dietary inclusion of 3 Nordic brown macroalgae on enteric methane emission and productivity of dairy cows

Macroalgae has gained increasing interest as anti-methanogenic feed additives for cattle, but most in vivo studies are limited to investigating effects of the red macroalgae Asparagopsis spp. Hence, this study aimed to investigate the CH 4 mitigating potential of 3 brown macroalgae from the Northern hemisphere when fed to dairy cows, and to study the effects on feed intake, milk production, feed digestibility, and animal health indicators. The experiment was conducted as a 4 × 4 Latin square design using 4 lactating rumen, duodenal, and ileal cannulated Danish Holstein dairy cows. The cows were fed a total mixed ration (TMR) without any macroalgae or the same TMR diluted with, on a dry matter basis, either 4% ensiled Saccharina latis-sima , 4% Ascophyllum nodosum (NOD), or 2% Sargas-sum muticum (MUT). Each period consisted of 14 d of adaptation, 3 d of digesta and blood sampling, and 4 d of gas exchange measurements using respiration chambers. Milk yield and dry matter intake (DMI) were recorded daily. Blood was sampled on d 13 and 16 and analyzed for health status indicators. None of the 3 species affected the CH 4 emission. Moreover, milk yield and DMI were also unaffected. Total-tract digestibility of crude protein was significantly lower for NOD compared with other diets, and additionally, the NOD diet also tended to reduce total-tract digestibility of neutral detergent fiber compared with MUT. Blood biomarkers did not indicate negative effects of the dietary inclusion of macroalgae on cow health. In conclusion, none of the 3 brown macroalgae reduced CH 4 emission and did not affect DMI and milk production of dairy cows, whereas negative effects on the digestibility of nutrients were observed when A. nodosum was added. None of the diets would be allowed to be fed in commercial dairy herds due to high contents of iodine, cadmium


INTRODUCTION
Enteric methane emission from ruminants is considered to be one of the major sources of greenhouse gas emissions from livestock (Myhre et al., 2013;Place et al., 2022).With a global warming potential 80 times higher than CO 2 in a 20-yr perspective and a lifetime of only 12 yr in the troposphere as compared with hundreds of years for CO 2 (Forster et al., 2021), reductions in enteric CH 4 from ruminants also represent an attractive and feasible opportunity to achieve acute effects on global warming mitigation.Recently, attention has been drawn to macroalgae as possible antimethanogenic feed additives.The most well-studied and potent CH 4 mitigating macroalgae belongs to the genus of red macroalga, Asparagopsis.Methane yield (CH 4 g/kg DMI) was reduced by up to 67% in a study by Roque et al. (2019), when A. armata was fed to dairy cows at an inclusion level of 1.0% on an OM basis; however, milk production and DMI were also significantly reduced.Similar findings have been reported by Stefenoni et al. (2021), when cows were fed A. taxiformis.High contents of brominated and chlorinated compounds, predominantly bromoform, are responsible for the antimethanogenic properties of Asparagopsis spp.(Machado et al., 2016).Unfortunately, several these halomethanes, including bromoform and chloroform, are suspected to be possible carcinogens (DeMarini, 2020).Additionally, the contribution to ozone depletion due to the emission of bromoform from Asparagopsis spp.during growth (Jia et al., 2022) could counteract the climate benefits of using Asparagopsis spp. as an antimethanogenic feed additive.Interestingly, Nørskov et al. (2021) found no detectable levels of halomethanes in the brown macroalgae Dictyota sp., Sargassum sp., Fucus vesiculosus, Fucus serratus, and Ascophyllum nodosum in contrast to Asparagopsis spp.Thus, the antimethanogenic effects of the species observed in several in vitro studies may have been caused by other, potentially safer, bioactive compounds (Machado et al., 2014;Pandey et al., 2022;Park et al., 2022;Thorsteinsson et al., 2023).Brown macroalgae contain a variety of phenolic compounds, including phlorotannins, which are believed to be the antimethanogenic compounds (Wang et al., 2008;Nørskov et al., 2021).Ensiling of brown macroalgae might increase the amount of bioactive CH 4 inhibitors in the alga tissue as a response to postharvest stress.However, studies on CH 4 mitigating properties of temperate brown macroalgae in vivo in dairy cows are limited (Antaya et al., 2019).
Hence, the aim of this study was to investigate the antimethanogenic potential of 3 Northern procurable brown macroalgae, A. nodosum, Sargassum muticum, and Saccharina latissima, when mixed into TMR of dairy cows.Furthermore, feed intake, milk production, nutrient digestibility, and animal health indicators were also investigated to evaluate the effects of the inclusion on animal welfare and performance.

MATERIALS AND METHODS
The experiments were conducted at Aarhus University, AU Viborg-Research Centre Foulum, Denmark.The handling and care of the cows complied with the guidelines set out by the Danish Ministry of Environment and Food (act 474 of 15th of May 2014 and executive order 2028 of 14th of December 2020) concerning animal experimentation and care of animals under experiments and under consideration of the ARRIVE Guidelines (Percie du Sert et al., 2020).A license was obtained from the Danish Animal Experiments Inspectorate.

Macroalgae
Based on previous in vitro studies, it was decided to investigate the CH 4 mitigating potential of ensiled Saccharina latissima, Sargassum muticum, and Ascophyllum nodosum in vivo (Thorsteinsson et al., 2021;Pandey et al., 2022;Park et al., 2022).Saccharina latissima (chemical composition after ensiling on a DM basis: ash: 46.5%; mannitol: 2.17%) was cultivated by Ocean Rainforest, Kaldbak, Faroe Islands.After harvest, S. latissima was packed in airtight bags and left for ensiling at room temperature for 14 d as a previous in vitro study indicated that ensiling remarkably increased the CH 4 mitigating potential of S. latissima (Thorsteinsson et al., 2021).In contrast, both A. nodosum and Sargassum spp.have shown antimethanogenic properties without any postharvest treatment.After ensiling, the S. latissima was oven-dried at 40°C and ground to 2-to 4-mm particle size.Ascophyllum nodosum (ash: 23.2% of DM; mannitol: 2.85% of DM) was purchased from a commercial producer (Thorverk HF, Reykhólar, Iceland; harvested from the shores of the islands of Rauðseyjar) and had a particle size of 2-to 4-mm.Sargassum muticum (ash: 35.8% of DM; mannitol: 3.22% of DM) was harvested from the wild along the shores of the Western part of Limfjorden, Denmark by the Department of Ecoscience at Aarhus University and hence not commercially available.After harvest, S. muticum was oven-dried at 40°C and ground to 2-to 4-mm particle size.Due to a limited amount of wild S. muticum available during the harvest season, the inclusion rate was reduced to 2% of DM instead of the 4% of DM for the 2 other macroalga-containing diets.All species were harvested in August to September as the total polyphenol content was believed to be highest in the autumn and ash content lower as compared with spring and winter (Pandey et al., 2022).

