Farm to Consumer: Factors Affecting the Organoleptic Characteristics of Coffee. II: Postharvest Processing Factors Ahsan Hameed , Syed Ammar Hussain, Muhammad Umair Ijaz, Samee Ullah, Imran Pasha, and Hafiz Ansar Rasul Suleria

Abstract: The production and consumption of coffee are increasing despite the roadblocks to its agriculture and global trade. The unique, refreshing, and stimulating final cupping quality of coffee is the only reason for this rising production and consumption. Coffee quality is a multifaceted trait and is inevitably influenced by the way it is successively processed after harvesting. Reportedly, 60% of the quality attributes of coffee are governed by postharvest processing. The current review elaborates and establishes for the first time the relationship between different methods of postharvest processing of coffee and its varying organoleptic and sensory quality attributes. In view of the proven significance of each processing step, this review has been subdivided into three sections, secondary processing, primary processing, and postprocessing variables. Secondary processing addresses the immediate processing steps on the farm after harvest and storage before roasting. The primary processing section adheres specifically to roasting, grinding and brewing/extraction, topics which have been technically addressed more than any others in the literature and by industry. The postprocessing attribute section deals generally with interaction of the consumer with products of different visual appearance. Finally, there are still some bottlenecks which need to be addressed, not only to completely understand the relationship of varying postharvest processing methods with varying in-cup quality attributes, but also to devise the next generation of coffee processing technologies. Keywords: coffee processing, cup quality, organoleptic characteristics, sensory attributes, postharvest processing

Introduction three continents (North America, Pacific Asia, and North Africa) Coffee is a valuable agricultural commodity of nonarid regions possess these least developed countries. On the other hand, coffee in the tropics. Over 50 countries are involved in the agricultural consumption mostly occurs in “first world” countries. Coffee is production of coffee, among which most, with few exceptions, regarded as a cash crop in these “second/third world” countries belong to the list of least developed countries. Geographically, and also brings long-term social strengths by generating a consid- erable amount of rural employment, businesses and environmen- tal benefits, and a significant range of globally consumed product CRF3-2018-0038 Submitted 2/28/2018, Accepted 4/27/2018. Author Hameed recipes and qualities that can end up as high-value specialty or is with the Laboratory for Yeast Molecular and Cell Biology, The Research Cen- gourmet coffees. However, these social assets of coffee-producing ter of Fermentation Technology, School of Agricultural Engineering and Food Science, Shandong Univ. of Technology, Zibo, Shandong 255000, China. Authors Hameed, countries are under great pressure owing to ever-increasing input Hussain, Ijaz, Ullah, and Pasha are with National Inst. of Food Science & Technology, costs, unstable coffee markets, a lack of incentives to improve qual- Univ. of Agriculture Faisalabad, Pakistan. Authors Hussain and Ullah are with Colin ity, increasing resistance of coffee pathogens, and climate change. Ratledge Center for Microbial Lipids, School of Agriculture Engineering and Food The lack of incentives, climate change, and increased resistance of Science, Shandong Univ. of Technology, Zibo, P.R. China. Author Ijaz is with Key Laboratory of Meat Processing & Quality Control, College of Food Sciences, Nan- coffee pathogens are considered the main causes of deterioration jing Agriculture Univ., Jiangsu, P.R China. Author Suleria is with UQ Diamantina in the physical and organoleptic quality attributes of coffee which Inst., Translational Research Inst. Faculty of Medicine, The Univ. of Queensland, are the main drivers for global coffee trade and consumption. 37 Kent Street Woolloongabba, Brisbane, QLD 4102, Australia. Author Suleria Consequently, newly introduced regulatory measures related to is with Dept. of Food, Nutrition, Dietetics and Health, Kansas State Univ., Man- environmental and public health and trade tariffs on value-added hattan, Kans. 66506, U.S.A. Author Suleria is with Centre for Chemistry and Biotechnology, School of Life and Environmental Sciences, Deakin Univ., Pigdons coffee products (specialty coffees) are further adding serious con- Road, Waurn Ponds, VIC 3216, Australia. Direct inquiries to author Suleria (E- cerns about the coffee trade (International Coffee Organization, mail: hafi[email protected]). 2014a, 2014b). Despite all these constraints, world coffee consumption is grow- Ahsan Hameed and Syed Ammar Hussain are equal first authors for this manuscript. ing steadily at an annual rate of 2.5%, with an amazing 17% per

C 2018 Institute of Food Technologists® r 1184 ComprehensiveReviewsinFoodScienceandFoodSafety Vol.17,2018 doi: 10.1111/1541-4337.12365 Farm to consumer . . . annum increase in its agricultural production too (International Roasting is next to fermentation in transforming the green Coffee Organization, 2017). The overall credit for increasing pro- coffee beans into roast coffee beans. The primary objective of duction and consumption of coffee goes only to its overwhelming roasting is to convert the green coffee beans into a brittle and organoleptic or final cup quality attributes. The quality of coffee extractable form. Green coffee beans behave as “mini bioreactors” is a multifaceted and complex trait. By and large, 2 comprehen- upon exposure to roasting temperature (200 ± 10 °C), during sive factors affect the physical, chemical, and sensory character- which various biochemical reactions (Maillard reactions, pyroly- istics of coffee in a number of complex ways. Based on nature sis, Strecker degradation, and so on) result in the production of and the sequence of events, these 2 factors can be termed as more than 1000 different types of aroma compound (Nebesny, “preharvesting” factors and “postharvesting” factors (Wintgens, Budryn, Kula, & Majda, 2007). Coffee aroma is the consequence 2008). It is established fact that 40% of the physical, chemical, of a complex balance of these many compounds. Any fluctuations and sensory features of coffee (beans) are defined by preharvesting in the roasting conditions can affect the aromatic profile of roast factors, and the remaining 60% of quality scores are determined beans by hampering the complex balance of these volatiles and by the method of postharvest processing (Richard, Charles, & their intermolecular interactions (Charles-Bernard, Roberts, & Mitiku, 2007). Preharvesting factors are mostly related to agri- Kraehenbuehl, 2005; Nagaraju & Bhattacharya, 2010). Roasting cultural variables ranging from the selection of a suitable geo- is followed by storage and grinding of roasted beans to equilibrate graphical location to harvesting. The most discussed agricultural the moisture and aromatics balance and to increase the bean sur- factors of coffee in the literature are altitude, latitude, slope of face area exposed to extraction or brewing, respectively (Akiyama land, coffee variety, seedlings, soil, and fertilization, rainfall and et al., 2003a). Grinding is a crucial step since it is the begin- irrigation, sun and shade ecosystems, frost and climate change, ning of coffee production, and it has been empirically optimized. coffee pathogens, and harvesting strategies. The interaction of The optimal grinding grade permits contact of the maximum sur- preharvesting variables in shaping the overall quality attributes face area with the hot water to attain a high-quality coffee brew of coffee has been comprehensively discussed and reviewed in (Akiyama et al., 2003b; Baggenstoss, Thomann, Perren, & Escher, the predecessor to this review article. This second article is in- 2010; Severini, Ricci, Marone, Derossi, & De Pilli, 2015). In addi- tended to be a successive serial review of its first part, focusing on tion to grinding, the beverage is also affected by variable roasting the postharvest processes influencing the organoleptic features of conditions. Roasting technology (that is, temperature, pressure, coffee. coffee/water ratio, roasting beds, roasting level, water quality), Following harvesting, coffee cherries go through a complex roasting kinetics, roasting indices and roasting physics are some series of postharvest processing steps to be in a more stable, of the key roasting aspects which have been thoroughly investi- transportable, and roastable form. Depending upon the nature gated with respect to aroma or volatiles extraction and their rela- of the product, the initial postharvest processing steps ensure the tion to final cupping quality (Gloess et al., 2013; Loapez-Galilea, safe transformation of the highly perishable green cherries into Fournier, Cid, & Guichard, 2006; Mestdagh, Davidek, Chaumon- somewhat stable green coffee beans with a moisture content of teuil, Folmer, & Blank, 2014; Salamanca et al., 2017; Severini et al., 10% to 12% wet basis (w.b.), to avoid unwanted fermentation. 2015). Therefore, the initial postharvest processing step is known as nat- In summary, it is obvious that each and every step in posthar- ural drying, wet drying, or semi-wet drying based on the mech- vest coffee processing is critical and has an obvious impact on anism adopted (Kleinwachter¨ & Selmar, 2010; Velmourougane, final cupping quality. The choice of postharvest processing system Shanmukhappa, Ventakesh, Prakasan, & Jayarama, 2008). Natu- is culture-driven and depends upon the coffee type and origin. ral drying involves drying the mature whole coffee fruit under However, despite the association of postharvest processing systems the sun followed by the manual or mechanical removal of outer with culture, there is also a trend to introduce fluctuations into unwanted layers. “Naturals” offer a coffee with heavy body and routine processing technologies in order to bring novel changes smooth, sweet, and complex cupping quality attributes (Poltron- in the flavor of coffee. The literature has much data about com- ieri & Rossi, 2016). The wet method, on the other hand, uses plex postharvesting processing systems and their relationship with the coffee cherries after the removal of pulp and mucilage; it organoleptic beverage characteristics. However, the literature is also has many variations in its operating procedure, and, as a still lacking a comprehensive review elaborating the mechanisms whole, all these variations significantly influence the taste, fla- by which postharvest processing systems influence the physical and vor, and aroma potential of coffee. Spontaneous fermentation oc- organoleptic characteristics of coffee. Therefore, the objective of curs in the pulp and mucilage of fruit during the natural dry- this review article is to collect and sum up all the previous and ing process. Uncontrolled and unknown indigenous (pectinolytic) up-to-date research knowledge about postharvest coffee process- and diverse microbes grow on the high concentration of glu- ing systems and to define their decisive affiliation in modeling final cose and fructose in fruit pulp and produce alcohols, organic cupping quality. acids, and other metabolites. In wet and semi-wet drying, spe- cific groups of microbial strains are intentionally inoculated for fermentation, which produces metabolic compounds that give Postharvest Processing Systems a metabolite profile different from that for dry processing. The Coffee processing begins just after the harvesting of coffee cher- production of metabolic compounds is highly species-specific, ries. Coffee cherry processing consists primarily of drying (natu- and different groups of bacteria, yeasts, and filamentous fungi ral, wet, or semi-wet), fermentation, roasting, storage, grinding, generate different classes of metabolites. These metabolic com- and brewing. For the sake of better understanding the nature, pounds can diffuse into the coffee grains and can affect the final significance, and sequence of processing events, initial and im- beverage quality by interfering with volatile and nonvolatile aro- mediate processing of cherries involving green bean production matic compounds (Masoud, Cesar, Jespersen, & Jakobsen, 2004; is here grouped into secondary processing, while processing after Pereira et al., 2014, 2015a; Pereira, Soccol, Brar, Neto, & Soccol, green bean production is grouped into the primary processing of 2015b). coffee.

r C 2018 Institute of Food Technologists® Vol.17,2018 ComprehensiveReviewsinFoodScienceandFoodSafety 1185 Farm to consumer . . .

Secondary processing et al. (2015) documented the same results, elaborating that average Secondary coffee processing aims to transform coffee cherries sun-drying rates on drying patios (temperature: 10 to 28 °Cand into green coffee beans and to lower the water content of fresh RH: 34.5% to 61.2%) are 0.002 kg/kg H2O/hr for fully washed cherries to a level which allows the preservation of beans (about coffee and 0.002 kg/kg H2O/hr for natural coffee. In agreement 11% to 12%); it involves the removal of all the covering which sur- with the results of Tsegaye, Mohammed, Shimber, Getachew, and rounds the beans and preparation of the beans according to market Getachew (2014), Taveira and colleagues also claimed better cup requirements. There are primarily 2 processes involved in process- quality conservation for slow drying (60/40 °C, details below) on ing or drying of coffee cherries, namely dry processing and wet patios with hot air rather than on flat-ground surfaces. Physical and processing. The dry process yields “naturals” while washed coffee biochemical quality attributes are lower for flat-ground surface- is produced by the wet process (Ameyu, 2017). The choice of dried coffee (Tsegaye et al., 2014). The reason for these lower drying method depends upon climatic factors, harvesting factors, total quality attributes are contact with foreign matter, rewetting, and production region. Terms commonly used in the trade world and consequent mold and bacteria growth, and black bean for- of green coffee beans are “dry natural,” “pulped natural,” “wet mation due to less air movement, whereas mesh wire or bamboo hulled,” or “fully washed” (Bytof, Knopp, Schieberle, Teutsch, mats not only protect the cherries from contamination, but also & Selmar, 2005; Poltronieri & Rossi, 2016). Pulped natural and guard them against rewetting due to more air movement (Selmar, “honey processing” have the least dependence on fermentation Bytof, Knopp, & Breitenstein, 2006). Additionally, higher drying and the least acidity, while dry naturals have the longest span rates at high temperatures can be harmful for coffee quality due of fermentation. Controlled fermentation involves wet-hulled or to damage to cell membranes (Marques, Borem, Pereira, & Biag- fully washed coffee beans and is more acidic than that used for gioni, 2008; Saath et al., 2010). Studies have revealed that coffee pulped natural and honey processing. Honey and pulped naturals membranes get damaged at a moisture level of 20% to 30% (w.b.) are the result of a third processing method called semi-wet pro- at temperatures of 60 °C or higher (Borem et al., 2008a, 2008b, cessing. Generally, coffee quality is a complex characteristic which 2014). Thus, a high drying temperature at the start of the drying depends on a series of preharvest and postharvest factors such as process followed by a low range of temperature can be a promising the genetic, environmental, and agronomical practices, harvesting way to maintain coffee quality. The 60/40 °C treatment (initially and postharvest handling, drying system, storage conditions, in- 60 °C followed by 40 °C) via hot air shows better results, with dustrial processing, roasting system, preparation of the beverage, lower values for germination, emergence, root protrusion, ESI, and prefernce of the consumer (Avelino et al., 2005; Decazy et al., and open cotyledonary conditions for natural and washed coffee 2003; Knopp, Bytof, & Selmar, 2006; Laderach, 2010; Leroy et al., (Taveira et al., 2015). This intermediate drying rate also favors the 2006; Silva et al., 2004). Inadequate systems of harvesting, process- accumulation of high-molecular-weight sugars and reducing sug- ing, storage, and transportation are responsible for the widespread ars (Borem et al., 2008a, 2008b; Marques et al., 2008). Coffee layer failure to maintain the inherent quality of coffee produced in thickness also matters to the final raw, cup, and total quality of cof- Ethiopia (Alemayehu et al., 2008). Thus, the major postharvest fee. Tsegaye and coworkers also stated that sun-drying using raised processing factors affecting coffee quality are described below. bed surfaces and a layer thickness of no more than 20 kg/m3 can be Dry processing. The processing variables involved in convert- an effective method for producing specialty-grade coffee with the ing coffee cherries into green beans appear to be of major impor- desired inherent characteristics. With an increase in coffee cherry tance. Of these processing variables, natural or sun-drying is the quantity on increasing the drying area (>20 kg/m2), the percent- most commonly used, and it is the oldest, cheapest, and easiest age of infected beans increases. Almost 80% of cherries were found way to transform the cherries into green beans. Dry processing to be infected when dried at 40 kg/m2 (Kouadio, Koffi, Nemlin, involves sun-drying/fermenting of the whole fruit, and produces & Dosso, 2012). A thicker layer of cherries (>20 kg/m2) causes coffee that is heavy in body, sweet, smooth, and complex. Sun- inefficient air movement and nonhomogeneous drying; conse- drying is a long process, and 95% of Arabicas from Brazil, Ethiopia, quently, more fermentation is required. This higher degree of Haiti, Indonesia, Paraguay, India, and Ecuador are sun-dried (Silva, fermentation causes the acidification of cherries and degradation Batista, Abreu, Dias, & Schwan, 2008). From a raw quality point of chlorogenic acids (Kouadio et al., 2012). A higher level of fer- of view, greater bean size and roast volume are observed for dry- mentation also brings the threat of more microbial growth which processed beans, whereas a higher moisture content is noted in can also increase the level of a mold mycotoxin called ochratoxin a wet-processed beans (Tadesse, Jemal, & Abebe, 2015). With re- (OTA) (Suarez-Quiroz´ et al., 2004). Further, the inclusion of de- spect to shape, size (appearance), and color, dry-processed beans graded/defective and infected cherries in processing causes a foul- are no longer considered better than wet-processed beans. A higher smelling and off-flavor coffee. This foul smell and taste may be due number of defective beans with earthy, musty, and greenish color to the production of some compounds at higher concentrations coffee defects are found in dry-processed coffee (Tadesse et al., than in mature and healthy seeds (Toci & Farah, 2014). Table 1 2015). Natural processing is the most difficult way to protect and shows details of the difference between defective and nondefective maintain the high quality of coffee, as whole coffee cherries are green coffee beans. Furthermore, Toci and Farah detected 159 sun- or machine-dried under direct sunlight on flat surfaces, raised totally different compounds in degraded/defective and infected beds (bare earth, cemented or bricked floor, tables with wire mesh, seeds after roasting, and, they noted that these were all alcohols, bamboo mats, asphalt, or wooden surfaces) or in drying machines carboxylic acids, esters, ketones (primary compounds during pri- along with the intact outer fruit layers, with Robusta preferably mary processing), pyrazines, furans, ketones, pyrroles, pyridines, being dried on bare earth, while special cemented or bricked and esters (secondary compounds) in roasted seeds. Whole dried terraces/floors are used for Arabicas (Food and Agriculture Or- coffee fruit is then dehulled mechanically to produce green coffee ganization [FAO], 2006). Coffee dried on a flat surface (brick beans. Dry-processed coffee is lower in acidity, and more sweet, terraces) dries more quickly than that dried on raised-bed surfaces tasty, smooth, and complex in flavor than wet-processed coffee (mesh wire and bamboo mats) due to prevailing lowland high (Uman et al., 2016). Only fully ripe coffee cherries are used for temperature and condensation effects. On the other hand, Taveira wet processing, whereas fruits at all stages of ripeness are utilized for

r 1186 ComprehensiveReviewsinFoodScienceandFoodSafety Vol.17,2018 C 2018 Institute of Food Technologists® Farm to consumer . . .

Table 1–Physical and biochemical differences between defective and nondefective beans.

Attributes Nondefective green beans Defective green beans Physical Attributes Shape Oval or/half a triaxial ellipsoid Irregular Size (inches) Larger (12/64 to 18/64 diameter) Smaller Color Blue to gray green (Arabicas)/brownish (Robusta) Black or dark brownish Volume x 109 (m3) 72–119 40–94 Bean density (kg m−3) 1210–1310 1133–1234 Bulk density (kg m–3) 625–630 595–648.8 Swelling during roasting (%) 65 35–41 Heat transfer rate/Roasting rate High Slow Roast weight loss (%) 13.9–16.1 10.3–14.5 Luminosity (L) (green beans/roast beans) 36–39/17–21 21–38/17–19 Chroma(a) (green beans/roast beans) 17–37/11–16 9–26/6–9 Hue angle (b) (green beans/roast beans) 84–88/68–78 79–98 Proximate attributes (roasted beans/green beans) Water (g/100g d.b) 1–3/8–13 1–2/9–10 Protein(g/100g d.b) 11–15/13–17 13–15/15–16 Lipids(g/100g d.b) 15–20/9–18 10–12/8–10 Minerals/Ash(g/100g d.b) 4–5/4–5 5–6/5–6 Carbohyderates(g/100g d.b) 40–71/60–76 66–71/59–71 Lipids(g/100g d.b) 10.4–11.5 9.2–10 Chlorogenic acid (5-CQA) (% d.b) 3.2 1.8 Caffeine (%) 0.9 1.3 Trigonneline (%) 1 0.8 Ph 6.2–6.7/5.52–5.9 6.04–6.5/5.3–5.5 Acidity (ml/100g) 98.1–111.3/196.4–242 101–114/207.2–263.3 Polygenic amine (Histamines) (mg/100g of beans) Traces >0.1 Polygenic amines (mg/100g of beans) 12.7 up to 10.3 Volatiles compounds (Signal to noise ratio) Benzene acetic acid – 104–213 1H-pyrrole – 80–226 4-Methylthiazole – 90–246 1,5-Dimethyl-2-pyrrolecarbonitrile – 592–863 methyl ester 2-Isoamyl-6-methylpyrazine – 310–521 Organoleptic attributes Smooth, pleasant, caramel, fruity, chocolate Heavy (black beans), sour, onion and astringent flavor Note: Here the term “defective beans” referred to all kind of coffee green beans or/and cherries which are considered infected, overfermented, black beans, immature beans and degraded. Data were compiled from Clark, 1985; Clarke and Macrae, 1985; Clarke and Macrae, 1987; Dutra et al., 2001; Franca et al., 2005a; Farah et al., 2006a; Mazzafera, 1999; Mendonc¸a et al., 2009; Mendonc¸a, 2006; Oliveira et al., 2006; Ramalakshmi et al., 2007; Rodrigues et al., 2003; Sivetz and Desrozier, 1979; Speer and Kolling-Speer,¨ 2001. dry processing (Maier, 1981; Sivetz & Desrosier, 1979). However, Toacknowledge the effect of different drying systems on cup qual- these factors play a minor role, as typical differences in cup qual- ity, Do Livramento et al. (2017) conducted proteomic analysis of ity between washed and unwashed coffees also exist, even though green coffee beans processed via different drying systems. They matured and unmatured coffee cherries are used in both processes deduced that natural coffee (naturals) sun-dried in the yard has the (Selmar, Bytof, & Knopp, 2002). Mean daily temperature, relative highest protein content (avg. 252 spots as compared to 360 for humidity/dew, air flow, layer thickness, rainfall, pressure, and stir- pulped) after wet-processed pulped coffee (pulped), as compared ring are the important factors that influence the rate of drying. to natural and pulped coffee dried in dryers and other facilities at Rain and higher relative humidity favor the growth of coffee bean 60 °C. Comparative proteomic analysis revealed 18 to 22 kinds of pathogens that can affect fermentation and the quality of the re- proteins which are 1.7- to 14-fold more abundant in yard-dried sulting product. Temperature is a core factor in deciding the drying beans; these proteins correspond to the storage proteins 11S glob- period (Illy & Viani, 2005; Uman et al., 2016). Over time, sun- ulin and glyceraldehyde-3-phosphate dehydrogenase, and proteins drying systems have been modified/combined with other facilities present in late embryogenesis, LEA, dehydrin, and others. So, the like solar-aided drying, mechanical drying, and hybrid drying sys- secret of high-quality yard-dried coffee also lies in higher amounts tems. Solar-aided drying pursues an increase in the drying rate by of protein and certain free amino acids which with reducing sugars attracting a high temperature and convective air flow using the engage in Maillard reactions during roasting, to produce extra fla- greenhouse effect and the sun as a source of energy. The applica- vor and aroma (Reineccius, 1995; Selmar, Bytof, & Knopp, 2008). tion of electric fan circulation (1 m3/min/m2) may increase the Greater expression of storage proteins in yard-dried beans is also a air temperature to 5 to 7 °C above ambient temperature (FAO, source of free amino acids (Andrade et al., 2011). The greater con- 2010a, 2010b). Static beds or silo dryers (both horizontal and ver- tent of the above-mentioned proteins is a function of minimizing tical) with solar-heated air forced (contra and concurrent flow) oxidative stress by creating thermal, desiccation, and dehydration through the coffee bed are used in mechanical drying. Hybrid tolerance, osmoprotection, stabilizing hydrophobic interactions, drying systems rely on sun-drying of coffee cherries during the and cellular maintenance (Berjak & Pammenter, 2007; Kosova, day, while heated air from a perforated duct is used to dry the V´ıtamvasa, Prasil, & Renaut, 2011; Mowla et al., 2006; Tun- cherry heap at night (FAO, 2010a, 2010b). Drying is the second nacliffe & Wise, 2007). The variation in total protein content most important step in the postharvest processing of coffee cher- between beans dried in a yard and those dried at other facilities ries and can produce premium quality aroma and flavor if properly may arise due to a higher rate of degradability or lesser expression, handled to minimize damage to proteins, enzymes, and other sen- asthedryingtemperaturecanbeupto60°C in other facilities sitive constituents (Taveira, Rosa, Borem,´ Giomo, & Saath, 2012). (Do Livramento et al., 2017). Different processing treatments also

r C 2018 Institute of Food Technologists® Vol.17,2018 ComprehensiveReviewsinFoodScienceandFoodSafety 1187 Farm to consumer . . . influence the pool of free amino acids, which are precursors of greater sensitivity to temperature and thermal degradability of the flavor and color of a coffee brew, by altering the metabolic CGA and trigonelline are possible causes of their low content response under different processing conditions; Bytof et al. (2005) in dry-processed beans (Duarte et al., 2010). The hypothesis of described that conversion of glutamic acid into γ -aminobutyric leaching down hexoses, as a leading cause of less sugar, during acid (GABA) mediated by enzymatic α-decarboxylation is related wet processing was cross-checked by Kleinwachter¨ and Selmar to a physiological drought stress situation and is specific to the (2010) who compared sun-dried and machine-dried (continuous mode of processing applied. The wet process leads to an increase and with rhythmic day and night pauses) wet-processed beans; in glutamic acid, whereas green beans from the dry process have in contrast, they noted a strong decline in glucose and fructose a higher concentration of GABA. The accumulated GABA is a contents even in the absence of water on the first day. The hexose result of a decarboxylation reaction which occurs only during dry content increases following the tremendous decrease on the first processing. GABA may also accumulate during extended wet pro- day of processing, and then again decreases in the latter half of cessing, but its quantity is always less than that in dry-processed processing. The increase in hexose content in the first half of beans. Bytof and coworkers also stated that accumulation of GABA processing may be due to hydrolysis of storage carbohydrates, also corresponds with the stress reaction that occurs in beans only for example, cell wall polysaccharides (CWP) (Bucheli, Meyer, when the water content of beans dips below 25%. This period is Pittet, Vuataz, & Viani, 1998; Joet¨ et al., 2010). This zigzag extended in dry processing due to the long time (2 to 3 weeks) fluctuation in hexose and galactose contents may be due to dif- required to reach the desired moisture level (for example, 12%) ferent metabolic activities and reactions during drying (Joet¨ et al., as compared to the 3 to 6 days in wet processing. It is also note- 2010; Kramer et al., 2010). CWP were further analyzed in coffee worthy that utilization of a greater quantity of unripe cherries beans processed by the three distinct processes and were found than mature ones may have toxicological effects on health, as un- to be in the descending order of wet > semi-dry > dry process. ripe cherries contain a greater amount of amino acids regardless Arabinose, mannose, and galactose are the main monosaccharides, of which processing method is applied. These amino acids may and galactomannans and arabinogalactans the predominant include asparagine, which may play a role with reducing sugars in polysaccharides (Fischer, Reimann, Trovato, & Redgwell, 2001; Maillard reactions to form the toxicological compound acrylamide Oosterveld, Harmsen, Voragen, & Schols, 2003a). Arabinose is or its precursors (Dias, Borem, Pereira, & Guerreiro, 2012). found in greater amounts in processed beans, whereas galactose After free amino acids, soluble carbohydrates are considered im- and mannose contents decrease in processed beans. Greater portant precursors for various volatile and nonvolatile compounds amounts of galactomannans and arabinose + galactose are found (Knoop et al., 2006). There are also some reports of loss of soluble in semi-dry and dry-processed beans, respectively. The presence of carbohydrates due to germination-related metabolism in beans a smaller amount of galactomannans in processed beans (dry and during processing (Selmar et al., 2005a, 2005b; Wootton, 1974). wet) is evidence of the germination process during postharvest Sucrose (7.07% d.m.), glucose (0.23% d.m.), and fructose (0.39% treatments. However, despite there being fewer polysaccharides in d.m.) are the major carbohydrates, whereas raffinose, stachyose, processed beans, no influence on the viscosity and surface tension galactose, arabinose, rhamnose, and mannose are the minor of coffee has been documented (Taveira et al., 2015). There is also (<0.1%) soluble carbohydrates detected (Knopp et al., 2006). The an indication that ecological stimuli like sun, shade, and altitude levels of the two hexoses (glucose and fructose), arabinose, and have a pivotal impact on the sugar composition of coffee beans mannose are highly influenced by postharvest treatment. Naturals (Geromel et al., 2008; Guyot et al., 1996). Further influences on have a greater amount of these hexoses, arabinose, and mannose green coffee carbohydrates occur during postharvest treatments. than washed coffee, whereas the levels of arabinose and mannose Unlike proteins, free amino acids and carbohydrates, and despite are intermediate between those in washed and unwashed coffee. the noninvolvement of lipids in the Maillard reaction, lipids are However, the levels of other sugars (sucrose, stachyose, raffinose, still considered to be the source of hydrophobic aroma compounds and rhamnose) remain the same irrespective of postharvest treat- that are responsible for the mouthfeel of the product (Illy & Viani, ment (Knopp et al., 2006), although, previously, Guyot, Petnga, 2005). Linalool enantiomers and trans-2-nonenal are two lipid fla- Lotod´e, and Vincent (1988) had obtained contradictory results, vor compounds responsible for “woody” off-notes and old-crop reporting a higher fructose content for wet processing than for dry coffee flavor, respectively (Bonnlander, Cappuccio, Suggi- processing. This contradiction in the literature may have arisen Liverani, & Winterhalter, 2006, 2007a, 2007b). For a long time, due to the use of different coffee species/hybrids/varieties and by the fatty acid profile of coffee was used as a coffee variety marker Guyot et al. (1988) not explicitly describing the sugar quantifi- until it was found that climate and environmental conditions also cation method they had used. The reason for a smaller amount influence the fat and fatty acid profiles (Villarreal et al., 2009). The of hexoses for the wet processing method may be because the literature contains very few attempts at investigating the impact of fermentation needs to capture more hexose molecules to produce processing on coffee lipids. These very few studies stated that the the same number of ATPs for respiration during dry processing method of coffee processing influences the enantiomeric ratio of (Knopp et al., 2006). Duarte, Pereira, and Farah (2010) also deter- linalool. During dry processing, an excess of S-(+)-linalool is de- mined the compositional differences of green beans to find out the tected, whereas wet processing elicits a racemic mixture of linalool causes of difference in the cup quality of beans processed by dry, (Bonnlander,¨ Cappuccio, Suggi-Liverani, & Winterhalter, 2006), semi-dry, and wet methods; they verified the above-mentioned with a significant increase in the lipid content (Joet¨ et al., 2010). results, stating a higher sucrose content in green beans processed Dry processing has its own microbial ecology, diversity, by the dry and semi-dry methods. Wet-processed green beans tend structure, and community nature which is unique from those of to have a higher chlorogenic acid (CGA) and trigonelline content, wet and semi-wet processing. This uniqueness may result in the whereas caffeine enjoys the status of being unaltered irrespective of variation of fermentation, its products, and finally, cup quality the processing treatment adopted (Duarte et al., 2010). According (Silva et al., 2000, 2008). A total of 940 microbial colonies or to the literature, better solubility of carbohydrates in water is the isolates were detected by Silva and coworkers during dry process- reason for their lower level in wet-processed coffee beans, whereas ing, among which there is variation in microbial populations and

r 1188 ComprehensiveReviewsinFoodScienceandFoodSafety Vol.17,2018 C 2018 Institute of Food Technologists® Farm to consumer . . . diversity with respect to variations moisture contents, but bacteria mucilaged coffee, in which the mucilage is specifically removed always represent the majority of microbes during processing. The by mechanical means. The chemical composition and final cup 4 distinct phases of moisture content (0 to 4, 6 to 12, 13 to 18, quality also depend upon the processing manner adopted in wet and 19 to 24 days) represent different microbial profiles. The processing (Gonzalez-Rios et al., 2007a, 2007b; Taveira et al., starting period shows a high bacterial population, which then 2015). Gonzalez-Rios et al. (2007b) studied the effect of all mod- decreases with decreasing moisture content, while yeasts show ified wet process methods on the final cup quality of coffee. The an inverse trend with respect to bacterial population (Silva et al., modifications studied included pulping via a disc pulper (T1 and 2000, 2008). Similarly, the ratio of fungi to yeast is also inverted, T2) or drum pulper (T3 and T4), and mucilage removal via natural which may be due to biological control of fungi by Debaryomyces fermentation with water (T1), natural fermentation without water hansenii (Silva et al., 2008). Species of that family predominate (T2 and T3), or mechanical removal (T4). These modifications in among Gram-negative bacteria, and species of the Bacillus genus processing lead to alterations in the amount of volatile compounds predominate among Gram-positive isolates. Bacteria identified of 15 classes (8 acids, 6 pyrroles, 4 aldehydes, 4 esters, 20 furans, during dry processing include B. subtilis, B. cereus, B. megaterium, 3 alcohols, 17 ketones, 2 lactones, 13 pyrazines, 3 pyridines, 1 B. macerans, B. polymyxa, Kurthia, Arthrobacter, Enterobacter ag- thiazole, 1 benzene compound, 4 phenolic compounds, 1 ter- glomerans, E. sakazakii, E. cloacae, E. aerogenes, Acinetobacter sp., pene compound, and 1 amine in trace form). Furans, ketones, and Serratia plymuthica, Tatumella ptyseos, Pseudomonas paucimobilis, P. pyrazines are the major classes detected. Thiazole is present for alcalifaciens, P. putrefaciens, Providencia mirabilis, Klebsiella oxytoca, K. all 4 treatments of light roast. Variations in wet processing bring ozaenae, Shigella dysenteriae,andYersinia sp. The 2 most-identified about random variations in the aldehyde and acid groups due to yeast species are D. hansenii and Pichia guilliermondii (Silva et al., their origin from postharvest treatments. Dry treatment (T3) gives 2000, 2008). The 364 filamentous isolates identified by Silva et al. more volatile compounds belonging to furans, ketones, pyrazines, (2000, 2008) were 132 Aspergillus spp., 101 Penicillium spp., 58 Cla- acids, and aldehydes. “Ecological dried wet processing” resembles dosporium spp., 44 Fusarium spp., 15 Pestalotia spp., 13 Paecilomyces mechanical wet processing (Gonzalez-Rios et al., 2007a, 2007b). spp. and Cladosporium cladosporioides. A high bacterial population Additionally, classical wet processing proves to be always superior, during fermentation suggests production of organic acids which providing a pleasant fruity note due to esters (light roast), and floral can percolate from pulp and mucilage onto bean surfaces, affecting and caramel notes due to ketones and furans for darker roasting. the organoleptic qualities of coffee. So, variation in microbial Mechanical demucilage treatment produces the most unpleasant ecology depends upon the postharvest process adopted and may coffee of all, with sour, toasted, bitter, burnt, and spicy notes. be another reason for variation in the cup quality of coffee, as Variations in pulping methods only yield indiscriminate variations microbial ecology structure primarily varies depending on the (Gonzalez-Rios et al., 2007b). Very similar results were docu- postharvest process, and secondarily, varies according to location, mented by Lymen et al. (2011), who also studied the effect of variety, and species (Hamdouche et al., 2016). Chryseobacterium different methods of drying beans, and the number of washings in and Sordariomycetes spp. are only detected in dry-processed coffee, wet processing. Sun-dried washed coffee has a more intense fruity while lactic acid bacteria (LAB) (Weissella, Lactobacillus, Leuconostoc flavor with a heavier body than mechanically dried washed cof- mesenteroides), Ralstonia and/or Acidovorax bacteria are not detected fee. ATR-FT-IR spectroscopy analysis revealed a greater content in dry processing, linking their nonrequirement to fermentation of esters, aldehydes, carbonyls (–(CH2)4–), and lipid esters, and during dry processing (Hamdouche et al., 2016). lower contents of vinyl esters and lactone carbonyls in sun-dried Conclusively, dry processing affects the composition of beans washed coffee beans than in washed coffee beans (Lymen et al., (that is, proteins, free amino acids, free fatty acids, low-and high- 2011) These extra esters, aldehydes, carbonyls (–(CH2)4–), and molecular-weight sugars and reducing sugars), due to which dry- lipid esters in sun-dried washed coffees may come from mucilage processed coffee has its own distinct quality features. Additionally, remnants. Aging of these mechanically dried fully washed beans for a detailed study of coffee beans and germination physiology also causes deterioration of the cup quality due to degradation during processing, recent comprehensive reviews and chapters are of aldehydes, which in turn can reduce their masking of ketones, available (Joet¨ et al., 2010, 2014; Selmar, Kleinwachter, & Bytof, lactones, and vinyl esters. Similarly, pulped natural (1 wash) coffee 2015). has a heavy body, lingering fruity sweetness, darker fruit notes, and Wet processing. Coffee fruit consists of an outer skin or peel muted acidity as compared to the vibrant acidity, medium body, called the exocarp, under which there is a layer of pulp. The pulp and chocolate-sweet taste of fully washed (2 washes) coffee. These is followed by the mucilaginous mesocarp (mucilage) which, in differences in cup quality may be attributed to the high lactone, turn, strongly adheres to the endocarp, also called parchment. This aliphatic acid, and ketone content detected in fully washed coffees parchment protects the two coffee seeds which are surrounded by (Lymen et al., 2011). silver skin (Evangelista, Miguel, Silva, Pinheiro, & Schwan, 2015). As discussed for dry processing, wet processing influences the Fermentation is often carried out in order to remove the pulp biochemical composition of coffee beans. Free sugars are the first and mucilage before processing. Compared to dry and semi-dry target of wet processing. Glucose, fructose, and myo-inositol con- methods, wet processing generally involves mechanical peeling and tents are reduced, while stachyose and sorbitol contents are in- depulping, leaving mucilage to adhere to beans. These depulped creased. CWP and lipids also increase, but the overall fatty acid coffee beans are then subjected to fermentation in large water profile is stable. CGA and caffeine are the compounds least affected tanks for 6 to 72 hr, during which the remaining adhering mu- by wet processing. However, a slight drop in 3,5-di-caffeoylquinic cilage is microbially degraded and solubilized, and the beans are acid (3,5-di-CQA) and an increase in 4,5-di-CQA and the ra- finally sun-dried (Silva, 2014a, 2014b). More specifically, in wet tio of CQA to di-CQA is seen (Joet¨ et al., 2010). Measurement processing, after removal of the exocarp, 3 kinds of coffee are of net loss and gain of compounds reveals a correlation between produced: (i) peeled cherry coffee, obtained after peeling and lowering sucrose and increasing CWP. Similarly, an increase in drying of beans; (ii) fully washed coffee, obtained after peeling sorbitol is coupled with an increase in glucose and myo-inositol. and subsequent mucilage removal by fermentation; and (iii) de- Targeting of these specific metabolites in wet processing clearly

r C 2018 Institute of Food Technologists® Vol.17,2018 ComprehensiveReviewsinFoodScienceandFoodSafety 1189 Farm to consumer . . . depicts the occurrence of metabolism, which leads to mobiliza- omyces hansenii, Meyerozyma caribbica, P. fermentans, T. delbrueckii, tion of energy reserves, a sign of germination. This hypothesis is Candida railenensis, C. quercitrusa, C. glabrata, Wickerhamomyces ci- reinforced by overexpression of the glyoxylate cycle-specific isoci- ferrii, and W. anomalus. T. delbrueckii was found to be the major trate lyase (ICL) gene (Selmar et al., 2006), followed by abundance yeast, with a population of 5.3 to 4.9 log CFU/g during fermen- of β-tubulin, a marker for cell division or elongation (Bytof et al., tation and drying, followed by H. uvarum (4.1 log CFU/g) and 2007) in wet processing. However, the results of Joet¨ et al. (2010) M. caribbica (3.8 to 4.0 log CFU/g) (Evangelista et al., 2015). On do not agree completely with the mobilization hypothesis as they the other hand, Feng et al. (2016) reported that the major yeasts documented the accumulation of storage compounds (CWP and were H. uvarum, W. anomalus, P. k l u y v e r i , and D. hansenii. Among lipids). However, their results give us a clue to stress metabolism the organic acids produced due to metabolism of these microbial in which glucose content is highly correlated to sorbitol biosyn- communities, lactic acid is predominant (up to 2.33 g/kg), fol- thesis (via the ethanol fermentative pathway), as sorbitol and other lowed by acetic acid (0.02 g/kg) and traces of citric, malic, and polyols play important roles as osmoprotectants. succinic acids (Evangelista et al., 2015; Feng et al., 2016). The Nutrient-rich pulp and mucilage are the cause of a good micro- highest value for lactic acids also verifies a greater population of bial population throughout the wet processing of coffee. However, bacteria than of yeasts, although quality-impairing organic acids as in dry processing, microbial diversity is subject to variation de- (that is, propionic and butyric acids) were found in the smallest pending upon regional characteristics, coffee fruit composition, amounts. The discrepancies for wet processing among different and the method of fermentation (Feng et al., 2016; Silva et al., microbial ecological works may have arisen due to coffee variety, 2014a, 2014b, 2014c). These microbes consume the nutrients of fermentation conditions, and ecological, regional, and operational pulp and mucilage and secrete organic acids and other metabo- variations. lites that percolate into coffee seeds and influence the final cup Classically, coffee quality attributes during processing refer to quality. Variation in microbial diversity and ecology may bring only 2 general factors, that is, usage of mature fruits exclusively for about changes in the type of these organic acids and metabolites wet processing, while fruits of all stages are used in natural process- or the extent to which they are secreted, and hence produce dif- ing. Secondly, wet processing needs much more persistence and ferent distinct coffee qualities (Massawe & Lifa, 2010; Silva et al., accuracy than dry processing (Brando, 2004; Illy & Viani, 1995; 2012). Classical morphological and biochemical methodologies Njoroge, 1998; Selmar et al., 2002; Selmar et al., 2006). Later, have identified Leuconostoc, Erwinia, and Klebsiella bacteria, and the findings about involvement of the metabolic and developmental yeasts Kloeckera, Candida, Saccharomyces bayanus, S. cerevisiae var. el- status of coffee beans, physiological conditions, germination, ex- lipsoideus, and Cryptococcus during fermentation in Hawaii, India, tent of stress, and microbial ecological variations were found to and Mexico (Agate & Bhat, 1966; Avallone, Guyot, Brillouet, Ol- play a major role in establishing the final cup quality of coffee guin, & Pierre, 2001; Frank, Lum, & Cruz, 1965). Masoud et al. (Selmar et al., 2015). So, in this regard, all the research studies (2004) identified Hanseniaspora uvarum, Pichia kluyveri, P. anomala, conducted in these contexts so far have been described in 2 sec- Kluyveromyces marxianus, Torulaspora delbrueckii, and Candida pseu- tions (that is, dry and wet processing). Any extra description is out dointermedia during wet coffee fermentation in Tanzania. Pichia of the context of this paper, as full, in-depth study chapters are fermentans (YC5.2) and Saccharomyces sp. (YC9.15) have also been also available (Kleinwchter, Bytof, & Selmar, 2015; Selmar et al., studied, and have potential for use as starter cultures for wet cof- 2015). fee fermentation (Pereira et al., 2014). Recently, phenotypic and Semi-dry/semi-wet processing. Semi-dry/wet processing (pul- genotypic characterization, followed by PCR-DGGR method- ped natural) is a transitional system of wet and dry processing to ologies, identified the same microbial diversity trend in wet pro- address the limitations of dry processing by utilizing a mechan- cessing too, that is, great diversity and populations of bacteria fol- ical flotation technique to separate under- or overripe cherries lowed by those of yeasts, but filamentous fungi were not detected from fully matured cherries (Wintgens, 2008). Usage of flotation even in samples from different geographical locations (Evangelista techniques enables the processor to process the 3 different kinds et al., 2015) until Feng et al. (2016) reported 50 isolates com- of cherry (underripe, mature, overripe) separately. In semi-dry prising 5 filamentous strains in wet processing, although it has processing, the exocarp and most of the mesocarp is generally re- been verified that microbial diversity and populations vary from moved in the depulping process (as for the wet process), and then region to region. Feng and coworkers also found 307 bacterial and beans are sun/air-dried until their moisture content reaches 11% 327 yeast isolates, as compared to a total of 184 to 251 microbial to 12% (Ribeiro et al., 2017). The remnants of the mesocarp are isolates reported by Evangelista et al. (2015), both of which are called mucilage. Mucilage remains clinging to the outside of the rather fewer than the number of isolates detected after dry pro- parchment during drying, or it may be removed immediately via cessing (Silva et al., 2000, 2008). The bacterial species detected by demucilagers. From a cup quality point of view, semi-dry process- Evangelista and coworkers were Klebsiella oxytoca, Enterococcus sp., ing with or without mucilage also produces different cup qualities. Ochrobactrum pseudogrignonense, Staphylococcus warneri, Chryseobac- The cup quality of semi-washed coffee is somewhat similar to that terium taichungense, C. bovis, Erwinia persicina, Paenibacillus amylolyti- of wet-processed coffee, having a bright, clean cup with a some- cus, Pseudomonas sp., Curtobacterium sp., Bacillus amyloliquefaciens, what less body than naturally processed coffee (Espresso & Coffee Escherichia hermannii, Enterobacter asburiae, E. ludwigii, Chryseobac- Guide, 2008; Stumptown, 2011). Semi-washed coffee is of supe- terium pallidum, Pantoea agglomerans, P. dispersa, Microbacterium sp., rior quality to naturals, with less input cost and a more constant and M. laevaniformans, Serratia marcescens, Weissella cibaria, Kocuria sp., cleaner flavor with less acidity, and which is potentially preferable and Leuconostoc mesenteroides. S. warneri (8.9 log CFU/g) was the for espresso coffee (DR Workefied Blog, 2013). The cup quality predominant bacterium during fermentation and drying, followed features of pulped naturals are also variable depending upon the by E. asburiae (7.4 log CFU/g), L. mesenteroides (6 CFU/g), and altitude of growth. Low-altitude-grown coffee has a resemblance E. persicina (5.5 CFU/g). Feng et al. (2016) reported Enterobacter to naturals, whereas high-altitude-grown coffee comes closer to species in the highest number followed by Pantoea sp. and Bacillus washed coffee (Wintegen, 2008). So, a grower may also find it ad- sp. The yeast population includes Hanseniaspora uvarum, Debary- vantageous to blend pulped naturals either with naturals or washed

r 1190 ComprehensiveReviewsinFoodScienceandFoodSafety Vol.17,2018 C 2018 Institute of Food Technologists® Farm to consumer . . . coffee. Moreover, pulped natural coffees are also strongly appreci- mination and stress-related metabolic processes (ICL expression, ated in blends for espresso coffee (Teixeira, Brando, Thomaziello, β-tubulin, and GABA accumulation status, dehydrins, change in & Teixiera, 1995). Much data are available in the literature (as 4C nuclei content, and so on) also need to be quantified in semi- discussed above) about the impact of wet or dry processing on dry processing. composition and final cup quality, while data about the influence In semi-dry coffee processing, cup quality also varies based of semi-dry processing are scarce. Duarte et al. (2010) conducted on the decision to eradicate or keep the mucilage clinging to a study to make a comprehensive comparison of the influence of the parchment. Mucilage is composed mainly of water (85% to wet and semi-wet processing on the major compositional primary 91%) and sugars (6.2% and 7.4%) of which 63% are reducing sug- and secondary metabolites of coffee. These metabolites are con- ars (Puerta-Quintero & R´ıos-Arias, 2011). Clifford and Willson sidered to be very important, because their degradation during (1985) demonstrated more detailed figures about its composition, roasting results in the formation of those volatile or nonvolatile and cited water (84.2%), protein (8.9%), reducing sugars (2.5%), compounds necessary for the development of characteristic coffee nonreducing sugars (1.6%), and pectin (1.0%) as the main com- bitterness, pigmentation, astringency, sweetness, aroma, body, and ponents. So, variation in semi-wet processing may bring about so on (Farah, 2005, Farah et al. 2006a; Farah, Monteiro, Calado, variation in fermentation pattern, fermentation products (organic Franca, & Trugo, 2006a; Toci& Farah, 2008). The distribution of 9 acids and other secondary metabolites), and microbial ecology, CGA classes (3-CQA, 4-CQA, 5-CQA, 3-FQA (feruloylquinic which ultimately produces coffee with varied cup quality. Addi- acid), 4-FQA, 5-FQA, 3,4-di-CQA, 3,5-di-CQA, and 4,5-di- tionally, the method of demucilization also has a strong impact CQA) is very similar between semi-wet and wet-processed coffee, on total cup quality. Recent studies have also claimed that demu- but coffee processed via the semi-wet method has, on average, cilization is a cost-effective way of producing high-quality cof- a statistically smaller amount of total CGA, except for di-CQA. fee with/without the need for fermentation and washing (FAO, So, the CGA content of semi-washed coffee lies between those 2010a, 2010b). Several methods are practiced in coffee growing re- of natural and washed coffee. The lower CGA content in semi- gions for this purpose. Important demucilization methods include washed coffee may be due to a shorter soaking time of seeds, natural fermentation, enzyme treatment, alkali washing, and di- which causes less loss of other water-soluble components by lix- rect machine washing (Rothfos, 1985). Natural fermentation is an iviation and fermentation (Teixeira et al., 1995; Wootton, 1974). old and widely used demucilization technique. Velmourougane Caffeine content remains unchanged for both processing methods, (2013) studied the impact of natural fermentation after depulp- but it is statistically insignificantly higher in semi-washed coffee ing on both Arabica and Robusta varieties, and claimed a better (1.12 ± 0.02 to 1.54 ± 0.02 g/100 g d.m.). Like CGA, trigonelline cup quality than for enzyme-treated, alkali-washed, and direct is found in considerably lower amounts in semi-washed Arabica machine-washed coffee. Comparatively, a high fermentation rate coffee (avg. 0.81 ± 0.02 g/100 g d.m.) than in washed coffee (avg. with a shorter time span (8 hr only), larger pH drop, and increase 0.90 ± 0.01 g/100 g d.m.). Sucrose content also shows the inter- in temperature is observed for Arabica varieties. The fermentative vention of other factors such as species, genetics, physiology, and mass of both varieties is predominantly yeast, followed by bacteria so on, its content varying with the processing method. Among and filamentous fungi. The yeast colonies identified are mainly of Arabica and hybrid samples, sucrose is found in greater amounts Saccharomyces sp. and Schizosaccharomyces sp., whereas Lactobacillus in semi-washed coffee (avg. 12.3 ± 0.07 g/100 g d.m.) than in sp., Bacillus sp., Pseudomonas sp., Leuconostoc sp., and Flavobacterium wet-processed coffee (9.0 ± 0.03 g/100 g d.m.). However, there sp. are the principal bacterial colonies. The majority of filamen- is no difference in the total sucrose content of semi-washed and tous fungi are Aspergillus sp., Fusarium sp., Penicillium sp., Mucor sp., washed Arabica samples (Duarte et al., 2010; Knopp et al., 2006). Rhizopus sp., and Cladosporium sp. (Velmourougane, 2013). Better Knopp and team also detected that levels of major sugars, that is, cup quality of naturally fermented coffee may be attributed to the sucrose and some low-molecular-weight sugars (stachyose, raffi- consumption of mucilage polysaccharides by the microbial con- nose, and rhamnose), remain unchanged during all methods of sortium and production of various volatile/nonvolatile compounds processing, whereas the levels of monosaccharaides (fructose and important for characteristic coffee aroma and taste (Masoud, Poll, glucose), arabinose, and mannose are between those of natural & Jakobsen, 2005, Masoud & Kaltoft, 2006; Velmourougane et al., and washed coffee. A long soaking time (ࣙ8 hr) and consequent 2008). Ameyu (2017) studied the combined effect of agricultural greater loss of sugars due to their high solubility, might be the rea- practices and postharvesting practices on the overall quality of cof- son behind the lower sucrose content in washed coffee (Wootton, fee. It is noteworthy that the mean raw coffee quality attributes 1974). The other possible reason for the lower sucrose content for semi-washed coffee (29.72%) are between those of washed in washed coffee is fermentation-enhanced glucose turnover due (30.33%) and dry-processed (32.33%) coffee. With respect to over- to anaerobic fermentation in the coffee endosperm (Knopp et al., all cup quality, statistically relatively low values are observed for 2006; Summers, Ratcliffe, & Jackson, 2000). Similarly, dry weight semi-dried coffee as compared to fully fermented wet-processed loss of coffee seeds is observed for all coffee processing methods, coffee (Ameyu, 2017). and it is more significant in pulped natural processing if coffee During semi-dry processing, most of the microbial ecology is seeds are dried on patios rather than in fermentation tanks (Selmar made up of yeasts (382 isolates) and bacteria (286 isolates), while et al., 2015). In addition to all these findings, research on semi-dry filamentous fungi (60 isolates) are in the minority, indicating unfa- processing has been given less attention due to its limited applica- vorable conditions for them during this process. The major bacte- tions in specific regions. Possible new themes in this context may rial species identified are Lactobacillus plantarum, Enterobacter agglom- include the influence of semi-dry processing on lipids, proteins, erans, Escherichia coli, Bacillus macerans, B. cereus, and B. megaterium. free amino acids, biogenic acids, and peptides. As it has been con- The yeast isolates identified are biochemically classified as Pichia cluded that these compositional factors play a partial role in the anomala, Hanseniaspora uvarum, Torulaspora delbrueckii, Rhodotorula organoleptic qualities of coffee, and coffee seed physiology and mucilaginosa, Saccharomyces bayanus, Kloeckera sp., Candida spp., and germination-related metabolic processes also have a major impact Arxula sp. Of the 60 isolates of filamentous fungi, the species in determining the final quality of coffee (Selmar et al., 2015), ger- most frequently found are Aspergillus tubingensis, A. versicolor,

r C 2018 Institute of Food Technologists® Vol.17,2018 ComprehensiveReviewsinFoodScienceandFoodSafety 1191 Farm to consumer . . .

