Predicting Feed Intake of the Individual Dairy Cow

What the hay: predicting equine voluntary forage intake using a meta-analysis approach

To properly formulate diets, the ability to accurately estimate feed intake is critical as the amount of feed consumed will influence the amount of nutrients delivered to the animal. Inaccurate intake estimates may lead to under- or over-feeding of nutrients to the animal. Individual differences in equine forage intake are well-known, but predictive equations based on animal and nutritional factors are not comprehensive. The objective of the present study was to consolidate the current body of knowledge in the published literature on voluntary forage DM intake (VFDMI) in equines and conduct a meta-analysis to identify driving factors, sources of heterogeneity, and develop predictive equations. Therefore, a systematic literature search was applied and identified 61 publications which met the inclusion criteria. From each study, the outcomes of interest (e.g., forage intake), diet composition (e.g., forage information, nutrient composition), and animal factors (e.g., sex, age, breed, BW, exercise level) were extracted. Forage intake was analyzed as two different outcome variables: (1) VFDMI in kg/d and (2) VFDMI in g/kg BW. Linear mixed model analysis treating study as a random effect was applied, using a backward-stepping approach to identifying potential driving variables for VFDMI (both units) where all terms have P < 0.1. The best fitting models for VFDMI included similar factors (also across kg/d and g/kg BW) such as forage quality (i.e., neutral detergent fiber or CP content), forage type (i.e., grass, legume, or mixed), the animals' size category (i.e., horses vs ponies), and some management factors (i.e., pasture access). As anticipated, forage intake increased when higher quality forages were fed (i.e., lower neutral detergent fiber or higher CP), potentially due to improved digestibility. Additionally, VFDMI increased as BW increased but ponies increased their VFDMI more per every kg increase in BW compared to horses. Lastly, pasture access (i.e., grazing) may influence VFDMI such that pastured animals consume less than stalled animals, possibly due to the time it takes to graze forage. In conclusion, equations to predict equine VFDMI with high accuracy and precision (concordance correlation coefficient  = 0.82 - 0.95; root mean squared error RMSE = 0.82-5.49) were developed which could be applied in practice by equine nutritionists or owners and managers. The results of this meta-analysis confirm that animal traits and forage quality have a significant impact on the VFDMI of equines and should be accounted for when formulating diets to ensure nutritional requirements are met.

Feed Conversion Ratio (FCR) and Performance Group Estimation Based on Predicted Feed Intake for the Optimisation of Beef Production

This paper reports on the use of estimates of individual animal feed intake (made using time spent feeding measurements) to predict the Feed Conversion Ratio (FCR), a measure of the amount of feed consumed to produce 1 kg of body mass, for an individual animal. Reported research to date has evaluated the ability of statistical methods to predict daily feed intake based on measurements of time spent feeding measured using electronic feeding systems. The study collated data of the time spent eating for 80 beef animals over a 56-day period as the basis for the prediction of feed intake. A Support Vector Regression (SVR) model was trained to predict feed intake and the performance of the approach was quantified. Here, feed intake predictions are used to estimate individual FCR and use this information to categorise animals into three groups based on the estimated Feed Conversion Ratio value. Results provide evidence of the feasibility of utilising the 'time spent eating' data to estimate feed intake and in turn Feed Conversion Ratio (FCR), the latter providing insights that guide farmer decisions on the optimisation of production costs.

Modern livestock farming under tropical conditions using sensors in grazing systems

The aim of this study was to evaluate a commercial sensor—a three-axis accelerometer—to predict animal behavior with a variety of conditions in tropical grazing systems. The sensor was positioned on the underjaw of young bulls to detect the animals’ movements. A total of 22 animals were monitored in a grazing system, during both seasons (wet and dry), with different quality and quantity forage allowance. The machine learning (ML) methods used were random forest (RF), convolutional neural net and linear discriminant analysis; the metrics used to determine the best method were accuracy, Kappa coefficient, and a confusion matrix. After predicting animal behavior using the best ML method, a forecast for animal performance was developed using a mechanistic model: multiple linear regression to correlate intermediate average daily gain (iADG) observed versus iADG predicted. The best ML method yielded accuracy of 0.821 and Kappa coefficient of 0.704, was RF. From the forecast for animal performance, the Pearson correlation was 0.795 and the mean square error was 0.062. Hence, the commercial Ovi-bovi sensor, which is a three-axis accelerometer, can act as a powerful tool for predicting animal behavior in beef cattle production developed under a variety tropical grazing condition.

