Comparison of Methane Emissions from Dairy Cows Quantified by Maximum Amplitude of Eructation Peaks and other Concentration Metrics

1. Introduction

During the last decade, several methods for quantifying CH₄ emissions from cattle have been proposed. A common approach used is the “sniffer method” which makes use of sensors that were initially created to find hazardous gas leaks [1]. The sniffer method samples gas at a place that the animal frequently visits such as a feed bin. The aim is to sample the air near the animal’s mouth and nostrils. Sniffer or breath sampling techniques have several advantages over other techniques for large-scale measurement of CH₄ emissions by individual animals. After the device is installed, animals are unaware of it and are in their normal surroundings, making sampling non-invasive. There is no need for training, handling, or dietary changes because animals follow their regular pattern, which includes feeding and milking. Equipment is often inexpensive, and operating costs are minimal [1].

Automatic milking stations provide an ideal location to measure CH₄ emissions on commercial farms given that all cows visit for milking several times per day and gas measurements can be obtained in a standard way at the feed bin [2]. Furthermore, infrared gas analysers are available that are generally mobile, durable and with high sensitivity (e.g. Nondispersive Infrared (NDIR), Fourier-transform infrared (FTIR), or photoacoustic infrared (PAIR)). Eructations produced by an animal can be detected as large peaks in CH₄ concentration by a gas analyser. Detection of peaks to quantify enteric CH₄ emissions shows good agreement with respiration chamber measurements of total CH₄ production from the same cows [3]. Bell et al. [4] found that CH₄ emissions for cows quantified by the integral of an eructation peak had higher repeatability and rank correlation compared to average CH₄ concentration and CH₄ to CO₂ ratio during milking. To test the precision and adoption of eructation peak detection for spot measurement of CH₄ emissions, there is a need to evaluate performance of alternative CH₄ metrics.

The objective of this study was to quantify CH₄ emissions from dairy cows during milking by maximum amplitude of eructation peaks and compare estimates with other metrics for CH₄ emissions.

2. Results

Cows in the study had an average lactation number of 2.6 ± 1.9, were milked 2.4 ± 0.8 times per day at robotic milking stations and produced 30.4 ± 9.7 kg of milk per day (mean ± standard deviation). Average CH₄ emissions for maximum amplitude of eructation peaks, average CH₄ concentration, integral of eructation peaks and ratio CH₄ to CO₂ are shown in Table 1. There was a higher coefficient of variation for CH₄ calculated by maximum amplitude of eructation peaks compared to other metrics.

There was a high correlation (r = 0.71 and CCC = 0.64) between CH₄ emissions calculated from integral of eructation peaks and maximum amplitude of eructation peaks (Figure 1).

There was no correlation between CH₄ emissions calculated from maximum amplitude of eructation peaks and average CH₄ concentration (Figure 2) or ratio of CH₄ to CO₂ (Figure 3).

3. Discussion

Using peak analysis techniques, enteric CH₄ obtained from integral and maximum amplitude of eructation peaks in the current study showed a strong association. The association between maximum eructation amplitude with average CH₄ concentration and ratio of CH₄ to CO₂ was weak. Both integral and maximum eructation amplitude use peak analysis approaches to extract features associated with the detection and shape of eructation peaks i.e. rise time and height. The correlation in the current study between average CH₄ concentration and ratio of CH₄ to CO₂ with maximum amplitude of eructation peaks was lower (r = 0.29 and 0.08 respectively) than reported by Bell et al. [4] assessing the rank correlation of CH₄ measurements for average CH₄ concentration and ratio of CH₄ to CO₂ in cows fed on a commercial ration (ranging from r = 0.35 to 0.62). The current study used a larger dataset of 2,206 cows across multiple farms compared to the single experimental study of 36 cows by Bell et al. [4].

