A meta-analysis of effects of seaweed and other bromoform containing feed ingredients on methane production, yield, and intensity in cattle

A meta-analysis of effects of seaweed and other bromoform containing feed ingredients on methane production, yield, and intensity in cattle | bioRxiv /* */ /* */ <!-- <!-- /*! * yepnope1.5.4 * (c) WTFPL, GPLv2 */ (function(a,b,c){function d(a){return"[object Function]"==o.call(a)}function e(a){return"string"==typeof a}function f(){}function g(a){return!a||"loaded"==a||"complete"==a||"uninitialized"==a}function h(){var a=p.shift();q=1,a?a.t?m(function(){("c"==a.t?B.injectCss:B.injectJs)(a.s,0,a.a,a.x,a.e,1)},0):(a(),h()):q=0}function i(a,c,d,e,f,i,j){function k(b){if(!o&&g(l.readyState)&&(u.r=o=1,!q&&h(),l.onload=l.onreadystatechange=null,b)){"img"!=a&&m(function(){t.removeChild(l)},50);for(var d in y[c])y[c].hasOwnProperty(d)&&y[c][d].onload()}}var j=j||B.errorTimeout,l=b.createElement(a),o=0,r=0,u={t:d,s:c,e:f,a:i,x:j};1===y[c]&&(r=1,y[c]=[]),"object"==a?l.data=c:(l.src=c,l.type=a),l.width=l.height="0",l.onerror=l.onload=l.onreadystatechange=function(){k.call(this,r)},p.splice(e,0,u),"img"!=a&&(r||2===y[c]?(t.insertBefore(l,s?null:n),m(k,j)):y[c].push(l))}function j(a,b,c,d,f){return q=0,b=b||"j",e(a)?i("c"==b?v:u,a,b,this.i++,c,d,f):(p.splice(this.i++,0,a),1==p.length&&h()),this}function k(){var a=B;return a.loader={load:j,i:0},a}var l=b.documentElement,m=a.setTimeout,n=b.getElementsByTagName("script")[0],o={}.toString,p=[],q=0,r="MozAppearance"in l.style,s=r&&!!b.createRange().compareNode,t=s?l:n.parentNode,l=a.opera&&"[object Opera]"==o.call(a.opera),l=!!b.attachEvent&&!l,u=r?"object":l?"script":"img",v=l?"script":u,w=Array.isArray||function(a){return"[object Array]"==o.call(a)},x=[],y={},z={timeout:function(a,b){return b.length&&(a.timeout=b[0]),a}},A,B;B=function(a){function b(a){var a=a.split("!"),b=x.length,c=a.pop(),d=a.length,c={url:c,origUrl:c,prefixes:a},e,f,g;for(f=0;f<d;f++)g=a[f].split("="),(e=z[g.shift()])&&(c=e(c,g));for(f=0;f<b;f++)c=x[f](c);return c}function g(a,e,f,g,h){var i=b(a),j=i.autoCallback;i.url.split(".").pop().split("?").shift(),i.bypass||(e&&(e=d(e)?e:e[a]||e[g]||e[a.split("/").pop().split("?")[0]]),i.instead?i.instead(a,e,f,g,h):(y[i.url]?i.noexec=!0:y[i.url]=1,f.load(i.url,i.forceCSS||!i.forceJS&&"css"==i.url.split(".").pop().split("?").shift()?"c":c,i.noexec,i.attrs,i.timeout),(d(e)||d(j))&&f.load(function(){k(),e&&e(i.origUrl,h,g),j&&j(i.origUrl,h,g),y[i.url]=2})))}function h(a,b){function c(a,c){if(a){if(e(a))c||(j=function(){var a=[].slice.call(arguments);k.apply(this,a),l()}),g(a,j,b,0,h);else if(Object(a)===a)for(n in m=function(){var b=0,c;for(c in a)a.hasOwnProperty(c)&&b++;return b}(),a)a.hasOwnProperty(n)&&(!c&&!--m&&(d(j)?j=function(){var a=[].slice.call(arguments);k.apply(this,a),l()}:j[n]=function(a){return function(){var b=[].slice.call(arguments);a&&a.apply(this,b),l()}}(k[n])),g(a[n],j,b,n,h))}else!c&&l()}var h=!!a.test,i=a.load||a.both,j=a.callback||f,k=j,l=a.complete||f,m,n;c(h?a.yep:a.nope,!!i),i&&c(i)}var i,j,l=this.yepnope.loader;if(e(a))g(a,0,l,0);else if(w(a))for(i=0;i (function(w,d,s,l,i){w[l]=w[l]||[];w[l].push({'gtm.start':new Date().getTime(),event:'gtm.js'});var f=d.getElementsByTagName(s)[0];var j=d.createElement(s);var dl=l!='dataLayer'?'&l='+l:'';j.src='//www.googletagmanager.com/gtm.js?id='+i+dl;j.type='text/javascript';j.async=true;f.parentNode.insertBefore(j,f);})(window,document,'script','dataLayer','GTM-M677548'); Skip to main content Home About Submit ALERTS / RSS Search for this keyword Advanced Search New Results A meta-analysis of effects of seaweed and other bromoform containing feed ingredients on methane production, yield, and intensity in cattle View ORCID Profile Ermias Kebreab , View ORCID Profile Eleanor May Pressman , View ORCID Profile John-Fredy Ramirez-Agudelo , View ORCID Profile André Bannink , View ORCID Profile Sanne van Gastelen , View ORCID Profile Jan Dijkstra doi: https://doi.org/10.1101/2025.05.17.654669 Ermias Kebreab 1 Department of Animal Science, University of California , Davis, CA 95616, USA Find this author on Google Scholar Find this author on PubMed Search for this author on this site ORCID record for Ermias Kebreab For correspondence: ekebreab{at}ucdavis.edu Eleanor May Pressman 1 Department of Animal Science, University of California , Davis, CA 95616, USA Find this author on Google Scholar Find this author on PubMed Search for this author on this site ORCID record for Eleanor May Pressman John-Fredy Ramirez-Agudelo 1 Department of Animal Science, University of California , Davis, CA 95616, USA Find this author on Google Scholar Find this author on PubMed Search for this author on this site ORCID record for John-Fredy Ramirez-Agudelo André Bannink 2 Wageningen Livestock Research, Wageningen University & Research , PO Box 338, 6700 AH, Wageningen, the Netherlands Find this author on Google Scholar Find this author on PubMed Search for this author on this site ORCID record for André Bannink Sanne van Gastelen 2 Wageningen Livestock Research, Wageningen University & Research , PO Box 338, 6700 AH, Wageningen, the Netherlands Find this author on Google Scholar Find this author on PubMed Search for this author on this site ORCID record for Sanne van Gastelen Jan Dijkstra 3 Animal Nutrition Group, Wageningen University & Research , PO Box 338, 6700 AH, Wageningen, the Netherlands Find this author on Google Scholar Find this author on PubMed Search for this author on this site ORCID record for Jan Dijkstra Abstract Full Text Info/History Metrics Supplementary material Preview PDF ABSTRACT Methane (CH 4 ) emissions from ruminants contribute significantly to global GHG emissions. Bromoform (CHBr 3 )-containing feed ingredients, such as Asparagopsis seaweed, have emerged as promising tools to reduce enteric CH 4 emissions. This meta-analysis quantitatively assessed the effects of CHBr 3 -containing seaweeds and synthetic CHBr 3 -based additives on CH 4 production (g/day), CH 4 yield (g/kg DMI), and DMI, as well as CH 4 intensity (g/kg product) and production (milk or daily gain in dairy and beef cattle, respectively). Data were collected from 14 studies, with 39 treatment mean differences for CH 4 production, comprising both beef and dairy cattle fed CHBr 3 -containing ingredients, while accounting for dose, DMI, and diet composition. The random-effects model’s estimates revealed that at an average CHBr 3 dose of ≈28.3 mg/kg DM, CH 4 production was reduced by 47.3%, CH 4 yield by 43.3%, and CH 4 intensity by 39.0%, with increases in CHBr 3 dose resulting in larger efficacy in mitigating CH 4 emissions. The efficacy of CHBr 3 was influenced by cattle type, with greater mitigation effects in beef cattle than dairy catle, and by dietary composition, with greater reductions observed in diets higher in starch, whereas higher NDF levels attenuate its effect. At the average CHBr 3 dose, estimated DMI was significantly reduced by 6.45% and 3.26% in dairy and beef cattle, respectively. The significant reduction in DMI did translate into a significant effect on milk yield (-4.60%) at average CHBr 3 dose. Carrier type (oil vs. biomass), measurement technique and cattle type influenced the results. Oil carrier potentially leading to more pronounced reductions, particularly for CH 4 intensity, respiration chambers yielded significantly greater CH 4 reductions compared to other methods, and beef cattle showed stronger mitigation effects than dairy cattle. This study highlights the CH 4 mitigating potential of CHBr 3 -containing feed ingredients, providing predictive models to optimize CH 4 reduction strategies under diverse production conditions. Future research should address long-term effects, dietary optimization, and practical implementation. INTRODUCTION Methane ( CH 4 ) is a potent GHG, with a global warming potential 27 or 81 times greater than carbon dioxide ( CO 2 ) over a 100-year or 20-year period, respectively (IPCC, 2021). The livestock sector, particularly large ruminant production, is a major source of anthropogenic enteric CH 4 emission, contributing substantially to global GHG emissions ( Gerber et al., 2013 ). Addressing enteric CH 4 emission from cattle is critical for mitigating climate change and meeting global sustainability goals ( Smith et al., 2014 ). One promising approach to reduce enteric CH 4 emission is the use of feed ingredients such as CHBr 3 -containing seaweed into cattle diets ( Honan et al., 2022 ). Bromoform ( CHBr 3 ) is a naturally occurring compound in certain types of seaweed, such as Asparagopsis spp. , which has been shown to considerably inhibit enteric CH 4 emissions during rumen fermentation ( Kinley et al., 2020 ). The key metalloenzymes are blocked by CHBr 3 via the Wolfe cycle ( Glasson et al., 2022 ), inhibiting the methyl transfer of the cobamide-dependent enzyme, which is needed to catalyze the final step in methanogenesis ( Fouts et al., 2022 ; Glasson et al., 2022 ). There are, however, strict safety measures for products fed to cattle that enter the human food supply, and not all CHBr 3 -containing seaweeds are FDA-approved, and in Europe CHBr 3 extracts or otherwise are feed additives and require EFSA approval ( Tricarico et al., 2025 ). Stabilized CHBr 3 -based additives have also been developed to achieve similar effects (e.g., Colin et al., 2024 ). The effectiveness of CHBr 3 -containing additives in reducing enteric CH 4 emission has been demonstrated in several studies, with reported reductions ranging from no significant reduction to over 80% ( Roque et al., 2019 ; Kinley et al., 2020 ). However, the effectiveness of CHBr 3 -containing seaweeds and CHBr 3 -based additives can vary widely due to factors such as the dose, cattle type, diet composition, and the methods used to measure CH 4 emission ( Roque et al., 2021 ). Therefore, developing robust predictive equations to quantify and elucidate the factors contributing to the variation in the effects of feed additives, including CHBr 3 -containing feed ingredients, on CH 4 emissions is essential. These equations can optimize the use of CHBr 3 -containing feed ingredients in various production systems and inform policy and management decisions to enhance sustainability in the livestock sector ( Dijkstra et al., 2025 ). Previous studies have attempted to quantify the effects of CH 4 -reducing additives through meta-analyses of in vivo and in vitro data (e.g., Appuhamy et al., 2013 ; Dijkstra et al., 2018 ; Martins et al., 2024 ). Lean et al. (2021) and Orzuna-Orzuna (2024) conducted a meta-analysis on the effects of dietary seaweed on cattle performance and CH 4 yield, which included all types of seaweed. However, our analysis is novel in its specific focus on CHBr 3 -containing seaweeds, providing a more targeted investigation compared to the broader scope of Lean et al. (2021) and Orzuna-Orzuna et al. (2024). Although multiple compounds in Asparagopsis spp. may act synergistically to inhibit methanogenesis, CHBr 3 is the most abundant natural product in Asparagopsis spp. ( Machado et al., 2016 ) and the anti-methanogenic properties of red seaweed are often attributed to CHBr 3 . Variability in CHBr 3 efficacy across studies may be due to variation in the CHBr 3 content of seaweeds, as well as potential instability of CHBr 3 in seaweed (e.g., Stefenoni et al., 2021 ; Angellotti et al., 2025 ). Thus, focusing our analysis on the effect of CHBr 3 (not red seaweed per se) allows greater standardization of the impact of CHBr 3 -containing seaweed supplementation. Most studies employ frequentist statistical methods, which, although valuable, have limitations in fully capturing the complexity and uncertainty inherent in the data, particularly when dealing with limited or noisy datasets. In contrast, Bayesian methods incorporate prior knowledge and thus potentially offering a more comprehensive quantification of uncertainty ( Sidebotham et al., 2023 ) and hence may complements these frequentist methods. The objective of this study was to develop predictive equations that quantify the effect of CHBr 3 -containing seaweed and other CHBr 3 -based additives on CH 4 production (g/d), yield (g/kg DM) and intensity (g/kg milk, ECM, ADG) in dairy and beef cattle. We hypothesized that supplementing CHBr 3 -containing seaweed as well as CHBr 3 -based additives reduce enteric CH 4 production, yield, and intensity in beef and dairy cattle, and that the level of CH 4 reduction depends on CHBr 3 dose and diet composition. Both frequentist and Bayesian meta-analysis approaches were employed to ensure the robustness of the present study and to provide a comprehensive quantification of the uncertainty associated with the estimates of CH 4 reduction achieved. MATERIALS AND METHODS Data Sources A priori inclusion criteria for the studies included in the meta-analysis were defined as follows: studies must have been conducted in vivo in either dairy or beef cattle fed either CHBr 3 -containing seaweed (e.g., Asparagopsis spp .), a seaweed-based product, or a CHBr 3 -containing additive for the purpose of CH 4 inhibition. The targeted and/or fed CHBr 3 dose (mg CHBr 3 /kg DMI) must be either reported or calculable from the provided information. The CHBr 3 products must have been fed during the same experimental timeframe as enteric CH 4 emissions were measured and not directly introduced in the rumen. The study must have included a control group that did not receive the CHBr 3 -based treatment. Enteric CH 4 emissions must have been measured (i.e., emission rate quantitatively assessed), not estimated through mathematical modeling, meta-analysis, or lifecycle analysis (LCA) or alike. Measurements of individual DMI must be provided. The basal diet’s chemical composition (% dietary DM) of starch, crude protein (CP), neutral detergent fiber (NDF), and crude fat (FAT) must either be reported or calculable from the basal diet’s ingredient composition. Studies must be original research articles published or under review in peer-reviewed journals (e.g., not theses, meeting abstracts, literature reviews, meta-analyses, or life cycle assessments) and written in English. Literature searches were conducted on 1/6/2025 using Web of Science, Scopus, and CAB Abstracts databases. Search terms included combinations of (“Asparagopsis” OR “seaweed” OR “red seaweed” OR “macroalgae” OR “algae” OR “bromoform” OR “CHBr3”) AND (“dairy cattle” OR “dairy cow” OR “beef cattle” OR “steer” OR “cow” OR “cattle”) AND (“enteric methane” OR “methane” OR “CH4”). Filters were applied to include only peer-reviewed journal articles. One co-author screened abstracts and full articles as needed to assess whether the inclusion criteria were met. The literature searches in the Web of Science, Scopus and CAB Abstracts databases yielded a total of 218, 143 and 78 entries, respectively. After removing duplicates, the remaining entries (268) underwent further screening based on the established inclusion criteria. The majority of entries (256) were excluded for the reason that studies: did not measure enteric CH 4 emission directly; CHBr 3 concentrations of the seaweed or CHBr 3 -based additive fed was not reported; involved in vitro experiments, bioreactor setups or co-digestion processes instead of in vivo studies; consisted of review articles, modelling studies, or meta-analyses; investigated seaweed species or other feed additives that did not contain CHBr 3 ; were conducted in species other than cattle; did not involve actual feeding trials, such as surveys of farmers’ or consumers’ perceptions. A total of 12 entries met the inclusion criteria ( Altman et al., 2024 ; Alvarez-Hess et al., 2023 , 2024 ; Colin et al., 2024 ; Cowley et al., 2024 ; Eikanger et al., 2024 ; George et al., 2024 ; Kinley et al., 2020 ; Meo-Filho et al., 2024 ; Roque et al., 2019 , 2021 ; Williams et al., 2024 ). Krizsan et al. (2023) and Stefenoni et al. (2021) studies were excluded because the CHBr 3 concentrations in the seaweed or in the CHBr 3 -based additive fed were not reported. Kinley et al. (2024) study was excluded because SEM of CH 4 production was not available. Although the CHBr 3 concentration of the additive fed was not given by Colin et al. (2024) , it was given in associated thesis (Colin, 2023). Hence, Colin et al. (2024) was included and CHBr 3 concentrations from the thesis were used, but all other data were from the published article. Two manuscripts in press were verified to meet inclusion criteria and included ( Angellotti et al., 2025 ; Kelly et al., 2025 ), leaving a total of 14 papers in the final dataset. The process of identifying, screening, and including studies is presented in a PRISMA (Preferred Reporting Items for Systematic Reviews and Meta-analysis; Page et al., 2021 ) flow diagram ( Figure 1 ). Download figure Open in new tab Figure 1. Study selection process. Literature searches were conducted using Web of Science, Scopus, and CAB Abstracts databases. Search terms included combinations of (“Asparagopsis” OR “seaweed” OR “red seaweed” OR “macroalga” OR “alga” OR “bromoform” OR “CHBr3”) AND (“dairy cattle” OR “dairy cow” OR “beef cattle” OR “steer” OR “cow” OR “cattle”) AND (“enteric methane” OR “methane” OR “CH4”). Data were manually collected for all included studies by one co-author using an Excel database. Data collected for all studies where available were: cattle type (beef, dairy); CH 4 production (g/day), CH 4 yield (g/kg DM), CH 4 intensity (g/kg milk, ECM, ADG), and their respective standard errors; average initial body weight (kg); milk and ECM yield (kg/day) or average daily gain (ADG, kg/day) and their respective standard errors; the CH 4 measurement technique (i.e., Greenfeed, SF 6 , or respiration chamber); CHBr 3 carrier (e.g., seaweed biomass, seaweed-steeped carrier oil, etc.). Dry matter intake and ADG (kg/d) for each treatment group and standard error were directly collected from reports of all studies except Meo-Filho et al. (2024) , as described below. Bromoform intake (g/day) and CHBr 3 dose (mg/kg diet DM) were collected or calculated for all studies as described below. Basal diet content of starch, CP, NDF, and FAT (% DM) were collected or calculated as described below for all studies. Numerical data tables were not provided in 2 articles. The online Plot Digitizer tool ( PlotDigitizer, 2024 ) was used to extract CH 4 production and yield data from Figure 1 and CH 4 intensity (g/kg milk yield) from Figure 2 in Roque et al. (2019) . Personal communication with R. Kinley clarified that the error bars in figures in Roque et al. (2019) are SE (not SD, as noted in the former), so SEM for these variables was extracted by subtracting the bottom from top error bar in each column in the figures and dividing by 2. For Kinley et al. (2020) , numerical data and SEM for CH 4 production and yield, DMI (intake during CH 4 measurement in chambers), and ADG (over treatment period) data were obtained through personal communication with the authors; SEM for CH 4 intensity (g/kg ADG) were not available. Only predicted DMI was available for Meo-Filho et al. (2024) because of the study’s setting (grazing beef cattle). Predicted DMI and ADG (kg/day) used in CH 4 yield and intensity calculations, and their respective SE, were obtained through personal communication with the authors. For George et al. (2024) , overall trial CH 4 emission and production data (1-200 d) were used and for Cowley et al. (2024) , overall trial DMI was used. Where SEM was reported as a range in the latter study, the average of the range extremes was recorded. For Altman et al. (2024) , only the data from the gas collection period in Experiment 1 were used. Experiment 2 quantified emissions reductions in steers that were previously, but not currently, supplemented with a CHBr 3 -containing feed additive, so data from this experiment were excluded. Gas emissions and SE given in L/d were converted to g/d using the ideal gas molar volume at standard temperature and pressure ( STP ; 22.4 L/mol) and gas molar mass. For gas emissions under the limit of detection in Altman et al. (2024) , the limit of detection divided by the square root of 2 was used ( Hornung and Reed, 1990 ). Production data were not available from Altman et al. (2024) . Download figure Open in new tab Figure 2. Forest plot showing bromoform (CHBr 3 ) dose (mg/kg DM) and relative mean difference values (RMD is calculated as (treatment mean − control treatment mean) / control treatment mean × 100) in CH 4 production (g/d), CH 4 yield (g/kg DMI) and CH 4 intensity (g/kg milk, ECM or ADG) for beef and dairy cattle studies. P1 and P2 = experimental periods; Wk = experimental week; HF = high forage diet; LF = low forage diet; MF = medium forage diet. The black squares represent the power of the study (i.e., greater sample sizes and smaller confidence intervals are indicated by a larger box). The summary values at the bottom of the Forest Plot (the diamond, showing -47.3%, -43.3%, -39.0%) represent the pooled estimate from the Random-Effects Meta-Analysis model. Doses of CHBr 3 in Eikanger et al. (2024) and Roque et al. (2019, 2021 ), where seaweed was fed on an OM basis, were calculated by first multiplying the reported seaweed application rate, provided on an OM basis (g seaweed OM/g diet OM), by the OM content of the basal diet DM. This product was then multiplied by the reported CHBr 3 concentration of the seaweed on a DM basis, and then by the OM content of the seaweed DM, to give (mg CHBr 3 /kg diet DM). Although seaweed was also fed on an OM basis in Angellotti et al. (2025) , seaweed OM content was not reported, so CHBr 3 doses reported in the article’s discussion were used as follows: the average of 20.7 and 16.8 mg CHBr 3 /kg DMI was used for the high group dose, and half of this was used for the low group dose. Total CHBr 3 ingested (g/d) was determined by multiplying the CHBr 3 dose by the DMI for each CHBr 3 -treated group. In Kinley et al. (2020) , the CHBr 3 dose was calculated by multiplying the daily as-fed TMR intake for each group by the Asparagopsis taxiformis inclusion rate in the as-fed ration. This result was then multiplied by the CHBr 3 content of the as-fed A. taxiformis to determine the daily CHBr 3 intake (g/d). Finally, the daily CHBr 3 intake was divided by each treatment group’s daily DMI to express CHBr 3 dose as mg CHBr 3 /kg DM. In Alvarez-Hess et al. (2024 , 2023 ) and Williams et al. (2024), CHBr 3 intake was calculated by multiplying the Asp-Oil intake by its CHBr 3 concentration. The CHBr 3 dose was then calculated by dividing the CHBr 3 intake by the DMI. The online Plot Digitizer tool was also used to extract per-week CHBr 3 intake (mg/day) from Figure 1 in Meo Filho et al. (2024). Bromoform dose was calculated by dividing CHBr 3 intake by DMI. In studies where grain was delivered via a concentrate station ( Alvarez-Hess et al., 2023 , 2024 ; Williams et al., 2024 ), the composition of the overall ingested diet was calculated by weighting each chemical component (starch, CP, NDF, and FAT as % DM) of the forage and concentrate portions according to their respective contribution to the treatment group’s overall DMI. Because dietary chemical composition was not provided by Altman et al. (2024) , starch, CP, NDF, and FAT content of the overall experimental diet was calculated by multiplying the starch, CP, NDF, and FAT content of each ingredient given by NASEM (2021) by the proportion of that ingredient in the diet and then summing for all ingredients. Dietary starch content was calculated in the same manner for George et al. (2024) and Kelly et al. (2025). For George et al. (2024) , information on nutrient composition was from the analyzed ration. Starch content for Meo-Filho et al. (2024) was assumed to be 2%, based on the starch content of the pellet ingredients, as the steers’ diet consisted of grass, which contributes negligible starch to the overall intake. The CP, NDF, and starch content of the diet in Angellotti et al. (2025) was determined by dividing the daily intake of each component by the total DMI for each treatment group. Because FAT intake was not reported, the FAT content of the basal diet was calculated by multiplying the FAT content of each ingredient given