Table 2–Advantages and disadvantages of different drying procedures.

Processing methods Advantages Disadvantages Dry  Simple process  Lengthy process (3-4 weeks)  Low cost  Over fermentation  Easy to perform and handle  Fungal/mold contamination  Less laborious  Ochratoxin A production  Higher drying rate  Higher rate of defected beans  both mature and immature cherries can be used  Less export market due to chances of foreign matter  higher level of (storage)proteins and hexoses  Less desirable quality attribuutes (Heavy body with (glucose and fructose) in brew sweetness and complex attributes)  Environmental friendly  Fetch less price Wet  Time saving  Expensive  Controlled fermentation  Laborious  Increase stachyose and sorbitol contents  Need extra arrangements  Higher contents of CGA and trigoneline  Only ripe/mature cherries needed  Higher organic acid production  Need extra arrangements  Attractive export market  Less environmental friendly  Less rate of defective beans  Less chances of OTA production  Desirable quality attributes Semi-Wet  Transition of dry and semi-wet system  Expensive  Under-ripe, mature and overripe cherries can use  Less practiced globally  Highest caffeine contents  Laborious  Intermediate level of CGA  Less environmental friendly  Lowest trigoneline contents  Cleaner flavor with less acidity and body

Cladosporium cladosporioides, Aspergillus sp., and Penicillium decum- and mucilage results in the production of various microbial sec- bens (Vilela, Pereira, Silva, Batista, & Schwan, 2010). The exis- ondary metabolites and aroma precursors which percolate into tence of fewer filamentous species is an indicator of minimized coffee beans and reappear after roasting and brewing in the form colonization of mycotoxins and OTA-producing fungi (Durand of volatile and nonvolatile compounds which have a major influ- et al., 2013; Vilela et al., 2010). Finally, a more generllized ence on sensory and organoleptic properties. Volatile compounds, comparison of 3 above mentioned secondary processing sections arguably the most important quality determinant, present in fer- (w.r.t advantages and disadvantages) has been presented here in mented and roasted coffee beans include hydrocarbons, alcohols, Table 2. aldehydes, ketones, carboxylic acids, esters, pyrazines, pyrroles, pyridines, furans, sulfur compounds, oxazoles, phenols, and fu- Fermentation and Starter Culture Technology ranones (Sunarharum, Williams, & Smyth, 2014). Furans are the Fermentation of coffee fruit is a complex process which has major volatile compounds quantitatively, contributing a big share the primary objective of removing the protective outer cover- of the aroma of roasted coffee (Akiyama et al., 2007). Furans are ings (pulp, mucilage, and skin) around the pair of seeds to make mostly formed from the degradation of sugars, acids, and unsat- them suitable for storage and transportation (Silva et al., 2000). urated fatty acids during roasting, and impart malty and sweet As discussed above, coffee is usually processed by 1 of 3 methods aromas (Akiyama et al., 2007; Burdock, 2010). Thiols are the (dry, wet, and semi-wet), and microbial growth dynamics have most vital sponsors of coffee flavor. Different thiols donate differ- also been discussed extensively in their respective sections. Yet the ent flavors, for example, 2-furfurylthiol, 2-methyl-3-furanthiol, 3- data about microbial growth dynamics is very descriptive and de- methyl-2-butene-1-thiol, and 2,4-dimethyl-5-ethylthiazole con- void of mechanisms of interaction. The relationship of microbial tribute roasted, and meaty flavors (Blank, Sen, & Grosch, 1992; growth to coffee processing, fermentation, and bean quality is also Ribeiro, Augusto, Salva, Thomaziello, & Ferreira, 2009; Semmel- not fully understood, which is a hindrance to attempts to con- roch & Grosch, 1995). Pyrazines are also the product of roasting, trol natural fermentation and fermentation using starter cultures. and they contribute towards roasted, nutty, green, and earthy aro- Microbial ecology, density, population, and diversity vary for each mas (Sunarharum et al., 2014). Furanones, products of the Mail- coffee processing methodology and have been presented in Table lard reaction and aldol condensation, are responsible for sweet and 3 with their illustrations in their sections too, so this section of the caramel aromas of roasted beans (Akiyama et al., 2007; Grosch, review will focus only on coffee fermentation control and coffee 2001). Phenolic compounds arise from the thermal degradation fermentation technology using starter cultures. of CGA and impart a spicy aroma in a quantity directly propor- Fermentation has an intricate and delicate relationship with the tional to the amount of CGA in seeds. Among nonvolatile com- aroma and organoleptic properties of coffee (Lee, Cheong, Cur- pounds, important classes include alkaloids, carboxylic acids, lipids, ran, Yu, & Liu, 2015). Coffee aroma and organoleptic properties proteins, carbohydrates, and polymeric polysaccharides, minerals, are regarded as the major factors affected by the fermentation pro- CGA, and melanoidins (Buffo & Cardell-Freire, 2004). Alkaloids cess which is carried out to remove mucilage. The organisms of the (caffeine and trigonelline) affect the strength, bitterness, and body microflora in fermentation mainly secrete the pectinolytic, cellu- of coffee. Similarly, the secondary metabolite CGA imparts as- lose, and polygalacturonase enzymes that attack the 1,4-glycosidic tringency and bitterness to coffee beverages (Oestreich, 2010). links of fully esterified (methoxylated) chains of pectin in mu- Acids are the product of microbial consumption of hexoses, and cilage. During fermentation, consumption of nutrient-rich pulp they contribute tartness or acidity to coffee beverages. Acids are

r 1192 ComprehensiveReviewsinFoodScienceandFoodSafety Vol.17,2018 C 2018 Institute of Food Technologists® Farm to consumer . . . ) Continued ( herbaceous, chocolate, astringent, acidic Bitter, caramel, fruity, sp., A. steynii, A. F. semitectum, Arthrobacter F. semitectum, sp. sp. sp. spp., B. subtilis, B. macerans, Brochothrix, Arthrobacter, B. megaterium, B. subtilis, sp., sp., E. sakazakii, E. aerogenes, Kurthia, E. gergoviae, Yersinia sp., spp., B. laterosporus, B. laterosporus, Microbacterium P. fellutanum, P. expansum, P. janthinelum, F. illudens, F. xylarioides, sp., spp., sp., K. ozaenae, P. vesicularis, Chromobacter violaceum,Salmonella choleraesuis, sp., Enterobacter agglomerans, Brochothrix spp., P. funiculosum, Paecelomyces sp., sp. Dermabacter P. brevicompactum, P. crustosum, P.westerdijkiae, roqueforti, A. P. lactofeatus, citrinum, and Cladosporium A. cladosporioides,P.viridicatum, sclerotioniger, Paecelomyces P.citrinum, A.flavus, P. P.brevicompactum, waksmanii, Penicillium,P.brevicompactum, P. crustosum,solitum, F. P. lateritium, brevicompactum, A. P. dimorphicus, corylophilum, F.Paecelomyces P. concolor, chrysogenum, P. F. roqueforti, illudens, P. F. citrinum,F. moniliforme, P. stilboides, F. P.crustosum, nivale, A. Pestalotia ochraceus, F. xylarioides, F. trincictum, P. aurantiogriseum amylolyticus, Enterobacter agglomerans, Bacillus cereus,B. Arthrobacter, megaterium, Yersinia B. polymyxa., Enterobacter cloacae, Acinetobacter spp., glaucus group, Cladosporium herbarum, FusariumAspergillus stilboides, flavus, Penicillium A. chrysogenum, fumigatus, P. A. viridicatum,C. niger, P. herbarum, A. islandicum, C. parasiticus, macrocarpum, A. Fusarium sydowii,citrinum, oxysporum, A. P. Penicillium tamarii, cyclopium, albidum, A. P. P. terreus, funiculosum, brevicompactum, Cladosporiumgranulatum, P. P. cladosporioides, Mucor, jensenii, chrysogenum, Cercospora, Aspergillus P. Phoma, fumigatus, Penicillium, A.P. A. niger, brevicompactum, carbonarius, Fusarium P. A. graminearum, crustosum, niger, Penicillium P. A. roqueforti, ochraceus, P. Penicillium citrinum, Cladosporium cladosporioides, Paecelomyces ofunaensis,P. sydowiorum, P. acaciae, P.fermentans,Blastobotrys anomala, proliferans, P. C. ciferii, glucosophila, P. Saccharomycopsis jadinii,C. fermentans, P. vartiovaarae, C. lynferdii, Citeromyces incommunis, P. matritensis, C. sydowiorium, C. paludigena, Geotrichum Sporopachydermia schatarii, cereana, Schizosaccharomyces S. pombe, fibuligera, A. D. adeninivorans.,oedocephales, polymorphus,Williopsis D. Pichia saturnus polymorphus, guilliermondii, var. P. sargentensis, guilliermondii, Trichosporonoides S. P. smithiae, subpelliculosa, Candida Debaryomyces saitoana, hansenii, C. P.smithiae, fermentati, burtonii, Arxula P. P. adeninivorans. Burtonii, anomala, C. membranifaciens, Saccharomyces cerevisiae, Stephanoascus P. pseudoalcaligenes, Pseudomonad paucimobilis, P.haemolytica, fluorescens, S. Serratia marcescens, liquefaciens, Cedecea S. plymuthica,P. Pasteurella aeruginosa,P. cepacia, Citrobacter freundi,oxytoca, Hafnia S. alvei, paratyphi, Flavobacterium Tatumella odoratum, ptyseos, S.stearothermophilus, Bacillus enterica B. subtilis, var. anthracis, B. arizonae, Cellulomonas cereus, Klebsiella Shigella dysenteriae, B. megaterium, B. Bacillus cereus, B. polymyxa, Acinetobacter Lactobacillus Cladosporium cladosporioides, Alternaria, Aspergillus flavus, A. niger, A. tamarii, A. ochraceus, A. wentii, A. versicolor, A. Arxula adeninivorans. A. adeninivorans. Candida auringiensis, Arxula adeninivorans, Saccharomyces cerevisiae, Pichia Aeromonas, Serratia rubidea, S. plymutica, Citrobacter, Providencia mirabilis, Chromobacterium, Cedecea, Paenibacillus Bacteria Lactic acid bacteria Filamentous Fungi Table 3–Detected microbial ecology during the dry, wet, and semi-wetProcessing processing method methods of coffee cherries. Dry Microbial isolates Yeasts Genus/species/strains Sensory attributes

r C 2018 Institute of Food Technologists® Vol.17,2018 ComprehensiveReviewsinFoodScienceandFoodSafety 1193 Farm to consumer . . . chocolate, fermented, buttery caramel notes with musty, spicy, bitter, fermented, mushrooms, citric, herbaceous like flavor. Bitter, astringent Frutiy, acidic, caramel, Distinct fruity, buttery and Pichia spp., sp. sp., sp., Shizosaccharomyces sp., Candida pseudointermedia, and Penicillium decumbens Kloeckera spp. sp. sp., B. cereus, B.macerans, B. megaterium, sp., Hanseniaspora uvarum, Issatchenkia Erwinia dissolvens, Paracolobactrum sp., sp. Proteus thodology. (Adapted from Durand et al., 2013; Evangelista et al., 2015; Feng et al., 2016; Gonzalez-Rios et al., 2007a, 2007b; Lactobacillus plantarum, L. brevis, Leuconostoc sp., 17; Silva, 2014a, b, c; Silva et al., 2008; Taveira et al., 2015; Silva et al., 2000; Velmourougane, 2013; Velmourougane, 2011; Vilela sp., , S. bayanus, S. cerevisiae, Rhodotorula mucilaginosa, C. lactis sp., sp. C. quercitrusa, Cryptococcus laurentii, C. albidus, Kloeckera apis apicuata, Flavobacterium sp. sp., Weissella soli ıcola, Morganella morganii, Klebsiella oxytoca, Erwinia toletana, Pantoea eucrina, ´ sp., Pseudomonas sp., Leuconostoc pseudomesenteroi, Lactobacillus plantarum, Enterococcus faecalis, Weissella confusa, W. Pseudomonas aeruginosa, B.subtilis, Pseudomonas sp., Enterobacter agglomerans, Acinetobacter sp., Klebsiella pneumoniae, Erwinia tasmaniensis, Escherichia coli, Serratia sp., sp., S. cerevisiae (UFLA YCN727) Candida ernobii, S. cerevisiae (UFLA YCN724), P. caribbica, Kluyveromyces Erwinia herbicola, Flavobacterium Mitchella repens, Trichosporon cavern Pantoea S. bayanus, S. cerevisiae var. ellipsoideus, S. cerevisiae var. ellipsoideus luteola.Streptococcus Hanseniaspora opuntiae, Saccharomyces Pichia kluyveri, P. Anomala, Candidaorientalis, guilliermondii, Saccharomyces Schizosaccharomyces marxianus (Kluyveromyces marxianus),Saccharomyces Kluyveromyces marxianus marxianus, (Kluyveromyces Eremothecium marxianus), coryli, S. bayanus, Schizosaccharomyces sp., anomala, C. membranifaciens, Saccharomyces carpophila,Torulaspora delbrueckii, C. fukuyamaensis, Hanseniaspora uvarum,Arxula aerogenoides, P. coliforme, P. intermedium, Escherichia intermedium. thailandensi, L. citreum, Lactococcus lactis sub Yarrowia lipolytica, Rhizopus oligosporus Aspergillus tubingensis, Aspergillus versicolor, Cladosporium cladosporioides, Aspergillus Enterococcus Leuconostoc mesenteroides, Lactococcus lactis, Lactobacillus C. parapsilosis, C. parapsilosis, C. parapsilosis, P. guilliermondii (UFLA YCN731), Saccharomyces P. fermentans, P. kluyveri, C. quercitrusa, P. guilliermondii, P. caribbica, Kluyveromyces, Pichia fermentans, Candida glabrata, Klebsiella pneumoniae, K. ozaenae, K. oxytoca, Enterobacter herbicola, Pseudomonas cepaciae, Chrysomonas Bacillus Bacteria Lactic acid bacteria (Filamentous)Fungi Bacteria Lactic acid bacteria (Filamentous)Fungi : The data are compiled from earlier publications. Most of authors used all 3 levels of roasting and brewing was stated to perform under standardized me Table 3–Continued. Processing methodWet Microbial isolates Yeasts Semi-wet Yeasts Genus/species/strainsNote Hamdouche et al., 2016; Leeet et al., al., 2016a, 2010.) 2016b; Lee et al., 2017a, 2017b; Masoud et al., 2004; Neto et al., 2017; Pereira et al., 2014; Ribeiro et al., 20 Sensory attributes

r 1194 ComprehensiveReviewsinFoodScienceandFoodSafety Vol.17,2018 C 2018 Institute of Food Technologists® Farm to consumer . . . also involved in reacting with chlorogenic and quinic acids dur- to check their antifungal properties against OTA-producing As- ing roasting, to form the lactones that can also influence coffee pergillus strains. Use of a starter culture composed of LAB and yeasts aroma (Sunarharum et al., 2014). (Figure 3A and B shows the showed a strong contribution towards the suppression/inhibition mechanisms of locatone and their precusors formations during of other microorganisms, including OTA-producing filamentous roasting in details below). Carbohydrates and polymeric polysac- fungi (Massawe & Lifa, 2010). The anaerobic, acidic and alcoholic charides are responsible for coffee sweetness and trapping volatile environmental conditions are created by yeasts and LAB, due to compounds, respectively, which, in turn, bring beverage viscosity production of organic acids, alcohols, hydrogen peroxide, cer- up to the desired level (Oestreich, 2010). Lipids are the largest tain volatile compounds, and diacetyl (Mbugua & Njenga, 1991; component of coffee beans and are important for formation of Wood & Hodge, 1985), which lead to overall unfavorable condi- the crema/cream emulsions carrying fat-soluble flavors and vita- tions for mycotoxin-producing filamentous fungi. However, the mins in espresso coffee. Amino acids play an important role in authors did not incorporate sensory analysis or the volatile and the formation of nitrogen/sulfur heterocyclic compounds called nonvolatile compound profiles of beans fermented by this yeast melanoidins, via the Maillard reaction or caramelization, which and LAB starter culture. Moreover, the ecology and diversity of are thought to be important for coffee flavor (Shibamoto, 1983). fermentative microflora is also influenced by the environmental Minerals are important catalysts for the various biochemical reac- conditions of fermentation, and 6 factors are considered crucial tions responsible for the generation of various aroma and flavor for the desirable attributes of fermentation. These 6 attributes compounds (Oestreich, 2010). After describing the major volatile include water activity (Aw), sugar concentration (measured in de- and nonvolatile compounds and their contribution to sensory and grees Brix), availability of oxygen, acidity, temperature, and time organoleptic attributes, it is also worth noting that, besides fer- (Poltronieri & Rossi, 2016). To control natural fermentation, it mentation, levels of these compounds are also influenced by coffee is imperative to understand and control microbial ecology and variety, genetics, agricultural practice with geographical location, diversity with respect to these environmental factors, and their climatic conditions, and variations in processing technology and combined relationship, in the final determination of coffee qual- microbial ecology. However, the extent of fermentation is the ity. In addition, elevated moisture levels in beans with uncontrolled main driving force for determining the levels of these volatile fermentation/overpopulation of microflora are the cause of ethyl and nonvolatile compounds in beverages. Unfortunately, studies esters of short-chain fatty acids (overproduction of acetic, butyric, published so far on coffee fermentation have agreed upon the un- and propionic acids) and short-chain acid production considered controllability/overfermentation of coffee cherries which results responsible for overfermented, acidic, and musty flavor defects in the production of black and foul-smelling beans with alco- (Flament, 2002). holic, floral, and sour flavor defects (Jackels & Jackels, 2005a). Generally, broad microbial diversity is observed in coffee fer- Prevention of overfermentation largely depends upon the identi- mentation, and most indigenous microbes are not considered nec- fication of a predefined fermentation end-point. The end-point essary for high-quality coffee. Further, this microbial ecology and of fermentation is very subjective, and it was analyzed for the diversity is also coffee process-specific, so, in recent studies of each first time by Jackels and Jackels (2005a) by creating a gap between processing technique, only small ranges of aromatic microbes have the fermentation process and fermentation mass. In this study, been used as starter cultures in coffee fermentation. For ease of under-fermentation was characterized by incomplete degradation readers’ understanding, we have discussed starter culture technique of mucilage, which could successively encourage microbial growth in 3 portions according to coffee processing technique. and vice versa. In addition, the end-point of fermentation was es- In dry processing, fermentation is characterized by aerobic con- tablished on the basis of measuring the pH of the fermenting mass; ditions, with a high concentration of sucrose, glucose, and fructose a pH range of 4 to 5.5 is a suitable indictor of the end-point of substrates for microflora (Knopp, Bytof, & Selmar, 2005). This coffee fermentation (Jackels & Jackels, 2005b). This pH range and availability of high substrate levels in drying processing offers a its impact on coffee quality attributes were further investigated kind of opportunistic access for most epiphytic yeasts and bacteria and it was found that the lower end-point of pH leads to a drop with pectinolytic and cellulolytic enzymes which produce distinct in the cup quality due to overfermentation (Jackels et al., 2006). aromatic microbial metabolites specific to dry processing. So, most The FAO (2006) has also declared pH, time, and acidity as cru- authors selected and used those kinds of LAB or yeast strains that cial parameters in coffee fermentation. Besides pH, starter culture are not only indigenous to the process but also to the variety being technology and the addition of pectinolytic enzymes to the fer- used. Similarly, Evangelista et al. (2014a, 2014b) used indigenous mentation mass can also be used for better control of fermentation baker’s yeast (Saccharomyces cerevisiae UFLA YCN727 and S. cere- and, additionally, OTA-producing fungal strains (Massawe & Lifa, visiae UFLA YCN724) and bacteria (Candida parapsilosis UFLA 2010). The simultaneous usage of cellulase enzymes with suit- YCN448 and Pichia guilliermondii UFLA YCN731) as a starter cul- able starter cultures results not only in a reduction in fermentation ture for both washed and nonwashed coffee processing. This mixed time (30 min) and OTA-producing fungal strains but also increases starter culture positively influenced the profile of volatile and non- the level of Maillard and caramelization reaction substrates (that volatile compounds in roasted beans. After roasting, 48 volatile is, reducing sugars), contributing to improved sweetness in taste, aromatic compounds (17 alcohols, 11 esters, 1 hydrocarbon, 3 ke- with caramel and slight acidic notes (Lin, 2010; Massawe & Lifa, tones, 5 furans, 5 aldehydes, 5 acids, and 1 phenol) were detected 2010; Velmourougane, 2011). A comprehensive study in this re- both in washed and nonwashed coffee beans, of which 18 com- spect was performed by Massawe and Lifa (2010), who selected pounds were found to significantly differ. Nonwashed coffee was and used 6 strains of yeast (that is, 3 strains of Pichia anomala: P.aS12, found to have a higher amount of these compounds, which include P.aS14, and P.aS16, and 3 strains of P. k l u y v e r i : PS7Y1, PkS4Y3, hexanal, 2,3-butanedione, linalool, propionic acid, 3-methyl- and PkS13Y4) and LAB (that is, Leuconostoc/Weissella BFE 6997 butanol, hexanoic acid, and so on. Most of these compounds and 6989, homofermentative Lactobacillus BFE 6853 and 6823, are involved in flavor defect notes (that is, overfermentation, heterofermentative Lactobacillus BFE 6812 and Enterococcus BFE insect damage, and high acidity). Similarly, nonwashed coffee also 6803) as a starter culture in the fermentation of wet processing, showed a higher content of acetic acid (10.52 g/kg in pulp +

r C 2018 Institute of Food Technologists® Vol.17,2018 ComprehensiveReviewsinFoodScienceandFoodSafety 1195 Farm to consumer . . . seed; seed 2.07 g/kg) when inoculated with P.guilliermondii UFLA The wet method of processing is broadly employed in Colom- YCN73. In sensory analysis, all the washed treatments with se- bia, Hawaii, and Central America for Arabica coffee. The mature lected yeasts showed a greater sensation of caramel, fruity, and fruit cherries are mechanically depulped and then subjected to acidic notes than the controls, whereas nonwashed coffee sam- spontaneous tank fermentation consisting of a more complex mi- ples exhibited chocolate, acidic, and bitter notes. So, depending crobiological process for 24 to 48 hr until the moisture content is upon the processing method, inoculated yeasts behave differently reduced to 10% to 11% (Vilela et al., 2010). Microbial ecological in both green and roasted coffee. studies have shown that the frequently occurring yeasts and bacte- Semi-dry processing, as mentioned above, involves the mechan- ria are Saccharomyces cerevisiae, Pichia kluyveri, Debaryomyces hansenii, ical depulping of coffee fruit followed by natural drying (for 12 P. anomala, Hanseniaspora uvarum, and Torulaspora delbrueckii (Silva to 15 days) on some selected setup, during which natural fer- et al., 2008; Vilela et al., 2010), and Erwinia, Klebsiella, Aerobac- mentation occurs. The microbiota associated with semi-dry pro- ter, Escherichia, and Bacillus, respectively (Silva et al., 2012). Since cessing (Vilela et al., 2010) has also been discussed earlier. Af- the beginning of starter culture technology, Pereira et al. (2014) ter this initial work, Evangelista et al. (2014a, 2014b) studied were the first to use a targeted approach for the selection of yeast the inoculation of 4 yeast strains (S. cerevisiae UFLA YCN727, species with stress tolerance and the production of desirable aro- S. cerevisiae UFLA YCN724, Candida parapsilosis UFLA YCN448 matic compounds in wet processing. The wet process-specific mi- and Pichia guilliermondii UFLA YCN731) in Acaia (Coffea arabica crobial isolates detected were selected on the basis of their survival L. var. Acaia). The starter cultures S. cerevisiae UFLA YCN724, S. in the hostile environment and stress conditions (that is, acidity cerevisiae UFLA YCN727, C. parapsilosis UFLA YCN448, and P. tolerance, heat tolerance, growth capacity, polygalacturonase ac- guilliermondii UFLA YCN731 do not favor the production of un- tivity, and metabolite accumulation tolerance) prevailing during desirable acids such as butyric and propionic acids, while lactic and wet process fermentation. Of a total of 144 isolates, only 9 mi- succinic acids are the predominant organic acids. But, strangely, crobial strains (Saccharomyces sp. YC9.15, YC9.13, and YC.8.10, the authors not only found a higher population of bacteria than of Pichia fermentans YC5.2 and YC8.8, P. k l u y v e r i YH7.16, P. g u i l - yeasts but also identified microbial isolates of yeasts other than those liermondii (YC1.2), Hanseniaspora opuntiae (YC1.4), and Candida in the starter culture: Trichosporon cavernicola, Debaryomyces hansenii, glabrata YH1.5) were found to cope with this situation and produce and Cystofilobasidium ferigula. After fermentation and roasting, out distinct aromatic compounds. These strains were verified by grow- of 47 aromatic volatile compounds identified, 23 were from green ing on a pulp-stimulated medium where they produced 14 volatile coffee. The volatile aromatic compounds were 3 ketones, 15 al- compounds (acetaldehyde, ethyl acetate, benzaldehyde, ethyl lau- cohols, 5 aldehydes, 7 acids, 10 esters, 5 furans, 1 phenol and 1 rate, isoamyl acetate, caprylic acid, ethanol, 2,3-butanedione, diether. Green coffee was found to be rich in aldehydes, esters, 1-decanol, 3-methyl-1-butanol, 2-hexanol, 2-octanol, 2-methyl- and alcohols, whereas furans, acids, and alcohols are abundant in 1-butanol, and 1-octanol) responsible for the characteristic fruity roasted beans. Sensory analysis revealed that all 4 strains of this flavor of washed coffee (Czerny & Grosch, 2000; Evangelista et al., coffee variety produce a caramel sensation in the first instance, 2014a, 2014b; Gonzalez-Rios et al., 2007a, b; Sanz, Maeztu, Za- followed by bitterness (Evangelista et al., 2014a, 2014b). Notably, pelena, Bello, & Cid, 2002). P. fermentans YC5.2 and P. k l u y v e r i coffee inoculated with UFLA YCN727 and UFLA YCN448 had YC7.16 produce considerable amounts of isoamyl acetate and ethyl a cup score > 80 and can be declared as a specialty coffee, whereas acetate which are responsible for pineapple and banana-like fruity the coffee made with the other 2 strains (UFLA YCN724 and flavors (Pereira et al., 2014), whereas Saccharomyces strains produce UFLA YCN731) scored < 80 and so can be used in a blend with a greater amount of alcohols than any other strain, followed by specialty coffees. The strains UFLA YCN448 and UFLA YCN727 acetaldehyde, which account for floral flavors (Pereira et al., 2014). have also been used previously in natural coffee fermentation, The highest isoamyl acetate and ethyl acetate concentrations (59.5 where they influence the coffee with a different taste. Coffee in- and 171.8 μmol/L, respectively) were determined in a pure culture oculated with UFLA YCN448 produces caramel and herbaceous of P. fermentans at 28 °C, whereas the production of acetaldehyde flavors, whereas coffee inoculated with UFLA YCN727 has a and ethanol was greater in mixed fermentations at 28 and 37 °C, fruity flavor (Evangelista et al., 2013). The same strains produce respectively. Inoculation of these verified strains also produces the caramel and bitter flavors in semi-washed coffee (Evangelista et al., expected volatile aromatic compounds (that is, ethanol, ethyl ac- 2014a). In another recent work, Ribeiro et al. (2017) used 2 strains etate, 2,3-butanedione, acetaldehyde, isoamyl acetate and hexanal) of baker’s yeast (S. cerevisiae CCMA 0200 and CCMA 0543) to in washed coffee with their desirable intense fruity, buttery, and study the effects of inoculation on the chemical composition of fermented aroma (Pereira et al., 2014). After the screening and beans of 2 Arabica varieties, namely Ouro Amarelo (OA) and inoculation of selected yeast strains, Pereira and colleagues also Mundo Novo (MN), processed using the semi-dry method. OA carried out a similar kind of study on LAB in wet processing had higher final attributes with CCMA 0543 while, unfortunately, (Pereira et al., 2014, 2015a, 2015b). However, screening and se- MN failed to produce higher quality attributes with both strains. lection were merely based on the lactic acid production potential of A higher population of CCMA 0543 in the later stage also showed various strains, as lactic acid has been found to endow coffee with a its suitability for these coffee varieties. OA coffee inoculated with pleasant acidic body as compared to the oniony and foul flavors due CCMA 0543 also contained more sugars, and organic acids except to other acids, that is, butyric, acetic, and propionic acids (Pereira for maleic acid. Variety RO was found to be positively influenced et al., 2015c; Silva et al., 2013). Based on these criteria, pulped by the starter culture technology, since it showed better results in coffee was inoculated with the selected LAB strain Lactobacillus sensory analysis than controls and counter-species. Ribeiro and plantarum LPBR01 and compared with a spontaneously fermented coworkers also detected 96 volatile compounds, of which 54 were control group. It is noteworthy that this strain was found to reduce detected in roasted coffee. The predominant compounds found fermentation time to 12 hr, with efficient conversion of hexoses were acids, hydrocarbons, alcohols, furaldehydes, aldehydes, ke- to lactic and fumaric acids. This strain also enhances the volatile tones, esters, pyrans, pyrroles, furans, triazoles, thiophenes, and compounds bestowing a fresh fruity flavor (that is, butyl acetate, sulfur compounds. ethyl acetate, ethyl hexanoate, 1-octanol, ethyl succinate, isoamyl

r 1196 ComprehensiveReviewsinFoodScienceandFoodSafety Vol.17,2018 C 2018 Institute of Food Technologists® Farm to consumer . . . acetate, and ethyl octanoate). From a sensory point of view, the which also influence the coffee aroma profile. With the initiation inoculated coffee can be declared as a specialty coffee, with overall of starter culture technology, as discussed above, many authors scores of 88.3 for aroma, body, acidity, aftertaste, balance, and over- have not only tried to address this problem, but have also used all quality (Pereira et al., 2014). Amid controlling the fermentation and characterized the yeast and bacterial starter cultures being process, the same team (Pereira et al., 2015a, 2015b) moved a step used for the production of distinct and peculiar flavor and aroma forward and chose a dominant yeast, that is, P. fermentans YC5.2 notes. After this, the concept of biotransformation of green beans with peculiar fermentative functional properties from the sponta- (substrate) to volatile and nonvolatile aromatic compounds, by neous wet process and inoculated pulped Arabica (Catui) cherries. using microfungi in solid state fermentation (SSF), was introduced The effect of excessive carbon source on cup quality was also as- by Li, Li, and Li (2010). The authors used Antrodia camphorata sessed by supplementing one of the inoculated fermentation group and Hericium erinaceum on green coffee beans (SSF), but their with 2% glucose. The inoculated yeast lowered the microbial di- reported work was devoid of qualitative and quantitative data versity both in supplemented and unsupplemented groups while about improvements/changes in the coffee aroma profile. Later, maintaining its population at over 80%. After fermentation, the this concept was further cross-checked by Lee et al. (2015), 2016a, major volatile compounds included ethanol, acetaldehyde, ethyl 2016b) by using the mold Rhizopus oligosporus; they identified acetate, isoamyl acetate, and 2-heptanone, with greater amounts that fungal or mold fermentation of other coffee matrices for inoculated (both supplemented and unsupplemented) treat- produces novel aroma profiles. R. oligosporus has been widely ments, while propyl acetate, ethyl hexanoate, n-butyl acetate, and used in fermented foods due to its extracellular proteins and n-butanol were not detected in these treatments (Pereira et al., polysaccharide-degrading enzymes (De Reu, Linssen, Rombouts, 2015a, 2015b). In roasted coffee beans, benzaldehyde, isobutyl & Nout, 1997). So, hydrolysis of macromolecules coupled with acetate, 2-pentanone, acetaldehyde, ethyl acetate, isoamyl acetate, its metabolism inevitably brings about changes in the biochemical and 2-octanone were found at higher levels in inoculated coffee composition of green coffee and its aroma precursors. Lee, beans, while propyl acetate, ethyl hexanoate, and n-butyl acetate Cheong, Curran, Yu, and Liu (2016a) studied and compared were not detected. However, the authors’ work is devoid of data the influence of R. oligosporus on the volatile and nonvolatile about the influence of the subject strains on the detailed profile compounds of both fermented and unfermented green coffee. of nonvolatile compounds after fermentation and roasting. From In volatile compounds, the major classes of compounds detected a cup quality point of view, this yeast strain also acts positively included alcohols, furans, phenols, acids, pyrroles, carbonyls, and peculiarly in developing special flavor notes. All the coffees esters, pyrazines, and terpenes. Among these, compounds such inoculated with this strain scored more than 80 and were declared as 2,3-butanediol, hexanal, 1-octen-3-ol, 2-pentyl-furan, and specialty coffees. The inoculated coffee supplemented with sugars nonanal were the only result of R. oligosporus fermentation (Lee had a vanilla taste with intense lemon and floral aroma notes. On et al., 2016a). Furthermore, the levels of buttery and fruity note- the other hand, inoculated coffee without sugar supplementation imparting volatile compounds (2,3-butanediol, 2-phenylethyl exhibited a caramel-like taste with lactic, citric, and phosphoric alcohol, guaiacol, p-vinylguaiacol, ethyl palmitate, and ethyl acid perception (Pereira et al., 2015a, 2015b, 2015c). Regarding 3-hydroxybutanoate) rose substantially in R. oligosporus-fermented fermentation, Silva and colleagues (2014a, 2014b, 2014c) con- green coffee beans. Among nonvolatile compounds, the level of cluded that coffee microbiome and altitude are correlated, and a sugars decreased enormously after fermentation, which in turn change in altitude brings about changes in the coffee fermenta- caused a rise in lactic acid concentration. However, the level of tion microbiome too. Variations in the coffee microbiome in turn acid also depends upon the method of drying (Steiman, 2013). are reflected in fermentation temperature and pH profile which Besides lactic acid, R. oligosporus was also found to increase the create uniqueness in the flavor notes of coffee from different alti- quantities of fumaric acid, quinic acid, and caffeic acid, whereas tudes. Recently, this hypothesis was cross-checked by Neto et al. the levels of CGA, ferulic acid, and maleic acid decreased. (2017) who wet-processed Arabica coffee from a large specialty Among proteins and amino acids, R. oligosporus induces a higher coffee-producing region in Brazil. The microbial load detected concentration of amino acids, especially glutamic acid, aspartic was found to be the same as that described in earlier publications acid, alanine, proline, phenylalanine, and ammonia (Lee et al., (Masoud et al., 2004; Silva et al., 2008; Velmourougane, 2013), 2016a). As roasting is the key process during which hundreds except for the identification of a new strain of yeast, Hansenias- of aroma compounds are generated from their precursors in pora vineae. H. vineae had previously been identified as increasing fermented green beans via a series of complex reaction pathways the fruity aroma in wines due to its good production of acetate (Cheong et al., 2013), Lee, Cheong, Curran, Yu, and Liu esters. So, the presence of H. vineae is the main reason for ex- (2016b) also checked the influence of roasting R. oligosporus- tra fruit notes in coffee and could potentially be used in future fermented green beans on the modulation of aroma profile. starter culture studies. The volatile compounds detected in the The effect of R. oligosporus fermentation was more obvious fermented beans included hydrocarbons and alcohols originating in light-roasted green beans than in medium- or dark-roasted either from beans or the fermentation process (Neto et al., 2017). beans, so we will focus only on light-roasted fermented beans Among the nonvolatile compounds in fermented beans, the au- in our descriptions. In nonfermented green beans, 30 common thors only mentioned acids, with lactic and acetic acids dominant. compounds, both of thermal and nonthermal or processing So, this paper lacks a detailed profile of volatile and nonvolatile origin, were detected, among which the compounds of thermal compounds both in fermented and roasted beans, but it reports for origin (p-vinylguaiacol, furfural, furfuryl alcohol, guaiacol, the first time the mineral profile of fermented and roasted beans. 2-formylpyrrole, and γ -butyrolactone) increased, while com- The profile of major minerals was found to be in the descending pounds of processing origin (1-octen-3-ol, 2,3-butanediol and order of potassium (K) > phosphorus (P) > magnesium (Mg) > 2-methoxy-3-isobutylpyrazine) decreased or were not detected. calcium (Ca) (Neto et al., 2017). The levels of pyrazines, furans, phenols, alkanes, acids, and terpenes Traditional fermentation processing of coffee fruit is full of decreased, whereas the level of carbonyls, esters, and pyridines microbial diversity, with uncontrollability and irregularity issues increased with roasting degree severity (Lee et al., 2016b). In

r C 2018 Institute of Food Technologists® Vol.17,2018 ComprehensiveReviewsinFoodScienceandFoodSafety 1197 Farm to consumer . . .

