Meta-analysis of the relationship between dietary starch intake and enteric methane emissions in cattle from in vivo experiments

Objective: Different sources and levels of starch in the total mixed ration might result in different rumen fermentation profiles, rumen microbial composition, and enteric methane production. The current study aimed to evaluate the effect of dietary starch intake on enteric methane production through a meta-analytical approach by integrating related published studies. Materials and Methods: Papers that provided study results on enteric methane production from cattle fed different sources and levels of starch were selected. A total of 52 publications were filtered based on some specified criteria, comprised of 73 studies focused on enteric methane production in cattle fed a basal diet supplemented with starch. The collected data were subjected to statistical meta-analysis through a mixed methodology model. The p-value and root mean square error (RMSE) were applied as the statistical models. Results: Results showed that increasing the level of dietary starch intake and its digestibility in the rumen as well as increasing propionate significantly decreased the acetate to propionate (A/P) ratio as well as enteric methane production (p < 0.01). Conclusions: It is concluded that the addition of starch to diets mitigates enteric methane production from cattle, primarily through two ways, i.e., manipulation of the protozoa population and alteration to a lower A/P ratio.


Introduction
The FAO data report for 2020 shows that enteric fermentation emissions contribute 17.50% of total gas emissions in agrifood systems [1], and 71.67% of enteric fermentation emissions came from cattle [2].Enteric fermentation emissions, as a product of methanogen activities in the rumen, have an unfavorable impact on the cattle and the environment.This is because the methane produced during methanogenesis in the rumen reflects the amount of dietary energy consumed by animal losses during the rumen microbial fermentation process [3].The lower the methane produced during methanogenesis, the more dietary energy is used for animal production [4].This lost energy will be released in the form of methane (CH 4 ) into the environment, thereby worsening global warming.
Many approaches have been widely used to mitigate enteric methane production from ruminants and are classified into several groups.The first approach is rumen fermentation modification using various types of materials, such as bacteria [5,6], chemicals [7,8], oil [9], and plant active ingredients, namely tannins and saponins [10][11][12][13][14], that are added to animal feed as a feed additive.The second approach is high-quality feed sources for low-quality diet supplementation, such as high-protein or starch feed [15,16] and macroalgae supplementation [17][18][19].The last approach is a nutrient balance in the complete ratio to ensure high feed efficiency that improves animal productivity [20].Evaluations through meta-analysis of mitigation approaches using a modification of enteric methane production have been done previously [21,22].
One of the approaches used to mitigate enteric methane is starch supplementation in feed.Starch is commonly used as a source of energy in ruminant feed.Also, it improves the utilization of structural carbohydrates, which then increases the amount of protein flow to the small intestine [23].Sources of starch mainly used in ruminant feeds are cereal grains, legumes, and tubers.Cereal grains such as corn, wheat, oats, and barley are primarily starch sources due to their high starch contents (60%-80%).Tubers like cassava contain around 60%-90% of starch, while legume beans contain 25%-50% of starch [24].Grains as a source of feed energy have commonly become the main component in concentrate, improving fermentation kinetics, improving microbial protein supply flow into the small intestine, and decreasing methane formation in the rumen [25,26].A high-starch diet in the rumen changes rumen bacteria composition by promoting propionic acid bacteria growth over methanogens [27].Feeding ruminant animals with a high-starch diet depresses methanogen growth due to the lower rumen pH as a result of the rapid digestion process of starch diets [25,28].
A similar meta-analysis of starch used as feed in ruminants has been published previously by Moharrery et al. [29].This paper discusses the relationship between starch intake used as feed for ruminants and its digestibility in the rumen, small intestine, and hindgut of dairy cows.Meanwhile, this study observes the relation of starch intake to feed digestibility, short-chain fatty acid profiles, rumen microorganism population, and enteric methane production.Furthermore, the present study aims to analyze the relationship between animal productivity and feed modification using starch to reduce enteric methane production through the meta-analysis method.

Ethical approval
This is a meta-analysis study, and no live animals were used in this study, so ethical approval is not necessary for this type of study.

Search strategy
A database was developed from publishing papers mentioning the role of high starch content in diets to reduce enteric methane yield.A total of 302 publications were used in developing the database.The papers collected consisted of 200 publications found in Google Scholar using the keyword "Reducing enteric methane production using high starch contents on feed;" 40 publications found in Science Direct using the keyword "High starch feed to mitigate methane production," and 76 publications found in Scopus using the keywords "starch" and "methane" and "cattle" or "cow."This process was shown in the preferred reporting items for systematic reviews and meta-analysis (PRISMA) flow chart (Fig. 1).

