Ruminal CO 2 holdup monitoring, acidosis might be caused by CO 2 poisoning

The ruminal buffering system is composed of bicarbonate (HCO 3) and dissolved CO 2 (dCO 2 ). While low pH indicates high dCO 2 formation, the pH scale is a ratio between acids and bases in a solution, i.e. HCO 3-and dCO 2 , and fail to provide individual component concentrations. For instance, modern feeding practices can reduce CO 2 gas fugacity from the ruminal fluid or "CO 2 holdup". Under those conditions, not only dCO 2 can reach critical concentrations, but the buffering system might favour HCO 3-formation resulting in normal ruminal pH values, the quotient, regardless of the harmful dCO 2 accumulation. Consequently, subacute ruminal acidosis (SARA), traditionally associated with low or variable pH, might be triggered by CO 2 holdup. This observational study aimed to continuously monitor ruminal dCO 2 and characterised CO 2 holdup within the ruminal fluid targeting the specific infrared signal of dCO 2 with an attenuated total reflectance infrared (ATR-IR) spectrometer. Three lactating dairy cattle were longitudinally exposed to diets designed to elevate both ruminal dCO 2 and SARA risk. Indwelling pH sensors and ruminal fluid samples served as references for dCO 2 analysis, while a categorical analysis detected CO 2 holdup from the output of the ATR-IR sensor. Milk yield, milk components, and feed intake supported the known positive role for high dCO 2 in rumen function. However, SARA was associated with ruminal CO 2 holdup, suggesting that prolonged exposure to critical dCO 2 concentrations during extended postprandial periods might trigger SARA. Continuous dCO 2 monitoring with the proposed methodology and analysis may offer a valuable tool for optimising rumen function and prevent SARA risk.

Traditionally, the ruminal short-chain fatty acid (SCFA) concentrations are blamed for pH variations, however the primary driver of these fluctuations is the CO2 buffer system (Turner and Hodgetts, 1955).SCFA concentrations are seemingly constant (Dijkstra et al., 1993), are dominated by the bases not the acid forms of SCFA, pKa ~4.7, even lactic acid that is commonly associated to ruminal acidosis is mainly found as lactate, the base, pKa ~3.8 (Russell and Hino, 1985).As bases, SCFA play only a buffering role when the pH is below 5.4 and HCO3 -is depleted (Turner and Hodgetts, 1955;Hille et al., 2016).Moreover, the threshold for ruminal acidosis, pH 5.5 (Nocek et al., 2002), the rumen fluid equilibrium or pKa' 6.1 (Hille et al., 2016) and the pH scale range, 5 to 7 (De Veth and Kolver, 2001), coincides with the CO2 species equilibrium described by the Bjerrum equations (Buchholz et al., 2014).In fact, the relationship between SCFA formation and the pH scale decline may simply be a consequence of dCO2 released during fermentation (Wolin, 1960;Dijkstra et al., 2012).Therefore, increased SCFA production leads to greater dCO2 formation, with a concomitant ruminal pH decline.
Another aspect explaining the spurious relationship between pH, SCFA, and dCO2 is the capnophilic nature of ruminal bacteria (Gladstone et al., 1935;Dehority, 1971).Ruminal succinate and lactate-producing bacteria thrive in high dCO2 environments (Wright, 1960;Samuelov et al., 1991).Under these conditions, ruminal propionate production also increases, as succinate and lactate are the main precursors (Mizrahi et al., 2021).Consequently, a rumen environment rich in dCO2 enhances propionate production, manifested as a low acetate to propionate ratio (A/P ratio) and low pH (Lana et al., 1998;Russell, 1998).Moreover, high ruminal dCO2 stimulates lipopolysaccharides (LPS) formation in Streptococcus bovis (Dain et al., 1956;Cheng et al., 1976), a major factor described in the pathogenesis of ruminal acidosis (Li et al., 2012).Similarly, elevated lactate during ruminal acidosis may result from bacteria favouring the acrylate pathway, which does not involve decarboxylation reactions that can be limited by high ruminal dCO2 (Wright, 1960;Counotte and Prins, 1981;Blombach and Takors, 2015).Therefore, the common clinical signs of ruminal acidosis may be a consequence of high dCO2 concentrations.
Therefore, the variable pH bouts observed during SARA onset may be consequential to CO2 holdup, as it might also increase HCO3 -formation.