Approximately 87% to 93% of enteric CH₄ emissions, along with other volatile compounds such as CO₂, are released by dairy cows through eructation from their mouths and nostrils. Ruminant animals expel rumen gas from digestion of food as large eructations on a regular basis. These emissions from fermentation of food are highest after feeding [5,6]. In the current study, the proximity of the animals’ head to the sampling tube was not measured but the maximum peak amplitude assumes that the eructation produced by a cow, and with the greatest amplitude during a milking, represents the time when the mouth and nostrils of the cow are closest to the sampling tube. This approach therefore accounts for cow head position to obtain a representative spot measurement from the animal being sampled. This approach to eructation peak detection is based on the theory that CH₄ pulses expelled by the animal can produce a repeatable and reliable measure of individual CH₄ emissions from spot measurements when compared to respiration chamber measurements [3,7,8].

Garnsworthy et al. [3] noted differences in CH₄ concentration and frequency of eructation peaks between high and low CH₄-emitting cows. Other studies using signal processing and detection of eructation peaks found that dairy cows with high and low enteric CH₄ production have distinct CH₄ profiles [9]. Cows that produce more CH₄ exhibit higher average CH₄ concentrations and more frequent eructation peaks than cows with lower CH₄ production [10]. To effectively analyse large datasets, identify hidden trends in CH₄ emissions, develop accurate estimation models, and facilitate real-time monitoring and updates, machine learning and artificial intelligence techniques hold significant potential [11,12,13,14,15].

4. Materials and Methods

Approval for this study was obtained from the Animal Welfare and Ethical Review Body of the University of Nottingham (approval number 30/3210). Data used were from previous research collecting CH₄ measurements across commercial farms [16].

4.1. Data

A total of 105,701 CH₄ spot measurements were collected from 2,206 dairy cows across 17 commercial farms (Farms A to Q) and the University of Nottingham (Farm R) in the UK (Table 2). Measurements were obtained between December 2009 and December 2013. The majority of the study's cows were Holstein-Friesian breeds. The diets provided to cows were categorised as either a partial mixed ration (PMR; conserved forage and concentrate feed) or a PMR with pasture that had been grazed. The CO₂ and CH₄ concentrations were only recorded simultaneously at farm R. The cows were fed ad libitum. During milking, concentrate feed was supplied to each cow to correspond with her daily milk production. Live weight, milk output, and robot concentrate consumption were recorded automatically at every milking. Eleven farms fed PMR and 7 farms fed a PMR with grazing (farms A to R).

4.2. Gas Sampling

A sniffer approach was used to measure CH₄ and CO₂ concentrations in the feed bin of robotic milking stations whilst dairy cows was being milked (Figure 4). Gas measurements were recorded for at least 7 days continuously at 1 second intervals using a data logger (Simex SRD-99; Simex Sp. Z o.o., Gdańsk, Poland) connected to an infrared gas analyser (Guardian SP; Edinburgh Instruments Ltd., Livingston, UK). Before each monitoring period, each analyser was calibrated using standard mixtures of CH₄ in nitrogen (0.0, 0.25, 0.50, 0.75, and 1.0% CH₄; Thames Restek UK Ltd., Saunderton, UK).

Time series gas measurement during milking was aligned to when each individual cow visited the robotic milking station using milking station records. A single infrared CH₄ analyser was used with each automatic milking station. Each gas analyser had a range of 0 to 1% for CH₄ or 0 to 5% for CO₂ concentrations. Between the gas entry port and the analyser, an integrated pump drew air through the apparatus. A 3 m long, 8 mm diameter polyethylene tube attached to the analyser’s gas inlet port was used to constantly sample air at a rate of 1 L/min from the feed bins in the milking stations. The analyser’s exhaust port was vented into the atmosphere at least 3 metres away from every sample location. An inline combination particle filter and water separator (Air-Pro IF-14; Flotec Industrial Ltd., Loughborough, UK) was installed about 50 cm from the analyser inlet port on the sample tube. The sample tube’s inlet end was placed at the back of the milking station’s feed bin. Although feed bin design can vary in milking stations, all stations had a trough that was about 50 cm wide by 50 cm long and 25 cm deep into which concentrates were delivered by a chute that was about 8 cm in diameter.

