Effect of number of measurement days on variance in methane and carbon dioxide emissions measured using GreenFeed units in grazing dairy cows and growing heifers
M. M. Della Rosa

A
B
Handling Editor: James Hills
Abstract
The minimum number of days needed to measure gas emissions from cattle by using spot sampling methods is the result of the visit frequency, within animal variation and among animal variations.
To estimate (a) the effect of the length of the measurement period on the variation in methane (CH4) and carbon dioxide (CO2) emissions and (b) the number of animals required to detect a difference of 10% between two treatment means for CH4 and CO2.
Gas emissions from 72 dairy cows, supplemented with different concentrate diets, and 72 heifers, weaned at different ages, in two separate experiments, were measured for 3–5 weeks using GreenFeed units. In all four experiments, the animals grazed ryegrass-based pasture. The cows received various concentrate treatments twice daily during milking. The gas emissions in heifers were measured at 280 and 370 days of age. Data from 76 cows and 77 heifers were used in the data analysis. The coefficient of variation (CV) and number of animals required to detect a difference of 10% between the two means were modelled for periods of 3–36 days at 3-day steps.
The CV of CH4 emissions became stable between Days 12 and 18 of measurements in the cows and heifers, respectively (17–37 visits for cows and 43–73 visits for heifers) and then 13–19 cows and 9–11 heifers were required per treatment to detect differences of 10% between means. The CV of CO2 emissions became stable few days earlier than did the CH4 emissions and the variation was smaller.
A minimum of 12 and 18 measurement days are recommended to estimate CH4 emission in grazing lactating cows and heifers respectively, and 9–19 animals per treatment were required to detect differences of 10% between means for the conditions of the current studies.
The current analysis has provided information about among-animal variation of gas emissions when performing GreenFeed measurements with grazing cattle, within the experimental conditions of the data sets used for the current study, which can be used to design future cattle studies.
Keywords: cattle, enteric CH4 measurement, experimental design, grazing systems, greenhouse gases, power analysis, spot samples, visits.
Introduction
Agriculture is the largest contributor to anthropogenic methane (CH4) emissions, and approximately 72% of the agricultural CH4 emissions come from ruminants’ enteric fermentation (Gerber et al. 2013). Agricultural CH4 emissions must be reduced by 43% by 2030 to meet the targets of Paris Agreement and limit global warming to 1.5°C (Rogelj et al. 2018), and this has increased the research into ways to reduce enteric CH4 emissions. The consequent increased need for CH4 emission measurements from ruminants has resulted in the development of several new methods in recent decades. One of the main new methods that has seen a large increase in its use in research to measure CH4 emissions in the past few decades has been the GreenFeed automated emission monitoring system (C-Lock Inc., Rapid City, South Dakota, SD, USA) (Della Rosa et al. 2023a), which can provide accurate CH4 emission data (Huhtanen et al. 2019; Jonker et al. 2020a) as long as a sufficient number of spot samples are collected to enable reduction within and among animal variability. Determining daily CH4 emissions using GreenFeed units relies on averaging data from multiple spot samples of exhaled air from animals voluntarily visiting the units multiple times per day and across days (e.g. 10–21 days; Jonker et al. (2020b)).
The GreenFeed units have been mostly tested and validated in housed systems (Hristov et al. 2016; McGinn et al. 2021) for ranking animals, and in many cases with a small number of animals per unit, and are also used with cattle under grazing conditions, with up to 40 animals per unit (Waghorn et al. 2016). During grazing studies, the GreenFeed units are exposed to weather conditions, especially wind speed and direction, which can affect the capture rate of gases into the system and increase data exclusion (Huhtanen et al. 2015). In addition, the number of animals that do not visit a GreenFeed is generally greater, and visits per animal per day are generally less in grazing animals than in housed animals (Hammond et al. 2015; Waghorn et al. 2016; Jonker et al. 2017a; Hristov and Melgar 2020; Starsmore et al. 2024). Last, grazed pasture composition and availability are more variable within and across days than are conserved diets and grazing cattle have generally more distinctive feeding patterns with intensive grazing bouts around dusk and dawn (Iqbal et al. 2022, 2023), which might increase the diurnal variance in emissions. These factors may affect the number of measurement days required to determine the gas emissions accurately and precisely of an individual cow and the number of animals required to detect a difference in CH4 emissions.
In addition to CH4 measurements, the GreenFeed units also simultaneously determine carbon dioxide (CO2) emissions. The CO2 emissions have a strong relationship with dry-matter intake (DMI) (Pereira et al. 2015; Kjeldsen et al. 2024), and the CH4 to CO2 ratio as well as the residual CH4 production predicted from CO2 have been used as a similar trait for CH4 yield (g/kg DMI) when intake is not known (Madsen et al. 2010; Herd et al. 2016; Della Rosa et al. 2024a) such as in grazing studies. The objective of the current data analysis was to determine the decrease in variance (coefficient of variation) for CH4, CO2 emissions and CH4 to CO2 ratio with an increasing number of measurement days in grazing dairy cows and growing heifers, and to model the number of animals required to detect a difference of 10% between two treatment means.
Materials and methods
Data were extracted from two dairy cow studies and two growing heifer studies measuring gas emissions using GreenFeed under grazing conditions. All the animal manipulations were approved by the AgResearch Animal Ethics Committee (AE approvals, #288 for Experiment 1, #717 for Experiment 2, #781 for Experiments 3 and 4) and animal use and manipulations adhered to the guidelines of the 1999 New Zealand Animal Welfare Act and the AgResearch Code of Ethical Conduct.
Methane and carbon dioxide emissions
The experiment was conducted from mid-October to mid-November 2022 at Massey University Dairy Farm No. 4 (Palmerston North, New Zealand; S40°23′49″, E175°36′46″). Seventy-two early lactation cows [Friesian × Jersey, (mean ± s.d.) 113 ± 17 days in milk, 512 ± 48 kg bodyweight] blocked by parity were allocated to receive 0, 2, 4 and 6 kg dry matter (DM) of mixed concentrate pellets per day. The concentrate pellets (on DM basis) were composed of 30% barley grain, 15% maize grain, 10% wheat brollard, 25% palm kernel expeller, 8% soya hulls, 7% soybean meal and 5% molasses (Denver Stock Feeds, Palmerston North, New Zealand). The experiment was described in detail by Bosher et al. (2024). The cows received their respective allocation of concentrates equally divided into two meals during milking at approximately 07:00 hours and 15:00 hours. The cows of all treatments were grazed as one group ad libitum ryegrass-based pasture (~22 kg DM/cow.day), with a new strip being provided after each milking (no back fencing). The cows were also offered ~2 kg DM/cow ryegrass silage with mineral mix before afternoon milking on a feed pad.