Ruminal dCO2 and CO2 blood pools rapidly equate (Whitelaw et al., 1972;Veenhuizen et al., 1988) due to the positive ruminal gradient with the blood, ~60 vs. 2.5 mM (Turner and Hodgetts, 1955;Kohn and Dunlap, 1998) and the preferential use of ruminal CO2 for SCFA uptake (Ash and Dobson, 1963;Rackwitz and Gabel, 2018).Consequently, the proposed timescale and risk for SARA onset might involve prolonged exposure of the ruminal epithelium to critical dCO2 concentrations and SARA signs might be caused by CO2 poisoning.
Ruminal dCO2 concentrations are partially characterised and CO2 holdup has never been observed in situ (Chou and Walker, 1964b, a;Laporte-Uribe, 2019;Wang et al., 2019).This observational study aimed to use a wired attenuated total reflectance infrared spectrometer (ATR-IR) that detected the distinctive IR signal of dCO2 at 4.27 µm within the ruminal liquor (Schädle et al., 2016).The more reliable ATR-IR technique and output evaluation might help us to confirm the following hypotheses (first) to confirm the ruminal dCO2 range, (second) to unveil the relationship between CO2 species and pH, and (third) to disclose the role of CO2 holdup on disease (SARA) and rumen function.

Ethical and experimental guidelines
The experimental protocols were approved and licensed by the Animal Care and Ethics Committee of Wageningen University and Livestock Research, WUR Dairy Campus, according to the Experiment in Animals Act, WOD, The Netherlands with permit AVD401002015298.The care of all cattle involved in this experiment adhered to the guidelines of the ethical committee for the use of fistulated cattle in tied stall facilities.

The experimental setup
The diets and cattle performance were described previously (Laporte-Uribe, 2019).In brief, three fistulated lactating dairy cattle (Bar Diamond Inc., Ida., USA; 10 cm diameter), ~100 days in milk (DIM), were housed in tied stalls.The cattle were milked twice daily with ad libitum access to drinking water.Three total mixed ratio (TMR) diets were prepared daily using an automatic feeding system (Trioliet Feeding Technology, Oldenzaal, The Netherlands) and were served in equal parts, three times per day.The SARA-prone diets were a low physically effective neutral detergent fibre (Low-peNDF), a high ruminally degradable starch (High-RDS) and a combination of both (Combined); please see Laporte-Uribe (2019) for details on the formulation.All cattle were fed the same diet simultaneously for two weeks (run): the second week was for ruminal sampling and sensor deployment.The cattle had a three-days rest period between runs on a standard production TMR diet (Dairy Campus, Wageningen University).Indwelling pH sensors and manual ruminal samples were used as references (Diagram 1).
Diagram 1.The longitudinal observational design aimed to continuously monitor ruminal dCO2 concentration.Total inorganic carbon (TIC) from rumen fluid samples and indwelling pH sensors were employed to corroborate the output of the attenuated total reflectance infrared (ATR-IR) sensor.The diets were the low physically effective neutral detergent fibre (Low-peNDF) in the 1 st run, the high ruminally degradable starch (High-RDS) in the 2 nd run, and a combination of both previous diets (Combined) in the 3 rd run.

Sensor deployment
The pH from the ventral ruminal sac was recorded every 15 sec for three days with indwelling pH sensors in all cattle (DASCOR, Inc., CA, USA).For continuously monitoring of the ruminal dCO2 concentrations in one random sentinel cow per run, a wired ATR-IR sensor, VS-3000/3000E Sensor System was employed (BevSense LLC, MA, USA, formerly VitalSensors Technologies LLC).The ATR-IR was placed into the ventral ruminal sac, and the dCO2 was recorded every 10 seconds for three days.The wire was exteriorised through the cannula, sealed to reduce CO2 losses, and connected to the sensor Management Station, VS-300 (BevSense LLC, MA, USA).
All pH sensors were calibrated before and after placement using a three-point calibration protocol (DASCOR, Inc., CA, USA).The ATR-IR sensor came calibrated for sensing dCO2 specific IR signal at 4.27 µm in liquids ranging from 0 to 273 mM with a resolution of 0.02 mM, a repeatability of 0.36 mM and an accuracy of 0.89 mM; see the product specification for details (BevSense LLC, MA, USA).Nevertheless, validation of the ruminal dCO2 values and range was advised using a three-point alignment protocol developed for steady fermentative processes (BevSense LLC, MA, USA).The following modified protocol was adopted due to the dynamic nature of the ruminal environment.