Are subjectively scored linear type traits suitable predictors of the genetic merit for feed intake in grazing Holstein-Friesian dairy cows?

Measuring dry matter intake (DMI) in grazing dairy cows using currently available techniques is invasive, time consuming, and expensive. An alternative to directly measuring DMI for use in genetic evaluations is to identify a set of readily available animal features that can be used in a multitrait genetic evaluation for DMI. The objectives of the present study were thus to estimate the genetic correlations between readily available body-related linear type traits and DMI in grazing lactating Holstein-Friesian cows, but importantly also estimate the partial genetic correlations between these linear traits and DMI, after adjusting for differences in genetic merit for body weight. Also of interest was whether the predictive ability derived from the estimated genetic correlations materialized upon validation. After edits, a total of 8,055 test-day records of DMI, body weight, and milk yield from 1,331 Holstein-Friesian cows were available, as were chest width, body depth, and stature from 47,141 first lactation Holstein-Friesian cows. In addition to considering the routinely recorded linear type traits individually, novel composite traits were defined as the product of the linear type traits as an approximation of rumen volume. All linear type traits were moderately heritable, with heritability estimates ranging from 0.27 (standard error = 0.14) to 0.49 (standard error = 0.15); furthermore, all linear type traits were genetically correlated (0.29 to 0.63, standard error 0.14 to 0.12) with DMI. The genetic correlations between the individual linear type traits and DMI, when adjusted for genetic differences in body weight, varied from -0.51 (stature) to 0.48 (chest width). These genetic correlations between DMI and linear type traits suggest linear type traits may be useful predictors of DMI, even when body weight information is available. Nonetheless, estimated genetic merit of DMI derived from a multitrait genetic evaluation of linear type traits did not correlate strongly with actual DMI in a set of validation animals; the benefit was even less if body weight data were also available.

Design a system for measuring individual cow feed intake in commercial dairies

Monitoring individual cow feed intake is necessary for calculating the cow individual feed efficiency. The cost and maintenance time necessary for research systems make them impractical for most of the commercial producers. We developed a measurement system with producer convenience and low investment as key design criteria. The goal of this study was to design the system and validate its ability to rank cows by their feed conversion efficiency in commercial farms. The new system consisted of three principal parts: (a) a hanging weighing system, (b) a visual cow identification system and (c) an automatic cleaning system. The weighing system consisted of hanging a single load cell to provide feed mass measurements. The image-based cow identification system (replacing Radio-Frequency Identification) entailed cameras installed above the feeding area and an image processing algorithm that recognized cows by their collar numbers. The new system worked within normal farm routines: the feed supplying truck distributed the animal feed, and a tractor cleaned feed residual. To validate the accuracy and convenience of the system and to rank the cows by their efficiency, an experiment with six scales and 12 cows was conducted in a research barn, succeeded by eight-scale system in a commercial farm with 16 cows. The feed intake of each cow participating in the experiments was monitored for one month. The validation experiment showed that the system had the following specification: scales were accurate within 120 g; the visual cow identification rate was greater than 96%; feeding duration was accurate to 52 s; and routine farm practices (feed distribution, pushing, and residual removal) continued as usual. The cost for a feeding station (utilized consequently for a number of cows) was about 1 500 USD. An example of application of the system to rank cows by their efficiency under commercial conditions was shown. The system can potentially be used for ranking cows by their efficiency in commercial facilities.

