Seasonal relationship between liver abscesses and feeding behavior in crossbred beef cattle

Abstract

This retrospective study evaluated how feeding behavior during the last 60 to 90 days on feed related to liver abscess prevalence in crossbred beef steers finished in summer or fall. Liver abscesses were scored at slaughter as a binary trait with 10.8% prevalence, and individual feed intake data were collected from 1,114 steers using an automated feeding system. Feeding behavior traits included daily dry matter intake, eating time, meal frequency, meal size, meal length, and variability measures. Meals were defined by grouping all feeding events that occurred within 49 min of each other into a single meal. Generalized linear mixed models assessed associations between liver abscess presence and feeding behavior; accounting for finishing season, beef breed type, heterosis, slaughter weight, slaughter age, and contemporary group. Across seasons, steers with less average daily gain were more likely to have liver abscesses (P = 0.01). Slow-gaining steers may benefit from earlier slaughter to reduce feed costs and days on feed, which could also reduce liver abscess risk. Dry matter intake and intake variability were not associated with abscess prevalence. Several feeding behaviors differed by season. In summer, steers with more consistent eating time tended to have greater liver abscess prevalence (P = 0.06), but these differences were more a function of mean differences rather than variability. In fall, steers with more frequent meals (P = 0.02), smaller meal sizes (P = 0.08), or greater day-to-day changes in eating time (P = 0.04) or meal frequency (P = 0.05) had increased liver abscess risk. Although managing cattle for more, smaller meals is a strategy for stable rumen conditions, these steers were self-selecting into these behaviors, which could be a function of underlying physiology. Maintaining more consistency in eating time and meal frequency may reduce liver abscess risk in the fall, but potential management strategies to promote these behaviors were unclear. This study provided initial evidence that late-finishing growth performance and fall feeding behavior patterns were associated with liver abscess risk in crossbred beef steers. Identifying slow-gaining cattle for early marketing or targeted nutritional strategies may help reduce economic losses and improve feedlot management.

Growth performance and certain fall feeding behaviors - like more meals and greater day-to-day variation - were associated with greater liver abscess risk in crossbred feedlot steers.

Introduction

Liver abscesses, pus-filled cysts caused by bacterial infection, are significant health and economic concerns in beef cattle. Prevalence varies widely, affecting 10% to more than 50% of slaughtered cattle depending on region and cattle type (Reinhardt and Hubbert 2015; Herrick et al. 2022; Grimes et al. 2024). Economic losses result from reduced weight gain, fewer pounds of saleable product including condemned livers, and increased carcass trimming, with estimated costs ranging from $40 to $400 million annually (Herrick et al. 2022; Broadway et al. 2024; Taylor et al. 2025). Common mitigation strategies include feeding subtherapeutic antibiotics and increasing dietary roughage (Reinhardt and Hubbert 2015; Broadway et al. 2024).

The biological mechanisms underlying liver abscess development are complex. The traditional theory on their development is that high-concentrate diets lower rumen pH, resulting in ruminal acidosis. Acid damage to the rumen lining allows bacteria to cross into portal circulation and reach the liver, where abscesses form (Nagaraja and Chengappa 1998). Recent research suggests an additional pathway may involve bacterial translocation from the small intestine (de Jong et al. 2012; Amachawadi and Nagaraja 2016; Lindholm-Perry et al. 2024).

Feeding behavior may have a relationship with rumen pH and ruminal acidosis (Schwartzkopf-Genswein et al. 2003a; DeVries et al. 2009; González et al. 2012) and has been hypothesized to affect liver abscesses (Amachawadi and Nagaraja 2016). Inconsistent feed delivery, which increases variation in intake, has been linked to reduced rumen pH and greater ruminal acidosis risk (Schwartzkopf-Genswein et al. 2004). However, a 2-d/wk disruption in feed delivery timing and quantity did not induce liver abscesses (Schneid et al. 2024). Feed intake often decreases following a ruminal acidosis event (DeVries et al. 2009; González et al. 2012), indicating variation in feed intake can both cause ruminal acidosis and result from it. Ruminal acidosis risk decreases when cattle eat more frequent meals rather than a few large meals (González et al. 2012). Feeding behavior and ruminal acidosis risk are also related to season, with changes in meal patterns between hot and cold weather (Brown-Brandl et al. 2003; Schwartzkopf-Genswein et al. 2003b; DelCurto-Wyffels et al. 2021) and greater ruminal acidosis risk during hot weather (Elam 1976; Tajima et al. 2007).