In Vivo Experiment
Experimental Design.The experiment was conducted as 4 × 4 Latin square design using 4 lactating Danish Holstein dairy cows, which received TMR with or without addition of 3 different macroalgae: control (CON) diet without any macroalgae, or the same ration diluted with, on a DM basis, either 4% ensiled S. latissima (LAT), 4% A. nodosum (NOD), or 2% S. muticum (MUT).Each of the 4 experimental periods lasted 3 weeks, where the first 14 d were assigned to adaptation to the diet, followed by 3 d of digesta sampling, and finally 4 d of gas exchange measurements.Based on the lower CH 4 mitigating potential of the 3 brown macroalgae observed in vitro as compared with Asparagopsis spp., it was decided to include the macroalgae in the TMR at higher inclusion levels compared with that used in previous in vivo studies with the potent Asparagopsis spp.(up to 1% on a OM basis; Roque et al., 2019;Stefenoni et al., 2021).
Animals and Housing.Before the experiment, the cows (2 third parity and 2 fourth parity) were fitted with rumen, duodenal, and ileal cannulas to allow the collection of digesta for determination of diet digestibility in different segments of the digestive tract.The cows were milked twice daily at 05.15 and 16.30 h.Average ± SD milk yield was 31.1 ± 4.99 kg/d, days in milk was 174 ± 63.2 d, BCS was 3.0 ± 0.20, and BW was 739 ± 54.0 kg at the beginning of the experiment.All cows were housed in individual pens (400 × 450 cm) with slatted floor and a cubicle bed with mattress and sawdust Diet and Feeding.The total mixed rations were prepared once a day before noon and fed to the cows on an ad libitum basis, divided in 2 daily meals given at 6.15 and 17.15 h with approximately 40% of the daily ration provided in the morning and 60% in the afternoon.Spring growth grass/clover silage (perennial ryegrass, hybrid ryegrass, red clover, and white clover) and corn silage were included in the rations as roughages, while barley, sugar beet pulp, sugar beet molasses, rapeseed meals, mineral supplements, and macroalga were included as concentrates (CON diet without macroalgae; Table 1).Feed residues were weighed daily before afternoon feeding.Titanium dioxide (TiO 2 ) and chromic oxide (Cr 2 O 3 ) were weighed into degradable paper bags (10 g of Cr 2 O 3 and 13 g of TiO 2 ) and used as external markers to determine nutrient flow in the digestive tract by dosing the markers directly to the rumen twice daily coinciding with milkings (Myers et al., 2006;Titgemeyer et al., 2013).The cows had free access to water, and the amount of ingested water was measured by a water-meter (Brødrene Dahl, Brøndby, Denmark) during the sampling and chamber periods.
Sampling and Recordings.The rumen was visually inspected for any signs of damages such as rumen papilla atrophy and ulcers on d 1 in each period and after the last period.This was done after rumen evacuation.Dry matter intake of cows was measured from d 13 to 21 on a daily basis by weighing the amount of allocated feed and feed residues followed by determination of feed and residue DM contents, while intake of feed as-fed was recorded throughout the experiment.Milk yield was recorded daily throughout the experiment.The composition of the milk was determined on d 18 to 21 in each period.
In each period, rumen liquid, duodenal and ileal content, urine, and feces were collected at 8 sampling times over a 3 d period (i.e., at 1800 h on d 14; at 0300, 1200, and 2100 h on d 15; at 0600, 1500, and 0000 h on d 16; and at 0900 h on d 17).At each sampling time, duodenal (0.50 L) and ileal (0.20 L) samples were collected in plastic bags attached to the cannulas.Fecal samples  (0.35 L) were collected during voluntary defecation or by grab sampling from the rectum.Digesta samples were pooled across all sampling times and stored at −20°C until analyzed.Rumen fluid (30 mL) was sampled from the ventral ruminal sac using a syringe attached to a rumen sampler device (Bar Diamond Inc., Parma Idaho).
The pH in rumen fluid was measured immediately after sampling using a digital pH-meter (Meterlab PHM 220, Radiometer, Brønshøj, Denmark).Samples were stored at −20°C for later analysis of VFA, L-lactate, glucose, and NH 3 concentration.Redox potential was measured after 2 min of stabilization in the ventral rumen sac simultaneously with the sampling of rumen liquid by using a redox meter with a platinum electrode (Intellical MTC101 ORP/redox electrode, Hach, Germany).Urine was also collected at all sampling times during voluntary urinations or upon manual stimulation of the pelvic region, and the pH was measured immediately after collection using the same digital pH-meter as for the rumen fluid.Serum and plasma were sampled by venipuncture from the tail vein on d 14 at 1600 h and on d 17 at 0800 h.Plasma was collected in lithium-heparin stabilized vacutainers (Greiner Bio-One GmbH, Kremsmünster, Austria) for subsequent determination of thyroxine (T4), thyroid-stimulating hormone (TSH), urea, glucose, BHB, and nonesterified fatty acids (NEFA).The tubes were centrifuged at 3,000 × g at 4°C for 20 min.Additionally, blood was drawn into serum vacutainers (Greiner Bio-One GmbH, Kremsmünster, Austria) for determination of bile acids, total protein, albumin, aspartate aminotransferase (AST), gammaglutamyl transferase (g-GT), glutamate dehydrogenase (GLDH), and total bilirubin.The samples were left to coagulate for at least 1 h at room temperature before being centrifuged at 1,300 × g at 20°C for 10 min.Plasma and serum samples were transferred to cryotubes and stored at −20°C until analyzed.
Four individual transparent polycarbonate respiration chambers based on open-circuit indirect calorimetry, modified from Hellwing et al. (2012), were used for measurement of gas exchange on d 18 to 21. Inside dimensions of the chambers measured 415 (length) × 270 (width) × 234 cm (height), resulting in a volume of 28.4 m 3 .The chambers were placed in a separate barn from the pens used during the sampling period.To allow visual constant between the cows, the chambers were placed in a square.The cows were assigned to the same specific respiration chamber for the first 48 h of gas measurements throughout the experiment.To counteract any differences in background air composition, the cows were changed to the chamber along the diagonal for the latter 48 h of gas measurement.Airflow was measured using a mass flow meter (HFM-200 with laminar flow element, Teledyne Hastings Instruments, Hampton, Virginia).Additionally, concentrations of gases (CH 4 , CO 2 , O 2 , and H 2 ; Columbus Instruments, Columbus, Ohio) in outlet air, and temperature, humidity, and differential pressure (Veng Systems, Roslev, Denmark) in the chambers were also measured.Before, during, and after the experiments, recovery tests (n = 55 for CO 2 and n = 54 for CH 4 ) were performed by infusing a known amount of pure CO 2 or CH 4 into the chambers and comparing it with the amount of gas measured by the system.Across chambers, average recovery values ± SD were 100.2 ± 1.18% for CO 2 and 99.65 ± 1.65% for CH 4 .Recovery tests were used to correct the measured gas concentrations.The average of CH 4 and CO 2 recoveries was used to correct O 2 and H 2 .

