Tomato Waste from Processing Industries as a Feedstock for Biofuel Production

Tomato pomace, a solid by-product of tomato processing industries, was assessed for the first time as a feedstock for acetone-butanol-ethanol-isopropanol (ABEI) fermentation. After the pretreatment with hydrothermal process and enzymatic hydrolysis of tomato pomace with a 20% solid-to-solvent ratio, a broth containing 44.1 g/L total sugars was obtained (61.0% saccharification yield). Twelve Clostridium strains were compared to find an appropriate microorganism for the transformation of tomato pomace hydrolysate. Four of these strains produced 0.22–5.95 g/L acetone, 5.82–7.00 g/L butanol, 0.17–0.39 g/L ethanol, 0.05–8.28 g/L isopropanol and 11.03–15.07 g/L ABEI. The strain C. beijerinckii DSM 6423 obtained 2.67 g butanol, 2.99 g isopropanol and 5.92 g ABEI from 100 g dry tomato pomace (containing 30.6 g carbohydrates). In addition, another pretreatment was performed with a 30% solid-to-solvent ratio in order to obtain higher sugar concentrations, obtaining a hydrolysate with 85.1 g/L total sugars (66.8% saccharification efficiency), which was employed for bioethanol production by comparing twelve different yeast and bacterial strains. Kluyveromyces marxianus DSM 5422, Saccharomyces cerevisiae Ethanol Red®, S. cerevisiae Hércules and Lachancea thermotolerans DSM 3434 produced 20.1–21.7 g/L ethanol (about 5.21 g ethanol per 100 g dry tomato pomace). According to these results, tomato pomace could be an interesting feedstock for ABEI biorefineries.


Introduction
Nowadays, in an environmentally challenging context, it is necessary to tackle certain problems, such as greenhouse emissions, management of landfill wastes and the assurance of the food supply to a growing human population. Within this situation, the improvement in waste treatment during the foodchain production is mandatory in order to reduce its environmental repercussion [1]. Over recent years, the idea of a biorefinery concept has emerged as an approach to deal with waste management problems by converting agricultural and food processing wastes into high-valued compounds such as alcohols (butanol, ethanol and isopropanol) that can be utilised by bioenergetic or chemical industries [2]. In this sense, lignocellulosic agri-food wastes could become an interesting feedstock for both butanol and ethanol fermentation processes.
Between 1910 and 1960, butanol and acetone were produced worldwide at industrial scale by the acetone-butanolethanol (ABE) fermentation [3,4]. In this process, sugar-or starch-rich feedstocks, such as maize, molasses, grain or potatoes, were transformed by solventogenic Clostridium strains through a two-phase fermentation route whose main products are acetone (or alternatively isopropanol), n-butanol, ethanol, CO 2 , H 2 , acetate and butyrate [3,5,6]. The elevated price of those feedstocks and the rapid development of petrochemical industry caused the decline of ABE fermentation plants [3,6,7]. Currently, butanol is mainly produced via synthesis from petroleum-derived propylene [3].
Regarding alcoholic fermentation, 96% of the ethanol produced in the European Union in 2017 was obtained from corn, cereals and sugars, whereas only 4% came from lignocellulosic biomass [8]. Most industrial fermentations use the yeast Electronic supplementary material The online version of this article (https://doi.org/10.1007/s12155-019-10016-7) contains supplementary material, which is available to authorized users.
Saccharomyces cerevisiae due to its high ethanol yield, osmotolerance and resistance to inhibitors. However, S. cerevisiae wild strains can only metabolise hexose sugars (such as those present in traditional sucrose-rich and starchrich feedstocks) [9], but they are unable to ferment pentoses, which are common in lignocellulosic biomass. Therefore, it is necessary to unearth new microorganisms that are capable of transforming lignocellulosic-derived sugars into ethanol and thus avoid the use of expensive feedstocks that could be devoted to human or animal nutrition.
Upon completion of the alcoholic or ABE fermentation in conventional industrial plants, the solvents were generally recovered and purified by distillation [4]. However, distillation implies high economic and energetic costs, which makes this process financially unfeasible in biorefineries, where tight profit margins and hard competition with petrochemical industries are expected. Recently, other recovery methods for solvents, such as gas stripping, two-stage gas stripping, pervaporation, liquid-liquid extraction, adsorption, non-ionic surfactants, pervaporation, perstraction and reverse osmosis, have been reviewed and proposed as alternatives to distillation [6,[10][11][12][13][14].
Lignocellulosic agri-food by-products are composed mainly of cellulose, hemicellulose and lignin. Conventional fermentation microorganisms are not able to directly metabolise these substrates and a pretreatment is necessary to degrade the complex lignocellulosic structure and release the sugars contained in cellulose and hemicellulose [10]. These pretreatments involve physical, chemical, physicochemical and/or biological procedures, among which the most common are milling, acid or alkaline pretreatment at high temperatures, ammonia fibre explosion (AFEX), organosolv pretreatment, hydrothermal processes, steam explosion and enzymatic hydrolysis [5,9,12,[15][16][17]. However, lignin-and sugar-degradation products, such as acetic acid, formic acid, levulinic acid, furfural or 5-hydroxymethylfurfural (5-HMF), can also be released during the pretreatment and they are known to inhibit fermentative microorganisms [14,18]. Therefore, in some cases, depending on the pretreatment type, on the nature of the substrate and on the sensitivity of the microbial strain, detoxification methods (e.g. extraction, evaporation, adsorption, electrodialysis, neutralisation, overliming, steam stripping or enzymatic treatment) are recommended to remove inhibitors before fermentation [19], which raises the global price of the production process.
Tomato pomace has been used mostly as animal feed [26] or it has been left on the soil causing environmental problems [29]. However, recently tomato pomace has been proposed as a source of bioproducts, like polysaccharides [30], lycopene, β-carotene, phenolic compounds, flavonoids, levulinic acid, phytosterols, vegetal oil, cutin, polyhydroxyalkanoates and vitamin B12 [27,29,31,32], and as plant fertiliser or microbial growth medium [31]. The generation of biofuels like ethanol and butanol from a substantial number of lignocellulosic and food processing wastes has been extensively assessed [15,[33][34][35]. However, the application of tomato waste in biofuel production by fermentation is not well studied [26], in spite of the recent methods described to release simple sugars from tomato by-products [36]. So far, bioethanol production from tomato wastes has yielded poor results [24,[37][38][39]. On the other hand, to the best of our knowledge, no information about biobutanol production from tomato pomace has been published yet.
In this work, tomato pomace was subjected to a soft pretreatment including hydrothermal process and enzymatic hydrolysis to release its fermentable sugars and obtain a liquid hydrolysate. The suitability of this hydrolysate as a feedstock for butanol production by acetone-butanol-ethanol-isopropanol (ABEI) fermentation and for ethanol production by alcoholic fermentation was assessed with twenty-four different bacterial and yeast strains (Fig. 1).

