Meta-analysis: effects of exogenous fibrolytic enzymes in ruminant diets

ABSTRACT There are unknown interactions between supplements of exogenous fibrolytic enzymes (EFE) and the cell walls of feedstuff in ruminal conditions. The quantitative effects of using EFE in ruminant diets were evaluated using meta-analysis. Records (586) were extracted from 74 journal articles from a list of published papers (2000–2012). Statistical analyses were performed considering fixed [type of forage-based diet, forage-to-concentrate ratio (F:C ratio) and primarily enzyme activities in the EFE], and random effects [Experiment(Article)]. In dairy cows fed high-forage (F:C ≥50%), the supplementation of primarily mixtures of cellulases and xylanases (Cel:Xyl: 1:4–1:1) increased milk production and milk composition of legume-based diets, and primarily xylanases (Xyl) EFE improved those variables of grass-based diets. In F:C <50% grass-based diets, Cel:Xyl improved the average daily gain (ADG) and feed conversion [FC:DM intake (DMI)/ADG] of beef cattle. DMI of dairy cows was not affected by EFE supplementation, but EFE improved the DMI of beef cattle. EFE effects were inconsistent in sheep productive performance variables. Cellulases (Cel) and Xyl enhanced in vivo dry matter (DM) digestibility (DMD) in low-forage (F:C <50%) grass-based diets. In F:C ≥50% legume-based diets, EFE enhanced the in situ DM disappearance (ISDMD), and mainly Cel:Xyl improved the in situ neutral detergent fibre (NDF) disappearance (ISNDFD), but there were no effects in those variables in F:C ≥50% grass-based diets. Regardless of the type of ruminal liquid (RL) or forage, in F:C ≥50% diets, in vitro DM degradability (IVDMD) was improved mainly by Cel, but fibre degradability only was improved by Cel:Xyl when sheep RL was used for in vitro evaluations. Overall, EFE could improve the productive performance of dairy cows and beef cattle, but the response depends upon the proper mixture of Cel and Xyl according to the diet composition. Abbreviations: ADF: acid detergent fiber; ADG: average daily gain; A:P: acetate:propionate ratio; BW: initial body weight; Cel:Xyl: cellulases:xylanases; DM: dry matter; DMD: in vivo dry matter digestibility; DMI: dry matter intake; EA: enzyme activities; EFE: exogenous fibrolytic enzymes; F: type of forage; FC: feed conversion; F:C: forage-to-concentrate ratio; ISDMD: in situ dry matter disappearance; ISNDFD: in situ neutral detergent fiber disappearance; IVADFD: in vitro acid detergent degradability; IVDMD: in vitro dry matter degradability; IVNDFD: in vitro neutral detergent fiber degradability; NDF: neutral detergent fiber; VFA: in vitro volatile fat acids


Introduction
Supplementation with exogenous fibrolytic enzymes (EFE) is thought to enhance ruminal fermentation and to increase the degradability of forage cell walls, potentially reducing feed costs and sustaining the productive performance of ruminants; however, the underlying interactions are unknown and the effects of using EFE are highly variable.
Within the rumen, EFE hydrolyse certain components of the cell wall and produce substrates that favour selected populations of microorganisms, even with low-forage diets (Beauchemin, Colombatto, Morgavi, Yang, and Rode 2004;Bedford and Cowieson 2012). In the ruminal environment, EFE can affect bacterial attachment and colonization, and microbial populations (Colombatto and Mould et al. 2003;De Souza et al. 2008;Wang et al. 2012), affecting in vitro Ranilla et al. 2008;Srinivas et al. 2008), in situ (Tirado-Estrada et al. 2015) and in vivo NDF digestibility.
Some models have suggested that the increase in in vitro digestibility of forage NDF is associated with a higher production (Oba and Allen 1999;Jung et al. 2004), which could reduce feed costs Allen 2000a, 2000b;Oba and Allen 2005;Staton et al. 2007).
To improve the consistency of the results, it is important to know the interaction within the ruminal ecosystem, the cell walls of plants and the type of EFE Meale et al. 2014). Previously, the authors have considered factors such as Cel:Xyl ratio Eun and Beauchemin 2008b), and dose (Dean et al. 2008), stability and structure of the enzymes in the EFE preparations (Morgavi et al. 2000) along with their interactions with the different types of forages (Jililvand et al. 2008). However, despite the studies and reviews of the effects of EFE supplementation, some conclusions remain unclear Beauchemin, Colombatto, Morgavi, Yang, and Rode 2004;Bedford and Cowieson 2012;Tirado-González et al. 2015).
Some interfering or confounding factors Bala et al. 2009) can be evaluated using a meta-analysis of different studies (Eugéne et al. 2004(Eugéne et al. , 2008Khiaosa and Zebeli 2013) in order to determine the effect of EFE supplementation and the optimal situations (Desnoyers et al. 2009).
Using meta-analysis, the aim of the present study was to quantify the effects of using EFE in ruminant diets with varying proportions of legumes and grasses on in vitro, in situ and in vivo digestibility, and the productive performance of lactating dairy cows, sheep, and growing beef cattle.