The GreenFeed units (C-Lock, Rapid City, SD, USA) were trailer-mounted and powered by batteries charged through a fuel-powered generator and were used to measure CH4 and CO2 emissions. The cows were attracted to the GreenFeed units by receiving lucerne based pellets as bait (Alpaca and Llama pellets, Country Harvest, Hamilton, New Zealand) via the integrated automated concentrate feeder. Two wings and an alleyway attached to each GreenFeed unit prevented more than one cow from having access to the hood during the gas measurements. The 72 cows were given access to two GreenFeed units for 14 days for training purposes before the measurement phase of 33 days. During the training, the settings of the GreenFeed units were the same as in the measurement phase, but the GreenFeed units did not have the wings and alleyway attached. Most of the training and the first half of the measurement period was conducted with one GreenFeed unit, owing to technical problems with the second GreenFeed unit. The units were moved every day and placed in between the current and next strip to be grazed. GreenFeed settings during the study are summarised in Table 1.
Item | Experiment 1 | Experiment 2 | Experiment 3 | Experiment 4 | |
---|---|---|---|---|---|
Number of training days | 14 | 14 | 14 | 7 | |
Number of measurement days | 33 | 36 | 22 | 38 | |
Number of animals per unit | 36–72A | 36 | 36 | 36 | |
GreenFeed settings | |||||
Amount of pellets per drop (g as is)B | 34.65 | 33.84 | 32.04 | 32.97 | |
Maximum drops per visit | 7 | 7 | 7 | 7 | |
Interval between drops (s) | 15–25 | 25 | 20–25 | 20–25 | |
Maximum number of visits allowed per day | 12 | 12 | 12 | 8–12 | |
Interval between visits (h) | 2 | 2 | 2 | 2–3 | |
GreenFeed shifting | Every day | Every second day at least | Every third day | Every third day | |
Minimum visit duration of data analysed (min) | 2 | 1.5 | 1.5 | 1.5 |
The experiment was conducted from early March to early April 2023 at Massey University Dairy Farm No 4. Seventy-two late lactation cows, largely the same ones as used in Experiment 1 (six cows of the previous study were replaced with new cows from the same spring calving herd), [Friesian × Jersey, (mean ± s.d.) 206 ± 9.5 days in milk, bodyweight of 531 ± 50 kg, 4.1 ± 0.52 months pregnant, except for seven cows that were empty for the duration of this study] were allocated to receive 0 kg/days of concentrate, 5 kg/days of high starch concentrate pellets, 5 kg/days of high fibre concentrate pellets, and 50:50 high starch to high fibre concentrate pellets as supplements (expressed as DM; Della Rosa MM, Khan MA, Jonker A, AgResearch Ltd., unpubl. data). The high starch pellets were composed of 10% maize grain, 13% wheat grain, 30% barley grain, 30% bollard, 7.5% palm kernel expeller, 5% molasses and 4.5% of soybean meal (on DM basis; Denver Stock Feeds). The high fibre concentrate pellets comprised 30% brollard, 40% palm kernel expeller, 5% molasses, 23% soybean hulls and 2% soybean meal (on DM basis; Denver Stock Feeds). The cows received their respective allocation of concentrates equally divided into two meals during milking at approximately 06:30 hours and 13:30 hours. The cows of all treatments were grazed as one group ad libitum ryegrass-based pasture (~22 kg DM/cow.day) with a new strip provided after morning milking (no back fencing). The cows received ~2 kg DM/cow ryegrass silage with mineral mix after morning milking in the paddock. Two GreenFeed units were located in the paddock starting 14 days before the measurement phase of 36 days for training purposes as described in Experiment 1. The units were moved every second day to the two new pasture strips to be grazed. GreenFeed settings are summarised in Table 1.
The experiments were conducted from mid-January to early February 2023 (Experiment 3) and from late April to late May 2023 (Experiment 4) at AgResearch Aorangi Farm (near Palmerston North, New Zealand; S39°12′17″, E176°2′38″; Jonker A, Khan MA, AgResearch Ltd., unpubl. data). Seventy-two 3-week-old Hereford × Friesian heifer calves born in Autumn 2022 were enrolled in a study to evaluate the effect of weaning age on CH4 emissions. The calves were randomly allocated to one of two weaning ages (weaned from milk at 10 weeks of age or at 20 weeks of age). The 72 calves were individually fed 1 kg of milk replacer (as is) per calf per day divided into two meals (Sprayfo Rosso, Trouw Nutrition, Deventer, The Netherlands). The weaning strategy was gradual, starting 3 weeks before reaching the weaning age of 10 or 20 weeks of life. During the milk-feeding phase and after weaning, the heifers were managed in a grazed ryegrass-based pasture system without supplementation of other solid feeds. The heifers were managed as two grazing groups (half of the heifers from each weaning age in each group) managed as blocks. Methane measurements on the same animals were performed at 280 (s.d. 6.39) days of age (Experiment 3) and at 370 (s.d. 6.39) days of age (Experiment 4). During both experiments, the heifers were grazing in two blocks (18 early and 18 late weaned heifers per block) that were established at the end of the weaning. One GreenFeed unit was placed in each paddock with the heifers for 14 and 7 days before gas emission measurements for training purposes (22 and 38 days) in Experiments 3 and 4 respectively. During the training and measurement phases, the same pellets as in Experiments 1 and 2 (Alpaca and Llama pellets, Country Harvest) were used as bait. The training was performed as explained in Experiment 1. Every 7th day, the heifers were rotated to a new paddock of pasture. GreenFeed settings are summarised in Table 1.
Data processing and analyses
Processed data provided by C-Lock included visits of ≥2.0 min duration for Experiment 1 and ≥1.5 min duration for Experiments 2–4. The CH4 emissions of every spot sample were plotted against the visit duration to visually assess the dispersion of the data, similar to that performed by Bennett et al. (2021). The variability of CH4 emissions for visits between 1.5 and 2 min was similar to that of visits of ≥2 min only, therefore visits of 1.5–2 min were kept in the dataset (Table 2). Data from the cows that started visiting the units three or more days later than the beginning of the measurement period and cows that were removed from the study before the end of the experiment because of health reasons were excluded from the database before data analysis.
