Hydrolysis of the β-1,4-glycosidic bonds in cellulose can be achieved by many different enzymes known as cellulases which use two different catalytic mechanisms, the retaining and the inverting mechanisms. All GH 12 cellulases, for example, hydrolyze glycosidic bonds by the retaining mechanism whereas family 6 cellulases use the inverting mechanism 33,36. In both mechanisms, two catalytic carboxylate residues are involved and catalyze the reaction by acid-base catalysis. Many different fungal species have the ability to degrade cellulose by producing extracellular fungal cellulose-degrading enzymes including endo-cleaving (endoglucanases) and exo-cleaving (cellobiohydrolases). Endoglucanases can hydrolyze glycosidic bonds internally in cellulose chains whereas cellobiohydrolases act preferentially on chain ends. The products of the enzymatic reaction are mostly a disaccharide known as cellobiose and, to a lesser extent, cello-oligosaccharides, which will be further hydrolyzed by the third group of enzymes called β-glucosidases 56. Cellulases mostly have a small independently folded carbohydrate binding module (CBM) which is connected to the catalytic domain by a flexible linker. The CBMs are responsible for binding the enzyme to the crystalline cellulose and thus enhance the enzyme activity 33. Currently many CBMs have been identified and classified into 54 families, however only 20 families (1, 13, 14, 18, 19, 20, 21, 24, 29, 32, 35, 38, 39, 40, 42, 43, 47, 48, 50 and 52) have been found in fungi. Different fungal cellulolytic enzymes and their main features are summarized in Table 3.

Several different enzymes are needed to hydrolyze hemicelluloses, due to their heterogeneity 10. Xylan is the most abundant component of hemicellulose contributing over 70% of its structure. Xylanases are able to hydrolyze β-1,4 linkages in xylan and produce oligomers which can be further hydrolyzed into xylose by β-xylosidase. Not surprisingly, additional enzymes such as β-mannanases, arabinofuranosidases or α-L-arabinanases are needed depending on the hemicellulose composition which can be mannan-based or arabinofuranosyl-containing 71. Similar to cellulases, hemicellulases are usually modular proteins and have other functional modules, such as CBM, in addition to their catalytic domains. Also similarly to cellulases, most of the hemicellulases are glycoside hydrolases (GHs), although some hemicellulases belong to carbohydrate esterases (CEs) which hydrolyze ester linkages of acetate or ferulic acid side groups 71,72. Hemicellulases belong to 20 different GH families (1, 2, 3, 4, 5, 8, 10, 11, 26, 27, 36, 39, 43, 51, 52, 53, 54, 57, 62 and 67) and all of them except for 4 (families 4, 8, 52 and 57) have been found in fungi. All but 1 (family 7) of the 7 different CE families (1, 2, 3, 4, 5, 6 and 7) reported for hemicellulases have been found in fungi 35. Similarly to cellulases, aerobic fungi such as Trichoderma and Aspergillus secrete a wide variety of hemicellulases in high concentrations (8 and 12 hemicellulases, respectively) and these work in a synergistic manner 71.

Lignin, the most abundant renewable aromatic polymer on the Earth, is composed of non-phenolic (80-90%) and phenolic structures 73. It has been shown that fungi degrade lignin by secreting enzymes collectively termed “ligninases”. These include two ligninolytic families; i) phenol oxidase (laccase) and ii) peroxidases [lignin peroxidase (LiP) and manganese peroxidase (MnP)] 74. White-rot basidiomycetes such as Coriolus versicolor 73, P. chrysosporium and T. versicolor 75 have been found to be the most efficient lignin-degrading microorganisms studied. Interestingly, LiP is able to oxidize the non-phenolic part of lignin, but it was not detected in many lignin degrading fungi. In addition, it has been widely accepted that the oxidative ligninolytic enzymes are not able to penetrate the cell walls due to their size. Thus, it has been suggested that prior to the enzymatic attack, low-molecular weight diffusible reactive oxidative compounds have to initiate changes to the lignin structure (as discussed below) 76,77.