Tomato Waste Description
Wet tomato wastes from a processing industry were provided by Conservas Vegetales de Extremadura S.A. (Badajoz, Spain) in August 2018. The raw biomass was dried in the open air during 6 days and then dried further in an oven at 45°C during 48 h. The dried biomass was ground in a rotary mill SM100 Comfort (Retsch GmbH, Haan, Germany) and kept in an airtight container until use.

Hydrolysis of Biomass
In order to obtain sugar-rich fermentable hydrolysates, tomato waste was pretreated by hydrothermal process (i.e. using water as the only solvent and reagent). For the experiments of ABEI fermentation, the biomass was pretreated in a highpressure 2-L reactor made of alloy Carpenter 20 (Parr Instrument Company, Moline, IL, USA) at 121°C during 20 min, with a solid-to-solvent ratio of 20% (w/w). The total mass introduced in the reactor vessel for each batch was 400 g. The reactor temperature was increased from 22 to 121°C in 21.5 min; it was kept at 121°C during 20 min, and then the reactor was cooled down again in 5 min with a water refrigeration system. For the experiments of ethanol fermentation, the biomass was pretreated in an autoclave at 121°C during 20 min, with a solid-to-solvent ratio of 30% (w/w). It was necessary to employ a greater biomass load in the case of ethanol experiments (30%) in order to guarantee a hydrolysate with a higher initial concentration of simple sugars, which is essential for a profitable ethanol fermentation process.
The samples were cooled down and then the solid/liquid mixture was subjected to enzymatic hydrolysis at pH 5.0 (citrate buffer 50 mM), 50°C and 180 rpm in an orbital shaker (HT Minitron, Infors AG, Bottmingen, Switzerland). The cellulase cocktail Cellic CTec 2 (activity 100 FPU/mL) was provided by Novozymes (Tianjin, China) and the endo-β-1,4-glucanase enzyme Viscozyme L (activity 41 CMC/mL) was obtained from Novozymes (Bagsvaerd, Denmark). Both of them were applied to the sample at a dose of 3.6 FPU/g biomass and 0.41 CMC/g biomass, respectively. These doses were chosen based on previous works in the case of Cellic CTec2 [41] and on experimental data in the case of Viscozyme L. The hydrolysis time was set at 48 h or 120 h, depending on the biomass load (20% or 30%, respectively). Saccharification efficiency was calculated as the ratio between the mass of dissolved sugars present in the hydrolysate and the initial mass of carbohydrates from tomato pomace present in the sample just before the hydrothermal pretreatment, expressed as a percentage (Eq. 1): where m RS is the amount of released sugars after the enzymatic hydrolysis (expressed in g), calculated taking into account sugar concentrations and the recovered total liquid volume; and m C is the amount of total carbohydrates present in the dry tomato pomace sample before any pretreatment (expressed in g). The hydrolysates of the different batches were mixed and homogenised to guarantee the same initial composition of all the samples. These hydrolysates were directly employed (without filtration) in fermentation experiments.

