The macroalgae species were selected to represent the major groups of marine macroalgae (red, brown, green) based on their demonstrated ability to decrease enteric CH₄ andimprove fermentation [13] and are listed in Table 1. The biomass was sourced from either large scale culture at the Centre for Macroalgal Resources and Biotechnology (MACRO) James Cook University (JCU) in Townsville QLD (19.33˚S; 146.76˚E), Pacific Reef Fisheries in Ayr QLD (PRF; 19.58˚S, 147.40˚E), or intertidal flats near Townsville Qld at Nelly Bay (NB; 19˚16'S; 146˚85'E) with GBRMPA permit GO2/20234.1, or Rowes Bay (RB; 19˚23'S; 146˚79'E) with DPIF permit 103,256. The individual algae and sources were Caulerpa taxifolia ([M.Vahl] Agardh 1817; NB), Cladophora patentiramea ([Montagne] Kützing 1849; PRF), Cystoseria trinodis ([Forsskål] Agardh 1820; NB), Dictyota bartayresii (Lamouroux 1809; RB), Padina australis (Hauck 1887; RB), Sargassum flavicans ([Mertens] Agardh 1820; NB) and Ulva ohnoi (Hiraoka and Shimada 2004; MACRO).

Macroalgae were rinsed and centrifuged in a commercial washing machine at 1000 rpm for 5 min at ambient temperature to remove excess water. The biomass was freeze- dried (VirTis Benchtop 2K; Warminster PA, USA) at −55˚C and 120 µbar for minimum 48 h, then milled to pass a 1 mm sieve and stored at −20˚C. The basal substrate was Rhodes grass (Chloris gayana) with neutral detergent fibre (NDF) of 645 g∙kg⁻¹ dry matter (DM) and acid detergent fibre (ADF) of 315 g∙kg⁻¹ DM. The DM content of the various biomass was determined by drying at 105˚C to constant weight and the OM content was determined by the combustion for 8 h at 550˚C [14] . Crude protein and

a. MJ∙kg⁻¹ DM.b. Parameters presented as g∙kg⁻¹ unless otherwise stated. DM, dry matter; OM, organic matter; CP, crude protein.

gross energy quantification were determined as previously described by [13] . Crude protein of the macroalgae was based on total nitrogen (wt %) content using a nitrogen factor of 5 for the macroalgae [15] and 6.25 for the Rhodes grass.

Rumen fluid inoculum (RF) was collected from four fistulated Brahman steers (Bos indicus; LW 460 ± 20 kg) fitted with 10 cm Bar Diamond (Parma, OH, USA) rumen cannulas. The steers were maintained at the College of Public Health, Medical and Veterinary Sciences of JCU according to current guidelines [16] and approved by the CSIRO animal ethics committee (A5/2011). The steers were maintained on Rhodes grass hay ad libitum for 6 months before the collection of RF, which was extracted 2 h after morning feeding by sampling from four quadrants of the rumen and hand squeezing to completely fill pre-warmed 1 L stainless steel thermos flasks.

The RF was pooled and immediately processed by filtration through a 0.5-mm sieve and combined with incubation buffer (GVB) [17] at a ratio of 1:4 (RF:GVB). Through- out the inoculation process the RF buffer fermentation media (RFB) was maintained at 39˚C and continuously mixed at 30% - 35% of maximum vortex speed (Major Science SWB 20 L⁻³; Saratoga, CA, USA) to ensure RFB homogeneity between individual fermentations. Anaerobic conditions were maintained with high purity N₂ (HPN₂; BOC, Wetherill Park NSW, AUS). A Dose-It pump (Integra Biosciences, Hudson NH, USA) was used to aspirate 125 mL of RFB into each incubation bottle containing the macroalgae and Rhodes grass substrates. The substrate and macroalgae were preloaded prior to RF collection into 250 mL Simax incubation bottles (Sázava, CZE). Each bottle was purged HPN₂, capped, and warmed to 39˚C. The bottles were sealed with an Ankom RF1 gas production module (Macedon NY, USA), placed in mixing incubators (Ratek OM11; Boronia VIC, AUS), and maintained at 39˚C and oscillating at 85 RPM for 72 h.