A few decades ago, non-enzymatic degradation mechanisms for plant cell-wall polysaccharide degradation were also considered and over the time more evidence for these was found. The non-enzymatic degradation mechanism is mostly assisted by oxidation through production of free hydroxyl radicals (•OH). In fact, many white and brown-rot fungi have been shown to produce hydrogen peroxide (H₂O₂) which enters the Fenton reaction and results in release of •OH 78,79. These free radicals attack polysaccharides as well as lignin in plant cell walls in a nonspecific manner providing some cleavages which make it easier for the lignocellulolytic enzymes to penetrate 80,81. Three different pathways have been found for the generation of free radicals (discussed below) including cellobiose dehydrogenase (CDH) catalyzed reactions, low molecular weight peptides/quinone redox cycling and glycopeptide-catalyzed Fenton reactions (Table 4) 44.

CDH, an extracellular monomeric protein with some glycosylation has been identified in a number of wood- and cellulose-degrading fungi including basidiomycetes (mostly white-rot fungi) and ascomycetes growing on cellulose. The enzyme is able to oxidize cellobiose, higher cellodextrins and other disaccharides or oligosaccharides with β-1,4 linkages. In addition, CDH with (in ascomycetes) or without CBM (in basidiomycetes) have been identified however even in the absence of CBM they are able to bind to cellulose through hydrophobic interactions 82. It has been shown in some fungi that under cellulolytic conditions CDH production increases which helps cellulases and hemicellulases 83,84. It is now widely accepted that CDH are able to degrade and modify all three major components of the lignocellulosic residues (cellulose, hemicelluloses and lignin) by producing free hydroxyl radicals in a Fenton-type reaction (for detailed information please refers to the review by Baldrin and Valaskova, 2008 44).

It has been shown that white and brown-rot fungi produce low molecular weight chelators which are able to penetrate into the cell wall. For example Gloeophyllum trabeum produces a low molecular weight peptide (known as short fiber generating factor, SFGF) which can degrade cellulose into short fibers by an oxidative reaction 81,85. It has also been reported that some of these low molecular weight compounds are quinones which have to be converted to hydroquinones by some fungal enzymes (Table 4) and then through Fenton reaction, free hydroxyl radicals will be produced 73.

Different glycopeptides with different molecular weight (ranging from 1.5 to 12 kDa) have been found in many brown-rot fungi such as G. trabeum 86 and white-rot fungi such as P. chrysosporium 77,87. Similar to the other mechanisms, glycopeptides are able to catalyze redox reactions and thus produce free hydroxyl radicals.

Anaerobic fungi represent a special group of microorganisms inhabiting the gastro-intestinal tract of ruminants and most non-ruminant herbivores. These fungi, along with some anaerobic bacteria (mainly from the class Clostridia e.g. Clostridium thermocellum 92), produce a range of cellulolytic and hemicellulolytic enzymes in a multienzyme complex known as cellulosome. The first anaerobic gut fungi able to break down ingested lignocellulosic residues were identified in 1975 by Orpin 93 and since then 6 genera and 18 species have been identified some of which are shown in Table 2. The cellulosome, however, was initially discovered in anaerobic bacteria (Clostridium thermocellum) in 1983 94, and then first described in anaerobic fungi in 1992 (Neocallimastix frontalis) 95. In anaerobic bacteria, the cellulosome is usually comprised of 20 or more different cellulolytic/hemicellulolytic enzymes. However, in anaerobic fungi such as N. frontalis and Piromyces, cellulosomes include at least six or ten polypeptides, respectively (cellulosome-type complex) 55,95,96. All hydrolytic enzymes in the cellulosome are bound together by noncatalytic scaffolding proteins. In addition to catalytic subunits, all enzymes have noncatalytic subunits known as “fungal dockerin domains” (FDD), which allow binding to cohesin modules of the scaffolding proteins. Interestingly, 50 fungal FDDs have been identified so far which present different amino acid sequences than those found in bacterial dockerins 31,97.

Anaerobic fungi efficiently hydrolyze cellulose and hemicellulose by producing many lignocellulolytic enzymes. Most of the enzymes are associated with the cellulosome; however, some free enzymes also have been identified. In Piromyces sp. PC2, a cellulosome-producing anaerobic fungus for example, 17 lignocellulolytic enzyme encoding genes have been isolated including ten cellulases, one β-glucosidases, five hemicellulases and one enzyme facilitating protein folding. Interestingly, FDD has been reported only for 11 of the genes, which indicates that these cellulases are cellulosome-associated. Moreover, CBM has been identified only in three of those 17 genes, including two cellulases and one hemicellulase 31. Interestingly, the major products of cellulose digestion by fungal cellulosomes are glucose which eliminate the costly addition of β-glucosidase, whereas in the case of bacterial cellulosomes, cellobiose is the major product 98. Despite many advantages of cellulosomes such as synergistic activity between the components and efficient hydrolytic activity on both cellulose and hemicellulose, fungal cellulosomes are much less well characterized compared to bacterial cellulosomes.