Sampling method
A list of articles published between 2000 and 2012 was generated as follows: five combinations of words were introduced as the search criteria in the CabAbstracts browser: (1) 'fibrolytic' and 'enzymes'; (2) 'cellulases', 'xylanases' and 'ruminant'; (3) 'exogenous', 'fibrolytic', 'enzymes' and 'ruminant'; (4) 'exogenous', 'fibrolytic' and 'enzymes'; and (5) 'exogenous', 'enzymes' and 'ruminant'. The search procedure identified 226 articles after a list of non-recurring items was generated. These articles were published in the Journal of Dairy Science (48), Journal of Animal Science (44), Canadian Journal of Animal Science (19), Asian-Australasian Journal of Animal Science (14), Animal Feed Science and Technology (8) and other journals (93). The final analysis was performed with 74 articles chosen from the random list after eliminating those articles which were not performed under the set conditions, or did not analyse variables or factors included in the present study experiments (see Appendices 1 and 2).

Evaluated variables
From the articles considered, the averages for numerous variables were extracted and entered into a database (see Appendices 1 and 2). For in vitro studies (24 and 48 h incubations), the variables were: dry matter (DM), NDF, acid detergent fibre (ADF) degradability (IVDMD, IVNDFD and IVADFD, respectively), total VFA, propionate proportion (Gurbuz et al. 2008) and A:P ratio. For in situ studies, the variables were: disappearance after 24, 36 and 48 h incubation times for DM and NDF (ISDMD and ISNDFD). For in vivo studies, the variables were: initial body weight (BW), DMI, feed conversion [FC (DMI/ ADG)], ADG, and milk production and composition, dry matter and neutral detergent fibre digestibility (DMD and NDFD). Prior to tabulation in the database, all data were transformed into similar units of measurements to allow direct analysis within certain parameters.

Experiment definition
According to Desnoyers et al. (2009), papers included mean treatments which were individually coded. Each article contained two or more treatments (control: without EFE supplementation; treatment means: with EFE supplementation), defined as experiments.

Classification of the experiments
In situ, in vivo or in vitro experiments were classified according to: (1) type of study: in situ, in vitro or in vivo; (2) animal species used in the experiment: sheep, lactating dairy cows and beef cattle; (3) primary forage in the diet: grasses or legumes; (4) dietary forage-to-concentrate ratio (F:C): <50% or ≥0%; (5) type of enzyme product (Promote, Roxazyme G2, Zado, Econase, Naturazyme, Liquicell, Novozyme, GNC, Maxicel, Nutreco, Cattle-Ase-P, etc.); (6) primary supplemented enzyme activity (EA): no enzyme (control), Cel, Xyl or a combination of both (Cel:Xyl ratio from 1:1 to 1:4); and (7) application time of EFE to the feed or diet (added to feed in a liquid form as a pre-treatment or provided as a powder in the concentrate or mixed with the diet): <1 h, 1-24 h, 25 h-10 d, >10 d prior to evaluation.

Inclusion criteria for experiments
Experiments carried out an insufficient number of times (with minimal experimental representation), or where EFE effects could not be separated from other effects were not included in the analyses. The final analysis included 74 and 586 articles: 160 in vivo studies (94 EFE treatments/66 controls), 120 in situ studies (66 EFE treatments/54 controls) and 306 in vitro studies (206 EFE treatments/100 controls) (see Appendices 1 and 2).

Categorizing of factors
To categorize the included factors and to find a general model, multiple linear regression analysis was performed (stepwise) using the Proc REG module of SAS statistical software v. 9.4 (SAS 2013). Data were subdivided to generate subgroups of studies carried out under the same conditions but with varying levels of the following factors: (1) type of study: in situ, in vivo and in vitro; (2) type of forage: legume or grass; (3) F: C: <50% or ≥50% and (4) primary type of supplemented EA: cellulases (Cel), cellulases:xylanases proportion (Cel:Xyl: 1:4 to 1:1), xylanases (Xyl) or without enzymes (control).