Ruminal fluid samples and calculations
The ventral ruminal sac fluid was manually sampled five consecutive times postprandially, feeding started at 07:00h (0.5 h, 1 h, 2 h, 4 h, 6 h), during the first three days of the experimental week in all cattle.The pH of the samples was recorded with a temperature-corrected handheld system (Seven2Go ProS8, Mettler-Toledo).Approximately 30 ml of rumen fluid was alkalised by the addition of 1 ml of 5 M sodium hydroxide (NaOH) solution and was frozen for subsequent total inorganic carbon (TIC) analysis (-20 °C).The goal was to retain TIC in HCO3 -form by increasing the pH of the sample (pH ~10), according to the protocols given by the reference laboratory (Buchholz et al., 2014).TIC was determined by gas chromatography at the Institute of Biochemical Engineering, University of Stuttgart.
Calculations of CO2 species.The ruminal dCO2 concentrations were computed from the TIC using the Bjerrum plot equation (Eq. 1) and described as the observed dCO2.The calculated dCO2 was derived from the TIC as if only HCO3 -was recovered.The calculated HCO3 -was derived from the average pH and dCO2 sensor reading for each minute in a day (1,440 records).
Raw values from the ATR-IR sensor were expressed in parts per million per 100 g of H2O (ppm/100 g H2O) and the following formulas were used to convert these values to millimole per litre (mM) of ruminal dCO2. and, where x is the ruminal dCO2 concentration in parts per million per 100 g of H2O (ppm/100 g of H2O) y is the dCO2 in milligrams per one hundred grams of water (mg/100 g H2O), and z is the dCO2 in millimoles per litre (mM).

Analysis and statistics
All values from the dCO2 and pH sensors were used in the development of the categorical analysis except for the records made one hour after deployment.A histogram method was used to detect outliers in the sensors' output (Gebski and Wong, 2007).The pH sensors yielded no outliers, and ATR-IR yielded only few values.Values for CO2 and HCO3 -from the ruminal manual samples were compiled together.All descriptive statistical analyses and graphics were conducted in Origin 2020 (Origin Lab Corporation, MA, USA).

Categorical analysis to observe ruminal CO2 holdup.
The area under the curve for ruminal pH (AUC, pH units per min) emphasises the duration of the acidotic bouts at specific thresholds (Nocek et al., 2002).AlZahal et al. (2007a) employed the cumulative time under the curve to define a cut-off point for half-day exposure.More recently, Villot et al. (2018) normalised ruminal pH recordings and described two optimal thresholds, 30 th and 50 th percentile, for SARA detection.Previously, a "categorical analysis" was proposed to observe ruminal pH in the New Zealand pastoral system (Gibbs and Laporte Uribe, 2009).Our assumption was that changes in sensor location, due to the mixing movements and by the influx and outflow of nutrients led to the recording of distinct pH values.Nevertheless, with sufficient "iterations", the pH category with the highest frequency was consistently identified, such as in several cattle, days, and short recording intervals (<15 seconds).The four categories for ruminal pH values were "Critical," (pH <5.4), "Acidic" (pH between 5.4 and 5.8), "Optimal" (pH between 5.8 and 6.4), and "Suboptimal" (pH> 6.4), reflecting the state of the art on ruminal pH effect.For instance, cattle with pH values lower than 5.4 and 5.8 for 3 to 5 h/d have a high risk of ruminal acidosis and SARA (Dohme et al., 2008;Villot et al., 2018).Bacterial protein synthesis and fibre digestion diminish when the pH falls below 5.8, which is also recognised as a sign of ruminal dysfunction (Russell, 1998;De Veth and Kolver, 2001).Values around 6.4 are in the upper range in cattle given a TMR and are optimal for fermentation in pasture-based diets (Russell, 1998;De Veth and Kolver, 2001).
To my knowledge this is the first time that continuously recordings of ruminal dCO2 have been performed, and thresholds for ruminal dCO2 function remain undefined.However, CO2 holdup can be identified by assigning a probability value derived from the normal cumulative distribution function (Eq.4).Accordingly, four categories for ruminal dCO2 were defined: "Low" for values below the 10 th percentile, "Normal" for values between the 10 th and 50 th percentiles, "High" for values between the 50 th and 90 th percentiles, and "Critical" for values above the 90 th percentile.
where "x" is the recorded dCO2 value, "µ" is the overall dCO2 mean, and "σ" is the overall standard deviation for the experiment.