Pretreatment of the lignocellulosic residues is necessary because hydrolysis of non-pretreated materials is slow, and results in low product yield. Some pretreatment methods increase the pore size and reduce the crystallinity of cellulose (Figure 1). Pretreatment also makes cellulose more accessible to the cellulolytic enzymes, which in return reduces enzyme requirements and thus the cost 99. Many different pretreatment methods have been used, but they can be categorized into three broad groups: chemical (e.g. acid or alkali), physical/ physicochemical (e.g. physical ball milling or physicochemical steam explosion) and biological pretreatment by microorganisms 100. In the chemical pretreatment method using acid for example, hemicelluloses will be targeted whereas in alkali-catalyzed pretreatment mainly lignin is removed 22. It has been suggested that, there will probably not be a general pretreatment procedure and that different raw materials will require different pretreatments 22. Since many fungal cellulolytic enzymes (fungal-derived cellulases and β-glucosidases) work at lower pH (usually 4-5) acidic pretreatment seems a preferred option when fungal enzymes are chosen for the hydrolysis 22.

Biological pretreatment uses microorganisms and their enzymes selectively for delignification of lignocellulosic residues and has the advantages of a low-energy demand, minimal waste production and a lack of environmental effects 101. White-rot basidiomycetes possess the capabilities to attack lignin. P. chrysosporium, for example, has been shown to non-selectively attack lignin and carbohydrate 102. P. chrysosporium was successfully used for biological pretreatment of cotton stalks by solid state cultivation (SSC) and results have shown that the fungus facilitates the conversion into ethanol 101. Other basidiomycetes such as Phlebia radiata, P. floridensis and Daedalea flavida, selectively degrade lignin in wheat straw and are good choices for delignification of lignocellulosic residues 103. Ceriporiopsis subvermispora, however, lacks cellulases (cellobiohydrolase activity) but produces manganese peroxide and laccase, and selectively delignifies several different wood species 104.

After pretreatment, cellulose and hemicelluloses are hydrolyzed to soluble monomeric sugars (hexoses and pentoses) using cellulases and hemicellulases, respectively (Figure 1). As mentioned earlier, many fungal species such as Trichoderma, Penicillium, Aspergillus and T. emersonii are able to produce large amounts of extracellular cellulases and hemicellulases. High temperature and low pH tolerant enzymes are preferred for the hydrolysis due to the fact that most current pretreatment strategies rely on acid and heat 105. In addition, thermostable enzymes have several advantages including higher specific activity and higher stability which improve the overall hydrolytic performance 106. Ultimately, improvement in catalytic efficiencies of enzymes will reduce the cost of hydrolysis by enabling lower enzyme dosages. Some fungal strains such as T. emersonii 107, T. aurantiacus 29, T. terrestris 30, S. thermophile 28, Chaetomium thermophilum 108 and Corynascus thermophilus 109 can produce thermostable enzymes which are stable and active at elevated temperatures (˃60°C) well above their optimum growth temperature (30 to ~55) 110. Due to the promising thermostability and acidic tolerance of thermophilic fungal enzymes, they have good potential to be used for hydrolysis of lignocellulosic residues at industrial scales.

In the fermentation process, the hydrolytic products including monomeric hexoses (glucose, mannose and galactose) and pentoses (xylose and arabinose) will be fermented to valuable products such as ethanol (Figure 1). Among these hydrolytic products, glucose is normally the most abundant, followed by xylose or mannose and other lower concentration sugars. Saccharomyces cerevisiae is the most frequently and traditionally used microorganism for fermenting ethanol from starch-based residues at industrial scales 22. S. cerevisiae has a few advantages such as its wide public acceptance, high fermentation rate and high ethanol tolerance that make it a good candidate for fermentation processes. However, S. cerevisiae is unable to efficiently utilize xylose as the sole carbon source or ferment it to ethanol 111. To make industrial lignocellulosic bioconversion more economically feasible, it is necessary to choose microorganisms capable of fermenting both glucose and xylose. Therefore, many successful attempts have been made to improve xylose fermentation in S. cerevisiae since the first discovery of pentose-fermenting yeasts in 1981 by a Canadian group 112. These efforts can be classified within two major groups: recombinant (e.g. metabolic engineering of S. cerevisiae with genes from other xylose-fermenting yeasts) and non-recombinant (e.g. adaptation) techniques. These improvements have reached the point where the deficient xylose-fermenting S. cerevisiae can now convert xylose to ethanol at an efficiency close to its theoretical value of 0.51 g g^-1 (for extensive review please read Chu and Lee, 2007 111).