R. oligosporus-fermented coffee beans, 21 carbonyls, 4 acids, 27 volatile phenols (4-vinylguaiacol and 4-vinylphenol) and acid furans, 2 alcohols, 2 alkanes, 5 esters, 5 furanones, 6 pyridines, metabolism. Variations in the impact of yeast and mold fermenta- 7 phenols, 11 pyrroles, 26 pyrazines, 11 sulfur-containing com- tions may arise from their different sets of fermentative pathways pounds, and 4 terpenes were detected, with changes to volatile which also modulate the aroma precursor and volatile compound compounds in respective classes (Lee et al., 2016b). The levels of profile after fermentation and roasting. pyrazines, pyrazine-derived amino acids (phenylalanine, aspartic Besides SSF, the other recently published novel ways of modu- acid, and glutamic acid), acids (especially malic and succinic acids), lating coffee aroma are digestive bioprocessing and monsooning. phenols, methyl, and ethyl palmitate, which impart fruity and In digestive bioprocessing, mature coffee cherries are consumed nutty flavor notes to coffee, increased in R. oligosporus-fermented by civet cats (Indonesia) or elephants, and beans are collected from and light-roasted coffee (Kesen, Kelebek, & Delli, 2014). The the dung of these animals and brewed to produce the world’s most levels of hexoses and caramel-endowing furanones (4-hydroxy- expensive specialty coffees called Kopi Luwak and Black Ivory, 2,5-dimethyl-3-furanone, dihydro-2-methyl-3(2H)-furanone, respectively (Main, 2014; Marcon, 2004). In digestive bioprocess- 2,5-dimethyl-3(2H)-furanone, 3,4-dimethyl-2,5-furandione, and ing, the coffee beans pass through a series of unique acidic/gastric, 2-ethyl-3-methyl maleic anhydride) decreased in fermented proteolytic, and fermentation processes while passing through the coffee beans (Lee et al., 2016b). So, fermentation with R. gastrointestinal tract of the animal, which brings about unique oligosporus undoubtedly imparts fruity and sweet attributes to changes in amino acid composition. These characteristic changes light-roasted fermented coffee, and roasted, spicy and sweeter in amino acid composition, in turn, bring about changes in aroma notes to fermented dark-roasted coffee. precursors on roasting. Protein hydrolysis is also a cause of a small Recently, the same authors (Lee et al., 2017a, 2017b) also degree of bitterness in the final brew, with less body and acidity used the proteolytic and lipolytic “dairy yeast” Yarrowia lipolyt- overall (Marcon, 2004). Research work is needed in this regard ica on green coffee beans to generate novel aromas. Compari- to determine the influence of intestinal microflora on final cup son of volatile compounds before and after fermentation revealed quality. Moreover, with advancements in medical studies, it is also a drop in volatile acids (hexanoic and nonanoic acids), alka- possible to study the influence of the intestinal microflora starter nes (dodecane, tridecane), and aldehydes (3-methylbutanal and culture technology in a pseudo-gastrointestinal tract by selecting 2-phenylethanal), whereas total alcohols (2-phenylethanol), ke- specific bacterial species. It will definitely help us to understand the tones (gamma-butyrolactone), volatile phenols (4-ethylguaiacol, role of these microorganisms in generating unique aroma precur- 4-vinylguaiacol, and 4-vinylphenol), and gamma-terpinene in- sors, and add to progress towards lower-cost production of these creased. Among nonvolatile compounds, the levels of sucrose, specialty coffees. Monsooning is another form of fermentation, citric acid, succinic acid, proteins, amino acids, and nitrogenous in which fermentation usually occurs during transportation in the compounds were significantly reduced, while the glucose level rainy season and is responsible for the generation of specific fla- rose after fermentation. These featured changes in volatile and vor notes (Ahmed et al., 2003). High bacterial (108 CFU/g) and nonvolatile compounds are due to the unique de novo biosynthesis fungal (105 CFU/g) populations are considered responsible for the and hydrophobic substrate-metabolizing pathways of Y. lipolytica generation of these specific flavor notes. However, the mechanisms (Celinska, Kubiak, Bialas, Dziadas, & Grajek, 2013; Fickers et al., underlying production of these specific aroma precursors are still 2005). These Y. lipolytica-fermented coffee beans were further sub- to be uncovered both at fermentation and roasting levels. jected to roasting (Lee et al., 2017b), as aroma precursors gener- Consequently, as reviewed in this section of the review, we re- ated during fermentation influence the volatile profile of roasted alize that variance in microbial communities is observed among coffee via several complex biochemical pathways. Similarly, like the different fermentation methods, and this variance brings about their previous findings (Lee et al., 2016a, 2016b), the influence of variations in the volatile and nonvolatile compound profile scores Y. lipolytica fermentation was found to diminish on increasing the of the coffee, so these results infer that the microbiota can po- roasting degree. The most obvious difference between fermented tentially influence coffee quality. However, compositional data and unfermented coffee beans was seen at the light-roast level. alone are not enough to elaborate microbial contribution to cof- Three aldehydes, 3 acids, 2 alcohols, 4 esters, 22 furans, 4 alkenes, fee flavor. Future comprehensive research on linking or matching 19 ketones, 2 lactones, 7 furanones, 26 pyrazines, 15 pyridines, aroma compounds with their contribution to coffee flavor and 13 phenols, 17 pyrroles, 15 sulfur-containing compounds, and coffee quality attributes will broaden our current understanding. 2 terpenes were detected in light-, medium-, and dark-roasted Understanding the flavor contribution of each compound, and coffees. The levels of aldehydes, alcohols, acids, esters, lactones, subsequent tracking back to the individual processes involved in ketones, pyrazines, and terpenes decreased on increasing the roast its formation, will help the coffee industry to control desirable degree (Lee et al., 2017b). Among volatile compounds, the levels flavor outcomes of coffee through processing or other farm man- of acids (such as acetic acid), most diketones (2,3-butanedione, agement techniques. According to microbial ecology, in starter 2,3-pentanedione), and amino acids decreased, whereas the lev- culture technology using specific yeast or bacterial strains, the els of certain aromatic alcohols (2-phenylethanol, 4-vinylguaiacol populations of those strains remain more persistent and present and 4-vinylphenol), ketones (such as gamma-butyrolactone), sul- in greater amounts (about 80% of that species) than in controls fur compounds and cysteine increased (Lee et al., 2017b). Contrary throughout fermentation, which is not necessarily a sign of better to R. oligosporus fermentation, the levels of pyrazines, furan and flavor and aroma notes. Microbial diversity decreases with the pas- furanones in Y. lipolytica-fermented coffee were changed insignif- sage of fermentation time, due to a drop in pH and other hostile icantly after roasting. Among nonvolatile compounds, the levels conditions, except for some resistant species of yeast (S. cerevisiae) of hexoses, CGA, ferulic acid, and caffeic acid decreased, whereas which have an acid-tolerant nature. Almost 20 species of bacteria the levels of succinic acid, quinic acid, and p-coumaric acid in- and yeasts have been identified in coffee fermentation, but it is not creased in fermented light-roast coffee (Lee et al., 2017b). So, clear which species or strains have the most impact on coffee cup Y. lipolytica modulates the aroma profile of fermented coffee which quality. Different coffee varieties are influenced differently by dif- is preserved after roasting, such as generation of 2-phenylethanol, ferent yeast and bacterial species. All descriptions of starter culture

r 1198 ComprehensiveReviewsinFoodScienceandFoodSafety Vol.17,2018 C 2018 Institute of Food Technologists® Farm to consumer . . . technology research should mention the complete biochemical ity of coffee. An impermeable packaging system is called hermetic profile (volatile and nonvolatile compounds) of fermented and packaging/hermetic storage and is of 3 kinds: (a) organic-hermetic roasted beans. Future studies must focus on elucidation of the mi- storage, (b) vacuum-hermetic fumigation, and (c) gas-hermetic fu- gration mechanism of unique flavor-endowing aromatic metabo- migation. Organic-hermetic storage and gas-hermetic fumigation lites into seeds during fermentation, and also elaborate on methods are the only mostly used as packaging systems for green coffee for their conservation during roasting. bean storage. The former packaging system relies on the respira- tion and metabolic activities of microflora, insects, and the com- Storage modity itself to form a nonlife-sustaining low-oxygen atmosphere. The viability and quality of living coffee beans with their own Gas-hermetic fumigation packaging systems use an external gas active physiological system are highly influenced by storage condi- source (such as CO2) to create the nonlife-sustaining low-oxygen tions. Temperature, relative humidity (RH), moisture, and gas/air atmosphere. Both packaging systems are applicable for prolonged composition are the main driving forces of storage conditions. storage of green coffee beans, or transportation ranging from sev- Storage conditions may vary from region to region; however, the eral months to a year, without compromising the quality attributes choice of minimum optimal storage conditions largely depends of coffee as seen in jute sacks (Villers et al., 2010). Some advanced upon the local and international food safety laws and regulations, studies (Borem et al., 2013; Ribeiro et al., 2011; Villers et al., 2010) market conditions, environmental factors, period of storage, and also cited the use of impermeable packaging systems (for exam- type of coffee. Temperature is a most crucial factor influencing ple, GrainPro) for lining jute sacks, to control the inward/outward both the viability and quality of coffee beans. Coffee beans need flow of gases by injection and maintenance of 60% CO2. This to be stored at a low temperature to slow down metabolism, res- kind of packaging prevents a gain in moisture by hygroscopic cof- piration, and heat generation as a result of metabolic activities. fee beans for almost 12 months (Borem et al., 2013; Ribeiro et al., A high storage temperature combined with self-generated heat 2011; Villers et al., 2010). Coffee deterioration and quality loss is not only a major cause of weight loss, but it is also involved happen as a result of cellular membrane disintegration/injury and in the deterioration of green coffee aroma precursors. Storage of the resulting lixiviation/leaching of solutes/nutrients. Electrical high-moisture coffee beans (13% to 15%) at a lower temperature conductance and P lixiviation are the standard tests for check- (10 °C) is also helpful in preserving the quality of beans for up to ing for cell injuries (Abreu et al., 2017). Hermetic packaging 1 year, whereas the quality of coffee beans with a standard moisture systems, including GrainPro, also show low values for electrical content (11% to 12%) deteriorates after just 6 months when stored conductance and P lixiviation (that is, 160 μSiemens/cm/g as at 35 °C. RH is another factor with a direct influence on the qual- compared to 220 μSiemens/cm/g for jute sacks). However, pro- ity of coffee. Normally, it is recommended to store coffee below gressive loss of reducing sugars occurs in all kinds of packaging 60% RH, as a higher RH is a source of microbial infestation and systems throughout storage, but, luckily, according to Selmar et al. weight gain due to the hygroscopic nature of coffee seeds. Initial (2008), progressive loss of coffee aroma is independent of varia- coffee bean moisture content also matters in preserving the aroma tions in sugar and amino acid contents during prolonged storage. precursors of green coffee beans, and coffee beans are generally Storage in hermetic packaging without the injection of CO2 re- stored with a moisture content of 11% to 12%. A moisture content sults in more uniform sensory attributes, because no difference in higher or lower than this range brings about variations in the color, sensory quality grades is measured at low, medium, and high sam- appearance, consistency, aroma profile and, hence, cup quality of pling positions. Injection of CO2 creates sensory quality grades at beans. Table 4 represents the most aroma-active compounds highly the 3 different sampling positions, and the highest sensory quality influenced by the storage or/and storage conditions. grade is found for the medium sampling position. However, the Following production, storage is a crucial step that strongly in- grade for coffee stored without the injection of CO2 is inferior fluences coffee commercialization, product quality, meeting har- to the grade of coffee stored with the injection of CO2 (Borem vest demands, and securing the highest market price. For this et al., 2008b; Borem et al., 2013; Ribeiro et al., 2011). However, purpose, green coffee beans are commonly stored in 2 kinds of the work of Ribeiro and coworkers is devoid of the impact of bag, namely small (20 kg) jute bags and big jute bags (1200 kg). hermetic packaging on the aroma profile and sensory attributes Irrespective of the many advantages of this type of packaging, of hermetically stored green coffee beans. These limitations were it still provides free access to atmospheric oxygen and moisture further supported by Borem et al. (2013), who noted that the at ambient temperature which accelerate respiration and other color of hermetically sealed green coffee beans injected with 60% metabolic activities in green coffee beans, leading to physical, CO2 remains standard (that is, green) as compared to the yel- chemical, and sensory quality losses (Correa, Afonso Junior, Silva, lowish, bleached and mottled color of coffee seeds stored in jute & Ribeiro, 2003; Vieira, Tekrony, Egli, & Rucker, 2001).The sacks. It is noteworthy that hermetic packaging, with or without moisture content of jute bag-stored green coffee beans also rises the injection of CO2, preserves fruity, chocolaty, and vanilla-like with a rise in the RH of the environment, and this rising mois- aroma precursors, while jute sack-stored beans exhibit earthy and ture content is a source of microbial infestation/growth, causing a jute-like flavors (Borem et al., 2013). Full body, sweetness, and deterioration in the aroma profile and producing cancer-causing acidity features are also preserved in hermetically sealed coffee in- mycotoxins (Villers, Navarro, & De Bruin, 2010). Furthermore, jected with CO2. Moreover, this hermetically sealed coffee also the free oxygen availability in jute storage systems increases res- has “current crop” status as compared to jute sack-stored coffee piration and metabolic activity which result in the production beans which have old-crop characteristics (Borem et al., 2013). of byproducts that generate earthy, jute-like and woody aroma As discussed in the fermentation section, the profile of volatile defects. To overcome this issue, some authors (Borem, Nobre, and nonvolatile compounds is critical for determining organolep- Fernandes, Pereira, & Oliveira, 2008b; Nobre, Borem, Fernandes, tic coffee quality. Since coffee seeds are considered a living entity & Pereira, 2007) put forward the idea of storing coffee beans under with required metabolic activities, various studies have reported controlled/altered atmospheric conditions in impermeable pack- on the influence of storage time, temperature, and atmosphere on aging, and they concluded it is a viable way of preserving the qual- the profile of volatile and nonvolatile compounds (Table 4), and

r C 2018 Institute of Food Technologists® Vol.17,2018 ComprehensiveReviewsinFoodScienceandFoodSafety 1199 Farm to consumer . . .

Table 4–List of odor-active odorants most influenced by the storage or/and different storage conditions.

Pyrazines Acetates Pyrazine 4-Hydroxy-3-methoxybenzaldehyde Methyl acetate Methylpyrazine 3-(methylthio)propanal Ethylacetate 2-Ethenyl-5-methylpyrazine Ketones 1,2-Ethanedioldiacetate 2,6-Diethylpyrazine 2-Butanone Furfurylacetate 2-Ethyl-5-methylpyrazine 3-Hydroxy-2-butanone Alcohols 2-Ethyl-6-methylpyrazine γ -Butyrolactone 2-Furanmethanol 2,3-Dimethylpyrazine 2,3-Butanedione 2-phenylethanol Ethylpyrazine (E)-a-damascenoneˆ Pyrroles 2,6-Dimethylpyrazine 3-Hydroxy-4,5-dimethyl-2(5H)-furanone 1-Methylpyrrole 2,5-Dimethylpyrazine 4-Hydroxy-2,5-dimethyl-3(2H)-furanone Furans 3-Ethyl-2,5-dimethylpyrazine Esters Furan Aldehydes 3-Methylbutanoate 2-Methylfuran Acetaldehyde 2-Methylbutanoate 2,5-Dimethylfuran Propanal Ethyl 2-methylbutanoate 2-methyl-3-(methyldithio)furan 2-Methylpropanal Ethyl 3-methylbutanoate Carboxylic acids 2-Methylbutanal Phenols 2- and 3-Methylbutanoic acid 3-Methylbutanal Phenols Pyridines Hexanal Guaiacol Pyridine 2-Thiophenecarboxaldehyde 2-Methoxy-4-vinylphenol Thiols 2-Furancarboxaldehyde 2-Methoxy-5-vinylphenol Methanethiol 5-Methyl-2-furancarboxaldehyde 3-Methoxyphenol 2-Furfurylthiol (E,E)-2,4-nonadienal Thiophenes Thiozoles (E)-2-nonenal 2-Methylthiophene 4-Methylthiazole

Note: Data compiled from Czerny et al., 2000; Holscher, 1995; Illy and Viani, 1995; Marin et al., 2008; Scheidig et al., 2007. also the shelf-life and eventual cup quality of beans. Scheidig and age. From a sensory point of view, the authors quoted the same Schieberle (2006) studied the storage of commercial green coffee typical old, woody, raspy, and stale notes, even without quantifi- beans for up to 3 years and noted a typical flatness and slackness cation of lipid oxidation products. Vitzthum et al. (1975) were the in the cup quality of green coffee beans stored for prolonged pe- first to analyze the volatile aroma fractions of raw coffee beans, and riods. The “off-notes” also observed during the course of storage they identified two methoxypyrazines (3-isopropyl-2-methoxy- are mainly caused by the degradation and oxidation of the non- and 3-isobutyl-2-methoxypyrazine) as important aroma contrib- volatile fraction of green coffee beans, especially lipids (Kurzrock, utors out of 4. These methoxypyrazines also do not decrease dur- Klling-Speer, & Speer, 2005; Speer & Kolling-Speer, 2006). These ing roasting (Czerny & Grosch, 2000). Czerny and Grosch (2000) off-notes can be prevented by storing green coffee beans under op- also identified 2-methoxy-3,5-dimethylpyrazine as an important timum storage conditions, but flatness in cup quality is also noted aroma-active compound in raw coffee. After this, several studies even under optimum storage conditions (Abreu et al., 2017; Sel- reported ethyl 2-methylbutanoate and ethyl 3-methylbutanoate as mar et al., 2008). Some authors also related flatness, slackness, key odorants solely responsible for the off-flavor notes in long- and cup quality of coffee to the viability of coffee seeds (Sivetz & stored and overfermented coffee beans (Bade-Wegner et al., 1997; Desrosier, 1979), which was later approved by many studies (Abreu Stirling, 1980). Becker Do¨hla, Nitz, & Vitzthum, (1988) also et al., 2017; Rendon, Gratao, Salva, Azevedo, & Bragagnolo, 2013; highlighted pea-like off-flavor notes due to the production of 3- Selmar et al., 2008). However, some recent studies have presented isopropyl-2-methoxypyrazine from insects and methoxypyrazine- a different approach and concluded that coffee beans stored in producing bacteria in stored coffee. A comprehensive study re- parchment show little variation in color and in the organoleptic garding changes in the profile of key volatile odorants during attributes of coffee (Abreu et al., 2017; Selmar et al., 2008), which storage of green coffee beans was published by Scheidig, Czerny, may be due to a shielding effect and having less exposure to the and Schieberle (2007) who noted intense earthy, bell pepper- outside environment. Selmar and team also investigated the rela- like, pea-like, and flowery flavor defects in unstored raw coffee tionship of viability, processing type, and nonvolatile compound beans. The authors also carried out GCO (gas chromatogra- profile with storage (22 °C, 63% RH for 2 years) and established phy olfactometry) experiments with mass spectrometry to iden- that the level of hexoses (that is, glucose and fructose) decreases tify the key odorants for these off-flavor notes. They noted over time in all types of processed coffee beans; however, fewer cor- that 3 volatiles (2-methoxy-3-isopropylpyrazine, 2-methoxy-3- responding concentration differences were noted in wet-processed isobutylpyrazine, and 2-phenylethanol) are responsible for the green coffee beans. Contrary to these results, Bucheli et al. (1998) earthy, bell pepper-like, pea-like, and flowery flavor defects in found an increase in the glucose level during storage at 85% RH, fresh unstored coffee beans. However, odorants for fruity (ethyl which might be due to hydrolysis of sucrose as most metabolic 2-methylbutanoate and ethyl 3-methylbutanoate), grassy (hex- reactions are not retarded at 85% RH. A slight decrease in sucrose anal), potato-like (methional), meaty (2-methyl-(3-methyldithio level at 63% RH confirmed the retardation of metabolic activities furan), fatty ((E,E)-2,4-nonadienal), vanilla-like (vanillin), and at this humidity level, which is why Selmar and team noted a resid- clove-like (2-methoxy-4-vinylphenol) flavors have a low flavor ual loss of sucrose (7.5 to 7.0 g/100 g d.w.) during storage. Among dilution factor (Scheidig et al., 2007; Toledo, Pezza, Pezza, & the minor sugars, galactose did not show storage-dependent Toci, 2016). The sensory evaluation of the same green cof- change, whereas the level of rhamnose increased 10 times (0.25 fee beans stored at 40 °C for 9 months with a 20% oxygen to 2.5 mg/100 g d.w.) in 24 hr. Arabinose showed a different supply revealed more sweaty, fruity, spicy, and baked apple- pattern by displaying an increase (2 to about 4.5 mg/100 g d.w.) like flavors due to high flavor dilution factors of 2- and 3- only in dry-processed coffee (Selmar et al., 2008). No significant methylbutanoic acid (sweaty), methyl 2- and 3-methylbutanoate, changes were cited in the total content of amino acids, including ethyl 2-methylbutanoate, ethyl 3-methylbutanoate (fruity), GABA, except for that of glutamine which decreased during stor- 3-hydroxy-4,5-dimethyl-2(5H)-furanone (sotolon) (spicy), and

r 1200 ComprehensiveReviewsinFoodScienceandFoodSafety Vol.17,2018 C 2018 Institute of Food Technologists® Farm to consumer . . .

(E)-β-damascenone (baked apple) odorants. Besides this, 2 phe- months of storage is attributed to stress conditions due to water nols, β-damascenone and the methyl esters of 2- and 3- loss, a higher rate of respiration, and ROS accumulation (Dussert methylbutanoic acid showed a marked increase in stored cof- et al., 2006; Rendon et al., 2013, 2014). A higher rate of carbonyl fee beans. Quantitatively, 5 odorants (2-methoxy-4-vinylphenol, group production as result of protein breakdown also occurs in the 2-methoxy-5-vinylphenol, methyl 2-methylbutanoate, methyl 3- early months of storage, most probably due to lipid oxidation and methylbutanoate, and (E)-β-damascenone) were found to differ oxidative stress during drying (Rendon et al., 2013; Toledo et al., markedly in stored and unstored coffee beans (Scheidig et al., 2016). ROS accumulation during oxidative stress is also a cause 2007). The volatiles 2-methoxy-4-vinylphenol, 2-methoxy-5- of disintegration of cell membranes and structures, due to which vinylphenol, 2 methyl esters, and (E)-β-damascenone increased various enzyme systems can easily meet up with their substrates, in stored coffee beans. On the other hand, (E,E)-2,4-nonadienal which results in the formation of compounds that affect both the decreased considerably, and 2-methoxy-3-isopropylpyrazine re- color and cup quality of coffee (Scheidig et al., 2007). The level mained unchanged. The authors further elaborated that 2- of CGA (5-CQA), an antioxidant compound, also decreases with methoxy-4-vinylphenol, all esters, and (E)-β-damascenone the passage of storage time, due to possible participation in radical- increased significantly during the initial months of storage of cof- scavenging activities, that is, lipid and protein oxidation. However, fee seeds with 13.5% moisture at 40 °C, while 2-methoxy-5- Rendon et al. (2014) detected a stable TBARS (thiobarbituric vinylphenol increased considerably after 9 or 12 months. In con- acid-reactive substances) value after the third month of storage; trast, when storage temperature and coffee bean moisture were possible reasons behind this are interruption of the respiration reduced to 25 °C and 11.5% (constant 20% oxygen), respectively, process and, hence, of the lipid oxidation process during storage the generation of 2-methoxy-4-vinylphenol slowed down con- (Rojast, 2006), and most probably the reaction of lipid oxidation siderably, and 2-methoxy-5-vinylphenol and (E)-β-damascenone products with protein-forming polymers not detectable by the were not formed/detected even for 18 months of storage. A fur- TBARS test (Rendon et al., 2014).Thus, from a sensory point of ther decrease in storage temperature (25 to 12 °C) and coffee bean view,a high level of FFAs, carbonyl groups and ROS, together with moisture content (11.5% to 6.5%) not only lowered vinylphe- lipid oxidation, can be considered responsible for the characteristic nols and (E)-β-damascenone but it also inhibited the formation woody and “rested coffee” flavor in long-stored coffee beans. Or- of methyl and ethyl esters. Thus, it was revealed that water con- tala, Gutierrez, Chiralt, and Fito (1998) pinpointed the influence tent is the most critical factor during storage at high temperatures of storage time on bean lipids and highlighted that storage temper- (Toledo et al., 2016). This work also suggests the generation of ature is the main cause of hydroperoxides in low-moisture coffee 2-methoxy-5-vinylphenol from decarboxylation of isoferulic acid beans, but storage temperature is independent of hydroperoxide as an indicator of improper storage of coffee beans (Scheidig et al., production in hydrated coffee beans. This may be due to water 2007; Toledo et al., 2016).Controversially, the same study also blunting the effects of temperature and atmosphere on a bigger claimed the nonoccurrence of lipid oxidation during roasting in- fraction (74% to 75%) of coffee bean lipids, the triacylglycerols stead of during storage. The authors detected a very small amount (TAGs). of unsaturated fatty acid oxidation products ((E)-2-nonenal and After the preservation of green coffee beans, the preservation (E,E)-2,4-nonadienal), even at elevated temperatures and for ex- of roasted whole or ground coffee with inherent quality attributes tended storage periods, considered responsible for the character- is also crucial for its commercialization. Different factors which istic woody flavor in long-stored coffee beans (Rendon, Salva, & may influence the quality of roasted coffee include its moisture Bragagnolo, 2014). This contradiction may arise due to the fact content, grinding size, roast level, storage temperature and period, that the authors supplied only 20% oxygen during all kinds of ex- true density (ρu), bulk density (ρap), intergranular porosity (ε), periments, which is quite low compared to the free air/oxygen and and so on. Moisture is the most important physical factor for moisture access to jute sack-stored coffee beans. More specifically determining the quality and shelf-life of various food items, for the lipid fractions of green coffee beans, physical and sensory including coffee, during storage. The legal moisture limit for changes in green coffee beans during storage are also associated storing and conserving the quality of whole and ground coffee with lipid oxidation due to greater activity of lipases, carbonyl beans is 5% to 5.26% (dry basis) (Brasil, 2010). The moisture level group production from the breakdown of proteins, ROS genera- of roasted whole or ground coffee also depends upon the roast tion, and cellular structure/membrane disruption due to many en- level and particle size of ground coffee (Melo, 2004; Oliveira, zymatic and nonenzymatic reactions. For physical changes, color is Correa, Oliveira, Baptestini, & Vargas-elias, 2017; Schmidt et al., an indicator of the storage duration/age of coffee, and a criterion 2008). Similarly, storage period, along with the variables of roast- for deciding the price of coffee. The green color of fresh cof- ing and particle size, significantly influences the moisture content fee beans normally fades to a lighter and more yellow hue (CCI, of all coffee types at both low and high storage temperatures. 2002; Rendon et al., 2014; Rojas, 2004). Color variations are During storage, ground coffee beans (coarse: 1.19 mm, medium: also an indication of bad drying and storing practices leading to 0.84 mm and fine: 0.59 mm) from medium to dark roasting are nonuniformity and poor appearance of roasted beans. Color vari- more hygroscopic, with a higher moisture content than light- ations are more pronounced in hulled coffee seeds (washed and to medium-roasted coffee (Correa et al., 2016a; Oliveira et al., semi-washed) than in nonhulled (naturals) coffee beans (Godinho, 2017). However, the equivalent moisture level of both light- Vilela, Oliveira, & Chagas, 2000; Toledo et al., 2016); however, and dark-roasted ground coffee after 180 days indicates that the storing coffee beans in their natural form is not a commercially moisture level of roasted coffee is a function of storage time adopted methodology. Changes in color during storage are mainly (Correa et al., 2016a). The initial difference in moisture content correlated with an increase in free fatty acids (FFAs) and carbonyl between light-medium to dark-roasted coffee beans may be due groups, and a decrease in 5-CQA. FFA content is also higher in to a high degree of moisture loss during prolonged roasting semi-wet/wet-processed coffee beans (3.84 mg/g oil) than in nat- to produce a dark roast (Baggenstoss, poisson, Kaegi, Perren, ural coffee beans (3.76 mg/g oil) (Rendon et al., 2014; Ribeiro & Escher, 2008a, 2008b). In contrast, some authors have also et al., 2011). The higher rate of FFA formation during the early deduced that roasting degree does not affect the hygroscopic

r C 2018 Institute of Food Technologists® Vol.17,2018 ComprehensiveReviewsinFoodScienceandFoodSafety 1201 Farm to consumer . . . ability of whole and ground coffee (Fernandes, Pereira, Borem, to unknown coffee blends with unknown roasting conditions. Nery, & Padua, 2006), but storage at low temperature markedly Moreover, staling mostly depends upon roasting conditions, reduces the adsorption of moisture from the environment, in storage time, temperature, moisture, packaging system, and con- addition to varying other influential physical factors like ρu, tact of coffee material with the external environment (Oliveira ρap, and intergranular porosity (Correa et al., 2016a), because et al., 2017; Toledo et al., 2016). Marin et al. (2008) described a higher storage temperature causes more excitation of water the staleness of coffee stored in open and punctured packages molecules, leaving sorption sites for the retrieval of moisture from after 4 months, whereas hermetically sealed coffee showed little the environment (Correa, Oliveira, Botelho, Goneli, & Carvalho, sign of staleness. The same authors also mentioned the loss of 2010). As compared to that of roasted ground coffee, storage of key odorant compounds, and not off-odor oxidation products, whole roasted coffee beans is preferable for conserving coffee being responsible for staleness. Toci et al. (2013) investigated the constituents, moisture, and quality, without agglomeration and combined effect of storage time, temperature, and atmosphere on compactness, probably due to intact cells working as a physical TAGs and FFAs in light- and dark-roasted beans. A continuous barrier (Toledo et al., 2016). In ground roast coffee, light- to decrease in TAGs (42% to 51%) was reported, both in light- and medium-roasted coffees with high ρu and ρap usually show less dark-roasted beans, for up to 6 months. To better visualize this agglomeration and compactness (Correa et al., 2016b). These phenomenon, the authors also stated that the ratio of unsaturated results are based on 2 facts; one is that more finely ground coffee to saturated fatty acids (ࢣUFA/SFA) remains nearly constant is more tightly agglomerated for a given volume, resulting in a (1.24 to 1.70) in light-roasted coffee beans, but it tends to decrease higher weight per volume and lower ρu. The second is that dark (1.06 to 1.38) from the third to sixth months of storage due to roasting causes bigger volume increases with higher weight losses esterification of UFAs (Toci et al., 2013). Some authors have also followed by lower ρu (Licciardi, Pereira, Mendonc¸a, & Furtado, reported the loss of aromatic viability of coffee beans after 5 to 2005). With reference to ρap, dark-roast ground coffee has a lower 6 months of storage, so measurement of the ratio (ࢣUFA/SFA) ρap due to a great loss of volatiles and release of CO2, and volume could prove to be a promising tool in assessing the quality and changes due to intense heat transfer and pyrolysis during a longer shelf-life of coffee (Baggenstoss et al., 2008a, 2008b; Kreuml, roasting period (Borges, Mendonc¸a, Franca, Oliveira, & Correa, Majchrzak, Ploederl, & Koenig, 2013; Toci et al., 2013).The 2004; Toledo et al., 2016). More than 1000 volatile compounds FFA fraction of coffee lipids increases up to the third month of involved in the aroma profile formation of coffee beverages have storage and then dramatically decreases up to the sixth month been identified so far, but recent research work has revealed that of storage in both light- and dark-roasted coffee. With respect only 25 and 28 odorant compounds actually constitute the aroma to individual FAs, UFAs are more prone to oxidation (Frankel, of fresh coffee beverage and ground coffee, respectively (Czerny, 2005; Kreuml et al., 2013). It has been observed that storage Mayer, & Grosch, 1999; Mayer & Grosch, 2001; Mayer, Czerny, time, temperature, atmosphere, and their interactions significantly & Grosch, 2000). Storage of roasted coffee beans results in a influence TAGs, and determine the rise of FFAs both in green loss of freshness, causing staleness which usually occurs due to and roasted beans (Speer & Kolling-Speer, 2006; Kreuml et al., loss and oxidation of these key odorant compounds as indicated 2013; Toci et al., 2013). Speer and Kolling-Speer (2006) also in Table 4 (Holscher & Steinhart, 1992; Radtke-Granzer & reported that temperature and bean moisture content are major Piringer, 1981; Nicoli et al., 1993). Many authors have described factorsinanincreaseinFFAs(3.8to4.8g/kg)andfreediterpenes the ratios of several pairs of these compounds as changing with in storage for up to 18 months, whereas Kreuml et al. (2013) the passage of storage time and under varying storage conditions, described the continuous decrease of sensory qualities in roasted leading to staleness (Arackal & Lehmann, 1979; Vitzthum stored Arabica and Robusta coffee beans even with vacuum et al., 1979). The ratios of 2-methylfuran to 2-butanone (M/B) packaging. and of methanol to 2-methylfuran (M/M) are indices of the After discussing variations in the volatile and nonvolatile com- freshness of coffee, and they decrease with the passage of time. pounds in green and roasted coffee beans during storage and Kallio, Leino, Koullias, Kallio, and Kaitaranta (1990) and Leino, under the influence of different storage conditions, some re- Kaitaranta and Kalli, (1992) also found that the level of some searchers have also investigated the same theme in coffee brews, odorants, termed the s-index (2-methylfuran, 2-methylpropanal, as the outcomes from green and roasted coffee beans are not 3-methylbutanal, and 2,3-butanedione), decreases in stale coffee applicable to coffee brews. The first to report work on cof- during storage. Holscher and Steinhart (1992) also wrote about fee brews was Steinhart and Luger (1995), who investigated the low level of the freshness-endowing odorant methanethiol the influence of oxygen on the volatiles in coffee brew during in stale coffees. Czerny and Schieberle (2001) claimed that brief storage (4 hr) at 80 °C. In addition to preexisting ag- the degradation of 2-furfurylthiol is an indicator of staleness. ing indices, that is, M/M (methanol/2-methylfuran) and M/B According to various studies in the literature, many aroma (methanol/2-butanone), Steinhart and coworkers also presented 2 and freshness indices have been proposed, including M/B new indices, that is, DMS/B (dimethyl sulfide/2-butanone) and (2-methylfuran/2-butanone), M/BD (2-methylfuran/2,3- DMDS/B (dimethyl disulfide/2-butanone), called oxidative in- butanedione), M/P (2-methylfuran/propanal), MT/BD dices, based on delivering reliable information about oxidative (methanethiol/2,3-butanedione), MT/HE (methanethiol/ and stress reactions. The authors also noted the occurrence of hexanal), MT/B (methanethiol/2-butanone), and FT/HE (2- hydrolytic reactions at such high storage temperatures, and de- furfurylthiol/hexanal). A linear relationship between a decrease noted methylformate as the main indicator of hydrolytic reactions. in these indices and staleness over time has also been detected However, the authors focused on only 14 volatiles, as compared (Czerny & Schieberle, 2001; Kallio et al., 1990; Leino, Kaitaranta, to the 25 to 28 volatiles documented to form the aroma pro- & Kallio, 1992; Marin, Pozrl, Zlatic, & Plestenjak, 2008; Toledo file of coffee beverages (Czerny et al., 1999; Mayer & Grosch, et al., 2016). However, these aroma and freshness indices were 2001; Mayer et al., 2000). The other drawbacks of this work investigated in familiar blends of coffee prepared from known are the inapplicability of its results to coffee brews stored at a roasting conditions, so these quality indices may not be applicable lower temperature for longer periods, and unavailability of sensory

r 1202 ComprehensiveReviewsinFoodScienceandFoodSafety Vol.17,2018 C 2018 Institute of Food Technologists® Farm to consumer . . . evaluation data. These limitations were later covered by the work Araki, Sagara, & Tambunan, 2017; Summa, dela Calle, Brohee, of Perez-Martınez, Sopelana, Paz de Pena, and Cid (2008), who Stadler, & Anklam, 2007; Wang & Lim, 2014). For physical identified and quantified 47 compounds (2 sulfur compounds, 5 changes, color, bulk density, apparent density, roast volume, and pyrazines, 3 esters, 7 aldehydes, 5 ketones, 5 furans, 1 alcohol, 2 roast loss are important features. Roast volume generally increases thiophenes, 1 pyridine, 4 pyrroles, 1 alkene, and 1 acid) in the during roasting due to the softening of cellulose and release headspace of Arabica coffee brews throughout 30 days of stor- of pyrolysis products and gases. The increase in roast volume age, both at 4 and 25 °C. Freshness aroma odorants (sulfur com- consequently results in the decrease of apparent and bulk densities, pounds, for example, methanethiol and dimethyl sulfide) started but defective or poor-quality roasted beans have lower values for to degrade even at the start of storage, and diminished earlier, ex- roast volume and apparent and bulk densities than normal-quality cept for dimethyl sulfide, which did not show significant variation roasted beans. These kinds of defective beans also need to be at low temperature. Fruity and malty flavor odorants (aldehy- roasted slowly and to a lesser degree than healthy beans (Franca, des, propanal, and Strecker aldehydes) increased during storage, Oliveira, Mendonc¸a, & Silva, 2005). Roast or weight loss exhibits both at 25 and 4 °C. Among aromatic esters, formic acid methyl two trends; less weight loss occurs during the initial stages of esters decreased considerably at high storage temperature (Perez- roasting, while more weight loss is noted in the later stages. The Martınez et al., 2008; Toledo et al., 2016). Furans and ketones, initial lower weight loss is due to dehydration and slow release of responsible for burnt, burnt sugar, caramel, and buttery flavors, volatiles, while higher weight loss in the later stages is attributed dropped enormously at 25 °C compared to 4 °C. Thiophenes and to intensive volatile and CO2 loss due to pyrolysis (Rodrigues, burnt flavor-endowing pyrroles were the volatiles most affected at Borges, Franca, Oliveira, & Correa, 2003). The development high temperature, and they decreased with the passage of storage of color (light to dark) is one of the most important physical time. Smoky and earthy/musty odorants (pyridines and pyrazines) changes and is frequently used for assessing the degree/end-point showed the same trend and decreased considerably during earlier of roasting. Analysis of L∗a∗b∗,whereL represents brightness or days of storage at both temperatures (Perez-Martınez et al., 2008; luminosity, +a to −a represents reddish to greenish color, and +b Toledo et al., 2016). Generally, odorant loss is less intense at re- to −b yellowish to bluish color, for this purpose is an approach frigerated temperatures as compared to the greater aroma loss with widely reported in the literature. Luminosity and chroma rancid notes at room temperature. Moreover, the 7 aroma indices usually decrease during roasting due to Maillard reactions and proposed in this study could be useful tools to monitor both the caramelization, and this decrease is more obvious for whole coffee age and sensory quality of stored coffee brews. beans than for ground coffee. Variations in bean color from light green to dark brown also bring about changes in hue angle which Primary Processing can be divided into three steps corresponding to color changes Roasting from light green to light yellowish and from light yellowish brown Physicochemical variations. There are many kinds of com- to dark brown (Rodrigues et al., 2003). Color measurements mercially available coffee products on the market, such as whole have been successfully applied to predict the optimal temperature roasted coffee, ground roasted coffee, soluble (instant) coffee, and period of roasting (Mendes, De Menezes, Aparecida, & Da and ready-to-drink coffee. The production technology for Silva, 2001), acrylamide level (Gokmen¨ & S¸enyuva, 2006), and to each coffee product depends upon the kind of coffee, but for discriminate between raw and roasted Arabica and Robusta beans production of any kind of coffee product, roasting is a mandatory (Mendonca, Franca, & Oliveira, 2009). Fine color correlations step as it involves the production of coloring, flavoring, and with sucrose content (Santos, Viegas, & Pascoa, 2016b), acidity aromatic compounds (Eggers & Pietsch, 2001). Roasting involves (Santos, Lopo, Rangel, & Lopes, 2016a), moisture, density a complex series of chemical reactions (that is, Maillard reactions, (Alessandrini, Romani, Pinnavaia, & Rosa, 2008), and caffeine caramelization and pyrolysis) inside the seeds that results in bean content (Pizarro, Esteban-Diez, Gonzalez-Saiz,´ & Forina, 2007) expansion, loss of seed mass and moisture, and generation of have also been established. Online roasting control systems coloring, flavoring, and aromatic compounds. Roasting can via online image monitoring also work on the principle of be divided into 3 stages: a) preroasting (<150 °C), b) roasting evaluating variation in surface color/brightness and surface area (>150 to 180 °C), and c) overroasting. Preroasting involves of coffee beans (Hernandez, Heyd, & Trystram, 2008). However, the drying of seeds and the beginning of a complex series of the color-based assessment of roast degree has also become chemical reactions. Actual roasting occurs in the second stage controversial after results were published describing that coffee of roasting, whereas overroasting comprises a reduction in the roasting under different conditions may present the same color extent of aroma- and flavor-producing reactions and burning of (Dutra, Oliveira, Franca, Ferraz, & Afonso, 2001). So, after visual beans (Hernandez,´ Heyd, Irles, Valdovinos, & Trystram, 2007). inspection of color changes, light reflectance is another way Production of aromatic and flavoring compounds mostly occurs used to assess the degree of roast but, unfortunately, coffee with in the second stage of roasting. The generation of flavoring varying degrees of roast also presents the same values for light and aromatic compounds follows path-dependent biochemical reflectance, which raises concerns/questions about the credibility reactions which are greatly under the influence of roasting of using the robust and noninvasive near-infrared spectroscopy settings/levels (roasting time, roasting temperature, hot air flow technique (Bertone, Venturello, Giraudo, Pellegrino, & Geobaldo, speed, and so on) In industry, roasting settings are a function of 2016) for the authentication of coffee varieties and roast degree. dedicated artisanal/personal expertise from hit and trial methods, Furthermore, it has been observed that dark and sour beans are and there is much literature disclosing the effects of different less luminous than nondefective and mature beans. It has also been roasting settings on the physical, chemical, structural, mechanical, deduced that, under the same roasting conditions, low-quality and compositional changes to coffee beans during roasting (Bicho, coffee will have a lower roasting degree with lower weight loss Leitao,˜ Ramalho, deAlvarenga, & Lidon, 2013; Franca, Oliveira, than high-quality coffee (Franca et al., 2005). Poor mixing and Oliveira, Agresti, & Augusti, 2009; Moon & Shibamoto, 2009; nonuniformity of green coffee beans while roasting is another Petisca et al., 2013; Pittia, Dalla Rosa, & Lerici, 2001; Pramudita, factor in unexpected colorations (Clark, 1987). Roast (weight)

r C 2018 Institute of Food Technologists® Vol.17,2018 ComprehensiveReviewsinFoodScienceandFoodSafety 1203 Farm to consumer . . .

Table 5–A comprehensive list of major identified odorant compounds with their contributing sensory aroma notes generated and identified upon the (light, medium, and dark) roasting of coffee green beans.

Furans Pyrroles 2-Methylfuran Unpleasant 1-Methylpyrrole Coffee 2,5-Dimethylfuran Coffee 2-Acetylpyrrole Unpleasannt 2-Acetylfuran Spicy 2-Acetyl-1-methylpyrrole Coffee 2-Vinyl-5-methylfuran Coffee 2-Formyl-1-methylpyrrole Buttery Furfuryl alcohol Burnt Pyrrole Toasted 5-Methylfurfural Caramel Pyridines Furfuryl acetate Nutty Pyridine Burnt 2-Propionyl-furan Fruity 2-Methylpyridine Coffee 1-(2-furyl)-2-propanone Pleasant 3-Ethylpyridine Toasted Furfuryl formate Floral 3-Hydroxypyridine Fishy 2-Furfuryl methyl sulfide Coffee 6-Methyl-3-pyridinol unwanted notes Furfural Almond/Bitter Phenols 4-4-Hydroxy-2,5-dimethyl-3[2H] furanone Roasty, sweet, caramel Phenol astringent, unpleasant (Furaneol) Kahweofuran Coffee, smoky, burnt Guaiacol medical, adhesive 2-Methyl-3-furanthiol Sulfury, meaty, roasted 4-Ethyl guaiacol Flower, spicy 2-Methyltetrahydrofuran-3-One Nutty 4-Vinyl guaiacol sweet, flowers Pyrazines Vanilline Vanilla like Pyrazine Coffee Cyclopentenes/Ketones 2-Methylpyrazine Toasted 2-2-Hydroxy-3-methyl-2-cyclopenten-1-one Spicy, celery, roasted, caramel 2,5 Dimethylpyrazine Toasted, roasty, nuts 3-Ethyl-2-hydroxy-2-cyclopenten-1-one Spicy, celery 2,6-Dimethylpyrazine Nutty, sulfur like 3,4-Dimethylcyclopentenol-1-one Cramel like, sweet 2-Ethylpyrazine Toasted 1-Octen-3-one mushroom, fungus, hay 2,3-Dimethylpyrazine Toasted 2,3-Hexadione mushroom 2-Ethyl-6-methylpyrazine Toasted, caraway, cheese 2,3-Butanedione buttery, oily 2-Ethyl-3-methylpyrazine Nutty 2,3-Pentanedione buttery, oily 2-Ethyl-5-methylpyrazine Toasted, caraway 4-(4-hydroxyphenyl)-2-butanone sweet, caramel 2-Vinyl-5-methylpyrazine Toasted 1-(2-furanyl)-2-butanone Fruity 2,6-Diethylpyrazine Toasted 2,4-Dimethyl-3-pentanone Pleasant 2-Methyl-6-vinylpyrazine Coffee 3-Hydroxy-2-butanone Buttry 2-Acetylpyrazine Toasted, Coffee 1-Hydroxy-2-propanon Nutty 2,3-Diethyl-5-methyl pyrazine Earthy, Roasty 2-Ethyl-4-methyl-2,5-furanedione Coffee 2-Ethenyl-5-methyl pyrazine musty, burnt 1-(6-methyl-2-pyrazidine)-1-ethanone Coffee 2-Ethyl-3,5-dimethyl pyrazine Earthy, Roasty, potatoes -Methyl-2-(3H)-furanone Coffee 3-Ethyl-2,5-dimethyl pyrazine Earthy, Roasty, potatoes 1-(2-furyl)-2-propanone Pleasant Ethyl pyrazine Toasted, caraway 3-Methyl-2-cyclopenten-1-one Floral 2-Isobutyl-3-methoxy pyrazine Herbs, smoky, peperika 1-Acetyloxy-2-propanone unpleasant M Methyl dihydro cyclopenta pyrazine Roasty sweet 2-Methyl-cyclopenten-1-one Toasted Propyl pyrazine Potatoes, vegetables 1-Hydroxy-2-butanone Toasted 2,3,5-Trimethyl pyrazine Herbs, earthy, musty 2-Ethyl-4-methyl-2,5-furanedione Coffee Aldehydes 1-(6-methyl-2-pyrazidine)-1-ethanone Coffee 2-Methylbutanal Sweet 2-Ethyl-4-hydroxy-5-methyl-3(2H)-furanone sweet, caramel 3-Methylbutanal Sweet 3-Hydroxy-4,5-dimethyl-2(5H)-furanone sweet, caramel 4-Methoxy-benzaldehyde grass, hay, mint 4-Hydroxy-2,5-dimethyl-3(2H)-furanone sweet, caramel 2-Methylpropanal buttery, oily 5-Ethyl-3-hydroxy-4-methyl-2(5H)-furanone sweet, caramel 3-Methylbutanal Malty 5-Ethyl-4-hydroxy-2-methyl-3(2H)-furanone sweet, caramel Phenylacetaldehyde sweet, fruity 1-Acetyloxy-2-propanone unpleasant Propanal unpleasant Sulfur-containing compounds Esters Dimethyl trisulfide Meaty Ethylisovalerate Fruity Bis(2-methyl-3-furyl)disulphide boiled potato like Methyl acetate Pleasant Methional Ethyl acetate Fruity Alcohols Ethyl-2-butenoate Floral Ethanol Sweet Ethyl-3-methylbutyrate Fruity 3-Methylbutan-1-ol unpleasant Ethyl-2-methylbutyrate Fruity Thiols 3-3-Methyl mercapto-3-methyl butyl Herbs Methanethiol amines like formate Acids 3-Methyl-2-butene-1-thiol vegetables 2-Methylbutyric acid Sweaty 2-Methyl-3-furanthiol boiled meat 3-Methylbutyric acid Sweaty Miscellaneous compounds Acetic acid pungent, sour 3-Methylthiophene Roasty Formic acid unpleasant 2,4-Dimethyl-5-ethylthiazole earthy, roasty Terpenes (E)-β-damascenone floral, honey Linalool floral, grass, vegetables Toluene unpleasant Limonene Lemon, Thiazole Toasted Geraniol Astringent 4 Methoxy benzenamine unpleasant γ -Butyrolactone Sweet

Note: Data were compiled from Akiyama et al., 2008, 2003; Baggenstoss et al., 2008a,b; Craig et al., 2015; Czerny et al., 1999, 2008; Czerny & Grosch, 2000; Dawidowicz & Typek, 2017; Dorfner et al., 2004; Gloess et al., 2014; Gonzalez-Rios et al., 2007a,b; Grosch et al., 1999, 2001a,b; Hertz-Schunemann¨ et al. 2013a, 2013b; Lee et al., 2016a,b; Mayer & Grosch, 2001; Moon and Shibamoto et al., 2009; Nebesny et al., 2007; Pramudita et al., 2017; Ribeiro et al., 2010; Rousi et al., 2012; Santos et al., 2016b; Schenker et al., 2002; Steen et al., 2017; Toledo et al., 2017; Wang & Lim, 2014.