In Vitro Experiment
Experimental Design.After the in vivo experiment was finalized, we decided, based on the outcomes, to further investigate the CH 4 mitigating potential of the macroalgae in an in vitro system, simulating rumen fermentation, where the possible interferences of continuous feed intake and outflow of material from the rumen were eliminated and with the possibility to increase the proportion of macroalgae in the substrate compared with the in vivo experiment (17% in vitro vs. up to 4%, on a DM basis in vivo).For assessment of CH 4 mitigating properties of the macroalgae, 0.5 g ± 0.02 g of basal feed was added to a Duran bottle (capacity: 132 ± 1.1 mL) together with 0.1 g ± 0.02 g of macroalga.Corn silage (ash: 3.54%; CP: 7.77%; NDF: 32.9%; starch: 35.1% on a DM basis) and grass/clover silage (ash: 9.18%; CP: 17.9%; NDF: 36.0%;sugars: 1.76% on a DM basis) were used separately as basal feeds.To compare CH 4 mitigating properties of the TMR, 0.5 g ± 0.02 g of TMR from in vivo experiment was also added to Duran bottles.For the determination of rumen degradability of the pure macroalga biomass, 0.5 g ± 0.02 g of macroalga was added to Duran bottles without any basal feed.Each sample type was included as triplicates and tested in 2 separate runs.Three replicates of pure corn silage and grass/clover silage were included in each run as control samples.Similarly, 2 replicates of buffered rumen fluid without any addition of feed or algae were included which served as blank samples.All samples were incubated for 48h.
Sampling and Recordings.Rumen fluid for the in vitro experiment was collected 30 min before the morning feeding from 3 rumen cannulated nonlactating Danish Holstein cows (average BW ± SD 805 ± 28 kg) fed at maintenance level on a diet consisting of grass/ clover hay, barley straw, and a pelleted concentrate mixture (40% barley grain, 40% oat grain, 10% soybean meal, 3% rapeseed meal, 3% sugar beet molasses, and 4% of a commercial mineral mixture per kg fresh mixture).The liquid was separated from particles by filtration through one layer of cheesecloth and collected in preheated thermos flasks.Buffer, redox indicator, reducing agent, macro and micro mineral solutions were prepared and mixed as described by Menke and Steingass (1988), and kept under anaerobic conditions by continuously flushing with N 2 until rumen fluid was added in a 2:1 ratio (buffer solution: rumen fluid).The wireless, fully automated ANKOM Gas Production System (ANKOM Technology, Macedon, New York) was used for measurement of in vitro gas production as described by Thorsteinsson et al. (2023).
After 24 h of incubation, 10 mL of gas was extracted from each gasbag using a gastight syringe with a twist valve (Hamilton Bonaduz AG, 7402 Bonaduz, Switzerland).Following, the gas was transferred into Exetainer vials (Labco Limited, Ceredigion, United Kingdom) for later CH 4 analyses.After 48 h of incubation, the fermented rumen inoculum was filtered through F57 fiber bags (ANKOM Technology, Macedon, New York) for the collection of undegraded feed residues.A sample of the fermented liquid was collected during the filtration for later analysis of VFA.