Experiment 1 was conducted to rank macroalgae according to their antimethanogenic potency and effects on rumen fermentation parameters. This was accomplished using in vitro fermentation batch culture (Ankom) to determine effects of macroalgae inclusion on total gas (TGP), CH₄, and VFA production, and IVD-OM. A series of four incubation periods were completed and consisted of 1.0 g of Rhodes grass substrate as control and appropriate quantities of each macroalga to achieve a dose concentration of 5% of the substrate OM according to the biomass composition described in Table 1. Fermentations were also completed as controls (no macroalgae), a positive-control (P- control) of Asparagospis at 2%, and RFB blanks. The inclusion of the macroalgae at 5% was determined as an arbitrary feasible level of feeding for livestock and based on results previously described for inclusion approaching 20% [13] . The inclusion of Asparagopsis at 2% was set based on previously determined optimum dose for this seaweed in rumen fermentations [10] [11] . The seven macroalgae and Asparagopsis were randomly assigned to the four incubation periods (n = 3), and controls and blanks were included in all periods (n = 4). Within each incubation period, there was duplication of each macroalga, and controls and blanks at each sampling time point (12, 24, 48, and 72 h).

Experiment 2 was conducted to determine if the seven macroalga when combined with Asparagopsis studied in Experiment 1 would benefit in vitro rumen fermentation. This was accomplished using the same fermentation techniques as Experiment 1. A series of five incubation periods were completed and consisted of 1.0 g of Rhodes grass substrate as control and appropriate quantities of each macroalga to achieve a dose concentration of 5% (OM basis) in combination with Asparagopsis at 2% according to the biomass composition described in Table 1. Fermentations were also completed as controls (no macroalgae), a P-control of Asparagospis at 2%, and RFB blanks. The seven combinations and Asparagospis were randomly assigned to the five periods (n = 3), and controls and blanks were included in all periods (n = 5). Within each incubation period, there was duplication of each macroalgae combination, and controls and blanks at each sampling time point (12, 24, 48, and 72 h).

The fermentation methods used in this study were similar to [18] but using the Ankom RF1 system and gas analysis as described by [10] . Pressure accumulation in the incubation bottles was measured continuously and recorded every 20 min. The maximum fermentation pressure inside the incubation bottle was set to 3 psi which, when exceeded, caused venting for 250 ms and the pressure change accounted in the cumulative pressure recording. Gas pressure was measured every 60 s and cumulative pressure was recorded at 20 min intervals. The TGP was expressed as mL of gas produced per gram of substrate OM (mL∙g⁻¹ OM) by application of the natural gas law to the accumulation of the recorded gas pressure while accounting for individual bottle volume.

In vitro CH₄ production was quantified in time series at multiple time points of 12, 24, 48 and 72 h during the fermentation. The mL of CH₄ g⁻¹ of substrate OM was estimated using concentrations in headspace at the time series points while assuming homogeneity of headspace gas. The headspace samples were collected into 10 mL Labco Exetainer vials (Lampeter, GBR) and quantified using gas chromatography (GC) according to parameters described by [10] . Concentrations of CH₄ in headspace gas were converted to mL∙g⁻¹ substrate OM based on TGP at the relative time series points and assuming headspace homogeneity at system venting [19] .

The IVD-OM and VFA production was determined as described by [10] and quantified to coincide with CH₄ determinations at each time series point. Each fermentation was chilled to cease bacterial activity and the in vitro fluid (IVF) was then vacuum filtered through a Duran No. 1 porosity glass fritted crucible containing a 0.5 cm layer of sand filtration aid. The pH was measured on the filtrate. Crucibles and wet residue were oven-dried to constant weight at 105˚C for DM determination. Residue OM was determined as loss on ignition in a muffle furnace at 550˚C for 8 h [14] .

The TVFA accumulated in the IVF were quantified after termination of fermentation. Preparation of IVF prior to GC analysis consisted of addition 4 mL of IVF to 1 mL of 20% metaphosphoric acid containing 11 mM of 4-methylvaleric acid (Sigma-Al- drich; Castle Hill NSW, AUS) providing 2.2 mM internal standard. The samples were mixed and stored at −20˚C until a 1.5 mL subsample was centrifuged for 15 min at 13,500 rpm and 4˚C (Labnet Prism R; Edison NJ, USA). The supernatant was filtered through 0.2 µm PTFE syringe tip filters (Agilent; Santa Clara, CA, USA) and quantified by GC according to parameters described by [10] .

