Currently, various attempts are being made to
implement breeding schemes aimed at producing low
methane (CH4) emitting cows. We investigated the
persistence of differences in CH4 emission between
groups of cows categorized as either low or high emitters
over a 5-mo period. Two feeding regimens (pasture
vs. indoors) were used. Early- to mid-lactation Holstein
Friesian cows were categorized as low or high emitters
(n = 10 each) retrospectively, using predictions from
milk mid-infrared (MIR) spectra, before the start of the
experiment. Data from MIR estimates and from measurements
with the GreenFeed (GF; C-Lock Technology
Inc., Rapid City, SD) system over the 5-mo experiment
were combined into 7-, 14-, and 28-d periods. Feed
intake, eating and ruminating behavior, and ruminal
fluid traits were determined in two 7-d measurement
periods in the grazing season. The CH4 emission data
were analyzed using a split-plot ANOVA, and the repeatability
of each of the applied methods for determining
CH4 emission was calculated. Traits other than CH4
emission were analyzed for differences between low and
high emitters using a linear mixed model. The initial
category-dependent differences in daily CH4 production
persisted over the subsequent 5 mo and across 2 feeding
regimens with both methods. The repeatability analysis
indicated that the biweekly milk control scheme,
and even a monthly scheme as practiced on farms,
might be sufficient for confirming category differences.
However, the relationship between CH4 data estimated
by MIR and measured with GF for individual cows was
weak (R2 = 0.26). The categorization based on CH4
production also generated differences in CH4 emission
per kilogram of milk; differentiation between cow categories
was not persistent based on milk MIR spectra
and GF. Compared with the high emitters, low emitters
tended to show a lower acetate-to-propionate ratio
in ruminal volatile fatty acids, whereas feed intake and
ruminating time did not differ. Interestingly, the low
emitters spent less time eating than the high emitters.
In conclusion, the CH4 estimation from analyzing the
milk MIR spectra is an appropriate proxy to form and
regularly control categories of cows with different CH4
production levels. The categorization was also sufficient
to secure similar and persistent differences in emission
intensity when estimated by MIR spectra of the milk.
Further studies are needed to determine whether MIR
data from individual cows are sufficiently accurate for
breeding.