In addition to xylose, S. cerevisiae is also unable to ferment arabinose, unless supplemented with rich media 113. Therefore, recombinant S. cerevisiae harbouring xylose-fermenting genes have been engineered with arabinose-metabolizing genes from other microorganisms. The latest recombinant S. cerevisiae (TMB 3400) has been shown to successfully ferment both xylose and arabinose in addition to glucose 114.

During fermentation of lignocellulosic-based biomass, S. cerevisiae faces yet another challenge: the presence of inhibiting compounds including low molecular weight organic acids, furan derivatives, phenolics and inorganic compounds. These compounds are released and formed during pretreatment and/or hydrolysis of the lignocellulosic residues 115. Thus, it is necessary to detoxify hydrolytic products before the fermentation which increases process cost in addition to sugar loss 116. Interestingly, S. cerevisiae is one of the least sensitive microorganisms to the inhibitory effect of lignocellulolytic hydrolysate inhibitors. In a recent study for example, glucose and xylose, the hydrolytic products of steam-pretreated corn stover were efficiently co-fermented to ethanol without detoxification using the recombinant S. cerevisiae strain TMB 3400 117. It is also possible to adopt recombinant xylose-fermenting S. cerevisiae to the hydrolysate inhibitors by continuous cultivation in the presence of the inhibitors 118.

Attempts have been taken to reduce by-product inhibition. In a recent study for example, wheat straw pellets were subjected to wet explosion pretreatment using three different oxidizing agents, H₂O₂, O₂ and air 119. Interestingly, the pretreatment with O₂ has been shown to be the most efficient in enhancing conversion of the raw material to sugars. Using the method also has been minimized the production of furfural as a by-product which improved enzymatic hydrolysis and minimized the enzyme loading to 10 FPU/g with conversion rate of 70 and 68% for cellulose and hemicellulose respectively 119. Ammonia fiber explosion (AFEX) pretreatment also has been shown to be a good candidate since it does not produce some inhibitory by-products such as furans. However, the disadvantage of the method is that some of the phenolic compounds in lignin may remain on the pretreated material, which then needs to be washed. This creates wastewater, which causes the process to become environmentally unfriendly 100.

The last two steps of bioconversion of pretreated lignocellulolytic residues to ethanol (hydrolysis and fermentation) can be performed separately (SHF) or simultaneously (SSF) (Figure 1). In the separate hydrolysis and fermentation (SHF), the hydrolysate products will be fermented to ethanol in a separate process. The advantage of this method is that both processes can be optimized individually (e.g. optimal temperature is 45-50 °C for hydrolysis, whereas it is 30 °C for fermentation). However, its main drawback is the accumulation of enzyme-inhibiting end-products (cellobiose and glucose) during the hydrolysis. This makes the process inefficient, and the costly addition of β-glucosidase is needed to overcome end-product inhibition 120. In simultaneous saccharification and fermentation (SSF), however, the end-products will be directly converted to ethanol by the microorganism. Therefore, addition of high amounts of β-glucosidase is not necessary and this reduces the ethanol production costs 121. However, the main drawback of SSF is the need to compromise processing conditions such that temperature and pH are suboptimal for each individual step. However, the development of recombinant yeast strains (i.e. improved thermotolerance) is expected to enhance the performance of SSF 1. Further process integration can be achieved by a process known as consolidated bioprocessing (CBP) which aims to minimize all bioconversion steps into one step in a single reactor using one or more microorganisms. CBP operation featuring cellulase production, cellulose/hemicellulose hydrolysis and fermentation of 5- and 6- carbon sugars in one step have shown the potential to provide the lowest cost for biological conversion of cellulosic biomass to fuels, when processes relying on hydrolysis by enzymes and/or microorganisms are used (Figure 1) 122.