by (NASEM, 2021) by the proportion of that ingredient in the diet and then summing for all ingredients. Given they were fed a finishing diet, the nonlactating, nonpregnant Jersey cattle in Colin et al. (2024) were categorized as beef. Production data were not available from this study. Effect size calculation The data were analyzed using methods similar to those of Kebreab et al. (2023) , focusing on a frequentist meta-analysis of CH 4 emission. This analysis centered on 2 fundamental calculations: the effect size and its variance. We defined our custom effect size as the relative mean difference ( RMD ) in CH 4 production (g/d), CH 4 yield (g/kg DMI), CH 4 intensity (g/kg milk, ECM, ADG), DMI (kg/d), and product output (kg/d milk, ECM, ADG), expressed as the percentage change from the control to the treatment group. This approach allowed for comparisons across studies with varying baseline values. Importantly, whether CH 4 intensity is measured per milk output, per ECM, or per ADG, the analysis outcome remains consistent. This consistency is because the RMD framework inherently accounts for baseline variations. As a result, it ensures comparability between dairy and beef systems by standardizing proportional changes between treatment and control groups. For example, Roque et al. (2019) report CH 4 yield per milk output rather than per ECM. Nevertheless, the RMD framework guarantees consistency in effect size calculations across different measurement units. The RMDs were calculated using the equation: where = mean for the treatment (T) and control (C) groups. Since the calculation of RMD is essentially a ratio of 2 random variables, the variance calculation had to account for the variability in both the numerator and the denominator. We applied the Delta Method (e.g., Deng et al., 2018 ), using the same underlying statistical principle employed by Veneman et al. (2016) and Purba and Sangsawad (2025) for the variance of the response ratio, to ensure the variability in the treatment and control groups was accurately incorporated into the RMD uncertainty. We assumed that the groups in the RMD were independent, with zero covariance between their means. The variance was calculated using the following equation: where for the treatment (T) and control (C) group means, assuming that the reported SEM represents the variability within each group ( Ramirez-Agudelo and Kebreab, 2024 ). The equation takes the uncertainty in the treatment group and scales it by how much the control group influences the RMD . It also takes the uncertainty in the control group and scales it by how much the interaction of the treatment and control values affects the RMD . These 2 contributions are then added to combine the total uncertainty. Finally, the result of uncertainty of RMD (Var(RMD)) is obtained by multiplying by 100 2 because the RMD is expressed as a percentage. The treatment group’s influence is simpler because it only affects the numerator in the RMD calculation, while the control group’s influence is more complex because it affects both the numerator and denominator. The sampling variances are crucial for weighting the effect sizes in the meta-analysis. Studies with smaller variances (more precise estimates) are given more weight in the analysis than studies with larger variances ( Viechtbauer, 2010 ). Statistical analysis Both frequentist and Bayesian approaches were employed to meta-analyze the database and quantify the uncertainty associated with the effect of covariates on the reduction of CH 4 emission. We evaluated several categories: cattle type (included as 2 binary category: Dairy was coded as 1 if the animal was dairy cattle and as 0 if otherwise, and Beef was coded as 1 if the animal was beef cattle and as 0 if otherwise), CHBr 3 dose (mg/kg DM), quadratic CHBr 3 dose ((CHBr 3 dose) 2 ), DMI (kg/d), starch (% of DM), CP (% of DM), NDF (% of DM), and FAT (% of DM). In models that included starch, data from Roque et al. (2019) were excluded because the starch proportion was not reported and dietary data were not available to calculate it. Covariates, except cattle type, were centered within each cattle type by subtracting their mean values to reduce multicollinearity, specifically between CHBr 3 dose and its squared term, and to improve model’s results interpretability. Before centering, natural log transformation was applied to CHBr 3 to reduce its skewness (Supplemental Figure S1). This transformation normalizes the distribution, particularly for beef, which exhibited significant skewness (Shapiro-Wilk P -value = 0.031 before transformation). Equations derived from our analysis assume that CHBr 3 doses are first natural log-transformed. Using untransformed data in these equations would yield incorrect results, as the relationships were defined on the log scale. Biological systems often exhibit diminishing returns at higher treatment doses, where the effect plateaus or even reverses due to physiological saturation or adverse impacts (e.g., Calabrese et al., 2003). To capture this behavior, we included the (natural log of CHBr 3 dose) 2 term as covariate. We assumed that both the linear and the quadratic terms are important because they represent different biological processes: the linear term captures the direct inhibitory effect of CHBr 3 , whereas the quadratic term reflects saturable effect of CHBr 3. Including both ensures the model is both biologically realistic and mathematically robust. In our frequentist meta-analysis, we explored all possible combinations of up to 5 covariates. These models consistently included cattle type (i.e., the 2 binary covariates: Dairy and Beef) and natural log of CHBr 3 dose as primary predictors, along with 1, 2 or 3 additional covariates as fixed effects. The models were structured to include an intercept-free formulation via the mods argument in the ‘rma’ function (i.e., mods = ∼ 0 + Dairy + Beef + CHBr 3 + additional covariates), using Maximum Likelihood (ML) estimation. To avoid multicollinearity, among covariates in the model, Pearson correlation coefficients were calculated, and any model containing covariate pairs with an absolute correlation coefficient greater than 0.50 was excluded. Each candidate model was fitted using mixed-effects meta-regression analysis with the ‘rma’ function from the ‘metafor’ package ( Viechtbauer, 2010 ) in R ( R Core Team, 2021 ). Many studies in our database included multiple treatment groups compared to a common control group, likely resulting in correlations between observations within each study. To address this within-study correlation, we applied Robust Variance Estimation ( RVE ), clustered by study, using the ‘robust’ function. Models were evaluated based on Akaike and Bayesian information criteria ( AIC , BIC ), concordance correlation coefficient ( CCC ) between reported and predicted RMDs, and the root mean square error ( RMSE ) of leave-one-out cross-validation ( LOOCV ; Allen, 1974 ), ensuring statistical significance ( P ≤ 0.05) for all covariates. During model fitting, Cook’s distances ( Cook, 1977 ), calculated with the ‘cooks.distance’ function, were used to reveal outliers which influenced the regression (i.e., how much the estimated regression coefficients change when a particular observation is removed from the dataset). A common rule of thumb is to consider observations as influential if their Cook’s distance values exceed 4/N, where N is the total number of observations (e.g., Altman and Krzywinski, 2016 ). However, applying this to our database resulted in the removal of an excessive quantity of data points (between 25 to 55%, depending on the model structure), which would lead to a considerable loss of information and reduce the robustness of the analysis. To address this issue, we adopted a data-driven approach by defining the threshold as 3 times the mean Cook’s distance. This method provided a more balanced criterion for identifying influential data points whilst preserving the integrity of the database, and is used in the main document of this study. Cook’s distance-based outlier identification was performed separately for each candidate model. Because a data point’s influence depends on the model’s covariate set, the specific outliers identified varied slightly across models. To evaluate the sensitivity of the results to outlier elimination, we also implemented an approach using the CCC metric as an alternative to the Cook’s distance approach. Unlike Cook’s distance, which identifies outliers based on their influence on regression coefficients, the CCC-based method targets observations with the largest residuals (i.e., discrepancies between predicted and actual values of the response variable, RMD). The algorithm iteratively removes the most extreme residual (in absolute value), refits the model, and recalculates predictions until the number of removed data points matches the count identified by Cook’s distance in the respective model. Critically, the CCC method does not necessarily remove the same data points as Cook’s distance, as it prioritizes prediction accuracy rather than coefficient stability. The results obtained with the CC metric in the outlier elimination method are shown in the supplementary document of this study. To estimate how well each model might generalize and predict outcomes on new, unseen data, we calculated the RMSE derived from LOOCV. In the RMSE-LOOCV process, after outliers elimination, 1 observation was omitted at a time, the model was refitted using the remaining data, and the omitted observation was predicted. The RMSE was calculated for each prediction, and the average RMSE across all iterations. Additionally, the amount of residual heterogeneity ( τ 2 ) and its proportion ( I 2 ) were extracted from the summary of the models. The degree of heterogeneity measured by the I 2 statistic can be classified as follows: no heterogeneity (0% < I 2 ≤ 25%), low heterogeneity (25% < I 2 ≤ 50%), moderate heterogeneity (50% 75%) ( Borenstein et al., 2017 ). After removing outliers detected with the Cook’s distance method, we constructed funnel plots to visually assess the presence of publication bias ( Appuhamy et al., 2013 ). Our analyses involved 2 types of meta-analytic models, both fitted with the ‘rma’ function. First, we estimated a random-effects model to obtain an overall pooled effect size while accounting for between-study heterogeneity. We specified the observed effect sizes (RMD) and their variances (i.e., res_random <-rma(yi = RMD, vi = var(RMD), data = data, method = “ML”)). Next, we fitted a mixed-effects meta-regression model using the same function but including cattle type (i.e., the 2 binary covariates: Dairy and Beef), log-centered CHBr 3 dose and centered starch as moderators (i.e., res_mixed <-rma(yi = RMD, vi = var(RMD), mods = ∼ 0 + Dairy + Beef + log-centered CHBr 3 + centered starch, data = data, method = “ML”)). Egger’s regression test was performed using the ‘regtest’ function on both type of models. To determine the effect of CHBr 3 supplementation on RMD of DMI and products (milk, ECM, or ADG), we employed a similar meta-analytic mixed model approach as used for the CH 4 outcomes, utilizing the ‘rma’ function. The used model included the 2 binary covariates (Dairy and Beef) and the CHBr 3 dose term, centered within each cattle type. Key differences for the DMI and products analysis were that the CHBr 3 dose was used on its original scale (mg/kg DM) without log transformation, and only the model containing the linear CHBr 3 term was evaluated. The relationship between CHBr 3 dose and production outcomes was assumed to be linear within the studied dose range, unlike CH 4 outcomes which showed a nonlinear response. Pairwise differences in RMD for CH 4 production, yield, and intensity between carrier methods (biomass vs. oil), techniques (respiration chamber vs. Greenfeed vs. SF 6 ), and cattle type (dairy vs. beef) were estimated using a mixed-effects meta-regression model. To get a holistic view of the combined influence of factors, a single comprehensive model was fitted for each CH 4 outocome including carrier, technique, cattle