Individual dairy cow management: achievements, obstacles and prospects

This review deals with the prospects and achievements of individual dairy cow management (IDCM) and the obstacles and difficulties encountered in attempts to successfully apply IDCM into routine dairy management. All aspects of dairy farm management, health, reproduction, nutrition and welfare are discussed in relation to IDCM. In addition, new IDCM R&D goals in these management fields are suggested, with practical steps to achieve them. The development of management technologies is spurred by the availability of off-the-shelf sensors and expanded recording capacity, data storage, and computing capabilities, as well as by demands for sustainable dairy production and improved animal wellbeing at a time of increasing herd size and milk production per cow. Management technologies are sought that would enable the full expression of genetic and physiological potential of each cow in the herd, to achieve the dairy operation's economic goals whilst optimizing the animal's wellbeing. Results and conclusions from the literature, as well as practical experience supported by published and unpublished data are analyzed and discussed. The object of these efforts is to identify knowledge and management routine gaps in the practical dairy operation, in order to point out directions and improvements for successful implementation of IDCM in the dairy cows' health, reproduction, nutrition and wellbeing.

Advancements in sensor technology and decision support intelligent tools to assist smart livestock farming

Remote monitoring, modern data collection through sensors, rapid data transfer, and vast data storage through the Internet of Things (IoT) have advanced precision livestock farming (PLF) in the last 20 yr. PLF is relevant to many fields of livestock production, including aerial- and satellite-based measurement of pasture’s forage quantity and quality; body weight and composition and physiological assessments; on-animal devices to monitor location, activity, and behaviors in grazing and foraging environments; early detection of lameness and other diseases; milk yield and composition; reproductive measurements and calving diseases; and feed intake and greenhouse gas emissions, to name just a few. There are many possibilities to improve animal production through PLF, but the combination of PLF and computer modeling is necessary to facilitate on-farm applicability. Concept- or knowledge-driven (mechanistic) models are established on scientific knowledge, and they are based on the conceptualization of hypotheses about variable interrelationships. Artificial intelligence (AI), on the other hand, is a data-driven approach that can manipulate and represent the big data accumulated by sensors and IoT. Still, it cannot explicitly explain the underlying assumptions of the intrinsic relationships in the data core because it lacks the wisdom that confers understanding and principles. The lack of wisdom in AI is because everything revolves around numbers. The associations among the numbers are obtained through the “automatized” learning process of mathematical correlations and covariances, not through “human causation” and abstract conceptualization of physiological or production principles. AI starts with comparative analogies to establish concepts and provides memory for future comparisons. Then, the learning process evolves from seeking wisdom through the systematic use of reasoning. AI is a relatively novel concept in many science fields. It may well be “the missing link” to expedite the transition of the traditional maximizing output mentality to a more mindful purpose of optimizing production efficiency while alleviating resource allocation for production. The integration between concept- and data-driven modeling through parallel hybridization of mechanistic and AI models will yield a hybrid intelligent mechanistic model that, along with data collection through PLF, is paramount to transcend the current status of livestock production in achieving sustainability.

Short communication: Prediction of hyperketonemia in dairy cows in early lactation using on-farm cow data and net energy intake by partial least square discriminant analysis

The objectives of this study were (1) to evaluate if hyperketonemia in dairy cows (defined as plasma β-hydroxybutyrate ≥1.0 mmol/L) can be predicted using on-farm cow data either in current or previous lactation week, and (2) to study if adding individual net energy intake (NEI) can improve the predictive ability of the model. Plasma β-hydroxybutyrate concentration, on-farm cow data (milk yield, percentage of fat, protein and lactose, fat- and protein-corrected milk yield, body weight, body weight change, dry period length, parity, and somatic cell count), and NEI of 424 individual cows were available weekly through lactation wk 1 to 5 postpartum. To predict hyperketonemia in dairy cows, models were first trained by partial least square discriminant analysis, using on-farm cow data in the same or previous lactation week. Second, NEI was included in models to evaluate the improvement of the predictability of the models. Through leave-one trial-out cross-validation, models were evaluated by accuracy (the ratio of the sum of true positive and true negative), sensitivity (68.2% to 84.9%), specificity (61.5% to 98.7%), positive predictive value (57.7% to 98.7%), and negative predictive value (66.2% to 86.1%) to predict hyperketonemia of dairy cows. Through lactation wk 1 to 5, the accuracy to predict hyperketonemia using data in the same week was 64.4% to 85.5% (on-farm cow data only), 66.1% to 87.0% (model including NEI), and using data in the previous week was 58.5% to 82.0% (on-farm cow data only), 59.7% to 85.1% (model including NEI). An improvement of the accuracy of the model due to including NEI ranged among lactation weeks from 1.0% to 4.4% when using data in the same lactation week and 0.2% to 6.6% when using data in the previous lactation week. In conclusion, trained models via partial least square discriminant analysis have potential to predict hyperketonemia in dairy cows not only using data in the current lactation week, but also using data in the previous lactation week. Net energy intake can improve the accuracy of the model, but only to a limited extent. Besides NEI, body weight, body weight change, milk fat, and protein content were important variables to predict hyperketonemia, but their rank of importance differed across lactation weeks.