To comprehend the daily variation in these parameters and monitor CO2 holdup, the day was divided into discrete segments of 10-min, e.g., 0:00, 0:10..., 23:50, or 144 segments.The interval was visually chosen, i.e., details were lost with longer intervals, and intervals smaller than 10-min might require shorter sampling frequencies or more iterations.Therefore, the "frequency" for each category was calculated by adding all the recorded values throughout the experiment for the 10-min interval.The AUC (%) for each category was the frequency divided by the total number of observations within that 10-min segment, multiplied by one hundred.The graphical representation, a 100% staked area, provided a succinct overview of the calculated AUC for pH,

Results and Discussion
Repeated acidosis challenges can lead to SARA (Dohme et al., 2008), and the diets were fed in subsequent two-weeks periods.Cattle during the first run on the Low-peNDF diet experienced increased milk yield, they developed SARA when fed the High-RDS diet in the second run and returned to a pre-trial performance when fed the Combine diet, third run (Laporte-Uribe, 2019).Accordingly, this report focuses on ruminal dCO2 monitoring with the ATR-IR spectrometer and does not reiterate on these previously established facts which are again summarised in Table 2.This is the first time that ATR-IR was used to monitor continuously in situ ruminal dCO2 concentrations.It was uncertain whether ruminal dCO2 would exceed ~60 mM (Kohn and Dunlap, 1998), whether CO2 holdup would develop, or if the dietary treatments would produce signs of SARA.Early work revealed high and varied ruminal dCO2 (Ash and Dobson, 1963;Chou and Walker, 1964b, a), but confirming its presence by manually sampling the rumen was challenging, Table 1 (Hille et al., 2016;Laporte-Uribe, 2019;Wang et al., 2019).The TIC sampling protocols adhered to the laboratory's recommendations (Buchholz et al., 2014), recognizing that freezing and transporting ruminal samples could lead to dCO2 losses (Hille et al., 2016).However, this report relies instead on the more established and accurate ATR-IR technique targeting the specific IR signal of dCO2 to confirm the ruminal dCO2 range and presence (Schädle et al., 2016).Moreover, the widespread use of the ruminal pH scale (Nocek et al., 2002;AlZahal et al., 2007b) has obscured the well-established significance of dCO2 in rumen function (Ash and Dobson, 1963;Gabel et al., 1991).As you are about to observe, ruminal dCO2 did exist in substantial quantities, and rather than solely attributing changes in rumen function to the diet or feeding sequence, we should also consider the role that these large variations in ruminal dCO2 concentrations might elicit on the epithelium and bacterial activity.
Manual sampling versus continuous ruminal CO2 monitoring.Table 1 summarises the values for pH, total inorganic carbon, TIC, and the observed dCO2 obtained by through manual sampling the ventral ruminal sac.Previously, it was stated that manual TIC sampling protocol used in these experiments primarily recovered ruminal HCO3 -(Laporte-Uribe, 2019).For instance, the marked difference between manual (0.5 points higher) and continuous pH monitoring (Duffield et al., 2004;AlZahal et al., 2007b) is attributed to dCO2 losses during manual sampling (Turner and Hodgetts, 1955;Kohn and Dunlap, 1998).To verify this assumption, calculated HCO3 -was derived by averaging the continuous pH and dCO2 measurements (Eq.2).The calculated HCO3 - closely resembled the TIC values for all diets, which confirmed that mostly HCO3 -was recovered via manual sampling.Subsequently, calculated dCO2 was computed from TIC (now HCO3 -) and compared to the continuous dCO2 values derived from the ATR-IR sensor, as presented in Table 1 (Eq.1).
Discrete manual sampling and continuous measurement represent different techniques with distinct outcomes (Duffield et al., 2004;AlZahal et al., 2007b) are not readily comparable due to variations in time scales and sampling locations.Acidification of ruminal fluid samples in the past has yielded substantial TIC recovery (Chou and Walker, 1964b, a); however, alkali addition cannot be recommended for manual TIC sampling (Laporte-Uribe, 2019).Nevertheless, the good agreement between the calculated HCO3 -and TIC values, as well as between the calculated and continuous dCO2 values (Table 1), highlights the suitability of the ATR-IR technique and sensor for continuously monitoring CO2 holdup and dCO2 concentrations.