Chemical Analyses
In Vivo Experiment.Dry matter content of fresh feed and residue samples was determined by daily drying at 60°C for 48 h (AOAC International, 2000).Milk samples were analyzed for contents of fat, protein, lactose monohydrate, urea, and composition of fatty acids (FA; Schwarz et al., 2022) by mid-infrared reflection (MilkoScan 7 RM; Eurofins Steins Laboratorium A/S, Vejen, Denmark).
Before the chemical analysis, TMR and digesta samples were freeze-dried and ground on a 1-mm screen, except for a 0.5-mm screen was used for analysis of starch (Ultra Centrifugal Mill ZM 200,Verder Scientific,Hann,Germany).Ash content in the samples was determined by combustion at 525°C for 6 h.Total N and C in digesta, basal feeds, macroalgae and TMR was analyzed using the Dumas principle (Hansen, 1989) in a Vario Max CN (Elementar Analysesysteme GmbH), and CP was calculated as total nitrogen × 6.25.Crude fat in TMR was determined by Soxhlet extraction with petroleum ether (Soxtec 2050, Foss Analytical, Hillerød, Denmark) after hydrolysis with HCl (Stoldt, 1952).Starch in digesta and TMR was digested with heat-stable α-amylase and amyloglucosidase.The reaction was subsequently assayed for glucose (Kristensen et al., 2007) by using a YSI model 2900 analyzer (YSI Inc., Yellow Springs, OH).Neutral detergent fiber was treated with heat-stable amylase and sodium sulfite using a Fibertec M6 System (Foss Analytical, Hillerød, Denmark) and reported as ash-free NDF (Mertens, 2002).Content of TiO 2 in digesta and TMR samples was analyzed according to Myers et al. (2004), while content of Cr 2 O 3 was analyzed by oxidation to chromate and afterward determined spectrophotometrically using Lamba 900 equipment (PerkinElmer Inc., Waltham, Massachusetts; Schürch,1950).Mineral content in TMR was measured according to Boderskov et al. (2021).
Concentration of NH 3 in ruminal fluid was measured using Randox Ammonia Kit-AM1015 and Cobas Mira Plus (Roche) after being diluted by phosphate buffer (100 mmol/L), while concentrations of VFA were determined in stabilized rumen fluid after methanolchloroform extraction with 2-ethylbutyrate as internal standard, using a gas chromatography (GC; Trace 1310, Thermo Scientific, Germany) with split/splitless injector at 225°C and a flame ionization detector at 250°C.A 30 m × 0.53 mm × 1 µm HP-FFAP column (Agilent Technologies Inc., Wilmington, DE) was used with helium as carrier gas at 0.3405 atm.The GC oven was programmed to increase from 100 to 200°C at 10°C/min.L-lactate and glucose in rumen fluid were analyzed using the immobilized glucose oxidase electrode technique (Mason, 1983; YSI 2900D, YSI Inc., Yellow Springs).
Bile acid, total protein, albumin, AST, g-GT, GLDH, and total bilirubin in serum were analyzed by Laboklin Laboratory for Clinical Diagnostics GmbH and Co. KG (Bad Kissingen, Germany) using the photometric method on a Roche Cobas 8000 (Roche Diagnostics, Indianapolis).Thyroxine and TSH in plasma were analyzed at the Central Veterinary Laboratory at the University of Copenhagen, Denmark, and measured using the Immulite 2000 immunoassay system (Immulite 2000, Siemens Healthineers, Erlangen, Germany).The concentrations of glucose, L-lactate, and urea in plasma were measured by a spectrophotometric assay, following the manufacturer's guidelines (Siemens Medical Solutions, Tarrytown, New York).β-OH-butyrate was determined using a method involving oxamic acid in the media to inhibit lactate dehydrogenase followed by measurement of the absorbance at 340 nm due to the production of NADH (Harano et al., 1985), while NEFA was determined using the Wako, NEFA C ACS-ACOD assay method.All analyses were performed using an autoanalyzer, ADVIA 1800 Chemistry System (Siemens Medical Solutions, Tarrytown, New York).
In Vitro Experiment.Basal feeds and TMR were freeze-dried and ground on a 2-mm screen (Ultra Centrifugal Mill ZM 200, Verder Scientific, Hann, Germany), while the macroalgae were ground on a 1-mm screen using the same mill.Nutrient content was analyzed using near-infrared spectroscopy at a commercial feed testing laboratory (Eurofins Agro Testing A/S, Vejen, Denmark).Mannitol was measured by extraction in 66% ethanol and by using enzymatic-fluorometric methods, principally such as analyses of D-lactate (Larsen, 2017), except mannitol dehydrogenase (EC 1.1.1.67,Megazyme Ltd., Ireland) was used as oxidoreductase instead.Ash-free NDF in the macroalgae was determined using heat-stable α-amylase and sodium sulfite (Mertens, 2002) in the ANKOM 2000 Fiber Analyzer (ANKOM Technology, 2017).Content of DM was determined in incubated samples by oven-drying the samples at 103°C for 24 h, while ash content was determined by combustion at 525°C for 6 h.Concentration of VFA and NH 3 in incubated rumen fluid was determined as described for the in vivo experiment, while concentration of CH 4 in gas collected during fermentation was determined from vials using a GC (Trace 1310, Thermo Scientific, Germany) as described by Thorsteinsson et al. (2023).

Calculations
In Vivo Experiment.Dry matter intake was calculated as the amount of DM of feed residues subtracted from DM offered.Gross energy contents in TMR were calculated according to NorFor (Volden, 2011).Nutrient intakes of OM, CP, carbon, NDF, and starch ash were calculated by multiplying DMI with the respective nutrient content in the TMR.
Duodenal, ileal and fecal DM flows were calculated separately and subsequently averaged across markers, assuming concentrations in pooled digesta samples were representative for the average daily flow of digesta.The flows of OM, NDF, CP, C, and starch in duodenum, ileum, and feces were calculated from the DM flow and their respective concentrations in each section of the digestive tract.Nutrient intake and flow were used to calculate apparent nutrient digestibility in different sections of the digestive tract.Ruminal redox potential was not converted relative to a standard hydrogen electrode.
Gas exchange was measured as flows at standard temperature and pressure (STP; 0°C/ 273.15K and 101.325 kPa).Following, the exchange of gas in L/d was converted to g/d using the density of each gas at STP, which were 0.716, 1.963, 0.0899, and 1.428 (L/g) for CH 4 , CO 2 , H 2 , and O 2 , respectively.Data were deleted when chambers were open and cows were milked and fed.The cows were assumed to have a similar gas production for deleted minutes as the average for all minutes for each measuring period.The respiratory coefficient was calculated as the ratio between CO 2 produced and O 2 consumed (L/L).
In Vitro Experiment.Cumulative gas pressure from the in vitro experiment was converted from psi to mL gas produced at STP (IUPAC, 2014), assuming that the ideal gas law applies to the produced gases (predominantly CO 2 , CH 4 and H 2 ).Methane (mL) production was calculated from TGP after 24h and CH 4 concentration (%) in the collected gas.
Rumen degradable DM (dDM; %) was calculated as: Rumen degradable OM (dOM) was determined similarly to dDM.Predicted dOM (%) of macroalga and basal feeds in combination was calculated as: Production of VFA and NH 3 were calculated from concentration and total inoculum volume in bottles.Total gas production, VFA, NH 3 , dDM and dOM response parameters were blank corrected before the statistical analyses.