Biobutanol Production
Twelve bacterial strains were compared for their performance in tomato waste fermentation. The strains Clostridium C. pasteurianum DSM 526; C. saccharobutylicum DSM 13864; and C. saccharoperbutylacetonicum DSM 2152 and DSM 14923 were purchased from DSMZ (Braunschweig, Germany), whereas the strain C. beijerinckii CECT 508 was supplied by CECT (Paterna, Spain). Seed cultures for fermentation were prepared from stocks of pure spore solutions of each strain maintained at 4°C in sterile distilled water. These spores were heat activated in reinforced clostridial medium (RCM) broth or in the case of strains DSM 2152 and DSM 792 in a potato-based medium [42] as described previously [43]. Inocula from the non-spore forming strain DSM 6228 were performed by using a single bacterial colony from a 48h RCM plate culture (2% agar) obtained from a bacterial stock maintained at − 80°C in glycerol (80% v/v). Liquid inocula were maintained at 35°C until reaching a cell density of 5·10 8 cells/mL (24-48 h) as determined by counting in a Bürker chamber (Paul Marienfeld GmbH & Co. KG, Lauda-Königshofen, Germany).
For fermentation experiments, 50 mL of tomato hydrolysates were placed in 100-mL glass bottles, where 5 g/L yeast extract, 2.1 g/L NH 4 Cl, 1 g/L KH 2 PO 4 and 5 g/L CaCO 3 were added. The bottles were sterilised at 121°C for 15 min to eliminate any possible bacterial contamination typical of ABEI processes [44] and, after adding 0.01 g/L FeSO 4 · 7H 2 O, 0.2 g/L MgSO 4 ·7H 2 O and 0.5 g/L cysteine, the pH was adjusted to 6.0 with a solution of NaOH 40%. The bottles were closed in an airtight manner with rubber caps and they were inoculated with 3% (v/v) of the corresponding bacterial strain. Gaseous N 2 was bubbled into the bottom of the sample for 5 min to guarantee anaerobic conditions. Then, the bottles were incubated at 35°C and 100 rpm during 96 h in an orbital shaker (Infors HT Minitron, Infors AG, Bottmingen, Switzerland). Control fermentations were carried out for all the strains, with an aqueous solution containing glucose and xylose mixtures at similar concentrations to those of tomato hydrolysates (37 g/L glucose and 12 g/L xylose) and the abovementioned nutrients. All experiments were performed in triplicate.

Strain Cultivation and Inocula Preparation
Twelve different bacterial and yeast strains were compared for ethanol production. Yeast strains, from a cryopreserved solution (glycerol 80% v/v), were inoculated on culture plates (10 g/L glucose, 3 g/L yeast extract, 3 g/L malt extract, 5 g/L soy peptone, 20 g/L agar) and maintained at 20°C until obtaining colonies of 1-2 mm. Subsequently, seed cultures for fermentation were prepared in liquid medium from a single yeast colony and kept at 30°C and 150 rpm during the time necessary to reach a concentration of 1·10 8 cells/mL (incubation time varied from 7 to 24 h depending on the strain used, with the exception of strain DSM 3434 which required 48 h). In the case of Z. mobilis DSM 3580, cryopreserved stock solutions were cultured as described previously [45] during 24 h. Then, a single colony was inoculated in 50 mL of liquid medium in a 100-mL glass bottle capped with a rubber stopper in anaerobic conditions by injecting gaseous nitrogen in the headspace during 5 min. The inoculum was kept 24 h at 30°C (cell density 1·10 8 cells/mL). For all strains, cell density was determined by using a Bürker counting chamber (Paul Marienfeld GmbH & Co. KG, Lauda-Königshofen, Germany).