type, and log-centered CHBr 3 dose in the model (i.e., mods <-∼ Carrier + Technique + Cattle Type + log-centered CHBr 3 ) using the ‘rma’ function. The model parameters were estimated using restricted maximum likelihood (REML). Cluster-robust standard errors were calculated, clustering by Study, using the ‘robust’ function (i.e., RVE). Pairwise differences between the levels of technique, carrier, and cattle type were then conducted using estimated marginal means (emmeans package; Lenth, 2022), applying Tukey’s method for multiple comparison adjustment, based on the fitted robust model. Some studies were excluded from this part of the analysis due to the limited data available on the reported types of carriers and technique. Specifically, Two papers using the headbox-type indirect calorimeter technique ( Altman et al., 2024 ; Colin et al., 2024 ) were not used in this part of the analysis. Two studies using respiration chambers ( Cowley et al., 2024 ; Kinley et al., 2020 ), did not report SEM of CH 4 intensity and were therefore not included in the analysis on CH 4 intensity for technique comparisons. Altman et al. (2024) utilized a Proprietary Kelp Blend Product as the carrier of CHBr 3 , whereas Kelly et al. (2025) employed a powder carrier in 1 treatment. Although Colin et al. (2024) did not report specific details about the carrier within the bromoform containing product (Alga 1.0) used, we classified this study in the biomass carrier group. Finally, in our Bayesian analysis, the ‘brms’ package ( Bürkner, 2017 ) in R was employed to perform a comprehensive Bayesian meta-analysis to quantify the uncertainty surrounding the effects of (LN(CHBr3 dose)) 2 , DMI and dietary covariates on CH 4 production, yield, and intensity. This analysis leveraged the flexibility and robustness of Bayesian modeling to provide probabilistic interpretations of parameter estimates and their associated uncertainties. The 6 covariates (LN(CHBr 3 dose)) 2 , DMI, starch, CP, NDF, and FAT) were evaluated across the CH 4 reduction outcomes. A natural log transformation was applied to CHBr 3 concentration (mg/kg DM) and used as co-variate, and covariates were mean-centered within each cattle type. The 6 Bayesian models were structured with the RMD as the response variable, alongside its variance, which was explicitly modeled using the ‘se’ function. Fixed effects included the 2 cattle type binary covariates (Dairy and Beef) and LN(CHBr 3 dose), as well as one of the 6 covariates. Random effects were incorporated to explicitly models between-study heterogeneity through a random intercept term (1 | Study). The formula for the Bayesian model was structured as: RMD | se(sqrt(var (RMD)), sigma = TRUE) ∼ 0 + Dairy + Beef + LN(CHBr 3 dose) + one of the 6 covariates + (1 | Study). The Bayesian modeling was implemented using the ‘brm’ function with 4 Markov chains to ensure robust sampling. Each chain consisted of 10,000 iterations, including a 3,000-iteration warm-up (burn-in) period. Convergence was facilitated by setting the control parameter adapt-delta to 0.999, and a fixed random seed (i.e., 123) ensured reproducibility. Covariate-specific models results were saved as RDS files for further analysis. Key metrics, including posterior means, credible intervals, and R-hat values, were examined to assess model convergence and the reliability of the results ( Bürkner, 2017 ). RESULTS AND DISCUSSION Methane Production, Yield, and Intensity The meta-analysis assessed the effectiveness of CHBr 3 -containing seaweeds and CHBr 3 -based feed additives on CH 4 emission in both dairy and beef cattle, using a database covering a range of experimental conditions, doses, DMI, and key dietary covariates. The database comprised information from 14 studies, with 39 unique comparisons between experimental and control groups. Among these, 6 studies were conducted on dairy cattle and 8 on beef cattle. Altman et al. (2024) reported only CH 4 production, Colin et al. (2024) did not report CH 4 intensity, and Cowley et al. (2024) and Kinley et al. (2020) lack SEM data for CH 4 intensity. Consequently, 38 treatment comparisons were included in the analysis of CH 4 yield, and 30 were used for CH 4 intensity. Table 1 presents descriptive statistics of studies, the mean and SEM for CHBr 3 dose, DMI, dietary components, CH 4 emissions and production outcome, categorized into 3 groups: all data (encompassing both dairy and beef cattle) across the 3 CH 4 outcomes, dairy only, and beef only. Table 2 presents descriptive statistics of studies, the mean and SEM for CHBr 3 dose, DMI, and product output (milk, ECM, or ADG) again categorized into 3 groups: all data (encompassing both dairy and beef cattle), dairy only, and beef only. The results of the Random-Effects Meta-Analysis model for CH 4 outcomes are presented in Figure 2 . The RMD values for CH 4 production and yield were all negative, and for CH 4 intensity only 2 data points comparisons— Eikanger et al. (2024 ; low dose level) and Kelly et al. (2025, low dose level in P2)—had RMD values of 0 and +24.3%, respectively. This indicates that CHBr 3 supplementation (i.e., in the form of seaweed or stabilized feed additives) generally demonstrated an anti-methanogenic effect. View this table: View inline View popup Download powerpoint Table 1. Descriptive statistics of studies and CH 4 emission for All (encompassing both dairy and beef cattle), Dairy, and Beef across the 3 CH 4 outcomes datasets (mean ± SEM) View this table: View inline View popup Download powerpoint Table 2. Descriptive statistics of studies, DMI (categorized into 3 groups: all data, dairy only, and beef only), and product output (categorized into dairy or beef) mean ± SEM) The detection and removal of outliers in this study were critical to ensure the accuracy and reliability of the results. By identifying and eliminating these outliers, we ensured that any lack of significant effect observed in the analysis was not due to the distortion caused by anomalous values, but rather to the absence of a true effect. The results of this study are presented in 3 parts. The primary analysis, which uses Cook’s distance to identify and address outliers, is detailed in the main document ( Tables 3 , 4 , and 5). To offer a more comprehensive perspective, 2 supplementary analyses are included in the supplementary material. The first supplementary analysis shows the results without any outlier removal (Supplemental Table S1 to Table S4), serving as a baseline to assess the influence of outliers on the outcomes. The second supplementary analysis (Supplemental Table S5 to Table S8) applies the CCC-based method for outlier removal, providing an alternative approach to handling outliers. Together, these supplementary analyses highlight the sensitivity of the results to different outlier treatment methods, allowing readers to evaluate the robustness and reliability of our conclusions across different methodological approaches. Cook’s distance was selected as the primary outlier diagnostic because it quantifies each observation’s influence on regression coefficients, ensuring model stability ( Cook, 1977 ). This standard regression tool (e.g., Díaz-García and González-Farías, 2004) combines leverage and residual information to identify observations that disproportionately affect model predictions. View this table: View inline View popup Download powerpoint Table 3. Comparison of regression models evaluating the effect of bromoform dose (CHBr 3 ; in LN (mg CHBr 3 / kg DM) and additional dietary (% DM) or animal covariates (cattle type) on the relative mean difference in CH 4 production (estimates ± SE) View this table: View inline View popup Download powerpoint Table 4. Comparison of regression models evaluating the effect of bromoform dose (CHBr3; in LN mg CHBr3/kg DM) and additional dietary (% DM) or animal covariates (cattle type) on the relative mean difference in CH 4 yield (estimates ± SE) Our analysis revealed distinct patterns in the significance of variables related to CH 4 production, yield, and intensity, depending on the treatment of outliers. When outliers were not removed, only NDF was a significant covariate ( P < 0.05) for CH 4 production and yield, and starch exhibited a trend ( P < 0.10). For CH 4 intensity, NDF, starch, and (LN(CHBr 3 dose)) 2 were significant covariates ( P < 0.05). However, when the CCC-based method was applied to eliminate outliers, the significance of other covariates emerged (Supplemental Tables S5, S6, and S7). For CH 4 production, FAT, starch, and NDF were significant ( P < 0.05); for CH 4 yield, FAT and starch were significant ( P < 0.05), and NDF showed a trend ( P = 0.071); and for CH 4 intensity, starch were significant ( P < 0.05); and LN (CHBr 3 ) 2 and NDF showed a trend ( P < 0.10). Tables 3 to 5 present the results obtained using Cook’s distance, the reference method for outlier detection, for CH 4 production, yield and intensity, respectively. When utilizing the estimators presented in the tables (in both the main document and the supplementary material), 2 key considerations must be addressed to ensure accurate application. First, the tables include separate columns for Dairy and Beef, which are mutually exclusive. This design requires that the appropriate column value be selected based on the cattle type under analysis. Specifically, equations intended for dairy cattle must exclusively incorporate values from the Dairy column, while those for beef cattle should use values from the Beef column. Second, the CHBr 3 dose as reported in studies in mg/kg DM must be natural log-transformed prior to its inclusion in any of the CH 4 emission calculations made with the equations presented here. This transformation is essential because all models were developed and validated based on this natural log scale. Adhering to these guidelines is critical for the proper interpretation and application of the reported estimators. View this table: View inline View popup Download powerpoint Table 5. Comparison of regression models evaluating the effect of bromoform dose (CHBr3; in LN mg CHBr3/kg DM) and additional dietary (% DM) or animal covariates (cattle type) on the relative mean difference in CH 4 intensity (estimates ± SE) The model selection process identified key dietary characteristics influencing CHBr 3 efficacy. In CH 4 production, the starch-inclusive model was the most impactful, in terms of model efficiency, as evidenced by the lowest AIC (298), BIC (306), LOOCV-RMSE (16.4%), and I 2 values (76.1%), and highest CCC (0.863) value. RMSE is expressed as a percentage of the observed mean, representing the average magnitude of prediction errors relative to the actual values.. Similar was observed for CH 4 yield, were the starch-inclusive model (Type + CHBr 3 + Starch) emerged as the most impactful, with the lowest AIC (275), BIC (283), LOOCV-RMSE (15.2%), and I 2 values (71.6%), and highest CCC (0.868) value. In line with these results, also for CH 4 intensity, the starch-inclusive model demonstrated the best good fit statistics with the lowest AIC (192), BIC (198), low LOOCV-RMSE (13.0%), and I 2 values (9.05%), and highest CCC (0.866) value. NDF was not a significant ( P = 0.076) explanatory variable for CH 4 intensity in any model that included this variable. Below we present the best models (i.e., lowest AIC, BIC, and RMSE-LOOCV and highest CCC values) to predict the percentage change in CH 4 production, yield, and intensity using the within each cattle type means reported in Table 1 : where the estimates for Dairy or Beef are -24.8 ± 2.47 or -61.2 ± 5.22, respectively; the mean LN(CHBr 3 dose in mg/kg DM) for dairy and beef are 2.685 ± 0.1111 and 3.373 ± 0.1667, respectively; and the mean starch content (% DM) for dairy and beef are 12.8 ± 0.22 and 29.7 ± 3.61, respectively. where the estimates for Dairy or Beef are -22.5 ± 2.49 or -56.0 ± 5.46, respectively; the mean LN(CHBr 3 dose in mg/kg DM) for dairy and beef are 2.685 ± 0.1111 and 3.349 ± 0.1719, respectively; and the mean starch content (% DM) for dairy and beef are 12.8 ± 0.22 and 30.0 ± 3.75, respectively. where