Selection criteria
All publications were selected based on the following criteria: 1) the article was published in English; 2) the year of publication was limited from 2010 to the newest; 3) the kinds of experiments were in vivo; 4) the animal used in the experiment was cattle; and 5) enteric methane production was not estimated using an equation but measured directly using the green feed, respiratory chamber, and SF6 systems.Some experiments measuring the effect of feeding starch on enteric methane emissions were eliminated from the list of publications collected since the animals used in the experiment were sheep, goats, and buffalo.

Database development
A total of 52 publications were selected based on the above criteria, consisting of 73 studies that focused on the cattle chosen (Fig. 1).When the experiment consisted of more than one study, each respective study was encoded separately.The number of cattle used in each experiment reported in those publications started from three heads and up to 28 heads, with a mean of 7,8 cows in each experiment.Variables used and integrated into the database developed were methane production, dry matter (DM) intake, starch intake, volatile fatty acid (VFA) total and partial production, bacteria, and protozoa population, DM, neutral detergent fiber (NDF), and acid detergent fiber (ADF) digestibility.The dietary starch as the main parameter used in the analyses was selected based on the type of sources.They are mixed grain or concentrate, corn meal or grain as a single feedstuff, wheat, barley, rapeseed meal, and soybean meal.
The level of starch supplement calculated in the database was expressed in grams per kg of DM.When the amount of starch given to the animals in the experiments was expressed in different units, they were then calculated to make the same unit (gm/kg DM).The level of starch supplemented with the daily rations was variable, ranging from 5 to 603.23 gm/kg DM.The population of bacteria and protozoa variables were logarithmized to allow a linear relationship with dependent variables.The units of bacteria and protozoa population were expressed as log 10 cells/gm DM rumen content and log 10 cells/ml rumen fluid.
There was also variation in the enteric methane production measurement period among the different systems.The methane production was measured from one to seven days into the period when the respiratory chamber was used.Measurements were conducted over a period of 4-112 days when greenfeed units were used.When the SF6 method was used in the experiment, the measurements were taken for 4-15 days.Once we evaluated all the selected publications to meet the specified criteria, we listed the results in Table 1.

Statistical analysis
The data collected in the database was processed using statistical meta-analysis based on mixed model methodology.Different studies were classified as random effects; the intake of starch was classified as a fixed effect.This study uses a linear mixed model with a statistical model based on a p-value (<0.05).All statistical analyses were carried out using SAS University Edition.The following model was applied [82,83]: (1) where: Y ij = Dependent variable expected on study i and level j of X (independent variable), B 0 = The overall intercept across all studies (fixed effect), Data were analyzed to express: 1) the effect of feed-starch intake and metabolic BW on enteric methane production (gm/kg DMI) and short-chain fatty acids; 2) the correlation among feed digestibility, enteric methane production (gm/kg DMI), and rumen microorganism; and 3) the effect of VFA concentration on enteric methane production (gm/ kg DMI) and rumen microorganism.The model statistics presented are the p-value and root mean square error (RMSE).The data were then reported following PRISMA guidelines [84].

The effect of feed-starch intake and metabolic BW on enteric methane production and short-chain fatty acids
The linear regression generated based on data collected from in vivo studies between feed-starch and methane production is presented in Table 2.The feed starch was expressed as total intake (gm/head/day), intake per body weight (BW) gain (gm/BW 0.75 ), and starch contained in the feed (gm/kg DM intake).The results indicated that there   is a positive correlation between metabolism, BW, and enteric methane production.Increasing metabolic BW significantly increased enteric methane produced (p < 0.05).Meanwhile, results showed a negative correlation between starch intake, both as a total and per metabolic BW, and enteric methane production.Increasing starch content in the diet and starch intake significantly decreased enteric methane (p < 0.01).When starch intake was expressed per metabolic BW, there was a considerable decrease in enteric methane production.The enteric methane production was decreased by 0.05343 gm/kg DMI for every 1 gm starch intake per BW 0.75 .The results of linear regression to indicate the correlation between starch contained in the diet and starch intake with VFA concentration are presented in Table 3. Increasing starch content in the diet and consumption of starch, both total and per metabolic BW, significantly increased the concentration of propionate (p < 0.01).The opposite results were found on the correlation between starch intake and concentrations of acetate and butyrate, as well as a ratio of acetate to propionate (A/P).Increasing starch content in the diet and starch intake significantly decreased the A/P ratio (p < 0.01).