Statistical analysis
Statistical analyses were performed using SAS statistical software (SAS 2013). Normal distribution of the information for all variables was verified using the Shapiro-Wilk, Kolmogorov Smirnov, Cramer Von Mises and Anderson Darling tests, using the Univariate procedure. Experiments were carried out using completely randomized statistical designs and factorial treatment arrangements, considering fixed effects and the random effects of the number of records within the paper in which they were originally reported [Record (Paper)], according to models (1 and 2) (number of replicates are indicated in tables). Significant values for the model fixed effects were obtained using the GLIMMIX procedure, the weight factor was the number of experiments per article minus one. Correct standard errors were obtained from the adjusted means (LsMeans/ pdiff) using the GLIMMIX procedure. The GLM procedure was used to obtain the coefficients of variation and of determination. Least squares means were estimated (LSMEANS) and reported.

Productive animal performance
where Y is the average daily gain, dry matter intake, feed conversion (ADG, DMI, FC) milk protein, milk fat; μ is the general mean; [Exp (Art)] i( j) is the random effect of the ith experiment within the jth article; EFE k is the effect of the kth exogenous fibrolytic enzyme (EFE); F l is the effect of the lth type of forage; FC m is the effect of the mth F:C ratio; (FE × F) kl is the interaction between the kth EFE by the lth type of forage; (FE × FC) km is the interaction between the kth EFE by the mth F:C ratio; (F × FC) lm is the interaction between the lth type of forage by the mth F:C ratio; (FE × F × FC) klm is the interaction among the kth EFE, the lth type of forage and the mth F:C ratio; β1(x−x i ) is the effect of the covariate (initial weight, days in milk production, or the initial NDF in experimental diets); and E ijklm is the experimental error.

In situ and in vivo DM and NDF digestibility
where Y is the in vivo dry matter digestibility (DMD), in vivo neutral detergent fibre digestibility (NDFD), in situ dry matter disappearance (ISDMD), in situ neutral detergent fibre disappearance (ISNDFD); μ is the general mean; [Exp(Art)] i( j) = random effect of the ith experiment within the jth article; EEF k = effect of the kth EFE; RF l is the effect of the lth type of ruminal fluid; (EFE × RF) kl is the interaction between the kth EFE by the lth type of ruminal fluid; and E ijklm is the experimental error.

Effect of supplementation with EFE on animal performance
3.1.1. Dairy cows Table 1 shows that in low-forage diets (F:C <50%), EFE supplementation did not have positive effects on milk production and milk solid contents. EFE applied to high-forage diets (F:C ratio ≥50%) had positive effects on milk production, milk protein (1.96 kg/d and 99.44 g/d, increases, respectively) (P = .06) and milk fat (83 g/d) (P = .015). EFE negatively affected the DMI in all diets (P = .003). However, correlation coefficients between DMI and the production of milk, fat and protein were r 2 = 0.58, r 2 = 0.61 and r 2 = 0.60, respectively (P < .0001) (data not presented in Table 1).  Table 2 presents data from experiments performed using diets at high-and low-forage concentration and based on grasses. The ADG increased by 0.30 kg/d (P = 0.01) in animals fed diets containing <50% grasses and treated with EFE. In low-forage diets (F:C <50%), EFE also had positive effects on feed conversion (FC) (control = 5.8 vs. EFE treatment = 5.6) (P = .02).

Beef cattle
Although improvements were seen in DMI when EFE were included in diets containing ≥50% grasses (0.9 kg/d) (P = .03), and the simple linear regression (data not presented in tables) between DMI and ADG was r 2 = 0.49 (P < .0001), there were no positive effects on ADG and FC when EFE was used as a supplement of high-forage diets (F:C ≥50%).

Sheep
EFE supplementation did not have significantly positive effects on ADG and FC in both low-and high-forage diets (F:C <50% and F:C ≥50%) (P > .2) ( Table 2). DMI was not affected by EFE supplementation (P > .77), but the simple linear regression between DMI and ADG was r 2 = 0.38 (P < .0001) (data not presented in tables).