Statistical Analyses
All statistical analyses were conducted in R 4.1.2(R Core Team, 2021).
In Vivo Experiment.For the in vivo experiment, observations of all variables were averaged within cow and period.The effect of diet on the various animal responses was analyzed with the following linear mixed model fitted with REML and the "lmer" function from the "lme4" package (Bates et al., 2015): where Y dpc is the dependent response variable, µ is the overall mean, α is the fixed effect of diet (d = CON, LAT, MUT, or NOD), γ is the fixed effect of period (P = 1 to 4), A is the random effect of cow (c = 1 to 4), and Ε dpc is the random residual error assumed to be independent with constant variance and normally distributed.
In Vitro Experiment.Effects of the macroalgae and TMR in vitro on the various response parameters were analyzed with the following linear mixed model: where Y tse is the dependent response variable, µ is the overall mean, α is the fixed effect of treatment, A is the random effect of experimental run, and Ε te is the random residual error, assumed to be independent with constant variance and normally distributed.
All data were evaluated for normality of the residuals by using the Shapiro-Wilk test and by evaluating the QQ-plots constructed in R, while homogeneity of the variance was tested by evaluating plots of residuals and by using Bartlett's test.All presented data were normally distributed, except for ruminal redox potential where one outlier was removed (mean ± 3 × SD).Data are presented in tables as estimated marginal means (EMS) and standard error of means.Differences between EMS were evaluated using Tukey's method for comparison.Statistical significance was declared when P ≤ 0.05 and statistical tendencies were declared when 0.05 < P ≤ 0.10.

In Vivo Experiment
Gas Exchange, Milk Yield, and Composition of Milk FA.The cows consumed and produced significantly lower amounts of O 2 and CO 2 , respectively, when fed the NOD as compared with CON and LAT diets (Table 2), which was associated with a numerically lower DMI during the periods of gas exchange measurement (Table 3).However, inclusion of macroalgae in the diets did not affect gas yield (g/kg DMI) or gas intensity (g/kg ECM) of CH 4 , H 2 , O 2 , or CO 2 (Table 2).
Milk production and milk composition were unaffected by the treatments, except for NOD which had significantly lower levels of urea and C16:0 of total FA in milk as compared with CON and MUT and higher proportions of PUFA in milk than the other treatments (Table 3).Feed Intake, Nutrient Digestibility, and Rumen Fermentation Pattern.The LAT diet tended to have a lower DMI compared with MUT (P = 0.07; Table 4) during the sampling period, while no differences were observed during the chamber period.The lower DMI in combination with slightly higher ash content resulted in a significantly lower intake of OM in LAT, while MUT had the highest intake of OM.Furthermore, cows tended to consume a higher amount of water on the LAT as compared with CON diet (P = 0.08).
NOD tended to decrease the ruminal digestibility of starch compared with MUT (P = 0.10), but this was compensated by a significantly higher digestibility of starch in the small intestine in NOD compared with MUT.The lowest small intestinal digestibility of CP was observed in NOD compared with CON, and additionally total-tract digestibility of CP was reduced in NOD compared with all the remaining diets.NOD also reduced total-tract digestibility of NDF compared with MUT (P = 0.08).Rumen fermentation parameters (Table 5) were unaffected by dietary treatments, except for increased proportions of acetate and lowered proportions of propionate in total VFA when the NOD diet was fed.
Metabolic and Health Indicators.Urea concentrations in blood plasma and urine were reduced on the NOD compared with CON and MUT diets (Table 6), as it was also observed in milk (Table 3).Cows fed the LAT diet had significantly lower urinary pH than on CON and NOD diets.
The highest levels of total T4 were observed when cows were fed the LAT diet (P = 0.02), which also has the highest level of iodine (Table 1) compared with MUT, and hence the T4: TSH ratio in blood serum also tended to be highest in cows fed the LAT and NOD diets (P = 0.07).Plasma levels of albumin, g-GT and GLDH were above the reference ranges from the commercial laboratory for all dietary treatments, except for GLDH when cows were fed the LAT diet.However, the plasma levels were unaffected by dietary treatment.Plasma levels of total bilirubin was highest when cows were fed the LAT diet and lowest on MUT (P < 0.001), but levels were within the reference range for all dietary treatments.
No indications of rumen papilla atrophy or rumen ulcers were observed on any of the dietary treatments upon visual inspections after evacuation of rumen contents (results not shown).Values within the same line with different superscripts differ (P < 0.05). 1 CON = control diet, LAT = 4% ensiled Saccharina latissima (LAT), MUT = 2% Sargassum muticum, and NOD = 4% Ascophyllum nodosum.All inclusion rates were on DM basis. 2 Short chain FA include C4, C6, C8 and C10 FA; medium chain FA include C12, C14 and C16, and long chain FA include C18 and longer FA.Values within the same line with different superscripts differ (P < 0.05).