Item | Experiment 1 | Experiment 2 | Experiment 3 | Experiment 4 | |
---|---|---|---|---|---|
All animals that visited the GFs | |||||
Number [percentage] A of animals that visited GF | 54 [75%] | 65 [90%] | 52 [72%] | 65 [90%] | |
Number of GF visits per animal | |||||
Range of total number of visits | 1–147 | 10–275 | 1–150 | 1–247 | |
Mean ± s.d. of total number of visits | 56 ± 35 | 101 ± 54 | 62 ± 46 | 131 ± 63 | |
Animals included in analyses B | |||||
Number of measurement days | 33 | 36 | 21 | 36 | |
Number and [percentage] A of animals that visited GF | 25 [34%] | 51 [70%] | 30 [42%] | 47 [65%] | |
Number of GF visits per animal | |||||
Range of total number of GF visits per animal | 27–145 | 13–177 | 24–150 | 15–229 | |
Mean ± s.d. of total number of visits/animal | 75 ± 31 | 83 ± 36 | 90 ± 34 | 145 ± 38 | |
Range of visits per day per animal | 0.82–4.39 | 0.36–4.92 | 1.14–7.14 | 0.42–6.36 | |
Mean ± s.d. of visits per day per animal | 2.28 ± 0.93 | 2.31 ± 1.00 | 4.28 ± 1.64 | 4.04 ± 1.05 | |
Total number of spot samples | 1881 | 4241 | 2696 | 6836 | |
Percentage of spot samples >1.5 but <2.0 min | 0 | 14 | 10 | 5 | |
Percentage of spot samples >2.0 but <3.0 min | 49 | 33 | 27 | 16 | |
Percentage of spot samples >3.0 min | 51 | 53 | 63 | 79 |
The dataset used for analyses included 25 animals from Experiment 1 (8, 7, 5 and 5 cows that were offered 0, 2, 4, 6 kg DM concentrate pellets per day), 51 animals from Experiment 2 (10, 12, 15, 14 cows that were offered no concentrate or 5 kg DM/day of pellets high in starch, high in fibre or a 50:50 mix of high starch and high fibre concentrates respectively), 30 heifers from Experiment 3 (16 and 14 heifers weaned at 10 and 20 weeks respectively), and 47 heifers from Experiment 4 (26 and 21 heifers weaned at 10 and 20 weeks respectively). The CH4 to CO2 ratio was calculated for each spot sample prior to further calculations.
The CH4, CO2 and CH4 to CO2 ratio (average) were calculated for each animal for each interval of 3 days by using two approaches, namely (a) ‘arithmetic averaging method’ defined as the sum of the emissions per visit divided by the number of visits accumulated over intervals of 3, 6, 9, … n days within animal, and (b) ‘time of day method’ calculating the average emission per hour of the day within animal and then average the 24 h over intervals of 3, 6, 9, … n days within animal. Then the pooled mean for each experiment was calculated using the ‘weighted.mean’ function in R Core Team (2023; ver. 4.3.0) considering the number of treatments (dietary treatments in Experiment 1 and Experiment 2, and weaning age and grazing block in Experiment 3 and Experiment 4) and number of animals within treatment. The pooled standard deviation was calculated as
where n is the sample size for each treatment group and s.d. is the standard deviation (Borenstein et al. 2021). The number of experimental units per treatment group needed to detect a significant difference of 10% between two treatment means for CH4, CO2 emissions and CH4 to CO2 ratio (α = 0.05, power = 0.8) was calculated using the ‘power.anova.test’ function in R Core Team (2023). The ‘power.anova.test’ function returns the sample size required to detect the specified differences among means in a balanced one-way ANOVA and uses as inputs the number of treatments (four in this case) to compare, the mean and standard deviation.
The homogeneity of variance between treatments for each interval of 3, 6, 9, … n days was tested using the Bartlett’s test. The among-animal CV was considered to be stable when the among-animal CV decreased less than 15% compared with the previous 3-day interval.
Results
Gas emissions and visits to GreenFeed units: frequency and patterns
The proportion of animals that visited the GreenFeed units at least once represented 72–90% of all animals across the four experiments. After excluding animals that did not meet all the criteria for inclusion in the data analysis, the proportion of animals included represented 34–70% of the animals used in the experiments. The dairy cows in Experiments 1 and 2 visited the GreenFeed units, on average, 2.3 times per day, whereas the growing heifers in Experiments 3 and 4 visited the GreenFeed units on average 4.0 and 4.3 times per day. The minimum number of total visits for an individual animal included in the data analysis ranged from 13 to 27 visits after the exclusion of animals that did not meet the criteria.
The visit frequency per hour to the GreenFeed units was variable during the 24-h period but the cows on the different dietary treatments and the heifers with different weaning ages had a similar GreenFeed visitation pattern across the day (Fig. 1). The accumulated number of visits to the GreenFeeds was 20–40% greater in non-supplemented cows in Experiment 1 and Experiment 2 than in cows offered concentrates, whereas the accumulated number of visits to the GreenFeed units was similar for heifers of the two weaning ages (Fig. 2). The visit frequency decreased steadily from midnight until 05:00 hours to 06:00 hours in all experiments (Fig. 1). Lactating cows (Experiments 1 and 2) were milked twice daily, whereas the GreenFeed units remained in the paddock and, therefore, there were negligible visits around these times. Visit frequency to the GreenFeed unit was high in the period between morning and afternoon milking and after afternoon milking until ~18:00 hours, and then the visit frequency decreased during the next 2 h after the sunset, after which the visit frequency increased steadily until midnight. In growing heifers, the visit frequency to GreenFeed units was high around sunrise and again towards sunset, with moderate visitation throughout the rest of the day.
Distribution of the average of accumulated number of visits/animal to GreenFeed units during 24 h from two experiments performed with dairy cows (Experiment 1 and Experiment 2) and two experiments with growing heifers (Experiment 3 and Experiment 4). Coloured bars represent the mean and black bars represent the standard error. Legend for Experiment 2:, S, high starch; F, high fibre; SF, 50:50 S:F mix.

Distribution of the average accumulated number of visits/animal to GreenFeed units during the gas measurement phase from two experiments with dairy cows (Experiment 1 and Experiment 2) and two experiments with growing heifers (Experiment 3 and Experiment 4). Coloured bars represent the mean and black flags represent the standard error. Legend of Experiment 2: S, high starch; F, high fibre; SF, 50:50 S:F mix.

The hourly CH4 and CO2 emissions and CH4 to CO2 ratio were variable during the 24-h period and decreased steadily from midnight until 05:00 hours to 06:00 hours in all experiments (Fig. 3). Lactating cows in Experiments 1 and 2 had stable CH4 and CO2 emissions from 08:00 hours until midnight, with noticeable peaks after morning and afternoon milking. The CH4 to CO2 ratio from cows was stable during the 24 h, except just before morning milking. The CH4 emissions from heifers in Experiment 3 and Experiment 4 differed between experiments after the sunrise. In Experiment 3, the CH4 and CO2 emissions reached the minimum rate at sunrise; after that, the emission rate increased slightly until close to midnight. The CH4 to CO2 ratio in Experiment 3, started high and decreased until noon; after noon, it increased steadily until close to midnight. The CH4 and CO2 emissions in Experiment 4 increased from sunrise until peaking at ~18:00 hours. After the peak, the emissions decreased until reaching an intermediate value. The CH4 to CO2 ratio in Experiment 4 followed the pattern of both gases. In the four experiments, the ratio between the maximum and minimum emission rate within 24 h ranged from 1.48 to 1.74 for CH4 emissions, from 1.18 to 1.42 for CO2 emissions and from 1.23 to 1.35 for the CH4 to CO2 ratio (Fig. 2).