Natural feeding behavior is difficult to measure and is a complex interaction between the animal, pen mates, and feeding environment. Although pen-level intake can be measured in commercial feedlots, measuring individual animal intake requires specialized facilities, which may alter animal behavior. Feeding behavior can be modulated through management factors such as diet, feeding frequency, feeding timing, bunk space, and stocking density (González et al. 2012). In addition, feeding behavior depends on social dominance (González et al. 2012), and as cattle compete for feed, they eat more frequent, shorter meals at a faster rate (Corkum et al. 1994). Given these connections, clarifying how feeding behavior interacts with seasonal conditions to affect liver abscesses could provide useful insights for prevention. The objective of this retrospective analysis was to evaluate these relationships in crossbred beef cattle. We hypothesized that steers with larger and less frequent meals, and more erratic feeding behavior would be more likely to have liver abscesses, particularly in summer.

Methods

Animal data came from the existing U.S. Meat Animal Research Center (USMARC) database. All animal procedures were approved by the USMARC Institutional Animal Care and Use Committee, and cattle were harvested at a federally inspected packing plant. This study was a retrospective analysis of existing liver abscess and feeding behavior data at USMARC. Cattle included in the study had phenotypic observations for both traits through ongoing data collection protocols.

Animals

Steers from the USMARC Germplasm Evaluation (GPE) project in Clay Center, Nebraska, were used in this retrospective study. The GPE is a long-term, designed breeding program to evaluate breed and heterosis effects in beef cattle. The project currently evaluates 18 beef breeds with representation proportional to industry use in the U.S. This population was most recently described in Engle et al. (2025). Breeds were grouped into the following breed types:

• British (Angus, Hereford, Red Angus, Red Poll, Shorthorn, South Devon),

• Continental (Braunvieh, Charolais, Chianina, Gelbvieh, Holstein, Limousin, Maine Anjou, Pinzgauer, Salers, Simmental, Tarentaise), and

• Heat tolerant (Bonsmara, Brahman, Romosinuano).

Proportions by breed type are shown in Fig. 1. The average steer was 50% British, 45% Continental, and 5% heat tolerant breed types. Half of steers had at least 90% pedigree-predicted heterosis, and 5% had less than 10% heterosis.

Figure 1.

Proportions of British, continental, and heat tolerant breed types in crossbred beef steers.

Contemporary group definitions included slaughter date and experimental treatment groups such as implant types and dietary differences. Steers represented 24 contemporary groups with an average of 46 steers (range = 17 to 180 steers) per group (Table S1). The average steer was 467 d of age (SD = 15 d) and weighed 607 kg (SD = 60 kg) at slaughter.

Management

Calves were born in fall 2017 and 2018, and spring 2018, 2019, and 2020. Bulls were castrated after birth. Steers were comingled from different calving groups and entered the feedlot after weaning at 179 d of age (SD = 14 d, range = 130 to 216 d) and weighed 214 kg (SD = 33 kg). Fall-born calves were vaccinated with a 7-way clostridial vaccine and a modified-live vaccine for bovine rhinotracheitis, bovine virus diarrhea, parainfluenza 3, and bovine respiratory syncytial virus at pre-breeding and weaning. Spring-born calves received the 7-way clostridial vaccine at pre-breeding and pre-conditioning (∼1 month before weaning), and the modified-live vaccine at pre-breeding, pre-conditioning, and weaning. All calves received an injectable dewormer for internal parasites before weaning. Cattle health was monitored daily by farm operations staff, and illnesses were treated according to USMARC protocols under supervision of the veterinary staff.