r 1204 ComprehensiveReviewsinFoodScienceandFoodSafety Vol.17,2018 C 2018 Institute of Food Technologists® Farm to consumer . . . loss has also been proposed as a suitable indicator of degree of tanedione, 2,5-dimethylfuran/2,3-pentanedione, 2-methylfuran/ roast (Clark, 1987), but aberrant variation in roast loss in the furfural, 2,3-pentanedione/2,3-butanedione, pyridine/methylpy- works of Franca et al. (2005), 2009) also raised questions about razine, pyridine/2,6-dimethylpyrazine, pyridine/2,5-dimethyl- the credibility of this method. So, color variation and weight pyrazine, pyridine/2,5-dimethylpyrazine, pyridine/ethylpyrazine, loss, individually, may not be reliable tools to assess roast degree, pyridine/2-ethyl-6-methylpyrazine, and 5-methylfurfural/2- and Franca et al. (2009) proposed a parallel approach of assessing acetylfuran) having a good relationship with roast color, roasting degree via comparative analysis of the volatiles of both and their measurement is a reliable tool for predicting roast green and roasted coffee beans at 200 and 300 °C. This study levels. Among 13 aroma indices, 11 are common to Ara- showed that only 30% of compounds persist during roasting, bica, Robusta, and their 50/50 blend, whereas only 2 aroma with the remaining 70% being degraded and only detected in indices, 2-methylfuran/furfural and 2,3-pentanedione/2,3- raw coffee beans. The compound 3-isobutyl-2-methoxypyrazine butanedione, are specific for Arabica, and the guaiacol/2- is only detected in raw beans, whereas furans, pyrazines, ketones, ethyl-3,6-dimethylpyrazine index is specific for Robusta and pyridines, and pyrroles are only detected in roasted coffee 50/50 Arabica/Robusta blends. Among all these indices, 5- beans, and the concentrations of these compounds increase methylfurfural/2-acetylfuran gives the best color indications and with roasting degree (Franca et al., 2009). Multivariate statistical is the most reliable (r = 0.9989) marker of degree of roast (Ruosi analysis (principal component analysis (PCA) and hierarchical et al., 2012). All the aromatic compounds involved in composing cluster analysis (HCA)) on the extensive number of volatiles has these aroma indices are very similar to previously identified and revealed the highest caffeine content as an indicator of (unroasted) listed aromatic compounds (Sanz et al., 2002) from different Ara- green coffee beans, whereas the intensity of other discriminating bica/Robusta (A80/R20 to A20/R80) blends. Generally, more compounds (2-methyl pyrimidine, furfural, 3-furanmethanol, sulfur compounds are found in A20/R80 blends, while Strecker 5-methyl-2-furancarboxaldehyde, 2-furanmethanol, acetate, 2,6- aldehydes, furans, pyrazines, and α-dicarbonyls are abundant in diethyl pyrazine, and 4-vinylguaiacol) increase with an increase A80/R20 blends. Specifically, the key aroma-denoting odorants in roasting temperature, and can be used to assess roasting degree like carbon disulfide, methanethiol, and 2-methylfurans are up to 200 °C or light roast. From 200 to 300 °C, concentrations found in abundance in Robusta-dominant blends, and dimethyl of the odorants pyridine, furfural, 3-furanmethanol, 5-methyl-2- sulfide, dimethyl disulfide, 2-methylpropanal, 2-methylbutanal, furancarboxaldehyde, 2-(n-propyl)-pyrazine, and 4-vinylguaiacol 3-methylbutanal, α-dicarbonyls, pyrazines, 2,3-butanedione, and increase, and these compounds could be a marker of dark roast 2,3-pentanedione are abundant in Arabica-rich blends (Sanz et al., (Franca et al., 2009). Similar results were also reported by Toledo 2002). Roasting of coffee blends with added sugars (torrefacto et al. (2017) and Dutra et al. (2001), who proposed 2-hydroxy- coffee) also results in the excess production of furans, pyridines, pyridine, 5-methyl-2-furfural, and caffeine as markers of the and pyrazines, owing to an increase in Maillard and caramelization optimal end-point of roasting. Toledo and colleagues noted that reactions, which is why torrefacto coffee always has more roasted, the pyrazine class of volatiles comprises half of all markers, with burnt, and caramel notes than natural coffee (Loapez-Galilea et al., 2,3-dimethylpyrazine and 2-methylpyrazine as discrete chemical 2006; Sanz et al., 2002). A more advanced way of determining discriminators for dark/French roast, and dihydro-2-methyl- roast level is online or real-time chemical (volatiles) monitoring 3(2H)-furanone as a specific marker for light roast. Conclusively, of roasting with specific or a combination of analytical methods the work of Franca et al. (2009) emphasized the consideration of (that is, direct inlet mass spectrometry (DIMS), atmospheric temperature with color and roast loss for determining roasting pressure chemical ionization mass spectrometry (APCI-MS), level. In contrast, time–temperature conditions to attain the same proton-transfer reaction mass spectrometry (PTR-MS), proton- roast color do not imply the same coffee aroma and physical transfer reaction time of flight mass spectrometry (PTR-ToF-MS), properties (discussed below) (Baggenstoss et al., 2008a, 2008b). resonance-enhanced multiphoton ionization (REMPI)-ToF-MS, This may be due to the fact that the origin and nature of aroma REMPI, vacuum UV single-photon ionization (VUV-SPI), compounds also depend upon the chemical reactions to which SPI-ToF-MS, and so on) in which novel markers (for example, compositional components are submitted during roasting, and sugar, protein, and CGA degradation products, kahweol, caffeine, their formation and abundance strongly rely on roasting degree. phenol, and cafestol) for optimal roasting are identified (Dorfner, Various authors have pointed out several aroma compounds as Ferge, Yeretzian, Kettrup, & Zimmermann, 2004; Gloess, Vietri, markers for discriminating roasting degree, variety, blend, and & Wieland, 2014; Hertz-Schunemann,¨ Dorfner, & Yeretzian, so on without the roast color and roast loss conflicts (Table 5). 2013a; Hertz-Schunemann,¨ Streibel, Ehlert, & Zimmermann, Aroma compounds which are designated for the discrimination of 2013b). DIMS causes fragmentation and overlapping of parent roast level include furan, 2-methylfuran, 2,3-pentanedione, 2,3- ions, which cause difficulty in identifying and measuring volatiles. butanedione, 5-methylfurfural, pyridine, 3-ethylpyridine, furfural, However, besides DIMS, all other analytical tools have soft and guaiacol, whereas 2-acetylfuran, 2-vinyl-5-methylfuran and or little ionization, high sensitivity and resolution devoid of 2-propylpyrazine are used for differentiation of coffee varieties fragmentation. But these online chemical (volatiles) monitoring after roasting (Ruosi et al., 2012). The simultaneous discrimi- approaches have proved to be expensive, time-consuming, nation of roast degree and variety is also possible by identifying laborious, and hard to adopt at industrial levels. So, an analogous 2,3-butanedione, 1-methylpyrrole, 2,3-pentanedione, furfural, approach, called an electronic nose, has also been proposed in pyridine, 1-furfurylpyrrole, 5-methylfurfural, guaiacol, and the literature, with applications in the monitoring of optimal acetylpyrrole. Ruosi et al. (2012) also directly correlated aroma roasting, and as a substitute for the cupping test. An electronic compounds with roast degree through roast color, by applying nose is composed of an array of sensors which produce an array a multiple linear regression (MLR) model which shows a good of signals (conductivity or resistance) to identify/quantify specific linear relationship (r = 0.9387) between the analytes detected and classes of volatiles for which they are made sensitive (Brattoli roast color. PCA proposed 13 aroma indices (that is, pyridine/ et al., 2003). Many studies have successfully applied e-nose 5-methylfurfural, pyridine/furfural, 2-methylfuran/2,3-pen- technology for distinguishing and classifying ground and instant

r C 2018 Institute of Food Technologists® Vol.17,2018 ComprehensiveReviewsinFoodScienceandFoodSafety 1205 Farm to consumer . . . coffee, commercial coffee blends, coffee mixtures, varieties, pure the first to second stage is also important for the development of varieties, and espresso coffee aroma from different origins and darkness in the coffee roasting process. The textural characteristics roasting degrees (Romani, Cevoli, & Fabbri, 2012). Romani and of coffee beans have been elaborated by a force–deformation curve coworkers also compared e-nose technology with an artificial which shows a sudden and well-defined fracture point, leading to a neural network for the prediction of roast degree, on the basis of decrease in breaking force and stress at fracture (at which emission the physical characteristics of roast degree. The authors clearly of moisture, volatiles, and gases occurs, resulting in porous struc- declared the successful prediction of roast degree using e-nose tures, also called roast loss) at high-yield roasting conditions, while technology and emphasize its importance as an offline system prolonged dehydration conditions lead to brittleness of the prod- for integration of quality checks in the coffee industry. The uct. A further increase in temperature beyond the fracture point success rate of using electronic noses with cross-validation was results in a further drop in breaking force, which is an indication found to be 80% to 89.9%, with a need to lessen the number of of progressive reduction in the strength of beans, and an advance sensor arrays (Falasconi, Pardo, & Sberveglieri, 2005). Later, the towards roast loss (Pittia et al., 2001). Like color kinetics, roast loss correlation of signals from an electronic nose with a reduced array kinetics is also a 2-stage phenomenon. Roast loss in the first phase of sensors for roast loss, L∗a∗b∗, moisture, and density was set is characterized by a rapid loss of water, and weight loss in the with the help of complex algorithms on large datasets (Romani second phase is mainly attributed to the evolution of volatiles and et al., 2012). But still, the robust and reliable applicability of CO2 (Wang & Lim, 2014). The roast loss occurring in the second electronic noses has been found to be limited due to particular stage is more sensitive to temperature than that in the first phase. features like a large number of (descriptors) arrays (with respect The similarity in the kinetic curves of L and roast loss suggest to samples) having the same production technology, which can their relevance. Different trends in breaking force decrease seen create the chance of correlations of fitting, leading to a poor at varying roasting conditions correspond to varying bean-drying predictability model (Roberts & Cozzolino 2016). Recently, methods. A lower breaking force may be reached by beans during Giungato, Laiola, and Nicolardi (2017) correlated the response of dehydration, which may result from irregular bean shape and size, an electronic nose for brightness and mean density via generalized and structural variations within each group of coffee beans. The least squares regression with stepwise backward selection of reduction in bean strength during dehydrating conditions appears predictors. By using both brightness and density as indicators of to be less affected by a decrease in density, whereas textural bean roasting degree, the backward selection methodology showed the strength variations during roasting seem to be related to a decrease same correlations with the response of the chosen sensors, with in density. So, we can deduce that dehydrating beans with the same R2 = 0.99 to 0.9994 (Giungato et al., 2017). So, the combination density may possess different breaking forces, while roasting beans of an electronic nose with least squares regression using stepwise with normalized density may show different texture properties. backward predictors can so far claim to be a cheap, reliable, and Roasted beans also show a reduction in strain and work at fracture feasible online roasting monitoring system at industrial level. But, point as compared to green beans, which is also an indication of before the application of electronic nose technology in industry, textural changes, whereas dehydrating beans have relatively higher research work still needs to focus on improving its sensitivity and strain and work due to which they have a tough, brittle, and frag- resolution and to establish relationships between observed signals ile texture. The greater water content, acting as a plasticizer, of and relevant quality parameters. dehydrating beans is probably the main reason behind the brit- Among physical and mechanical variation studies, the relation- tleness, as a high water content contributes towards the stiffness ship of mechanical and textural properties of coffee beans at de- and toughness of coffee beans until a glassy state is reached. Af- hydrating (90 to 105 °C) and high-yield roast conditions (190 to ter the glassy state, a fracture point occurs in the roasting, which 240 °C, 4 min; Italian-style roasting) in a laboratory air-circulating leads to a drop in density, stress, breaking force, and overall bean oven was first presented by Pitia et al. (2001). After dehydration integrity, allowing vast textural changes in the bean (Pittia et al., (moisture < 2 g/100 g), a more rapid loss of mass and density oc- 2001). Among other textural changes, porosity and pore structure curs at high-yield roasting conditions than at dehydrating condi- also have a critical impact on final product quality. Textural fac- tions, which leads to an increase in the volume and porous texture tors like pore structure and porosity become more obvious after of beans. The change in hue angle is also quicker and steeper for roasting and storage, due to their controllability over mass trans- high-yield roasting conditions (Pittia et al., 2001). Color kinetic fer, gas desorption and degassing. Migration of oil droplets, loss studies under isothermal conditions also show the same results, of volatiles, and oxygen accessibility during storage also relate to with a quicker decrease in lightness (L) at higher temperature, and porosity and pore structure. The extent of this porosity and mi- a decrease in lightness for degree of roasting, temperature, and time cropore systems is also reported to depend upon roasting (Ortola, (Wang & Lim, 2014). The decrease in lightness is a 2-stage phe- 1998; Schenker, Handschin, Frey, Perren, & Esche, 2000). Many nomenon, where the rate of change of lightness is more sensitive authors have studied porosity and pore structure via image analy- to temperature in the later phase. The decrease in rate of change in sis, scanning electron microscopy, and mercury porosimetry while the later phase may be due to slowed Maillard reactions caused by roasting at different temperatures to the same roast level (Figure 1) dehydration and the transition of coffee beans from a rubbery to (Schenker et al., 2001). These authors described that the density of a glassy state (Craig, 2001). Further, most low-molecular-weight coffee relies on bean porosity. The unanimous description of these carbohydrates degrade quickly during the first stage, and only studies unveiled the fact that high temperature-roasted (HTST) polysaccharides remain a source of nonenzymatic browning in the coffee possesses a significantly higher pore area than LTLT-roasted later stage (Oosterveld, Voragen, & Schols, 2003b). Besides L,the coffee. However, the documented pore size in roasted coffee beans values of a (+a:redand−a: green) and b (+b: yellow and −b: may vary from 20 to 100 nm (Schenker et al., 2001). Variation blue) increase during the early phase of roasting to +15 and +30 in pore size may depend upon variations in gas/volatiles pressure maxima (red and yellow), respectively, but as roasting advances, the and water evaporation rate, which in turn may vary depending a and b values decrease and stabilize to the same hue regardless of upon roasting conditions. Despite the supportive role of water temperature (Wang & Lim, 2014). Transition of coffee beans from vapor and gases on expansion, structural resistance is also a factor

r 1206 ComprehensiveReviewsinFoodScienceandFoodSafety Vol.17,2018 C 2018 Institute of Food Technologists® Farm to consumer . . .

Figure 1–Oil migration process due to merging of small oil droplets in the presence of extensive microspore network. (A and B) SEM micrograph of green coffee beans cells. (C and D) SEM micrograph of roasted coffee beans intruded with mercury. (E) Cryo-SE micrograph of HTST roasted coffee beans after one day of storage. (F) TEM micrograph of cell wall of partially roasted beans. The continuous dark line represents the middle lamella separating the 2 adjacent cell walls and cytoplasm of 2 adjacent cells. The lines perpendicular to middle lamella are plasmodesmata channels through the walls. Sources: Adapted from Schenker et al. (2000). to be overcome for this expansion which is controlled by a shift ing which occurs between 220 and 250 °C. Bigger gaps in the in the balance towards expansion due to dehydration and the glass kinetic curves happened between 220 and 260 °C, showing the transition cycle (Figure 1). mass conversion of solids (that is, carbohydrates, lipids, proteins, All the studies about roasting stated so far have provided us with and so on) into volatiles and CO2. At higher temperatures (260 a basic understanding of the relationship between roasting param- to 300 °C), massive breakage of oleosomes (spherosomes rich in eters and bean characteristics, but, unfortunately, they all defined lipids that are found in plants), cracking, and swelling of beans fol- roasting parameters and levels differently. So, recently, a broader lowed by oil migration and a substantial drop in mass were clues to and inevitable research study, further supporting and elucidating the initiation of overroasting. Of the 2 roasting stages, the first was the roasting data discussed above, was published by Pramudita et al. found to be important for the generation of desirable volatiles and (2017) who pointed out the relationship between the dehydration color. Although unreliable, color is still considered a useful tool process, roasting time, and roasting temperature. This study de- to determine quality parameters and roasting degree. Despite the noted that a reduction in mass started even before the start of fact that different temperatures give different colorations, a longer the dehydration process, but a major reduction in mass was seen time at a lower temperature is required for the same level of color when the moisture content dropped to 2%. At 140 °C, only a generation as that at a higher temperature, except for 260 and 5% reduction in mass was noted besides the dehydration process, 360 °C where the final dark color is the same (Pramudita et al., which indicates that the reactions for aroma and flavor generation 2017; Wang & Lim, 2014). The exact reasons for these differences had not yet started. Cracking followed by the emission of volatiles in color generation still need to be investigated, but 2 possible and CO2 occurred at 14% to 15% roast loss, which corresponds reasons are given in the literature. The first is that the operation to previous findings (Schenker et al., 2000; Wang & Lim, 2014). of pathways involving melanoidins could be different under dif- Roast loss of 15% is also sometimes referred to as medium roast- ferent reaction conditions (Tressl, Nittka, & Kersten, 1995), as

r C 2018 Institute of Food Technologists® Vol.17,2018 ComprehensiveReviewsinFoodScienceandFoodSafety 1207 Farm to consumer . . . low-temperature roasting conditions favor low-molecular-weight and nondefective coffee beans and to predict the share of defec- melanoidins and vice versa (Bekedam, Loots, Schols, van Boekel, tive coffee beans both in crude-roasted beans and in their ground & Smit, 2008). However, some papers have also claimed that form (Craig et al., 2015; Mendonca et al., 2010; Santos, Sarraguc¸a, high-molecular-weight melanoidins may also result from the accu- Rangel, & Lopes, 2012). Mendonca and colleagues described the mulation of low-molecular-weight melanoidins during prolonged physical differences between defective and nondefective beans af- roasting (Fogliano, Monti, Musella, Randazzo, & Ritieni, 1999). ter roasting. According to them, defective coffee beans are smaller The second is that the melanoidin population may vary with varia- than nondefective ones due to less expansion, and hence can be tions in roasting conditions. Different melanoidin populations pos- separated by sieving in the case of Arabica coffee. Color variations sess different hue characteristics, so even if their eventual yield/sum are feasible for separating only black and dark sour defective coffee is the same, it is not necessarily the case that these melanoidins will beans based on electronic sorting (Mendonca et al., 2012). Other produce the same coloration. This fact shows us that biochemical authors employed mid- and near-infrared spectroscopy (FTIR and activities involving color changes are less dependent on tempera- NIR) successfully by using elastic net models and partial least ture changes and are not related to aroma- and flavor-producing squares regression analysis as predictive tools to visualize discrete reactions. So, roasting degree can also not be determined on the spectral bands linked to the correct classification of defective and basis of color, since roasted beans of the same color can have differ- nondefective coffees. ent roast loss and compound profiles. The kinetics of roast loss and Relationship of volatiles and roasting factors. Roasting is a brightness of coffee beans are similar. At 100 to 140 °C, brightness crucial step in the generation of characteristic volatile flavor and of beans is enhanced by removal of moisture, and then it is reduced aroma compounds from a combination of pyrolysis, Maillard, and continuously upto 260 °C, at which point minimum brightness caramelization products and the compositional components of is achieved. At/after that temperature, beans become overroasted coffee beans. The number of these volatile compounds identified and dark. This is due to the fact that beans tend to darken faster has reached up to 1000, among which 30 to 50 volatiles have at temperatures at which there is a rapid loss of solid mass due to an impact on quality. Many authors have identified the various pyrolysis (that is, 260 to 300 °C) (Pramudita et al., 2017). volatiles produced during roasting as a function of degree of Emission gases. The release of CO2 during roasting and cool- roasting, while others have investigated their formation. Gretsch, ing/storage is also linked to the roasting process. CO2 makes up Sarrazin, and Liardon (1999) comprehensively correlated coffee 87% of all the gases released during roasting and cooling/storage. quality attributes with the concentration of aroma compounds Of this 87% CO2,afractionofCO2 is released during roasting, at different degrees of roasting. They detected a restricting while the remaining fraction is bound to the polar sites of beans correlation between the concentrations and their corresponding and dissolved in residual moisture and oil which are released dur- sensory intensities, and they also noted the masking effect of ing cooling/storage (Geiger, Perren, Kuenzli, & Escher, 2005). some compounds formed during the terminal stages of roasting. The entrapped CO2 fraction makes up only 2%, while complete Mayer et al. (1999) also looked into the effect of provenance degassing of CO2 occurs in up to 6 weeks (Illy & Viani, 1995). and degree of roasting on the odorants in 3 coffee varieties, The degassing process is also highly influenced by process and stor- and they found 2,3-butanedione, 3-isobutyl-2-methoxypyrazine, age conditions, structural variations, and roasting. The evolution 2,3-pentanedione, 4-vinyl- and 4-ethylguaiacol, 4-hydroxy- of CO2 and residual CO2 are also roasting degree- and time- 2,5-dimethyl-3(2H)furanone, 2-furfurylthiol, 3-methyl-2- dependent. The evolution rate is also higher in HTST than LTLT buten-1-thiol, and 3-mercapto-3-methylbutyl formate as major roasting conditions. However, during storage, higher degassing discriminating compounds for coffees of different origins. The values are noted for LTLT-roasted beans, and lower cumulative authors also proposed conflicting results showing variations in values of degassed CO2 are observed for HTST-roasted coffee propanal, guaiacol, methanethiol, 2(5)-ethyl-4-hydroxy-5(2)- upto 63 days of storage, from which point the cumulative value methyl-3(2H)furanone, 4-ethylguaiacol, 2-furfurylthiol, and of degassed CO2 levels off (Geiger et al., 2005). Residual CO2 3-methyl-2-buten-1-thiol odorants with different roasting levels. increases with increasing roasting temperature and roasting time Two volatiles, namely 2-furfurylthiol and guaiacol, develop in a until cracking of coffee beans (also called second cracking) occurs, continuous way in Colombian and Kenyan coffees, and greatly allowing degassing of CO2 and hence lowering of residual CO2. increase with increasing degree of roasting. 2,3-Pentanedione and Regardless of roasting conditions, the amount of residual CO2 2,3-butanedione form to a maximum concentration for a medium can also be assessed from the L value up to medium-roast degree degree of roasting and exhibit lower concentrations in dark- (Wang & Lim, 2014). roasted coffee beans (Mayer et al., 1999). Different roasting process Defective beans. The addition of defective beans to normal- technologies have also been studied to investigate the formation quality beans can change the sensory status of coffee from “strictly of different flavors; maximum formation of aroma compounds soft” to hard, and inclusion of 19.5%, 16.4%, and 14.3% immature, occurs up to medium roasting (Schenker et al., 2000). The roast- sour, and black beans, respectively, is enough for this transforma- ing process technologies observed include (a) high-temperature, tion (Craig, Franca, Oliveira, Irudayaraj, & Ileleji, 2015). Some short-time (HTST); (b) low-temperature, long-time (LTLT); (c) authors also deduced that addition of only 2.5% immature beans high temperature with a reduced final stage (HL); (d) continuous to nondefective beans can cause the rejection of 30% of produce temperature increase from low to high (LHC); (e) preheating by cuppers due to unacceptable quality parameters. However, in temperature with subsequent LHC process (PLHC); (f) preheat- the case of espresso coffee, the addition of only 1% defective beans ing temperature, high temperature at intermediate stage, and can result in a beverage with an astringent and metallic taste. This reduced temperature at final stage (PHL). The first isothermal con- may be due to the fact that defective beans have a wider spectrum ditions, that is, HTST, generate 11 major odorants, whereas LTLT of volatiles than nondefective beans. Discrimination and quan- results in the production of only six. Further, aroma concentrations tification are the main keys to distinguishing between defective in both isothermal processes tend to decrease on termination of and nondefective coffee after roasting. So, various authors have the process at a medium degree of roasting. So, HTST conditions developed quantitative models to distinguish between defective are not a guarantee of producing excessive amounts of aroma

r 1208 ComprehensiveReviewsinFoodScienceandFoodSafety Vol.17,2018 C 2018 Institute of Food Technologists® Farm to consumer . . . compounds, although they do favor the production of some and 12 miscellaneous). The concentrations of sweet, bread-like, odorants in excess (2-furfurylthiol, 2,3-butanedione and caramel, and coffee aroma-endowing furfural derivatives (that methyldihydro-cyclopentapyrazine) by reducing the production is, furfural, furfuryl alcohol, and 2-furfurylthiol), caramel-fruity of others when compared to LHC (Schenker et al., 2000). flavor-bestowing DMHF and DMP (2,5-dimethyl-4-hydroxy- Overall, the greatest quantities of aroma compounds are achieved 3(2H)-furanone and 3,5-dihydroxy-2-methyl-4H-pyran-4-one), with LHC, but these aroma compounds are not necessarily related astringent phenols (that is, 2-methoxyphenol and 2-methoxy- to superior sensory quality. LHC roasting also induces more severe 4-vinylphenol), furans and a single furfural derivative named degradation of some aroma compounds. A comparison of HL 5-hydroxymethylfurfural decrease significantly in dark-roast and LHC also shows that a high initial temperature followed by coffees (Maga, 1992; Nakama, Kim, Shinohara, & Omura, 1993; low temperature stages is not fruitful for the production of large Silwar et al., 1993). On the opposite side, the net content of amounts of aroma compounds, and a high temperature is needed pyridines (smoky/burnt odor), pyrroles (roasted/toasted flavor), for the intermediate and final roasting stages. Comparison of γ -butyrolactone (buttery/coconut flavor), and catechol (astrin- PLHC and LHC also shows that a predrying stage is not effective gent) increases manyfold on increasing the roasting temperature for producing a greater content of aroma compounds; however, a from 230 to 250 °C for 12 to 21 min (Akiyama et al., 2008; Moon reduced-temperature final stage results in a shift in sensory profile & Shibamoto, 2009). Reactions involving pyridine formation without affecting overall aroma quantities (Schenker et al., 2002). are time-dependent, whereas pyrrole and pyrazine formation After this pioneer work, Baggenstoss et al. (2008a) also explored reactions are temperature-dependent (Baggenstoss et al., 2008a). specific time–temperature roasting profiles (HTST and LTLT, Prolonged roasting at a temperature above 240 °C generally leads 180 to 260 °C and time 160 to 1170 s) up to the last degree to overroasting in which degradation reactions dominate forma- of roasting (L = 21) and found significant variations in aroma tion reactions. The dominance of degradation reactions results and flavor compounds with respect to specific time–temperature in decreased productions of major sulfide, Strecker aldehyde, roasting profile variations. Most variations occur in specific classes α-diketone, pyrrole, and pyrazine odorants, but the contents of of compounds, that is, sulfur compounds, pyridines, pyrazines, pyridines, hexanal, dimethyl trisulfide, 2,3,5-trimethylpyrazine, and Strecker aldehydes, which bring about profound changes and 2-ethyl-3,5-dimethylpyrazine rise and level off at different in the physical and sensory attributes of the resulting beverages. time intervals (Baggenstoss et al., 2008a). After roasting condi- Most sulfur compounds (that is, methanethiol, dimethyl sulfide, tions, roasting speed (hot air flow) also brings about considerable dimethyl trisulfide, 3-mercapto-3-methylbutyl formate, and variation in the profile of volatile compounds, especially furans, 2-furfurylthiol), which originate from sulfur-containing amino in roasted beans and their beverages. Roasting speed has been acids, increase during HTST conditions compared to LTLT found to affect the concentrations of 4 classes of volatiles, namely conditions, and their maximum level is detected in medium pyridines, pyrazines, pyrroles, and ketones (Petisca et al., 2013). roasts. Strecker aldehydes and ketones (that is, methylpropanal, The content of roasted/toasted flavor-endowing pyrroles and 2-methylbutanal, 3-methylbutanal, and α-diketones) show similar pyrazines increases at fast roasting speeds, whereas pyridines behavior, except for hexanal whose peak concentration is found and ketones are topmost under the slow and medium roasting at 220 °CandL = 30 to 40 both in HTST and LTLT profiles. speed conditions (Moon & Shibamoto, 2009; Petisca et al., Smaller amounts of α-diketones (that is, 2,3-butanedione and 2,3- 2013). Amazingly, different roasting speed does not change the pentanedione) are found in longer roasts, whereas the maximum total volatiles concentration, but disturb the net contents of level of α-diketones is detected for extended HTST treatment individual volatile classes. PCA of these roasted beans and their (Baggenstoss et al., 2008a). Among the above-stated 1000 volatile espresso brews shows that furan and pyrrole content increases compounds, only 300 are heterocyclic compounds. These hete- from ground beans to espresso while a negative relationship is rocyclic compounds may include pyrroles, oxazoles, imidazoles, seen for pyrazines and ketones. As furans are the largest class furans, thiazoles, thiophenes, and pyrazines, which impart roasted, of volatiles (Moon & Shibamoto, 2009) both in ground beans toasted, and coffee flavors to beverages (Czerny et al., 1999). and espresso, the furanic derivatives 2-methylfuran, 2-furfural, Different roasting conditions influence the concentrations of furfuryl formate, 5-methylfurfural, and furfuryl acetate increase these volatile compounds, which impart specific coffee flavors enormously in espresso, whereas 2-furfuryl alcohol suffers a severe and aromas, produced under specific roasting conditions. Volatile decline (Petisca et al., 2013). compounds (that is, pyridines and pyrroles) which arise from the Maillard reaction of amino acids and sugars tend to arise from Compositional compounds: precursors of flavoring and light to intense roasting conditions (Moon & Shibamoto, 2009), aroma compounds whereas odorants (that is, furanones and furfural derivatives) In addition to fermentation products, water, carbohydrates, arise from the compositional material of green coffee beans and lipids, proteins, CGA, caffeine, and so on are the major com- decrease with the transition from mild to intense roasting condi- positional components of coffee beans and are the precursors of tions. Also, chemicals (for example, phenols and lactones) arising coffee flavor and aroma compounds. There are 2 water sources from the degradation of CGA tend to accumulate under intense in coffee beans during roasting: the first is initial water content roasting conditions (Moon & Shibamoto, 2009). Therefore, a and the second is chemical reaction prouduct water. Evapora- difference in roasting conditions brings about a difference in the tion of initial water content and formation of chemical water formation of these volatiles, and, hence, modulates the aroma and are highly temperature-dependent reactions. Evaporation of ini- flavor characteristics of beverages. Moon and Shibamoto (2009) tial water content occurs and is completed during the dehydration also studied the influence of these different roasting conditions stage of roasting; after that, formation and evaporation of chemical (light: 230 °C for 12 min, medium: 240 °C for 14 min, city: water starts during the roasting stages. Analysis of exhaust air dur- 250 °C for 17 min, and French: 250 °C for 21 min), and they ing roasting reveals that it is composed exclusively of chemical re- identified 52 volatile compounds (that is, 4 pyridines, 9 pyrazines, action water. The rate of moisture loss is higher in HTST, whereas 5 pyrroles, 7 furanones, 7 furans, 4 cyclopentenes, 4 phenols, more moisture evaporates during LTLT (Geiger et al., 2005). A

r C 2018 Institute of Food Technologists® Vol.17,2018 ComprehensiveReviewsinFoodScienceandFoodSafety 1209 Farm to consumer . . . temperature range of 180 to 200 °Cisrequiredforthefor- mation and loss of chemical reaction water. Chemical reaction Polysaccharides (insoluble) water amounts to 36% for LTLT and 41% for HTST (Geiger Hydrolysis de-branching & de-esterificaon et al., 2005). The initial water content of beans determines roasting behavior, and a high initial water content (>10% to Polysaccharides (soluble) Proteins/peptides/amino acids 11%) is believed to increase the rate of dehydration and (or- Hydrolysis Addion of charge ganic) roast loss and to give a less uniform roast coloration and elevated density. These variations in roasting behavior are Oligosaccharides Milliard products more obvious in light roasting stages and tend to diminish at dark roasting levels. A high initial moisture content leads to Hydrolysis lower production of important volatile aroma compounds (that Monosaccharaides is, N-methylpyrrole, 2-furfurylthiol, 2,3,5-trimethylpyrazine, 2,3- Aroma components + butanedione, 2,3-pentanedione, dimethyl sulfide, dimethyl disul- organic acids fide, 3-mercapto-3-methylbutyl formate, pyridine, and 2-ethyl- Fufural-like compounds 3,5-dimethylpyrazine) at light-roast levels, whereas attenuated variation in all volatiles, except 2-ethyl-3,5-dimethylpyrazine, and 2,3,5-trimethylpyrazine is seen at dark roasting degrees (Baggen- Figure 2–Schematic diagram of the polysaccharide conversion reactions stoss et al., 2008a, 2008b). Some authors have also cited the im- occur during roasting and extraction of coffee (Adapted from Oosterveld pact of steam treatment, under pressure (2 bar), of green beans et al., 2003). before roasting; they exhibit smaller amounts of caffeine (10%), saccharose, and free amino acids, while the levels of 3-CQA, 3,4- et al., 2002). Also, based on solubility and extractability indices, the DCQA, and monosaccharides increase. Steam treatment is also whole class of polysaccharides can be divided into unextractable a cause of extraction/decomposition of precursor compounds, and extractable polysaccharides, oligomers and monomers (Redg- due to which a significantly reduced amount of various odor- well et al., 2011). Extractable polysaccharides contribute towards ants (that is, trigonelline, dimethyl sulfide, dimethyl trisulfide, the organoleptic properties of coffee such as viscosity, sweetness, 2-methylbutanal, dimethyl disulfide, 2,3,5-trimethylpyrazine, 2- mouthfeel, and cream/crema/foam stability in espresso coffee. furfurylthiol, 2-methylbutanal, N-methylpyrrole, 4-vinylguaiacol, Extractable polysaccharides are mainly composed of soluble pyridine, and 2-ethyl-3,5-dimethylpyrazine) is detected both at arabinogalactans of which the major part is converted into sugar light-and dark-roast levels (Baggenstoss et al., 2008a, 2008b). degradation products. There are a further 2 classes of extractable However, the literature is devoid of sensory analyses of data from polysaccharide, that is, high-molecular-weight arabinogalactans high moisture- and steam-treated roasted beans. Besides water, with a low degree of arabinosylation, and lower-molecular-weight other compositional components act as substrates for the produc- arabinogalactans with a high degree of arabinosylation (Oosterveld tion of different classes of volatile flavor and aroma compounds et al., 2003a, 2003b; Redgwell et al., 2002). Recovery studies during roasting. All the volatile compounds generated from these from brews have revealed that unextractable polysaccharides substrates during different roasting degrees have been studied are chiefly composed of insoluble galactomannans, and (light) (Wang & Lim, 2014). The rate and amount of all volatiles increase roasting conditions increase the solubility of galactomannans by with the transition from light-roast to dark-roast level, except debranching and converting (almost 12%) them into extractable for 2-methylfuran, 2,5-dimethylfuran, and pyridine which show polysaccharides, oligosaccharides, monosaccharides, and degra- the opposite trend. Three odorants, namely 2,3-butanedione, dation products (Figure 2) (Oosterveld et al., 2003a, 2003b; 2,3-pentanedione, and furfural, appear mainly at the light-roast Redgwell et al., 2002). Moreover, the scant recovery of oligomers level, disclosing only their contribution to light-roast aroma. and monomers from brews also reveals their quick conversion to On the opposite site, the compounds 2-methylfuran, pyridine, sugar degradation products under all roasting conditions. The 3 methylpyrazine, furfuryl acetate, dihydro-2(3H)-furanone, and 2- widely used roasting conditions (light: 11% roast loss, medium: furan ethanol only increase considerably during dark-roasting, 15% roast loss, and dark: 22% roast loss) also bring about mean- which suggests their possible characteristic role in developing dark- ingful differences in the extent of unextractable and extractable roast aroma (Wang & Lim, 2014). Nonetheless, this recent work polysaccharides, oligomers, and monomers (Oosterveld et al., is devoid of information regarding identification of the origin of 2003a). Light-roasted coffee has unextractable polysaccharides all these volatiles, and does not show sensory analyses. (36% w/w) as the predominant carbohydrate as compared Among carbohydrates, 2 kinds of polysaccharide predominate, to extractable polysaccharides (21% w/w) and oligomers + that is, type II arabinogalactans and galactomannans (Fischer monomers (11% w/w). Intensifying the roasting conditions from et al., 2001). Roasting lessens the degree of branching in complex light to dark surprisingly causes an increment in the conversion of insoluble polysaccharides and converts the insoluble form of polysaccharides to sugar degradation products (38% w/w) due to polysaccharides to soluble forms and other products, that is, loss of extractable polymeric arabinogalactans, while reducing the monomers and furfural-like compounds, which also react with content of unextractable polysaccharides and slightly increasing Maillard reaction products (produced from proteins, amino acids, oligomers and monomers. These outcomes suggest that the amines, peptides and organic acids) to generate featured aroma degradation of oligomers and monomers is much faster than the and flavor compounds (Figure 2) (Oosterveld et al., 2003a, 2003b; reactions of extractable polysaccharides to oligosaccharides. In ad- Redgwell et al., 2002). In this whole process, variations in roasting dition to roasting conditions, more intense extraction conditions conditions definitely influence the extent of debranching, degra- (90 °C, 1 hr to 170 °C, 20 min) have also been found to be helpful dation and solubility of polysaccharides, followed by oligomer in converting part of the insoluble/unextractable polysaccharides and monomer reactions with Maillard reaction products and, con- into soluble/extractable polysaccharides. This might be due to sequently, variations in the aroma and flavor of coffee (Redgwell structural changes in galactomannans during intense roasting and

r 1210 ComprehensiveReviewsinFoodScienceandFoodSafety Vol.17,2018 C 2018 Institute of Food Technologists® Farm to consumer . . .

Figure 3–(A) Lactone formation from chlorogenic acid. (B) Structure of lactones and their precursors while roasting. Source: Farah, Paulis, Trugo, and Martin (2005b). extraction, in such a way that extraction becomes easier (Figure 2) increase to twice their original amounts (Dawidowicz & Typek, (Oosterveld et al., 2003a, 2003b; Redgwell et al., 2002). 2017; Farah et al., 2006a; Ludwig et al., 2014). Loss of 5-CQA is a Irrespective of variations in genetic, geographical, and secondary sign of improvement in cup quality, as a higher level of CQAs (es- processing technology, phenolic acids, including CGA and their pecially 5-CQA) and their oxidized products reduces cup quality derivatives (that is, heterocyclic aromatic hydrocarbons –HAHCs, and increases Rio off-flavor (Farah et al., 2006a). Elevated levels chlorogenic lactones – CGLs, melanoidins, and so on), also form of CQAs (especially 5-CQA) in coffee brews are also markers of or are eliminated during the primary processing of coffee (Dawid- the inclusion of defective and immature beans in processing. Farah owicz & Typek, 2017). The existence of these metabolites directly et al. (2006a) also specified a level of 5-CQAs from 2 to 13 g/kg influences the quality of coffee brews and contributes to changes (from dark to light roast) for good cup quality, but later findings in color, aroma, and flavor during roasting (Farah et al., 2006a). of Zanin, Corso, Kitzberger, Scholz, and Benassi (2016) described The main groups of CGA found in green and roasted coffee beans the cup quality of coffee brews as independent of levels of 5-CQA, are CQAs (76% to 82%), FQAs (5.2% to 6.2%), and di-CQAs since the authors noted superior cup quality in samples having a (15% to 18%), with at least 3 isomers for each group, and one of higher 5-CQA content than specified. However, this discrepancy the main types of CGA isomers present in green and roasted cof- in results also clearly shows the involvement of other postharvest fee is CQAs, representing from 76% to 82% (Farah & Donangelo, processing factors in obtaining good cup quality. Longer periods 2006b). Roasting degree has been found to affect the different of roasting (that is, medium-dark and dark) decrease the levels isomers of CGA differently. Overall total chlorogenic content in- of both total CGAs (up to 5%) and their derivative CGLs (up creases during the initial moments of roasting. After that, at 15% to 20%) (Dawidowicz & Typek, 2017; Farah et al., 2006a). The roast loss (light-medium roast), the level of 5-CQAs decreases major CGLs identified in the literature belong to 4 classes of lac- considerably, while the levels of 3-CQAs, 4-CQAs, and FQAs tones, namely caffeoyl-1,5-lactones (CQLs), feruloyl-1,5-lactones

r C 2018 Institute of Food Technologists® Vol.17,2018 ComprehensiveReviewsinFoodScienceandFoodSafety 1211 Farm to consumer . . .

(FQLs), p-coumaroyl-1,5-lactones (pCoQLs) and 3,4-dicaffeoyl- essary to correlate the levels of HAHCs with those of CGAs 1,5-lactones (3,4-di-CQL). Among CQLs and FQLs, 3-CQL, and caffeine while discussing the influence of roasting degree on 4-CQL, 3-FQL, and 4-FQL are the dominant lactones, detected caffeine and CGAs. The most-discussed HAHCs in the literature at their maximum amount up to 14% roast loss. Robusta coffee are benzo(a)pyrene, benzo(b)fluoranthene, benz(a)anthracene, and contains more CGLs than Arabica at all roast levels. pCoQLs are benzo(k)fluoranthene (EFSA, 2008). These HAHCs have also the least-detected lactones, with usual amounts of 30 to 70 and 50 been classified as most potent carcinogen compounds; therefore, to 60 mg/100 g for green Arabica and Robusta coffees, respec- it is also indispensable to know their levels in roasted coffee beans tively (Clifford, 1997). pCoQLs and 3,4-diCQLs follow the same subjected to different roasting levels. Benzo(b)fluoranthene and trend during roasting as CQLs. Figure 3AB described the possible benz(a)anthracene are the most-detected HAHCs in all roasting mechanisms of CGA lactones and their precursors formation in conditions, while diminishing levels of benzo(k)fluoranthene and detail. benzo(a)pyrene are found in all roasting conditions (Houessou For espresso coffee, Kuceraˇ et al. (2016) recently published et al., 2007). Moreover, light roasting conditions favor the pro- the results of roasting impact on 4 bean constituents, that is, duction of low-molecular-weight HAHCs (3 to 4-ring structure), CGAs, CGLs, atractyloside derivatives (ATRs), and melanoidin whereas high-molecular-weight HAHCs are detected at dark-roast reaction products (MRP-Ms), and proposed 4 marker ratios (that levels (Houessou et al., 2007). Variety-wise, the highest average is, ࢣdi-CGAs/ࢣmarkers, ࢣCQLs + FQLs/ࢣmarkers, ࢣATR- levels of HAHCs are observed for Robusta at dark-roast levels, IV + ATR-V/ࢣmarkers, and ࢣMRP-Ms/ࢣmarkers) specific to whereas the opposite trend is noted in Arabica coffee, although each class of compound for the identification of roast degree. some authors have also documented a lack of correlation between Kuceraˇ and team deduced very similar results for CGAs to those HAHC level and roasting degree (Tfouni et al., 2012, 2013), previously published by Farah et al. (2006a) and Sanchez-Bridge which may be due to a high degree of variability in process- et al. (2016) for nonespresso and soluble coffees. For CGAs, the ing, coffee cultivar/variety, and roasting degree. After HAHCs, content of CQAs, FQAs, p-coumaryl-quinic acids (pCQAs), and there is also strong evidence of the inclusion of intact CGA di-CQAs decreases with prolonged roasting and dehydration, with molecules in the composition of Maillard end-products called the formation of corresponding lactones (Ludwig et al., 2014; melanoidins (Bekedam et al., 2008). Brown/dark-colored nitroge- Kuceraˇ et al., 2016). Among lactones, the amount of CQL-1, nous melanoidins, which are 25% of the dry matter of coffee CQL-2, and FQL-2 rises until dark roast is obtained, at which brews, are well known for their antioxidant, color-contribution, point a slight fall in their quantity is observed due to degradation. and flavor-binding capacity. These compounds are formed from Similar to that observed for some classes of polyphenols and lac- CGAs and polysaccharides (galactomannans) and arabinogalactan tones, ATR (ATR-I and ATR-II) and melanoidin markers are also proteins during roasting, as a result of Maillard and caramelization enhanced with roasting degree, with a maximum amount detected reactions. This hypothesis was further strengthened by the find- for dark roast (Kuceraˇ et al., 2016). ings of Bekedam et al. (2008) who noted the immediate loss of So, we can deduce that around 50% of the original CGAs are lost sucrose and proteins/amino acids, formation of covalent bonds during the entire roasting process and about half of the remain- between CGAs and proteins/amino acids, and the incorpora- ing CGAs are consumed in the production of quinic acid and tion of CGA, glucose, and melanoidin molecules during roast- hydroxycinnamic acids such as caffeic and ferulic acid (Crozier, ing. Roasting influences the extent of the whole chain of this Jaganath, & Clifford, 2009; Dawidowicz & Typek, 2017; Maatta, process, and various roasting degrees affect the yield of differ- Eldin, & Torronen, 2003). The remaining quarter of CGAs gen- ent melanoidin fractions (low, intermediate, and high molecu- erate CGLs and other metabolites from chemical pathways, that is, lar weight). Roasting degree significantly increases the levels of acyl migration, dehydration, epimerization and transesterification melanoidins, especially prolonged roasting which is the main cause with acetic, quinic, and shikimic acids during roasting, as eluci- of an increase in the levels of high-molecular-weight melanoidins dated by Jaiswal, Matei, Golon, Witt, and Kuhnert (2012) and (Bekedam et al., 2008), whereas intermediary-molecular-weight Dawidowicz and Typek (2017). Those authors used targeted and melanoidins dominate during light-roast conditions. Identification nontargeted approaches to find out the fate of the remaining quar- of melanoidin-type fractions derived from the reaction of CGAs ter of CGAs during roasting. In the targeted approach, synthetic and other coffee components was carried out for the first time standards were employed to compare and quantify the possible by Jaiswal et al. (2012) using a nontargeted ESI-FT-ICR-MS CGA derivatives detected. Acyl migration (that is, the migration approach. This nontargeted approach generated molecular mass of a hydroxycinnamoyl group from any of the 4 OHs of quinic acid formulas from the mass list with their S/N ratios. Of a total of 300 to another one) within CGAs was confirmed by the detection and molecular formulas, 127 were found to be identical to the molec- identification of 1-acyl CGAs and their 1-acyl-substituted lactone ular formulas from the standard CGA roasting system. Ninety- derivatives in extracted ion chromatograms (EICs) of the LC-MS eight molecular formulas were found to be identical to those of data for roasted coffee extracts. Moreover, after lactone forma- the roasting model for carbohydrates, whereas only 75 molecular tion was reported by Farah et al. (2006a), alternative dehydration formulas contained nitrogen atoms, representing the true Maillard pathways for CGAs during roasting leading to the formation of products. These findings clearly unveil the fact that CGAs and hydroxycinnamoyl shikimate derivatives were disclosed by Jaiswal carbohydrates tend to react to a limited degree (20%) to generate et al. (2012). The authors also noted 2 epimers of CGAs (that melanoidin-type compounds during roasting. The same authors is, 3-caffeoyl muco-quinic acid, and 3-feruloyl muco-quinic acid) also used van Krevelen analysis and further tandem MS and veri- during roasting, validating the occurrence of epimerization (Daw- fied the esterification of CGAs with organic acids during roasting. idowicz & Typek, 2017). After the loss/bioconversion of CGAs into other bioactive com- CGAs also act as precursors for carcinogenic HAHCs, and caf- ponents, the fate of polyphenolic antioxidants during roasting was feine is known to form a complex with these HAHCs which is also sketched by numerous authors (Delgado-Andrade & Morales, later transformed in coffee brews with their characteristic aro- 2005; Hecimovic et al., 2011; Tfouni et al., 2012). Roasting de- mas and flavors (Tfouni et al., 2012, 2013). So, it is also nec- gree influences the polyphenol content of different coffee varieties

r 1212 ComprehensiveReviewsinFoodScienceandFoodSafety Vol.17,2018 C 2018 Institute of Food Technologists® Farm to consumer . . . differently (Hecimovic et al., 2011), which might be the main rea- gree (Hecimovic et al., 2011). Overall, some authors have claimed son for irregularity/discrepancy in the antioxidant capacity results a drop in caffeine content for prolonged periods of roasting (Hec- of various authors (Cammerer & Kroh, 2006; Da Silveira-Duarte, imovic et al., 2011; Ludwig et al., 2014), while contrasting results De Abreu, Castle de Menezes, Dos Santos, & Paiva-Gouvea, 2005; of up to 21% increment in caffeine content during medium- and Sanchez-Gonzales et al., 2005; Wen et al., 2005) . Generally, total dark-roasted conditions are also presented in the literature (Casal, phenolic content (TPC) decreases with increasing roasting de- Oliveira, & Ferreira, 2000; Tfouni et al., 2012). This irregular- gree, exhibiting the thermo-sensitivity of phenolic compounds ity in the literature is also attributed to the differential response to temperature. At variety level, Brazilian Arabica has the high- of different coffee varieties to roasting conditions. Ludwig et al. est TPC and total flavonoid content (TFC) for light and dark (2014) also proposed the caffeine/CQA ratio as a good sign for roasts, respectively,whereas the opposite trend is noted for Robusta identifying roast degree. Like that of CGA and trigonelline, the and Cioccolatato varieties (Hecimovic et al., 2011). Additionally, protein content of coffee beans also decreases during roasting, and among the different polyphenolic classes, the content of proan- it usually drops from 11% to 3.5%, which implies that protein thocyanidins and flavan-3-ols follows the same trend (increasing content plays an active role in the formation of volatiles in coffee. with increasing roasting degree), whereas tannins show significant Generally, aldehydes and sulfur compounds are the only degrada- variability among different coffee varieties and different roast lev- tion products of amino acids/proteins, as compared to furans, cy- els. In Arabica coffee, the maximum content of tannins is found at clotenes, pyrans, acids, and carbonyl compounds, which are simple light-roast level while Robusta and Cioccolatato varieties have the sugar fragmentation products. But it has been found that protein highest tannin quantity at medium-roast level (Hecimovic et al., content also controls the amount of sugar degradation products, 2011). The grade of roasting is also in an inverse relationship with because proteins/amino acids usually catalyze various sugar degra- the amount of phenolic acids (Sanchez-Bridge et al., 2016), and dation reactions during the Maillard reaction (Hwang, Chen, & CGAs in soluble coffees. Even the bioavailability of these phyto- Ho, 2012). Detailed interactions of volatiles both from sugars and chemicals/metabolites in the small and large intestine varies with proteins have been discussed (above and below) from different per- varying degree of roast, because it mostly relies on their initial spectives to a lesser or greater degree, so here we will only describe content in the product (Sanchez-Bridge et al., 2016). Thus, dif- the role of protein-bound amino acids on the induction of sugar ferential absorption of these metabolites is driven by the different degradation. Protein-bound amino acids also influence the phys- grade of roast. Among CGA molecules, the level of CQAs (es- ical and sensory characteristics of coffee by altering its profile of pecially 5-CQA) decreases (42% to 94%) with a transition from volatiles. Different protein levels in green beans not only produce light roast to dark roast for Arabica and Robusta coffee (Farah, different levels of protein-based odorants, but they also produce DePaulis, Trugo, & Martin, 2005a, 2005b; Tfouni et al., 2012). different levels of furan and pyrazine volatiles (Hwang et al., 2012). Description of individual levels of CGAs and caffeine, and their Hwang and coworkers removed the protein from coffee beans and correlation with the levels of their derivative lactones, HAHCs, studied the effect of proteins by adding different proteins in an and melanoidins, is not enough to control the levels of these ideal roasting system. They highlighted that increasing the protein (carcinogenic) compounds, unless a thorough study to establish a content decreases furans and furan derivatives (2,5-dimethylfuran, relationship between these metabolites and their potent precursors furfural), with a rise in pyrazines and their derivatives (pyrazine, 2- at different roast levels is carried out. ethyl-3-methylpyrazine and 2,5-dimethylpyrazine). The decrease Besides caffeine and trigonelline, proteins, peptides, free amino in furans may be due to less degradation of polysaccharides, while acids, and amines are the main N-containing compounds in coffee higher protein levels lead to an increase in the breakdown of amino beans, comprising 9% to 13% of coffee dry material. Free amino acids that take part in Maillard reactions, and a rise in pyrazine acids only account for 1%, but their role is more obvious than that levels. As CGAs and proteins are also involved in melanoidin of others in the generation of odorants. Measurement of the total formation, protein content also influences roast color, and in- protein content of coffee is usually based on calculating the total creased protein content gives a lighter roast color (Hwang et al., nitrogen content of the beverage and or Gerry? sample and, mul- 2012). The reason behind this lighter color is that proteins partici- tiplying it by 6.25. This total (crude) protein content also includes pate in complex color pigment formation reactions by condensing caffeine and trigonelline contents, and is usually in the range 11 carbonyl compounds to yield cross-linked or polymerized color to 16 g/100 g in green beans. After subtraction of caffeine and products. The addition of proteins also gives rise to high lev- trigonelline content, the net level of pure protein content drops els of other volatiles such as methylacetate, acetic acid, 2-furfuryl to 9 to 13 g/100 g (Franca et al., 2005). Initially, it was evidenced alcohol, maltol, 2,5-dihydro-3,5-dihydroxy-6-methyl-4H-pyran- that protein level does not change significantly among beverages 4-one, 3,5-dimethyl-1,2,4-trithiolane, 2-cyclohexen-1-one until of different quality; then Franca et al. (2005)) cited higher levels of a further increase in protein content (beyond 50%) results in the protein in soft and high-quality coffees than in low-quality bev- suppression of sugar degradation products (Hwang et al., 2012). So, erages. For caffeine, roasting also reportedly leads to up to 30% the protein content of coffee plays a vital role in its sensory/quality degradation, but soft and higher-quality coffees still have a superior attributes by contributing a distinct type/level of volatiles after caffeine content than lower-quality ones. Trigonelline (including roasting. 5-CQA) shows the opposite trend during roasting, with a higher Amines are a group of bases found naturally in coffee and can be content detected in defective roasted beans and low-quality bever- classified into two subgroups: a) those formed via de novo biosyn- ages. Caffeine content varies significantly among different coffee thesis, called polyamines, and b) those formed due to thermal or varieties, and this varietal caffeine difference is also depicted during enzymatic decarboxylation of free amino acids during roasting, roasting. Likewise, polyphenolic content and caffeine content are called biogenic amines. The polyamines commonly found in cof- also influenced by the coffee variety during roasting. Some coffee fee are agmatine, cadaverine, putrescine, spermidine and spermine, varieties (for example, cherry coffee and Cioccolatato) have the whereas biogenic amines include agmatine, cadaverine, histamine, highest caffeine content for light roast while others (for example, phenylethylamine, putrescine, serotonin, tyramine, and tryptamine green Minas) show the lowest caffeine level for the same roast de- (Cirillo, 2003). Polyamines were first detected by Amorim, Basso,

r C 2018 Institute of Food Technologists® Vol.17,2018 ComprehensiveReviewsinFoodScienceandFoodSafety 1213 Farm to consumer . . .