Two-factor repeated-measures permutational analysis of variance (PERMANOVA) was used to test for significant differences in TGP, CH₄ production, and IVD-OM over time. A one-factor PERMANOVA was used to test for significant differences in the production of VFA between treatments (fixed factor) using Primer 6 (version 6.1.13; [20] statistical software and PERMANOVA+ (version 1.0.3) [21] . For PERMANOVA, Bray-Curtis similarity matrices were produced using the untransformed raw data and dummy variables (0.0001) were used to account for zero values. The P-values were calculated from 999 (TGP) and 9,999 (CH₄, IVD-OM, and VFAs) random permutations. Pair-wise a posteriori comparison was used to determine significant groupings, where applicable. For PERMANOVA, differences were considered significant if P < 0.05. The TGP data were also fitted with generalized additive models (GAM) using cubic regression spline smoothers to predict the relationship and examine differences between TGP over time. The generalized additive models were produced using the mgcv package within the R language (version 3.0.1) [22] . Goodness-of-fit of the individual smoothers was quantified using the hydroGOF package within the R language and was assessed from the proportion of variance in the data that was accounted for by the model (r²) [23] .

Based on predictions of the GAM, which had a high goodness-of-fit (r² > 0.97), there was a significant difference (P < 0.001) in TGP in the fermentations over 72 h (Figure 1). In addition, the TGP reached its plateau approaching 48 h in all fermentations and the Asparagopsis induced a lower TGP rate compared with all other treatments (P < 0.001). There was no significant difference in cumulative TGP between the seven macroalgae with inclusion at 5% (OM basis) and the Rhodes grass control after 72 h of fermentation. The TGP for all macroalgae other than Asparagopsis ranged between 171 and 176 mL∙g⁻¹ OM, representing an insignificant 2.4% - 4.8% reduction in TGP com-

pared to the Rhodes grass control (180 mL∙g⁻¹ OM). However, the Asparagopsis at 2% had significantly less TGP (127 mL∙g⁻¹ OM) representing a 26% - 29% reduction compared to all other macroalgae and controls.

The CH₄ production as affected by inclusion of the seven macroalgae other than Asparagopsis was not altered significantly compared to the Rhodes grass control during the course of the 72 h incubations (Figure 1). The rate of CH₄ production as defined by time series headspace sampling was variable in the same manner as for TGP. The only significant reduction in CH₄ was induced by Asparagospis in the same way as previously demonstrated [10] and no detectable CH₄ was produced. The CH₄ production from the other fermentations ranged from a low of 14 mL∙g⁻¹ OM to a high of 16 mL∙g⁻¹ OM for U. ohnoi and the Rhodes grass control, respectively.

A typical pattern of IVD-OM increasing with time was demonstrated with or without macroalgae (Figure 1), however, the rate of digestion of the substrate varied between treatments in the first half of the fermentation period (P = 0.004). Asparagospis induced the earliest onset of substrate degradation in the same fashion as demonstrated by [10] for Asparagopsis at dose concentrations ≤ 5% of substrate OM. The C. patentiramea was the slowest to achieve maximum rate of IVD-OM. Approaching 48 h, IVD-OM in all the fermentations coalesced, and after 72 h had similar IVD-OM ranging from a low of 70% to a high of 72% for C. patentiramea and the Asparagopsis, respectively. In the absence of macroalgae the IVD-OM for the Rhodes grass control was 72%.

In Experiment 1 the series of in vitro fermentations of the seven macroalgae or Asparagopsis P-control did not induce significant differences between any of the treatments or Rhodes grass control for TVFA (P = 0.093) or propionate (P = 0.098). However, fermentations including the Asparagopsis were significantly lower for acetate (P = 0.002) and higher for butyrate (P = 0.004) compared to the other fermentations (Table 2). A significant increase in propionate and butyrate with 2% Asparagopsis inclusion has been previously reported [10] , however this feature was slightly dampened in the present study for propionate but confirmed the previously observed butyrate concentrations. Compared to the Rhodes grass control, after 72 h of fermentation the macroalgae induced marginal reductions in TVFA on a molar concentration basis from a low of 9% to a high of 23% for D. bartayresii and C. patentiramea, respectively, and Aspara- gopsis reduced TVFA by 22%. The production of acetate was marginally reduced by a low of 10% to a high of 24% by inclusion of D. bartayresii and C. patentiramea, respectively, and a significant 44% with Asparagopsis, respectively. Conversely, propionate was marginally decreased by a low of 5% to a high of 24% with inclusion of D. bartayresii and C. patentiramea, respectively, and increased by 9% with Asparagopsis. The production of butyrate was marginally decreased by a low of 11% to a high of 37% for D. bartayresii and C. patentiramea, respectively, and significantly increased by 76% with Asparagopsis.