In the feedlot, steers were fed a total mixed ration once per day ad libitum at approximately the same time each day. All diets included rumensin. Steers were fed a grower diet upon entry to the feedlot for 55 to 84 d, followed by a transition period for 28 to 49 d, and then a finishing diet for 146 to 190 d. For the first 19 to 30 d after weaning, coccidiostat was included in diets. Chlortetracycline was fed at 0.45 kg/animal for 5 d beginning approximately 10 d after weaning for steers slaughtered in summer 2019 (n = 162). Two groups were fed tylosin phosphate around the transition period to test targeted, short-term feeding. Steers slaughtered in winter 2018 (n = 159) were fed for 10 d beginning 12 d after the transition period. Steers slaughtered in winter 2019 (n = 141) were fed for 12 d with feeding occurring more than halfway through the 45-d transition period. Transitions included 3 steps of 70% grower 30% finisher, 50% grower 50% finisher, and 30% grower 70% finisher diets and were typically 1 to 2 wk/step. Diets fed during the growing and finishing periods are shown in Tables S2 to S6. Dietary differences were captured within contemporary group definitions and were modeled accordingly. Nutrient concentrations were calculated using feed nutrient values from NASEM (2016) and feed tag values for supplements (Table S2 to S6).

Liver abscesses

After slaughter, livers were scored by one trained veterinarian (Keele et al. 2016). Scoring was originally based on the Elanco liver grading system with no abscess, small abscesses, or severe abscesses (Elanco Animal Health 2025). Later, livers were scored based on presence or absence of abscesses because only 17 animals were ever scored with small abscesses. For analysis, a binary trait was created based on presence or absence of liver abscesses regardless of scoring system. Only 10.4% of steers had liver abscesses, which was less than industry averages. Liver abscess prevalence by year and slaughter season is shown in Fig. 2. The number of steers varied by year and season ranging from 141 to 357 steers (Fig. 2).

Figure 2.

Liver abscess prevalence by year and slaughter season for crossbred beef steers. Bars are labeled with n.

Feeding behavior

Data

Individual animal feed intake was recorded using an Insentec system (Insentec B.V., Marknesse, The Netherlands). On average, there were 5 steers/bunk (range = 2.9 to 6.4 steers/bunk). Intake test timing and length were optimized to phenotype more animals while also balancing the ongoing needs of designed research projects. Based on previous research, intake test lengths were at least 42 d to capture individual animal differences in feed intake (Thallman et al. 2018). After preliminary review of intake tests with associated liver abscess phenotypes, tests were selected if data were recorded until 1 to 13 d before slaughter. Hence, the study captured feeding behavior preceding slaughter. Figure 3 shows the timing of feed intake data relative to slaughter date for summer and fall tests, and Table 1 shows actual test dates and lengths. Figure 4 shows mean daily temperature humidity index across feed intake test days. Temperature humidity index was calculated as

Figure 3.

Timing of feeding behavior data collection relative to slaughter date in crossbred beef steers.

Table 1.

Summary of feed intake tests used to capture feeding behavior data on crossbred beef steers.

Season	Start	End1	Days	Steers
Fall 2018	8/29	12/1	94	80
		12/2	95	79
Summer 2019	5/1	7/7	67	32
		7/8	68	48
		7/14	74	38
		7/15	75	44
Fall 20192	8/21	12/4	93	58
		12/5	94	83
Summer 2020	5/6	7/18	73	354
Summer 2021	5/5	7/6	62	295

¹Tests can have multiple slaughter dates creating variation in end date.

²No data captured from 9 November 2019 to 21 November 2019.

Figure 4.

Mean temperature-humidity index across feed intake test periods in Clay Center, NE.

$$\mathit{THI}=\mathit{temp}-\left(0.55-\left(0.0055\mathrm{*}RH\right)\right)\mathrm{*}(\mathit{temp}-14.5)$$

where temp was the dry-bulb temperature in °C and RH was the relative humidity as a percent. Weather data were collected from an on-site weather station (South Central Station of the Automated Weather Data Network, operated by the High Plains Regional Climate Center Central Weather Station).

