Beef cattle spend their days feeding, ruminating, lying, standing and moving, and every one of those behaviors leaves a measurable trace in a wearable sensor. A new study of 1,032 Japanese Black cattle has now traced how much of that daily behavior is written in the animals’ genes, and whether those genetic tendencies connect to the traits that matter most at slaughter: carcass weight, rib eye area, fat thickness, yield rate and marbling. The answer, published in Genetics Selection Evolution, is nuanced. Behavior is heritable but only weakly so, and its genetic links to carcass traits, while real and interpretable, are too faint to sharpen genomic predictions of breeding values.
The research team, led by Akio Onogi, monitored cattle at two research stations of the Livestock Improvement Association of Japan, one in Hiroshima and one in Hokkaido. Behavior was recorded automatically every ten minutes by a commercial monitoring system called U-motion, which classifies each ten-minute window into one of six classes based on acceleration and barometric sensor outputs: feeding, lying, moving, standing, ruminating while lying, and ruminating while standing. Combining the two rumination categories produced a seventh class, simply called ruminating. Each animal carried 144 records per day, and monitoring ran from June 2021 to March 2024, averaging 439 days per animal.
From these dense records the researchers constructed six metrics for each behavioral class: daily total duration, daily frequency, average duration, longest duration, average interval, and longest interval, calculated over weekly windows. That yielded 42 non-ratio behavioral traits. They then went a step further and computed ratio-based traits, dividing one behavior by another, for example the ratio of daily feeding duration to ruminating duration. With 21 possible pairs of classes and six metrics, this added 126 ratio traits, for a total of 168 behavioral traits analyzed. Ratios were treated as missing whenever either component was zero, and preliminary checks showed the choice of denominator barely affected the estimated genetic parameters.
All 1,032 animals were genotyped, and after quality pruning 36,572 SNPs remained for estimating a genomic relationship matrix. Behavioral phenotypes were Box-Cox transformed to approximate normality before analysis. Repeatability models fitted to weekly records estimated both permanent environmental and additive genetic effects, and bi-trait models estimated genetic correlations among behaviors. For the carcass comparison, behavioral values were first adjusted for fixed and permanent environmental effects and averaged per animal, then analyzed jointly with carcass records in Bayesian bi-trait models, with significance judged by whether the 95 percent highest density interval excluded zero.
The headline genetic finding is that heritability of behavior is generally low. Across all behavioral traits, estimates ranged from 0.02 to 0.37, and heritability accounted for only 22.7 to 67.5 percent of repeatability, meaning much of the consistent variation between animals was environmental rather than genetic. Pattern mattered more than behavior type: daily total duration, daily frequency, average duration and average interval were the most heritable metrics, with ranges up to 0.37, while longest duration and longest interval were the least, dipping to 0.02. A two-way analysis of variance confirmed that the metric, not the behavioral class, was the dominant source of variation in both repeatability and heritability. These feeding-behavior estimates aligned well with earlier studies in Nellore, multi-breed European, and Holstein cattle, although rumination and lying time heritabilities here, at 0.16 and 0.22, were lower than previously reported in dairy cows.
Among the 544 animals slaughtered and graded by the Japan Meat Grading Association, genetic correlations between behavior and carcass traits ranged from −0.28 to 0.29, modest in magnitude but statistically detectable for every carcass trait except rib thickness. Of 252 non-ratio behavior-carcass combinations, 20 showed significant genetic correlations; among 756 ratio-trait combinations, 62 did, and the component behaviors behind those significant ratios were strikingly uneven, dominated by feeding, moving and lying, suggesting the signals were not random.