Distribution of hourly average methane (CH4) and carbon dioxide (CO2) emissions and methane to carbon dioxide ratio (CH4:CO2) across 24 h measured using GreenFeed units from two experiments with dairy cows (Experiment 1 and Experiment 2) and two experiments with growing heifers (Experiment 3 and Experiment 4). Flags represent the standard error. m/m, ratio between maximum to the minimum emission rates averaged per hour.

Among-animal variations in CH4 and CO2 emissions
Both the ‘arithmetic mean’ and the ‘time of day’ method used to estimate the daily gas emissions per animal every 3 days resulted in similar among-animal CV and number of animals required to detect 10% of differences between means (Table 3, Supplementary Table S1).
Item | Measurement days | ||||||||||||
---|---|---|---|---|---|---|---|---|---|---|---|---|---|
3 | 6 | 9 | 12 | 15 | 18 | 21 | 24 | 27 | 30 | 33 | 36 | ||
Average accumulated number of visits per animal | |||||||||||||
Experiment 1 | 4 | 9 | 16 | 24 | 29 | 37 | 46 | 52 | 62 | 71 | 75 | ||
Experiment 2 | 5 | 8 | 12 | 17 | 22 | 31 | 39 | 49 | 59 | 67 | 76 | 83 | |
Experiment 3 | 10 | 23 | 39 | 50 | 60 | 73 | 84 | ||||||
Experiment 4 | 10 | 19 | 30 | 43 | 60 | 72 | 86 | 100 | 111 | 122 | 135 | 145 | |
Methane emissions (g/days) | |||||||||||||
Experiment 1 | |||||||||||||
Mean | 274 | 289 | 292 | 302 | 306 | 316 | 325 | 331 | 341 | 345 | 344 | ||
CV | 25.8 | 21.2 | 16.5 | 15.8 | 14.6 | 13.1 | 13.9 | 13.1 | 13.0 | 12.6 | 12.6 | ||
Min Number of animals | 49 | 34 | 21 | 19 | 17 | 14 | 15 | 14 | 13 | 13 | 13 | ||
Experiment 2 | |||||||||||||
Mean | 377 | 380 | 384 | 374 | 366 | 355 | 350 | 353 | 356 | 354 | 353 | 354 | |
CV | 19.5 | 19.5 | 17.3 | 15.5 | 14.0 | 15.1 | 14.7 | 14.6 | 14.1 | 14.0 | 14.1 | 13.6 | |
Min Number of animals | 29 | 28 | 23 | 19 | 15 | 18 | 17 | 17 | 16 | 15 | 15 | 14 | |
Experiment 3 | |||||||||||||
Mean | 151 | 153 | 168 | 166 | 164 | 167 | 164 | ||||||
CV | 20.9 | 14.5 | 11.8 | 11.0 | 10.9 | 10.4 | 10.7 | ||||||
Min Number of animals | 33 | 16 | 11 | 10 | 10 | 9 | 9 | ||||||
Experiment 4 | |||||||||||||
Mean | 151 | 167 | 168 | 174 | 176 | 172 | 175 | 176 | 177 | 176 | 175 | 175 | |
CV | 17.0 | 14.1 | 13.9 | 11.6 | 10.7 | 11.4 | 11.2 | 9.5 | 8.7 | 8.6 | 8.5 | 8.5 | |
Min Number of animals | 22 | 15 | 15 | 11 | 9 | 10 | 10 | 8 | 7 | 6 | 6 | 6 | |
Carbon dioxide emissions (g/days) | |||||||||||||
Experiment 1 | |||||||||||||
Mean | 10,937 | 11,191 | 11,306 | 11,576 | 11,687 | 11,918 | 11,978 | 12,123 | 12,242 | 12,353 | 12,307 | ||
CV | 11.9 | 10.3 | 9.0 | 8.8 | 8.5 | 7.9 | 8.6 | 8.4 | 8.5 | 8.4 | 8.3 | ||
Min Number of animals | 11 | 9 | 7 | 7 | 6 | 6 | 6 | 6 | 6 | 6 | 6 | ||
Experiment 2 | |||||||||||||
Mean | 12,115 | 12,088 | 12,179 | 12,013 | 11,893 | 11,756 | 11,696 | 11,737 | 11,877 | 11,882 | 11,875 | 11,923 | |
CV | 10.3 | 9.8 | 9.5 | 9.0 | 8.8 | 9.1 | 8.7 | 8.9 | 8.8 | 8.9 | 9.0 | 8.8 | |
Min Number of animals | 9 | 8 | 8 | 7 | 7 | 7 | 7 | 7 | 7 | 7 | 7 | 7 | |
Experiment 3 | |||||||||||||
Mean | 5836 | 5690 | 6013 | 6058 | 6012 | 5997 | 5854 | ||||||
CV | 9.7 | 9.2 | 7.5 | 7.1 | 7.1 | 6.7 | 6.7 | ||||||
Min Number of animals | 8 | 7 | 5 | 5 | 5 | 4 | 4 | ||||||
Experiment 4 | |||||||||||||
Mean | 5713 | 6109 | 6067 | 6272 | 6316 | 6211 | 6316 | 6382 | 6457 | 6479 | 6498 | 6525 | |
CV | 8.7 | 8.3 | 8.0 | 7.4 | 7.1 | 7.9 | 7.8 | 7.2 | 7.1 | 6.9 | 7.0 | 7.0 | |
Min Number of animals | 7 | 6 | 6 | 5 | 5 | 6 | 5 | 5 | 5 | 5 | 5 | 5 | |
Methane to carbon dioxide ratio (mol/mol) | |||||||||||||
Experiment 1 | |||||||||||||
Mean | 0.0684 | 0.0704 | 0.0704 | 0.0711 | 0.0715 | 0.0724 | 0.0739 | 0.0744 | 0.0757 | 0.0760 | 0.0761 | ||
CV | 18.6 | 14.3 | 10.8 | 10.4 | 9.8 | 8.5 | 8.4 | 7.6 | 7.3 | 7.1 | 7.1 | ||
Min Number of animals | 26 | 16 | 10 | 9 | 8 | 6 | 6 | 5 | 5 | 5 | 5 | ||
Experiment 2 | |||||||||||||
Mean | 0.0844 | 0.0852 | 0.0856 | 0.0848 | 0.0835 | 0.0816 | 0.0811 | 0.0813 | 0.0813 | 0.0808 | 0.0806 | 0.0844 | |
CV | 16.7 | 14.7 | 13.4 | 11.6 | 9.2 | 10.1 | 9.6 | 9.2 | 9.0 | 9.0 | 8.8 | 8.6 | |
Min Number of animals | 21 | 17 | 14 | 11 | 7 | 8 | 8 | 7 | 7 | 7 | 7 | 6 | |
Experiment 3 | |||||||||||||
Mean | 0.0704 | 0.0737 | 0.0764 | 0.0749 | 0.0746 | 0.0760 | 0.0768 | ||||||
CV | 18.5 | 8.6 | 7.0 | 6.6 | 6.9 | 6.9 | 7.1 | ||||||
Min Number of animals | 26 | 6 | 5 | 4 | 5 | 5 | 5 | ||||||
Experiment 4 | |||||||||||||
Mean | 0.0721 | 0.0747 | 0.0759 | 0.0760 | 0.0761 | 0.0754 | 0.0756 | 0.0753 | 0.0751 | 0.0743 | 0.0739 | 0.0735 | |
CV | 13.4 | 11.1 | 10.7 | 9.1 | 8.5 | 8.6 | 8.4 | 7.7 | 7.0 | 7.1 | 7.0 | 6.9 | |
Min Number of animals | 14 | 10 | 9 | 7 | 6 | 6 | 6 | 5 | 5 | 5 | 5 | 5 |
Results are shown for two experiments with dairy cows (Experiments 1 and 2) and two experiments with growing heifers (Experiments 3 and 4) by using the ‘arithmetic mean method’ to estimate daily emissions.