Effect of type of enzyme activity
3.2.1. Effect of using EFE with different enzymatic compositions on animal productive performance Table 3 shows the effect of the addition of different EFE products in diets.
In low-forage diets (F:C < 50%) (Table 3) based on grasses, Xyl and Cel:Xyl EFE positively affected the beef cattle productive performance (P < .02). EFE primarily Xyl improved the ADG (50.9 g/d) and FC (0.5 units), but increasing amounts of Cel of EFE using a Cel:Xyl product had better results on the ADG (114 g/d) and FC (0.9 units) than control and Xyl treatments. All types of EFE products negatively affected the DMI of beef cattle (200-400 g/d) (P = .02).
A trend to improve the ADG in sheep (14.7 g/d) was observed when low-forage diets (F:C <50%) based on grasses were supplemented with EFE primarily Cel (P = .096). However, the use of EFE with primarily Xyl activities slightly decreased the ADG (−14.3 g/d) (P = .096), and both Cel and Xyl EFE products negatively affected the sheep FC (−0.6 to −0.8 units) (P = .02). DMI was not affected by adding EFE products to low-forage diets (P = .74). Table 4 shows the improvements of in vivo DMD and NDFD following EFE supplementation of different diets composition. Among the rumen fluid sources, the addition of primarily Xyl and Cel EFE products to low-forage (F:C <50%) grass-based diets improved the DMD (8 and 30 g/kg DM) (P = .02). NDFD was not affected by the addition of EFE treatments (P = .86).

Effect of using EFEs with different enzymatic compositions on in vivo and in situ digestibility
In high-forage diets (F:C ≥50%), using Xyl EFE products in legume-based diets enhanced in situ DMD (25 g/kg DM) (P = .0005), but Cel:Xyl EFE supplementation primarily improved the in situ NDFD (154.2 g/kg DM) (P = .008). Despite the species of ruminant, EFE in grass-based diets did not affect either ISDMD or ISNDFD (P > .25).   Table 5 presents in vitro evaluations of high-forage diets (F:C ≥50%) supplemented with EFE. Cel enzyme activities primarily supplemented either legume-or grass-based diets, increased the IVDMD (P < .0001) regardless of the type of ruminal fluid used during in vitro evaluation (P > .19). Among the rumen fluid sources and diets, the addition of EFE composed primarily of Cel improved the IVDMD by an average of 90 ± 30.1 g/kg DM. Adding primarily Cel:Xyl also enhanced the DMD of grassbased diets according in vitro experiments performed with sheep ruminal fluid (80.7 g/kg DM); however, this improvement was not consistent with other rumen fluids.
The A:P ratio was not affected when EFE supplemented diets composed primarily of grasses (P > .25), there were also negative or null effects related to EFE use on in vitro VFA. EFE treatments comprised primarily of Xyl and Cel had less VFA than control (P = 0.0001) in ruminal fluid from dairy cows (-9.6 ± 8.6 mM/100 mM), beef cattle (-11 ± 28.6 mM/100 mM) and sheep (-2.5 ± 5.5 mM/100 mM).

Effect of supplementation with EFE on animal performance
We analysed the productive performance separately in dairy cows, beef cattle and sheep fed with four different diets: <50% and ≥50% of forage content in grass-or legume-based diets. The proportions and populations of ruminal bacteria vary according to the type of diet (Petri et al. 2013;Zhao et al. 2014), stage of production (Li et al. 2012) and type of ruminant (Lee et al. 2012).

Dairy cows
EFE supplementation increased the production of milk and milk solids with high-forage-based diets. These results agree with studies published subsequently to those included in our meta-analysis (Kholif and Aziz 2014). Mohamed et al. (2013) found that the use of a preparation with primarily Xyl activity increased milk production by 1.5 kg/d (3.8%). It might be possible to increase the amount of forage in the diets of dairy cows if it is treated with fibrolytic enzymes. The use of EFEs as a supplement for ruminants could increase the digestible energy of high-fibre and forage-based diets and reduce the amount of feed required per unit of milk or live weight (Meale et al. 2014).
Maximizing the amount of forage included in the diet not only might reduce feed costs (Oba and Allen 2005;Mendoza et al. 2014), but might also have beneficial effects on the health and welfare of dairy cows. For example, including more forage in diets increases the diversity of rumen microorganisms up to 3.45 times, and reduces the potential for ruminal acidosis caused by Acetitomacullum, Lactobacillus, Prevotella and Streptococcus (Petri et al. 2013). Higher forage diets also reduce the occurrence of abnormal metabolites in the rumen (Saleem et al. 2012;.
In beef cattle experiments, DMI improvements could be related to how enzymes act to increase the availability of fibre (Bradford and Allen 2004). Using EFE as supplement in highgrain diets can contribute to breaking seed hulls which contain an average of 96% cellulose; EFE acts not only through the direct hydrolysis of cellulose and xylose links, but also by changing the structure of cell walls and the successive populations of microorganisms (Yu et al. 2005;Wang et al. 2012;Vyver and Cruywagen 2013).

