Beef cattle methane emission estimation using the eddy covariance technique in combination with geolocation

https://doi.org/10.1016/j.agrformet.2020.108249Get rights and content

Highlights

  • Free-ranging cattle methane emissions were measured using eddy covariance.

  • A footprint model was used to estimate cattle contribution to the measured flux.

  • Estimated cattle methane emissions were of 220 ± 10 g CH4 LU−1 day−1.

  • No relation between methane emission and cattle behavior was observed.

  • The present technique can be used for other gases, including CO2.

Abstract

Methane emissions of a grazing herd of Belgian Blue cattle were estimated per individual on the field by combining eddy covariance measurements with geolocation of the cattle and a footprint model. This method allows the measurement of outdoor non-invasive methane emissions but is complex and subject to methodological issues. Estimated emissions were 220 ±35 g CH4 LU−1 day−1 (grams of methane per livestock unit per day), where the uncertainty corresponds to the random error and does not include any possible systematic error. Cattle behavior was also monitored and presented a clear daily pattern of activity with more intense grazing after sunrise and before sunset. However, no significant methane emission pattern could be associated with it, the diurnal emission variation being lower than the measurement precision.

Introduction

Ruminants are able to digest cellulose which makes them incredibly apt to transform raw forage, like grass, into high quality products. This digestive characteristic is due to an association with a very specific microbial flora present in the rumen or hindgut which allows the transformation of complex plant material into digestible fatty acids (acetate, lactate, propionate or butyrate). However, this transformation is accompanied by the co-production of methane, a potent greenhouse gas, which is mostly released through eructation (Broucek, 2014).

The current standard measurement method for cattle methane emissions is the metabolic chamber. This method calculates a mass balance between methane entering and leaving a sealed ventilated chamber containing an animal. Tracer methods are the major alternative for grazing ruminants; they involve the use of an external (e.g., SF6 released by an ingested canister) or internal (e.g., metabolic CO2 emissions) tracer released at a known rate from the animal's rumen. Measuring tracer and methane concentration ratios in excreted gases allows the computation of methane fluxes. Both techniques are accurate with a precision commonly higher than 90%, but require lots of animal handling (Storm et al., 2012), are rather invasive and could impact the natural grazing behavior of cattle. Emerging methods rely on the use of proxies; they are based on the relationship between methane emissions and the composition of matrices that are easy to sample such as feces or milk (Dehareng et al., 2012; Vanlierde et al., 2018). This method is valid as long as the composition of the proxies and the characteristics of the sampled animals (i.e., breed, intake level, physiological status, etc.) remain within the range of variability of the database that was used to develop the relationship. In addition to these animal-centered approaches, measurement methods have been developed that work at the scale of the environment in which the animals evolve. Some of these techniques simply reproduce lower scale methods (i.e., by considering the barn or the feeding trough as a chamber or by adding a tracer gas in a ventilated barn at a known rate and measuring the methane/tracer ratio) while others involve micro-meteorological methods (Johnson and Johnson, 1995; Storm et al., 2012). The latter are promising because they allow measurements to be recorded of the emission rate of the whole herd, on the field, with a half-hour time resolution, little animal handling and without disturbing the cow's natural behavior. Among micrometeorological methods, eddy-covariance (EC) is well suited for measurements in a pasture with low cattle density over large areas, and has become more affordable with the release of fast and precise optical methane analyzers. Nevertheless, applying this measurement method to grazed pastures is challenging due to a combination of source complexity (i.e., spatial and temporal variation in animal locations and emission intensities) and limitations in methodology specific to EC (Dumortier et al., 2017; Wohlfahrt et al., 2012).

Cattle emissions are not constant over time. Most of the CH4 produced escapes through the mouth, with 83% of emissions associated with eructation and 15% associated with respiration (Hammond et al., 2016). Cattle eruct 15 to 28 times each hour (every 130 to 230 s) according to the composition of their diet, feed intake levels and physiology (Blaise et al., 2018). Moreover, methane emissions vary throughout the day, peaking approximately 2 hours after feeding followed by a decrease until the next feeding event (Blaise et al., 2018). Cattle methane emissions thus present a 24-hour emission pattern which can be related to their feeding behavior (Hammond et al., 2016; Hegarty, 2013).