The correlation among feed digestibility, enteric methane production (gm/kg DMI), and rumen microorganisms
Analyzed results on the effect of DM, OM, NDF, and ADF digestibility on enteric methane production are presented in Table 4.There was a positive correlation between the digestibility of DM, OM, NDF, and ADF and enteric methane production.Increasing digestibility of DM, organic matter (OM) NDF, and ADF significantly increases the methane  production in the rumen (p < 0.01).A different result was shown on the correlation between starch digestibility and methane production in the rumen.There was a negative correlation between starch digestibility and enteric methane production.When the amount of starch digested in the rumen increases, the amount of methane produced in the rumen tends to decrease.The result of analyses the correlation between rumen microbial population and feed digestibility is shown in Table 5. Rumen bacteria and methanogen populations have a positive correlation with DM digestibility but have a negative correlation with OM, NDF, and ADF digestibility.However, based on the data collected and being meta-analyzed, the increase in the total population of rumen bacteria and methanogen did not significantly increase DM digestibility (p > 0.05).Moreover, those increases also did not significantly decrease NDF and ADF digestibility (p > 0.05) but decreased OM digestibility (p < 0.05).A different pattern was found for the correlation between protozoa population and digestibility.Increasing the rumen protozoa population significantly increased ADF and NDF digestibility (p < 0.05), while significantly decreasing DM, OM, and starch digestibility (p > 0.05).
Table 6 shows the results of the linear correlation between the rumen microbial population and the production of methane in the rumen.When rumen bacteria and methanogen were expressed based on DM content in the rumen, there was a negative correlation indicated between enteric methane production and the population of rumen bacteria and methanogen.Data showed that a decrease in enteric methane production occurred when the population of bacteria and methanogen bacteria increased (p < 0.05).A different result was identified in the correlation between the protozoa population and the methane produced in the rumen.When the population of protozoa was expressed based on ml of rumen fluid, increasing the population significantly increased the amount of methane produced in the rumen (p < 0.01).

The effect of VFA concentration on enteric methane production and rumen microorganisms
The linear correlation between VFA concentration and enteric methane production is presented in Table 7.An increase in total VFA production caused a rise in enteric methane production (p < 0.05).However, when it was expressed per VFA individual, such as acetate, propionate, butyrate, minor VFA, and A/P, the results showed different patterns.Increasing the concentration of acetate, butyrate, minor VFA, and ratio A/P raised enteric methane production significantly (p < 0.01); except for acetate, the increase was not significantly different (p > 0.05).A negative correlation was shown for propionate concentration.
Increasing propionate concentration resulted in a significant decline in enteric methane production (p < 0.01).
The results presented in Table 8 indicated a negative correlation between the four variables, namely the concentration of total VFA, acetate, butyrate, A/P ratio, and the rumen microbial population.On the other hand, the opposite results were found in the correlation between the propionate concentration, the minor VFA concentration, and the rumen microbial population.Increasing concentrations of total VFA, acetate, and butyrate tend to decrease the population of rumen bacteria and methanogen but tend to increase the population of rumen protozoa.However, the increase in acetate did not strongly increase the rumen protozoa population (p > 0.05).On the other hand, when the analyses were based on propionate, the increase in propionate concentration significantly increased the rumen bacteria and methanogen population but decreased the rumen protozoa population (p < 0.05).