Alcoholic Fermentation
Because of the lower nutritional requirements of yeasts in comparison with Clostridium, tomato waste hydrolysate was not supplemented with any nutrients for alcoholic fermentation. In addition, due to the competitive capacity of yeasts, the broth was not sterilised before fermentation, which implies energy savings. Alcoholic fermentations with the bacterium Z. mobilis DSM 3580 were also performed without nutrient supplementation and without sterilisation for comparison purposes. The pH of tomato waste hydrolysate was adjusted to 5.0 with NaOH and it was inoculated with 3% (v/v) of liquid inoculum containing yeasts or bacteria. All yeast fermentations were performed in 100-mL Erlenmeyer flasks containing 50 mL tomato hydrolysate, plugged with foam stoppers, under aerobic conditions. Fermentations with Z. mobilis DSM 3580 were carried out in 100-mL rubber-capped bottles containing 50 mL tomato hydrolysate, where gaseous N 2 was bubbled during 5 min to guarantee anaerobic conditions. Fermentation controls were prepared with aqueous solutions at pH 5.0 containing glucose and xylose mixtures at similar concentrations to those of tomato hydrolysates (60 g/L glucose, 22 g/L xylose), and supplemented with nutrients and salts (10.4 g/L yeast extract and 1.47 g/L KH 2 PO 4 for yeasts; and 7 g/L yeast extract, 2.5 g/L K 2 HPO 4 , 1.6 g/L (NH 4 ) 2 SO 4 and 1 g/L MgSO 4 ·7H 2 O for bacteria). All samples and controls were fermented in triplicate in an orbital shaker at 30°C and 150 rpm during 72 h.

Chemical Analyses
Samples of hydrolysates and fermented broths were centrifuged, filtered and analysed according to Hijosa-Valsero et al. [40] for the quantification of sugars (cellobiose, maltose, glucose, xylose, galactose, mannose, rhamnose and arabinose), potential fermentation inhibitors [formic acid, acetic acid, levulinic acid, 5-hydroxymethylfurfural (5-HMF) and furfural] and fermentation products (acetone, butanol, ethanol, isopropanol, acetic acid and butyric acid). Butanol and ethanol fermentation yields (Y B/S , Y E/S ) and productivities (W B , W E ) were calculated as reported elsewhere [40], based on total sugar consumption (Eqs. 2 and 3, respectively): where Y i/S is the fermentation yield of metabolite i (expressed in g/g); i t is the mass (g) of metabolite i at the end of the fermentation; S 0 is the mass (g) of total sugars at the beginning of the fermentation, S t is the mass (g) of total sugars at the end of the fermentation; W i is the fermentation productivity of metabolite i [expressed in g/(L·h)], C it is the concentration (g/L) of metabolite i at the end of the fermentation and t is the fermentation time (h). In addition, the liquid and solid fractions of the hydrolysates were separated and measured after the enzymatic pretreatment and after the fermentation in order to perform a mass balance.

Statistical Analyses
Fermentation samples were compared with a one-way ANOVA and Tukey's HSD test using the software Statistica 7 (StatSoft Inc., Tulsa, OK, USA).