the estimates for Dairy or Beef are -25.6 ± 2.04 or -45.6 ± 3.22, respectively; the mean LN(CHBr 3 dose in mg/kg DM) for dairy and beef are 2.685 ± 0.1111 and 3.342 ± 0.2099, respectively; and the mean starch content (% DM) for dairy and beef are 12.8 ± 0.22 and 25.7 ± 4.87, respectively. Figure 3 illustrates the observed vs. predicted percentage change in CH 4 production, yield, and intensity for beef and dairy cattle, using these best-fitting models. Download figure Open in new tab Figure 3. Observed (x-axis) vs. predicted (y-axis) percentage change in CH 4 production, yield, and intensity for beef (circles) and dairy (triangles) cattle, using the best-fitting models (starch-inclusive models) from Tables 3 , 4 , and 5 and data without outliers detected by the Cook’s distances method. The dashed diagonal line (y = x) denotes the line of identity, indicating perfect agreement between observed and predicted values. The Concordance Correlation Coefficients (CCC) are presented in the panels of each unit CH 4 outcome. All the best-fitting models include starch as a key explanatory variable, reflecting its critical role in shaping CH 4 responses to CHBr 3 supplementation. The CCC reported, 0.863 for CH 4 production, 0.868 for CH 4 yield, and 0.866 for CH 4 intensity, demonstrate strong concordance between observed and predicted CH 4 emissions. In all outcomes, several data points lie near the line of identity, indicating that these starch-inclusive models generally capture both the direction and magnitude of CH 4 changes across different diets and cattle types. Notably, depending on CH 4 unit of expression, alternative models that included NDF (for CH 4 production and yield), FAT (for CH 4 intensity), and (LN(CHBr 3 dose)) 2 (when included in models for CH 4 production and yield) also demonstrated statistically significant effects ( P ≤ 0.05). These findings indicate that although multiple dietary components contribute to CH 4 emissions variability, dietary starch is the most robust predictor within our modeling framework, underscoring its importance in refining CH 4 mitigation strategies in both beef and dairy production systems. In the beef dataset, the CHBr 3 dose (mg/kg DM) distribution was left-skewed (Supplemental Figure S1). Due to this asymmetry, a natural log transformation of CHBr 3 dose before centering was necessary to better represent the data and improve models fit. Figure 4 visualizes the predicted dose-response relationships from the best-fitting models (incorporating cattle type, natural log-transformed CHBr 3 dose, and starch content, with the latter two centered at their mean values for each cattle type). The curves demonstrate diminishing returns, where incremental CHBr 3 additions yield progressively smaller CH 4 reductions, despite the absence of a quadratic term in those models. This non-linearity arises from the natural log-transformed predictor’s inherent properties: when plotted against the linear x-axis scale (mg/kg DM), the relationship manifests as curvilinear, naturally capturing the biological saturation effect without requiring explicit quadratic terms. Beef cattle showed substantially greater percentage reductions in CH 4 production and yield than dairy cattle at equivalent doses. However, this difference diminished for CH 4 intensity, suggesting that emission mitigation effectiveness becomes more comparable between systems when normalized to product output (milk/ECM or ADG). The models predict near-zero or slightly positive relative mitigation differences at very low doses (<3-6 mg/kg DM); however, due to limited data in this range (only 2 observations, at 2 and 6 mg/kg DM), and the theoretical expectation that the inhibitor should not increase CH4 emissions, model predictions at very low doses should be interpreted with caution. The models demonstrate strong CH 4 reduction potential within practical application ranges (>10 mg/kg DM). Although one might argue that adding a quadratic term on a log-transformed variable could be redundant, our results indicate for CH 4 production and yield that both the linear and quadratic terms provide unique and important insights. In fact, our frequentist approach revealed that the (LN(CHBr 3 dose)) 2 -NDF-inclusive model (i.e., Type + LN(CHBr 3 dose) + (LN(CHBr 3 dose)) 2 + NDF; Tables 3 and 4 ) significantly represents the variability in CH 4 production and yield. Download figure Open in new tab Figure 4. Predicted percentage change (%) in CH 4 production, CH 4 yield, and CH 4 intensity relative to control as a function of CHBr 3 dose (mg/kg DM). Points represent individual study comparisons for dairy (triangles) and beef (circles). Curves represent predictions for beef (red line) and dairy (cyan line) cattle derived from the best-fitting frequentist meta-regression models (incorporating cattle type, natural log-transformed CHBr 3 dose, and starch content; see Tables 3 - 5 ), holding CHBr 3 dose and starch at their mean values for each cattle type. Data points do not include outliers detected by the Cook’s distance method. Across studies, dairy cattle were fed diets with less variation in nutritional composition compared to beef. The nutritional composition of the beef cattle diets varied considerably across studies, with CP ranging from 10.2 to 17.4%, NDF varying between 17.5 and 51.8%, FAT content spanning 3.90 to 8.33%, and starch levels fluctuating from 2.0 to 50.7%. Therefore, the marked variability in beef cattle diets likely underpins the significant impact of dietary covariates in the models, as they capture critical differences in nutritional content that directly affect CH 4 mitigation outcomes. The greater CH 4 mitigation efficacy of CHBr 3 in high-starch diets compared to high-NDF diets likely arises through interconnected mechanisms centered on hydrogen (H 2 ) dynamics and microbial ecology. High-starch fermentation preferentially drives propionate production, a major H 2 sink, thereby reducing the available H 2 pool for methanogenesis. In this H 2 -limited environment, CHBr 3 inhibits a larger proportion of the total methanogenic potential, resulting in greater percentage reductions. In contrast, high-NDF fermentation elevates H 2 production, creating a larger substrate pool for methanogens and possible larger relative mass of methanogens. Here, while CHBr 3 maintains similar absolute inhibitory effects, its relative impact diminishes as it acts on a larger potential CH 4 formation pool. The reduced mitigating effect of CHBr 3 at higher NDF levels aligns with similar findings for 3-nitrooxypropanol ( Dijkstra et al., 2018 ; Kebreab et al., 2023 ) Publication Bias Funnel plots ( Figure 5 ) were used to visualize the potential presence of publication bias. These plots display individual study RMDs of CH 4 emissions on the horizontal axis against their corresponding SEs on the vertical axis. In the random-effects models (i.e., intercept-only model without specifying a mods argument), the central vertical line represents the pooled estimate and the triangular confidence region delineates ±1.96 SE, indicating the range within which 95% of studies would be expected to lie in the absence of bias. A symmetric, inverted funnel shape suggests no substantial publication bias, whereas asymmetry implies that smaller studies might be missing on one side of the funnel, potentially skewing the overall estimate ( Afonso et al., 2024 ). In the random-effects meta-analysis, the funnel plot reveals that many observations fall outside the expected inverted funnel, indicating a high degree of dispersion. This pronounced spread is a hallmark of substantial heterogeneity among the studies, and it may also signal potential small-study effects, such as publication bias or other systematic influences. When moderators (Dairy, Beef, log-centered CHBr 3 dose, and centered Starch) are subsequently incorporated in a mixed-effects model, the model accounts for some of the between-study variability by adjusting the predicted effect sizes. Consequently, the dispersion observed in the funnel plot is altered; rather than plotting raw effect sizes, the funnel plot often displays standardized residuals that are more closely centered around their predicted values. Download figure Open in new tab Figure 5. Funnel plots for CH 4 emissions. Top row: Funnel plots of RMDs vs. SEs for random-effects models of CH 4 production (g/d), CH 4 yield (g/kg DMI), and CH 4 intensity (g/kg milk, ECM, ADG), with pooled estimates and ±1.96 SE regions. Bottom row: Standardized Residuals funnel plots for mixed-effects models, accounting for cattle type (Dairy and Beef), log-centered CHBr 3 dose and centered starch as moderators. After removing outliers detected with the Cook’s distance method in both models. Symmetry suggests low publication bias. Egger’s test p-values, displayed in each panel, indicate no significant funnel plot asymmetry ( P ≥ 0.05). Summaries of the random-effects analyses revealed substantial heterogeneity ( I 2 > 85%), indicating that the variation in reported RMDs across studies exceeded what would be expected by chance. In the random-effects model I 2 was 94.0, 88.4, and 87.3% for CH 4 production, yield, and intensity, respectively. In the mixed-effects model (including Dairy, Beef, log-centered CHBr 3 dose, and centered starch as moderators), I 2 decreased to 76.1, 71.6, and 9.05% for CH 4 production, yield, and intensity respectively. Notably, the residual heterogeneity for the CH 4 intensity mixed-effects model dropped substantially ( I 2 = 9.05%), indicating that the included moderators (cattle type, log-centered-CHBr3 dose, and starch) explained almost all the between-study variability for this outcome. The resulting standardized residual for the mixed-effects model did not display pronounced asymmetry, suggesting that once moderators were accounted for, the evidence for publication bias diminished. Egger’s regression tests were conducted to statistically evaluate funnel plot asymmetry for each CH 4 outcome in both random-effects and mixed-effects models. The results indicated no significant ( P ≥ 0.05) evidence of publication bias across all the CH 4 outcomes. However, Egger’s test yielded a borderline P -value ( P = 0.050) for the mixed-effects model for CH 4 yield, suggesting potential slight asymmetry even after accounting for moderators, although visual inspection of the residual plot did not show strong asymmetry. These results suggest that publication bias is unlikely to have substantially influenced the pooled RMD in any of the analyses. Bayesian analysis To ensure a robust and comprehensive analysis, we employed both frequentist and Bayesian statistical approaches. The frequentist meta-regression models allowed us to estimate fixed effects and assess statistical significance using well-established hypothesis testing frameworks. However, frequentist methods have limitations, particularly in handling uncertainty. To address this limitation, we complemented our analysis with a Bayesian framework. Bayesian models allowed us to quantify uncertainty more explicitly through posterior distributions and credible intervals, offering a probabilistic interpretation of the results. By integrating both approaches, we ensured that our conclusions were not solely dependent on P -values and frequentist assumptions but were reinforced by Bayesian estimates that account for uncertainty in effect sizes. The comparison of these methodologies provides a more nuanced understanding of how CHBr 3 , DMI, and dietary covariates influence CH 4 reductions, particularly in cases where frequentist and Bayesian results diverged. The posterior distributions of the estimated effects on CH 4 production, yield, and intensity are presented in Figure 6 . Each density plot represents the distribution of parameter estimates for different predictors, including (LN(CHBr3 dose)) 2 , DMI, starch, FAT, NDF, and CP. The x-axis denotes the estimated effect size and each predictor blue-shaded region illustrates the probability density function of the posterior estimate, with the peak of the curve representing the most probable estimate or the highest