When using EC, the measured covariance corresponds to the vertical flux at one specific point that is representative of exchanges within the footprint, the area “sensed” by the flux measurement device. This footprint can be modeled through a set of functions that weight the respective contribution of each element of the surface to the measured vertical flux (Rannik et al., 2012), known as a footprint model. However, animals act as moving CH4 sources which may wander in or out of the footprint. Therefore, fluxes measured through eddy covariance must be combined with a footprint model as well as information about the cattle's location on the pasture in order to estimate the animals’ contribution to the measured flux. The ability of this approach to provide reliable emission estimates was previously tested using artificial sources (Dumortier et al., 2019). Previous investigations by Heidbach et al. (2017) showed that the FFP (Flux Footprint Prediction) model presented by Kljun et al. (2015) was the most efficient of the four tested models as long as the artificial source was located further from the mast carrying the sensors than the footprint peak (maximum of the footprint function). One of the main drawbacks of this model is that sources are assumed to be at ground level, while cattle emissions are emitted at muzzle height (i.e., up to 1 m height). To tackle this issue, Coates et al. (2017) simulated free-range cattle with artificial methane sources scattered on a field at a height of 0.8 m. They were able to estimate artificial source emissions with an error of 10% regardless of the distance between the source and the mast by using a Lagrangian stochastic model which could consider source heights. Because stochastic approaches require high computational power, Dumortier et al. (2019) tried to assess to what extent ready-to-use footprint models, that do not consider source height, could be stretched beyond the conditions for which they were designed in order to estimate methane emissions from elevated artificial sources. They concluded that emissions could be correctly estimated (error of less than 15%) using the analytical Kormann & Meixner (2001) footprint model when the artificial source was located further from the mast than the footprint peak.

These results strengthen the idea that EC can be used to estimate point source emissions of methane from cattle in field conditions. Felber et al. (2015) were the first to put this idea into practice. They calculated an emission per dairy cow by combining EC with cow geolocation data and the Kormann & Meixner (2001) footprint model. The experiment was run on a 3.6 ha pasture divided into 6 paddocks which were either very close to or distant from the mast. Every few days animals were transferred from one paddock to another (rotational grazing). This resulted in high stocking densities at the pasture level (5.5 LU ha−1; LU, livestock unit) but very high stocking densities in the occupied paddock (up to 33 LU ha−1). For paddocks close to the mast (less than 60 m), measured methane emission levels compared reasonably well (difference of less than 5%) with those obtained from metabolic chambers hosting dairy cows with similar milk production levels and body weights. However, for paddocks more distant from the mast, measured emissions per animal were lower and compared poorly to metabolic chambers, suggesting an imprecision of the footprint model. Other authors have successfully used a similar approach in different contexts (Prajapati and Santos, 2017), researching different gases (Gourlez de la Motte et al., 2019) or using different footprint tools (Coates et al., 2018).

In this work, free ranging cattle methane emissions on the pasture are estimated by combining eddy covariance with geolocation. This approach provides a variety of situations with the herd at rest, gathered at various distances from the mast, and cows more dispersed on the pasture during grazing. Moreover, we are able to rely on a methane emission estimation method previously validated on the same site with an artificial tracer (Dumortier et al., (2019). Our main objectives are:

  • To adapt an existing method combining the EC technique and a footprint model (Dumortier et al., 2019) with cattle geolocation data in order to estimate mean enteric emissions per livestock unit (LU). The validity of this approach is estimated by the internal consistency of the results (stability of emissions, uncertainties and impact of meteorological conditions).

  • To estimate methane emissions of Belgian Blue cattle on a typical Belgian commercial farm and to compare these with existing estimates (including IPCC default values).

  • To investigate the relation between methane emissions and cattle behavior.