Review: Synergy between mechanistic modelling and data-driven models for modern animal production systems in the era of big data

Mechanistic models (MMs) have served as causal pathway analysis and 'decision-support' tools within animal production systems for decades. Such models quantitatively define how a biological system works based on causal relationships and use that cumulative biological knowledge to generate predictions and recommendations (in practice) and generate/evaluate hypotheses (in research). Their limitations revolve around obtaining sufficiently accurate inputs, user training and accuracy/precision of predictions on-farm. The new wave in digitalization technologies may negate some of these challenges. New data-driven (DD) modelling methods such as machine learning (ML) and deep learning (DL) examine patterns in data to produce accurate predictions (forecasting, classification of animals, etc.). The deluge of sensor data and new self-learning modelling techniques may address some of the limitations of traditional MM approaches - access to input data (e.g. sensors) and on-farm calibration. However, most of these new methods lack transparency in the reasoning behind predictions, in contrast to MM that have historically been used to translate knowledge into wisdom. The objective of this paper is to propose means to hybridize these two seemingly divergent methodologies to advance the models we use in animal production systems and support movement towards truly knowledge-based precision agriculture. In order to identify potential niches for models in animal production of the future, a cross-species (dairy, swine and poultry) examination of the current state of the art in MM and new DD methodologies (ML, DL analytics) is undertaken. We hypothesize that there are several ways via which synergy may be achieved to advance both our predictive capabilities and system understanding, being: (1) building and utilizing data streams (e.g. intake, rumination behaviour, rumen sensors, activity sensors, environmental sensors, cameras and near IR) to apply MM in real-time and/or with new resolution and capabilities; (2) hybridization of MM and DD approaches where, for example, a ML framework is augmented by MM-generated parameters or predicted outcomes and (3) hybridization of the MM and DD approaches, where biological bounds are placed on parameters within a MM framework, and the DD system parameterizes the MM for individual animals, farms or other such clusters of data. As animal systems modellers, we should expand our toolbox to explore new DD approaches and big data to find opportunities to increase understanding of biological systems, find new patterns in data and move the field towards intelligent, knowledge-based precision agriculture systems.

A decision-tree model to detect post-calving diseases based on rumination, activity, milk yield, BW and voluntary visits to the milking robot

Early detection of post-calving health problems is critical for dairy operations. Separating sick cows from the herd is important, especially in robotic-milking dairy farms, where searching for a sick cow can disturb the other cows' routine. The objectives of this study were to develop and apply a behaviour- and performance-based health-detection model to post-calving cows in a robotic-milking dairy farm, with the aim of detecting sick cows based on available commercial sensors. The study was conducted in an Israeli robotic-milking dairy farm with 250 Israeli-Holstein cows. All cows were equipped with rumination- and neck-activity sensors. Milk yield, visits to the milking robot and BW were recorded in the milking robot. A decision-tree model was developed on a calibration data set (historical data of the 10 months before the study) and was validated on the new data set. The decision model generated a probability of being sick for each cow. The model was applied once a week just before the veterinarian performed the weekly routine post-calving health check. The veterinarian's diagnosis served as a binary reference for the model (healthy-sick). The overall accuracy of the model was 78%, with a specificity of 87% and a sensitivity of 69%, suggesting its practical value.