Many fungal strains have been subjected to extensive mutagenesis studies due to their ability to secrete large amounts of cellulose-degrading enzymes. It has been four decades since Mandels and Weber (late 1960s) screened over 100 wild-type strains of Trichoderma species to isolate the best cellulolytic strain and came up with T. reesei (initially called T. viride QM6a) 123. Cellulolytic activity of T. reesei QM6a has been improved by using different mutagenesis techniques including UV-light and chemicals at the US Army Natick Laboratory, resulting in the mutant QM 9414 with higher filter paper activity (FPA) 124. Other studies in different laboratories have also made significant contributions to strain improvements using mutagenesis techniques, leading to development of the mutant strains M7, NG14 125 and RUT-C30 126. T. reesei RUT-C30 is one of the best known mutants, producing 4-5 times more cellulase than the wild-type strain (QM 6a). A recent study by Kovács and et al. (2008) has shown that wild-type Trichoderma atroviride (F-1505) produces the most cellulase among 150 wild-type Trichoderma. Moreover, T. atroviride mutants were created by mutagenesis using N-methyl-N′-nitro-N-nitrosoguanidine (NTG) as well as UV-light. These T. atroviride mutants (e.g. T. atroviride TUB F-1724) produce high levels of extracellular cellulases as well as β-glucosidase when they are grown on pretreated willow 127.

Cellulase and xylanase activities in Penicillium verruculosum 28K mutants were improved about 3-fold using four cycles of UV mutagenesis. The enzyme production was further improved by 2- to 3-fold in a two-stage fermentation process using wheat bran, yeast extract medium and microcrystalline cellulose as the inducer 128. However, caution has to be taken during strain improvement by mutagenesis. Studies have shown that the best T. reesei mutant (RUT-C30) lacks an 85 Kb genomic fragment and is consequently missing 29 genes which include transcription factors, metabolic enzymes and transport proteins. In fact, these genotype changes are correlated with phenotypic changes such as poor growth on α-linked oligo- and polyglucosides or disturbance of osmotic homeostasis 129.

On the other hand, site-directed mutagenesis (SDM) has played a central role in the characterization and improvement of cellulases including their putative catalytic and binding residues. Different site-directed mutagenesis methods such as saturation mutagenesis, error-prone PCR and DNA shuffling have been used to improve specific enzyme properties. For example, by the application of SDM it was found that Glu 116 and 200 are the catalytic nucleophile and acid-base residues in Hypocrea jecorina (anamorph T. reesei) Cel12A, respectively. In the study, mutant enzymes were produced where Glu was replaced by Asp or Gln at each position (E116D/Q and E200D/Q). The specific activity of these mutants was reduced by more than 98%, suggesting the critical role of these two residues in the catalytic function of the enzyme 130.

In another study, the thermostable endo-1,4-β-xylanase (XynII) mutants from T. reesei were further mutated to resist inactivation at high pH by using SDM. All mutants were resistant to thermal inactivation at alkaline pH. For example, thermotolerance for one mutant (P9) at pH 9 was increased approximately 4-5 °C, resulting in better activity in sulphate pulp bleaching compared to the reference 131. Also, the catalytic efficiency and optimum pH of T. reesei endo-β-1,4-glucanase II were improved by saturation mutagenesis followed by random mutagenesis and two rounds of DNA shuffling. The pH optimum of the variant (Q139R/L218H/W276R/N342T) was shifted from 4.8 to 6.2, while the enzyme activity was improved more than 4.5-fold 132. Moreover, the stability of T. reesei endo-1,4-β-xylanases II (XynII) was increased by engineering a disulfide bridge at its N-terminal region. In fact, two amino acids (Thr-2 and Thr-28) in the enzyme were substituted by cysteine (T2C:T28C mutant) resulting in a 15 °C increase in thermostability 133.