Section snippets

Experimental site

The ICOS-candidate Dorinne Ecosystem Station (BE-Dor) is a 4.2 ha pasture located in Dorinne, Belgium (location: 50˚18’42.84”N; 4˚58’4.8”E; 248 m above sea level). The site is the location of previous investigations and is fully described in Dumortier et al. (2017) and in Gourlez de la Motte et al. (2019). The pasture is situated on a loamy plateau with a calcareous and/or clay substrate. Its species composition is: 66% grasses, 16% legumes and 18% other species. The dominant species are

Cattle behavior and distribution

For each campaign, cattle were found to be well spread over the whole pasture when grazing, while they gathered near the water troughs and the trees bordering the pasture when ruminating or idling (Fig. 5). We also observed that grazing behaviors followed a diurnal pattern; animals grazed mainly during the day with peak activities just after sunrise and before sunset (Fig. 6). This behavior was confirmed by GPS trackers which revealed a strong correlation between cattle movement and grazing

Validity of the method

The first objective was to provide estimates of the mean enteric CH4 emissions per livestock unit by combining the EC technique with a footprint model and cattle geolocation data. The combination of EC with geolocation allows stable and realistic estimations of cattle methane emissions to be made with measurement campaigns as short as one month (197 to 229 g CH4 LU−1 day−1). Obtained methane emissions were realistic and the regression slope 95% uncertainty range was estimated between 18 and 40%

Conclusions

Estimated methane emissions from cattle raised at the BE-Dor site were 220 ±35 g CH4 LU−1 day−1, where the uncertainty corresponds to the random error and does not include any possible systematic error. This figure corresponds to previous estimates and should be representative of common rearing practices in south Belgium.

The present technique is not limited to methane and, provided the appropriate analyzers are available, can be used to estimate other gaseous animal emissions like CO2 (

Declaration of Competing Interest

None.

Acknowledgments

The research site activities were supported by the Walloon region (Direction Générale Opérationnelle de l'Agriculture, des Ressources naturelles et de l'Environnement, Département du Développement, Direction de la Recherche, Belgium), through projects D31-1235 and D31-1278. The authors wish to thank Frédéric Wilmus who was in charge of the site maintenance during the experiment, Yves Brostaux who provided statistical insight into functional analysis and error quantification and Adrien Paquet

References (41)

  • G. Fratini et al.

    Relative humidity effects on water vapour fluxes measured with closed-path eddy-covariance systems with short sampling lines

    Agricultural and Forest Meteorology

    (2012)
  • L. Gourlez de la Motte et al.

    Herd position habits can bias net CO2 ecosystem exchange estimates in free range grazed pastures

    Agricultural and Forest Meteorology

    (2019)
  • L. Gourlez de la Motte et al.

    Carbon balance of an intensively grazed permanent grassland in southern Belgium

    Agricultural and Forest Meteorology

    (2016)
  • K.J. Hammond et al.

    Review of current in vivo measurement techniques for quantifying enteric methane emission from ruminants

    Animal Feed Science and Technology

    (2016)
  • R.S. Hegarty

    Applicability of short-term emission measurements for on-farm quantification of enteric methane

    Animal

    (2013)
  • K. Heidbach et al.

    Experimental evaluation of flux footprint models

    Agricultural and Forest Meteorology

    (2017)
  • D.R. Lockyer

    Methane emissions from grazing sheep and calves

    Agriculture, Ecosystems & Environment

    (1997)
  • A. Vanlierde et al.

    Short communication: Development of an equation for estimating methane emissions of dairy cows from milk Fourier transform mid-infrared spectra by using reference data obtained exclusively from respiration chambers

    Journal of Dairy Science

    (2018)
  • A.L.H. Andriamandroso et al.

    A review on the use of sensors to monitor cattle jaw movements and behavior when grazing

    Biotechnol. Agron. Soc. Environ.

    (2016)
  • Arrêté ministériel exécutant l'arrêté du Gouvernement wallon du 3 septembre 2015 relatif aux aides...
  • Cited by (5)

    View full text