Regardless of the averaging method used to estimate the daily CH4 emissions, the variance of the daily CH4 emissions at 3-day intervals was homogeneous among treatment groups. The among-animal CV for CH4 emissions and the CH4 to CO2 ratio became stable at between 12 and 18 days of measurements across the four experiments and the average total number of visits reached during this period ranged from was 17–37 visits in dairy cows and 43–73 in growing heifers. At these number of measurement days, 13–19 cows, and 9–11 heifers per treatment were required to detect differences of 10% in CH4 emissions between the two treatment means.
The among-animal CV for CO2 emissions became stable between 6 and 15 days of measurement and for CH4 to CO2 ratio between 12 and 18 days of measurements in cows and heifers, which was earlier than for CH4 production. Therefore, the number of animals required to detect a 10% difference between treatment means for CH4 emissions was also sufficient for detecting similar or larger differences in CO2 production and CH4 to CO2 ratio. The variance of CO2 and CH4 to CO2 ratio at 3-day intervals was homogeneous among treatment groups.
Discussion
The main findings of the current analyses were that CH4 emission measurements by using GreenFeed stabilised after 12–18 days in grazing lactating dairy cows (17–37 visits per cow) and grazing heifers (43–73 visits per heifer). So, to measure CH4 emissions precisely, it was necessary to collect more spot samples over several days compared with measuring CO2 emissions because the variability of CH4 emissions from day to day was greater than the variability of CO2 emissions and the CH4 to CO2 ratio. The GreenFeed visitation frequency per day was less in the cow studies than in the heifer studies and, therefore, the among-animal CV stabilized slightly later in those cow studies. In grazing experiments with a relatively low number of visits per day, measuring the CH4 emissions for less than 18 days would increase the between animal CV (decrease precision) and the likelihood of Type II errors (i.e. false negative). The among-animal variability is the main factor affecting the number of animals required to detect differences between treatment means. In this data analysis, 13–19 cows and 9–11 heifers were required to detect a 10% difference in CH4 emissions between treatment means.
The number of spot samples required to get a stabilised among-animal CV for cows was within the range reported by other studies and, for heifers, it was slightly higher than the minimum number of spot samples that resulted from other studies (Manafiazar et al. 2016; Renand and Maupetit 2016; Arthur et al. 2017; Gunter and Bradford 2017; Dressler et al. 2023). Other studies have suggested that 15–50 spot measures are needed per animal to generate a precise individual CH4 emission measurement across a range of cattle types and diets (Manafiazar et al. 2016; Renand and Maupetit 2016; Arthur et al. 2017; Gunter and Bradford 2017; Dressler et al. 2023). The animal visits to the GreenFeed units are voluntary and increasing the number of visits increases the likelihood that each hour of the day will have gas emission measurements. Because the CH4 emissions follow diurnal patterns that are highly associated with feed intake events (Jonker et al. 2014), it is expected that a greater number of visits/spot samples reduce the variation in the daily CH4 emission estimate within days and the variation among animals within the same treatment group.
The among-animal CV stabilised with fewer visits in cows than in heifers; however, in absolute terms, the CV remained greater in cows than in heifers. The greater absolute among-animal CV observed in cows than in heifers was consistent with what is generally observed in these two cattle categories, namely an increasing variance with an increasing DMI concentration across growing and lactating cattle (Charmley et al. 2016; Jonker et al. 2017b). Besides, the greater among-animal CV observed in cows may be due to the different daily CH4 emissions and diurnal variation associated with the dietary treatments (Della Rosa et al. 2023b, 2024b; Della Rosa et al., in press). In contrast, the earlier stability in among-animal CV observed in cows was likely to be related to the more consistent daily diet owing to the daily provision of new pasture breaks compared with the once-a-week provision of new pasture in the heifer experiments. The dairy cows were offered a new pasture strip daily, which reduced pasture selection within a day and reduced variation of the grass quality eaten across days. In contrast, the heifers were given access to fresh pasture weekly. As a result, it is likely that they grazed on higher pasture at the beginning, with feed quality declining as the week progressed. This decline occurs because the leafy parts at the top of the canopy have a greater quality than do the stems at the bottom of the canopy (Rook and Tallowin 2003).
The among-animal CV for CO2 and CH4 to CO2 ratio became stable with a similar or smaller number of spot samples/sampling days than for CH4 emissions, and both, CO2 emissions and CH4 to CO2 ratio were less variable (i.e. smaller CV) than were CH4 emissions. Therefore, getting the minimum number of spot samples required to estimate CH4 emissions will also satisfy the minimum number of spot samples required for measuring CO2 emissions. These are likely to be due to the difference in the primary source of the two gases; CO2 emissions are largely related to energy expenditure for body maintenance and production, which remain stable over time. (Manafiazar et al. 2016), whereas CH4 emissions are affected by feeding events because emissions peak after feeding and then gradually decline until the next feeding event (Jonker et al. 2014; Muetzel et al. 2024). Renand and Maupetit (2016) also found smaller variability in CO2 emissions and the CH4 to CO2 ratio than in CH4 emissions measured with GreenFeed units from heifers; similarly, they found that the variation of the CH4 to CO2 ratio was greater than the variation of CO2 emissions.