In Vitro Evaluation of the Effect of Dietary Algae Addition on Rumen Fermentation
The pure macroalgae, A. nodosum, used in formulation of the NOD diet had by far the lowest degradability of DM and OM, and this was associated with the lowest production of total VFA and the lowest NH 3 concentration in the fermented fluid after the 48 h of incubation in buffered rumen fluid.The highest in vitro degradability were observed for ensiled S. latissima, used in formulation of the LAT diet (Table 7).
Among the 4 formulated TMR used in the in vivo cow feeding trial (CON, LAT, MUT, NOD), there were overall treatment effects on CH 4 production (Table 7) with the lowest production observed for the LAT and MUT diets.However, the differences between the TMR were not large enough to become significant in the pairwise-comparisons.There were no differences in in vitro rumen degradability of DM and OM between the 4 TMR, and differences in total VFA production and NH 3 during fermentation were small, however highest in CON.
When the effect of the pure macroalgae were tested in vitro upon addition to either of 2 basal feeds, corn silage and grass/clover silage, the inclusion rate was higher (16.7%) than in the TMR.In these in vitro tests, A. nodosum significantly reduced CH 4 production expressed per g incubated OM on grass/clover silage, and also tended to reduce the CH 4 production per g incubated OM on corn silage.However, no effects on CH 4 production were observed when expressed per g dDM or dOM on either of the 2 basal feeds, because the addition of A. nodosum significantly reduced dDM and dOM as well as NH 3 concentration in the fermented fluid after fermentation (Table 7).Production of VFA per g dOM was not affected by addition of any of the 3 macroalgae to the 2 basal feed silages.The observed dOM, when A. nodosum or ensiled S. latissima were added to either of the 2 basal feeds, was slightly lower than the predicted in vitro dOM (calculated from the individual degradability of the macroalga and basal feed), while the observed and predicted in vitro dOM were in agreement for S. muticum (Supplemental Figure S1).

DISCUSSION
The addition of 3 Northern procurable macroalgae in diets for dairy cows at inclusion rates of 2% (S. muticum) or 4% (A.nodosum and ensiled S. latissima) of dietary DM had no effect on enteric CH 4 emission.Values within the same line with different superscripts differ (P < 0.05). 1 CON = control diet, LAT = 4% ensiled Saccharina latissima (LAT), MUT = 2% Sargassum muticum, and NOD = 4% Ascophyllum nodosum.All inclusion rates were on DM basis.Similar findings have been reported by Antaya et al. (2019) and Muizelaar et al. (2023) when A. nodosum and S. latissima, respectively, when supplemented to dairy cows, although the inclusion levels in the current study were remarkably higher (Antaya et al., 2019: 113 g DM/d;Muizelaar et al., 2023: 150 g DM/d).Due to the relatively small particle size of the macroalgae biomass, it could be speculated that the macroalgae particles may have had a too short retention time in the rumen (Kaske and Engelhardt, 1990) to exert antimethanogenic effects before being flushed out of the reticulorumen.To follow up on this hypothesis, CH 4 mitigating properties of the 4 TMR from the in vivo experiment were also tested in an in vitro system simulating rumen fermentation, where the possible interferences of continuous feed intake and outflow of material from the rumen were eliminated (Ramin and Huhtanen, 2012).Some suppression of the methanogenesis was observed in vitro when A. nodosum was added to corn silage, but no effect was detected when expressed per g of dDM or dOM even at a higher inclusion rate (17%) than it was possible to achieve in the TMR (2%-4%) fed to the cows.In contrast, Sargassum muticum increased CH 4 production per g incubated g DM and OM when added to grass/clover compared with A. nodosum.The reason for this is not obvious, but the higher mannitol content in S. muticum might be part of the explanation (Maneein et al., 2021), although no difference was observed on corn silage.The exact nature of antimethanogenic compounds in brown macroalgae is unknown.Nørskov et al. (2021) found that neither of the species tested in our study contained detectable amounts of the halomethanes, which as previously mentioned are responsible for the antimethanogenic actions of the red Asparagopsis species (Machado et al., 2016).In contrast, the brown macroalgae contained a range of other phenolic compounds and for numerous of those the chemistry is still to be determined (Nørskov et al., 2021).Some of the antimethanogenic and antimicrobial effects of brown macroalgae have been ascribed to the presence of phenolic compounds, including phlorotannins (Wang et al., 2008;Kadam et al., 2015;Vissers et al., 2018).These compounds are produced by macroalgae to mitigate environmental stressors such as changes in nutrient availability, UV radiation or interactions with other organisms or the abiotic environment (Tierney et al., 2013;Li et al., 2017).Hence, the ensiling of S. latissima was believed to increase the concentration of bioactive compounds as the process was thought to induce a stress response in the alga tissue.The in vitro results imply that the contents of antimethanogenic compounds were not high enough in the 3 macroalgae to exert a CH 4 mitigating effect or that the macroalgae contained ad-ditional compounds with a broader antimicrobial effect also because rumen degradability of DM and OM was suppressed.This could have masked or exceeded any CH 4 mitigating effects when relating CH 4 emission to dDM or dOM.Moreover, high concentrations of certain minerals in the macroalgae such as arsenic (Forsberg, 1978), cadmium (Jalc et al., 1994), and iodine (Oldick and Firkins, 2000) may also have had a suppressing effect on microbial growth and hence fermentation.