Data edits

Records were filtered based on data from steers with liver abscess phenotypes and their pen mates (n = 1,445 animals, 1,874,501 records) to better identify data quality issues. Data were edited based on system functionality. First, if there were 2 overlapping visits at the same feeder and one was inconsistent with surrounding data, the record with the anomaly was removed (n = 2). Second, if one steer was recorded eating from 2 feeders simultaneously, the first intake and eating rate were set to missing (n = 25,246), and the events were considered different meals if the time between visit starts was less than the 49-min meal interval (meal interval described in the next section). Missing data were ignored in subsequent calculations.

Records were removed for the following reasons:

• Length of the feeder visit was less than 0 min (n = 512) or greater than 23.75 min (99^th percentile, n = 18,741) (Ternman et al. 2022),

• As-fed intake was less than −0.2 kg (n = 10,611),

• As-fed intake divided by length of visit was greater than 1.06 kg/min (99^th percentile, n = 18,527), or

• As-fed intake was between −0.2 to 0 kg and the length of visit was longer than the 49-min meal interval (n = 161).

Any remaining negative dry matter intakes (DMI) were set to 0. The last day of data in each test period was removed because of incomplete data (n = 39,637). Steers were removed if they had fewer than 56 d of data (n = 271). A 35-d or 42-d test length accurately evaluated feed intake (Archer et al. 1997; Wang et al. 2006; Culbertson et al. 2015; Cassady et al. 2016; Manafiazar et al. 2017; Ahlberg et al. 2018; Thallman et al. 2018). Preliminary analysis showed a 56-d test length better characterized the CV for the feeding behavior traits (results not shown). Final test lengths ranged from 62 to 95 d (Table 1). After cleaning, there were 1,114 animals and 1,804,434 individual feeder visits across 403 d.

Meals

Feeder visits can be summarized into meals. For a meal, an animal might access feeder(s) with break(s) to drink water, to wait for an open feeder, to socialize, etc. A meal ends when a certain amount of time passes without any feeder activity. Meals were defined as feeding events separated by less than 49 min. This interval was defined based on the probability of starting a new meal in the next min (Howie et al. 2009). The 49-min meal interval was verified using a second method based on inflection points of first derivatives for meal traits (Brown-Brandl and Eigenberg 2015). The meal interval was longer than most literature estimates (Schwartzkopf-Genswein et al. 2002; Brown-Brandl and Eigenberg 2015; Kelly et al. 2020). Meals were allocated to the day that the meal started and ended when the animal did not access a feeder for at least 49 min. Hence, a single meal can contain multiple feeder visits. For each meal, meal DMI was the total DMI for all visits within the meal. Meal length was the time from the start of the first visit to the end of the last visit within the meal.

Traits

Phenotypes were calculated for each test day. Daily DMI was the total DMI for each day. Daily eating time was the total time spent in a feeder and did not include the gaps that were included in meal length. Daily meal frequency was the number of meals. The meal traits (meal DMI, meal length) were averaged within test day. Feed intake variability may contribute to ruminal acidosis (Cooper et al. 1999; González et al. 2012) and thus liver abscesses (Broadway et al. 2024). Variation across the entire intake test was quantified by the CV for each trait. Short-term variability was assessed by computing the mean absolute difference between each test day and the preceding day.

After all calculations, this study included the daily mean, CV across the test period, and mean absolute differences from the prior day for DMI, eating time, meal frequency, meal DMI, and meal length. In addition, average daily gain was the gain across the test period divided by the number of test days. Gain: feed was the average daily gain divided by average daily DMI. Summary statistics are shown in Table S7.

Statistical analysis

Data transformation

Many feeding behavior traits were right skewed and had a lower boundary at 0. The distribution of each trait was inspected for deviations from normality. Non-normality led to violations of residual assumptions when pre-adjusting the trait as detailed next. Traits were transformed based on the boxcox function in the MASS package in R (Box and Cox 1964; Venables and Ripley 2002). Transformations are listed in Table 2.