Fungal co-culturing offers a means to improve hydrolysis of lignocellulosic residues, and also enhances product utilization which minimizes the need for additional enzymes in the bioconversion process. In the case of cellulose degradation, for example, all three enzymatic components (EG, CBH and β-glucosidase) have to be present in large amounts. However, none of the fungal strains, including the best mutants, are able to produce high levels of the enzymes at the same time. T. reesei for example produces CBH and EG in high quantities whereas its β-glucosidase activity is low 134. A. niger however, produces large amounts of β-glucosidase, but has limited EG components 56. In addition, hemicellulose hydrolysis must also be considered when lignocellulosic residues are subjected to biomass conversion. However, this will be determined by the pretreatment methods. Specifically in an alkali pretreatment method, a part of lignin will be removed and thus hemicellulose has to be degraded by the use of hemicellulases, whereas in acid-catalyzed pretreatment, the hemicellulose layer will be hydrolyzed 22. Again, some fungal strains have been shown to work more efficiently on cellulosic residues whereas others produce more hemicellulolytic enzymes and efficiently hydrolyze hemicellulosic portions 20,135. Conversion of both cellulosic and hemicellulosic hydrolytic products in a single process can be achieved by co-culturing two or more compatible microorganisms with the ability to utilize the materials. In fact, in nature, lignocellulosic residues are degraded by multiple co-existing lignocellulolytic microorganisms. Co-culturing of two or more fungal strains in mixed culture fermentation is widely used in many biological processes including the production of antibiotics, enzymes and fermented food 136. Mixed fungal cultures have many advantages compared to their monocultures, including improving productivity, adaptability and substrate utilization. Improving fungal cellulolytic activity of T. reesei and A. niger by co-culturing was the subject of extensive research including studies done by Maheshwari 137, Ahmad 138 and Juhász 139. Moreover, other fungal strains have been co-cultured to obtain better cellulolytic activity such as co-culturing of T. reesei RUT-C30 and A. phoenicis 140 or A. ellipticus and A. fumigatus (Table 5) 141. There are a few examples of co-culturing fungal strains for the purpose of combining cellulose and hemicellulose hydrolysis such as co-culturing T. reesei D1-6 and A. wentii Pt 2804 in a mixed submerged culture 142 or co-culturing T. reesei LM-UC4 and A. phoenicis QM329 using ammonia-treated bagasse 143. In the both cases, enzyme activity for cellulases and hemicellulases was significantly increased. The main drawback of co-culturing however is the complexity of growing multiple microorganisms in the same culture 144.

Alternatively to co-culture, microorganisms can be metabolically engineered, which enable one microorganism to complete an entire task from beginning to end. This can be done by altering metabolic flux by blocking undesirable pathway(s) and/or enhancement of desirable pathway(s). For example by application of homologous recombination, the production of T. reesei β-glucosidase I was enhanced using xylanase (xyn3) and cellulase (egl3) promoters which improved β-glucosidase activity to 4.0 and 7.5 fold compared to the parent, respectively. This will permit one fungal strain such as T. reesei to be more efficient on hydrolysis of cellulose to glucose which improve the yield and therefore lower the cost 70.

Heterologous expression is a powerful technique to improve production yield of enzymes, as well as activity. In order to make a robust lignocellulolytic fungal strain, many different fungal cellulases with higher and/or specific activity based on the need for a functional cellulase system in the organism have been cloned and expressed. For example, thermostable β-glucosidase (cel3a) from thermophilic fungus T. emersonii was expressed in T. reesei RUT-C30 using a strong T. reesei cbh1 promoter. The expressed enzyme has been shown to be highly thermostable (optimum temperature at 71.5 °C) with high specific activity 69. In the study for the improvement of biofinishing of cotton, T. reesei cellobiohydrolase (I & II) were overexpressed using additional copy(s) of the genes cloned under T. reesei cbh1 promoter. The results have shown that the expression of CBHI was increased to 1.3- and 1.5-fold with one or two additional copies of the gene, respectively. In the case of CBHII, however, the expression was increased to 3- to 4-fold using just one additional copy of the gene 145. In addition, chimeric proteins with specific applications have been designed using recombinant DNA technology. For example, an endoglucanase from Acidothermus cellulolyticus was fused to T. reesei cellobiohydrolase and expressed in T. reesei. This bi-functional endo- & exo-acting cellulolytic enzyme has been shown to improve saccharification yields 146. Moreover, the structural and biochemical information obtained from family GH 12 homologues was used to create a wide range of H. jecorina Cel12 A variants which were heterologously expressed as secreted proteins in A. niger displaying temperature stability changes ranging from none to an increase of 3.9 °C (the most stable variant, P201C) (Table 5) 36.