Hydrolysis of Tomato Waste
High solvent-to-solid ratios were employed during the pretreatment (20-30% biomass, w/w) in order to guarantee a sufficient sugar concentration in the hydrolysate for fermentations, since the total carbohydrate content of dry tomato waste was relatively low (30.6% carbohydrates). It was observed that the enzymatic hydrolysis step was slowed down and prolonged from 48 to 120 h when increasing the solid biomass load from 20 to 30% w/w. Therefore, the hydrolysis of the samples destined for ABEI fermentation (20% solid load) was faster than that of the samples prepared for ethanol fermentation (30% solid load).
The chemical composition of tomato waste hydrolysates is shown in Table 1. The broth used for ABEI fermentation contained 44.1 g/L total sugars, which implies a saccharification efficiency of 61.0% of the carbohydrates present in tomato waste biomass. On the other hand, the broth prepared for ethanol fermentation was composed of 85.1 g/L total sugars, entailing a saccharification efficiency of 66.8%. Other tomato pomace hydrolysates reported in literature and destined for ethanol fermentation contained 36-57 g/L total sugar [24,39]. Therefore, the applied hydrothermal pretreatment in the present work (121°C, 20 min) was efficient for sugar release. In fact, previous works have evidenced that the use of reagents, such as acids, is not necessary for the pretreatment of tomato wastes and that the hydrothermal process, followed by enzymatic hydrolysis, is an efficient pretreatment for tomato wastes [36,46]. Working temperatures for the physicochemical pretreatment of tomato pomace have been reported between 110 and 135°C [36,46]. Treatment times for hydrothermal processes of tomato and other vegetable wastes are in the range of 5-60 min [35,36,40,46].
Regarding the generation of fermentation inhibitors, only formic acid (0.23-0.36 g/L) and acetic acid (1.64-2.79 g/L) exhibited relevant values, especially in the case of the hydrolysate allocated to ethanol fermentation (Table 1). Levulinic acid, 5-HMF and furfural were present at very low concentrations.

Biobutanol Production from Tomato Waste
The Clostridium species selected to perform the fermentation screening corresponded to typical solventogenic Clostridia widely mentioned in the scientific literature [6,10]. Some strains, such as DSM 51, DSM 552, DSM 791, DSM 1820, DSM 6423, DSM 13864, DSM 2152, DSM 792, DSM 6228,  [34,35,43,47,48]. In the present work, some of the tested strains produced acetone (A), butanol (B) and ethanol (E) as main solvents, whereas other strains transformed part of the acetone into isopropanol (I), according to their typical metabolic pathways [3]. The highest acetone production was achieved by C. saccharobutylicum DSM 13864 (5.95 ± 0.36 g/L acetone), the best butanol concentration was obtained by C. beijerinckii DSM 1820 (7.00 ± 0.05 g/L butanol), whereas the greatest isopropanol value was provided by C. acetobutylicum DSM 792 (8.28 ± 0.85 g/L isopropanol) (Fig. 2, Table S1). Ethanol concentrations were below 1 g/L for all the tested strains. Regarding the total production of solvents (sum of concentrations, ABEI), the strains C. acetobutylicum DSM 792 and C. beijerinckii DSM 6423 were significantly superior to the rest of strains (p < 0.05), attaining concentrations of 14.5-15.1 g/L ABEI (Fig. 2,Table S1). On the other hand, some strains, such as C. beijerinckii CECT 508, DSM 51, DSM 552, DSM 791; C. pasteurianum DSM 526; and C. saccharoperbutylacetonicum DSM 2152 and DSM 14923, were unable to properly ferment tomato pomace (Fig. 2, Table S1). Although isopropanol concentration was slightly higher for C. acetobutylicum DSM 792 than for C. beijerinckii DSM 6423 (p < 0.05), the fermentation broths of the former were difficult to handle during the posterior analytical procedures due to their gelatinous nature and high viscosity. This fact should be taken into account for potential industrial applications.
It must be noted that the isopropanol:butanol ratio of the two abovementioned strains (DSM 792 and DSM 6423) was remarkably high for the fermentation of tomato wastes (A:B:E:I was approximately 1:28:1.6:38 for DSM 792, and 1:24:1.4:27 for DSM 6423), a fact which could be related to the chemical composition of tomato pomace hydrolysate. In fact, isopropanol:butanol ratios were clearly lower in the case of control fermentations of simple synthetic media (Table S2). It has been observed that the acetone:butanol ratio in C. acetobutylicum can vary depending on the nutrient composition of the fermentation medium (mainly iron and phosphorus) and on fermentation conditions like pH [49]. In addition, the ratios of isopropanol:acetone are affected by the concentrations of acetate, butyrate and vitamin B3 [50].
Total sugar consumption varied between 18 and 86% depending on the strain (Fig. 2, Table S1). Sugar consumption was related to a successful ABEI fermentation. Actually, all the strains with a sugar consumption above 70% (DSM 6228, DSM 6423, DSM 1820, DSM 13864, DSM 792) were able to produce at least 5 g/L butanol. It was not possible to measure cell density in tomato pomace hydrolysates and fermented broths due to the presence of microscopic biological debris. The highest butanol yields (Y B/S ) attained were about 0.20 g/g (Table S1). This value is lower than the yields of 0.27 g/g reported in literature for the fermentation of other food wastes like apple pomace or coffee silverskin [34,40], but similar to that of potato peel [35].  Taking the strain C. beijerinckii DSM 6423 as a model, a mass balance for ABEI fermentation was proposed (Fig. 3). From 100 g dry tomato pomace (containing 30.6 g carbohydrates), it is possible to obtain 2.67 g butanol, 2.99 g isopropanol and 5.92 g ABEI.
The butanol (6.2-7.0 g/L) and isopropanol (7.3-8.3 g/L) concentrations obtained in the best cases represent a promising starting point for future research focused on process optimisation. In addition, it must be highlighted that tomato pomace hydrolysate was directly fermentable without the need of any detoxification step to remove potential inhibitors, which simplifies the process. The possibility of using tomato wastes for ABEI fermentation could enlarge the variety of potential lignocellulosic feedstocks and contribute to the engineering development of biorefineries. According to Jiang et al. [7], feedstock costs would represent 30% of the total process costs in an ABE biorefinery based on cellulosic biomass, whereas in the traditional corn-based ABE fermentation this cost attained 66%. The cost of the petrochemical process to produce nbutanol is about 1.23 US$/L, while that of traditional ABE fermentation from corn is about 1.51-1.90 $/L, and 0.80-1.07 $/L from a lignocellulosic by-product like corn stover [7]. In order to sustain an economically viable biorefinery, minimum selling prices for lignocellulosic biobutanol have been calculated to be in the range of 0.40-1.38 $/L [51][52][53]. However, technoeconomic drawbacks still have to be overcome before the establishment of commercial ABEI biorefineries based on lignocellulosic by-products [10,14].