Table 2.

P-Values for the interaction between feeding behavior traits and season in the analysis of liver abscess prevalence in crossbred beef steers finished during summer or fall.

Trait	Trait × season5	Trait5
DMI, mean	0.34	0.29
DMI, CV1	0.12	0.19
DMI, change	0.39	0.41
Eating time, mean	0.12	0.68
Eating time, CV1	0.08 5
Eating time, change	0.02
Meal frequency, mean	0.06
Meal frequency, CV	0.46	0.06
Meal frequency, change	0.10
Meal DMI, mean	0.06
Meal DMI, CV2	0.16	0.95
Meal DMI, change1	0.63	0.71
Meal length, mean1	0.69	0.95
Meal length, CV3	0.89	0.78
Meal length, change4	0.46	0.72
Average daily gain	0.62	0.01
Gain: feed3	0.53	0.07

For feeding behavior × season interactions with P ≤ 0.10, results are presented within summer and fall feeding seasons. For feeding behavior traits with P ≤ 0.10 and no season interaction, results are presented across both seasons.

¹Log transformed.

²Negative square root transformation.

³Inverse transformation.

⁴Negative fourth root transformation.

⁵Bold values are P <= 0.10.

Pre-adjustment for feeding behavior

Because feed intake and feeding behavior can be systematically affected by animal characteristics, these traits were adjusted before analysis. The following linear model was fitted with the lm function in R (R Core Team 2024):

$$y_{ij}={\beta }_0+{\beta }_1{\mathit{age}\mathrm{ }on\mathrm{ }\mathit{test}}_i+{\beta }_2{\mathit{heterosis}}_i+{\beta }_3{\mathit{Continental}\mathrm{ }\mathit{breed}}_i+{\beta }_4{\mathit{heat}\mathrm{ }\mathit{tolerant}\mathrm{ }\mathit{breed}}_i+{\mathit{sire}}_j+e_{ij},$$

where y was a single feeding behavior phenotype and 1 model was fitted for each phenotype. β₀ was the intercept. The other β were regression coefficients for the animal’s age in d at the start of the feed intake test, the percent retained heterosis, the proportion of Continental breeds, and the proportion of heat tolerant breeds. The proportion of British breeds was the linear combination of the intercept minus the proportion of the 2 other breed types. Sire was the random effect for the animal’s sire and was independent and identically distributed as N(0, σ_s²). The random residual, e, was independent and identically distributed as N(0, σ_e²) with different residual variance, σ_e², per trait. Residuals were checked for normality, non-linear relationships with model effects, and homoscedasticity. The residual from the above model was the adjusted feeding behavior phenotype.

Model

Binary liver abscess was analyzed with generalized linear models. One model was fitted for each feeding behavior trait. Models were fit using the glmmTMB function in the glmmTMB package (Brooks et al. 2017; McGillycuddy et al. 2025) in R (R Core Team 2024). The model was:

$$y_{\mathit{ijk}}={\beta }_0+{\beta }_1{\mathit{feeding}\mathrm{ }\mathit{behavior}}_i\mathrm{\hspace{0.17em}*\hspace{0.17em}}{\mathit{season}}_j+{\beta }_2{\mathit{heterosis}}_i+{\beta }_3{\mathit{Continental}\mathrm{ }\mathit{breed}}_i+{\beta }_4{\mathit{heat}\mathrm{ }\mathit{tolerant}\mathrm{ }\mathit{breed}}_i+{\beta }_5{\mathit{slaughter}\mathrm{ }\mathit{weight}}_i+{\beta }_6{\mathit{slaughter}\mathrm{ }\mathrm{age}}_i+{\mathit{group}}_k+e_{\mathit{ijk}},$$

where y was the binary liver abscess phenotype (1 = abscess, 0 = no abscess). The model included the interaction between a single continuous adjusted feeding behavior trait and season (summer or fall intake test) to assess seasonal differences in behavior. Regressions were included for normalized slaughter weight and normalized slaughter age. Slaughter age was used instead of days on feed because these variables were correlated and slaughter age had more variability within contemporary group than days on feed. Group was a random effect for the contemporary group consisting of slaughter date and any experimental treatments during the animal’s life. This effect accounted for all dietary and management differences within and across time. Group was a random effect with different variance per season (summer, fall). All other effects were the same as in the prior model. For each model, residuals were assessed using the DHARMa package (Hartig 2024).