Crocomo, and Teixeira (1977) in green and roasted coffee. Interaction between volatile & nonvolatile fractions. In addition to the The authors showed the existence of spermine, putrescine, and detailed discussion above regarding the individual role of volatile spermidine before roasting and declared that amines degrade at and nonvolatile (compositional components) coffee fractions on 240 °C after 12 min. On the opposite side, Cirillo (2003) re- beverage quality, some authors have also identified interaction ported the presence of both polyamines and biogenic amines, and between these two large coffee fractions which influence the claimed that the level of biogenic amines decreases, but they are stability and quality of coffee beverages after roasting. Coffee not degraded completely even after 12 min at 300 °C. The author volatiles are affected by various environmental factors, intrinsic also reported a variation in amine levels among different varieties stability index and interaction with nonvolatile bean matrix. due to vast differences in cultivation conditions. Among all com- Screening of these volatiles on the basis of their interaction monly detected amines, only spermidine, putrescine and spermine activity with nonvolatile components revealed that 21 thiol, are found both in green and roasted beans. These 3 amines are sulfide, pyrrole, and diketone odorants are significantly influenced also considered the most commonly prevailing amines in roasted by these nonvolatile coffee matrices, while aldehydes, esters, beans, with mean ranges of 50% to 64% (putrescine), 23% to 29% pyrazines, and guaiacols are the least affected classes of volatiles (spermidine) and 12% to 21% (spermine) (Amorim et al., 1977; (Charles-Bernard, Kraehenbuehl, & Roberts, 2005a). Changes in Cirillo, 2003). The levels of spermidine and spermine do not have aroma and even staling after brewing are also considered to be due a significant relationship with cup quality, although higher lev- to the instability of volatiles caused by the nonvolatile fraction of els of putrescine are found during roasting in low-quality coffees. coffee (Milo, Londez & Fumeaux, 2006). The nonvolatile fraction The presence of histamine, cadaverine, tryptamine, and tyramine also traps volatile aroma compounds and affects their stability. is also associated with low-quality coffee (Oliveira, Franca, Gloria, The mechanisms for destabilizing volatile compounds by their & Borges, 2005) Overall, a decrease in the total amine content is entrapment, and rates of reaction with nonvolatile components, registered up to brown roast, and only 5% to 22% of amines re- were proposed by Charles-Bernard et al. (2005a). These authors main detectable in low- and high-quality beverages, respectively, also showed that thiols (2-furfurylthiol and ethanethiol), pyrroles, before dark roast. Amorim et al. (1977) and Cirillo (2003) also sulfides (dimethyl sulfide), and diketones (2, 3-pentanedione) proposed a 90% to 97% reduction in total amines up to light-roast are the classes of volatiles most affected by the coffee solution as level. Among the most dominant amines, the levels of putrescine compared to sulfides, pyrroles, and diketones. Thiols are the most and spermine start to decrease significantly, even from the drying affected group of compounds; their reactivity with coffee matrix stage, whereas spermidine proves to be stable, as 10% of its orig- follows a pseudo first-order reaction, suggesting the chemical inal content is still detected after 16 min of roasting. The level reaction of thiols with the matrix. So, thiol losses occur through of low beverage quality-associated amines also decreases, although physicochemical interactions with the coffee matrix, which may low cup-quality coffee has a higher content of these amines than include physical/hydrophobic trapping, salting out and reversible high-quality beverages (Oliveira et al., 2005). Further research or irreversible covalent binding. Thiols also react with the Maillard is needed to specifically explore the relationship of amines with reaction product pyrazinium and oxidize CGA (Negishi, Negishi, specific kinds of coffee defect. Aoyagi, Sugahara, & Ozawa, 2001). Generally, thiol losses occur The lipid fraction of roasted coffee beans depends upon the in two widely discussed ways, that is, radical reactions and ionic roasting and preparation methods and may vary from 11 to reactions. Radical reactions usually happen during transition 20 g/100 g of roasted coffee (Gloess et al., 2013). The relation- metal catalysis, heating, light exposure, and oxidation with ship between lipids and roasting degree is uncontroversial, and an reactive oxygen species generation which can abstract hydrogen increased amount of lipids has been observed from light to dark from sensitive aroma compounds like thiols. Disintegration of the roasting stages due to their stability and the degradation of polysac- odorant 2-furfurylthiol is a good example of a radical reaction; charides and proteins (Dias, de Faria-Machado, Mercadante, Bra- a positive relation has been noted between the degradation of gagnolo, & de Toledo Benassi, 2014). The unsaponifiable fraction 2-furfurylthiol and the radical activity in a coffee solution (Blank of these lipids are called diterpenoid alkaloids, which are mainly et al., 2002). Ionic reactions involve the transfer of electrons from composed of 2 physiologically important compounds named kah- electron donors or nucleophiles, such as thiols, to electrophilic weol and cafestol (Campanha, Dias, & Benassi, 2010). The liter- sites in the coffee matrix. Binding of 2-furfurylthiol to pyrazinium ature lacks consensus about the relationship of these diterpenoid and the addition of polyphenolic-generated quinones are good alkaloids with roasting degree, as some authors noted stability in examples of ionic reactions. The mechanisms of thiol degradation the diterpenoid alkaloid content with prolonged roasting (Cam- have been deeply investigated, and two pathways have been panha et al., 2010), while others registered disintegration of these found to be mainly responsible for all thiols (Charles-Bernard diterpenoid alkaloids into their dehydrated derivatives (that is, de- et al., 2005b). Based on their nucleophilic reactivity, thiols can hydrocafestol and dehydrokahweol) or isomers (that is, isokahweol be divided into 4 groups: aromatic or benzylic thiols > primary and isocafestol) (Dias et al., 2014). In fact, diterpenoid alkaloids thiols > secondary thiols > tertiary thiols. The degradation rate are thermolabile (especially in C. arabica), and their degradation of aliphatic thiols (primary, secondary, and tertiary thiols) increases starts only during light roasting. The kahweol and cafestol content with increasing nucleophilicity, and it also requires oxidizing of diterpenoid alkaloids also declines after 2 to 8 min of roasting conditions, whereas aromatic thiols undergo a parallel mechanism at 230 °C both in Arabica (up to 70%) and Robusta (up to 60%). of degradation involving oxidation, which can be prevented Degradation of kahweol and cafestol results in disproportionate solely by adding strong reducing agents (Charles-Bernard et al., production of their dehydrated products, that is, dehydrokahweol 2005b). The degradation of aliphatic thiols can be prevented by and dehydrocafestol, respectively (Speer & Kolling-Speer, 2006). the removal/modification of electrophilic sites, whereas removal Speer and Kolling-Speer (2006) also mentioned the ratio of cafestol of both oxygen and electrophilic sites is mandatory for inhibiting to dehydrocafestol as a good indicator of roasting degree, with val- the disintegration of aromatic thiols. In other words, preventing ues below 15 representing dark-roasted coffee. the addition of nucleophilic sites is not enough to tackle aromatic degradation. Moreover, aliphatic thiols with long chains have

r 1214 ComprehensiveReviewsinFoodScienceandFoodSafety Vol.17,2018 C 2018 Institute of Food Technologists® Farm to consumer . . . greater hydrophobicity, which in turn has a positive relationship roast volume and soluble solids content in the beverage (Eggers with the nucleophilicity of thiols. Thiols not only endow coffee & Pietsch, 2008). In addition, the net soluble solids content of with roasted and burnt flavors but also provide acidity. So, Arabica coffee is 3% to 4% lower than that of Robusta for the degradation of coffee thiols during storage or processing results same roasting degree in the same roasting system, and that per- in reduced burnt, roasted, and acidic features. Dimethyl sulfide centage is also lower for light-roast levels. Similarly, all fast roast decay also obeys pseudo first-order reactions in addition to systems currently employed also provide higher total soluble solids oxidation of sulfides to sulfoxides and sulfones, leading to their content in the final beverage due to greater porosity, which allows incorporation in the coffee matrix (Charles-Bernard et al., 2005a). greater extraction of solids (Clark, 1987). Generally, coffee roasters Pyrroles and diketones also react with the coffee matrix, but at a use all three mechanisms of heat transfer, conduction, convection, slower rate (Charles-Bernard et al., 2005b). Variation in the rate of and radiation; however, convection is the most valuable mode of reaction between volatiles and nonvolatiles largely depends upon heat transfer which determines the rate and uniformity of roast- the composition of the coffee matrix. Nonvolatile phenols can ing (Baggenstoss et al., 2008a, 2008b). Nonetheless, it is hard to react with volatile 2-furfurylthiol only in the presence of water and achieve a precise convection/conduction ratio in any roast sys- oxygen. Similarly, degradation of CGA and formation of quinones tem. A FBR exclusively uses the convection mechanism of heat to interact with thiols are also water- and oxygen-dependent transfer, resulting in a high roast volume and yield. The common reactions. Carbohydrates offer a negligible contribution towards benefits of FBR are good control over the process, uniformity of interaction between aroma compounds and the coffee matrix. product, less maintenance, ease of cleanliness, and high mechan- Most proteins are usually denatured during roasting; however, ical reliability. Various roaster manufacturing companies market a some studies have also remarked on exchange reactions between wide variety of FBRs on the basis of operation (batch or con- disulfide aroma compounds and protein sulfhydryl groups. tinuous) and design/mechanism of roasting (rotating or jet zone, Nevertheless, Charles-Bernard and colleagues did not witness or multiple-zone quasi-continuous belt process FBR, and so on). this interchange reaction, whether they found degradation of FBRs provide the most efficient heat transfer mechanism by si- sulfhydryl groups into sulfur compounds and melanoidins or multaneous tumbling and heating of individual coffee beans by not. Strong interaction between these melanoidins and sulfur hot air in the roast chamber. High-velocity hot air is usually di- compounds, 2,3-pentanedione and pyrrole, has also been docu- rected from the bottom of the roaster, which heats and fluidizes mented. Some authors have suggested that low-molecular-weight the motion of beans concurrently. The nonnecessity of mechan- melanoidins are worse at trapping volatiles than high-molecular- ical agitation, reduced loss of coffee particles, cleanliness, ease of weight melanoidins (Charles-Bernard et al., 2005a, 2005b). removal of coffee skin, and less smoke are some of the added ben- Roasting technology/roasting systems. Selection of good roasting efits of FBRs (Eggers & Pietsch, 2008). The other most important technology is also essential for ensuring good cup characteristics advantage related to cup quality is that FBR-roasted coffees have of roasted coffee, as different roasting approaches influence the a higher soluble solids content than other roasted coffee beans. individual content of each odorant in each class of volatiles. Nor- On the other hand, Nagaraju et al. (1997) claimed that HTST mally, commercially available roasters are used for roasting green SBR coffee extracts have a higher soluble solids content but their beans, but each roaster has its own design, time–temperature pro- study lacks evidence-based comparison of soluble solids between file, and mechanism of heat transfer. Usually, the roasters which FBR- and SBR-roasted beans. The fluidization velocity of hot have been employed in all the roasting investigation studies so far air (200 °C) is usually set at 2.1 m/s for roasting green beans, can be classified into 3 classes on the basis of the above-mentioned and it usually decreases to 1.7 m/s due to expansion of coffee 3 features: a) fluidized bed roasters (FBRs), b) spouted bed roasters beans. This variation clearly illustrates that control over fluidiza- (SBRs), and c) microwave roasters (MWRs). Besides these, hor- tion velocity is crucial. However, there is a specific air-to-bean izontal rotating drum roasters and packed-bed roasters have also ratio used in the roasting process which depends upon blend type, been used and patented in the last century. Horizontal rotating roasting time, degree of roast, roasting systems, coffee variety, etc. drum roasters are a conventional type of roaster with a horizontal Furthermore, a small quantity of air needs a higher temperature rotating drum and a vertical drum with paddles. These conven- to transfer the same extent of heat, since a higher quantity of tional drum roasting systems work on the principle of HTLT air can transfer the heat at specific lower roasting temperatures. (high-temperature, long-time) with hot gas (450 °C) to heat the SBR is the alternate form of FBR and works on the principle beans for 8 to 12 min. However, continuous-type conventional of HTST. A steady axial jet causes the agitation of particles in drum roasters use a high air-to-bean ratio which permits roast- SBR and also results in a clean and smooth roast product due to ing at a lower temperature (that is, 250 °C) and shorter time (3 the uniform mixing of solids facilitated by HTST. The minimum to 6 min). Commercial usage of conventional drum roasters has spouting velocity required is 1.2 to 5.5 m/s for the proper spouting been abandoned due to high energy input costs, loss of product, of materials, for which the air/solid ratio is set to 1.1 to 1.5 (Na- degradation of aromatic compounds, less total soluble solids and garaju et al., 1997). SBR-roasted coffee has more extractable total volatiles in the beverage, smoky environment, high degree of burnt solids with less acids, which is a desirable trait of roasting. The flavor, and so on (Eggers & Pietsch, 2008; Nagaraju, Murthy, Ra- HTST working principle of SBR also favors less loss of aroma malakshmi, & Srinivasa, 1997; Nagaraju, Ramalakshmi, & Sridhar, compounds, which results in rich coffee quality attributes. Also, 2016). Packed-bed roasters are batch-type roasting systems with a the caffeine/CGA ratio of SBR-treated coffee beans is 0.31, which tangential flow of hot gas through specially designed louvers which is very near to the value for conventional drum roasting (CDR), whirl the beans in a spinning packed-bed pattern. The high ve- that is, 0.32 (Nagaraju et al., 1997). So, SBR-roasted coffee has its locity of the hot gas makes it possible to transfer the maximum own physical characteristics which normally differ from those of quantity of heat in less time so that roasting can be performed FBR- and CDR-roasted coffee. Nagaraju and Bhatacharya (2010) at a lower temperature (that is, 250 to 260 °C) in a shorter time. also determined the physical characteristics of SBR-roasted Ara- The reduction in roasting temperature and time ultimately reduces bica beans and noted that the brightness of beans decreases with an loss of product and aromatic compounds to give a slightly higher increased roasting period, while brightness variations ranges from

r C 2018 Institute of Food Technologists® Vol.17,2018 ComprehensiveReviewsinFoodScienceandFoodSafety 1215 Farm to consumer . . .

5.2% to 20.4% for a designated time (30 to 300 s) and temperature same classes of compounds but the levels of each class of com- (150 to 250 °C). The lowest chroma values are detected at termi- pounds differ significantly for each roasting method. Furans are the nal time and temperature, with a curvilinear decrease with both most abundant (21% to 26%) class of compound, which imparts parameters. It seems that chroma variations are more dependent sweet, burnt and caramel aromas. Among furans, various odor- on temperature than on brightness variations, which rely more on ants, that is, 2-furanmethanol, 2-furaldehyde acetate, 5-methyl-2- time in SBR. Hue was found to be time–temperature-dependent, furaldehyde, and 2-furaldehyde acetate, are the dominant furans in increasing with increasing time–temperature of roasting (576 to CCM roasting, whereas 2,5-dimethyl-4-hydroxy-3(2H)-furanone 603 nm) (Nagaraju & Bhatacharya, 2010). So, roasting in SBR with sweet and caramel notes is more abundant in Mw roasting reduces brightness drastically, but the sensory quality evaluation of (Nebesny & Budryn, 2006; Nebesny et al., 2007). The concen- dark brown coffee is better than that of dark-roasted coffee. Na- tration of pyrazines (17% to 22%), which impart nutty, burnt, garaju and Bhatacharya (2010) also declared a temperature of 237 and nut-skin flavors, follows that of furans, and it is higher in to 250 °C and time of 180 to 240 s as the most suitable combina- Mw-roasted coffee due to their moderately volatile nature. Car- tion for desirable color and textural features. Roast characteristics bonyl compounds are important ingredients of coffee aroma, giv- of SBR with Arabica beans have also been studied, disclosing that ing flowery, buttery, and rancid/rotten impressions. As for furans, the internal temperature of beans has exponential behavior below the highest amounts of α-diketones and 2,3-butanedione are de- 250 °C, but deviation from this exponential kinetics is observed tected in CCM-roasted coffee, whereas volatile aldehydes (ethanol, when bean temperature reaches 260 °C or beyond, when the propanal, 2-methylpropanal, and butanal), and cyclic ketones (that exothermic reactions responsible for aroma and flavor production is, 2-hydroxy-2,3-dimethyl-2-cyclopenten-1-one and 2-hydroxy- stop and burning of beans starts (Hernandez´ et al., 2007). Roast 3-methyl-2-cyclopenten-1-one) are found in greater amounts for loss is 3% to 12% after 10 min of roasting in SBR, depending on Mw and CCM roasting methods. The high content of carbonyl temperature (Hernandez´ et al., 2007). SBR technology has been derivatives in convective heating may be due to wax deposition further extended and improved by accompanying it with the cryo- and leaching-out of lipids, which lead to production of numer- genic treatment (−40 ± 5 °C) of roasted beans. The key physical ous carbonyl compounds that usually occurs during prolonged aspects of coffee roasted by cryogenic SBR (CSBR) have also been roasting. Benzene derivatives (phenols) usually arise due to the evaluated and compared with those of noncryogenic SBR and degradation of CGA and are found at up to 8% to 11% in the 3 CDR coffee beans (Nagaraju et al., 2016). It was found that cryo- roasting systems employed (Nebesny et al., 2007). These bioactive genic treatment has a more significant impact on bulk density than compounds usually bless coffee with spicy, smoky, and clover-like on texture, dehydration, and color degradation properties. CSBR- aromas. These phenolic compounds (2-methoxyphenol, 4-ethyl- roasted beans (both cryo- and noncryo-treated) have a higher rate 2-methoxyphenol, p-hydroxybenzenesulfonic acid benzaldehyde, of dehydration and lower bulk density than CDR beans. Changes and 2,6-dimethyl-p-benzoquinone) have a high boiling tempera- in color development between these 3 roasting techniques clearly ture, and prolonged roasting at a high temperature favors their for- indicate overall variations in the reaction patterns of pyrolysis, mation, which is why they are found in superior quantities in co- caramelization, and Maillard reactions (Nagaraju et al., 2016). roasted coffees. Sulfur compounds make up 2% of the headspace Comparative volatile analysis revealed that CSBR is efficient in in the three roasting systems, and most of their derivatives do not retaining more of the major flavor volatile compounds, which are vary significantly; the exceptions are methanethiol and dimethyl mostly furans (that is, 2-[(methylthio)methyl]-, 3-furan methanol, sulfide, the lowest amounts of which are detected in Mw roasting. acetate, 2,3-pentanedione), pyrazines (that is, pyrazine-methyl), Sulfur compounds usually impart rotten, garlic-like, onion-like, pyrroles (that is, 1H-pyrrole), and thiols (that is, 1-propen-2-ol, meaty and turnip-like aromas to beverages. Like sulfur compounds, 2-furfuryl thiol), due to enveloping of volatiles at such a low tem- pyridines are another less desirable group of compounds in coffee, perature. The retention of such aroma compounds makes CSBR having a bitter and astringent aroma, and their contents are also coffee sweeter, with nutty notes devoid of the smoky and burnt lowest in CCM roasting. Mushroomy and smoky aroma-endowing flavors observed in counterpart noncryo-treated and CDR-roasted pyrroles (1H-pyrrole and 1-methylpyrrole) are also found to be coffee. The bitter, smoky, and burnt notes in noncryo-treated and scarcest in Mw roasting. Volatile alcohols (2-methyl-2-buten-1- CDR-roasted coffee may be due to the presence of higher levels ol, 3-methyl-3-buten-2-ol, and 3-methyl-1-butanol) and esters of the relevant aroma compounds, that is, furans, pyridines, and (2,4,5-trimethyloxazole) are found at higher levels on roasting for thiazoles, due to longer thermal processing (Nagaraju et al., 2016). the shortest possible time, that is, Mw and CCM (Nebesny et al., Microwave roasting also provides uniformity of roast with high 2007). aroma indices and fewer burnt flavors (Nebesny & Budryn, 2006; Grinding. Grinding is another very important step in coffee Nebesny et al., 2007). Microwave roasting is usually used for the processing, as size reduction, particle distribution, and unifor- worst quality of coffee beans which need a longer roasting time mity determine the final quality attributes of coffee. Grinding is using other roasting methods. Comparison of traditional con- a 2-phase process involving crushing and fragmentation of brit- vective heating and roasting by microwave has also disclosed the tle coffee beans in the first phase, while the grinding phase in- better influence of the latter heating system on the overall physic- volves fine grinding using shear force (Baggenstoss et al., 2010). ochemical and sensory properties of coffee (Nebesny & Budryn, Grinding is usually performed with grinding mills (roller or disc) 2006). Later, a detailed study found that significant variations in or blenders, but specific and uniform grind-size distribution can the odorant profile of beverages roasted by microwave (Mw), cou- be easily achieved by using mill grinding, whereas a larger par- pled convective–microwave (CCM) or convective (Co) heating ticle size and the broadest size distribution are seen in blender methods are the main reason for ameliorated physicochemical and grinding (Baggenstoss et al., 2010). A figure 4B show a com- sensory properties of coffee. Mw heating systems significantly re- parison of 2 grinding systems with respect to desired grind- duce roasting time due to intensive heating of every element in ing size for EC.Grind size or grinding degree is usually mea- beans which also brings about a change in the overall pattern sured by a traditional method called sieve analysis and is stated in of biochemical reactions. The 3 methods employed provide the mesh or microns (25,400 micron = 1 inch). The extractability of

r 1216 ComprehensiveReviewsinFoodScienceandFoodSafety Vol.17,2018 C 2018 Institute of Food Technologists® Farm to consumer . . .

Figure 4–(A) Particle distribution limits in different kinds of coffee. (B) Comparison of 2 grinding systems with respect to desired grinding size forEC. (C and D) The reverse relationship between the brewing time compared with number of particles/gram of coffee. Sources: Adapted from Modern process equipment Inc. Chicago, Ill., U.S.A. soluble solids, acids, and aroma compounds largely depends upon soluble solids, leading to more concentrated or less intense coffee, the fineness/grind size of the particles which will later form the respectively. Too fine a grind can also impose 2 kinds of quality body of the beverage (Severeni et al., 2015). The acceptable range risk: the first is clogging of bigger particles and filters, which can of extractability is from 18% to 22%, whereas extractability per- dramatically increase the extraction and percolation time, and the centages of 16% and 24% are registered as underdeveloped and second is that they can end up in the beverage. Figure 4A symbol- overextracted beverages, respectively (Andueza, Pena, & Cid, izes the schematic representations of the particle size distribution 2003a, 2003b). Underextracted coffee beverages have less body, speciefied for each coffee type, whereas figures 4C and 4D are whereas overextracted beverages have a high degree of bitter- establishing a reverse association of brewing time and number of ness and astringency. Grinding increases the surface area of the particles per gram of coffee grind. coffee beans, which in turn enhances the degree of interaction Roasting also plays a decisive role in grinding. Different roast- between water and coffee grinds/particles. This increased inter- ing degrees have different residual moisture contents, porosity, action is the basis for good release of extractable solids and aroma and brittleness, which influence grinding. Brittleness and poros- compounds from coffee. Grinding also ruptures coffee seed tissues ity of coffee beans increase with roasting degree, which is why and cells, accelerating the release of CO2 and volatiles which also medium-dark and dark-roasted beans can be easily ground, and permits easier extraction of the remaining aroma compounds (An- such grinds usually tend to have a wide range of particle distri- dueza et al., 2003a, 2003b). Although the literature lacks consensus bution with a high content of fine fractions. On the other hand, about the exact particle size required to constitute a coffee type, light-roasted coffee beans have a high moisture content, which various authors have used screen analysis to differentiate the grind- make the beans plastically more deformable and ultimately results ing grade of coffee into three particle sizes, depending on coffee in a coarser grind. Moreover, coffee which is roasted slowly shows type: coarse (1.19 mm), medium (0.84 mm), and fine (0.59 mm) a homogeneous matrix and pore size and higher density, which, (Oliveira et al., 2017). Coarse grinding is usually performed for fil- consequently, provide more uniform grinding results. Also, it is ter coffees, boiled coffees and napoletana-type coffees. Fine grinds advisable to give beans a resting time of 6 to 12 hr just after are used for espresso coffees, whereas Turkish-style coffees require the roasting for a) even redistribution of residual moisture after extremely small grinds. Extremities in both coarseness and fineness quenching, b) degassing, as freshly roasted coffees can retain up can decrease the extractability of soluble solids, organic acids, and to 20 g/kg gases, and c) moving the oil inside from the surface. volatiles when the volume-specific surface area is too small or large After roasting, grinding is considered the most critical step in the to retain water for solubilization and emulsification. The finer (up primary processing of coffee. Grinding must be performed just be- to 0.59 mm) and more circular the particles are, the easier their fore the start of intended brewing as important volatiles are found percolation and extraction. On the contrary, irregularity with too to disperse into the surroundings from the ground coffee as early much fineness or coarseness may lead to faster or slower release of as 1 to 12 min after grinding (Akiyama et al., 2003a, 2003b).

r C 2018 Institute of Food Technologists® Vol.17,2018 ComprehensiveReviewsinFoodScienceandFoodSafety 1217 Farm to consumer . . .

Furthermore, ground coffee is more hygroscopic than whole less volatile odorants, that is, dimethyl sulfide (cabbage), pyri- beans, which can increase its moisture level above the legal limit dine (pungent), 4-vinylguaiacol (spicy), hexanal (grassy), 2-ethyl- (that is, 4% to 5%) and hence influence quality attributes neg- 3,5-dimethylpyrazine, and 2-ethyl-3,6-dimethylpyrazine (earthy). atively. The moisture content of ground coffee increases with This higher volatiles content in (cold) wet grinding is due to en- storage time, and Correa et al. (2016a, 2016b) found that the trapment of aroma compounds by water. However, the advantages moisture level can reach up to 3% to 4% (d.b.) from 1% to 2% of using wet grinding can be avoided by using hot water (that is, (d.b.) in 180 days of storage. Illy and Viani (1995) also found that hot water wet grinding) whether grinding in a mill or blender water exists within the product as a monolayer at this level of (Baggenstoss et al., 2010). The only drawback of (cold water) wet moisture. Despite the fact that products with different roast levels grinding is that it can only be applied in facilities where extrac- (light-, medium-, and dark-roasted coffee) have different mois- tion is required just after grinding because wet grinding yields a ture levels, the hygroscopicity of the resulting roast products is the coffee suspension due to greater solubility of coffee both in hot same, as all roast products adsorb the same amount of moisture and cold water. Additionally, all the studies mentioned above lack after 180 days of storage (Correa et al., 2016a, 2016b). An increase quantitative data about how much of these volatiles is lost, how in moisture content is a major reason for agglomeration in ground much is retained by ground beans, and their overall impact on coffee beans, which consequently decreases the flowability with sensory attributes. There are only very few studies in the literature an increase in the product’s angle of repose (the steepest angle be- which disclose the relationship between grinding and the qual- tween the descent/dip of the slope of flowable granular material ity characteristics of beverages. These few studies have exclusively with respect to its horizanetal plane). This decrease in flowability established the relationship (individually and combined) between (due to an increase in the angle of repose) is greater for fine grinds, different grinding grades, extraction times, and the sensory and followed by that for medium-coarse and coarse grinds. An increase quality attributes of coffee (Andueza et al., 2003a, 2003b; Sev- in the angle of repose is also a symbol of rejuvenation of the cohe- ereni et al., 2015). Andueza and colleagues refined fine-ground sive forces among ground coffee particles which makes penetration coffee beans through sieve analysis (710, 600, 500, 400, 300, 200, of water molecules harder and slower and thus makes extraction and 100 μm). They reported the same pattern in natural and tor- longer and less smooth. The storage of whole and ground coffee refacto coffee up to 710 μm, except for coarseness and brittleness at lower temperatures decreases the hygroscopicity of the product, in natural and torrefacto coffee, respectively. Particles in the size as lower moisture adsorption values were noted by Fernandes et al. range 300 to 400 μm were unexpectedly found in fine-ground (2006). In addition, studies dealing with the sensory and shelf life beverages rather than in very fine ones. The extractability per- attributes of ground coffee also found less coffee aroma even in centage was also higher (>24%) for A20:R80 (a blend of 20% coffee grounds stored for 2 weeks. Storage at ambient conditions Arabica and 80% Robusta) 50% torrefacto coffee than for A20: causes a more obvious difference in the sensorial aroma profile R80 natural coffees; however, overextractability in torrefacto cof- than that of frozen ground coffee. However, bitterness has the op- fee did not make it bitter and astringent, which may be due to the posite trend, increasing with storage time, which may be due to addition of extra sugar that masks those effects. Extraction yield the degradation of phenolic compounds (Ross, Pecka, & Weller, and solids concentration also increase with a reduction in particle 2006). Previously, Mayer and Grosch (2001) also cited a decrease size up to certain a limit (Andueza et al., 2003a, 2003b; Oliveira in sweet/caramel flavor with a rise in earthy and smoky notes et al., 2017). However, contrary outcomes are also recorded in during storage of ground coffee. Thus, storage, packaging, and the literature, highlighting the insignificant relationship between transportation of ground coffees should be strongly discouraged different grinding grades and total soluble solids (Severeni et al., to avoid loss of volatiles and gain of moisture. Almost 47 types of 2015). This contradiction may arise due to the fact that authors volatiles have been cited as dissipating during the grinding process. might have used different roasting grades, roasted coffee blends, Most of these dissipated volatiles belong to 9 classes of odorants: extraction temperatures, varieties, and so on. Fine grinding also furans, aldehydes, diketones, pyrazines, phenols, pyrroles, thiols, shows better extraction of caffeine content (Bell, Wetzel, & Grand, acids, and pyridines. Nevertheless, furans, pyrazines, diketones and 1996). Total soluble solids and odor intensity have also been found aldehydes are the classes most affected by grinding (Akiyama et al., to fall with increasing grind size (Andueza et al., 2003a, 2003b; 2003a, 2003b), as it is clear that odorants which are highly volatile Severeni et al., 2015); however, no authors have described the spe- are affected more by the grinding process as the temperature dur- cific range of grind size most suitable for optimum odorant and ing grinding may reach > 100 °C. To counter this, many authors soluble extraction. Severeni and coworkers also highlighted the in- have described the use of condensers or absorbing devices to en- significant relationship between grinding grade and pH/titratable trap floating volatiles which can be added back during later stages acidity. Nonetheless, they mentioned lower pH/high acidity for of processing or packaging (Clark, 1987). However, this approach fine-ground coffees. They also mentioned no significant effect of may be expensive and requires an extension in the manufactur- longer extraction times (>9 s) on pH/acidity for the same grind ing technology of coffee. Alternatively, Baggenstoss and colleagues level, due to quick withdrawal of almost all organic acids during presented a novel approach of “wet grinding” to overcome the loss shorter extraction times (Severeni et al., 2015). Studies dealing of volatiles. They performed dry and (hot/cold water) wet grind- with the combined effect of grind size and extraction time have ing both in milling and blender grinders and compared the results. revealed that extraction of volatiles is really a function of grind Cold water wet grinding produced the smallest fraction of coarse size rather than extraction time. Severeni et al. (2015) also doc- particles in mill grinding, while this fraction was relatively abun- umented the clear discrimination of brews from medium-coarse dant in dry/hot water blender grinding. Analysis of aroma profiles and coarse grinds extracted for different time periods, that is, 9 also revealed the highest content of 2-methylbutanal (coca), 3- to 21 s, while poor discrimination was seen among the brews of methylbutanal (malty), dimethyl trisulfide (sulfuric or cabbage-like fine grinds extracted for the same time intervals. In other words, aroma), 2,3 butanedione (buttery), and 2,3-pentanedione (but- it takes less time (9 s) to extract almost all available soluble solids, tery) using cold water wet grinding whereas negligible differ- organic acids, aroma compounds, and so on, for finer grinds than ences between dry- and wet-ground coffees were detected for for medium-coarse or coarse grinds so there is no discrimination

r 1218 ComprehensiveReviewsinFoodScienceandFoodSafety Vol.17,2018 C 2018 Institute of Food Technologists® Farm to consumer . . . between extraction periods. In addition, sensory analysis of bever- method; it makes use of a perforated bottom, having a filter and ages prepared from different grind sizes reveals fermented, brewed, coffee on it, for pouring down boiling water under the influence woody, buttery, peppery, burnt, and roasted notes in both fine and of gravity. Coarser particles are also used in the drip technique; finer grinds for espresso and torrefacto coffee. On the other hand, nonetheless, various filter sizes are available commercially, which rubbery, burnt and acrid notes are found for coarse-grind espresso, allows the barista to control the extent of filtration. Control over and torrefacto coffee. Some other authors have also documented particle size and filtration is also advantageous in terms of control- a malty flavor in fine-grind espresso beverages due to overextrac- ling percolation speed and contact time between hot water and tion of 2-methylpropanal, 2-methylbutanal, and 3-methylbutanal ground coffee. The water drips from the water column by force (Holscher, Vitzthum, & Steinhart, 1990; Semmelroch & Grosch, of gravity, running through the ground coffee bed held by the 1995). Similarly, the buttery flavor, well perceived in fine-grind filter. Much data is available in the literature about the effect of coffee beverages, is also attributed to good withdrawal of 2,3- coffee bed depth and filter shape on the quality of coffee. On butanedione and 2,3-pentanedione (Maeztu et al., 2001). More- the other hand, pressurized brewing/extraction is a somewhat dif- over, off-notes (for example, peppery, earthy, musty, and woody) ferent procedure to boiling brewing methods. Generally, a pres- due to larger amounts of the pyrazine class of odorants are found in surized brewing/extraction method is used to produce espresso both espresso and torrefacto coffee (Blank, Sen, & Grosch, 1991; coffee (EC) with the aid of a steam machine. This steam machine, Maeztu et al., 2001). now called an espresso machine, is used to induce the percolation of hot water (90 ± 5 °C) at a pressure of 7 ± 2 atm through Extraction/brewing a ground coffee cake (6 ± 1.5 g/40 mL) for a specified period Extraction techniques. Since the 15th century, various coffee of time (30 ± 5 s). Various authors in the literature have investi- extraction and brewing methods have been in use depending upon gated the factors affecting the quality of EC which include water cultural, social, geographical, financial, and life-style factors, as quality, EC extraction technology, and related technical conditions well as personal preferences. All these extraction and brewing like coffee/water ratio, extraction temperature (both inlet water methods differ on the basis of extraction time, extraction pressure, and coffee cake temperature), pressure, temperature application extraction tools, extraction temperature, yield, strength, body, and profile, and so on (discussed below) (Caprioli, Cortese, Sagratini, volume of the resulting extract. Yield is defined as the mass per- & Vittori, 2015). After boiling water (TC), immersion (FPC), centage of roasted and ground coffee dissolved in the brew. Yield and pressurized brewed coffees, there is another kind of beverage between 18% and 22% is the standard quality criterion above or called lungho or lungo coffee, characterized as a mild beverage below which a beverage is considered over- or underextracted. with a large volume (100 to 125 mL), which is usually consumed Coffee strength can be measured by drying a certain amount of in combination with milk and/or cream. Whereas espresso cof- beverage and weighing the residual solids. The body of a coffee fees are intense coffee beverages with a low extract volume (25 to beverage is really a matter of balance among the volatile and non- 30 mL) usually extracted from fine grinds, some lungo beverages volatile compounds being extracted by different methods. Broadly, can also be extracted in the espresso fashion, but their sensory these methods have been categorized into three types, that is, de- and quality characteristics definitely differ from those of espresso coction (boiling), infusion (immersion), and pressure methods. (Amanpour & Selli, 2015). Decoction and infusion are old-fashioned coffee brewing styles Since the beginning of the 21st century, many studies have been but now coffee-drinking societies have been overwhelmed by the published so far addressing the influence of a wide range of extrac- pressure-brewing method. Despite vast variation in the mecha- tion/brewing methods on the chemical, sensory and quality as- nisms of these methods, each method sticks to the liquid–solid ex- pects of coffee. The most comprehensive studies in this regard were traction phenomenon with water as the main extraction “player.” presented by Gloess et al. (2013), Parenti et al. (2014), and Lopez-´ The earliest method of brewing was extraction of coffee with boil- Galilea et al. (2006; Lopez-Galilea,´ Paz de Pen, & Cid, 2007). ing water (decoction), in which ground coffee is mixed with water Gloess et al. (2013, 2014) compared the widely used espresso and and then brought to boiling. Turkish coffee (TC) is an example lungo extraction methods (semi-automatic espresso and lungo – of this oldest brewing method. A higher extraction temperature SE and SL, automatic espresso and lungo – AE and AL, Nespresso (ࣙ100 °C) and longer extraction time make TC a bit harsh in – NE, moka/espresso Bialetti – MOK, lungo/French press – BO, taste. Furthermore, water evaporation is a usual phenomenon at lungo/Karlsbader Kanne – KK, and filter coffee – F) and cor- ࣙ100 °C, due to which fewer water-soluble components (for ex- related analytical data with sensory profiles. Different extraction ample, sugars, organic acids, and amino acids) are extracted, which methods have a different extraction yield of total soluble solids, renders the coffee with an intense, bitter body and chocolaty fla- acids, caffeine, CGA volatiles, and so on, which ultimately renders vor. French press coffee (FPC), also called immersion brew, coffee the beverages with different body, strength, and other organoleptic press, coffee plunger or cafetiere,` is a variation of Turkish-style features. Gloess et al. (2013) cited extraction temperature and time coffee in which a cylindrical vessel is used for boiling/brewing for as the most influential factors on extraction efficiency, which was a preset time interval (2 to 5 min). Relatively coarser particles are found to decrease in descending order: MOK > lungi > espressi. used in FPC, which remain in contact with hot water depend- Generally, the highest coffee extraction yield is obtained for lungo ing upon the body (sweet and acidic or bitter) preferences of the extraction, due to a higher water-to-coffee ratio. The greatest ex- brew. Short extraction makes FPC sweet and acidic while longer traction yield variability is seen in NE (capsule extraction) whereas extraction times turn it into a harsh and bitter beverage. A plunger the highest coffee washing efficiency (31.2%) is shown by MOK usually holds a filtering device which separates the ground coffee followed by BO (26%) and filter coffee (19%). The higher water- and brew. However, “fines” in ground coffee can pass through to-coffee ratio in lungo is also synonymous with better extrac- this filtering device and appear as sediment in coffee. At the end, tion of acids, fatty acids, and volatiles per gram of coffee, espe- pressure is applied by a piston which squeezes more brew, oil cially in the case of the BO extraction method. Irrespective of and fines from the coffee, making the coffee to have more sedi- the lowest detected yield, EC has the most intense aroma com- ment than drip-filter coffee. The drip technique is also a boiling pared to lungo, which may be attributed to its low volume. The

