. Tetrahedron Lett. 45: Loiseau, G., C. Cuinier, J.L. Berger, and D. Peyron Fixation des anthocyanes par les levures œnologiques au cours de la vinification en rouge. In Proceedings for the Société Francaise de Microbiologie, pp Dijon, France. Morata, A., M.C. Gomez-Cordoves, J. Suberviola, B. Bartolome, B. Colomo, and J.A. Suarez. 2003a. Adsorption of anthocyanins by yeast cell walls during the fermentation of red wines. J. Agric. Food Chem. 51: Morata, A., M.C. Gomez-Cordoves, B. Colomo, and J.A. Suarez. 2003b. Pyruvic acid and acetaldehyde production by different strains of Saccharomyces cerevisiae: Relationship with vitisin A and B formation in red wines. J. Agric. Food Chem. 51: Ribereau-Gayon, P The anthocyanins of grape and wines. In Anthocyanins as Food Colors. P. Markakis (Ed.), pp Academic Press, New York. Ribereau-Gayon, P., Y. Glories, A. Maujean, and D. Dubourdieu Compuestos Fenólicos. In Tratado de Enología. 2. Química del vino. Estabilización y Tratamiento, pp Editorial Hemisferio Sur, Buenos Aires. Rivas-Gonzalo, J., S. Bravo-Haro, and C. Santos-Huelga Detection of compounds formed through the reaction of malvidin 3-monoglucoside and catechin in the presence of acetaldehyde. J. Agric. Food Chem. 43: Schwarz, M., T. Wabnitz, and P. Winterhalter Pathway leading to the formation of anthocyanin-vinylphenol adducts and related pigments in red wines. J. Agric. Food Chem. 51: Vasserot, Y., S. Caillet, and A. Maujean Study of anthocyanin adsorption by yeast lees: Effect of some physicochemical parameters. Am. J. Enol. Vitic. 48: Vivar-Quintana, A., C. Santos-Buelga, and J. Rivas-Gonzalo Anthocyanin-derived pigments and color of red wines. Anal. Chim. Acta 458: Zoecklein, B.W., K. Fugelsang, B.H. Gump, and F.S. Nury Wine analysis and production. Chapman & Hall, New York.