Visits to the GreenFeed units
Planning a trial with GreenFeed measurements requires careful consideration of the minimum number of animals needed to detect differences between treatment means. This includes accounting for animals that may be excluded from data analysis because of low or no visitation to the GreenFeed unit or other reasons for exclusion before the end of the measurement phase (e.g. health issues). In the four experiments reported here, 10–28% of the animals exposed to the GreenFeed units did not visit the units after the 14 days of training, which is greater than the 10% estimated by Jonker et al. (2020b), and greater than the 10–17% of cows that did not visit the units in a grazing study (Waghorn et al. 2016). It is worth mentioning that the proportion of animals that did not visit the GreenFeed units was 2.5 times greater in the first experiment when the animals were exposed to the unit (Experiments 1 and 3) than when the same animals were used in the second experiment (Experiments 2 and 4). The visit frequency per day is a key factor that determines the number of measurement days required to estimate gas emissions. Additional exclusion criteria were imposed in the current study for the requirements of the current analysis, with the main additional criteria being that animals had to be visiting the GreenFeed units from the beginning of the measurement period, which increased the number of animals excluded, especially in Experiment 1. The greater exclusion of cow in Experiment 1 than in the other experiments is likely to be due to the technical problems encountered during the study with one of the GreenFeed units. Thus, there were more animals per unit when one unit was not working than the recommended maximum of 40 animals per unit, which will reduce the number of animals visiting the GreenFeed unit. Anecdotally, on the basis of ICAR Feed and Gas Working group Zoom meetings (Benzoni et al. 2023), it has been observed by GreenFeed users that the daily visit frequency and number of animals visiting the GreenFeed unit gradually increases if the number of animals per GreenFeed unit are gradually reduced from 40 to 10 animals per unit.
While the measurement period of Experiment 1 was underway, it was decided to increase the measurement period to have sufficient measurement days with two GreenFeed units in the paddock (Bosher et al. 2024). Measurement of 65% of the cows that visited the units was used in that study; however, for the current study, only data from 34% of the cows were used as one data exclusion criterion was that cows had to visit the units from the beginning of the study. The number of animals excluded from the current analyses from Experiments 2 to 4 ranged from 30% to 58%, which was bigger than the non-frequent GreenFeed visitors (40%) excluded in the studies of Velazco et al. (2017) and Waghorn et al. (2016). The four experiments reported here all had a large number of animals per GreenFeed unit (≥32), which is likely to have contributed to some of the small number of cattle that visited the GreenFeed units. Generally, the number of cattle that visited the GreenFeed unit was greater when the cattle had exposure to GreenFeed units in previous studies, as observed in Experiments 2 and 4.
The lactating cows visited the GreenFeed approximately twice per day on average, whereas heifers visited the unit four times per day on average. Both heifers and cows were in the range of visits per day that is usually observed in non-lactating and lactating cattle in grazing systems. (Hammond et al. 2015; Jonker et al. 2017a; Starsmore et al. 2024). The exact drivers for GreenFeed visitation frequency are unknown, but the number of animals per unit, bullying by dominant cattle, the proximity of the GreenFeed to the animals, the feed quality of the base diet, and age and energy requirements of the animals may affect the visit frequency. Most of the dairy cows (75%) received concentrates during milking (2–6 kg DM/day) and had a smaller number of accumulated visits to the GreenFeed units than did the non-supplemented cows (Fig. 2). Besides, all the cows were away from the units for approximately 2–3 h per day for milking and, for periods of time, there was only one unit working in Experiment 1, which doubled the stocking rate for periods. All these factors together might have contributed to the lower daily visitation in dairy cows than in heifers and the greater among-animal variability in cows than in heifers.
Diurnal variations in CH4 and CO2 emissions
The method to estimate the daily gas emissions (‘arithmetic mean’ or ‘time of the day’) did not affect the among-animal variability, i.e. the precision of the estimates, or the number of animals required to detect 10% of the mean differences. These results are in agreement with those of Manafiazar et al. (2016) who did not find differences in the repeatability and variability of CH4 and CO2 emissions calculated with any of the two averaging methods.
By definition, the mean is affected by extreme values and a high number of records containing extreme values will bias the mean, and that is the reason why some authors (Waghorn et al. 2016; Jonker et al. 2017a) include the hourly variation to estimate the gas emissions per day. The diurnal variations in CH4 emissions (maximum to minimum emission rate ratio) were greater than were the variations in CO2 emissions. In addition, the visit frequency to GreenFeed varied among hours, showing higher visit frequency at sunrise, before sunset and at midnight. A similar visitation pattern has also been found in other studies in housed animals and grazing animals (Alemu et al. 2017; Jonker et al. 2019), which supports the idea of factoring in the hourly variations to calculate the daily CH4 emissions. The variations between the maximum to minimum CH4 emission ratio detected by GreenFeed units for all experiments, were similar to the peak emissions to pre-feeding CH4 emissions ratio detected in growing heifers fed ad libitum when CH4 emissions were measured in respiration chambers (Jonker et al. 2014) and smaller than the ratio observed in non-lactating cows and heifers at restricted feeding levels in respiration chambers (Jonker et al. 2014; Della Rosa et al. 2023b). Animals fed ad libitum generally have more frequent and smaller meals than do those fed restricted, which reduces the diurnal variation of CH4 emission (Jonker et al. 2014). However, the CH4 emissions that occurred just after concentrate feeding during milking of the dairy cows were not captured because the cows were away from the GreenFeed units in the paddock (~1.5 h/milking), and therefore peak emissions might not have been captured, which might have contributed to the smaller diurnal variation of CH4 emissions.
The data analyses of the current study suggest that including the hourly variation to estimate the daily CH4 emissions from multiple spot samples did not increase the precision of the gas emission estimates compared with not including the hourly variation. However, if visit distribution and maximum to minimum emission ratio are not checked, then accounting for the hourly variation during the estimation of CH4 emissions from spot samples would be advisable.
Implications
The absolute gas emissions, daily GreenFeed unit visit frequency and measurement period duration were key parameters that defined the among-animal variation and minimum number of animals required to detect 10% of differences between treatment means. In total, 12–18 days of GreenFeed measurements were required to obtain a stable estimate of among-animal variation in CH₄ emissions in the current grazing cattle experiments. This duration allowed for the collection of 17–37 spot samples per cow (for lactating cows without supplementation or supplemented with different types or amounts of concentrates during milking) and 43–73 spot samples per heifer (for heifers grazing ryegrass-based pasture). However, when applied to future experiments, one needs to remember that this might be different under different management and feeding conditions, with different visit frequencies to the GreenFeed units, with a different number of animals per unit, and the findings presented here are for treatment comparison, not individual animal ranking.
In practical terms, generally, a minimum of 2 weeks of GreenFeed gas measurements would be advisable with grazing cattle; however, this period might have to be changed if GreenFeed visit frequency is lower or higher. The low visitation frequency can be mitigated by selecting animals with a high number of visits to the GreenFeed units after the training period, but it requires starting with a much greater number of animals. Furthermore, reducing the number of animals per GreenFeed unit generally increases the visitation per animal. Experiments with cows needed more animals (13–19 cows) to detect 10% of the mean differences between treatment means because of the greater among-animal variation in gas emissions than for heifers (9–11 heifers would be required).
The among-animal variation for the CH4 to CO2 ratio became stable during a similar number of measurement days as for CH4 emissions, whereas the among-animal variation for CO2 emissions became stable earlier than for CH4 emissions. The variability of both CO2 emissions and the CH4 to CO2 ratio was smaller than for CH4 emissions. The number of cattle required to start a trial should consider that a proportion of the animals (10–28% in the current studies) might not visit the GreenFeed units and therefore additional animals need to be included in a GreenFeed study to ensure that the study has sufficient statistical power to detect significant differences among treatments.