Sheep
Although in the present analysis we found some negative and inconsistent effects because of using EFE in diets for sheep, some studies have demonstrated improvements in ADG and FC following EFE supplementation related to an increase in in vivo DMD and increases in the proportions of propionic acid and VFA Miller, Granzin, Elliot and Norton 2008;Gado et al. 2011). Tirado-Estrada et al. (2011 found similar results when using EFE in diets with more than 56% corn stover.

Effect of supplementation with EFE on DMI
The results of present meta-analysis suggest that, although EFE had negative of null effects on DMI of dairy cows and sheep, DMI was positively correlated with the production of milk, fat and protein, and sheep ADG (r 2 = 0.58, r 2 = 0.61 and r 2 = 0.60, and r 2 = 0.39, respectively). In high-grain diets (F:C ≥50%), EFE supplementation improved the beef cattle DMI, which was also positively correlated with ADG (r 2 = 0.48). The correct selection of the type and application of EFE could contribute to improve DMI by enhancing DM and NDF digestibility. Improving in vitro NDF digestibility of feedstuff should be reflected in the increment of passage rate, dry matter intake and yield of fat corrected milk Allen 1999, 2005;Jung et al. 2004).
EFE effects on DMD and NDFD depend upon the enzymesubstrate specificity. Before, authors have considered the DMD improvement (Faramarzi-Garmroodi et al. 2013) or positive changes in patterns of fermentation (Yang et al. 2011) to make the selection of EFE products; nevertheless, some authors have hypothesized that NDFD and ADFD improvements should be considered in a proper EFE selection (Phakachoed et al. 2013). Independently from VFA, DMD or NDFD improvements, certain types of enzyme preparations change the structure of cell walls of some forages (Vyver and Cruywagen 2013), enhance colonization of the substrate by bacteria (De Souza et al. 2008;Mao et al. 2013) and promote the activity of certain endogenous enzymes (Colombatto and Mould et al. 2003;De Souza, Figueiredo and Berchielli et al. 2006), which could change the passage rate (De Souza, Figueiredo and Teresinha et al. 2006) and therefore the DMI (Balci et al. 2007;Arriola et al. 2011). 4.3. Effect of type of enzyme activity 4.3.1. Effect of cellulases:xylanases proportion on dry matter and neutral detergent fibre digestibility Multi-enzyme cocktails may work better than extracts of almost pure enzymes (Yu et al. 2005); the correct mixture of xylanases and cellulases improves glucose release, first by removing the xylose , which increases the accessibility to cellulose (Grabber et al. 2002).
Although the excess of Cel in EFE could limit the access of the enzyme to the hydrolysable portion of carbohydrates and reduce microbial adherence (Morgavi et al. 2001;Wang et al. 2004), previous studies suggest that the deficiency of Cel activities is the main limitation for the release of reducing sugars, and the excess of Xyl activities could also adversely affect the ruminal microorganism population Eun and Beauchemin 2008a).
It seems that there is a relationship between the activity of certain types of cellulases (endoglucanases, exoglucanases and β-glucosidases) or the combination of Cel:Xyl, and the improvement of alfalfa or corn silage IVNDFD, according to assays with mixtures of cellulases:xylanases (ranging from 1:10 to 1.5:1) Eun and Beauchemin 2008b).
In many cases, the products evaluated for ruminant feed applications do not contain the appropriate mixture of enzymes, which compromises the consistency of the results. However, supplementing the correct dose and EFE product depends on the type of diet (Tirado-González et al. 2015).
For example,  analysed the effect on in vitro DM and NDF degradability of enzyme activities of extracts from T. longibrachiatum, regression models showed that the units of enzyme activities of Cel were linearly associated with alfalfa IVNDFD (r=0.26), and the improvement of corn silage IVNDFD (r=0.72).  also performed an in vitro study with 22 EFE products; using multiple linear regression modelling, they found that Xyl activities were positively correlated with an improvement in alfalfa hay IVNDFD and negatively correlated with an improvement in corn silage IVNDFD.
Cel:Xyl balance in EFE should be different for supplementing grass or legume-based diets; thus, optimal doses would vary according to EFE composition (Fortes et al. 2010). EFE for legume-based diets could contain more Xyl than EFE for grass-based diets. Yang et al. (2011) evaluated 26 enzyme products in alfalfa forage; after the first selection, they choose EFE including 1:7.4 and 1:2.3 endoglucanase:xylanase ratios; EFE with a 1:7.4 endoglucanses:xylanases ratio increased the IVNDFD and IVADFD of alfalfa hay, while a ratio of 1:2.3 did not affect the digestibility. Moreover, Yu et al. (2005) reported a multiple linear regression model (r 2 = 0.74) that related the IVDMD of oat hulls with the activities of Xyl, two types of endoglucanases, ferulic acid esterases, β-glucosidases and Cel, but 55% of the DMD was explained by the activity of Cel and β-glucosidases of an enzymatic product with 1024 units of Cel and 4096 units of Xyl (Cel:Xyl = 0.25:1).
The present meta-analysis shows some positive results on animal productive performance, and in vivo, in situ and in vitro DM digestibility, mainly due to adding primarily Cel EFE. Although generally there were negative or null effects of EFE on in vivo, in situ and in vitro NDF and ADF digestibility.  observed that the optimum Cel:Xyl ratio for improving alfalfa hay and corn silage IVDMD ranged from 1:4 to 1:2, while Cel:Xyl mixtures ranging from 1:10 to 1:1.6 increased NDF degradability in tests including ruminal fluid , suggesting that mixtures of Cel:Xyl affect in a different manner the fibre and total DM degradability.