Effect of Dietary Supplementation with Macroalgae on Feed Digestibility
In our in vivo study, DMI was unaffected by the inclusion of up to 4% macroalgae in dietary DM.Similar findings have been reported by Kidane et al. (2015) and Antaya et al. (2019) when A. nodosum was fed to cows, and by Marín et al. (2009) and Gülzari et al. (2019) when Sargassum spp.and S. latissima, respectively, were fed to sheep.Likewise, total-tract digestibility of DM were unaffected by the macroalgae supplementation in our trial and in the studies by Gülzari et al. (2019), Marín et al. (2009), and Kidane et al. (2015).Although total-tract digestibility of DM and OM was similar across treatments, there were several indications in our study that A. nodosum from the NOD diet affected the nutrient digestibility negatively in different segments of the gastrointestinal tract.The in vitro studies showed that addition of A. nodosum to both basal feeds at higher inclusion rates than in the NOD diet fed to cows reduced overall rumen degradability of DM and OM.This was also associated with lower NH 3 concentrations in the liquid after 48 h of fermentation.The same tendency for a lowered rumen NH 3 concentration was also observed in the cows fed the NOD diet, indicating a lowered rate of feed protein degradation in the rumen (Herremans et al., 2020).However, rumen CP digestibility was unaffected by treatments.The significantly lower concentration of urea in milk and plasma from cows fed the NOD diet may directly be caused by this change in rumen fermentation.
Phlorotannins from macroalgae have been reported to exhibit protein and fiber-binding effects similar to what has been documented in tannins from terrestrial plants, and with highest affinity for the protein fraction (Le Bourvellec and Renard, 2012;Belanche et al., 2016;Vissers et al., 2018).The binding of tannins to feed proteins, ruminal microbiota, or microbial enzymes results in formation of molecular complexes that are poorly degraded under the anaerobic conditions in the rumen, hence leading to reduced ruminal digestibility of CP (Waghorn, 2008).In the abomasum, the low pH resulting from secretion of hydrochloric acid may induce dissociation of any formed phlorotannin-protein complexes.However, as the pH rises thought out the small intestine, the complexes might re-associate, and phlorotannins may bind not only to feed protein but also endogenous proteolytic enzymes excreted to the small intestine (Waghorn, 2008).This may explain the significantly lower small intestinal and total-tract digestibility of CP when cows were fed the NOD diet.Binding of phlorotannins to fiber or suppression of cellulolytic bacteria could also contribute to explain the tendency (P = 0.09) for a lower total-tract NDF digestibility when cows were fed the NOD compared with MUT diet (Wang et al., 2008;Pandey et al., 2022).Wang et al. (2008) speculated that cellulolytic bacteria may be more sensitive to phlorotannins and other phenolic compounds than amylolytic bacteria.However, the strongest tendency for dietary effect on rumen fermentation in our study was observed for starch, where the NOD diet tended to have a lower ruminal starch digestibility compared with the other diets (P = 0.08).This decrease in ruminal starch digestibility was also reflected in a lower proportion of propionate in total VFA in NOD fed cows, similar to the expected outcome upon reduction of dietary contents of starch (Lechartier and Peyraud, 2011).These results imply that there must be differences in contents and potency of algae bioactive components interfering with digestive processes between species because no negative effects on digestibility of CP, NDF, and starch were observed in cows fed the LAT and MUT diets in contrast to when the NOD diet was fed.Interestingly, all the diets with inclusion of macroalgae had higher large intestinal digestibility of most of the measured nutrients compared with the CON diet.However, it should be noted that ruminal NDF digestibility was higher for CON, MUT, and NOD than total-tract NDF digestibility.Similar problems have been reported in other studies using multicannulated cows (Lund et al., 2007;Brask et al., 2013;Olijhoek et al., 2016) and different types of duodenal cannulas (Stensig and Robinson, 1997).

Effect of Dietary Supplementation with Macroalgae on Milk Production
Milk production and composition were generally unaffected by dietary treatments.Similar findings were reported by Kidane et al. (2015) and Qin et al. (2023) when A. nodosum and S. latissima, respectively, were supplemented to dairy cows.The only changes in the present study were observed for cows on the NOD diet, where the level of urea in milk was increased, as previously discussed, and proportions of C16:0 and PUFA in milk FA were reduced and increased, respectively.However, numerically the differences in the proportion of C16:0 were minor.Larger differences were observed for the proportion of PUFA in milk with NOD having the highest proportion.Polyunsaturated FA are essential nutrients, which cannot be synthesized by mammals, and hence they are derived from the diet (van Ginneken et al., 2011).Crude fat content in A. nodosum of up to 4.5% of DM has been reported by van Ginneken et al. (2011) with as high a proportion of PUFA in total FA as 86%.In comparison, Marinho et al. (2015) and Santos et al. (2020) reported lower crude fat contents of S. latissima and S. muticum, respectively, and only around half of FA was PUFA.In the rumen, biohydrogenation of dietary unsaturated FA into saturated FA is extensive but usually incomplete and absorbed PUFA and intermediate metabolites can be incorporated into milk fat after absorption (Toral et al., 2018).The possible higher intake of PUFA in NOD fed cows may explain the higher proportion of PUFA in total milk FA, but altered patterns of rumen biohydrogenation could also play a role as discussed above.

Effect on Dietary Supplementation with Macroalgae on Indicators of Animal Health
High dietary contents of certain minerals such as iodine can lead to excessive excretion of these minerals into milk, which may have a negative effect on consumer nutrition and health (Antaya et al., 2019;Newton et al., 2021;Qin et al., 2023).All macroalgae-containing diets exceeded the daily maximum total iodine limit (50 mg I/d) set by the US Food and Drug Administration regulations (4,356, 143, and 1,246 mg/d for LAT, MUT, and NOD, respectively; NRC, 2021).Moreover, the LAT and NOD diet also exceeded the maximum allowed level defined by the European Commission (5 mg/kg mixed feed with a DM content of 88%; iodine content: 184, 2.86, and 49.8 mg/kg mixed feed DM for LAT, MUT, and NOD, respectively;European Commission, 2015) and therefore, it would not be allowed to fed the rations with the current inclusion level of macroalgae in commercial herds.An excessive iodine intake can alter thyroid function (Chung, 2014), and hence the production of the thyroid hormones triiodothyronine and T4, because iodine is an essential element in their synthesis.Although ensiled S. latissima had a very high concentration of iodine, and the level of T4 was highest in LAT fed cows, the level remained well within the normal range (54-110 nmol/L) for healthy cattle (Jackson and Cockcroft, 2002).There also were no signs of any dietary induced changes in plasma concentrations of the pituitary hormone, TSH, subject to negative feedback regulation induced by dietary induced changes in circulating levels of T4 (Chiamolera and Wondisford, 2009).However, it should be stressed that this was a short-term feeding trial and it cannot be ruled out that more substantial changes in thyroid function could occur after long-term exposure to the LAT diet.The European Commission also has restrictions on the contents of potentially harmful minerals in individual feedstuffs.Hence, macroalgal biomass is allowed to maximum contain 40 mg/kg DM of arsenic, 1 mg/kg DM of cadmium, 15 mg/kg DM of lead, and 0.01 mg/ kg DM of mercury.The accumulation in alga tissue is affected by the concentration in water during growth as macroalgae take up critical minerals from water more efficiently than terrestrial biomasses (Romera et al., 2006).Thus, all the macroalgae in the present study exceeded the limit for at least one mineral as the LAT and MUT diets exceeded the limit for arsenic while the LAT and NOD diets exceed the limit for cadmium (European Commission, 2013, 2019).Thus, none of the macroalgae would be allowed to be fed to dairy cows in commercial herds irrespective of the inclusion level.
Heavy metals such as arsenic and cadmium, which are also naturally present in macroalgae, have been reported to cause liver damage in cattle (Bertin et al., 2013;Lane et al., 2015).Serum levels of hepatic enzymes such as AST, g-GT and GLDH are often used as indicators of hepatic injury or disease (Whitfield, 2001).On all dietary treatments, cows had serum levels of g-GT and albumin slightly above the normal reference range indicated by the commercial laboratory responsible for the analyses.The same applies for GLDH, except when cows were fed the LAT diet.Nevertheless, serum levels of all indicators of hepatic function and health for all treatment were within the range reported for clinically healthy cows in other studies (Cozzi et al., 2011;Imhasly et al., 2014;Smuts et al., 2019).Arsenic from herbicide residues has been reported to also cause ulcers and necrosis of the rumen wall of sheep on pasture (Gonçalves et al., 2017).Additionally, in a study by Muizelaar et al. (2021), pronounced abnormalities of rumen wall papillae were identified in cows fed the red macroalga A. taxiformis.Thus, evacuations of rumen contents were performed in the present study after each period shift to allow inspection of the rumen epithelium, and in no case any indications of rumen papilla atrophy or rumen ulcers were observed.Hence, the inclusion of 2% or 4% of the 3 macroalgae in dietary DM did not seem to have negative implications for thyroid, liver, or rumen function or health for the duration the macroalgae were fed to the cows.However, it should again be stressed that this study only investigated the short-term effects of the 3 macroalgae on animal health and on a few animals were included in the study, and that none of the diets from the present study will be allowed to be fed to dairy cows in commercial herds in the US or the European Union.