Ethanol Production from Tomato Waste
The selection of the yeast and bacterial strains for alcoholic fermentation was based on previous studies in the case of K. lactis DSM 70799, K. marxianus DSM 5422, DSM 5418, DSM 7239, S. cerevisiae Ethanol Red, Hércules-green, CECT 1383 [54], S. stipitis DSM 3651 [55] and Z. mobilis DSM 3580 [56]. On the contrary, the available information about fermentation of agri-food by-products with L. thermotolerans is scarce [57]. The strains employed in this study for ethanol production differ in their abilities to ferment or assimilate various carbon sources ( Table 2). The species S. stipitis, K. marxianus and K. lactis var. lactis have the potential to grow on a broad variety of carbohydrate substrates.
According to the results obtained in control fermentations, all the tested strains were capable of fermenting glucose/ xylose mixtures (Table S3). However, not all of them were able to cope with tomato pomace hydrolysate, although cell densities at the end of the fermentation were similar to those of Fig. 3 Mass balance of ABEI and alcoholic fermentation (normalised to an initial amount of 100 g dried tomato pomace) for two model microorganisms. Mass losses due to evaporation have been considered for the calculation. Note: Xyl, xylose; Man, mannose; Gal, galactose control samples in most cases (Table S4). In this regard, the four best-performing strains for tomato samples were K. marxianus DSM 5422, S. cerevisiae Ethanol Red®, L. thermotolerans DSM 3434 and S. cerevisiae Hércules, with ethanol values of 20.1-21.7 g/L (Fig. 4, Table S4). Other strains like K. marxianus DSM 5418, K. marxianus 7239, K. lactis DSM 70799 and Z. mobilis DSM 3580 offered ethanol values above 16 g/L. On the contrary, the two strains of S. stipitis (DSM 3651, DSM 3652) were unable to ferment tomato pomace hydrolysate, in spite of their ability to assimilate and ferment numerous sugar sources (Table 2), a fact which could indicate their sensitivity to inhibitory compounds present in tomato hydrolysate. It has been reported that S. stipitis is less tolerant to inhibitors than S. cerevisiae [70].
Bellido et al. [55] observed that a concentration of 2.5 g/L acetic acid caused an inhibition of 60% in growth and ethanol production in S. stipitis DSM 3651, an acetic acid value similar to that of the tomato pomace hydrolysate (Table 1).
Total sugar consumption was 78-90% for all the yeasts that fermented tomato pomace successfully (Fig. 4, Table S4). Nevertheless, the bacterial strain Z. mobilis DSM 3580 had a clearly lower (p < 0.05) sugar consumption, which could be related to its limited carbohydrate utilisation ( Table 2).
According to metabolic stoichiometry, 1 mol of glucose produces 2 mol of ethanol, and 1 mol of xylose produces 1.6 mol of ethanol (or even 0.96 mol ethanol) [61]. This implies theoretical ethanol yields (Y E ) of 0.511 g/g for glucose and 0.294-0.491 g/g for xylose. Under experimental   Table S4 conditions, the observed ethanol yields Y E/S for glucose:xylose mixtures are temperature dependent and lie in the range of 0.37-0.49 g/g [62]. The ethanol yields based on total sugars (Y E/S ) calculated in the present study are slightly lower than the abovementioned results; for instance, the four best-performing yeasts for tomato pomace hydrolysate obtained yields of 0.266-0.323 g/g (Table S4). The best ethanol concentration of 21.7 g/L (obtained by K. marxianus DSM 5422) is clearly higher than those reported in literature for the fermentation of similar tomato wastes. Patle and Lal [39] employed a co-culture of Z. mobilis MTCC 92 and Candida tropicalis TERI SH 110 to ferment a tomato waste hydrolysate containing 36 g/L sugar, and obtained 14 g/L ethanol. Kasavi et al. [38] fermented untreated tomato peels with S. cerevisiae strains BC187, L-1374, L-1528, K11 and Y9, but with an extremely low sugar initial concentration (1 g/L carbon), which resulted in an ethanol concentration of 0.16-0.27 g/L and a sugar consumption of 42.3-54.8%. Lenucci et al. [24] subjected tomato pomace to enzymatic hydrolysis and obtained a broth with 28.6 g/L total sugars, which was fermented by S. cerevisiae strain Cispa 161 to about 10 g/L ethanol in 12 h; in addition, they prepared a more concentrated broth with 57 g/L total sugars, which yielded 15 g/L ethanol in 72 h.
The strain K. marxianus DSM 5422 was selected to calculate the mass balance of the alcoholic fermentation (Fig. 3). It was estimated that 5.21 g ethanol could be obtained from 100 g dry tomato pomace.
In order to obtain profitable concentrations of ethanol, it is necessary to start the fermentation with a sufficient sugar concentration [54,71,72]. Because of that, the biomass-tosolvent ratio during the pretreatment has to be increased. The minimum solids concentration for economic ethanol production is estimated at about 15% [73]. As mentioned in section Hydrolysis of Tomato Waste, the biomass ratio of 30% resulted in a slow enzymatic hydrolysis. Taking into account this problem and the low ethanol titres obtained, tomato pomace does not seem a good feedstock for bioethanol production under the conditions tested in this study.