Conclusions
The use of fibrolytic enzymes could allow increments in the amount of forage in the diets of dairy cows without compromising their productive performance, EFE treatments applied in high-forage-based diets increased milk production, and its protein and fat contents by an average of 1.96, 0.99 and 0.83 kg/d, respectively. For beef cattle experiments, EFE supplementation to low-forage diets increased the ADG by 0.30 kg/d. However, overall, the effects of adding EFE to diets offered to sheep were inconsistent. EFE supplementation did not affect DMI in dairy cows, but EFE improved the DMI in beef cattle. However, DMI was positively correlated with the production of milk, fat and protein, and sheep and cattle ADG (r 2 = 0.58, r 2 = 0.61 and r 2 = 0.60, and r 2 = 0.39 and r 2 = 0.48, respectively). In high-forage legume-based diets (F:C ≥50%), primarily Cel:Xyl (from 1:4 to 1:1) increased milk production and its fat and protein content (2.3, 0.118 and 0.083 kg/d, respectively), and Xyl EFE had similar results in cows fed with F:C ≥50% grass-based diets (3.10, 0.11 and 0.13 kg/d, respectively). In low-forage diets (F:C <50%) based on grasses, Cel:Xyl EFE supplementation had the best results on the ADG (114 g/ d) and FC (0.9 units) of beef cattle, and Cel EFE supplementation tended to improve the ADG of sheep (14.7 g/d). EFE could improve sheep DMI but Xyl negatively affected FC of sheep fed high-grain (F:C <50%) diets. Primarily Xyl and Cel activities in low-forage (F:C <50%) grass-based diets improved the DMD (8 and 30 g/kg DM), but NDFD was not affected by the addition of EFE treatments. In high-forage diets (F:C ≥50%), Xyl enzyme activities on legume-based diets enhanced ISDMD (25 g/kg DM), but Cel:Xyl supplementation improved ISNDFD (154.2 g/kg DM); however, in grass-based diets, ISDMD and ISNDFD were not affected by the addition of EFE. Among different types of ruminal fluid, IVDMD was improved by the addition of primarily Cel EFE in either legume-or grass-based diets (average, 90 ± 30.1 g/kg DM) (F:C ≥0%). Regardless of the diet base, the proportions of Cel, Xyl and Cel:Xyl of EFE composition considered in this meta-analysis had negative or null effects on IVNDFD, IVADFD, A:P ratio and total VFA in in vitro studies carried out with dairy cows or beef cattle ruminal liquid (RL). However, fibre in vitro degradability was improved by using Cel and Cel:Xyl in evaluations with sheep RL. However, fibre digestibility could be enhanced by supplementing the correct Cel:Xyl balance. The present meta-analysis suggests that EFE supplementation could improve dairy cows, beef cattle productive performance, DMD, ISDMD and IVDMD; however, the consistency of the results depends on the proper selection of Cel:Xyl balance in enzymatic products, according to the diet composition.

Appendices
Appendix 1. References used for meta-analysis.

N
Refs. N Refs. N Refs. the productive performance of lactating dairy cows, sheep and growing beef cattle.