r C 2018 Institute of Food Technologists® Vol.17,2018 ComprehensiveReviewsinFoodScienceandFoodSafety 1219 Farm to consumer . . . extraction yield of caffeine and CGA follow the same fashion as Among phenolic compounds, the major spicy/vanillin/astringent- that of total soluble solids, but filter methods have the highest endowing ones, that is, the phenols 4-vinyl guaiacol and guaiacol, extraction of caffeine and CGA (including 3-CQA and 5-CQA), are also abundant in TC. The major representative ketonic volatile which may be due to solids trapped in the filter (Gloess et al., is acetol (1-hydroxy-2-propanone), also constituting a greater part 2013). Perez-Martınez et al. (2010) also found higher extraction of TC (479 μg/kg) than of FPC (452 μg/kg). Similarly, pyri- yields of antioxidants in filter coffee and MOK than in espresso. dine, 2-methylpyrazine, and 1H-pyrrole, the major constituents MOK has the lowest extraction efficiency for CGA due to using of pyridines, pyrazines, and pyrroles, are also found in greater dark roast considered responsible for the degradation of CGA. BO amounts in FPC (Amanpour & Selli, 2015). After these exten- is the most efficient method of extracting the greatest amounts of sive reports on decoction, immersion, and pressurized brewing 3-CQA and 5-CQA due to long extraction times at relatively methods, some reports also solely compared the more sophisti- higher temperatures (Zanonni et al. 1992). In terms of net extrac- cated, ready-to-use and user-friendly pressurized EC extraction tion content, espresso shows higher values for total solids, degree methods, that is, capsule (that is, hyperespresso method – HIP, Brix, pH, and titratable acidity in sip (that is, 10 mL) and cup and espresso system – EST) and pod extraction methods, with (30 to 240 mL). As content per sip may be greater in espresso, it traditional EC (TEC) (Illy et al., 2005; Parenti et al., 2014; Wang is also necessary to state the content in a cup. Only total soluble et al., 2016). The comparison was made for sensory, physical, and solid contents show wide variation between sip and cup mea- volatile profiles. All capsular EC extraction systems have higher to- surements, and the same trend is exhibited by all CGA, caffeine, tal solids (66.16% to 66.69%), extraction yields (24.4% to 25.1%), 3-CQA, 5-CQA, titratable acidity, Brix and refractive index, both concentrations (6.6% to 7%), and so on, due to better thermal in- in sip and cup content (Gloess et al., 2013). In addition to capsule sulation mechanisms. TEC has higher scores for pH (5.32), density (NE) extraction, both semi- and full- automatic espresso methods (1.029 G/mL), viscosity (1.36 mNs/m2), and selected aroma com- show espresso characteristics for total soluble solids content (that pound classes. Besides inconsiderable variations between TEC and is, ࣙ4%) per sip, whereas espresso coffee and MOK show the low- HIP systems for density, viscosity, and body, no single system has est content (2.31%) in a sip. On the other hand, lungo beverages significant variation in caffeine and CGA content. The key aroma have total soluble solids content of around 1%, with a maximum marker compounds selected, that is, aldehydes (2-methylbutanal value detected in FPC (1.43%) and a minimum in filter coffee and 3-methylbutanal), ketones (diacetyl and 2,3 pentanedione), (1.03%). Total soluble solids content also shows a positive correla- and pyrazines (2-ethylpyrazine, 2-ethyl-3,5-dimethylpyrazine, and tion with degree Brix, and refractive index which in turn have a 2-ethyl-6-methylpyrazine), corresponding to toasty and malted, positive linkage with the concentration of CGAs (also 3-CQA and roasted, and burnt, and buttery tastes respectively, are more abun- 5-CQA), caffeine, titratable acidity, and aroma intensity. In a cup, dant in TEC. Capsular systems, especially HIP, excel in total lipid no significant difference is found between espresso and lungho concentration and, hence, foam index and foam persistence (Par- coffee (Gloess et al., 2013). Headspace aroma intensity decreases enti et al., 2014). The findings of Parenti et al. (2014) are somewhat from SE and AE to MOK and lungi. Longer extraction time and different from the previous findings of Gloess et al. (2013) in terms dilution factors are the two main factors considered responsible for of capsular EC brewing systems. This conflict in the literature may the least intense headspace aroma in lungi. All extraction classes have arisen because the former authors did not specify whether have an esterified fatty acid content below 0.2 w%, but the highest they used a semi-automatic or automatic EC brewing mechanism. esterified fatty acid content is found in FPC and the lowest in filter Additionally, there would have been a difference in the coffee va- coffee. Espresso coffees have a higher esterified fatty acid/total fat rieties and their roasting and washing conditions. Besides capsular content than filter coffee but lower than that of FPC (Gloess et al., extraction systems, other recently patented EC extraction tech- 2013). Caffeine and CGA follow almost the same trend: SE and nologies include pod machines and Caffe` Firenze (CF) (Masella AE have the highest content, followed by NE, MOK, and lungi. et al., 2015; Parenti et al., 2014; Vanni et al., 2009). Compared SE has the highest caffeine content (21.4 mg/sip), whereas filter to traditional espresso machines and capsular and pod extraction coffee has the lowest caffeine extraction yield (4.7 mg/sip). Sim- techniques, CF is a recently patented EC extraction technique. ilarly, SE has the highest CGA content (5.8 mg/sip) and KK the CF machines are designed somewhat on TEC machine principles, lowest (0.78 mg/sip). Sensory evaluation of all extraction methods but their operation is quite different. The main difference lies in also found that SE and AE have the most desireable texture/body, the hermetically sealed extraction chamber where combined air aroma and flavor profile, astringency, and so on, whereas filter and and hot water pressure (15 to 20 bar) is used for pressurized per- KK coffee are poorest for all these quality attributes (Gloess et al., colation of water. Extraction takes place in a partially static mode 2013). Recently, Amanpour and Selli (2015) differentiated well under a wide range of pressures at a temperature of 75 to 85 °C, between boiled (that is, TC) and immersion (that is, FPC) cof- lower than in TEC machines, regulated by incoming water and fees which are different on the basis of their brewing/extraction the chamber itself (Masella et al., 2015). In CF machines, pressure techniques. These different extraction techniques not only bring variations tend to influence the physicochemical characteristics about changes in the volatile profile of the 2 coffees but also bring of EC more markedly than temperature, and a high pressure to about variations in the odor activity values (OAV) of 13 selected low temperature approach produces more foam and greater vis- maker volatiles. The analytical data from this study showed that cosity with high stability of foam and emulsion. By using high TC has a greater amount of total volatiles (48641 μg/kg) than pressure tactics, higher values for total solids (60 mg/mL) and FPC (48153 μg/kg). TC is rich in furans, lactones, phenols, and brew density (1.022 to 1.071 g/mL) are recorded (Masella et al., ketones, whereas FPC has higher amounts of pyrazines, pyridines, 2015) as compared to TEC (Illy et al., 2005). Surprisingly, no and pyrroles. Among furans, furfuryl alcohol and furfuryl acetate apparent effect of the high pressure to low temperature tactic is are the main volatiles in both coffees, but a greater amount is found for the caffeine, CGA, trigonelline, or aromatic profile of detected in TC. Two lactones (that is, γ -butyrolactone and δ- CF-prepared coffee. Only a few key odorants (that is, 2-ethyl- valerolactone) contributing a butter/coconut flavor are slightly 3 6-dimethylpyrazine, 2-ethyl-6-methylpyrazine, and guaiacol), less abundant in FPC (8761 μg/kg) than in TC (8880 μg/kg). which are responsible for toasted, woody, and spicy sensations,

r 1220 ComprehensiveReviewsinFoodScienceandFoodSafety Vol.17,2018 C 2018 Institute of Food Technologists® Farm to consumer . . . respectively, increase on increasing the extraction temperature in larger surface area. A positive effect of longer brewing time and a CF (Masella et al., 2015). Lopez-Galilea´ et al. (2006) investigated high coffee-to-water ratio is also noticed on the caffeine extrac- torrefacto and traditionally roasted coffees by using 2 brewing tion rate. Drip extraction with a filter is superior for extracting a techniques, that is, espresso and filter/infusion extraction. Both higher caffeine content in beverages, provided a high coffee-to- conventionally roasted coffees (that is, EC and filter coffee) pro- water ratio is employed for brewing (Bell et al., 1996). Hinz et al. duce higher yields (44% and 47%, respectively) than torrefacto (1997) worked on filter coffee extracted over a period of time and roasted coffees. Chromatographic areas for headspace aromatic explained their results by presenting a mechanistic model. Stud- intensity are higher for EC in both roasted coffees (44% and 40% ies on the wetting dynamics of coffee particles were concluded for conventional roasting and torrefacto roasts, respectively) than by Mateus et al. (2007). Morphological changes in roasted coffee in filter coffees. The higher aromatic intensity detected in EC is matrices are noted as a result of the solubilization of coffee com- a result of high-pressure brewing in EC machines, whereas filter ponents. Wetting of roasted coffee also enhances the diameter of coffee is an infusion coffee brewed at relatively lower pressures. coffee particles due to absorption of water through concentration The other possible reason for the lower aroma content of tor- gradient or capillary transfer mechanisms. The processes of ab- refacto roasted coffees may be the replacement of 15% of roasted sorption and swelling result in deformation of the coffee matrix coffee by sugar, due to which fewer aroma compounds are gen- and mobility of constituents. The results of Mateus et al. (2007) erated compared to those in EC. Moreover, the addition of sugar suggest coffee swelling, water counterflow, agglomeration of sol- during torrefacto roasting may result in excessive caramelization ubles, and physical constraints as the main factors influencing the and darkening of coffee. To avoid the darkening and burnt flavor, diffusion coefficients of coffee components. Extraction of caffeine companies usually carry out torrefacto roasting at a lower temper- is 2 to 3 times faster from preswollen coffee beans. Similarly, the ature, which could be another possible reason for the presence of extraction kinetics of some selected coffee compounds were wit- fewer volatiles in torrefacto roasted coffees. Lopez-Galilea´ and col- nessed by Booth and coworkers, although their findings are not leagues also highlighted the significant interaction of all 47 volatiles related to prevalent process conditions such as grind size or flow detected for both roasting and extraction processes, except for N- rate. The extraction kinetics of antioxidants/bioactives of coffee furfurylpyrrole, 2-methoxyphenol, hexanal, and difurfurylether (that is, caffeine, trigonelline, and nicotinic acid) have also been which are less influenced by extraction method. The degrada- measured under various process conditions of pressure and tem- tion of sugars also results in a higher content of Maillard reaction perature (Caprioli et al., 2014). The extraction kinetics of caffeine products like furans (furfural, furfuryl acetate, 5-methylfurfural, and trigonelline are affected more by extraction temperature than and furanmethanol), aldehydes (acetaldehyde, hexenal, and 2- by pressure. Overall, an increase in temperature from 88 to 92 °C methylbutanal), pyrroles (1H-pyrrole, 2-formyl-1-methylpyrrole, leads to an increase in the extraction of caffeine and trigonelline and 1H-pyrrole-2-carboxaldehyde), ketones (2,3-pentanedione), but a further increase in temperature (92 to 98 °C) causes a spike pyridines, and pyrazines (2-ethyl-3,5-dimethylpyrazine, ethyl-6- in the extraction kinetics of both components. On the other hand, methylpyrazine, and 2-ethylpyrazine). A difference in extraction an increase in water pressure, at any fixed temperature, has min- method is more critical for torrefacto coffees than for conven- imal impact on the extraction efficiency of these components. tional EC. Furans and phenolic compounds are more abundant in Nonetheless, maximum extraction kinetics are detected at lower conventionally brewed coffees than in torrefacto coffees. Among pressure (that is, 7 to 9 bar) for Robusta blends (Caprioli et al. conventionally brewed coffees, furans (furfural, 2-acetylfuran, fur- 2014). More recently, a time-resolved extraction approach was furyl acetate, 5-methylfurfural, 2-furfuryl-propanoate, and furan- used to find out the extraction trends for caffeine and trigonelline methanol), and phenolics (2-methoxyphenol) are abundant in EC. on varying particle size and tamping pressure (Kuhn et al., 2017). The combined impact of grinding size, percolation time and filtra- The 2 substances have different extraction trends, with trigonelline tion type was investigated by Severini, Derossi, Fiore, Ricci, and being extracted faster than caffeine. Of the two variable process Marone (2016) who cited that only the physicochemical prop- conditions (particle size and tamping pressure), particle size in- erties (total soluble solids, pH, titratable acidity, caffeine content, fluences extraction rate significantly, whereas tamping pressure has and so on) of coffee prepared by 2 types of filter holder vary sig- no remarkable effect on the extraction behavior of these two com- nificantly with all grinding sizes and percolation times. Variations pounds. Both coffee compounds are extracted at a faster rate from become more obvious while using fine/fine-coarse grinds at all smaller particles (250 μm) than from larger particles (355 μm). percolation times, and the lowest values are detected when using Tamping pressure also has poor detectable results for extraction 2-cup filters. However, the variations in physicochemical features kinetics, flow rate, and coffee composition (Kuhn et al., 2017). seem not to influence the aroma profile of espresso even when Mestdagh et al. (2014) and Sanchez-Lopez et al. (2014) worked using 2 kinds of filter holder (Severini et al., 2016). successfully on extraction dynamics by using 2 techniques called Extraction kinetics. After the inclusive reports on extraction solid-phase microextraction (SPME)-GC/MS and proton-transfer techniques, some authors also worked on the extraction kinet- reaction time-of-flight mass spectrometry (PTR-ToF-MS), re- ics of EC brewing (Bell et al., 1996; Booth, Cummins, Dalwadi, spectively. According to these studies, the extraction kinetics of & Dellar, 2012; Caprioli et al., 2014; Hinz, Steer, Waldmann, odorants depend upon their polarity, and their volatility seems not Cammenga, & Eggers, 1997; Kuhn, Lang, Bezold, bMinceva, & to be influential on their extraction trend (Mestdagh et al., 2014). bBriesen, 2017; Mateus et al., 2007; Mestdagh et al., 2014; Mo- Furthermore, maximum extraction intensities can be reached in 2 roney et al., 2015a, 2005b, 2016; Sanchez-Lopez et al., 2014). to 24 s, depending upon blends, after which extraction slows down Bell and colleagues worked on coffee brewing in realistic con- at different rates depending on volatile compounds. The rate at ditions and proposed a high cumulative caffeine cup content by which the extraction intensity of a compound decreases discloses using a finer grind size and a greater amount of coffee powder. the extraction timing of that compound. A fast decrease in extrac- Grind size was found not to be influential on the extraction rate tion intensity shows fast extraction of the volatile compound over of caffeine in the case of large amounts of coffee used for brewing. a short period of time and vice versa. Moreover, volatile organic The higher extraction rate of caffeine from fine grinds is due to a compounds are extracted faster than colorants, and, among these,

r C 2018 Institute of Food Technologists® Vol.17,2018 ComprehensiveReviewsinFoodScienceandFoodSafety 1221 Farm to consumer . . . volatile organic compounds and more polar compounds tend to to the 8 to 12 min of other regular brewed coffees. These short- be extracted faster than less polar ones. The time–intensity data of brewed single-serve coffees tend to have smaller amounts of acidic extraction kinetics is not only helpful for studying extraction in- and phenolic compounds. An increased amount of coffee in the tensities but can also, prospectively, be used to distinguish among packaging increases the brewing time and brewing pressure. The coffee types by using HCA and PCA tools. The time–intensity increased amount of packaging also substantially increases total data show the short extraction times (15 s) for Ristretto Forte; dissolved solids, phenolics, and color, with an insignificant rela- Espresso Intenso and Lungo Fortissimo causes them to remain un- tionship to extraction yield. With respect to brewing volume, the derextracted, and Espresso Classico and Lungo Crema are similar first 4 oz of water extracts almost 96% of solutes, whereas the to each other over this time course. Expansion of the time window remaining solutes are extracted with the remaining 4 oz of water. to “full extraction time” leads to separation of Espresso Intenso, An increase in brewing volume leads to an increase in brewing and Lungo Fortissimo into their full respective clusters which is an time and hence extraction yield. The increase in brewing time is indication of full extraction. Booth et al. (2012) focused mainly on due to an increase in air pump purging time, while increased water the extraction dynamics of drip-filter brewing systems, which in- flow resistance is the main cause of brewing pressure. An increase volve pouring hot water over coffee grounds contained in a conical in grind and roast level decreases the mean volume diameter. A filter. The authors correlated evolution of the shape of the coffee decrease in the mean volume diameter increases the extraction bed during brewing from start to end. A significant relationship yield and eventually, of course, total dissolved solids. Alternatively, was found between the final bed shape and quality attributes of finer grinds and medium-dark roasts increase the air purging time coffee, and the use of single or multiple jets for supplying hot wa- and brewing time as finer grinds generate more hydraulic pres- ter to the coffee bed. Recently, the extraction dynamics of filtered sure, an indicator of increased water flow resistance, than medium coffee have also been explained by a validated multiscale model and coarse grinds. Similarly, water pump pressure also shows a of physics (Moroney et al., 2015a). This approach was modeled pressure-dependent trend which means a substantial rise in pres- using a double porosity model, conducted in 2 situations (that sure while using finer grinds, which also brings about an increase is, well-stirred diluted suspension and packed coffee bed) via 2 in the working time of the pump on reducing the pressure. An in- mechanisms, initial rapid extraction from damaged cells followed crease in the working time of the pistons accompanied by reduced by slower extraction from intact cells. The authors also verified pressure actually protects the integrity of capsules/pods. Navarini grind size as the main manipulator of fluid flow direction through et al., (2009) compared the physics of moka and EC. Moka and the grind and of extraction dynamics. Additionally, coarse grinds espresso brews share some similarities like extraction under high also show a higher extraction mass of coffee (that is, 32%) than fine pressure and temperature; nevertheless, moka is considered more grinds (that is, 28%) in the initial 5 hr, after which no remarkable harsh, bitter and burnt, and lacks foam. The thermodynamic be- changes are noticed in either batchwise or cylindrical brewing in havior of a moka device was linked to its extraction kinetics by a a fixed water volume. The model in Moroney’s work was derived work presented by Navarini and colleagues. It is interesting that by forming a balanced equation for soluble coffee compounds the extraction process in a moka can also be split into 2 phases, that and extracting water in different phases of the coffee bed. The is, the regular extraction phase where liquid–solid extraction oc- parameters of microscale and macroscale models were related via curs, followed by the “strombolian phase” in which the remaining a volume-averaging procedure. Transport of water or coffee sol- water undergoes intense evaporation due to excessive heating in ubles across the phase boundaries on the microscale model appear the final stages of extraction. The strombolian phase can be con- as source terms on the macroscale model. The general model is sidered to be responsible for the harsh, bitter, burnt, and astringent then specialized based on observations in coffee extraction ex- flavor of moka, as high-temperature extraction fluids (water and its periments. The phenomenon of a fast extraction followed by a vapors) can solubilize the less soluble aroma compounds, leading slow one is explained by reduced mass transfer resistance in fine to a deterioration in the organoleptic properties of coffee. Voil- particles and broken cells during the initial extraction moments. ley and Simatos (1979) carried out a number of experiments on The mass transfer resistance offered by intact coffee kernel cells a well-mixed system consisting of water and coffee grounds. A is considered responsible for the slower extraction rate at later simple model, based on the assumption of the spherical and sus- stages. The coffee extraction model analyzed in the paper by Mo- pended nature of coffee grounds, was used to explain the coffee roney et al. (2015a) was based on extraction from a diluted coffee extraction process. Extraction was then simulated as diffusion of suspension; however, the same authors also applied the same ex- a single component from a sphere with a diameter equal to the traction model on a packed coffee bed (Moroney et al., 2015, mean grain diameter. The model was tailored to the data using 2016a,b). the diffusion coefficient as a fitting factor and was found to deliver As compared to the amount of work published on the chemistry judicious agreement with the experimental extraction curve. It or biochemistry of coffee brewing kinetics, as discussed above, lit- was recognized, however, that the initial extraction ensued at a tle attention has been paid to the physics of coffee extraction. A faster rate in the experiment compared to the model extraction true understanding of the physics of coffee extraction will also curve. unveil the influences of various factors and assist in the forma- tion of the next generation of coffee extraction/brewing equip- Influence of EC processing technique on cup quality of EC ment. Some rare work about the physics of coffee brewing has EC is a widely consumed beverage prepared by passing hot been published by Fassano and Farina (2010), Fassano and Tala- water under pressure through ground coffee as the basic princi- mucci (2000), Gianino (2007), Moroney et al. (2015a, 2015b), pal procedure. Several aspects of EC brewing have been investi- Navarini et al. (2009), Wang et al. (2016), and Voilley and Simatos gated, mainly the influence of hot water temperature, pressure, (1979). The physics of single-serve brewing machines (that is, cap- ratio of coffee/water, brewing technique, and botanical variety. sules and pods) and capsule parameters were correlated with the These variables are all under the control of the barista (except organoleptic attributes of coffee by Wang et al., (2016). Single- variety) and have been deeply explored. The barista usually han- serve coffees are brewed for a short time (8 to 12 s) compared dles the roasted coffee with a potential understanding of its flavor

r 1222 ComprehensiveReviewsinFoodScienceandFoodSafety Vol.17,2018 C 2018 Institute of Food Technologists® Farm to consumer . . . and further follows the instructions of the coffee roaster to opti- tion process. Generally, the invasion of water causes initial wetting mize the organoleptic features. A barista is usually aware of the of the coffee bed followed by swelling and solubilization of sol- brewing methodologies that can undergo further amendments, ubles and low-molecular-weight compounds including aromatic practiced according to the priorities of local consumers. Several odorants. Insoluble coffee polysaccharides are the main culprits of researchers have studied the impact of these practiced amend- coffee bed swelling, which causes a decrease in coffee bed porosity ments/adjustments (that is, water temperature, pressure, ratio of and increases the water pressure on the coffee bed. Similarly, a finer coffee/water, brewing techniques). Pressure is the core driving grind size may cause clogging of coffee bed pores, resulting in very force for the extraction of EC and in the production of its hall- high pressure, and an inadequate flow of water (Navarini et al., mark crema or foam. Pressure forces the CO2 of roasted ground 2009). coffee into the water phase from which a stable and dense crema Some recent studies have reported the combined influence of is formed after absorbing some solids from the surroundings. A 3 coffee bed parameters (that is, pressure applied to the coffee stable and dense crema is also a token of preventing odorants from cake called tamping, grinding grade, and coffee-to-water ratio) evaporating. That is why pressurized coffee (EC) tends to have on the overall physical, chemical, and organoleptic properties of more volatiles than other coffees, whereas nonvolatiles are rel- EC (Andueza et al., 2007; Baggenstoss, Poisson, Kaegi, Perren, & atively less sensitive to pressure (Mestdagh et al., 2017). In the Escher, 2008b; Severini et al., 2015). Brews prepared from fine case of nonvolatiles, application of pressure is only a matter of grinds have a higher aromatic profile than those from coarse ones, homogeneous extraction via a stable water flow rate. Application but too fine a grind size hampers water distribution and perco- of pressure also forces the adsorbed oil in the coffee bed to flow lation. Additionally, finer grind sizes have small interstitial spaces, with the water so it ends up in the coffee cup and provides a leading to the build-up of extra pressure by hindering the normal strong body/mouthfeel. Extraction pressure is the name of the water flow rate. Severini et al. (2009) documented that microstruc- equilibrium created by water percolation force and the resistance tural changes in the coffee cake are a consequence of particle size offered by the coffee bed against water percolation. Each extrac- distribution. These microstructural variations in the coffee cake tion method has its own ranges of driving force and coffee bed significantly affect the EC extraction rate from the “log phase.” A properties which ultimately brew coffees with their own specific slower EC extraction rate is observed for fine particle sizes (grade aroma, taste, flavor, and color (Mestdagh et al., 2017). Driving 6 or below) for which 21.2 s is required to get a 22% extrac- forces are generated by a pressurized pump, a vapor-pressurized tion, whereas 10.1 and 15 s, respectively, is required for coarser chamber, and the force of gravity in espresso, moka, and filter grades (6.5 and 7, respectively) (Baggenstoss et al., 2008b; Severini coffees, respectively. One report suggested that the pressure of the et al., 2015). So, a very fine grind size (ࣘ6) needs more time to extracting hot water also influences the physicochemical, textural, reach a standardized extraction limit. Tamping has an insignificant and sensory attributes of EC (Andueza et al., 2002). A total of influence on EC quality as compared to grinding grade (Kuhn 40.71% variability was found in the overall physical, sensory and et al., 2017; Severini et al., 2015), although opposing results were compositional attributes of EC while extracting from 7 to 11 atm claimed by Illy (2005), signifying the role of tamping and em- pressure. Extraction with the highest hot water pressure resulted phasizing its crucial importance for porosity of the coffee cake, in the highest foam index and foam consistency, but unfortunately and pathways of water during percolation. Selection of the right this coffee was rejected in cup quality tests, which may be due to coffee-to-water ratio is very important for porosity, permeability, a significant increase in total solids, total lipids, and CGAs. The and desired water flow rate. A very high coffee-to-water ratio can coffee prepared at the highest extraction pressures was bitter and cause overcompaction of the coffee bed and disturb percolation astringent with a more intense aftertaste. From an odorant point of of the brew and deposition of solids in coffee (Andueza et al., view, some key odorants, like 2-methylbutanal, 3-methylbutanal, 2007), whereas a small amount of coffee can lead to overextrac- and 2-ethyl-3,5-dimethylpyrazine, are also found at their highest tion, making EC unacceptably bitter and astringent. Many suthors levels in EC, which brings about cereal/malty and burnt/roasted in the literature have proposed 5 to 8 g of coffee per cup (40 to off-notes in EC (Andueza et al., 2002). Coffee extracted at the 42 mL) depending on EC blend (Andueza et al., 2007; Petracco, lowest water pressure (that is, 7 atm) has an unstable/inconsistent 2005), but the study of Petracco et al. (2005) lacks data about the foam index, while coffee prepared at 9 atm has a foam consis- physicochemical and organoleptic characteristics of EC. Overall, tency similar to that of coffee prepared at 11 atm. The significant an increase in the coffee/water ratio (6 to 9 g/42 mL) leads to difference in EC prepared at low pressure and at 9 atm is the vari- an increase in acidity, foam index, and foam persistence, viscosity, ation in foam index and foam consistency. On the other hand, density, total lipids, total solids, total solids in filtrate, and surface ECs prepared at 9 and 11 atm differ mainly in sensory and com- tension. An increase in acidity, total solids and viscosity is more positional features (Andueza et al., 2002). Coffee prepared at 9 obvious in Arabica and torrefacto EC, whereas Robusta EC shows atm is ideal with respect to foam index and foam consistency, a greater increase in foam index, bitterness, and astringency. The and it has a desired level of key odorants related to cup quality. bigger increase in astringency and bitterness in Robusta EC is Extraction pressure has a significant link with coffee bed prop- due to the higher trigonelline and caffeine content of Robusta erties which rely on grind size and the particle size distribution varieties. Higher coffee/water ratios (>8g/40mL)makeArabica system. Grinding and the particle size distribution system con- EC unacceptable due to the higher perception of burnt/roasted, trol water permeability and water flow through the coffee bed. A fermented, and acrid off-notes. However, the perception of off- very high bed permeability endorses a very low extraction pres- note flavors is more obvious in natural Robusta (A20:R80) and sure, higher flow rates, and very short extraction times, leading to is even judged at 7.5 g/40 mL (Petracco et al., 2005). Percep- underextraction and vice versa (Corrochano et al., 2015). Coffee tion of more off-notes even at a lower coffee/water ratio may bed permeability also varies during the extraction process. Ba- be due to the masking effect of positive odorants in Arabica EC. sically, coffee bed permeability is a multifactorial phenomenon Extraction of volatiles related to malty flavor (2-methylbutanal, 2- in which grind size, coffee-to-water ratio, tamping and particle methylpropanal, 3-methylbutanal, Strecker degradation products) size distribution system play key roles depending upon the extrac- increases on increasing the coffee quantity for all coffee types.

r C 2018 Institute of Food Technologists® Vol.17,2018 ComprehensiveReviewsinFoodScienceandFoodSafety 1223 Farm to consumer . . .

Fewer freshness notes are observed in torrefacto coffee at all cof- temperature (Salamanca et al., 2017). The density values of Ara- fee/water ratios, although acrid off-notes become more obvious bica/Robusta EC have an insignificant relationship with ramp-up, at >6.5 g/40 mL (Andueza et al., 2007). All these findings show ramp-down, and fixed temperature profiles. From a sensory point that the increase in off-notes is greater than the increase in good of view, color, crema, roast flavor, and unpleasant notes are within flavor notes with an increasing coffee/water ratio for all coffee standard limits, whereas some organoleptic properties like positive types, provided that Arabica and natural Robusta (A20:R80) are flavor notes, body, aroma intensity, and acidity are lower for ramp- extracted at/lower than 7.5 g/40 mL (Andueza et al., 2007). For down and fixed temperature modes. Use of a ramp-up temperature torrefacto coffees, addition of sugar leads to the production of profile is a good way of enhancing crema texture, well-balanced more off-note odorants as compared to other coffee types, so an body, aroma intensity, bitterness, and astringency (Salamanca et al., optimistic coffee/water ratio given in the literature is 6.5 g/40 mL 2017). After the much-discussed inlet water temperature (Andueza (Andueza et al., 2007). et al., 2003a, 2003b; Salamanca et al., 2017), some work about Extraction temperature is another factor considered crucial for the temperature of the coffee bed is also found in the literature the adequate extraction (19% to 22%) of combined soluble (for (Albanese, Matteo, Poiana, & Spagnamusso, 2009). Albanese and example, caffeine, sugars, acids, CGA, and so on), insoluble solid colleagues monitored the temperature profile of the coffee bed and (for example, polysaccharides, proteins, melanoidins), and volatile wrote about faster heating in core zones of the coffee bed com- (aldehydes, ketones, pyrazines, and so on) compounds which, con- pared to that of peripheral parts. This nonhomogeneous heating sequently, form the taste, body/foam, and aroma of EC. Previously, is linked to the preferential direction of heat transport which relies a wide range in the extraction temperature of water (85 to 98 °C) on the percolating water. Total solids content, acidity, caffeine, and was outlined by Illy and Viani (1995), but this wide range causes CGA content increase on increasing the extraction temperature. variation in the extraction percentage of EC from weaker and A temperature of 110 °C is optimal for the desirable extraction less aromatic to overextracted and bitter/astringent EC (Andueza of total solids content (52.5 to 58.2 mg/mL), acidity, caffeine (2.6 et al., 2003a, 2003b; Mestdagh et al., 2017). Generally, total solids, to 4.65 mg/mL), and CGA content (Albanese et al., 2009). This total lipids, foam index, and caffeine increase on increasing the better extraction of the stated components becomes more obvious extraction temperature of water from 86 to 96 °C; however, the with an increase of Robusta percentage in coffee blends. However, content of trigonelline and CGA tends to decrease above 92 °C an increase in the Robusta content may result in an increase in (Andueza et al., 2003a, 2003b), which may be due to their degra- bitterness, acidity, and astringency notes. At 110 °C, foam index dation into other compounds. In the case of Robusta torrefacto is also more than 10% for 20 min, which is standard for EC. Body EC, greater extraction of total solids is observed at all water tem- and foam index values are higher in Arabica and Arabica-dominant peratures, which is due to overextraction of the caramelized sugars blends, which may be due to their high lipid content (Albanese added during roasting. Optimum amounts of featured odorants of et al., 2009). Arabica or natural blend of Robusta EC responsible for freshness or sulfurous (methanethiol), fruity (acetaldehyde and propanal), Water quality malty/roasted (methylbutanal, Strecker degradation products), and Water is the second most important component of coffee, and buttery flavors (2,3-butanedione and 2,3-pentanedione) are found cannot be neglected for good cup quality. The quality of coffee at 92 °C. Only pyrazines and hexanal odorants increase with in- is directly influenced by the quality of the water which affects creasing water temperature (up to 98 °C), which could be the the brewing process; however, some previous works proposed that main reason for a burnt/roasted flavor. For acceptability index and this influence is greater on physical attributes than on organoleptic cup quality, Arabica or natural blend of Robusta EC prepared at properties (Pangborn, 1982). According to SCAA, the quality of 92 °C exhibits the desired levels of odor, taste, flavor, color, foam water for coffee brewing purposes is usually measured in terms index and foam persistence, and body intensity (Andueza et al., of total hardness and alkalinity which depend upon the min- 2003a, 2003b). EC prepared at temperatures lower (for example, eral/electrolyte content (especially carbonates and bicarbonates of 88 °C) or higher (for example, 98 °C) than 92 °C has lower or calcium and magnesium) of water. The SCAA defines optimum higher levels of the aforementioned organoleptic characteristics, extraction at a total hardness of 68 ppm CaCO3 (with an accept- and lower acceptability index and cup quality, respectively (An- able range of 17 to 85 ppm), total alkalinity of 40 ppm of CaCO3, dueza et al., 2003a, 2003b). The percentage of negative flavor and pH 7 (acceptable range 6.5 to 7.5). The SCAA also set out notes also increases on increasing the water temperature; however, the maximum limits for sodium (10 mg/mL), total chlorine (0), certain limitations observed like low odor, taste, flavor, color, foam and total dissolved solids (150 mg/L) for optimum extraction. To- index and foam persistence, and body intensity identify 92 °Cas tal dissolved solids in water are mainly composed of carbonates the optimum water temperature for extraction. These initial find- and bicarbonates of calcium and magnesium. These electrolytes of ings of Andueza et al. (2003a, 2003b) were further verified by water are the main players which decide the terms of interaction Salamanca, Fiol, Gonzalez,´ Saez, and Villaescusa (2017) by brew- between water and coffee during brewing. These electrolytes are ing/extracting at ramp-up (88 to 93 °C), ramp-down (93 to 88 °C) considered the worst offenders for reducing the quality of EC as and fixed temperature (that is, 90 °C). Robusta EC has a decrease they neutralize the acidity of EC; for each 100 ppm of alkalinity in foam index, acidity, lipid content, viscosity, total polyphenolics, of water, a 0.22 unit pH rise is shown in the literature (Sivetz, caffeine, and CGA (including 5-CQA, and 3-CQA) at a ramp- 1972). Carbonates and bicarbonates of sodium ions are regarded down temperature profile and vice versa. Arabica EC has the same as notorious in slowing down extraction, while carbonates and bi- results, except for caffeine content which decreases with increasing carbonates of other monovalent and divalent ions have little effect and fixed temperature profiles. Washed Arabicas have the highest on extraction rate (Fond, 1995; Gardener, 1958). The addition values for CGA and polyphenols at fixed temperature, whereas of hard or alkaline water to the acidic coffee cake disturbs the the contents of polyphenols, caffeine, and CGA are lower for bicarbonate–carbonic equilibrium, which causes a rapid drop in fixed and ramp-up profiles. Some physicochemical characteristics extraction water pH from 7–7.5 to 5.5–5, with the majority of like turbidity, total solids, and filtered solids are highest at fixed ions switching from bicarbonate to carbonic acids to neutralize

r 1224 ComprehensiveReviewsinFoodScienceandFoodSafety Vol.17,2018 C 2018 Institute of Food Technologists® Farm to consumer . . . the acids from coffee. This neutralization releases carbon dioxide of odorants, 4 effects have been suggested by Dold et al. (2011): and swells the coffee bed, which results in a compaction of the (A) for more volatile odorants and a less stable crema, the crema coffee bed and, hence, a slower extraction process (Fond, 1995). acts as an aroma pump, that is, fast release of odorants; (B) for less Wellinger and Yeretzian (2015) interrogated the direct relationship volatile odorants and a less stable crema, a new interface is created of water composition to the sensory profile of resulting brews. The upon bubble rupture which enhances the release of less volatile highest sensory score was awarded to brews prepared from water odorants; (C) for more volatile odorants and a more stable crema, with a low electrolyte content. Even a small variation in hardness the crema acts as a barrier; (D) for less volatile odorants and a stable or alkalinity significantly affects the quality attributes of coffee. crema, diffusion is the ultimate process for the release of aromas Water with the lowest hardness and alkalinity values is best for the from the interface. For EC with crema, utilization of highly min- preparation of coffee with superior organoleptic properties. On eralized water lessens the stability of crema, leading to increased the other hand, Rivetti et al. (2001) showed conflicting results, released of volatiles (after t = 2.5 min of extraction) and vice versa. nullifying the outcomes of Wellinger and Yeretzian. Rivetti and PCA of the physical properties of crema and the release pattern of coworkers used ultra-pure reagent-grade and tap water in the ex- volatiles highlights a negative relationship of surface fraction area traction of roasted ground coffee and established that both kinds with selected marker volatiles at t = burst. This negative relation- of water interact in a similar way with coffee for the same brewing ship is symbolic of the fact that the thinner the lamella film, the time. This contradiction in the literature may be due to the fact greater the aroma burst. Gravitation and evaporation account for that the ultra-pure reagent-grade water was softened by sodium thinning of the lamella film of crema. Thus, volatiles entrapped softening methods which replace calcium and magnesium ions in the gas phase of the foam are expelled into the headspace as with sodium. So, water treated with these sodium carbonates or bubbles due to rupture of the thin lamella by gravitation and evap- bicarbonates behaves identically to tap water in brewing (Navarini oration forces. PCA data taken at t = 6 min show the negative & Rivetti, 2010; Rivetti et al., 2001). Carbonates and bicarbonates association of volatiles release and foam volume and foam surface of water are also involved in foam (crema) formation, as EC pre- area fraction. This strongly negative relationship suggests a higher pared from Milli-Q reagent-grade water has foam with less volume concentration of volatiles above the cup in the case of thinner but with a better persistence. Probably, the thermal decomposition lamella films, low foam volume, and more foam drainage. So, we of carbonates/bicarbonates, in addition to their interaction with can deduce that release of volatiles from EC prepared with poorly coffee acids, leads to the formation of carbon dioxide that aids in mineralized water is less than that from EC prepared with highly foam formation (Navarini & Rivetti, 2010). So EC prepared from mineralized water (Dold et al., 2011). hard water has a high volume of foam which is less persistent, Conclusively, it is crucial to have comprehensive insight into and an undesirable texture,while foam formed by using Milli-Q the interaction of all the variables of coffee (that is, coffee-to- reagent-grade water has a smaller volume but is more persistent water ratio, grind size and shape, particle size distribution, water and has a desirable texture (Navarini & Rivetti, 2010). Dold et al. quality and quantity, extraction temperature and pressure, coffee (2011) also studied the impact of water composition, with spe- bed properties, and water flow rate). All of these variables have cial reference to crema formation and stability. They established an obvious effect on extraction, and many are interlinked with the unique relationship of water quality (poorly to moderately each other in this brewing or extraction system. Water quality mineralized water and highly mineralized water) with crema and and composition also have an effect on extraction parameters and the release pattern of volatiles from EC. The physical structure coffee bed properties, and a barista also needs to select/adjust the of the foam of EC, prepared from poorly to moderately mineral- water quality parameters using self-testing and expertise over the ized water and from highly mineralized water, has an insignificant course of the extraction. relationship with the mineralization extent of water just after ex- traction (t = 0). After 5 min, foam surface area fraction, foam Postprocessing Conditions volume, and the stability of EC prepared from poorly mineral- Serving temperature and crema ized water have higher values than those for EC prepared from Visual cues are the depiction of interisic and extresic quality highly mineralized water. The area fraction is an indication of attributes of any product to its consumers. Lunning and Marce- light reflected by the crema surface. At t = 0, the insubstantial les (2009) present a model in this regard which represent the relationship of surface fraction to mineralization extent bespeaks interaction of consumers with the intrinsic and extrinsic quality the same thickness of the lamella film. But after 5 min, an abrupt attributes of coffee (Figure 5). Visual cues are the first criteria decrease in lamella thickness, as seen in EC prepared from highly which lure consumers to consume coffee. These physical cues mineralized water, could be a reason for a small surface fraction in play a vital role in a consumer’s decision about its purchase and beverages made from highly mineralized water (Dold et al., 2011). subsequent trends of consumption. A good visual cue actually gen- It is believed that the high concentration of ions from highly min- erates the expectations of consumers regarding the consumption eralized water may influence the hydrophobic and electrostatic experience of that product. For instance, the presence of foam, interactions in the surfactant–biopolymer complex which results in beer, sets up consumer expectation of its sweetness and fruiti- in the instability/fading of the crema or lamella of films (Piazza, ness. Visual cues are also believed to have a physiological impact Gigli, & Bulbarello, 2008). An abrupt increase in the area of liquid on the consumption experience of the consumer, as 2 orange (foam drainage) beneath the foam is noticed when using highly juices of different color hue are perceived differently irrespective mineralized water (Dold et al., 2011). For release of volatiles above of their equal sucrose content (Hoegg & Alba, 2007). Familiar- the cup, the effects of the extent of water mineralization are more ity index is another tool which is important for the decree of obvious while preparing EC with crema. However, crema is no acceptance or rejection of a meal by consumers. In the case of longer, generally, regarded as a “volatiles sealer”, as fewer volatiles coffee, sensory profile, aroma, (amount of) crema, serving tem- are detected above the cup in single liquid-phase EC, irrespective perature, color, and hue convey visual cues to consumers. The of water quality, at t = 2.5 after extraction (Dold et al., 2011). postprocessing sensory and aroma profiles of coffee brews are used Furthermore, depending upon crema stability and the volatility to judge final cup quality and make sure that every parameter is

r C 2018 Institute of Food Technologists® Vol.17,2018 ComprehensiveReviewsinFoodScienceandFoodSafety 1225 Farm to consumer . . .

Noticeable Product Qualities

Intrinsic Characteristics Extrinsic Characteristics

• Safety (microbial, chemical, • Product system characteristics’ Physical) • Assigned quality by marketing • Health (nutritional value, healthy communication compounds) • Sensory (texture, odor, taste, color) • Shelf Life (Keep ability, Freshness) • Conveniences (easy to use, to prepare

Physiochemical properties of Raw materials and products • Variable composition • Dynamic food Processes • Variable genetic characteristics

Technological Factors

• Processes parameters • Equipment properties • Building and facility characteristics • Environmental conditions

Legislave restricons and requirements

Figure 5–Extrinsic–intrinsic quality attributes model of coffee as perceived by consumers (Source: Luning and Marcels, 2009). just right. These sensory and aroma profiles of different coffee investigated the role of crema thickness and stability on aroma re- types have been thoroughly discussed in this article in the respec- lease above the cup (Dold et al., 2011; Petracco et al., 2005). The tive sections for each parameter. Impact of serving (amount of) release of volatiles above the cup is dependent on the time elapsed crema, serving temperature, and color/hue will be discussed here. since extraction. Within 2.5 min of extraction, brews with crema Crema is the hallmark of espresso that differentiates it from other generate a greater volatiles concentration above the cup than cof- coffees. It is an intrinsic quality factor and a significant visual cue fee without crema. This is because rupturing of the lamella also to consumers’ aroma sensory perceptions. Any misunderstand- causes the collapse of bubbles, releasing entrapped gases rich in ing of the physical cue of crema regarding a mismatch of actual volatiles (Dold et al., 2011; Parenti et al., 2014). In addition, evap- and likely expectations by the consumer can modulate perception oration of less volatile aroma compounds also occurs at the same of an existing product forever. The effect of the visual cue of an time on the surface of the liquid. In contrast, evaporation is the absence or different amount of crema on consumers’ expected and only phenomenon through which aroma compounds escape the actual organoleptic perception has been documented by Labbe liquid surface of a brew without crema. Despite all this informa- et al. (2016). The absence of crema significantly decreases the tion, detailed research revealing the mechanism of formation and expected and actual perception of a consumer about liking, qual- (in)stability of crema and its sponsor compounds has still not yet ity, taste intensity, bitterness, and smoothness of coffee. However, been published. different amounts of crema are not found to have an impact on With respect to serving temperature, different standard serving expected liking, quality, taste intensity, bitterness, and smooth- temperature recommendations exist across the world. The Nor- ness attributes of coffees (Labbe et al., 2016). This discrepancy wegian Coffee Association and the National Coffee Association might be due to a high familiarity index of consumers with the of America recommend a coffee serving temperature of 80 to 85 product. Furthermore, for beverages with the highest amount of and 82.2 to 85 °C, respectively. The preferred drinking temper- crema, actual consumer perception is much higher (both in blind ature of black coffees was also found to be 52.8 to 72.1 °Cin and standard tastings) than expected perception for liking, quality, many investigations. Variation in the serving temperature changes taste intensity, bitterness, and smoothness. the free energy, enthalpy, and entropy of a coffee brew system, Another aspect of crema is its function as a barrier against the which ultimately changes the release pattern of different volatile release of aroma compounds from the beverage. Many studies have aroma compounds. An elevated drinking temperature increases the

r 1226 ComprehensiveReviewsinFoodScienceandFoodSafety Vol.17,2018 C 2018 Institute of Food Technologists® Farm to consumer . . . orosensory pain threshold (Green, 1985), and the resultant turbu- growth dynamics is very descriptive and devoid of mechanisms of lent flow of air gets mixed with the intake of a small portion of cof- interaction. The relationship of microbial growth to coffee pro- fee. This mixing system decreases the temperature of coffee and re- cessing, fermentation and bean quality is also not fully understood, duces the air–liquid surface pressure, permitting a range of volatiles which hinders attempts to control natural fermentation and fer- to be released from the surface of the coffee. Seventy-six volatiles mentation using starter cultures. Comprehensive research linking (furans, pyrazines, ketones, aldehydes, thiophenes, pyrroles, and or matching individual aroma compounds with their contribution esters) are mostly influenced by serving temperature range (31 to coffee flavor and coffee quality attributes will broaden our cur- to 62 °C). Elevated temperature exponentially accelerates the rent understanding. Understanding the flavor contribution of each release of these aroma compounds (Steen et al., 2017). The profile compound, and subsequently tracking its origin back to individ- of released odorants varies significantly when the serving tem- ual processes, will help the coffee industry to control the desirable perature is higher than 37 °C. A lower serving temperature (31 flavor outcomes of coffee through processing or other farm man- to 37 °C) increases the release of 3-hexanone, benzaldehyde, 2- agement techniques. Storage of green coffee beans is important methyl-1-propanal, phenylethanal, 3-ethyl-2,5-dimethylpyrazine, for aging, as fresh green coffee beans have pea-like, earthy/musty, and 2-methyl-3,5-diethylpyrazine, whereas a drinking tempera- and bell pepper-like flavor off-notes without prescribed storage. ture of 31 to 44 °C increases the volatility of 3-hexanone, 4- But this storage must be in hermetic packaging systems, with the methylpyridine, benzaldehyde, and the alkylpyrazines. At temper- injection of 60% carbon dioxide, of green coffee beans with a uni- atures of 31 to 51 °C, alkylpyrazines are the most-released class form moisture content of 11% to 12%, at a temperature of 20 to of compounds, followed by ketones and pyridines. Release of fu- 25 °C for no more than 6 months. For roasted coffee beans, green rans is influenced less than the release of alkylpyrazines, ketones, coffee beans should be roasted to light-medium level, must not be pyrroles, and pyridines. Moderate release is noted for thiophenes ground, and should be stored at reduced temperatures. If possible, and thiazoles even at selected high temperatures. However, despite it is better to introduce a system capable of storing and transport- the release of all these aroma compounds at different serving tem- ing roasted beans with intact parchment as it offers a shielding peratures, all the drinks are actually perceived as expected with effect against the oxidation of lipids and loss of volatiles. Finally, roast, bitter, and tobacco flavor notes (Steen et al., 2017). for coffee brews, it is not advisable to store and sell it at com- In addition to product, the material of the cup or glass, or the mercial level, as commercialization may require specific packaging way the packaging material is designed, also has a positive rela- with a cold-chain system throughout the market, which will not tionship with expected and actual consumer perception. Unfortu- only raise the input cost of its production but also compromise its nately, research is still needed in this regard, with special reference quality attributes. to different coffee types. Furthermore, the impact of the nature Roasting is a very complex process, and no individual factor of the relationship of coffee coloration/hue and expected and ac- (color variation or roast loss) is enough to decide credibly the tual consumer perception still needs to be investigated. Finally, degree of roast and its quality attributes. The most promising ap- combining the human perceptions investigated with advanced in- proach in this regard could be the combination of physicochemical tegrated biological, biochemical, and physical approaches is a way and analytical approaches. Some analytical tools also need to be to ameliorate postprocessing brew conditions. improved in terms of their sensitivity and resolution, and to estab- lish relevance between their results and beverage quality attributes. Future Perspectives and Conclusions Different time–temperature roasting profiles also cause variation The final cup quality of coffee depends upon both the art and in the extent of the generation of specific classes of volatiles. The science of its manufacture. The overview of all the research con- greatest number of classes and amounts of aroma compounds are ducted on the processing technology of green coffee beans so far witnessed for HTST roasting profiles. Degradation reactions dom- illustrates that different coffee processing techniques not only in- inate during dark roasting, which leads to the elimination of major fluence the major quality-defining volatile and nonvolatile com- volatiles like sulfides, Strecker aldehydes, α-diketones, pyrroles, pounds, but also that coffee seed germination and stress-related and pyrazines, but not of pyridines, hexanal, dimethyl trisulfide, events play a significant role in shaping the final cup quality of 2,3,5-trimethylpyrazine, and 2-ethyl-3,5-dimethylpyrazine. The coffee. Processing conditions are the driving force behind these 30 to 50 odorants which have the greatest impact on quality be- germination and stress-related metabolic events, as well as their long mostly to six classes of odorants, namely furans, sulfur com- time course and amplitude with respect to prevailing processing pounds, pyrazines, pyridines, pyrroles, and α-diketones, whose conditions. Variations in the time course and amplitude of these rates of formation and degradation are highly path-dependent and germination and stress-related metabolic processes bring about vary on the basis of roasting conditions. Among the compositional changes in the organoleptic and sensory quality attributes of the components of coffee beans, extractable/soluble polysaccharides resulting coffee. This distinction and peculiarity in organoleptic (arabinogalactans), CGA, N-containing compounds, and lipids are and sensory quality attributes in coffee beverages are classified the main precursors for all odorant classes. Finally, for roasting sys- into dry/naturals, washed, and semi-washed/pulped naturals cof- tems, all fast-roasting systems currently employed (that is, FBR, fee. However, the entire impact and overall consequences of these SBR, and MWR) are comparatively more advantageous than tra- metabolic processes are still poorly understood. Apart from these ditional drum-type roasting systems in terms of cup quality and endogenous metabolic factors, certain exogenous factors (micro- yield. However, comprehensive studies to establish the relationship bial ecology and fermentation conditions) also exhibit a strong of beverage quality attributes to different roasting technologies still impact on coffee quality. Numerous species of bacteria, yeasts, and need to be carried out. filamentous fungi have been identified in coffee fermentation, but Roasting and grinding processes are interlinked, as the mois- it is not clear which species or strains have most impact on cof- ture, porosity and brittleness of roasted coffee beans are decided fee cup quality. A thorough understanding of metabolic events, by the roast level. Medium-dark roast beans are considered ideal microbial ecology and the occurrence of aroma compounds is a for grinding to any size of particle because light-roasted beans great challenge for future coffee research. The data for microbial have a higher moisture content, and dark-roasted beans are devoid