P-values less than or equal to 0.05 were considered significant and less than or equal to 0.10 were a trend. Seasonal effects on feeding behavior were assessed by the P-value for the feeding behavior × season interaction. For feeding behavior traits with at least seasonal trends, the behavior trait was evaluated for a non-zero regression coefficient within each season (2 tests). For feeding behavior traits without seasonal trends, the behavior trait was evaluated for a non-zero regression coefficient across seasons (1 test). Odds ratios were calculated for a 1-unit change in the feeding behavior trait. Odds ratios greater than 1 meant liver abscesses were associated with increases in the feeding behavior trait (transformed scale), and odds ratios less than 1 meant liver abscesses were linked with decreases in the feeding behavior trait.

Results and discussion

Model effects

Season tended to be associated with liver abscesses (P = 0.10) when not interacting with feeding behavior. Steers finished in fall tended to be more likely to have liver abscesses than steers finished in summer (odds ratio = 1.43). Reports from previous studies were inconsistent regarding the time of year with the highest prevalence. Grimes et al. (2024) reported more liver abscesses in July for 1 database and in December for another. Reinhardt and Hubbert (2015) summarized studies reporting the most liver abscesses in spring or summer. In the current study, seasonal diets differed in roughage and wet distillers grain concentration. The fall diets had less roughage and included wet distillers grains and the primary summer diet had more roughage and no wet distillers grains (Table S2 to S6). Because increasing roughage concentration (Reinhardt and Hubbert 2015) and decreasing distillers grains (Meyer et al. 2013) can reduce liver abscess risk, diet may partially explain the observed seasonal trend.

No other fixed effects were significant. Heterosis was not associated with liver abscess prevalence (P = 0.76), although heterosis has been linked to improved health outcomes in other contexts (Bunning et al. 2019). Likewise, breed type (P = 0.23) did not affect liver abscess risk, consistent with past results indicating minimal beef breed differences (Sperber et al. 2023). Slaughter age (P = 0.85) and slaughter weight (P = 0.44) were also not associated with liver abscesses, likely because most of the variation in these traits was already captured within contemporary groups.

Feeding behavior

For traits without significant feeding behavior × season interactions, P-values were shown in Table 2, and odds ratios averaged across season were shown in Table 3 (season = “both”).

Table 3.

P-Values and odds ratios for feeding behavior traits in crossbred beef steers.

Trait	Season	P-value	Odds ratio	Lower CI	Upper CI
Eating time, CV1	Summer	0.06	0.90	0.81	1.00
	Fall	0.37	1.10	0.89	1.37
Eating time, change	Summer	0.26	1.00	0.99	1.00
	Fall	0.04	1.01	1.00	1.02
Meal frequency	Summer	0.93	1.00	0.97	1.03
	Fall	0.02	1.08	1.01	1.15
Meal frequency, CV	Both	0.06	0.95	0.89	1.00
Meal frequency, change	Summer	0.99	1.00	0.90	1.11
	Fall	0.05	1.23	1.00	1.50
Meal DMI	Summer	0.49	1.02	0.97	1.07
	Fall	0.08	0.93	0.86	1.01
Average daily gain	Both	0.01	0.34	0.15	0.79
Gain: feed2	Both	0.07	1.22	0.99	1.51

¹Log transformed.

²Inverse transformation.