There were not adequate differences in TGP, CH₄, or VFA production to justify well defined ranking for the seven macroalgae evaluated. Clearly, none of the seven macro-

a. Significant at P < 0.05.

algae at 5% of substrate OM compares to Asparagopsis at 2% for CH₄ abatement without detriment to IVD-OM during in vitro fermentation. Therefore based solely on numerical CH₄ abatement the ranking order from lowest to highest CH₄ produced is as follows: 1) U. ohnoi; 2) C. taxifolia; 3) P. australis; 4) C. trinodis; 5) D. bartayresii; 6) S. flavicans; 7) C. patentiramea.

Similar to the results of Experiment 1, based on predictions of the GAM, which had a high goodness-of-fit (r² > 0.98), there was a significant difference (P < 0.001) in TGP in the fermentations over 72 h (Figure 2). In addition, the TGP reached its plateau approaching 48 h in all fermentations, however, in contrast with Experiment 1, the combination treatments were equivalent to the Asparagopsis P-control at 2% of substrate OM in the reduction of TGP, but all were significantly different from the Rhodes grass control (P < 0.001) after 72 h of fermentation. The TGP for the combinations ranged between 124 and 136 mL∙g⁻¹ OM representing a significant 28% - 34% reduction in TGP compared to the Rhodes grass control (188 mL∙g⁻¹ OM). The Asparagopsis induced a TGP of 134 mL∙g⁻¹ OM and therefore a 28% reduction, similar to the combinations demonstrating an overwhelming effect of Asparagopsis when combined with other macroalgae.

The CH₄ production as affected by inclusion of the seven macroalgae combined with Asparagopsis was not different compared to the Asparagopsis alone during the course of the 72 h incubations (Figure 2). In the absence of macroalgae the fermentation initiated production of CH₄ immediately after inoculation with RFB, however, the inclusion of the macroalgae combinations completely inhibited CH₄ until 48 h when trace

amounts were measured for the S. flavicans combination (0.2 mL∙g⁻¹ OM) representing 98% inhibition. After 48 h of fermentation S. flavicans and D. bartayresii combined with Asparagopsis accumulated small amounts of CH₄ such that at 72 h of fermentation these two combinations had reduced CH₄ by 95% (0.8 mL∙g⁻¹ OM) compared to the Rhodes grass control (16.0 mL∙g⁻¹ OM). The Asparagopsis alone reduced CH₄ by 99% (0.2 mL∙g⁻¹ OM) after 48 h and by 96% (0.6 mL∙g⁻¹ OM) at 72 h.

In the same way as Experiment 1 the typical pattern of IVD-OM increasing with time was demonstrated with or without macroalgae combinations (Figure 2). However, in Experiment 2 IVD-OM induced by the macroalgae combined with Asparagopsis was not as variable throughout the 72 h in vitro fermentations. Other than the Rhodes grass control, all fermentations contained 2% Asparagopsis which negated effects of the other macroalgae in the combinations. All the fermentations coalesced from onset of fermentation and after 72 h of fermentation the IVD-OM ranged from a low of 68% to a high of 72% for the D. bartayresii combined with Asparagopsis and Rhodes grass control, respectively.

In Experiment 2, in vitro fermentations with inclusion of the seven macroalgae combinations or Asparagopsis induced significant reductions in TVFA (P = 0.002) and acetate (P < 0.001), and significant increases in propionate (P < 0.001) and butyrate (P < 0.001) compared with the Rhodes grass control (Table 3). However, there was no significant difference between the Asparagopsis alone and the combinations, therefore improvements in VFA profiles due to combining Asparagopsis with any of the seven candidate macroalgae was not demonstrated. Compared with the Rhodes grass control, the macroalgae combinations induced reduction in TVFA on a molar concentration

a. Significant at P < 0.05. ‡Each of the seven macroalgae represents half of a combination pair with.

basis from a low of 7% to a high of 12% and 14% by inclusion of the P. australis and C. taxifolia combined with Asparagopsis and Asparagopsis alone, respectively. The production of acetate was reduced by a low of 24% to a high of 29% and 31% by inclusion of S. flavicans and C. taxifolia combined with Asparagopsis, and Asparagopsis alone, respectively. Conversely, propionate was increased by a low of 21% to a high of 26% and 21% by inclusion of the P. australis and C. taxifolia combined with Asparagopsis, and Asparagopsis alone, respectively. The production of butyrate was increased by a low of 75% to a high of 91% and 72% for the U. ohnoi and C. taxifolia combined with Asparagopsis, and Asparagopsis alone, respectively.