Data availability
The data that support the findings of this study are available from the corresponding author upon reasonable request.
Conflicts of interest
Arjan Jonker is a Associate Editor for the Australasian Dairy Science Symposium 2024 collection of Animal Production Science. To mitigate this potential conflict of interest, he had no editor-level access to this manuscript during peer review. The author(s) have no further conflicts of interest to declare.
Declaration of funding
The cow studies were funded by the Greenhouse Gas Inventory Research fund of the Ministry for Primary Industries (Wellington, New Zealand) and the heifer studies were funded by the AgResearch Strategic Science Investment fund (Ministry of Business Innovation and Employment, Wellington, New Zealand) and the New Zealand Agricultural Greenhouse Gas Research Centre (Palmerston North, New Zealand).
Acknowledgements
These animal studies would not have been possible without the support of the farm and technical staff of Dairy Farm No. 4 of Massey University, AgResearch Aorangi Research Farm and AgResearch Animal Nutrition and Physiology team.
References
Alemu AW, Vyas D, Manafiazar G, Basarab JA, Beauchemin KA (2017) Enteric methane emissions from low– and high–residual feed intake beef heifers measured using GreenFeed and respiration chamber techniques. Journal of Animal Science 95, 3727-3737.
| Crossref | Google Scholar | PubMed |
Arthur PF, Barchia IM, Weber C, Bird-Gardiner T, Donoghue KA, Herd RM, Hegarty RS (2017) Optimizing test procedures for estimating daily methane and carbon dioxide emissions in cattle using short-term breath measures. Journal of Animal Science 95, 645-656.
| Crossref | Google Scholar | PubMed |
Bennett CP, Ward JF, Jonker A (2021) Can GreenFeed automated emissions monitoring systems measure methane output from young red deer grazing pasture? New Zealand Journal of Animal Science and Production 81, 93-98.
| Google Scholar |
Benzoni L, Berry D, Dressler E, Hegarty R, Koning L, Mc Donnell C, Mcnaughton L, Ritchie G, Finocchiaro R, Van Breukelen A, García Rodríguez A, Gonzalez-Recio O, Richardson C, Villumsen TM, Gredler-Grandl B (2023) Greenfeed and sniffer standard operating procedure (SOP) in dairy and beef cattle. In ‘Proceedings of the ICAR Conference’. pp. 281–287. (ICAR: Arthur van Schendelstraat 650, 3511 MJ Utrecht, The Netherlands)
Bosher T, Della Rosa MM, Khan MA, Sneddon N, Donaghy D, Jonker A (2024) Methane emissions intensity in grazing dairy cows fed graded levels of concentrate pellets. New Zealand Journal of Agricultural Research 67, 296-302.
| Crossref | Google Scholar |
Charmley E, Williams SRO, Moate PJ, Hegarty RS, Herd RM, Oddy VH, Reyenga P, Staunton KM, Anderson A, Hannah MC (2016) A universal equation to predict methane production of forage-fed cattle in Australia. Animal Production Science 56, 169-180.
| Crossref | Google Scholar |
Della Rosa MM, Waghorn GC, Vibart RE, Jonker A (2023a) An assessment of global ruminant methane-emission measurements shows bias relative to contributions of farmed species, populations and among continents. Animal Production Science 63, 201-212.
| Crossref | Google Scholar |
Della Rosa MM, Duranovich FN, Pacheco D, Sandoval E, Khan MA, Biswas A, Jonker A (2023b) Forage type affects the temporal methane emission profiles in dairy cows fed fresh forages. Animal Feed Science and Technology 298, 115604.
| Crossref | Google Scholar |
Della Rosa MM, Pacheco D, Sandoval E, Jonker A (2024a) Simulating the number of spot-samples required to estimate the methane to carbon dioxide ratio in lambs and its relationship with methane yield. New Zealand Journal of Agricultural Research 67, 303-313.
| Crossref | Google Scholar |
Della Rosa MM, Duranovich FN, Pacheco D, Muetzel S, Janssen PH, Jonker A (2024b) Substituting ryegrass-based pasture with graded levels of forage rape in the diet of lambs reduced post-feeding variation in methane emissions. Animal Feed Science and Technology 308, 115862.
| Crossref | Google Scholar |
Della Rosa MM, Bosher TJ, Khan MA, Sandoval E, Molano G, Dobson Hill B, Duranovich FN, Jonker A (in press) Increasing concentrate feeding level in early lactation dairy cows fed pasture linearly decreased methane yield and intensity. Animal Nutrition. doi:10.2139/ssrn.4846996
Dressler EA, Bormann JM, Weaber RL, Rolf MM (2023) Characterization of the number of spot samples required for quantification of gas fluxes and metabolic heat production from grazing beef cows using a GreenFeed. Journal of Animal Science 101, skad176.
| Crossref | Google Scholar |
Gerber PJ, Hristov AN, Henderson B, Makkar H, Oh J, Lee C, Meinen R, Montes F, Ott T, Firkins J, Rotz A, Dell C, Adesogan AT, Yang WZ, Tricarico JM, Kebreab E, Waghorn G, Dijkstra J, Oosting S (2013) Technical options for the mitigation of direct methane and nitrous oxide emissions from livestock: a review. Animal 7, 220-234.
| Crossref | Google Scholar | PubMed |
Gunter SA, Bradford JA (2017) Technical Note: Effect of bait delivery interval in an automated head-chamber system on respiration gas estimates when cattle are grazing rangeland. The Professional Animal Scientist 33, 490-497.
| Crossref | Google Scholar |
Hammond KJ, Humphries DJ, Crompton LA, Green C, Reynolds CK (2015) Methane emissions from cattle: estimates from short-term measurements using a GreenFeed system compared with measurements obtained using respiration chambers or sulphur hexafluoride tracer. Animal Feed Science and Technology 203, 41-52.
| Crossref | Google Scholar |
Herd RM, Velazco JI, Arthur PF, Hegarty RS (2016) Proxies to adjust methane production rate of beef cattle when the quantity of feed consumed is unknown. Animal Production Science 56, 231-237.
| Crossref | Google Scholar |
Hristov AN, Melgar A (2020) Short communication: Relationship of dry matter intake with enteric methane emission measured with the GreenFeed system in dairy cows receiving a diet without or with 3-nitrooxypropanol. Animal 14, s484-s490.
| Crossref | Google Scholar | PubMed |
Hristov AN, Oh J, Giallongo F, Frederick T, Harper MT, Weeks H, Branco AF, Price WJ, Moate PJ, Deighton MH, Williams SRO, Kindermann M, Duval S (2016) Short communication: comparison of the GreenFeed system with the sulfur hexafluoride tracer technique for measuring enteric methane emissions from dairy cows. Journal of Dairy Science 99, 5461-5465.