The results also support, as a discrete sampling alternative, to target ruminal HCO3 -using the protocols described by Hille et al. (2016), in conjunction with in situ pH measurements, to indirectly estimate ruminal dCO2 using the equations described here (Eq.1).However, the manual sampling technique will have limited predictive value on detecting CO2 holdup formation or SARA onset compared with continuous ruminal dCO2 monitoring.concentration (continuous dCO2) measurements.The Observed dCO2 was calculated with Eq 1.The dCO2 derived from TIC (calculated dCO2) and the HCO3 -derived from the pH and dCO2 sensor (calculated HCO3 -) were computed using Eq 2. The median (50 th percentile) and the 10 th and 90 th percentiles, respectively.Normality of discrete manual samples was assessed using the Shapiro-Wilk test (p = 0.05).For continuous measurements, descriptive statistics provide a reliable assessment of normality due to the Central Limit Theorem.
Continuous ruminal dCO2 monitoring.The law of large numbers justified the reliance on descriptive statistics for analysing the continuous sensor data rather than solely statistical comparisons (Table 1).Multiple independent measurements of a physiological phenomenon typically follow a normal distribution, and the central value tends to be closer to the expected mean value, the Central Limit Theorem ( Van der Vaart, 2000).The agreement between discrete measurements of pH, calculated dCO2, and TIC by manual sampling, and continuous measurements of pH, dCO2, and calculated HCO3 -, respectively (Table 1), suggests a high likelihood that all parameters originated from the same population.The small kurtosis, skewness, and similar central values, both mean and median, for all diets indicated that the continuous dCO2 and pH recordings conformed to a Gaussian curve (Fig. 1 and Table 1).The normal distribution of these biological parameters justified normalisation for detecting disease, comparing diets, and eliminating drift or calibration errors (Nocek et al., 2002;AlZahal et al., 2007a;Villot et al., 2018).
Consequently, the goodness of fit of the output of the sensors employed in this study suggest that they accurately detected ruminal pH and dCO2 within the physiological and pathological range (Fig. 1 and Table 1), supporting the first hypothesis that ATR-IR is well-suited for continuous ruminal dCO2 monitoring.
The range of ruminal dCO2 concentrations detected by ATR-IR was 0 to 130 mM (Table 1).These results were comparable to the values for ruminal dCO2 described for sheep (Chou and Walker, 1964b, a).The average ruminal dCO2 values for the Low-peNDF, High-RDS, and Combined diets were 74.7, 73.0, and 59.1 mM, respectively (Table 1).These values mirrored those described for intact cattle (69.7 mM) and fistulated cattle (43.6 mM) and to the theoretical average ruminal dCO2 of ~60 mM (Kohn and Dunlap, 1998;Wang et al., 2019).The observed peak ruminal dCO2 values for the Low-peNDF (171 mM), Combined (151 mM), and High-RDS (117 mM) diets cannot be dismissed as biologically implausible.These findings challenge the previously proposed static view of the ruminal buffering system, and the saturation of the ruminal fluid at 60 mM of dCO2 (Russell and Chow, 1993;Kohn and Dunlap, 1998).
To provide context, human blood dCO2 rarely exceeds ~5% of the total CO2 content, with venous dCO2 levels at rest and during exercise being ~1.4 mM and ~2.4 mM, respectively (Geers and Gros, 2000).Cattle venous dCO2 levels, calculated from total CO2 using Eq. 1, might range from 2.2 to 2.5 mM under SARA (Gianesella et al., 2010).Further, dCO2 concentrations at rest in the inner lining fluid of the alveolar region are ~1.3 mM, corresponding to a 5% end-tidal CO2 gas content (Shao and Friedman, 2020).Ruminants exposed to over 5% CO2 gas in metabolic chambers develop tachypnoea (Blaxter, 1962) and at >10% CO2 gas exposure, alveolar dCO2 might exceed blood levels reaching over ~2.4 mM which is considered toxic (Abolhassani et al., 2009).In contrast, ruminal dCO2 values above 80 mM were routinely observed in all diets (Table 1, Fig. 2.2abc).These values are 30 times higher than blood and readily available for transepithelial absorption.