The interpretable patterns are the study’s most striking contribution. Animals that fed less frequently and for shorter durations, but ruminated more frequently and for longer, tended to have genetically greater carcass weight, and the same held for animals that moved less relative to ruminating. The authors suggest these ratios may reflect feed efficiency: cattle that convert feed efficiently ruminate more and move less, and finish heavier. Rib eye area showed the opposite flavor of association, rising genetically in animals that spent more time lying relative to moving or standing, a somewhat unexpected result given that physical activity generally promotes muscle development. Subcutaneous fat thickness was thinner in animals that spent more time feeding relative to ruminating, and yield rate rose in animals with more frequent, longer feeding relative to rumination, a pattern opposite to that seen for carcass weight, indicating that efficient feeders are not necessarily the highest-yielding animals. Beef marbling score was positively correlated with longer ruminating while lying and negatively with moving frequency.
Despite these meaningful genetic signals, behavior failed as a predictor. In five-fold cross-validation on the 544 carcass-recorded animals, bi-trait mixed-effect models that added a behavioral trait as a secondary trait produced prediction accuracies equivalent to single-trait carcass models, no matter which of four selection rules chose the secondary trait: highest heritability, highest absolute genetic correlation, the product of the two, or minimal prediction error variance. The explanation is arithmetic. Carcass traits in this population were highly heritable, at 0.44 to 0.72, while behavioral traits were weakly heritable and only weakly correlated with them. Multi-trait genomic prediction helps most when the focal trait has low heritability and the correlated trait is strongly linked to it, conditions that do not hold here. The authors conclude that behavioral traits are unlikely to help select bull candidates for carcass merit under current measurement devices and models.
The study also acknowledges its own limits. The U-motion system’s classification accuracy has not been formally validated in the scientific literature, although a third-party assessment by the Danish institute SEGES Innovation tested it on 655 cows for heat and pregnancy detection, and more than 200,000 units had been sold in Japan by July 2024. Simplified statistical models that ignore pen area, stocking density and social interactions may inflate residual variance and depress heritability estimates, and long monitoring spans may violate the assumptions of a simple repeatability model; the authors suggest random regression on age as a future refinement. Genetic correlation estimates also carried wide uncertainty, with 95 percent highest density intervals of 0.23 to 0.48, and variance components in the bi-trait carcass analysis were not always consistent with single-trait results.
Even so, the work opens a door that earlier studies left closed. Feeding and rumination have long been scrutinized for their ties to feed intake and efficiency, but lying, standing and moving had almost no published genetic parameters, and genetic correlations between behavior and carcass traits had barely been characterized in beef cattle, particularly in a single breed where genetic influences may differ from the multi-breed populations studied elsewhere. Ratio-based behavioral traits, analyzed here as quantitative traits for the first time, proved heritable and repeatedly informative, offering finer interpretive resolution than raw durations alone. The authors suggest behavioral measurements may prove more useful as secondary traits for genuinely low-heritability targets, such as reproductive traits like sperm volume and sperm mortality in Japanese Black cattle, and they call for future work on genetic correlations between behavior and reproduction. For now, the message is that a cow’s daily routine carries a genetic signature, faint but legible, and that reading it well will require better models before it can pay off in the breeding program.
Subject of Research: Heritability of wearable-sensor behavioral traits and their genetic correlations with carcass traits in Japanese Black beef cattle
Article Title: Heritability of behavioral traits and their genetic correlations with carcass traits in beef cattle
Article References: Onogi, A., Oya, Y., Watanabe, T., Ogino, A., Shinomiya, M., & Kurogi, K. (2026). Heritability of behavioral traits and their genetic correlations with carcass traits in beef cattle. Genetics Selection Evolution, 58(1), Article 61. https://doi.org/10.1186/s12711-026-01082-5
Image Credits: AI Generated
DOI: 10.1186/s12711-026-01082-5
Keywords: beef cattle, behavioral genetics, wearable sensors, heritability, genetic correlations, carcass traits, Japanese Black, rumination, feeding behavior, genomic prediction, breeding values, behavioral
News Source: Juliet Wilcox. (October 29, 2026). Research report: Heritability of behavioral traits and their genetic correlations with carcass traits in beef cattle. Scienmag.