Conclusions
Tomato pomace can be successfully pretreated with a hydrothermal process followed by enzymatic hydrolysis. Saccharification efficiencies of 60-67% can be attained for biomass loads of 20-30% (w/w). However, to avoid long treatment times during the enzymatic hydrolysis, solid-tosolvent ratios of up to 20% (w/w) are recommended.
Although the ethanol concentrations obtained for tomato pomace fermentation (~20 g/L) are the highest hitherto reported, this value is insufficient for industrial exploitation. The fact that the ethanol values reached in control solutions were similar to those of tomato pomace hydrolysates could indicate that the glucose/xylose composition of the sample was not very favourable for yeast fermentation. In any case, the good performance demonstrated by some strains of K. marxianus, K. lactis or L. thermotolerans suggests that they could be interesting alternatives to S. cerevisiae as potential biocatalysts for other lignocellulosic biomass hydrolysates containing a different sugar composition.
In contrast, tomato pomace was a suitable feedstock for ABEI fermentation. The butanol (6.2-7.0 g/L) and isopropanol (7.3-8.3 g/L) concentrations attained by some strains are only slightly below industrial values for conventional ABEI processes. Another remarkable fact is the possibility of directly fermenting tomato pomace hydrolysates without the need of any detoxification technique by selecting the most appropriate microbial strains. However, before considering the use of this biomass in commercial biorefineries, further research would be necessary to optimise fermentation and nutrient conditions to reduce costs.