The rapid equilibrium between ruminal and blood dCO2 pools is well established (Whitelaw et al., 1972;Veenhuizen et al., 1988) primarily due to CO2 diffusion (Endeward et al., 2017;Arias-Hidalgo et al., 2018) and the utilization of ruminal CO2 for SCFA absorption (Ash and Dobson, 1963;Rackwitz and Gabel, 2018).These exceptionally high ruminal dCO2 concentrations suggest that ruminants are constantly exposed to hypoxemic/hypercapnic conditions, which explains several known unique physiological adaptations, such as the high ruminal epithelial cholesterol content (Steele et al., 2011;Jiang and Loor, 2023) which limits CO2 diffusion (Arias-Hidalgo et al., 2018); the low oxygen affinity of adult ruminant haemoglobin (Bunn, 1980) which improves peripheral tissues oxygenation, and the enhanced blood HCO3 - carrying capacity due to the chloride shift (Westen and Prange, 2003), blood CO2 is carried mainly as HCO3 - (Geers and Gros, 2000).Nevertheless, the development of CO2 holdup might enhance CO2 absorption and overwhelm the cellular buffering system, as the capacity to eliminate this dCO2 excess is impaired by the low CO2 gas fugacity from the fluid.Ruminal pH cannot predict dCO2 concentrations.In all the diets, High or Critical dCO2 levels were consistently observed postprandially, which were paralleled by a decline in ruminal pH (Fig.
Feeding the combined diet resulted in the lowest pH (Fig. 1.1c) and minimum dCO2 (Fig. 1.2c), whereas the high-RDS diet produced the highest pH (Fig. 1.1b) and maximum dCO2 (Fig. 1.2b), corroborating the statement.The reduced pH in the Combined diet can be attributable not only to lower dCO2 but also to decreased HCO3 -levels.Conversely, both HCO3 -and dCO2 concentrations were high in the High-RDS diet, bringing the pH closer to the equilibrium constant for CO2 (pKa1 ≈ 6.1).The distinctive feature of CO2 holdup is that both ruminal dCO2 and HCO3 -concentrations were elevated.Therefore, CO2 holdup explains why low ruminal pH does not always predict clinical SARA onset (Villot et al., 2018) or that SARA affected cattle present larger variation in ruminal pH than healthy cattle (Nocek et al., 2002;Penner et al., 2007;Dohme et al., 2008).The equilibrium between CO2 species dictates the pH of the solution, if both molecules are in high concentrations the ruminal pH might seems normal (Eq.2), even when critical dCO2 might be present.Therefore, while high dCO2 can coexist with low pH, it is only during CO2 holdup that critical dCO2 concentrations persist for prolonged postprandial periods, a condition that cannot be accurately predicted by the ruminal pH scale but can be effectively monitored by the ATR-IR technique.The positive effect of high ruminal dCO2.Ruminal CO2 species play a pivotal role epithelial metabolism.The majority of the intracellular HCO3 -and H3O + available for SCFA -and Na + exchange (Penner et al., 2011;Rabbani et al., 2021) are likely derived from ruminal dCO2 (Veenhuizen et al., 1988;Rackwitz and Gabel, 2018), which is most likely absorbed into the epithelial cell with H2O through aquaporins (Endeward et al., 2017).Aquaporins are abundantly expressed in the ruminal epithelia (Zhong et al., 2020).Therefore, high ruminal dCO2 increases epithelial H2O absorption (Dobson et al., 1970) and carbonic anhydrase bound to the intracellular aquaporin domains (Vilas et al., 2015;Rabbani et al., 2021) may expedite intracellular CO2 hydration, leading to the formation of HCO3 -and H3O + , which in turn enhances SCFA -uptake (Ash and Dobson, 1963;Gabel et al., 1991;Rackwitz and Gabel, 2018).The rehydration by ruminal carbon anhydrase of the secreted intracellular HCO3 -and H3O + into dCO2 provides the perfect (re)cycling system for nutrient uptake and explains the widespread expression of carbon anhydrase throughout the gastrointestinal tract (Carter and Parsons, 1971;Mau and Südekum, 2011).