density region. The red lines in Figure 6 mark the zero-effect threshold (i.e., no significant effect). If the red line (zero-effect) is outside this highest density region, it indicates stronger evidence for a meaningful estimated effect. If the highest density region is predominantly on one side of this red line, it suggests an effect in that direction (positive or negative). If the peak of the blue curve is very close to the red line, it suggests the effect size is uncertain or weak. The blue shaded areas in Figure 6 also represent the 95% CI, which provide a measure of uncertainty. The symmetric distribution observed in all predictors suggests that the number of Markov chains and iterations were optimal to obtain distributions without strong skewness, which would indicate bias in the estimated effect. Download figure Open in new tab Figure 6. Posterior distributions of covariate effects on CH 4 production ( A , g/d), CH 4 yield ( B , g/kg DMI), and CH 4 intensity ( C , g/kg milk, ECM, ADG). DMI = Dry Matter Intake (kg/day), starch (% DM), FAT = Crude Fat (% DM), NDF = Neutral Detergent Fiber (% DM), and CP = Crude Protein (% DM). Each blue-shaded region illustrates the probability density function of the posterior estimate for a predictor. The peak of each curve represents the most probable estimate (highest density region) of the effect size. Vertical red lines indicate a zero-effect threshold. Posterior distributions were visually categorized into three effect types: (a) No effect — the zero line lies within the highest density region, indicating no credible effect on CH 4 reduction; (b) Negative effect — the distribution lies entirely right of zero, suggesting increased CH 4 emissions despite a positive coefficient; and (c) Positive effect — the distribution lies entirely left of zero, indicating reduced CH 4 emissions, consistent with a negative and beneficial coefficient. Visual inspection of the red line’s position relative to the highest density region (peak) of the posterior distribution ( Figure 6 ) reveals that, for CH 4 production, yield, and intensity, the effects of starch and FAT consistently indicate a meaningful CH 4 -reducing effect (peaks to the left of zero) which allows these factors to be classified in the positive significant effect group. Similarly, DMI shows a meaningful and significant CH 4 -reducing effect for CH 4 yield, just like DMI and (LN(CHBr3 dose)) 2 for CH 4 intensity. In contrast, NDF consistently indicates a meaningful CH 4 -increasing effect across all CH 4 outcomes (peak to the right of zero) which allows this factor to be classified in the negative significant effect group. The other combinations of effects and outcomes can be classified as having no significant effect because the zero-effect line falls within the highest density region of their posterior distributions. This suggests that these covariates do not have a strong or consistent influence on CH 4 outcomes reduction. For almost all the covariates, Bayesian and frequentist analyses agreed regarding their significance or lack thereof. However, divergences emerged in the interpretation of (LN(CHBr 3 dose)) 2 , and FAT effects. While the frequentist analysis determined that including (LN(CHBr 3 dose)) 2 in the model (Dairy or Beef + LN(CHBr 3 dose) + (LN(CHBr 3 dose)) 2 ) was not a significant covariate ( P > 0.05) for CH 4 intensity, the most probable values of this effect (i.e., vertical blue line in the highest density region of the posterior distributions; Figure 6 ) were far from the zero-effect line in CH 4 intensity, suggesting that there is a significant effect of (LN(CHBr 3 dose)) 2 on this outcome. While the frequentist analysis determined that FAT was a non-significant covariate for CH 4 production and yield, the Bayesian analysis indicated greater certainty in the effect of FAT in all CH 4 outcome reductions, as its highest density region (peak) of the posterior distributions is clearly positioned within the range of negative values. Visually, in Figure 6 (FAT column), the blue shaded 95% CI area is positioned to the left of the zero-effect red line. All Bayesian models achieved strong convergence, as evidenced by Rhat values of 1 and high effective sample sizes, thereby confirming the robustness and reliability of the findings. The observed differences between the Bayesian and frequentist analyses underscore the nuanced ways in which each method handles uncertainty, model structure, and borderline effects, ultimately offering complementary insights rather than mutually exclusive conclusions. In particular, for covariates such as (LN(CHBr 3 dose)) 2 and FAT, the results differed between the Bayesian and frequentist approaches; suggesting that, for these borderline cases, small shifts in modeling assumptions or data structure can tip the results. However, both methods converged in identifying starch as consistently beneficial (i.e., reducing CH 4 emission) and NDF as consistently detrimental (i.e., increasing CH 4 emission), highlighting robust findings for these dietary components. In applied settings, different analytical methods offer complementary perspectives rather than one being inherently better. Park et al. (2024) used multiple linear regression, k-nearest neighbor, and artificial neural networks to model volatile fatty acid production and CH 4 emissions, showing that neutral detergent fiber intake was associated with increased acetate and butyrate concentrations, which correlated with higher CH 4 emissions, while starch intake was linked to lower emissions. Although the k-nearest neighbor method provided the most accurate predictions, the overall agreement across methods highlighted the importance of using multiple approaches to better understand complex biological processes. DMI and Product Output The results of the Random-Effects Meta-Analysis model for DMI and product output (milk, ECM, or ADG) are presented in Figure 7 . The frequentist analysis ( Table 6 ) assessed the impact of CHBr 3 dose on DMI and product output, this analysis revealed that, at mean CHBr 3 dose (15.8 mg/kg DM), DMI in dairy cattle was significantly ( P ≤ 0.05) reduced (-6.45%). The reduction in beef cattle (-3.26%) was also significant ( P ≤ 0.05), at the mean CHBr 3 dose (37.5 mg/kg DM). Furthermore, the significant negative slope for the CHBr 3 dose term (-0.269, P < 0.001) indicates that increasing the dose relative to these respective means is predicted to further decrease DMI for both cattle types. The model exhibited moderate heterogeneity ( I 2 = 39.5%), suggesting variability in DMI was partially explained by cattle type and CHBr 3 dose. Minor differences were observed depending on the method of outlier removal. When outliers were not removed, the CHBr 3 dose remained a significant predictor, and DMI reductions in both dairy and beef cattle at the mean CHBr 3 dose were statistically significant ( P < 0.05; Supplemental Table S4). Download figure Open in new tab Figure 7. Forest plot showing bromoform (CHBr 3 ) dose (mg/kg DM) and relative mean difference values (RMD is calculated as (treatment mean − control treatment mean)/ control treatment mean × 100) in DMI (kg/d) and Products (milk, ECM and ADG, kg/d) for beef and dairy cattle studies. P1 and P2 = experimental periods; Wk = experimental week; HF = high forage diet; LF = low forage diet; MF = medium forage diet. The black squares represent the power of the study (i.e., greater sample sizes and smaller confidence intervals are indicated by a larger box). The summary values at the bottom of the Forest Plot (the diamond, showing - 5.9%, -1.7%) represent the pooled estimate from the Random-Effects Meta-Analysis model. View this table: View inline View popup Download powerpoint Table 6. Comparison of regression models evaluating the effect of bromoform dose (CHBr 3 ) on the relative mean difference in DMI and products (milk, ECM, ADG) (estimates ± SE) In particular at higher inclusion levels, studies report a significant reduction in DMI, likely due to the cows sorting out the seaweed, reacting to its taste, or related to inflammation of rumen wall. For instance, dairy cows fed A. taxiformis at 0.25% of diet DM showed reduced DMI and increased time spent on eating and on eating in a head-down position, suggesting feed sorting or aversion. However, at lower inclusion levels (e.g., 0.125% of diet DM), the reduction in DMI was not present, indicating that palatability challenges may be dose-dependent ( Nyløy et al., 2023 ). Researchers have tried mixing seaweed with molasses or other carriers to mask taste; some success is noted, but at higher doses cattle often still detect the seaweed and reduce intake or sort it out. This underscores that there is an upper limit to how much CHBr 3 cattle willingly consume. Studies consistently show that cattle do not fully adapt to the taste or presence of A. taxiformis in their diet, and in some cases, prolonged supplementation has resulted in significant intake depression. For example, Stefenoni et al. (2021) observed decreased DMI and milk yield in lactating cows fed A. taxiformis during 28-day experimental period, while Muizelaar et al. (2021) reported reduced intake across all treatment levels in a 28 day trial. These findings suggest that although A. taxiformis is highly effective at reducing enteric CH 4 emissions, its long-term use may be limited by palatability issues and potential negative effects on feed intake and animal performance. Notably, dairy cows appear a bit more sensitive to seaweed’s effects on DMI and performance than beef cattle, perhaps because even a moderate intake drop in lactation can noticeably reduce milk outputs. There was a significant negative linear effect (-0.331; P ≤ 0.05) of the CHBr 3 dose on production outcomes (milk, ECM, or ADG). At the average dose of CHBr 3 , a reduction of 4.60% in milk yield was observed ( P ≤ 0.05), whereas beef cattle showed a non-significant 0.858% increase in ADG. The model exhibited low I 2 (22.6%), indicating that the observed effect of CHBr3 dose on production outcomes was relatively consistent across the different studies included in the analysis, after accounting for cattle type. Figure 8 illustrates the relationship between CHBr 3 dose and the RMD in DMI and products (milk, ECM, or ADG). The superimposed lines represent predicted linear dose-responses from meta-regression models after exluding outliers identified by Cook’s distance. Several data points indicated positive RMDs (i.e., increased ADG) including low doses (e.g., ID 18: +32.4% ADG at 2 mg/kg) and moderate-to-higher dose responses (e.g., ID 24: +21.5% at 27 mg/kg; ID 25: +32.3% at 32 mg/kg; ID 26: +35.8% at 33 mg/kg). However, these points were not flagged as outliers by Cook’s distance method. The presence of these variable responses contributed to poor model fit (CCC = 0.186) suggesting that a linear model may inadequately capture the ADG-dose relationship in beef cattle. Biological complexity or data noise could underlie this incosistency. Indeed, a negative RMD (i.e., decreased ADG) occurred, with the lowest response being -35.1% (ID 20, 18 mg/kg DM), but this data point was not flagged as an outlier by Cook’s distance method either. The inconsistent response (both increased and decreased ADG within a range of 2 to 35 mg/kg of CHBr 3 ) does not enable a model to adequately capture variation in response. When no outliers were removed, at the average dose of CHBr 3 , dairy cattle showed a non-significant decrease in milk production, and beef cattle showed a non-significant increase in ADG (Supplemental Table S4). An alternative outlier removal approach (CCC method, Supplemental Table S8) excluded IDs 18, 20, 25, and 26 (i.e., the changes in ADG with the greatest decrease (ID 20) and greatest increases (ID 18, 25, and 26)), modestly improving predictive metrics (CCC: 0.280; RMSE: 10.2%) compared to the Cook’s distance approach. Download figure Open in new tab Figure 8. Predicted percentage change (%) in DMI and products (milk, ECM, or ADG) relative to control as a function of CHBr 3 dose (mg/kg DM). Points represent individual study