| Crossref | Google Scholar | PubMed |
Huhtanen P, Cabezas-Garcia EH, Utsumi S, Zimmerman S (2015) Comparison of methods to determine methane emissions from dairy cows in farm conditions. Journal of Dairy Science 98, 3394-3409.
| Crossref | Google Scholar | PubMed |
Huhtanen P, Ramin M, Hristov AN (2019) Enteric methane emission can be reliably measured by the GreenFeed monitoring unit. Livestock Science 222, 31-40.
| Crossref | Google Scholar |
Iqbal MW, Draganova I, Morel PCH, Morris ST (2022) Factors affecting grazing and rumination behaviours of dairy cows in a pasture-based system in New Zealand. Animals 12, 3323.
| Crossref | Google Scholar |
Iqbal MW, Draganova I, Henry Morel PC, Todd Morris S (2023) Variations in the 24 h temporal patterns and time budgets of grazing, rumination, and idling behaviors in grazing dairy cows in a New Zealand system. Journal of Animal Science 101, skad038.
| Crossref | Google Scholar |
Jonker A, Molano G, Antwi C, Waghorn G (2014) Feeding lucerne silage to beef cattle at three allowances and four feeding frequencies affects circadian patterns of methane emissions, but not emissions per unit of intake. Animal Production Science 54, 1350-1353.
| Crossref | Google Scholar |
Jonker A, Scobie D, Dynes R, Edwards G, De Klein C, Hague H, McAuliffe R, Taylor A, Knight T, Waghorn G (2017a) Feeding diets with fodder beet decreased methane emissions from dry and lactating dairy cows in grazing systems. Animal Production Science 57, 1445-1450.
| Crossref | Google Scholar |
Jonker A, Molano G, Koolaard J, Muetzel S (2017b) Methane emissions from lactating and non-lactating dairy cows and growing cattle fed fresh pasture. Animal Production Science 57, 643-648.
| Crossref | Google Scholar |
Jonker A, Farrell L, Scobie D, Dynes R, Edwards G, Hague H, McAuliffe R, Taylor A, Knight T, Waghorn G (2019) Methane and carbon dioxide emissions from lactating dairy cows grazing mature ryegrass/white clover or a diverse pasture comprising ryegrass, legumes and herbs. Animal Production Science 59(6), 1063-1069.
| Crossref | Google Scholar |
Jonker A, Green P, Waghorn G, van der Weerden T, Pacheco D, de Klein C (2020a) A meta-analysis comparing four measurement methods to determine the relationship between methane emissions and dry-matter intake in New Zealand dairy cattle. Animal Production Science 60, 96-101.
| Crossref | Google Scholar |
Jonker A, Antwi C, Biswas AA, Gunter S, Hristov AN, Martin C, Minnée EMK, Renand G, Waghorn G (2020b) The ‘GreenFeed’ automated methane measurement system to determine enteric methane emissions from ruminants. In ‘Guidelines for estimating methane emissions from individual ruminants using: GreenFeed, ‘sniffers’, hand-held laser detector and portable accumulation chambers’. (Eds A Jonker, GC Waghorn) pp. 17–32. (Ministry for Primary Industries: Wellington, New Zealand)
Kjeldsen MH, Johansen M, Weisbjerg MR, Hellwing ALF, Bannink A, Colombini S, Crompton L, Dijkstra J, Eugène M, Guinguina A, Hristov AN, Huhtanen P, Jonker A, Kreuzer M, Kuhla B, Martin C, Moate PJ, Niu P, Peiren N, Reynolds C, Williams SRO, Lund P (2024) Predicting CO2 production of lactating dairy cows from animal, dietary, and production traits using an international dataset. Journal of Dairy Science 107, 6771-6784.
| Crossref | Google Scholar | PubMed |
Madsen J, Bjerg BS, Hvelplund T, Weisbjerg MR, Lund P (2010) Methane and carbon dioxide ratio in excreted air for quantification of the methane production from ruminants. Livestock Science 129, 223-227.
| Crossref | Google Scholar |
Manafiazar G, Zimmerman S, Basarab J (2016) Repeatability and variability of short-term spot measurement of methane and carbon dioxide emissions from beef cattle using GreenFeed emissions monitoring system. Canadian Journal of Animal Science 97, 118-126.
| Crossref | Google Scholar |
McGinn SM, Coulombe J-F, Beauchemin KA (2021) Technical note: validation of the GreenFeed system for measuring enteric gas emissions from cattle. Journal of Animal Science 99, skab046.
| Crossref | Google Scholar |
Muetzel S, Hannaford R, Jonker A (2024) Effect of animal and diet parameters on methane emissions for pasture-fed cattle. Animal Production Science 64, AN23049.
| Crossref | Google Scholar |
Pereira ABD, Utsumi SA, Dorich CD, Brito AF (2015) Integrating spot short-term measurements of carbon emissions and backward dietary energy partition calculations to estimate intake in lactating dairy cows fed ad libitum or restricted. Journal of Dairy Science 98, 8913-8925.
| Crossref | Google Scholar | PubMed |
R Core Team (2023) R: A Language and Environment for Statistical Computing. R Foundation for Statistical Computing, Vienna, Austria. Available at https://wwwR-projectorg/
Renand G, Maupetit D (2016) Assessing individual differences in enteric methane emission among beef heifers using the GreenFeed Emission Monitoring system: effect of the length of testing period on precision. Animal Production Science 56, 218-223.
| Crossref | Google Scholar |
Rogelj J, Shindell D, Jiang, K (2018) Mitigation pathways compatible with 1.5°C in the context of sustainable development. In ‘Global warming of 1.5°C. An IPCC special report on the impacts of global warming of 1.5°C above pre-industrial levels and related global greenhouse gas emission pathways, in the context of strengthening the global response to the threat of climate change’. (Eds V Masson-Delmotte, P Zhai, H-O Pörtner, et al.) pp. 93–174. (Intergovernmental Panel on Climate Change: Geneva, Switzerland)
Rook AJ, Tallowin JRB (2003) Grazing and pasture management for biodiversity benefit. Animal Research 52, 181-189.
| Crossref | Google Scholar |
Starsmore K, Lopez-Villalobos N, Shalloo L, Egan M, Burke J, Lahart B (2024) Animal factors that affect enteric methane production measured using the GreenFeed monitoring system in grazing dairy cows. Journal of Dairy Science 107, 2930-2940.
| Crossref | Google Scholar | PubMed |
Velazco JI, Herd RM, Cottle DJ, Hegarty RS (2017) Daily methane emissions and emission intensity of grazing beef cattle genetically divergent for residual feed intake. Animal Production Science 57, 627-635.
| Crossref | Google Scholar |
Waghorn GC, Jonker A, Macdonald KA (2016) Measuring methane from grazing dairy cows using GreenFeed. Animal Production Science 56, 252-257.
| Crossref | Google Scholar |