The effect of high ruminal dCO2 concentrations in this experiment confirm that CO2 hydration plays a crucial role in nutrient uptake.For instance, cattle fed the Low-peNDF diet produced more milk (ECM, 37.2 vs. 35.6kg/day) and lactose (1.62 vs. 1.48 kg/day) than cattle fed the Combined diet at a similar feed intake of 24.7 kg/day (Table 2).The diets were specifically formulated to provide similar amounts of energy and protein, and no significant differences in productivity were expected (Laporte-Uribe, 2019).The rumen AUC maps for the Low-peNDF diet revealed a balanced pH (Fig. 2.1a) and consistently high dCO2 levels (Fig. 2.2a) throughout the day.In contrast, cattle fed the Combined diet exhibited lower dCO2 levels (Figure 2.2c) and a more acidic ruminal pH (Fig. 2.1c).The lower ruminal pH in the Combined diet might indicate greater availability of SCFA, as they are passively absorbed as acids (Dijkstra et al., 1993;Penner et al., 2011).However, feeding the Combined diet did not result in a higher milk yield when compared with the Low-peNDF diet (Table 2).In fact, the reduced ruminal propionate levels with the Low-peNDF diet suggested enhanced SCFA absorption, as supported by a time series of propionate, see Laporte-Uribe (2019).Propionate absorption leads to glucose formation, which boosts lactose production and milk yield from the mammary gland (Aschenbach et al., 2010;Penner et al., 2011).
Consequently, the higher milk and lactose yields with the Low-peNDF diet can be attributed to increased ruminal propionate absorption (Table 2), which was likely promoted by the high dCO2 levels observed in the rumen AUC map (Figure 2.2a), confirming the positive effect of high dCO2 on ruminal absorption (Ash and Dobson, 1963;Gabel et al., 1991).
CO2 holdup might lead to clinical SARA signs.The ruminal AUC map for pH (Figure 2.1b) indicated that cattle had the lowest SARA risk when fed the High-RDS diet based on the conventional definition of SARA based on the pH scale (Nocek et al., 2002;Villot et al., 2018).
However, cattle consuming the High-RDS diet exhibited typical SARA symptoms: reduced feed intake and milk yield, Table 2 (Nocek et al., 2002;Dohme et al., 2008).The rumen AUC map revealed that cattle fed the High-RDS diet experienced critical dCO2 concentrations for extended postprandial periods, or CO2 holdup (Spikes of critical values in Figure 2.2b).The high ruminal SCFA levels and the lower milk yield suggested impaired activity of the sodium-hydrogen exchanger (NHE) with the High-RDS diet (Penner et al., 2011;Zhao et al., 2017).Otherwise, high ruminal SCFA production and undisturbed absorption should increase milk yield, besides low feed intake with High-RDS should reduce SCFA production and not enhance it, Table 2 (Dijkstra et al., 1993).
Notably, the epithelial response to SARA involves increased intracellular SCFA metabolism and enhanced NHE expression (Zhao et al., 2017), which might bolster intracellular H3O + and HCO3 -formation and SCFA absorption.Additionally, intracellular cholesterol synthesis and deposition are intensified (Steele et al., 2011;Zhao et al., 2017), likely as a response to high dCO2 exposure and as a mechanism to reduce dCO2 diffusion (Arias-Hidalgo et al., 2018).
Moreover, SARA courses with a strong inflammatory response (Penner et al., 2011) which mirror the inflammation pathways triggered by CO2 poisoning in the lungs (Liu et al., 2008;Abolhassani et al., 2009).Therefore, clinical SARA symptoms may come from CO2 holdup development, as shown by the rumen AUC maps.Prolonged exposure to these critical dCO2 conditions might elevate the risk SARA or CO2 poisoning.
Ruminal CO2 holdup monitoring.The "rumen AUC maps" depict the daily ruminal fermentation pattern associated with ruminal dCO2 and influenced by dietary components, daily feed intake, feed allowance and management routines (Gibbs and Laporte Uribe, 2009).Therefore, these maps enable the monitoring of ruminal dCO2 by classifying it into categories with biological significance.The dCO2 detected by ATR-IR sensor at these selected thresholds aligned with the established biological effect of CO2.For instance, ruminal bacterial growth starts at 12 to 20 mM dCO2 (Dehority, 1971), and the optimal succinate production, the primary ruminal propionate precursor, requires a greater than the ruminal average, > 60 mM (Dehority, 1971;Samuelov et al., 1991).A ruminal dCO2 threshold over 80 mM might signal an increased risk of hyperosmolarity (Lodemann and Martens, 2006;Steele et al., 2011), impaired buffering capacity (Turner and Hodgetts, 1955;Hille et al., 2016) and/or an increased risk of epithelial CO2 poisoning (Liu et al., 2008).Additionally, feeding consistent diets and adhering to stable feeding management routines enhance feed intake, milk yield, and lower the risk of nutritional disorders (Deming et al., 2013;Sova et al., 2013).Consequently, rumen AUC maps provide valuable insight on the health and productivity of dairy cattle subjected to diverse diets and management practices.Furthermore, the (cross) tabulation of frequencies on daily "contingency tables" with the proposed categorical analysis streamlines the statistical comparison of ruminal patterns, i.e., the 144-time segment and 4-category matrix can be analysed utilising the Pearson chi-square (X 2 ), G-test or Bayesian inference (Van der Vaart, 2000).Consequently, rumen AUC maps establish the foundation for "precision ruminal fermentation": the selection of diets and management practices that optimise ruminal fermentation, reduce waste products, and prevent nutritional diseases associated to SARA by continuously measuring dCO2 concentrations and CO2 holdup formation.