r C 2018 Institute of Food Technologists® Vol.17,2018 ComprehensiveReviewsinFoodScienceandFoodSafety 1227 Farm to consumer . . . of many classes of odorant and are only used for very strong ESI emergence speed index brews with a touch of bitter/astringent notes. Highly volatile EC espresso coffee classes of odorants are lost during grinding, and existing technolo- EICs extracted ion chromatograms gies (hot/cold water grinding) produce only coffee suspensions FQA feruloylquinic acid which need to be brewed immediately. Therefore, the next gen- FQLs feruloyl-1,5-lactones eration of grinding technologies needs to address these shortcom- FBRs fluidized bed roasters ings. Moreover, a comprehensive work investigating the extent FFAs free fatty acids of loss of volatiles with respect to grinding temperature profile FT-IR Fourier transform infrared and grinding grades will be most helpful in designing the new FPC French press coffee generation of grinding technologies. Grinding and brewing are GABA γ -aminobutyric acid related due to the phenomenon of extraction yield and to coffee GCO gas chromatography olfactometry type. An appropriate grind size is required for the recommended HAHCs heterocyclic aromatic hydrocarbons extraction yield (18% to 24%) and coffee type. Roasting time, one HCA hierarchical cluster analysis of the important parameters, which influences the organoleptic HTST high-temperature, short-time properties of a brew, is also dictated by grind size. ICL isocitrate lyase Organoleptic characteristics of coffee are one of the impera- LAB lactic acid producing bacteria tive attributes that helps to increase the global coffee consumption LC-MS liquid chromatography mass spectrometry and in the classification of novel specialty coffees. Coffee trade LTLT low-temperature, long-time and consumption revolves around its cup quality features whereas NIR near infrared these organoleptic features are highly a multifactorial trait de- PCR-DGGR polymerase chain reaction-denaturing gradi- pending upon both preharvesting (agricultural, environmental and ent gel electrophoresis meteorological factors) and postharvesting (processing methods). PTR-MS proton-transfer reaction mass spectrometry In short, the key theme of current review is centralized towards PTR-ToF-MS proton-transfer reaction time of flight mass the stimulating and refreshing organoleptic characteristics of cof- spectrometry fee being affected by postharvest processing and coffe processing pCoQLs p-coumaroyl-1,5-lactones factors. MWRs microwave roasters MLR multiple linear regression Acknowledgments PCA principal component analysis All the authors are highly indebted to the anonymous review- ROS reactive oxygen species ers of this manuscript whose comments/suggestion made this RH relative humidity manuscript scholarly/academically more valueable and interesting REMPI-ToF-MS resonance-enhanced multiphoton ionization for the readers. time of light spectrometry SFA saturated fatty acids Author Contributions SPME-GC/MS solid-phase microextraction-gass chromatog- A.H. and S.A.H. drafted this manuscript. H.A.R.S. and I.P. raphy mass spectrometry edited and reviewed the whole manuscript and provided sugges- SCAA specialty coffee association of America tions to main authors with critical input and corrections. M.U.I. SBRs spouted bed roasters and S.U. assisted in locating and interpreting the literature sources TBARS thiobarbituric acid-reactive substances whenever or/and wherever was necessary. All authors read and TPC total phenolic content approved the final manuscript. TFC total flavonoid content TAGs triacylglycerols Conflict of Interest TC Turkish coffee Authors declare no conflicts of interest. UFA unsaturated fatty acids VUV-SPI vacuum ultraviolet single-photon ionization Abbreviations w.b. wet basis ATRs atractyloside derivatives ATR-FT-IR attenuated total refelectance-fourier Refernces transform-infrared Abebe, M. (1987). Insect pests of coffee with special emphasis on antestia, APCI-MS atmospheric pressure chemical ionization Antestiopsis intricata,inEthiopa.Advances in Research on Tropical Entomolgy, mass spectrometry 8(4-6),977–998. Abreu, G. F. D., Rosa, S. D. V. F., Cirillo, M. A., Malta, M. R., Clemente, A. CQA caffeoylquinic acid C. S., & Borem,´ F. M. (2017). Simultaneous optimization of coffee quality CQLs caffeoyl-1,5-lactones variables during storage. Revista Brasileira de Engenharia Agr´ıcola e Ambiental, CWP cell wall polysaccharides 21(1), 56–60. https://doi.org/10.1590/1807-1929 CGA chlorogenic acid Agate, A. D., & Bhat, J. V. (1966). Role of pectinolytic yeasts in the CGALs chlorogenic acid lactones degradation of mucilage layer of Coffea robusta cherries. Applied Microbiology, 14(2), 256–260. CDR conventional drum roaster Ahmad, R., Tharappan, B., & Bongirwar, D. R. (2003). Impact of gamma CCM coupled convective–microwave irradiation on the monsooning of coffee beans. Journal of Stored Products CSBR cryogenic SBR Research, 39(2), 149–157. DIMS direct inlet mass spectrometry Albanese, D., Matteo, M., Poiana, M., & Spagnamusso, S. (2009). Espresso d.b. dry basis coffee (EC) by POD: Study of thermal profile during extraction process and influence of water temperature on chemical–physical and sensorial ESI-FT-ICR-MS electron spray ionization-Fourier transform properties. Food Research International, 42, 727–732. mass spectrometry

r 1228 ComprehensiveReviewsinFoodScienceandFoodSafety Vol.17,2018 C 2018 Institute of Food Technologists® Farm to consumer . . .

Alessandrini, L., Romani, S., Pinnavaia, G., & Rosa, M. D. (2008). time-temperature conditions. Journal of Agricuture and Food Chemistry, 56, Near-infrared spectroscopy: An analytical tool to predict coffee roasting 5836–5846.https://doi.org/10.1021/jf800327j degree. Annals of Chimeca Acta, 625, 95–102. Baggenstoss, J., Poisson, L., Kaegi, R., Perren, R., & Escher, F. (2008b). https://doi.org/10.1016/j.aca.2008.07.013 Roasting and aroma formation: Effect of initial moisture content and steam Akiyama, M., Murakami, K., Hirano, Y., Ikeda, M., Iwatsuki, K., Wada, A., treatment. Journal of Agriculture and Food Chemistry, 56, 5847–5851. . . . Iwabuchi, H. (2008). Characterization of headspace aroma compounds https://doi.org/10.1021/jf8003288 of freshly brewed Arabica coffees and studies on characteristic aroma Becker, R., Do¨hla, B., Nitz, S., & Vitzthum, O. G. (1988). Identification of compounds of Ethiopian coffee. Journal of Food Science, 73, 335–346. the “peasy” off-flavour note in central African coffees. In 12th International Akiyama, M., Murakami, K., Ikeda, M., Iwatsuki, K., Wada, A., & Tokuno, Conference on Coffee, Montreux. ASIC: Paris, France. 203– K. (2007). Analysis of the headspace volatiles of freshly brewed arabica 215. coffee using solidphase microextraction. Journal of Food Science, 72(7), Bekedam,E.K.,Loots,M.J.,Schols,H.A.,vanBoekel,M.A.J.S.,&Smit, 388–396. https://doi.org/10.1111/j.1750-3841.2007.00447.x G. (2008). Roasting effects on formation mechanisms of coffee brew Akiyama, M., Murakami, K., Ohtani, N., Iwatsuki, K., Sotoyama, K., Wada, melanoidins. Journal of Agriculture and Food Chemistry, 56(16), 7138– A., . . . Tanaka, K. (2003a). Analysis of volatile compounds released during 7145. the grinding of roasted co#ee beans using solid-phase microextraction. Bell, L. N., Wetzel, C. R., & Grand, A. N. (1996). Caffeine content in coffee Journal of Agricultural and Food Chemestry, 51, 1961–1969. as influenced by grinding and brewing techniques. Food Research Akiyama, M., Murakami, K., Ohtani, N., Iwatsuki, K., Sotoyama, K., Wada, International, 29(8), 185–189. A., . . . Tanaka, K. (2003b). Characterization of flavor compounds during Berjak, P., & Pammenter, N. W. (2007). From Avicennia to Zizania: Seed grinding of roasted coffee beans. In K. D. Deibler & J. Delwiche (Eds.), recalcitrance in perspective. Annals of Botony, 101(02), 1–16. “Handbook of Flavor Characterization:Sensory Analysis, Chemistry, and Physiology” (pp. 231–258). New York: Marcel Dekker, Inc. Bernard, M., Kraehenbuehl, K., & Roberts, D. D. (2005). Interactions between volatile and nonvolatile coffee components. 1. Screening of the Alemayehu, T., Esayas, K., & Kassu, K. (2008). Coffee development and nonvolatile components. Journal of Agriculture and Food Chemistry, 53, marketing improvement plan. In: Proceedings of a National Workshop Four 4417–4425. Decades of Coffee Research and Development in Ethiopia, (pp. 375–381), 14–17 August 2007, Ethiopian Institute of Agricultural Research (EIAR), Bertone, E., Venturello, A., Giraudo, A., Pellegrino, G., & Geobaldo, F. Addis Ababa, Ethiopia. (2016). Simultaneous determination by NIR spectroscopy of the roasting degree and Arabica/Robusta ratio in roasted and ground coffee. Food Amanpour, A., & Selli, S. (2015). Differentiation of volatile profiles and odor Control, 59, 683–689. https://doi.org/10.1016/j.foodcont.2015.06.055 activity values of turkish coffee and french press coffee. Journal of Food Processing and Preservation, 40(5), 1116–1124. Blank,I.,Pascual,E.C.,Devaud,S.,Fay,L.B.,Stadler,R.H.,Yeretzian,C., https://doi.org/10.1111/jfpp.12692 & Goodman, B. A. (2002). Degradation of coffee flavor compound 2-furfurylmercaptan in model Fenton-type reaction system. Journal of Ameyu, M. A. (2017). Influence of harvesting and postharvest processing Agriculture and Food Chemistry, 50, 2356–2364. methods on the quality of Arabica coffee (Coffeaarabica L.) in Eastern Ethiopia. ISAAB Journal of Food and Agricutlural Science, 7(1), 1–9. Blank, I., Sen, A., & Grosch, W. (1991). Aroma impact compounds of arabica https://doi.org/10.5897/ISABB-JFAS2016.0051 and robusta coffee.Qualitative and quantitative investigations. Proceedings of the 14th International Colloquium on the Chemistry of Coffee, San Amorim, H. V., Basso, L. C., Crocomo, O. J., & Teixeira, A. A. (1977). Francisco, CA; ASIC: Paris, France. 117–129. Polyamines in green and roasted coffee. Journal of Agricultural and Food Chemistry, 25(4), 957–958. Blank, I., Sen, A., & Grosch, W. (1992). Potent odorants of the roasted powder and brew of Arabica coffee. Zeitschrift fur¨ Lebensmittel-Untersuchung Andrade, A. C., Carazzolle, M. F., Colombo, C. A., Costa, G. G. L., und –Forschung, 195(3), 239–245. https://doi.org/10.1007/bf01202802 Mondego, J. M. C., & Parizzi, L. P. (2011). An EST-based analysis identifies new genes and reveals distinctive gene expression features of Coffea arabica Bicho, N. C., Leitao,˜ A. E., Ramalho, J. C., deAlvarenga, N. B., & Lidon, F. and Coffea canephora. BMC Plant Biology, 11–30. C. (2013). Impact of roasting time on the sensory profile of Arabica and https://doi.org/10.1186/1471-2229-11-30 Robusta coffee. Ecology of Food and Nutrition, 52(2), 163–177. Andueza, S., Pena, M. P. D., & Cid, C. (2003a). Chemical and sensorial Bonnlander, B., Cappuccio, R., Suggi-Liverani, F., & Winterhalter, P. characteristics of espresso coffee as affected by grinding and torrefacto roast. (2006). Analysis of enantiomeric linalool ratio in green and roasted coffee. Journal of Agricultural and Food Chemistry, 51, 7034–7039. JournalofFlavorFragrance, 21, 637–641. https://doi.org/10.1021/jf034628f Bonnlander, B., Colomban, and S., & Lonzarich, V. (2007a). Lipidomics and Andueza,S.,Maeztu,L.,Pascual,L.,Iba´nez,˜ C., de Pena,˜ M. P., & Cid, C. the influence on quality of coffee. Colloque Scientifique International sur le (2003b). Influence of extraction temperature on the final quality of espresso Cafe,´ Montpellier, France. 161–168. coffee. Journal of Science of Food and Agriculture, 83, 240–248. Bonnlander, B., Rivetti, D., Suggi, G., & Liverani, F. (2007b). Sugars in Andueza, S., Maeztu, L., Dean, B., de-Pena,˜ M. P., Bello, J., & Cid, C. green coffee beans—Important aroma precursors and their correlation to (2002). Influence of water pressure on the final quality of Arabica espresso post harvest method. 21eme` Colloque Scientifique International sur le Cafe,´ coffee. Application of multivariate analysis. Journal of Agricultural and Food Montpellier, France, 11–15, 461–468. Chemistry, 50, 7426–7427. Booth, C., Cummins, C., Dalwadi, M., & Dellar, P. (2012). Brewing of filter Arackal, T., & Lehmann, G. (1979). Measurement of the coffee. In Technical Report from MACSI’s Problem-solving Workshop with 2-methylfuran/2-butanone quotient of unground coffee during storage out Industry, Report MACSI/ESGI/0033. (Retrieved from of contact with air. Chemie, Mikrobiologie, Technologie der Lebensmittel, 6, https://www.maths-in industry.org/miis/605/1/Philips_Research_ESGI_ 43–47. 87_Report.pdf). (Accessed 21 October 2016). Avallone, S., Guyot, B., Brillouet, J. M., Olguin, E., & Pierre, G. J. (2001). Borem, F. M., Isquierdo, E. P., Oliveira, P. D., Ribeiro, F. C., Siqueira, V. C., Microbiological and biochemical study of coffee fermentation. Current & Taveira, J. H. D. S. (2014). Effect of intermittent drying and storage on Microbiology, 42, 252–256. parchment coffee quality. Biological Science Journal, 30, 609–616. Avelino,A.,Barboza,B.J.,Carlos,J.,Araya,J.C.,Fonseca,C.,Davrieux,F., Borem, F. M., Ribeiro, F. C., Figueiredo, L. P., Giomo, G. S., Fortunato, V. . . . Cilas, C. (2005). Effects of slope exposure, altitude and yieldon coffee A., & Isquierdo, E. P. (2013). Evaluation of the sensory and color quality of quality in two altitude terroirsof Costa Rica, Orosi and Santa Mar´ıa de Dota. coffee beans stored in hermetic Packaging. Journal of Stored Products Research, Journal of the Science of Food & Agriculture, 85, 52,1–6.https://doi.org/10.1016/j.jspr.2012.08.004 1869–1876.https://doi.org/10.1002/jsfa.2188 Borem, F. M. (2008a). Processamento do cafe.´ In Pos-Colheita´ do Cafe(F.M.´ Bade-Wegner, H., Bendig, I., Holscher, W., & Wollmann, R. (1998). Volatile Boremed.)´ Editora UFLA, Lavras. 203–240. compounds associated with the over-fermented flavor defect. In 17th Borem, F.M., Nobre, G., W., Fernandes, S. M., Pereira, R. G. F. A., & International Conference on Coffee, Nairobi, ASIC: Paris France. 176– Oliveira, P. D. (2008b). Avaliac¸ao˜ sensorial do cafe´ cereja descascado, 182. armazenado sob atmosfera artificial e convencional. Cienciaˆ e Agrotecnologia, Baggenstoss, J., Thomann, D., Perren, R., & Escher, F. (2010). Aroma 32, 1724–1729. Recovery from Roasted Coffee by Wet Grinding. Journal of Food Science, 75, Borges, M. L. A., Mendonc¸a,J.C.F.,Franca,A.S.,Oliveira,L.S.,& 9. https://doi.org/10.1111/j.1750-3841.2010.01822.x Correa, P. C. (2004). Efeito da torrac¸ao˜ em parametrosˆ f´ısicos de cafes´ de Baggenstoss, J., poisson, L., Kaegi, R., Perren, P., & Escher, F. (2008a). diferentes qualidades. Revista Brasileira de Armazenamento,v.EspecialCafe6–´ Coffee roasting and aroma formation. application of different 13.

r C 2018 Institute of Food Technologists® Vol.17,2018 ComprehensiveReviewsinFoodScienceandFoodSafety 1229 Farm to consumer . . .

Brando, C. (2004). Harvesting and green coffee processing. In: J. N. Craig, A. P., Franca, A. S., Oliveira, L. S., Irudayaraj, J., & Ileleji, K. (2015). Wintgens (ed.), Coffee: Growing, processing, sustainable production. 2004 Fourier transform infrared spectroscopy and near infrared spectroscopy for (pp. 604–715). Weinheim: Wiley-VCH. the quantification of defects in roasted coffees. Talanta, 134, 379–386. Brasil (2010). Ministerio´ da Agricultura Pecuaria´ e Abastecimento. Instruc¸ao˜ https://doi.org/10.1016/j.talanta.2014.11.038 Normativa n° 16 de 24 de maio de 2010. Bras´ılia: MAPA Diario´ Oficial. Craig, I. D., Parker, R., Rigby, N. M., Cairns, P., & Ring, S. G. (2001). Uniao.n˜ ° 98, terc¸a-feira, 25 de maio de. Maillard reaction kinetics in model preservation systems in the vicinity of Brattoli, M., Cisternino, E., & Dambruoso, P. (2013). Gas chromatography the glass transition: Experiment and theory. Journal of Agriculture and Food analysis with olfactometric detection (GC-O) as a useful methodology for Chemistry, 49(10), 4706–4712. chemical characterization of odorous compounds. Sensors, 13, Crozier, A., Jaganath, I. B., & Clifford, M. N. (2009). Natural Product Reports, 16759–16800. https://doi.org/10.3390/s131216759 26, 1001–1043. Bucheli, P., Meyer, I., Pittet, A., Vuataz, G., & Viani, R. (1998). Industrial Correa,P.C.,Oliveira,G.H.H.D.,Oliveira,A.P.L.R.D.,Vargas-El´ıas, G. storage of green Robusta coffee under tropical conditions and its impact on A., Santos, F. L., & Baptestini, F. M. (2016a). Preservation of roasted and raw material quality and ochratoxin A content. Journal of Agriculture and Food ground coffee during storage. Part 1: Moisture content and repose Chemistry, 46, 4507–4511. angleRevista Brasileira de Engenharia. Agr´ıcola e Ambiental, 20(6), 581–587. Buffo, R. A., & Cardelli-Freire, C. (2004). Coffee flavour: An overview. https://doi.org/10.1590/1807-1929/agriambi.v20n6p581-587 Journal of Flavour and Fragrance, 19(2), 99–104. Correa, P. C., Oliveira, G. H. H., Vasconcelos, W. L., Vargas-Elias, G. A., https://doi.org/10.1002/ffj.1325 Santos, F. L., & Nunes, E. H. M. (2016b). Preservation of roasted and Burdock, G. A. (2010). Fenaroli’s handbook of flavor ingredients (6th ed.). Boca ground coffee during storage. Part 2: Bulk density and intergranular Raton: CRC Press/Taylor & Francis. porosityRevista Brasileira de Engenharia. Agr´ıcola e Ambiental, 20(7), 666–671. https://doi.org/10.1590/1807-1929/agriambi.v20n7p666-671 Bytof, G., Knopp, S. E., Kramer, D., Breitenstein, B., Bergervoet, J. H. W., & Groot, S. P. C. (2007). Transient occurrence of seed germination processes Correa,P.C.,Oliveira,G.H.H.,Botelho,F.M.,Goneli,A.L.D.,& during coffee post-harvest treatment. Annals of Botony, 100(1), 61–66. Carvalho, F. M. (2010). Modelagem matematica´ e determinac¸ao˜ das propriedades termodinamicasˆ do cafe(´ Coffea arabica L.) durante o processo Bytof, G., Knopp, S. E., Schieberle, P., Teutsch, I., & Selmar, D. (2005). de secagem. Revista Ceres, 57, 595–601. Influence of processing on the generation of g-aminobutyric acid in green https://doi.org/10.1590/S0034-737X2010000500005 coffee beans. European Journal of Food Research and Technology, 220, 245–250. https://doi.org/10.1007/s00217-004-1033-z Correa, P. C., Afonso Junior, P. C., Silva, F. S., & Ribeiro, D. M. (2003). Qualidade dos graos˜ de cafe(´ Coffea arabica L.) durante o armazenamento em Cammerer, B., & Kroh, W. (2006). Antioxidant activity of coffee brews. condic¸oes˜ diversas. Revista Brasileira de Armazenamento, 7, 137–147. European Food Research and Technology, 223, 469–474. Czerny, M., & Schieberle, P. (2001).Changes in roasted coffee aroma during Campanha, F. G., Dias, R. C. E., & Benassi, M. T. (2010). Discrimination of storage-influence of the packaging, Proceedings of the 19th International coffee species using kahweol and cafestol effects of roasting and of defects. Conference on Coffee Science, Trieste, Italy. Coffee Science, 05, 87–96. Czerny, M., & Grosch, W. (2000). Potent odorants of raw Arabica coffee. Caprioli, G., Cortese, M., Sagratini, G., & Vittori, S. (2015). The influence Their changes during roasting. Journal of Agricultural and Food Chemistry, of different types of preparation (espresso and brew) on coffee aroma and 48-868-72. bioactive constituents. International Journal of Food Science and Nutrition, 66(5), 505–513. https://doi.org/10.3109/09637486.2015.1064871 Czerny, M., Mayer, F., & Grosch, W. (1999). Sensory study on the character impact odorants of roasted Arabica coffee, Journal of Agricultural and Food Casal, S., Oliveira, M. B., & Ferreira, M. A. (2000). HPLC diode-array Chemistry, 47, 697–699. applied to the thermal degradation of trigonelline, nicotinic acid and caffeine in coffee. Food Chemistry, 68, 481–485. Da Silveira-Duarte, S. M., De Abreu, C. M. P., Castle de Menezes, H., Dos Santos, M. H., & Paiva-Gouvea, C. M. C. (2005). Effect of processing and CCI (Centro de Comercio Internacional). (2002). Calidad del cafe.´ In roasting on the antioxidant activity of coffee brews.Cie˛ncia e Tecnologia de CCI-UNCTAD/OMC (Ed.), Cafe´ guia del exportador. (pp. 270–321). Alimentos, 25, 387–393. Ginebra: Centro de Comercio Internacional UNCTAD/OMC. Dawidowicz, A. L., & Typek, R. (2017). Transformation of chlorogenic acids Celinska, E., Kubiak, P., Bialas, W., Dziadas, M., & Grajek, W. (2013). during the coffee beans roasting process. European Food Resrach and Yarrowia lipolytica: The novel and promising 2-phenylethanol producer. Te c h no lo gy, 243, 379–390. https://doi.org/10.1007/s00217-016-2751-8 Journal of Industrial Microbiology and Biotechnology, 40, 389–392. Decazy, F., Avelino, J., Guyot, B., Perriot, J., Pineda, C., & Cilas, C. (2003). Charles-Bernard, M., Roberts, D. D., & Kraehenbuehl, K. (2005). Quality of different honduran coffees in relation to several environments. Interactions between volatile and non-volatile coffee components. 2. Journal of Food Science, 68(7), 2356-61. mechanistic study focused on volatile thiols. Journal of Agriculture and Food Chemistry, https://doi.org/10.1021/jf048020y Delgado-Andrade, C., & Morales, F. J. (2005). Unravelling the contribution of melanoidins to the antioxidant activity of coffee brews. Journal of Charles-Bernard, M., Kraehenbuehl, K., & Roberts, D. D. (2003). Influence Agricultural and Food Chemistry, 53, 1403–1407. of coffee brew non-volatiles on coffee brew stability. In J. L. Le Quere & P. X. Etievant (Eds.), Fla Vour Research at the Dawn of the Twenty-First Century De Reu, J. C., Linssen, V. A. J. M., Rombouts, F. M., & Nout, M. J. R. (pp. 552–555). Paris: Editions Tec et Doc. (1997). Consistency, polysaccharidase activities and non-starch polysaccharides content of soya beans during tempe fermentation. Journal of Cheong, M. W., Tong, K. H., Ong, J. J. M., Liu, S. Q., Curran, P., & Yu, B. Agricultural and Food Chemistry, 73(3), 357–363. (2013). Volatile composition and antioxidant capacity of Arabica coffee. Food Research International, 51(1), 388–396. Dias, R. C. E., de Faria-Machado, A. F., Mercadante, A. Z., Bragagnolo, N., & de Toledo Benassi, M. (2014). Roasting process affects the profile of Cirilo,M.P.G.,Coelho,A.F.S.,Ara,U.C.M.,Goncalves,F.R.B., diterpenes in coffee. European Food Research and Technology, 239, 961–970. Nogueira, F. D. E., & Gloria, M. B. A. (2003). Profile and levels of bioactive https://doi.org/10.1007/s00217-014-2293-x amines in green and roasted coffee. Food Chemistry, 82, 397– 402. Dias, E. C., Borem, F. M., Pereira, R. G. F. A., & Guerreiro, M. C. (2012). Amino acid proWles in unripe Arabica coVee fruits processed using wet and Clarke, R. J., & Macrae, R. (1987). Coffee. In Te c h no lo gy (first ed., Vol. 2). dry methods. European Food Research and Technology, 234, 25–32. London: Elsevier Applied Science. https://doi.org/10.1007/s00217-011-1607-5 Clarke, R. J. (1987). Roasting and Grinding. In R. J. Clarke & R. Macrae Workefied Blog, D. R. (2013). The difference between semi-washed and (Eds.), Coffee Tech (Vol. 2, pp. 73–108). London: Elsevier Applied Science. fully-washed methods. Retrieved from Clarke, R. J. (1985). Green coffee processing. In M. N. Clifford & K. C. https://news.drwakefield.com/coffee-trading/the-diffrence-betweensemiwas Willson (Eds.), Coffee botany, biochemistry and production of beans and beverage hedand-fullywashed-methods/80162181 (pp. 149–196). London: Croom Helm. Do Livramento, K. G., Borem, F. M., Jose, A. C., Santos, A. V., do Clarke, R. J., & Macrae, R. (Eds.) (1985). Coffee Technology (Vol. 2). Livramento, D. E., Alves, J. D., & Paiva, L. V. (2017). Proteomic analysis of Clifford, M. N. (1997). The nature of chlorogenic acids. Are they coffee grains exposed to different drying process. Food Chemistry, 221, advantageous compounds? Proc. 17th Int. Sci. Coll. Coffee (Nairobi), 1874–1882 https://doi.org/10.1016/j.foodchem.2016.10.069 Association Scientifique Internationale du Cafe´ (Paris) 79–91. Dold, S., Lindinger, C., Kolodziejczyk, E., Pollien, P., Ali, S., Germain, J. C., Clifford, M. N., & Willson, K. C. (1985). Coffee: Botany, biochemistry and . . . Hartmann, C. (2010). Influence of foam structure on the release kinetics production of beans and beverage. Westport, Connecticut: The AVI Publishing of volatiles from espresso coffee prior to consumption. Journal of Agricultural Company Inc. and Food Chemistry, 59, 11196–11203. https://doi.org/10.1021/jf201758h

r 1230 ComprehensiveReviewsinFoodScienceandFoodSafety Vol.17,2018 C 2018 Institute of Food Technologists® Farm to consumer . . .

Dold, S., Lindinger, C., Kolodziejczyk, E., Pollien, P., Ali, S., Germain, J. C., Feng, X., Dong, H., Yang, P., Yang, R., Lu, J., Lv, J., & Sheng, J. (2016). . . . Hartmann, C. (2011). Influence of foam structure on the release kinetics Culture-Dependent and -Independent Methods to Investigate the of volatiles from espresso coffee prior to consumption. Journal of Agricultural Predominant Microorganisms Associated with Wet Processed Coffee. and Food Chemistry, 59, 11196–11203. https://doi.org/10.1021/jf201758h Current Microbiology, 73, 190–195. Dorfner, R., Ferge, T., Yeretzian, C., Kettrup, A., & Zimmermann, R. https://doi.org/10.1007/s00284-016-1047-3 (2004). Laser mass spectrometry as on-line sensor for industrial process Fickers, P., Benetti, P. H., Wache, Y., Marty, A., Mauersberger, S., & Smit, analysis: Process control of coffee roasting. Annals of Chemistry, 76, M. S. (2005). Hydrophobic substrate utilisation by the yeast Yarrowia 1386–1402. lipolytica, and its potential applications.FEMS Yeast Research, 5, 527– Duarte, G. S., Pereira, A. A., & Farah, A. (2010). Chlorogenic acids and 543. other relevant compounds in Brazilian coffees processed by semi-dry and Fischer, M., Reimann, S., Trovato, V., & Redgwell, R. J. (2001). wet post-harvesting methods. Journal of Agricultural and Food Chemistry, 118, Polysaccharides of green Arabica and Robusta coffee beans.Carbohydrate 851–855. https://doi.org/10.1016/j.foodchem.2009.05.042 Research, 330, 93–101. Durand, N., Sheikha, A. F. E., Quiros, M. L. S., Oscar, G. R., Nganou, N. Flament, I. (2002). Coffee flavor chemistry. Chichester, England: John Wiley & D., Fontana-Tachon, A., & Montet, D. (2013). Application of PCR-DGGE Sons Ltd. to the study of dynamics and biodiversity of yeasts and potentially OTA Food and Agriculture Organization. (2006). Fermentation of coffee – producing fungi during coffee processing. Food Control, 34, 466– Control of operation. In Good hygiene practices along the coffee chain. 471. https://doi.org/10.1016/j.foodcont.2013.05.017 Franca, A. S., Oliveira, L. S., Oliveira, R. C. S., Agresti, P. C. M., & Augusti, Dussert, S., Davey, M. W., Laffargue, A., Doulbeau, S., Swennen, R., & R. (2009). A preliminary evaluation of the effect of processing temperature Etienne, H. (2006). Oxidative stress, phospholipid loss and lipid hydrolysis on coffee roasting degree assessment. Journal of Food Engineering, 92, during drying and storage of intermediate seeds. Physiologia Plantarum, 127, 345–352. 192–204. Franca,A.S.,Oliveira,L.S.,Mendonc¸a, J. C. F., & Silva, X. A. (2005). Dutra, E. R., Oliveira, L. S., Franca, A. S., Ferraz, V. P., & Afonso, R. J. C. F. Physical and chemical attributes of defective crude and roasted coffee beans. (2001). A preliminary study on the feasibility of using the composition of Food Chemistry, 90, 84–89. coffee roasting exhaust gas for the determination of the degree of roast. Journal of Food Engineering, 47(3), 241–246. Frankel, E. N. (2005). Lipid oxidation (2nd ed.). Bridgwater: The Oil Press. EFSA, European Food Safety Authority. (2008). Polycyclic aromatic Frank, H. A., Lum, N. A., & Cruz, A. S. D. (1965). Bacteria responsible for hydrocarbons in food, scientific opinion of the panel on contaminants in the mucilage-layer decomposition in kona coffee cherries. Applied Microbiology, food chain. The EFSA Journal, 724, 1–114. 13, 201–207. Eggers, R., & Pietsch, A. (2001). Technology I: Roasting. In R. J. Clarke & Fogliano, V., Monti, S. M., Musella, T., Randazzo, G., & Ritieni, A. (1999). O. G. Vitzthum (Eds.), Coffee recent developments 90–107. Oxford: Blackwell Formation of coloured Maillard reaction products in a gluten–glucose Science Ltd. model system. Food Chemistry, 66, 293–299. Espresso and coffee guide. (2008). Semi-washed processing. Retrieved from Fond, O. (1995). Effect of water and coffee acidity on extraction. Dynamics of coffee bed compaction in Espresso type extraction, ASIC. In Proceedings https://www.espressocoffeeguide.cm/all-about-coffee-2/coffee-processing/ th semi-washed-processing 16 colloquium Kyoto, Japan 413–421. Evangelista,S.R.,Miguel,M.G.D.P.C.,Silva,C.F.,Pinheiro,A.C.M.,& Gardner, D. G. (1958). Effect of certain ion combinations commonly found Schwan, R. F. (2015). Microbiological diversity associated with the in potable water on rate of filtration through roasted and ground coffee. spontaneous wet method of coffee fermentation. International Journal of Food Food Reseearch, 23, 76–84. Microbiology, 210, 102–112. Geiger, R., Perren, R., Kuenzli, R., & Escher, F. (2005). Carbon dioxide https://doi.org/10.1016/j.ijfoodmicro.2015.06.008 evolution and moisture evaporation during roasting of coffee beans. Journal Evangelista,S.R.,Miguel,M.G.C.P.,Cordeiro,C.S.,Silva,C.F., of Food Science, 70–72. Pinheiro, A. C. M., & Schwan, R. F. (2014a). Inoculation of starter cultures Geromel, C., Ferreira, L. P., Davrieux, F., Guyot, B., Ribeyre, F., Scholz, M. in a semi-dry coffee (Coffea arabica) fermentation process. Food Microbiology, B. S., . . . Marraccini, P. (2008). Effects of shade on the development and 44, 87–95. sugar metabolism of coffee (Coffea arabica L.) fruits. Plant Physiology and Evangelista, S. R., Silva, C. F., Miguel, M. G. C.P., Cordeiro, C. S., Pinheiro, Biochemistry, 46, 569–579. https://doi.org/10.1016/j.plaphy.2008.02.006 A. C. M., Duarte, W. F., & Schwan, R. F. (2014b). Improvement of coffee Giungatob, P., Laiola, E., & Nicolardi, V. (2017). Evaluation of industrial beverage quality by using selected yeasts strains during the fermentation in roasting degree of coffee beans by using an electronic nose and a stepwise dry process. Food Research International, 61, 183–195. backward selection of predictors.Food Analytical Methods, Farah, A. (2005). Distribuic¸ao˜ nos graos,˜ importanciaˆ na qualidade da bebida https://doi.org/10.1007/s12161-017-0909- e biodisponibilidade dos acidos´ clorogenicosˆ do cafe.´ Ph.D. Thesis, UFRJ Gloess, A. N., Vietri, A., & Wieland, F. (2014). Evidence of different flavor (in Portuguese). formation dynamics by roasting coffee from different origins: Online Farah, A., DePaulis, Trugo, L. C., & Martin, P. R. (2005a). Formation of analysis with PTR-ToF-MS. International Journal of Mass Spectrometery, chlorogenic acids lactones in roasted coffee. Journal of Agricultural and Food 365–366, 324–337. https://doi.org/10.1016/j.ijms.2014.02.010 Chemistry, 53, 1105–1113. Gloess, A. N., Schonb¨ achler,¨ B., Klopprogge, B., D’Ambrosio, L., Chatelain, Farah, A., Paulis, T., Trugo, L. C., & Martin, P. (2005b). Effect of Roasting K., Bongartz, A., . . . Yeretzian, C. (2013). Comparison of nine common on the Formation of Chlorogenic Acid Lactones in Coffee. Journal of coffee extraction methods: Instrumental and sensory analysis. European Food Agricultural and Food Chemistry, 53, 1505–1513. Research and Technology, https://doi.org/10.1007/s00217-013-1917-x https://doi.org/10.1021/jf048701t Godinho, R. P., Vilela, E. R., Oliveira, G. A., & Chagas, S. J. R. (2000). Farah, A., Monteiro, M. C., Calado, V., Franca, A. S., & Trugo, L. C. Var iac¸oes˜ na corena composic¸ao˜ qu´ımica do cafe(´ Coffea arabica L.) (2006a). Correlation between cup quality and chemical attributes of armazenado em cocoˆ e beneficiado. Revista Brasileira de Brazilian coffee. Food Chemistry, 98, 373–380. Armazenamento-Especial, 1, 38–43. Farah, A., & Donangelo, C. M. (2006b). Phenolic compounds in coffee. Gokmen, V., & S¸enyuva, H. Z. (2006). Study of colour and acrylamide Brazilian Journal Plant Physiology, 18(1), 23–36. formation in coffee, wheat flour and potato chips during heating. Food Chemistry, 99, 238–243. https://doi.org/10.1016/j.foodchem.2005.06.054 FAO. (2010).Introduction to Coffee Drying.‘Good hygiene practices in the primary production of coffee. Internet doc. Retrieved from Gonzalez-Rios, O., Suarez-Quiroz, M., Boulanger, R., Barel, M., Guyot, B., https://www.Zecuppa.com/coffeetermsbean-grading.htm#top. Accessed Guiraud, J. P., & Schorr-Galindo, S. (2007a). Impact of post-harvest on April 2017/ processing on the volatile fraction of coffee beans: I. Green coffee. Journal of Food Composition and Analysis, https://doi.org/10.1016/j.jfca.2006.07.009 Falasconi, M., Pardo, M., & Sberveglieri, G. (2005). The novel EOS835 electronic nose and data analysis for evaluating coffee ripening. Sensors Gonzalez-Rios, O., Suarez-Quiroz, M. L., Boulanger, R., Barel, M., Guyot, Actuators B: Chemical, 110, 73–80. B., Guiraud, J., & Schorr-Galindo, S. (2007b). Impact of ‘‘ecological’’ https://doi.org/10.1016/j.snb.2005.01.019. post-harvest processing on coffee aroma: II. Roasted coffee. Journal of Food Composition and Analysis, 20, 297–307. Fernandes,S.M.,Pereira,R.G.F.A.,Borem,F.M.,Nery,F.C.,&Padua,F. https://doi.org/10.1016/j.jfca.2006.12.004 R. M. (2006). Alterac¸oes˜ qu´ımicas em cafes´ torrados e mo´ıdos durante o armazenamento. Revista Brasileira de Armazenamento, v.Especial Cafe´, Gretsch, C., Sarrazin, C., & Liardon, R. (1999). Evolution of Coffee Aroma 12–18. Characteristics during Roasting. In 18th International Scientific

r C 2018 Institute of Food Technologists® Vol.17,2018 ComprehensiveReviewsinFoodScienceandFoodSafety 1231 Farm to consumer . . .

Colloquium on Coffee, Helsinki, Finland; Association Scientifique Illy, A., & Viani R. (2005). Espresso coffee: The science of quality (pp. 200–400) Internationale pour le Cafe´: Paris, France 27–34. Amsterdam: Elsevier academic press. Grosch, W. (2001). Chemistry III: Volatile compounds. In R. J. Clarke & O. Jackels, S. C., Jackels, C. F., Vallejos, C., Kleven, S., Rivas, R., & G. Vitzthum (Eds.), Coffee: Recent developments (pp. 68–89). Oxford: Fraser-Dauphinee, S. (2006). Control of the coffee fermentation process and Blackwell Science. quality of resulting roasted coffee: Studies in the field laboratory and on Guichard, E. (2002). Interactions between flavor compounds and food small farms in Nicaragua during the 2005–06 harvest. In 21st International ingredients and their influence on flavor perception. Food Review Scientific Colloquium on Coffee – Post-harvest processing and green coffee International, 18, 49–70. quality. Montpellier, France. Guyot, B., Gueule, D., Manez, J. C., Perriot, J. J., Giron, J., & Villain, L. Jackels, S. C., & Jackels, C. F. (2005a). U.S. Patent No. US20060204620. (1996). Influence de l’altitude et de l’ombrage des caf´es Arabica. Plant Rech Washington, DC: U.S. Patent and Trademark Office. D´evelop, 3, 272–280. Jackels, S. C., & Jackels, C. F. (2005b). Characterization of the coffee Guyot, B., Petnga, E., Lotod´e, R., & Vincent, J. C. (1988). Analyse mucilage fermentation process using chemical indicators: A field study in qualitative d’un caf´e Coffea canephora var. Robusta en function de la Nicaragua. JournalofFoodScience, 70(5), 321–325. maturit´e. Partie II.Application de l’analyse statistique multidimensionnelle. Jaiswal, R., Matei, M. F., Golon, A., Witt, M., & Kuhnert, N. (2012). Caf´e Cacao Th´e, 32, 229–242. Understanding the fate of chlorogenic acids in coffee roasting using mass Hinz, T., Steer, A., Waldmann, C., Cammenga, H. K., & Eggers, R. (1997). spectrometry based targeted and non-targeted analytical strategies, Food and Reostkaffee-Extraktion: Einflußparameter und Modellierung. Chemie Function, 3, 976–984. Ingenieur Technik, 69(5), 685–690. Joet,¨ T., Laffargue, A., Salmona, J., Doulbeau, S., Descroix, F., Bertrand, B., Holscher, W., Vitzthum, O. G., & Steinhart, H. (1990). Identification and . . . Dussert, S. (2014). Regulation of galactomannan biosynthesis in coffee sensorial evaluation of aroma impact compounds in roasted colombian seeds. Journal of Experimental Botany, 65, 323–337. coffee. Cafe, Cacao, 34, 205–212. Joet,¨ T., Laffargue, A., Descroix, F., Doulbeau, S., Bertrand, B., de kochko, Hamdouche, Y., Meile, J. C., Nganou, D. N., Durand, N., Teyssier, C., & A., & Dussert, S. (2010). Influence of environmental factors, wet processing Montet, D. (2016). Discrimination of post-harvest coffee processing and their interactions on the biochemical composition of green Arabica methods by microbial ecology analyses. Food Control, 65, 112–120. coffee beans. Food Chemistry, 118, 693–701. https://doi.org/10.1016/j.foodcont.2016.01.022 https://doi.org/10.1016/j.foodchem.2009.05.048 Hecimovic, I., Belscak-Cvitanovic, A., Horzic, D., & Komes, D. (2012). Kallio, H., Leino, M., Koullias, K., Kallio, S., & Kaitaranta, J. (1990). Comparative study of polyphenols and caffeine in different coffee varieties Headspace of roasted ground coffee as an indicator of storage time. Food affected by the degree of roasting. Food Chemistry, 129, 991–1000. Chemistry, 36, 135–148. https://doi.org/10.1016/j.foodchem.2011.05.059 Kesen, S., Kelebek, H., & Selli, S. (2014). Characterization of the key aroma Hernndez, J. A., Heyd, B., Irles, C., Valdovinos, B., & Trystram, G. (2010). compounds in Turkish olive oils from different geographic origins by Analysis of the heat and mass transfer during coffee batch roasting. Journal of application of aroma extract dilution analysis (AEDA). Journal of Agricultural Food Engineering, https://doi.org/10.1016/j.jfoodeng.2005.12.041 and Food Chemistry, 62(2), 391–401. Hernandez, J. A., Heyd, B., & Trystram, G. (2008). Prediction of brightness Kramer, D., Breitenstein, B., Kleinwachter, M., & Selmar, D. (2010). Stress and surface area kinetics during coffee roasting. Journal of Food Engineering, metabolism in green coffee beans (Coffea arabica L.): Expression of dehydrins 89, 156–163. https://doi.org/10.1016/j.jfoodeng.2008.04.026 and accumulation of GABA during drying. Plant Cell Physiology, 51(4), 546–553. https://doi.org/10.1093/pcp/pcq019 Hernandez,´ J. A., Heyd, B., Irles, C., Valdovinos, B., & Trystram, G. (2007). Analysis of the heat and mass transfer during coffee batch roasting. Journal of Kreuml, M. T. L., Majchrzak, D., Ploederl, B., & Koenig, J. (2013). Changes Food Engineering, 78, 1141–1148. in sensory quality characteristics of coffee during storage. Food Science and Nutrition, 1(4), 267–272. https://doi.org/10.1002/fsn3.35 Hertz-Schunemann, R., Dorfner, R., & Yeretzian, C. (2013a). On-line process monitoring of coffee roasting by resonant laser ionisation timeof- Kleinwchter, M., Bytof, G., & Selmar, D. (2015). Coffee beans and flight mass spectrometry: Bridging the gap from industrial batch roasting to processing. Coffee in health and disease prevention flavour formation inside an individual coffee bean. Journal of Mass https://doi.org/10.1016/B978-0-12-409517-5.00009-7 Spectrometery, 48, 1253–1265. https://doi.org/10.1002/jms.3299 Kleinwachter, M., & Selmar, D. (2010). Influence of drying on the content Hertz-Schunemann, R., Streibel, T., Ehlert, S., & Zimmermann, R. of sugars in wet processed green Arabica coffees. Food Chemistry, 119, (2013b). Looking into individual coffee beans during the roasting process: 500–504. https://doi.org/10.1016/j.foodchem.2009.06.048 Direct micro-probe sampling on-line photo-ionisation mass spectrometric Knopp, S., Bytof, G., & Selmar, D. (2006). Influence of processing on the analysis of coffee roasting gases. Analytical and Bioanalytical Chemistry, 405, content of sugars in green Arabica coffee beans. European Food Research and 7083–7096. https://doi.org/10.1007/s00216-013-7006-y Te c h no lo gy, 223, 195–201. https://doi.org/10.1007/s00217-005-0172-1 Holscher, W., & Steinhart, H. (1992). Investigation of roasted coffee freshness Knopp, S., Bytof, G., & Selmar, D. (2005). Influence of processing on the with an improved headspace technique. Zeitschrift fur¨ content of sugars in green Arabica coffee beans. European Food Research & Lebensmittel-Untersuchung und –Forschung, 195, 33–38. Technolology, 223, 195–201. Houessou, J. K., Maloug, S., Leveque, A. S., Delteil, C., Heyd, B., & Camel, Kosova, K., V´ıtamvasa, P., Prasil, T. I., & Renaut, J. (2011). Plant proteome V. (2007). Effect of roasting conditions on the polycyclic aromatic changes under abiotic stress — Contribution of proteomics studies to hydrocarbon content in ground arabica coffee and coffee brew. Journal of understanding plant stress response. Journal of Proteomics, 74(8), 1301–1322. Agricultural and Food Chemistry, 55, 9719–9726. Kouadio, I. A., Koffi, L. B., Nemlin, J. G., & Dosso, M. B. (2012). Effect of Hwang, C. F., Chen, C. C., & Ho, C. T. (2012). Contribution of coffee Robusta (Coffea canephora P.) coffee cherries quantity put out for sun proteins to roasted coffee volatiles in a model system. International Journal of drying on contamination by fungi and Ochratoxin A (OTA) under tropical Food Science &Technology, 47, 2117–2126. humid zone (Coteˆ d‘Ivoire). Food and Chemical Toxicology, 50, 1969–1979. https://doi.org/10.1111/j.1365-2621.2012.03078 https://doi.org/10.1016/j.fct.2012.03.042 International Coffee Organization (2017). “ICO Annual Review 2017.” Kurzrock, T., Klling-Speer, I., & Speer, K. (2005). Effects of controlled Retrieved from https://www.ico.org/ (accessed 24.11.2017). storage on the lipid fraction of green arabica coffee beans. In: 20e’me ´ International Coffee Organization (2014a). World coffee trade (1963–2013): A Colloque Scientifique International sur le Cafe , Bangalore, India, 11–15 ´ review of the markets, challenges and opportunities facing the sector. London: Octobre 2004 Paris: Association Scientifique Internationale du Cafe United Kingdom. (ASIC), 61–168. International Coffee Organization. (2014b). Breakdown of exports of green Kucera, L., Papousek, R., Kurka, O., Bartak, & Bednar, P. (2016). Study of Arabica and green Robusta for countries exporting significant volumes of composition of espresso coffee prepared from various roast degrees of Coffea both types of coffee. Retrieved from arabica L. coffee beans. Food Chemistry, 199, 727–735. , 2014b (accessed 24.11.2017). https://doi.org/10.1016/j.foodchem.2015.12.080 Illy, E., & Navarini, L. (2011). Neglected food bubbles: The espresso coffee Kuhn, M., Lang, S., Bezold, F., Minceva&, Briesen, H. (2017). foam. Food Biophysics, 6, 335–348. Time-resolved extraction of caffeine and trigonelline from finely-ground espresso coffee with varying particle sizes and tamping pressures. Journal of Illy, A. (2005). Quality. In A. Illy & R. Viani (Eds.), Espresso coffee: The science Food Engineering, 206, August 2017, 37–47. of quality, Chap 1 (2nd edn, pp. 1–20).London, UK: Elsevier Academic https://doi.org/10.1016/j.jfoodeng.2017.03.002 Press.

r 1232 ComprehensiveReviewsinFoodScienceandFoodSafety Vol.17,2018 C 2018 Institute of Food Technologists® Farm to consumer . . .