Lameness Affects Cow Feeding But Not Rumination Behavior as Characterized from Sensor Data

Using automatic sensor data, this is the first study to characterize individual cow feeding and rumination behavior simultaneously as affected by lameness. A group of mixed-parity, lactating Holstein cows were loose-housed with free access to 24 cubicles and 12 automatic feed stations. Cows were milked three times/day. Fresh feed was delivered once daily. During 24 days with effectively 22 days of data, 13,908 feed station visits and 7,697 rumination events obtained from neck-mounted accelerometers on 16 cows were analyzed. During the same period, cows were locomotion scored on four occasions and categorized as lame (n = 9) or not lame (n = 7) throughout the study. Rumination time, number of rumination events, feeding time, feeding frequency, feeding rate, feed intake, and milk yield were calculated per day, and coefficients of variation were used to estimate variation between and within cows. Based on daily sums, using each characteristic as response, the effects of lameness and stage of lactation were tested in a mixed model. With rumination time as response, each of the four feeding characteristics, milk yield, and lameness were tested in a second mixed model. On a visit basis, effects of feeding duration, lameness, and milk yield on feed intake were tested in a third mixed model. Overall, intra-individual variation was <15% and inter-individual variation was up to 50%. Lameness introduced more inter-individual variation in feeding characteristics (26–50%) compared to non-lame cows (17–29%). Lameness decreased daily feeding time and daily feeding frequency, but increased daily feeding rate. Interestingly, lameness did not affect daily rumination behaviors, fresh matter intake, or milk yield. On a visit basis, a high feeding rate was associated with a higher feed intake, a relationship that was exacerbated in the lame cows. In conclusion, cows can be characterized in particular by their feeding behavior, and lame cows differ from their non-lame pen-mates in terms of fewer feed station visits, faster eating, less time spent feeding, and more variable feeding behavior. Further, daily rumination time was slightly negatively associated with feeding rate, a relationship which calls for more research to quantify rumination efficiency relative to feeding rate.

Effect of breeding for milk yield, diet and management on enteric methane emissions from dairy cows

Enteric methane production from livestock is an important source of anthropogenic greenhouse gas emissions. The aim of the present study was to (1) assess the effect of long-term breeding for kilograms of milk fat plus protein production and (2) investigate the influence of parity, genetic line and diet on predicted enteric methane emissions of Holstein Friesian dairy cows. Analyses were based on 17 years of experimental data for lactating and dry cows, housed and at pasture. Restricted maximum likelihood (REML) was used to assess the effects of parity, genetic line and diet on the predicted enteric methane output of lactating and dry cows. A non-linear equation based on metabolisable energy intake (MEI) was used to predict daily enteric methane output. The present study found that selection for kilograms of milk fat plus protein production, zero-grazing low-forage diets and maintaining persistently high-yielding older cows can reduce a cow’s enteric methane emissions per kilogram milk by up to 12%, on average. Comparing the first 5 years to the most recent 5 years of the study period showed that large savings of 19% and 23% in enteric methane per kilogram milk were made in cows selected for milk fat plus protein or selected to remain close to the average genetic merit for milk fat plus protein production for all animals evaluated in the UK, respectively. Additionally, management to minimise the length of the drying-off period can help reduce enteric methane emissions during a cow’s lactation period.

Feeding soyhulls to high-yielding dairy cows increased milk production, but not milking frequency, in an automatic milking system