Further work, and current limitations
To date, it is not possible to predict the impact of a specific feeding regimen or diet on dCO2 concentrations or CO2 holdup.For example, it was expected that the Combined diet will produce a stronger effect on dCO2 retention, but the opposite happened (Table 1, Figure 2).Moreover, CO2 holdup might be transient in pastoral systems, while it could be persistent in concentrate and cornbased diets (Russell, 1998;Laporte-Uribe and Gibbs, 2009).It is this same unpredictability that might explain individual susceptibility to SARA (DeVries et al., 2009)

Conclusions
Dissolved CO2 is ubiquitous in the rumen environment, present in substantial and varied amounts.
For the first time ruminal dCO2 presence and dynamics including CO2 holdup formation were described in situ.Optimal rumen function relies heavily on dCO2 concentrations, as key component of the ruminal buffering system.This crucial contribution has gone unrecognized.
Conversely, disruption of the ruminal buffering system, leading to CO2 holdup, could potentially heighten the risk of CO2 poisoning and trigger clinical SARA signs.These results warrant further investigation.The novel methodology described here might help us to validate or refute this hypothesis.

Figure 1 .
Figure 1.Histograms and normal curve fits for continuous measurements of ruminal pH and 70.5±0.5b 71.5± 0.5 b *The diets were low physically effective neutral detergent fibre (Low-peNDF) in the 1 st run, high ruminally degradable starch (High-RDS) in the 2 nd run and the combination of both previous diets (Combined) in the 3 rd run.Dry matter intake (DMI), milk yield (MY), short-chain fatty acids (SCFAs) and the ruminal acetate to propionate ratio (A/P ratio).The energy corrected MY (ECM) = milk NEL output (Mcal/d)/0.7 Mcal of NEL/kg of milk, were milk NEL output (Mcal/d) = milk yield, kg/d × (0.0929 × milk fat % + 0.0563 × milk protein % + 0.0395 × milk lactose %).All comparison were made at the 95% confidence level (P < 0.05) and means that do not share a letter are significantly different (Bonferroni).
Further work in this area should focus on those challenges based in the knowledge gathered on this trial.The diagram below (Diagram 2) proposes a schematic overview of the ruminal buffering system's function and dysfunction based in the results of this pilot experiment.Diagram 2. During optimal ruminal fermentation (left), the buffering system relies on dCO2 formation from bicarbonate (HCO3 -) and protons (H3O + ) to reduce water (H2O) ionization and restore system equilibrium.Ruminal carbon anhydrase (CAr) catalyses this process.Impaired CO2 gas (gCO2) fugacity from the fluid results in dCO2 accumulation and the development of ruminal CO2 holdup (right).Accumulated dCO2 might contribute to various clinical signs of ruminal acidosis and subacute ruminal acidosis, including hyperosmolarity of the ruminal fluid, lipopolysaccharide (LPS) formation, decreased acetate/propionate ratio (A/P ratio), lactate (Lac) accumulation at the expense of succinate (Suc), and an overall shift in short-chain fatty acid (SCFA) metabolism away from decarboxylation pathways.

Table 1 .
Descriptive statistics of ruminal parameters measured by manual and continuous sampling methods*.

Table 2 .
Summary of longitudinal trial results in fistulated cattle.Mean (±SEM) values for performance, milk components, and ruminal parameters across three runs with three fistulated cattle fed three diets*.This table consolidates findings described in the previous report of this experiment (Laporte-Uribe, 2019).