Steers with less gain were more likely to have liver abscesses (P = 0.01), and the distributions of gain by liver abscess phenotype were shown in Fig. 5. Previous research has shown cattle with liver abscesses had less gain (Nagaraja and Chengappa 1998; Reinhardt and Hubbert 2015). Less efficient steers also tended to have more liver abscesses (P = 0.07), likely reflecting the overlap between slow growth and poor efficiency. In the last 2 to 3 months on feed, slow-gaining steers with liver abscesses started with similar weaning weights but possibly plateaued in their growth curves. Carcass traits, including fat thickness and yield grade, did not differ by liver abscess phenotype (results not shown), providing no evidence that reduced gain was due to earlier physiological maturity.

Figure 5.

Average daily gain distributions for crossbred beef steers with and without liver abscesses.

Although this study identified associations, causal mechanisms cannot be inferred. Reduced gain may precede abscess formation due to underlying ruminal or hepatic processes, or it may be a consequence of abscess pathology. Regardless, slow-gaining steers could be identified as greater risk for liver abscesses later in the finishing phase. Because these cattle were less efficient and more likely to develop liver abscesses, earlier marketing may improve profitability and reduce risk through fewer days on feed (Reinhardt and Hubbert 2015; Amachawadi and Nagaraja 2016). These cattle may also benefit from dietary interventions like increasing roughage late in the finishing phase (Reinhardt and Hubbert 2015; Broadway et al. 2024).

Steers with a smaller CV for meal frequency tended to have greater liver abscess prevalence (P = 0.06). The CV for meal frequency reflected the variability in the number of meals per day relative to the mean. In this study, the CV difference between phenotype groups was less than 1%, and the SD for meal frequencies were identical. Thus, all variation was due to minimal differences in the means. Because mean meal frequency interacted with season (P = 0.02), the CV alone was unlikely to be a reliable indicator of liver abscess risk.

Feed intake did not differ between liver abscess phenotypes (P = 0.29), consistent with some prior research (Nagaraja and Chengappa 1998; Batista and Holland 2022). Reduced feed intake has been reported (Brink et al. 1990; Torres et al. 2024), and greater feed intake has been hypothesized to increase liver abscess risk (Nagaraja and Chengappa 1998). The current study did not support intake differences late in the finishing period, especially with low liver abscess prevalence and ad libitum feeding in a limited number of individual-animal bunks.

Seasonal differences

P-values for the feeding behavior × season interactions are shown in Table 2, and within season significance for interactions with P ≤ 0.10 are shown in Table 3.

Summer

Steers with more consistent daily eating time tended to have greater liver abscess prevalence (P = 0.06). The mean eating time CV was 31% across seasons (Table S7), whereas the summer mean was 50%. Heat stress during summer feeding periods (Fig. 4) may contribute to the more variation in eating time (Brown-Brandl et al. 2005). However, within summer, the difference between phenotype groups was less than 1%, suggesting limited biological relevance. Prior research indicated that eating-time variability could be influenced by social competition (DeVries and von Keyserlingk 2009) and dominance (Haskell et al. 2019), suggesting a potential avenue for future research.

Fall

Multiple feeding behavior traits were associated with abscess prevalence in the fall. Cattle eating more (P = 0.02), smaller meals (P = 0.08) with more daily variation in meal frequency (P = 0.05) were more likely to have liver abscesses. These meal traits were correlated (|r| > 0.60). Although DMI was well studied (Pritchard and Bruns 2003; Schwartzkopf-Genswein et al. 2003a), far less was known about meal-level behaviors, particularly considering facility designs. Steers could be selectively modifying meal behavior because of underlying physiology related to liver abscesses. These cattle were fed once daily ad libitum, which was expected to create fewer, larger meals relative to more frequent daily feedings (Nagaraja and Chengappa 1998; González et al. 2012). Clean bunk management can influence meal patterns but has not been shown to affect liver abscess (Erickson et al. 2003). In addition, meal behavior traits were lowly repeatable (Kelly et al. 2020), complicating attempts to modify meal behavior directly.