CONCLUSIONS
None of the 3 Northern procurable macroalgae induced reductions in CH 4 emissions when fed to dairy cows at an inclusion rate of 2% (MUT) or 4% (LAT and NOD) of dietary DM.Moreover, all the macroalgae exceeded the maximum allowed content for feedstuffs or for the total ration, set by the European Commission, for at least one potentially harmful mineral and thus, are not possible to feed in commercial herds.The inclusion of Sargassum muticum and ensiled S. latissima resulted in no negative effects on the overall digestibility of dietary components in different segments of the gastrointestinal tract or on milk production for the duration of the short-term experiment, while the dietary addition of A. nodosum was associated with negative effects on the CP digestibility.
Jackson and Cockcroft (2002).Table7.In vitro degradability and the associated production of VFA and ammonia concentration after 48 h of incubation, and methane production after 24h of Superscripts are only shown when the overall F test for treatment resulted in P < 0.05.Values within the same column with different superscripts differ (P < 0.05). 1 0.5 g DM of pure macroalgae: LAT = ensiled Saccharina latissima; MUT = Sargassum muticum; and NOD = Ascophyllum nodosum. 2 0.5 g DM of TMR from the in vivo experiment: CON = control diet, LAT = 4% ensiled S. latissima (LAT), MUT = 2% S. muticum, and NOD = 4% A. nodosum.All inclusion rates were on DM basis.3 0.5 g DM of corn silage with addition of 0.1 g DM macroalgae: CS = corn silage; LAT = Corn silage and ensiled S. latissima; MUT = corn silage and S. muticum; and NOD = corn silage and A. nodosum.
Thorsteinsson et al.: METHANE EMISSION FROM COWS FED BROWN MACROALGAE Thorsteinsson et al.: METHANE EMISSION FROM COWS FED BROWN MACROALGAE

Table 1 .
Thorsteinsson et al.: METHANE EMISSION FROM COWS FED BROWN MACROALGAE Ingredient and chemical composition of a control diet without any macroalgae and the same diet diluted with one of 3 macroalgae Thorsteinsson et al.: METHANE EMISSION FROM COWS FED BROWN MACROALGAE Thorsteinsson et al.: METHANE EMISSION FROM COWS FED BROWN MACROALGAE

Table 2 .
Thorsteinsson et al.: METHANE EMISSION FROM COWS FED BROWN MACROALGAE Daily gas exchange in dairy cows fed either a control diet or the same diet diluted with one of 3 macroalgae 1 CON = control diet, LAT = 4% ensiled Saccharina latissima (LAT), MUT = 2% Sargassum muticum, and NOD = 4% Ascophyllum nodosum.All inclusion rates were on DM basis.

Table 3 .
Thorsteinsson et al.: METHANE EMISSION FROM COWS FED BROWN MACROALGAE Milk production and proportions of individual fatty acids (FA) as % of total FA of dairy cows fed either a control diet (CON) or the same diet diluted with one of 3 macroalgae

Table 4 .
Intake and apparent digestibility of nutrient entry in the different segments of the gastrointestinal tract in dairy cows fed either a control diet or the same diet diluted with one of 3 macroalgae

Table 5 .
Ruminal pH and redox potential, concentration of total VFA, and proportions of the individual VFA in dairy cows fed either a control diet or the same diet diluted with one of 3 macroalgae

Table 6 .
Thorsteinsson et al.: METHANE EMISSION FROM COWS FED BROWN MACROALGAE Metabolic and health status indicators of dairy cows fed either a control diet or the same diet diluted with one of 3 macroalgae OH-butyrate; NEFA = nonesterified fatty acids; T4 = thyroxine hormone; TSH = thyroid-stimulating hormone; AST = aspartate aminotransferase: g-GT = gamma-glutamyl transferase; and GLDH = glutamate dehydrogenase.Reference values were obtained from the commercial laboratory Laboklin Laboratory for Clinical Diagnostics GmbH and Co. KG (Bad Kissingen, Germany). a-b Thorsteinsson et al.: METHANE EMISSION FROM COWS FED BROWN MACROALGAE