comparisons for dairy (triangles) and beef (circles). Lines represent predictions for beef (red line) and dairy (cyan line) cattle derived from the best-fitting frequentist meta-regression models (incorporating cattle type and non-log-transformed CHBr 3 dose; see Table 6 ). Data points do not include outliers detected by the Cook’s distance method. Bromoform Carrier, Methane Measurement Technique, and Cattle Type Table 7 summarizes the estimated RMD differences in CH 4 production, yield, and intensity across carriers, measurement techniques, and cattle types. View this table: View inline View popup Download powerpoint Table 7. Estimated difference in relative mean difference for CH 4 production (g/d), CH 4 yield (g/kg DMI), and CH 4 intensity (g/kg milk, ECM, ADG) across carriers, techniques, and cattle types using a mixed-effects meta-regression model incorporating log-centered-CHBr 3 dose (mg/kg DM) Biomass-based studies used higher CHBr 3 doses (29.8–32.4 mg/kg DM) than oil-based studies (19.5–22.6 mg/kg DM). Despite this, oil was associated with greater CH 4 reductions, significantly so for CH 4 intensity ( P < 0.05), and nearing significance for production and yield ( P < 0.10), suggesting carrier type influences efficacy. Average CHBr 3 doses varied by measurement technique: GreenFeed studies used the highest dose (33.8 mg/kg DM), followed by respiration chamber (24.8 mg/kg DM for CH 4 production and yield), and SF 6 (15.6 mg/kg DM). Respiration chamber was associated with significantly greater CH 4 reductions than SF 6 for production and yield ( P < 0.05). Differences between GreenFeed and the other techniques were not significant for CH 4 production or yield ( P ≥ 0.05). For CH 4 intensity, GreenFeed resulted in significantly greater reductions compared to SF 6 ( P < 0.05); data were insufficient for comparisons involving respiration chamber. Although GreenFeed studies used higher doses, adjusted CH 4 reductions were not greater than those observed with respiration chamber, suggesting that measurement technique may affect outcomes beyond dose differences alone. Beef cattle received a substantially higher average CHBr 3 dose than dairy cattle (34.6 vs. 15.8 mg/kg DM for CH 4 production and yield; 38.9 vs. 15.8 mg/kg DM for CH 4 intensity). After adjusting for dose, carrier, and measurement technique, RMD was significantly lower ( P < 0.05) in beef cattle for CH 4 yield and intensity, indicating greater reductions. No significant difference ( P ≥ 0.05) was observed for CH 4 production between cattle types. CONCLUSIONS The study confirms that CHBr 3 -containing seaweeds and CHBr 3 -based additives significantly reduce CH 4 emission in dairy and beef cattle. The pooled estimates from the Random-Effects Meta-Analysis model show reductions of 47.3, 43.3, and 39.0% for CH 4 production, yield, and intensity, respectively, at average CHBr 3 dose level of ≈28.3 mg/kg DM, with increases in CHBr 3 dose resulting in larger efficacy in mitigating CH 4 emissions. The Bayesian framework provided critical confirmation of dietary starch and NDF effects. For starch, the entire posterior density for all CH 4 outcomes fell entirely to the left of the zero-effect threshold, indicating a high probability of its beneficial role in enhancing CHBr 3 efficacy. Conversely, NDF’s posterior density resided entirely to the right of zero, reflecting its consistent association with diminished CHBr 3 effectiveness. This Bayesian confirmation aligns with frequentist meta-regression results, solidifying starch and NDF as robust, directionally consistent predictors of CH 4 mitigation outcomes. At their respective average CHBr 3 doses, DMI was significantly reduced in dairy and beef cattle. Similarly, product output at the average dose was significantly reduced for dairy. The results highlight the potential for optimizing dietary management to maximize the environmental benefits of CHBr 3 . The detailed predictive equations provide a basis to integrate these findings into GHG accounting systems and practical farm management strategies. NOTES The study was financially supported by the UC Davis Sesnon Endowed Chair program (EK and JFRA). Special thanks to Robert Kinley, Breanna Roque, Leanna Kelly and Paulo Meo-Filho for their generous provision of additional data, which greatly enhanced the scope and depth of our research. Nonstandard abbreviations used CHBr 3 bromoform CH 4 methane CO 2 carbon dioxide RC respiration chamber GF GreenFeed I 2 heterogeneity LOOCV leave-one-out cross-validation MD mean difference RMD relative mean difference RMSE root mean square error SF 6 sulfur hexafluoride STP standard temperature and pressure REFERENCES ↵ Afonso , J. , R. Ramirez-Campillo , F. M. Clemente , F. C. Büttner , and R. Andrade . 2024 . The perils of misinterpreting and misusing “publication bias” in meta-analyses: An education review on funnel plot-based methods . Sports Med . 54 : 257 – 269 . doi: 10.1007/s40279-023-01927-9 . OpenUrl CrossRef PubMed Al-Adilah , H. , M. C. Feiters , L. J. Carpenter , P. Kumari , C. J. Carrano , D. Al-Bader , and F. C. Küpper . 2022 . Halogens in seaweeds: Biological and environmental significance . Phycology 2 : 132 – 171 . doi: 10.3390/phycology2010009 . OpenUrl CrossRef ↵ Allen , D. M . 1974 . The relationship between variable selection and data augmentation and a method for prediction . Technometrics 16 : 125 – 127 . doi: 10.1080/00401706.1974.10489157 . OpenUrl CrossRef Web of Science ↵ Altman , N. , and M. Krzywinski . 2016 . Analyzing outliers: Influential or nuisance? Nat. Methods 13 : 281 – 283 . doi: 10.1038/nmeth.3812 . OpenUrl CrossRef PubMed ↵ Altman , A. , E. Vanzant , S. Davis , D. Harmon , and K. R. McLeod . 2024 . Effects of a proprietary kelp blend product on enteric methane production and tissue residues in cattle . Animals 14 : 2411 . doi: 10.3390/ani14162411 . OpenUrl CrossRef PubMed ↵ Alvarez-Hess , P. S. , J. L. Jacobs , R. D. Kinley , B. M. Roque , A. S. O. Neachtain , S. Chandra , V. M. Russo , and S. R. O. Williams . 2024 . Effects of a range of effective inclusion levels of Asparagopsis armata steeped in oil on enteric methane emissions of dairy cows . Anim. Feed Sci. Technol . 310 : 115932 . doi: 10.1016/j.anifeedsci.2024.115932 . OpenUrl CrossRef ↵ Alvarez-Hess , P. S. , J. L. Jacobs , R. D. Kinley , B. M. Roque , A. S. O. Neachtain , S. Chandra , and S. R. O. Williams . 2023 . Twice daily feeding of canola oil steeped with Asparagopsis armata reduced methane emissions of lactating dairy cows . Anim. Feed Sci. Technol . 297 : 115579 . doi: 10.1016/j.anifeedsci.2023.115579 . OpenUrl CrossRef ↵ Angellotti , M. , M. Lindberg , M. Ramin , S. J. Krizsan , and R. Danielsson . 2025 . Asparagopsis taxiformis supplementation to mitigate enteric methane emissions in dairy cows – effects on performance and metabolism . J. Dairy Sci . 108 : 2503 – 2516 . doi: 10.3168/jds.2024-25258 . OpenUrl CrossRef PubMed ↵ Appuhamy , J. R. N. , A. B. Strathe , S. Jayasundara , C. Wagner-Riddle , J. Dijkstra , J. France , and E. Kebreab . 2013 . Anti-methanogenic effects of monensin in dairy and beef cattle: A meta-analysis . J. Dairy Sci . 96 : 5161 – 5173 . doi: 10.3168/jds.2012-5923 . OpenUrl CrossRef PubMed Berça , A. S ., L. O. Tedeschi, A. da Silva Cardoso, and R. A. Reis. 2023 . Meta-analysis of the relationship between dietary condensed tannins and methane emissions by cattle . Anim. Feed Sci. Technol . 298 : 115564 . doi: 10.1016/j.anifeedsci.2022.115564 . OpenUrl CrossRef ↵ Borenstein , M. , J. P. Higgins , L. V. Hedges , and H. R. Rothstein . 2017 . Basics of meta-analysis: I² is not an absolute measure of heterogeneity . Res. Synth. Methods 8 : 5 – 18 . doi: 10.1002/jrsm.1230 . OpenUrl CrossRef PubMed ↵ Bürkner , P. C . 2017 . brms: An R package for Bayesian multilevel models using Stan . J. Stat. Softw . 80 : 1 – 28 . doi: 10.18637/jss.v080.i01 . OpenUrl CrossRef PubMed Calabrese , E. J. , and L. A. Baldwin . 2003 . Toxicology rethinks its central belief . Nature 421 : 691 – 692 . doi: 10.1038/421691a . OpenUrl CrossRef PubMed Web of Science Colin , R. L. 2023 . The evaluation of feed additives on reducing enteric methane production from cattle. MS thesis. Univ. of Nebraska, Lincoln, NE . https://digitalcommons.unl.edu/animalscidiss/267/ . ↵ Colin , R. L. , J. L. Sperber , K. K. Buse , P. J. Kononoff , A. K. Watson , and G. E. Erickson . 2024 . Effect of an algae feed additive on reducing enteric methane emissions from cattle. Transl . Anim. Sci . 8 :txae109. doi: 10.1093/tas/txae109 . OpenUrl CrossRef ↵ Cook , R. D . 1977 . Detection of influential observation in linear regression . Technometrics 19 : 15 – 18 . doi: 10.1080/00401706.1977.10489493 . OpenUrl CrossRef Cooke , R. F. , L. R. Eloy , S. C. Bosco , P. V. Lasmar , J. M. de Simas , T. Leiva , and S. R. de Medeiros . 2024 . An updated meta-analysis of the anti-methanogenic effects of monensin in beef cattle. Transl . Anim. Sci . 8 :txae032. doi: 10.1093/tas/txae032 . OpenUrl CrossRef ↵ Cowley , F. C. , R. D. Kinley , S. L. Mackenzie , M. R. S. Fortes , C. Palmieri , G. Simanungkalit , A. K. Almeida , and B. M. Roque . 2024 . Bioactive metabolites of Asparagopsis stabilized in canola oil completely suppress methane emissions in beef cattle fed a feedlot diet . J. Anim. Sci . 102 :skae109. doi: 10.1093/jas/skae109 . OpenUrl CrossRef ↵ Deng , A. , U. Knoblich , and J. Lu . 2018 . Applying the Delta method in metric analytics: A practical guide with novel ideas . Pages 233–242 in Proc. 24th ACM SIGKDD Int. Conf. Knowl. Discov. Data Min., London, UK. ACM, New York, NY . doi: 10.1145/3219819.3219919 . OpenUrl CrossRef ↵ Dijkstra , J. , A. Bannink , J. France , E. Kebreab , and S. Van Gastelen . 2018 . Antimethanogenic effects of 3-nitrooxypropanol depend on supplementation dose, dietary fiber content, and cattle type . J. Dairy Sci . 101 : 9041 – 9047 . doi: 10.3168/jds.2018-14456 . OpenUrl CrossRef PubMed ↵ Dijkstra , J. , A. Bannink , G. F. S. Congio , J. L. Ellis , M. Eugène , F. Garcia , M. Niu , R. E. Vibart , D. R. Yáñez-Ruiz , and E. Kebreab . 2025 . Feed additives for methane mitigation: Modeling the impact of feed additives on enteric methane emission of ruminants— Approaches and recommendations . J. Dairy Sci . 108 : 356 – 374 . doi: 10.3168/jds.2024-25049 . OpenUrl CrossRef PubMed ↵ Eikanger , K. S. , S. T. Kjær , P. Dörsch , A. D. Iwaasa , A. W. Alemu , I. Schei , P. B. Pope , L. H. Hagen , and A. Kidane . 2024 . Asparagopsis taxiformis inclusion in grass silage-based diets fed to Norwegian red dairy cows: Effects on ruminal fermentation, milk yield, and enteric methane emission . Livest. Sci . 285 : 105495 . doi: 10.1016/j.livsci.2024.105495 . OpenUrl CrossRef Feng , X. Y. , J. Dijkstra , A. Bannink , S. Van Gastelen , J. France , and E. Kebreab . 2020 . Antimethanogenic effects of nitrate supplementation in cattle: A meta-analysis . J. Dairy Sci . 103 : 11375 – 11385 . doi: 10.3168/jds.2020-18541 . OpenUrl CrossRef PubMed ↵ Fouts , J. Q. , M. C. Honan , B. M. Roque , J. M. Tricarico , and E. Kebreab . 2022 . Enteric methane mitigation interventions. Transl . Anim. Sci . 6 : 1 – 16 . doi: 10.1093/tas/txac041 . OpenUrl CrossRef ↵ George , M. M. , S. V. Platts , B. A. Berry , M. F. Miller , A. M. Carlock , T. M. Horton , and M. H. George . 2024 . Effect of SEAFEED, a canola oil infused with Asparagopsis armata, on methane emissions, animal health, performance, and carcass characteristics of Angus feedlot cattle. Transl . Anim. Sci . 8 :txae116. doi: 10.1093/tas/txae116 . OpenUrl CrossRef ↵ Gerber , P. J. , A. N. Hristov , B. Henderson , H. Makkar , J. Oh , C. Lee , R. Meinen , F. Montes , T. Ott , J. Firkins , A. Rotz , C. Dell , A. T. Adesogan , W. Z. Yang , J. M. Tricarico , E. Kebreab , G. Waghorn , J. Dijkstra , and S. Oosting . 2013 . Technical options for the mitigation of direct methane and nitrous oxide emissions from livestock: A review . Anim . 7 ( Suppl. 2 ): 220 – 234 . doi: 10.1017/S1751731113000876 . OpenUrl CrossRef PubMed ↵ Glasson , C. R. , R. D. Kinley , R. de Nys , N. King , S. L. Adams , M. A. Packer , J. Svenson , C. T. Eason , and M. Magnusson . 