Greater day-to-day changes in eating time were associated with greater liver abscess prevalence in fall (P = 0.04), and only the day-to-day changes, not the test-wide CV, were significant. The absolute difference of 1 min/d was small relative to total eating time (approximately 1 h/d; Table S7) and may fall within measurement error. Although cattle were fed at similar times daily, day length decreased by 3.5 h over the fall test period, potentially affecting diurnal feeding patterns (Ray and Roubicek 1971; Rhind et al. 2002). Summer tests experienced less than 1 h change in day length. Characterizing diurnal feeding behavior may help disentangle seasonal differences and warrants further study.

Limitations

The present research provided important initial insights into feeding behavior and liver abscesses, particularly when cattle were fed during different seasons. However, all cattle were fed at the same temperate location in Nebraska with 3 years of replicates for summer and 2 years for fall feeding groups. Heat and cold stress effects were confounded with year, resulting in limited replication and environmental variability. This confounding restricted the ability to assess seasonal differences relative to temperature-humidity index, heat, or cold stress.

Although detailed individual animal records were available alongside liver abscess phenotypes, the prevalence of liver abscesses was quite low (11%), particularly relative to industry averages for beef x dairy cross cattle (Grimes et al. 2024). Over the years, steers were subject to similar management practices, including health and nutrition management. Although finishing diets had ingredient differences (Table S2 to S6), diets had some confounding with season. These factors limit the generalizability of the results and application to many types of production systems.

No other research is known to exist on feeding behavior and liver abscesses, and no repeatable indicator of liver abscesses in live cattle has been identified. Hence, it is unknown when feeding behavior data should be recorded to most closely relate to the physiological development of liver abscesses. All data were recorded in the 2 to 3 months before slaughter (Fig. 3), and all were evaluated after the post-weaning transition period. These factors could be important, but many groups are needed to more fully understand the relationship between feeding behavior observation and slaughter date. In addition, feeding behavior was measured in a facility with a limited number of individual animal bunks per pen. Facility design likely affects feeding behavior and limits extrapolation to commercial feedlot settings.

Conclusion

During the final 60 to 90 d before slaughter, crossbred beef steers with less gain were at a greater risk of liver abscesses, suggesting slow-gaining cattle could be prioritized for targeted management interventions. Earlier marketing of these steers could reduce both liver abscess risk and feed costs associated with extended days on feed. Feeding behaviors associated with liver abscesses differed by finishing season. In summer, no feeding behavior traits were related to liver abscess risk. In fall, more frequent meals and larger day-to-day changes in meal frequency and eating time were linked to more liver abscess prevalence. The cause of these behaviors and mechanisms to modulate feeding behaviors were unclear and needed further research. Overall, this retrospective study provided initial evidence that both late-finishing growth performance and season-specific feeding behaviors related to liver abscess risk. Additional research is needed to clarify the biological mechanisms connecting feeding behavior, season, and liver abscess development, and to determine whether feeding behavior can be practically leveraged to reduce liver abscess prevalence in diverse feedlot systems.

Glossary

Abbreviations

DMI

dry matter intake

GPE

germplasm evaluation

USMARC

U.S. Meat Animal Research Center

Supplementary data

Supplementary data is available at Translational Animal Science online.

Funding

This project was supported by the National Institute of Food and Agriculture Agriculture and Food Research Initiative competitive grant award no. 2023-67015-39781 from the U.S. Department of Agriculture.

Author contributions

None declared.

CRediT statement

Heather L. Bradford (Formal analysis, Methodology, Software, Validation, Visualization, Writing—original draft, Writing—reviewing & editing), Larry A. Kuehn (Conceptualization, Funding acquisition, Methodology, Resources, Writing—reviewing & editing), Brittney N. Keel (Conceptualization, Funding acquisition, Writing—reviewing & editing), Bryan W. Neville (Writing—reviewing & editing), Amanda K. Lindholm-Perry (Conceptualization, Funding acquisition, Project administration, Writing—reviewing & editing).

Conflict of interest

The authors declare no conflict of interest.