Kwasny, H., & Werkhoff, P. (1979). Relation of the MB aroma index of Massawe, G. A., & Lifa, S. J. (2010). Yeasts and lactic acid bacteria coffee roasted coffee to various parameters, Chemie, Mikrobiologie, Technologie der fermentation starter cultures. International Journal of Post Harvest Technology Lebensmittel, 6, 31–32. and Innovations, 2(1), 1. Laderach,¨ P., Lundy, M., Jarvis, A., Ramirez, J., Perez Portilla, E., Schepp, Masella, P., Guerrini, L., Spinelli, S., Calamai, L., Spugnoli, P., Illy, F., & K., & Eitzinger, A. (2010). Predicted Impact of Climate Change on Coffee Parenti, A. (2015). A newespresso brewing method. Journal of Food Supply Chains. In W. Leal Filho (Ed.), The economic, social and political Engineering, 146, 204–208. elements of climate change (pp. 703–723). Masoud, W., & Kaltoft, C. H. (2006). The effects of yeasts involved in the https://doi.org/10.1007/978-3-642-14776-0_42 fermentation of Coffea arabica in East Africa on growth and ochratoxin A Lee, L. W., Tay, G. W., Cheong, M. W., Curran, P., Yu, B., & Liu, S. Q. (OTA) production by Aspergillus ochraceus. International Journal of Food (2017a). Modulation of the volatile and non-volatile profiles of coffee Microbiology, 106, 229–234. fermented with Yarrowia lipolytica: I. Green coffee. LWT - Food Science and Masoud, W., Poll, L., & Jakobsen, M. (2005). Influence of volatile Te c h no lo gy, 77, 225–232. https://doi.org/10.1016/j.lwt.2016.11.047 compounds produced by yeasts predominant during processing of Coffea Lee, L. W., Tay, G. W., Cheong, M. W., Curran, P., Yu, B., & Liu, S. Q. arabica in East Africa on growth and ochratoxin A (OTA) production by (2017b). Modulation of the volatile and non-volatile profiles of coffee Aspergillus ochraceus. Yeast, 22, 1133–1142. fermented with Yarrowia lipolytica: II. Roasted coffee LWT - Food Science and Masoud, W., Cesar, L. B., Jespersen, L., & Jakobsen, M. (2004). Yeast Te c h no lo gy, 80, 32–s42. involved in fermentation of Coffea arabica in East Africa determined by Lee, L. W., Cheong, M. W., Curran, P., Yu, B., & Liu, S. Q. (2016a). genotyping and by direct denaturating gradient gel electrophoresis. Yeast, Modulation of coffee aroma via the fermentation of green coffee beans with 21, 549–556. Rhizopus oligosporus: I. Green coffee. Food Chemistry, 916–924. Marques, E. R., Borem, F. M., Pereira, R. G. F. A., & Biaggioni, M. A. M. Lee, L. W., Cheong, M. W., Curran, P., Yu, B., & Liu, S. Q. (2016b). (2008). Eficacia´ do teste de acidez graxa na avaliac¸ao˜ da qualidade do cafe´ Modulation of coffee aroma via the fermentation of green coffee beans with Arabica´ (Coffea arabica L.) submetido a diferente per´ıodos e temperaturas de Rhizopus oligosporus: II. Effects of different roast levels. Food Chemistry(50), secagem. Cienc.ˆ Agrotecnol, 32, 1557–1562. 925–936. Maatta, K. R., Eldin, K. A., & Torronen, A. R. J. (2003). Journal of Lee, L. W., Cheong, M. W., Curran, P., Yu, B., & Liu, S. Q. (2015). Coffee Agricultural and Food Chemistry, 51, 6736–6744. fermentation and flavor e an intricate and delicate relationship. Food Mayer, F., & Grosch, W. (2001). Aroma simulation on the basis of the Chemistry, 185, 182–191. odourant composition of roasted coffee headspace. Journal of Flavour and Leino, M., Kaitaranta, J., & Kallio, H. (1992). Comparison of changes in Fragrance, 16, 180–190. headspace volatiles of some coffee blends during storage, Food Chemistry, 43, Mayer, F., Czerny, M., & Grosch, W. (2000). Sensory study of the character 35–40. impact aroma compounds of a coffee beverage. European Food Research and Li, P. J., Li, C. C., & Li, C. H. (2010). U.S. Patent No.US 20100239711A1. Te c h no lo gy , 211, 272–276. Washington, DC: U.S. Patent and Trademark Office. Mayer, F., Czerny, M., & Grosch, W. (2003). Influence of provenance and Licciardi, R., Pereira, R. G. F. A., Mendonc¸a,L.M.V.L.,&Furtado,E.F. roast degree on the composition of potent odorants in Arabica coffees. (2005). Avaliac¸ao˜ f´ısico-qu´ımica de cafes´ torrados e mo´ıdos, de diferentes European Food Research and Technolology, 209(3-4), 242– marcas comerciais, da Regiao˜ Sul de Minas Gerais. Cienciaˆ e Tecnologia de 250. Alimentos, 25, 425–429. Mazzafera, P. (1999). Chemical composition of defective coffee beans. Food https://doi.org/10.1590/S0101-20612005000300006 Chemistry, 1999, 64(4), 547–554. Leroy, T., Ribeyre, F., Bertrand, B., Charmetant, P., Dufour, M., Mbugua, S. K., & Njenga, J. (1991). The antimicrobial activity of fermented Montagnon, C., . . . Pot, D. (2006). Genetics of coffee quality. Brazillian uji’. Ecology of Food and Nutrition, 28, 191–198. Journal of Plant Physiology, 18(1), 229–242. Melo, W. L. B. A. (2004). importanciaˆ da informac¸ao sobre do grau de torra Lin, C. C. (2010). Approach of improving coffee industry in Taiwan – do cafe´ e sua influencia nas caracter´ısticas organolepticas´ da bebida. Sao˜ Promote quality of coffee bean by fermentation. Journal of International Carlos: Empresa Brasileira de Pesquisa Agropecuaria.´ Comunicado Tecnico´ , 58. Management Studies, 5(1), 154–159. Mendes, L. C., De Menezes, H. C., Aparecida, M., & Da Silva, A. P. (2001). Loapez-Galilea, I., Paz de Pen, M., & Cid, C. (2007). Correlation of selected Optimization of the roasting of robusta coffee (C. canephora conillon) using constituents with the total antioxidant capacity of coffee beverages: acceptability tests and RSM. Food Quality & Preferences, 12, 153–162. Influence of the brewing procedure. Journal of Agricultural and Food https://doi.org/10.1016/S0950-3293(00)00042-2 Chemistry, 55, 6110–6117 Mendonca, J. C. F., Franca, A. S., & Oliveira, L. S. (2009). Physical Loapez-Galilea, I., Fournier, N., Cid, C., & Guichard, E. (2006). Changes in characterization of non-defective and defective Arabica and Robusta coffees headspace volatile concentrations of coffee brews caused by the roasting before and after roasting. Journal of Food Engineering, 92, 474–479. process and the brewing procedure. Journal of Agricultural and Food Chemistry, https://doi.org/10.1016/j.jfoodeng.2008.12.023 https://doi.org/10.1021/jf061178t.s Mendonca, J. C. F., Franca, A. S., & Oliveira, L. S. (2006). A comparative Ludwig, I. A., Mena, P., Calani, L., Cid, C., Rio, D. D., Lean, M. E. J., & evaluation of methodologies for water content determination in green Crozier, A. (2014). Variations in caffeine and chlorogenic acid contents of coffee. LWT- Food Science and Technology, 40(7), 1300–1303. coffees: What are we drinking? Food & Function, 5, 1718. https://doi.org/10.1016/j.lwt.2006.08.013 https://doi.org/10.1039/c4fo00290c Mestdagh, F., Davidek, T., Chaumonteuil, M., Folmer, B., & Blank, I. Lyman, D. J., Benck, R. M., & Merle, S. F. (2011). Difference spectroscopy (2014). The kinetics of coffee aroma extraction. Food Research International, in the analysis of the effects of coffee cherry processing variables on the 63, 271–274. https://doi.org/10.1016/j.foodres.2014.03.011 flavor of brewed coffee. International Journal of Spectrometery, ID 815304. https://doi.org/10.1155/2011/815304 Milo, C., Londez, A., & Fumeaux, R. (2002). Stability of coffee aroma compounds in model systems. Abstracts of Papers, 224th ACS National Maeztu, L., Sanz, C., Andueza, S., de Pena, M. P., Bello, J., & Cid, C. Meeting, Boston. Characterization of espresso coffee aroma by HS-GCMS and sensory flavor profile. Journal of Agricultural and Food Chemistry, 49, 5437–5444. Moroney, H., Moroney, K. M., Lee, W. T., O’Brien, S. B., Suijver, F., & Marra, J. (2016). Coffee extraction kinetics in a well mixed system. Journal of Maga, J. A. (1992). Pyrazine update. Food Reviews Intnational, 8, 479–558. Mathematics in Industry, 7(1), 1–19. Maier, H. G. (1981). Kaffee. Berlin Hamburg: Paul Parey Verlag. Moroney, H., Moroney, K. M., Lee, W. T., O’Brien, S. G. B., Suijver, F., & Main, D. The world’s most expensive coffee comes from elephant poo. Marra, J. (2015). Modelling of coffee extraction during brewing using Newsweek. Retrieved from , 2014, August 22 (accessed Engineering Science, 137(0), 216–234. on 12.03.2017). Mowla, S. B., Cuypers, A., Driscoll, S. P., Kiddle, G., Thomson, J., & Foyer, Marcone, M. F. (2004). Composition and properties of Indonesian palm civet C. H. (2006). Yeast complementation reveals a role for an Arabidopsis coffee (Kopi Luwak) and Ethiopian civet coffee. Food Research International, thaliana lateembryogenesis abundant (LEA)-like protein in oxidative stress 37, 901–912. tolerance. The Plant Journal, 48, 743–756. Marin, K., Pozrl, T., Zlatic, E., & Plestenjak, A. (2008). A new aroma index Moon, J. K., & Shibamoto, T. (2009). Role of roasting conditions in the to determine the aroma quality of roasted and ground coffee during storage. profile of volatile flavour chemicals formed from coffee beans. Journal of Food Technology and Biotechnology, 46(4), 442–447. Agricultural and Food Chemistry, 57, 5823–5831.

r C 2018 Institute of Food Technologists® Vol.17,2018 ComprehensiveReviewsinFoodScienceandFoodSafety 1233 Farm to consumer . . .

Merritt, M. C., & Proctor, B. E. (1959). Extraction rates for selected Petracco, M. (2001). Beverage preparation: Brewing trends for the new components in coffee brew. Food Research, 24, 735–743. millennium. In R. J. Clarke & O. G. Vitzthum (Eds.), Coffee: Recent advances Nakama, A., Kim, E. H., Shinohara, K., & Omura, H. (1993). Formation of (pp. 156). Oxford (UK): Blackwell Science Ltd. furfural derivatives in the amino-carbonyl reaction. Bioscience, Biotechnology Pereira, G. V. D. M., Neto, D. P. C., Medeiros, A. B. P., Soccol, V. T., Neto, and Biochemistry, 57, 1757–1759. E., Woiciechowski, A. L., & Socco, C. R. (2016). Potential of lactic acid Nagaraju, V. D., Ramalakshmi, K., & Sridhar, B. S. (2016). Cryo assisted bacteria to improve the fermentation and quality of coffee during on-farm spouted bed roasting of coffee beans. Innovative Food Science and Emerging processing. International Journal of Food Science and Technology, 51, 1689–1695. Te c h no lo gi e s, 37, 138–144. https://doi.org/10.1016/j.ifset.2016.08.016 https://doi.org/10.1111/ijfs.13142 Nagaraju, V. D., & Bhattacharya, S. (2010). Roasting green coffee beans using Pereira, G. V. M., Neto, E., Soccol, V. T., Medeiros, A. B. P., Woiciechowski, spouted bed roaster: Changes in physical characteristics. Journal of Food A. L., & Soccol, C. R. (2015a). Conducting starter culture-controlled Science & Technology, 47(6), 674–677. fermentations of coffee beansduring on-farm wet processing: Growth, https://doi.org/10.1007/s13197-010-0088-z metabolic analyses andsensorial effects. Food Research and International, 75, 348–356. https://doi.org/10.1016/j.foodres.2015.06.027 Nagaraju, V. D., Murthy, C.T., Ramalakshmi, K., & Srinivasa, P. N. R. (1997). Studies on Roasting of Coffee Beans in a Spouted Bed. Journal of Pereira, G. V. M., Soccol, V. T., Brar, S. K., Neto, E., & Soccol, C. R. Food Engineering, 31, 263–270. (2015b). Microbial ecology and starter culture technology in coffee processing. Critical Reviews in Food Science and Nutrition, Navarini, L., & Rivetti, D. (2010). Water quality for Espresso coffee. Food https://doi.org/10.1080/10408398.2015.1067759 Chemistry, 122, 424–428. https://doi.org/10.1016/j.foodchem.2009. 04.019 Pereira, G. V. M., Beux, M., Pagnoncelli, M. G. B., Soccol, V. T., Rodrigues, C., & Soccol, C. R. (2015c). Isolation, selection and evaluation of Nebesny, E., Budryn, G., Kula, J., & Majda, T. (2007). The effect of roasting antagonistic yeasts and lactic acid bacteria against ochratoxigenic fungus method on headspace composition of robusta coffee bean aroma. European Aspergillus westerdijkiae on coffee beans. Letters in Applied Microbiology, 61, Food Research & Technology, 225, 9–19. 96–101. https://doi.org/10.1007/s00217-006-0375-0006-0375-0 Pereira, G. V. M., Soccol, V. T., Pandey, A., Medeiros, A. B. P., Lara, J. M. Nebesny, E., & Budryn, G. (2006). Evaluation of sensory attributes of coffee R. A., Gollo, A. L., & Soccol, C. R. (2014). Isolation, selection and brews from robusta coffee roasted under different conditions. European Food evaluation of yeasts for use in fermentation of coffee beans by the wet Research and Technology, 224, 159–165. process. International Journal of Food Microbiology, 188, 60–66. https://doi.org/10.1007/s00217-006-0308-y https://doi.org/10.1016/j.ijfoodmicro.2014.07.008 Negishi, O., Negishi, Y., Aoyagi, Y., Sugahara, T. I., & Ozawa, T. (2001). Perez-Martınez, M., Caemmerer, B., Paz, de Pena, M., Cid, C., & Kroh, L. Mercaptan-capturing properties of mushrooms. Journal of Agricltural and Food W. (2010). Influence of brewing method and acidity regulators on the Chemistry, 49, 5509–5514. antioxidant capacity of coffee brews. Journal of Agricultural and Food Neto,D.P.C.,Pereira,G.V.M.,Tanobe,V.O.A.,Soccol,V.T.,Silva,B.J. Chemistry, 58, 2958–2965. G., Rodrigues, C. 1., & Soccol, C. R. (2017). Yeast diversity and Perez-Martınez, M., Sopelana, P., Paz de Pena, M., & Cid, C. (2008). physicochemical characteristics associated with coffee bean fermentation Changes in volatile compounds and overallaroma profile during storage of from the Brazilian Cerrado Mineiro region. Fermentation, 3, 11. coffee brews at 4 and 25 °c. Journal of Agricultural and Food Chemistry, 56, https://doi.org/10.3390/fermentation3010011 3145–3154. https://doi.org/10.1021/jf703731x Nicoli, M., C., Innocente, N., Pittia, P., & Lerici, C. R. (1993). Staling of Petisca, C., Perez-Palacios, T., Farah, A., Pinho, O., Ferreira, I., M & Lvo, P. roasted coffee: Volatile release and oxidation reactions during storage, (2013). Furans and other volatile compounds in ground roasted and espresso Proceedings of the 15th International Conferenceon Coffee Science, coffee using headspace solid-phase micro-extraction: Effect of roasting Montpellier, France. 557–565. speed. Food and Bioproduct Processing, 91(3), 233–241. Njoroge, J. M. (1998). Agronomic and processing factors affecting coffee Petracco, M. (2005). Percolation. Illy. In R. Viani (Ed.), Espresso coffee: The quality. Outlook on Agriculture, 27, 163–166. chemistry of quality (2nd ed., pp. 259–289). London: Elsevier Academic Press Nobre, G. W., Borem, F. M., Fernandes, S. M., & Pereira, R. G. F. A. (2007). London. Alterac¸oes˜ qu´ımicas do cafe-cereja´ descascado durante o armazenamento. Piazza, L., Gigli, J., & Bulbarello, A. (2008). Interfacial rheology study of Coffee Science, 2,1–9. espresso coffee foam structure and properties. Journal of Food Engineering, 84, Oestreich, J. S. (2010). Chemistry of coffee. In M. Lew & L. Hung-Wen 420–429. (Eds.), Comprehensive natural products (Vol. 2:3, pp. 1085–1117). Pittia, P., Dalla Rosa, M., & Lerici, C. R. (2001). Textural changes of coffee Oliveira,G.H.H.D.,Correa,P.C.,Oliveira,A.P.L.R.D.,Baptestini,F. beans as affected by roasting conditions. LWT- Food Science and Technology, M., & Vargas-elias, G. A. U. (2017). Roasting, grinding, and storage impact 34(3), 168–175. of thermodynamic properties and adsorption isotherms of arabica coffee. Pizarro, C., Esteban-Diez, I., Gonzalez-Saiz,´ J. M., & Forina, M. (2007). Use Journal of Food Process and Preservation, 41, 12779. of near-infrared spectroscopy and feature selection techniques for predicting https://doi.org/10.1111/jfpp.12779 the caffeine content and roasting color in roasted coffees. Journal of Oliveira, S. D., Franca, A. S., Gloria, M. B. A., & Borges, M. L. A. (2005). Agricultural and Food Chemistry, 55, 7477–7488. The effect of roasting on the presence of bioactive amines in coffees of https://doi.org/10.1021/jf071139x different qualities. Food Chemistry, 287–291. Poltronieri, P., & Rossi, F. (2016). Challenges in specialty coffee processing https://doi.org/10.1016/j.foodchem.2004.03.056 and quality assurance. Challenges, 7, 19. A, Viani R. (Eds.). Espresso coffee: Oliveira,L.S.,Franca,A.S.,Mendonc¸a,J.C.F.,&Barros-Junior,´ M. C. The science of quality. Academic Press. (2006). Proximate composition and fatty acids profile of green and roasted https://doi.org/10.3390/challe7020019Illy defective coffee beans. Lebensmittel-Wissenschaft & Technologie, 39, 235– Pramudita, D., Araki, T., Sagara, Y., & Tambunan, A. H. (2017). Roasting 239. and colouring curves for coffee beans with broad time-temperature Oosterveld, A., Harmsen, J. S., Voragen, A. G. J., & Schols, H. A. (2003a). variations. Food and Bioproduct Technology, Extration and characterization of polysaccharides from green and roasted https://doi.org/10.1007/s11947-017-1912-5. Coffea arabica beans. Carbohyderate Polymers, 52, 285–296. Radtke-Granzer, R., & Piringer, O. G. (1981). Problems in quality Oosterveld, A., Voragen, A. G. J., & Schols, H. A. (2003b). Effect of roasting evaluation of roasted coffee by quantitative trace analysis of volatile flavour on the carbohydrate composition of Coffea Arabica beans. Carbon Polymers, components, Deutsch Lebe Rund, 77, 203–210. 54, 192–193. Ramalakshmi, K., Kubra, I. R., & Rao, L. J. M. (2007). Physicochemical Ortala, M. D., Gutierrez, C. L., Chiralt, A., & Fito, P. (1998). Kinetic study characteristics of green coffee: Comparison of graded and defective beans. of lipid oxidation in roasted coffee. Food Science and Technology International, Journal of Food Science, 72, S333–S337. 4, 67–73. Reineccius, G. A. (1995). The maillard reaction and coffee flavor. In eme` Pangborn, R. M. (1982). Influence of water composition, extraction proceeding of the colloque scientifique international sur le cafe,´ 16, 1995, procedures, and holding time and temperature on quality of coffee beverage. Kyoto, Japan. Proceeding, Paris, France: ASIC. 249-257. Leben- Wissens and Techno, 15, 161–168. Rodrigues, M., Borges, M., Franca, A., Oliveira, L., & Correa, P. (2003). Parenti, A., Guerrini, L., Masella, P., Spinelli, S., Calamai, L., & Spugnoli, P. Evaluation of physical properties of coffee during roasting. Agricultural (2014). Comparison of espresso coffee brewing techniques. Journal of Food engineering international: The cigr journal of scientific research and Engineering, 121, 112–117.

r 1234 ComprehensiveReviewsinFoodScienceandFoodSafety Vol.17,2018 C 2018 Institute of Food Technologists® Farm to consumer . . .

development. Presented at the asae annual international meeting/CIGR Santos, J. R., Lopo, M., Rangel, A. O. S. S., & Lopes, J. A. (2016a). xvth world congress. Chicago, Ill. 2002. Manuscript FP 03 004. Vol. V. Exploiting near infrared spectroscopy as an analytical tool for on-line Romani, S., Cevoli, C., & Fabbri, A. (2012). Evaluation of coffee roasting monitoring of acidity during coffee roasting. Food Control, 60, 408–415. degree by using electronic nose and artificial neural network for offline https://doi.org/10.1016/j.foodcont.2015.08.007 quality control. Journal of Food Science, 77, 960–965. Santos, J. R., Viegas, O., & Pascoa, R. N. M.J. (2016b). In-line monitoring https://doi.org/10.1111/j.1750-3841.2012.02851.x of the coffee roasting process with near infrared spectroscopy: Measurement Rothfos, N. B. (1985). Coffee Production. Hamburg: Gordian–Max– Rieck. of sucrose and color. Food Chemistry, 208, 103–110. https://doi.org/10. 1016/j.foodchem.2016.03.114 Puerta, G. I. Q., & Rios, A. S. (2011). Composicion´ del muc´ılago de cafe,´ segun´ el tiempo de fermentacion´ y refrigeracion.´ Cenicafe´, 62, 23– Santos, J. R., Sarraguc¸a, M. C., Rangel, A. O. S. S., & Lopes, J. A. (2012). 40. Food Chemistry, 135, 1828–1835. Rendon, M. Y., Salva, T. J. G., & Bragagnolo, N. (2014). Impact of chemical Sanz, C., Maeztu, L., Zapelena, M. J., Bello, J., & Cid, C. (2002). Profiles of changes on the sensory characteristics of coffee beans during storage. Food volatile compounds and sensory analysis of three blends of coffee: Influence Chemistry, 147, 279–286. https://doi.org/10.1016/j.foodchem.2013.09.123 of different proportions of Arabica and Robusta and influence of roasting coffee with sugar. Journal of Science of Food and Agriculture, 82, 840–847. Rendon, M. Y., Gratao, P. L., Salva, T. J. G., Azevedo, R. A., & Bragagnolo, https://doi.org/10.1002/jfsa.1110 N. (2013). Antioxidant enzyme activity and hydrogen peroxide content during the drying of Arabica coffee beans. European Food Research and Scheidig, C., Czerny, M., & Schieberle, P. (2007). Changes in key odorants Te c h no lo gy, 236, 753–758. of raw coffee beans during storage under defined conditions. Journal of Agricultural and Food Chemistry, 55–5768–5775. https://doi.org/10.1021/ Reynolds, E. S. (1963). The use of lead citrate at high pH as an jf070488o electron-opaque stain in electron microscopy. Journal of Cell Biology, 17, 208–212. Scheidig, C., & Schieberle, P. (2006). Einfluss der Lagerung von Rohkaffee auf das Aroma von Rohkaffee. Ro¨stkaffee und Kaffeegetra¨nk. Rojast, J. (2004). Green coffee storage. In J. N. Wintgens (Ed.), Coffee: Lebensmittelchemie, 60, 55–56. Growing, processing, sustainable production (pp. 733–750). Weinheim: Wiley-VCH Verlag GmbH & Co. KGaA. Schenker, S., Heinemann, C., Huber, M., Pompizzi, R., Perren, R., & Escher, F. Impact of roasting conditions on the formation of aroma Ross, C. F., Pecka, K., & Weller, K. (2006). Effect of storage conditions on compounds in coffee beans. Journal of Food Science, 67(1), 60– the sensory quality of ground arabica coffee. Journal of Food Quality, 29, 66. 596–606. Schenker, S., Handschin, S., Frey, B., Perren, R., & Esche, F. (2000). Pore Ribeiro,L.S.,Miguel,M.G.D.C.P.,Evangelista,S.R.,Martins,P.M.M., structure of coffee beans affected by roasting conditions. Journal of Food van Mullem, J., Belizario, M. H., & Schwan, R. F. (2017). Behavior of yeast Science, 65(3), 452–457. inoculated during semi-dry coffee fermentation and the effect on chemical and sensorial properties of the final beverage. Food Research International, 92, Schulman, J. (2002). Insanely long water FAQ. Retrieved from 26–32. https://doi.org/10.1016/j.foodres.2016.12.011 https://www.big-rick.com/coffee/waterfaq.html Ribeiro, F. C., Borem, F. M., Giomo, G. S., De Lima, R. R., Malta, M. R., Schmidt, C. A. P., Miglioranza, E., & Prudencio,ˆ S. H. (200). Interac¸ao˜ da & Figueiredo, L. P. (2011). Storage of green coffee in hermetic packaging torra e moagem do cafe na preferenciaˆ do consumidor do oeste paranaense. injected with CO2. Journal of Stored Products Research, 47, 341–348. Ciencia Rural, 38, 1111–1117. https://doi.org/10.1590/S0103- https://doi.org/10.1016/j.jspr.2011.05.007 84782008000400032 Ribeiro, J. S., Augusto, F., Salva, T. J. G., Thomaziello, R. A., & Ferreira, M. Selmar, D., Kleinwachter, M., & Bytof, G. (2015). Metabolic responses of M. C. (2009). Prediction of sensory properties of Brazilian Arabica roasted coffee beans during processing and their impact on coffee flavor.431-75, In coffees by headspace solid phase microextraction–gas chromatography and Cocoa and Coffee Fermentations, https://doi.org/10.1016/B978-0-12- partial least squares. Analytica Chimica Acta, 634(2), 172–179. 409517-5.00009-7 https://doi.org/10.1016/j.aca.2008.12.028 Selmar, D., Bytof, G., & Knopp, S. E. (2008). The storage of green coffee Richard, M., Charles, A., & Mitiku, M. (2007). Primary coffee processing in (Coffea arabica): Decrease of viability and changes of potential aroma Ethiopia: Patterns, constraints and determinates. African Crop Science precursors. Annals of Botany, 101(1), 31–38. Confernce Proceedings, 8, 1417–1421. Selmar, D., Bytof, G., Knopp, S. E., & Breitenstein, B. (2006). Germination Rivetti, D., Navarini, L., Cappuccio, R., Abatangelo, A., Petracco, M., & of coffee seeds and its significance for coffee quality. Plant Biology, 8(17), Suggi-Liverani, F. (2001). Effect of water composition and water treatment 260–4. on espresso coffee percolation, ASIC. In Proceedings 19th colloquium.Trieste, Selmar, D., Bytof, G., Knopp, S. E., Bradbury, A., Wilkens, J., & Becker, R. Italy, CD-ROM. (2005a). Biochemical insights into coffee processing: Quality and nature of Roberts, J. J., & Cozzolino, D. (2016). An overview on the application of green coffees are interconnected with an active seed metabolism. In chemometrics in food science and technology-an approach to quantitative Proceedings of the 20’eme Colloque Scientifique International sur le ´ data analysis. Food Analytical Methods, 9, 3258–3267. Caf e”. ASIC, Paris. https://doi.org/10.1007/s12161-016-0574-7 Selmar, D., Hunecke, D., Jungh, Artchen, I., Breitenstein, B., Bytof, G., & Rojas, J. (2004). Green coffee storage. In J N. Wintgens (Ed.), Coffee: Knopp, S. E. (2005b). Expression of isocitrate lyase and the abundance of Growing, processing, sustainable production (pp. 733–750). Weinheim: ß-tubulin: Feasible markers for the estimation of germination processes in Wiley-VCH Verlag GmbH & Co. KGaA. differently processed green coffees. In: Proceedings of the 20’eme Colloque Scientifique International surle Caf´e”.ASIC,Paris. Ruosi, M. R., Cordero, C., Cagliero, C., Rubiolo, P., Bicchi, C., Sgorbini, B., & Liberto, E. (2012). A further tool to monitor the coffee roasting Selmar, D., Bytof, G., & Knopp, S. E. (2002). New aspects of coffee process: Aroma compos chem indi. Journal of Agricultural and Food Chemistry, processing: The relation between seed germination and coffee quality. In: 60, 11283−91. https://doi.org/10.1021/jf3031716 Proceedings of the “19eme Colloque Scientifique International sur le Caf´e”. ASIC Paris. Saath,R.,Borem,F.M.,Alves,E.,Taveira,J.H.S.,Medice,R.,&Coradi,P. C. (2010). Microscopia eletronicaˆ de varredura do endosperma de cafe´ Semmelroch, P., & Grosch, W. (1995). Analysis of roasted coffee powders and (Coffea arabica L.) durante o processo de secagem. Cienceˆ Agrotecnology, 34, brews by gas chromatography–olfactometry of headspace samples. LWT- 196–203. Food Science and Technology, 28(3), 310–313. https://doi.org/10.1016/ s0023-6438(95)94411-7 Salamanca, C. A., Fiol, N., Gonzalez,´ C., Saez, M., & Villaescusa, I. (2017). Extraction of espresso coffee by using gradient of temperature. Effect on Severini, C., Derossi, A., Fiore, A. G., Ricci, I., & Marone, M. (2016). The physicochemical and sensorial characteristics of espresso. Food Chemistry, electronic nose system: Study on the global aromatic profile of espresso 214, 622–630. https://doi.org/10.1016/j.foodchem.2016.07.120 coffee prepared with two types of coffee filter holders. European Food Research and Technology, https://doi.org/10.1007/s00217-016-2705-1 Sanchez, G. I., Jimenez, E. A., & Saura-Calixto, F. (2005). In vitro antioxidant activity of coffee brewed using different procedures (Italian, Severini, C., Ricci, I., Marone, M., Derossi, A., & De Pilli, T. (2015). espresso and filter). Food Chemistry, 90, 133–139. Changes in the aromatic profile of espresso coffee as a function of the grinding grade and extraction time: A study by the electronic nose system. Sanchez, B. B., Renouf, M., Sauser, J., Beaumont, M., & Actis, G. L. (2016). Journal of Agricultural and Food Chemistry, 63, 2321−7. The roasting process does not influence the extent of conjugation of coffee https://doi.org/10.1021/jf505691u chlorogenic and phenolic acids. BioFactors, 42(3), 259–267. https://doi.org/10.1002/biof.1268 Shibamoto, T. (1983). Heterocyclic compounds in browning and browning/nitrite model systems: Occurrence, formation mechanisms, flavor

r C 2018 Institute of Food Technologists® Vol.17,2018 ComprehensiveReviewsinFoodScienceandFoodSafety 1235 Farm to consumer . . .

characteristics and mutagenic activity. In G. Charalambous & G. Inglett submetidas adiferentes metodos´ de processamento e secagem. Pesquisa (Eds.), Instrumental analysis of food (Vol. 1, pp. 229–278). New York: Agropecuaria´ Brasileira, 47(10), 1511–1517. Academic Press. Taveira, J. H. D., Borem, F. M., Da Rosa, S. D. V. F., Oliveira, P. D., Giomo, Silva, S. A. (2014a). Characterization and delimitation of the terroir coffee in G. S., Isquierdo, E. P., & Fortunato, V. A. (2015). Post-harvest effects on plantations in the municipal district of Araponga, Minas Gerais, Brazil. beverage quality and physiological performance of coffee beans. African Revista Cien Agronom, 45, 18–26. Journal of Agricultural Research, 10(12), 1457–1466. https://doi.org/10.5897/ Silva, C. F. (2014b). Microbial Activity during Coffee Fermentation. In Cocoa AJAR2014.9263 and coffee fermentations (Vol. 430, pp. 397–382). Boca Raton, Fla.: CRC Teixeira, A. A., Brando, C. H. J., Thomaziello, R. A., & Teixiera, R. (1995). Press. In A. Illy & R. Viani (Eds.), Espresso coffee: The science of quality (pp. 91–95). Silva, C. F. (2014c). Microbial activity during coffee fermentation. In R. F., Italy: Elsevier Academic Press. 197, 198. Schwan & G. H., Fleet (Eds.), Cocoa and coffee fermentation (pp. 398–423). Tfouni, S. A. V., Serrate, C. S., Carreiro, L. B., Camargo, M. C. R., Teles, C. Boca Raton, Fla.: CRC Taylor & Francis. R. A., Cipolli, K. M. V. A. B., & Furlani, R. P. Z. B. (2012). Effect of Silva,C.F.,Vilela,D.M.,deSouza,C.C.,Duarte,W.F.,Dias,D.R.,& roasting on chlorogenic acids, caffeine and polycyclic aromatic hydrocarbons ´ Schwan, R. F. (2013). Evaluation of a potential starter culture for enhance levels in two Coffea cultivars: Coffea arabica cv. Catuaı Amarelo IAC-62 and ˜ quality of coffee fermentation. World Journal of Microbiology and Biotechnology, Coffea canephora cv. Apoata IAC-2258. International Journal of Food Science 29, 235–247. & Technology, 47, 406–415. https://doi.org/10.1111/j.1365-2621.2011. 02854.x Silva,C.F.,Vilela,D.M.,deSouza,C.C.,Duarte,W.F.,Dias,D.R.,& Schwan, R. F. (2012). Evaluation of a potential starter culture for enhance Tfouni, S. A. V., Serrate, C. S., Leme, F. M., Camargo, M. C. R., Teles, C. quality of coffee fermentation. World Journal of Microbiology and Biotechnology, R. A., Cipolli, K. M. V. A. B., & Furlani, R. P. Z. (2013). Polycyclic 29, 235–247. aromatic hydrocarbons in coffee brew: Influence of roasting and brewing procedures in two Coffea cultivars. LWT - Food Science and Technology, 50, Silva, C. F., Batista, L. R., Abreu, L. M., Dias, E. S., & Schwan, R. F. (2008). 526–530. https://doi.org/10.1016/j.lwt.2012.08.015 Succession of bacterial and fungal communities during natural coffee (Coffea arabica) fermentation. Food Microbiology, 25, 951–957. Toci, A. T., & Farah, A. (2014). Volatile fingerprint of Brazilian defective coffee seeds: Corroboration of potential marker compounds and Silva, M. D. C., Varzea, V., Guerra, G. L., Azinheira, H. G., Fernandez, D., identification of new low quality indicators. Food Chemistry, 153, 298–314. Petitot, A. S., . . . Nicole, M. (2006). Coffee resistance to the main diseases: https://doi.org/10.1016/j.foodchem.2013.12.040 Leaf rust and coffee berry disease. Brazillian Journal of Plant Physiology, 18, 119–147. Toci, A. T., & Farah, A. (2008). Volatile compounds as potential defective coffee seeds’ markers. Food Chemistry, 108, 1133–1141. Silva, C. F., Schwan, R. F., Dias, E. Q., & Wheals, A. E. (2000). Microbial diversity during maturation and natural processing of coffee cherries of Toledo,P.K.A.B.D.,deMelo,M.M.R.,Pezza,H.R.,Pezza,L.,Toci,A. Coffea arabica in Brazil. International Journal of Food Microbiology, 60, 251–260. T., & Silva, C. M. (2017). Reliable discriminant analysis tool for controlling the roast degree of coffee samples through chemical markers approach. Silwar, R., & Lullmann, C. (1993). Investigation of aroma formation in European Food Research Technology, https://doi.org/10.1007/s00217-016- Robusta coffee during roasting. Cafe Cacao, 37, 145–151. 2790-1 Sivetz, M., & Desrosier, N. W. (1979). In M. Sivetz & N. W. Desrosier (Eds.), Toledo, P. R. A. B., Pezza, L., Pezza, H. R., & Toci, A. T. (2016). Coffee technology (pp. 74–116). Westport, CN: Avi Publishing. Relationship between the different aspects related to coffee quality and their Sivetz, M. (1972). How acidity affects coffee flavour. Food Technology, 26, volatile compounds. Comprehensive Reviews in Food Science and Food Safety, 70–77. https://doi.org/10.1111/1541-4337.12205 Speer, K., & Kolling, S. I. (2006). The lipid fraction of coffee bean. Brazillian Tressl, R., Nittka, C., & Kersten, E. (1995). Formation of isoleucine-specific Journal of Plant Physiology, 18, 201–216, Maillard products from [1-13C]-D-glucose and [1-13C]-D-fructose. Journal Speer, K., & Kolling-Speer,¨ I. (2001). Non-volatile compounds—Lipids. In of Agricultural and Food Chemistry, 43, 1163–1169. R. J. Clarke & O. G. Vitzthum (Eds.), Coffee: Recent developments (pp. Trubey, R., Raudales, R., & Morales, A. (2005). Cafe´ britt hermetic cocoon 33–49). Oxford: Blackwell Science. storage trial – II. Report (pp. 1–14). Costa Rica: Mesoamerican Devel Insti Steiman, S. (2013). What is specialty coffee? In R. W. Thurston, J. Morris, & Corp. S. Steiman (Eds.), Coffee: A comprehensive guide to the bean, the beverage, and the Tsegaye, B., Mohammed, A., Shimber, T., Getachew, Y., & Getachew, E. industry (pp. 102–105). Lanham, MD: Rowman & Littlefield. (2014). The Influence of sun drying methods and layer thickness on the Stirling, H. (1981). Storage research on Kenya Arabica coffee. In 9th quality of lowland Arabica coffee varieties at Gomma I, Southewesten International Conference on Coffee, Londres, 1980; ASIC: Paris, France Ethiopia Research. Journal of Agricultural and Environmental Management, 189–200. 3(11), 547–554. Stumptown. (2011). Semi-washed coffee processing. Stumptown coffee Tunnacliffe, A., & Wise, M. J. (2007). The continuing conundrum of the roaster blog, Retrieved from https://stumptowncoffee.com/processing/ LEA proteins. Naturwissen, 94, 791–812. semi-washed/ Uman, E., Colonna, D. M., Colonna, D. L., Perger, M., Klatt, C., Leighton, Suarez-Quiroz, M. L., Gonzalez, R. O., Barel, M., Guyot, B., Schorr, G. S., S., . . . Hendon, C. H. (2016). The effect of bean origin and temperature & Guiraud, J. P. (2004). Study of 2ochratoxin A-producing strains in coffee on grinding roasted coffee. Scientific Reports, 6, 24483. https://doi. processing. International Journal of Food Science & Technology, 39, 501–507. org/10.1038/srep24483 Summa, C. A., dela Calle, B., Brohee, M., Stadler, R. H., & Anklam, E. Vanni, A. (2009). Sealed capsule for the preparation of a beverage, in (2007). Impact of the roasting degree of coffee on the in vitro radical particular espresso coffee European Patent 1 886 942 b1(12). scavenging capacity and content of acrylamide. LWT- Food Science & Velmourougane, K. (2013). Impact of natural fermentation on Te c h no lo gy, 40, 1849–1854. physicochemical, microbiological and cup quality characteristics of arabica Summers, J. E., Ratcliffe, R. G., & Jackson, M. B. (2000). Journal of and robusta coffee. Proceedings of the Nationall Academy of Sciences India Section Expermental Botany, 51, 1413–1422. B Biological Sciences, 83(2), 233–239. https://doi.org/10.1007/s40011-012- 0130-1 Sunarharum, W. B., Williams, D. J., & Smyth, H. E. (2014). Complexity of coffee flavor: A compositional and sensory perspective. Food Research Velmourougane, K. (2011). Effects of wet processing methods and subsequent International, 62, 315–325. https://doi.org/10.1016/j.foodres.2014.02.030 soaking of coffee under different organic acids on cup quality. World Journal of Sciences and Technology, 1(7), 32–38. Tadesse, F. T., Jemal, Y., & Abebe, H. (2015). Effect of green coffee processing methods and roastingtemperatures on physical and cup quality of Velmourougane, K., Shanmukhappa, D. R., Ventakesh, K., Prakasan, C. B., sidama coffee, southern ethiopia. Journal of Nutritional Ecology and Food & Jayarama. (2008). Use of starter culture in coffee fermentation - effect on Research, 3,1–7.https://doi.org/10.1166/jnef.2015.1119 demucilisation and cup quality. India Coffee, 72, 31–34. Tarzia, A., Scholz, M. B. S., & Petkowicz, C. L. D.O. (2010). Influence of the Vieira, R. D., Tekrony, D. M., Egli, D. B., & Rucker, M. (2001). Electrical postharvest processing method on polysaccharides and coffee beverages. conductivity of soybean seeds after storage in several environments. Seed International Food Science and Technology, 45, 2167–2175. https://doi.org/ Science & Technology, 29, 599–608. 10.1111/j.1365-2621.2010.02388.x Villarreal, D., Laffargue, A., Posada, H., Bertrand, B., Lashermes, P., & Taveira, J. H. S., Rosa, S. D. V. F., Borem,´ F. M., Giomo, G. S., & Saath, R. Dussert, S. (2009). Genotypic and environmental effects on coffee (Coffea (2012). Perfis proteicos e desempenho fisiologico´ de sementes de cafe´ arabica L.) bean fatty acid profile: Impact on variety and origin chemometric determination. Journal of Agricultural and Food Chemistry, 57, 11321–11327.

r 1236 ComprehensiveReviewsinFoodScienceandFoodSafety Vol.17,2018 C 2018 Institute of Food Technologists® Farm to consumer . . .

Vilela,D.M.,Pereira,G.V.,Silva,C.F.,Batista,L.R.,&Schwan,R.F. Wintgens, J. N. (Ed.) (2008). Coffee: Growing, processing, sustainable production. (2010). Molecular ecology and polyphasic character of the microbiota Weinheim: Wiley-VCH. associated with semi-dry processed coffee (Coffea arabica L.). Food Wootton, A. E. (1974). The dry matter loss from parchment coffee during Microbiology, 27, 1128–1135. field processing. In: Proceedings of the “16eme Colloque International sur la Villers, P., Navarro, S., & De Bruin, T. (2010). New applications of hermetic Chimie des Caf´es Verts” ASIC Paris. 316–324. storage for grain storage and transport, in: 446–452, navarro, s., riudavets, j. Wood, B. J. B., & Hodge, M. M. (1985). Yeast-Lactic Acid bacteria [eds.] fumigation, modified atmospheres and hermetic storage,proceedings interactions and their contribution to fermented foodstuffs’. In B. J. B., of the 10th international working conference on stored product protection, Wood (Ed.), Microbiology of fermented foods (pp. 263–293). London: Elsevier 27 june to 2 july 2010, estoril, portugal julius-kuhn-archiv,¨ 425, Applied Science publishers. bundesforschungsinstitut fur¨ kulturpflanzen, berlin 1077. Zanin, R. C., Corso, M. P., Kitzberger, C. S. G., Scholz, M. B. S., & Vitzthum, O. G., Werkhoff, P., & Ablanque, E. (1976). Volatile constituents Benassi, M. T. (2016). Good cup quality roasted coffees show wide variation of raw coffee beans (in German). In 7th International Conferenceon Coffee, in chlorogenic acids content. LWT Food Science and Technology, 74, 480– Hamburg, ASIC: Paris, France. 115–123. 483. https://doi.org/10.1016/j.lwt.2016.08.012 Wang, X., & Lim, L. T. (2014). A kinetics and modeling study of coffee Zanoni, B., Pagliarini, E., & Peri, C. (1992). Modelling the aqueous roasting under isothermal conditions. Food Bioprocess Technology, 7, 621–632. extraction of soluble substances from ground roasted coffee. Journal of the https://doi.org/10.1007/s11947-013-1159-8 Science of Food and Agriculture, 58, 275–279. Wen, X., Enokizo, A., Hattori, H., Kobayashi, S., Murata, M., & Homma, S. (2005). Effect of roasting on properties of the zinc-chelating substance in coffee brews. Journal of Agricultural Food Chemistry, 53, 2684–2689.

r C 2018 Institute of Food Technologists® Vol.17,2018 ComprehensiveReviewsinFoodScienceandFoodSafety 1237