2022 . Benefits and risks of including the bromoform-containing seaweed Asparagopsis in feed for the reduction of methane production from ruminants . Algal Res . 64 : 102673 . doi: 10.1016/j.algal.2022.102673 . OpenUrl CrossRef ↵ Honan , M. , X. Feng , J. M. Tricarico , and E. Kebreab . 2022 . Feed additives as a strategic approach to reduce enteric methane production in cattle: Modes of action, effectiveness, and safety . Anim. Prod. Sci . 62 : 1303 – 1317 . doi: 10.1071/AN20295 . OpenUrl CrossRef ↵ Hornung , R. W. , and L. D. Reed . 1990 . Estimation of average concentration in the presence of nondetectable values . Appl. Occup. Environ. Hyg . 5 : 46 – 51 . doi: 10.1080/1047322X.1990.10389587 . OpenUrl CrossRef ↵ Kebreab , E. , A. Bannink , E. M. Pressman , N. Walker , A. Karagiannis , S. van Gastelen , and J. Dijkstra . 2023 . A meta-analysis of effects of 3-nitrooxypropanol on methane production, yield, and intensity in dairy cattle . J. Dairy Sci . 106 : 927 – 936 . doi: 10.3168/jds.2022-22211 . OpenUrl CrossRef PubMed ↵ Kelly , L. , E. M. Pressman , J. F. Ramirez-Agudelo , H. Chernavsky , P. Alvarez-Hess , S. Jacques , M. Hess , and E. Kebreab . 2025 . The effect of Rumin8 investigational veterinary product—a bromoform-based feed additive—on enteric methane emissions, animal production parameters, and the rumen environment in feedlot cattle. Transl . Anim. Sci . 9 :txae015. doi: 10.1093/tas/txaf028 . OpenUrl CrossRef ↵ Kinley , R. D. , G. Martinez-Fernandez , M. K. Matthews , R. de Nys , M. Magnusson , and N. W. Tomkins . 2020 . Mitigating the carbon footprint and improving productivity of ruminant livestock agriculture using a red seaweed . J. Clean. Prod . 259 : 120836 . doi: 10.1016/j.jclepro.2020.120836 . OpenUrl CrossRef ↵ Kinley , R. D. , B. M. Roque , S. Mackenzie , M. Fortes , C. Palmieri , G. Tarr , H. Cuthbertson , R. Polkinghorne , and F. Cowley . 2024 . Productivity of commercial feedlot beef production significantly improved by Asparagopsis bioactives stabilized in canola oil . Am. J. Plant Sci . 15 : 899 – 929 . doi: 10.4236/ajps.2024.1510058 . OpenUrl CrossRef ↵ Krizsan , S. J. , M. Ramin , J. C. Chagas , A. Halmemies-Beauchet-Filleau , A. Singh , A. Schnürer , and R. Danielsson . 2023 . Effects on rumen microbiome and milk quality of dairy cows fed a grass silage-based diet supplemented with the macroalga Asparagopsis taxiformis. Front . Anim. Sci . 4 : 1112969 . doi: 10.3389/fanim.2023.1112969 . OpenUrl CrossRef ↵ Lean , I. J. , H. M. Golder , T. M. D. Grant , and P. J. Moate . 2021 . A meta-analysis of effects of dietary seaweed on beef and dairy cattle performance and methane yield . PLoS One 16 : e0249053 . doi: 10.1371/journal.pone.0249053 . OpenUrl CrossRef PubMed Lenth , R. 2022 . emmeans: Estimated marginal means, aka least-squares means. R package version 1.7.2. Available from CRAN : https://CRAN.R-project.org/package=emmeans . ↵ Machado , L. , M. Magnusson , N. A. Paul , R. D. Kinley , R. de Nys , and N. Tomkins . 2016 . Identification of bioactives from the red seaweed Asparagopsis taxiformis that promote antimethanogenic activity in vitro . J. Appl. Phycol . 28 : 3117 – 3126 . doi: 10.1007/s10811-016-0830-7 . OpenUrl CrossRef ↵ Martins , L. F. , S. F. Cueva , C. F. A. Lage , M. Ramin , T. Silvestre , J. M. Tricarico , and A. N. Hristov . 2024 . A meta-analysis of methane-mitigation potential of feed additives evaluated in vitro . J. Dairy Sci . 107 : 288 – 300 . doi: 10.1016/S0022-0302(23)00819-6 . OpenUrl CrossRef PubMed ↵ Meo-Filho , P. , J. F. Ramirez-Agudelo , and E. Kebreab . 2024 . Mitigating methane emissions in grazing beef cattle with a seaweed-based feed additive: Implications for climate-smart agriculture . Proc. Natl. Acad. Sci. U. S. A . 121 : e2410863121 . doi: 10.1073/pnas.2410863121 . OpenUrl CrossRef PubMed ↵ Muizelaar , W. , M. Groot , G. van Duinkerken , R. Peters , and J. Dijkstra . 2021 . Safety and transfer study: Transfer of bromoform present in Asparagopsis taxiformis to milk and urine of lactating dairy cows . Foods 10 : 584 . doi: 10.3390/foods10030584 . OpenUrl CrossRef PubMed National Academies of Sciences, Engineering, and Medicine (NASEM) . 2021 . Nutrient Requirements of Dairy Cattle: Eighth Revised Edition. Natl . Acad. Press , Washington, DC . ↵ Nyløy , E. , E. Prestløkken , M. Eknæs , K. S. Eikanger , L. Heldal Hagen , and A. Kidane . 2023 . Inclusion of red macroalgae (Asparagopsis taxiformis) in dairy cow diets modulates feed intake, chewing activity, and estimated saliva secretion . Animals 13 : 489 . doi: 10.3390/ani13030489 . OpenUrl CrossRef PubMed Orzuna-Orzuna , J. F. , A. Lara-Bueno , G. D. Mendoza-Martínez , L. A. Miranda-Romero , G. Vázquez Silva , M. E. de la Torre-Hernández , N. Sánchez-López , and P. A. Hernández-García . 2024 . Meta-analysis of dietary supplementation with seaweed in dairy cows: Milk yield and composition, nutrient digestibility, rumen fermentation, and enteric methane emissions . Dairy 5 : 464 – 479 . doi: 10.3390/dairy5030036 . OpenUrl CrossRef ↵ Page , M. J. , D. Moher , P. M. Bossuyt , I. Boutron , T. C. Hoffmann , C. D. Mulrow , L. Shamseer , J. M. Tetzlaff , E. A. Akl , S. E. Brennan , R. Chou , J. Glanville , J. M. Grimshaw , A. Hróbjartsson , M. M. Lalu , T. Li , E. W. Loder , E. Mayo-Wilson , S. McDonald , L. A. McGuinness , L. A. Stewart , J. Thomas , A. C. Tricco , V. A. Welch , P. Whiting , and J. E. McKenzie . 2021 . PRISMA 2020 explanation and elaboration: Updated guidance and exemplars for reporting systematic reviews . BMJ 372 : n160 . doi: 10.1136/bmj.n160 . OpenUrl FREE Full Text ↵ Park , M. , S. Cho , E. Jeon , and N. J. Choi . 2024 . Development of volatile fatty acid and methane production prediction model using ruminant nutrition: Comparison of algorithms . Fermentation 10 : 410 . doi: 10.3390/fermentation10080410 . OpenUrl CrossRef ↵ PlotDigitizer . 2024 . Version 3.1.5 . https://plotdigitizer.com (accessed October 10, 2024 ). ↵ R Core Team . 2021 . R: A language and environment for statistical computing. R Foundation for Statistical Computing, Vienna, Austria . https://www.R-project.org/ . Purba , R. A. P. , and Sangsawad , P . ( 2025 ). Meta-Analysis of Dietary Interventions for Enteric Methane Mitigation in Ruminants Through Methodological Advancements and Implementation Pathways . Vet. Sci . 12 : 372 . doi: 10.3390/vetsci12040372 OpenUrl CrossRef PubMed ↵ Ramirez-Agudelo , J. F. , and E. Kebreab . 2024 . Systematic review for optimizing sample size in dairy cow methane emission studies in temperate regions: A comprehensive methodological approach . J. Dairy Sci . 107 : 9442 – 9458 . doi: 10.3168/jds.2023-24529 . OpenUrl CrossRef PubMed ↵ Roque , B. M. , J. K. Salwen , R. D. Kinley , and E. Kebreab . 2019 . Inclusion of Asparagopsis armata in lactating dairy cows’ diet reduces enteric methane emission by over 50 percent . J. Clean. Prod . 234 : 132 – 138 . doi: 10.1016/j.jclepro.2019.06.193 . OpenUrl CrossRef ↵ Roque , B. M. , M. Venegas , R. D. Kinley , R. de Nys , T. L. Duarte , X. Yang , and E. Kebreab . 2021 . Red seaweed (Asparagopsis taxiformis) supplementation reduces enteric methane by over 80 percent in beef steers . PLoS One 16 : e0247820 . doi: 10.1371/journal.pone.0247820 . OpenUrl CrossRef PubMed ↵ Sidebotham , D. , C. J. Barlow , J. Martin , and P. M. Jones . 2023 . Interpreting frequentist hypothesis tests: Insights from Bayesian inference . Can. J. Anaesth . 70 : 1560 – 1575 . doi: 10.1007/s12630-023-02557-5 . OpenUrl CrossRef ↵ Smith , P. , H. Clark , H. Dong , E. A. Elsiddig , H. Haberl , R. Harper , J. House , M. Jafari , O. Masera , C. Mbow , N. H. Ravindranath , C. W. Rice , C. Roble do Abad, A. Romanovskaya , F. Sperling , and F. Tubiello . 2014 . Agriculture, forestry, and other land use (AFOLU) . In Climate Change 2014: Mitigation of Climate Change. Contribution of Working Group III to the Fifth Assessment Report of the IPCC . Cambridge Univ. Press . ↵ Stefenoni , H. A. , S. E. Räisänen , S. F. Cueva , D. E. Wasson , C. F. A. Lage , A. Melgar , M. E. Fetter , P. Smith , M. Hennessy , B. Vecchiarelli , J. Bender , D. Pitta , C. L. Cantrell , C. Yarish , and A. N. Hristov . 2021 . Effects of the macroalga Asparagopsis taxiformis and oregano leaves on methane emission, rumen fermentation, and lactational performance of dairy cows . J. Dairy Sci . 104 : 4157 – 4173 . doi: 10.3168/jds.2020-19686 . OpenUrl CrossRef PubMed ↵ Tricarico , J. M. , F. Garcia , A. Bannink , S.-S. Lee , M. A. Miguel , J. R. Newbold , P. K. Rosenstein , M. R. Van der Saag , and D. R. Yáñez-Ruiz . 2025 . Feed additives for methane mitigation: Regulatory frameworks and scientific evidence requirements for the authorization of feed additives to mitigate ruminant methane emissions . J. Dairy Sci . 108 : 395 – 410 . doi: 10.3168/jds.2024-25051 . OpenUrl CrossRef PubMed ↵ Veneman , J. B. , Saetnan , E. R. , Clare , A. J. , and Newbold , C. J . ( 2016 ). MitiGate; an online meta-analysis database for quantification of mitigation strategies for enteric methane emissions . Sci Total Environ . 572 : 1166 – 1174 . doi: 10.1016/j.scitotenv.2016.08.029 OpenUrl CrossRef PubMed ↵ Viechtbauer , W . 2010 . Conducting meta-analysis in R with the metafor package . J. Stat. Softw . 36 : 1 – 48 . https://www.jstatsoft.org/article/view/v036i03/ . OpenUrl CrossRef PubMed ↵ Williams , S. R. O. , A. S. O. Neachtain , S. Chandra , R. B. S. Burgess , S. Labaf , G. Aylward , P. S. Alvarez-Hess , and J. L. Jacobs . 2024 . The effects of feeding liquid or pelleted formulations of Asparagopsis armata to lactating dairy cows on methane production, dry matter intake, milk production, and milk composition . Anim. Feed Sci. Technol . 309 : 115891 . doi: 10.1016/j.anifeedsci.2024.115891 OpenUrl CrossRef View the discussion thread. Back to top Previous Next Posted May 20, 2025. Download PDF Supplementary Material Email Thank you for your interest in spreading the word about bioRxiv. NOTE: Your email address is requested solely to identify you as the sender of this article. Your Email * Your Name * Send To * Enter multiple addresses on separate lines or separate them with commas. You are going to email the following A meta-analysis of effects of seaweed and other bromoform containing feed ingredients on methane production, yield, and intensity in cattle Message Subject (Your Name) has forwarded a page to you from bioRxiv Message Body (Your Name) thought you would like to see this page from the bioRxiv website. Your Personal Message CAPTCHA This question is for testing whether or not you are a human visitor and to prevent automated spam submissions. Share A meta-analysis of effects of seaweed and other bromoform containing feed ingredients on methane production, yield, and intensity in cattle Ermias Kebreab , Eleanor May Pressman , John-Fredy Ramirez-Agudelo , André Bannink , Sanne van Gastelen , Jan Dijkstra bioRxiv 2025.05.17.654669; doi: https://doi.org/10.1101/2025.05.17.654669 Share This Article: Copy Citation Tools A meta-analysis of effects of seaweed and other bromoform containing feed ingredients on methane production, yield, and intensity in cattle Ermias Kebreab , Eleanor May Pressman , John-Fredy Ramirez-Agudelo , André Bannink , Sanne van Gastelen , Jan Dijkstra bioRxiv 2025.05.17.654669; doi: https://doi.org/10.1101/2025.05.17.654669 Citation Manager Formats BibTeX Bookends EasyBib EndNote (tagged) EndNote 8 (xml) Medlars Mendeley Papers RefWorks Tagged Ref Manager RIS Zotero Tweet Widget Facebook Like Google Plus One Subject Area Systems Biology Subject Areas All Articles Animal Behavior and Cognition (7618) Biochemistry (17635) Bioengineering (13859) Bioinformatics (41846) Biophysics (21401) Cancer Biology (18534) Cell Biology (25422) Clinical Trials (138) Developmental Biology (13352) Ecology (19860) Epidemiology (2067) Evolutionary Biology (24285) Genetics (15582) Genomics (22463) Immunology (17700) Microbiology (40298) Molecular Biology (17141) Neuroscience (88424) Paleontology (666) Pathology (2825) Pharmacology and Toxicology (4813) Physiology (7633) Plant Biology (15107) Scientific Communication and Education (2042) Synthetic Biology (4284) Systems Biology (9808) Zoology (2267)