4.3. Gas Measurements

The maximum amplitude of eructation peaks, area under eructation peaks (integral of eructation), average CH₄ concentration and ratio of CH₄ to CO₂ were extracted from the time series gas concentration (Figure 5).

Maximum amplitude of eructation peaks, area under eructation peaks and average CH₄ concentration were obtained for all spot measurements. Concentration of CO₂ was only measured at Farm R and provided 2,474 spot measurements with CO₂ concentration measured at the same time as CH₄ concentration. Background concentration of both gases during milking was assumed to be the lowest gas concentration and were subtracted from the average concentration recorded.

4.3.1. Maximum Eructation Amplitude

Maximum peak amplitude was extracted from measured CH₄ concentration using signal processing peak analysis tools in MatLab Signal Processing Toolbox (version R2018a, The MathWorks, Inc., Natick, United States). The rationale for this approach was that maximum amplitude should correspond to an eructation when the cow had its nose closest to the sampling tube, thus reducing variation in CH₄ measurements due to head position as proposed by Bell et al. [5]. The peak analysis tools identify eructation peaks using the findpeak function. Maximum peak amplitude of an eructation peak also removed background CH₄ concentration. Rise time to maximum peak was also determined. Filters were applied to data to prevent detection of the same peak more than once, with a minimum peak amplitude of 100 ppm and a minimum peak separation of 10 seconds. Given that the analyser needs 60 seconds to reach the "real" peak asymptote and finish processing the gas sample, all extracted emission rates (in grams per minute) during milking were scaled to estimated emissions based on the exponential increase in gas concentration and the extracted rise time for eructation peaks by Equation (1) as:

CH₄ emission rate (g/min) = maximum peak amplitude (ppm)/ [1- EXP ( -( peak rise for amplitude in seconds /60))] x 60 x 0.706 x 10^-6

(1)

Measurements were converted to emission rate in grams per minute by multiplying by 60 and density of CH₄, assumed to be 0.706 × 10⁻⁶ g/L.

4.3.2. Integral of Eructation

Average area under eructation peaks of three or more eructations per milking were calculated as the integral of peak concentrations in mg/L per minute multiplied by frequency of eructation peaks per minute to calculate concentration of CH₄ produced by the cow in mg/L. A custom-designed program developed by Garnsworthy et al. [3] extracted the integral of peaks and peak frequency. Milkings with less than three eructation peaks were excluded from analysis. Peaks that rose less than 200 mg/kg above baseline were excluded. Eructation frequency and integral were calculated for every milking. The product of calculated mean integral and eructation frequency was used to compute CH₄ emission index during each milking per cow in mg/L [3].

4.3.3. Average Concentration and Ratio of CH₄ to CO₂

Average concentrations of CH₄ and CO₂ emitted in ppm during each milking were calculated by subtracting minimum concentration at the start of the milking from average concentration during each milking. Ratio of CH₄ to CO₂ concentrations was then calculated by dividing average CH₄ concentration by average CO₂ concentration.

4.4. Statistical Analysis

Data were analysed using Genstat Version 21.1 (VSN International, Hemel Hempstead, UK) VSN, Hemel Hempstead, UK). Relationships between CH₄ spot measurement metrics were assessed using Pearson correlation coefficient (r), Lin’s bias correction factor (rb), and concordance correlation coefficient (CCC). The concordance correlation coefficient (CCC) was obtained by multiplying coefficient r by Lin’s bias correction factor (rb), which assesses how far the best-fit line deviates from the 45° line through the origin [17].

5. Conclusions

In this study, four enteric CH₄ emission metrics for dairy cow were compared. Methane emissions computed using maximum peak amplitude and integral of eructation peaks showed a high positive correlation. There was a low positive correlation between peak analysis measures and average CH₄ concentration or CH₄ to CO₂ ratio. The correlation between integral CH₄ peak analysis and maximum amplitude CH₄ peak analysis was anticipated given both methods detect and extract eructation peaks to estimate enteric CH₄ emissions.