Discussion
Regardless of the different sources of starch used and the variation in the level of starch offered to the cattle, the results show that dietary starch content on the basal diet and increasing starch intake by cattle increased propionate production but decreased production of acetate, butyrate, and the A/P ratio in the rumen (Table 3).The results also show that increasing starch intake reduced the formation of methane in the rumen (Table 2).Those findings were coherent with those reported by Bannink et al. [85] and Hristov et al. [86] that enteric methane production in the rumen is influenced by dietary factors.Daily intake in the form of starch produces more propionate during the fermentation process in the rumen.Increasing propionate production decreases the H available for methane production in the rumen.It is the reason enteric methane production decreases when using higher starch feed.A study by Hatew et al. [34] also suggested that increasing the intake of starch mitigates the amount of methane formed in the rumen.
Results of the current study also indicated that an increase in propionate concentration was followed by a decline in acetate, butyrate, and the A/P ratio and a decline in the amount of methane produced in the rumen (Tabel 7).This is because, during the formation of acetate in the rumen, H 2 was released and then became available in the rumen.Meanwhile, an amount of H 2 is required during the formation of propionate in the rumen.For each molecule of acetate formed, two H 2 molecules were also released.While four H 2 molecules were required to produce one molecule of propionate [20,87].This indicates that an increase in the amount of propionate formed will be followed by a decrease in the amount of acetate formed.On     the other hand, methanogenic archaea are other microbiota that live in the rumen and use an amount of H 2 to form enteric methane gas [88,89].
In line with the above results, methane produced in the rumen increased due to more DM, OM, NDF, and ADF digested in the rumen, as presented in the digestibility data shown in Table 4. Different results were demonstrated for a correlation between the amount of starch digested in the rumen and the methane produced.Increasing the amount of starch digested leads to a reduction in methane formed by Methanogen archaea.Since the digestion of starch in the rumen produces more propionate [90], less H 2 is available in the rumen.The proper availability of H 2 is required for methanogenesis by methanogen archaea [91].
Increasing the intake of starch or starch contents in the diet leads to an increase in propionate, which increases the bacteria and methanogen archaea but decreases the total protozoa in the rumen.The decrease in protozoa population when the starch contents in the diet increased was also caused by the high passage rate of feed in the rumen.It was noted that dietary fiber was used as a place to hide for protozoa, which were multiplying very slowly.Therefore, if fiber content is restricted in feed, then the number of washed-out protozoa becomes high [88].Some methanogens live in the rumen through association with protozoa; the range of association between protozoa and methanogen is about 0%-100% [92], so it is still possible that the number of methanogens increases and protozoa decreases with the inclusion of starch in the feed.Accordingly, protozoa provide hydrogen as a substrate for methanogenesis conducted by the methanogen archaea [93].Thus, a reduction in the protozoa population may lead to a decline in the population of methanogen archaea and, subsequently, reduce methane emissions as well.The fermentation end products that are produced by protozoa are acetate, butyrate, and H 2 .The H 2 is then used for methanogenesis by methanogen archaea to produce methane in the rumen [88].When the number of protozoa decreases, there is limited H 2 available in the rumen, resulting in a decrease in the amount of methane produced by methanogen [94].Meanwhile, the number of bacteria is increasing, caused by decreasing the number of protozoa that engulf bacteria besides the feed particles [88].
The correlation between protozoa and NDF digestibility was explained by Puniya et al. [88].Protozoa use the slow rate of passage of dietary fiber in the rumen as a place to hideout.Therefore, NDF digestibility will increase when the number of protozoa increases.
However, some datasets showed that starch inclusion in the feed does not affect enteric methane production.This is identified as a reason why the level of starch is not the only factor reducing enteric methane production.There may be different digestibilities of starch that can affect enteric methane production.Moharrery et al. [29] reported that total starch digestibility depended on the starch sources but did not affect starch intake.This study uses feed with a total starch digestibility higher than 90% (95.54% of the dataset), which results in increasing starch levels and decreasing enteric methane production.Therefore, a deeper observation of the starch digestibility effect on enteric methane production is needed, particularly for foods that have a total starch digestibility of less than 90% or come from different starch sources.
It should also be noted that what is included in the present meta-analysis study is starch, which comes from various sources.The degradation of starch in the rumen varied based on the type of starch and other nutrients contained in the feedstuff [95].This means that other nutrients present in the diet might have influenced the degradability of starch in the rumen.Sutaryono et al. [96] reported that the inclusion of Leucaena in corn stover silage affected DMD and OMD in the in vitro study.This is due to the fact that cellulose is more resistant to digestion compared to starches [97].Despite this, cellulose [98] and trace minerals [99,100] also affect feed digestibility.Similarly, the presence of other nutrients may also have different interactions with starch, resulting in variations in rumen microbial composition, VFA production, and methane generated in the rumen.This means that the presence of other nutrients cannot be neglected regarding their roles in the different results of the study presented.

Conclusion
This current meta-analysis study demonstrated that, based on many experiments with different sources and levels of starch supplemented to the basal diets of cattle, the addition of starch in the diets mitigates methane produced in the rumen.There are two ways dietary starch mitigates enteric methane production: first, through manipulation of the protozoa population or defaunation, and second, through changing the proportion of acetate and propionate as expressed as a low A/P ratio in the rumen.The two mechanisms decrease H available in the rumen, which is required for methanogenesis by methanogen archaea.

B 1 X
ij = The overall regression coefficient of Y on X ij (X on i study and j level), also called slope, b i X ij = The random effect of i study on the regression coefficient of Y on X ij , e ij = The unexplained residual error.

Figure 1 .
Figure 1.Process of paper selection according to PRISMA.

Table 1 .
In vivo experiments (Expt.)included in this meta-analysis.

Table 2 .
The effect of feed-starch intake and metabolic BW on enteric methane production (gm/kg DMI) in cattle.

Table 3 .
The effect of feed-starch intake on short-chain fatty acid profiles in the rumen.
n = Number of samples, SE = Standard error, RMSE = Root mean square error, A/P = Acetate/Propionate, BW = Body weight, BW 0.7 = Metabolic body weight

Table 4 .
The effect of nutrient digestibility on enteric methane production (gm/kg DMI).

Table 5 .
The effect of rumen microorganism population on nutrient digestibility.

Table 6 .
The effect of rumen microorganism population on enteric methane production (gm/kg DMI).DMI = dry matter intake, n = number of samples, SE = standard error, RMSE = root mean square error, DM = dry matter

Table 7 .
The effect of VFA profile on enteric methane production (gm/kg DMI) in cattle.

Table 8 .
The effect of VFA concentration on rumen microorganism population.