To attract a cow into an automatic milking system (AMS), a certain amount of concentrate pellets is provided while the cow is being milked. If the milking frequency in an AMS is increased, the intake of concentrate pellets might increase accordingly. Replacing conventional starchy pellets with nonstarchy pellets increased milk yield, milk fat, and milk protein and decreased body weight. The hypothesis was that a nonroughage by-product rich in digestible neutral detergent fiber, such as soyhulls and gluten feed, could replace starchy grain in pellets fed in an AMS. Sixty cows were paired by age, milk yield, and days in milk, and were fed a basic mixture ad libitum [16.2 +/- 0.35 (mean +/- SE) kg of dry matter intake/d per cow] plus a pelleted additive (6 to 14 kg of dry matter/d per cow) that was consumed in the AMS and in a concentrate self-feeder, which could only be entered after passing through the AMS. The 2 feeding regimens differed only in the composition of the pelleted additives: the control group contained 52.9% starchy grain, whereas the experimental group contained 25% starchy grain, plus soyhulls and gluten feed as replacement for part of the grain. Wheat bran in the control ration, a source of fiber with low digestibility, was replaced with more digestible soyhulls and gluten. During the first 60 d in milk, a cow received 10 to 12 kg of concentrate pellets. After 60 DIM, concentrate feed was allocated by milk production: < or =25 kg/d of milk entitled a cow to 2 kg/d of concentrate feed; >25 kg/d of milk entitled a cow to receive 1 kg/d of additional concentrate feed per 5 kg/d of additional milk production, and >60 kg/d of milk entitled a cow to receive 9 kg of concentrate. The concentrate feed was split between the AMS and concentrate self-feeder. The 2 diets resulted in similar frequencies of voluntary milking (3.12 +/- 0.03 to 2.65 +/- 0.03 visits/d per cow vs. 3.16 +/- 0.00 to 2.60 +/- 0.01 visits/d per cow). Average milk yields were higher in the experimental group (42.7 +/- 0.76 to 39.09 +/- 0.33 kg/d per cow vs. 39.69 +/- 0.68 to 37.54 +/- 0.40 kg/d per cow) and percentages of milk protein (3.02 +/- 0.06 to 3.12 +/- 0.05% vs. 3.07 +/- 0.04 to 3.20 +/- 0.04%) and milk fat (3.42 +/- 0.17 to 3.44 +/- 0.08% vs. 3.38 +/- 0.13 to 3.55 +/- 0.06%) were similar in the 2 groups. The results suggest that the proposed pellets high in digestible neutral detergent fiber can be allocated via the AMS to selected high-yielding cows without a negative effect on appetite, milk yield, or milk composition while maintaining a high milking frequency.

Comparing two concentrate allowances in an automatic milking system

This study investigated the potential for applying an automatic milking system (AMS) to the management of high-yielding cows offered a total mixed ration (TMR). The null hypothesis was that it is desirable to maintain even in AMS, the TMR feeding management practice recommended for high-yielding cows and therefore it can be attained by ‘reducing the concentrate allocation in the robot without reducing the number of milkings’. Two feeding regimes were used: the ‘candy concept’, with only 1·2 kg of food concentrate – the minimum to attract the cow – provided at each visit to the milking robot; and the provision of a maximum of 7 kg of food concentrate per day. Approximately 100 cows were subjected to one or other of these two treatments. Although the cows in the first treatment consumed approximately 3·5 kg of concentrate per day and those in the second treatment approximately 5 kg per day, no significant differences were observed in the numbers of voluntary milkings.

Feeding of Pellets Rich in Digestible Neutral Detergent Fiber to Lactating Cows in an Automatic Milking System

If the milking frequency in an automatic milking system (AMS) is increased, the intake of concentrated pellets in the robot may be raised accordingly. Consumption of a large quantity of starchy grains within a short time can impair the appetite, decrease voluntary visits to the milking stall, and lower intakes of dry matter (DM) and neutral detergent fiber (NDF). Therefore, the hypothesis to be tested in this study was whether conventional starchy pellets fed in the AMS could be replaced with pellets rich in digestible NDF without impairing the cows' motivation to visit a milking stall voluntarily. Fifty-four cows were paired according to age, milk yield, and days in milk, and were fed a basic mixture along the feeding lane (19.9 kg of DM/cow per d), plus a pelleted additive (approximately 5.4 kg of DM/cow per d) that they obtained in the milking stall and in the concentrate self-feeder that they could enter only after passing through the milking stall. The 2 feeding regimens differed only in the composition of the pelleted additive, which, for the control group, contained 49% starchy grain, and for the experimental group contained 25% starchy grain plus soy hulls and gluten feed as replacement for part of the grain and other low-digestible, NDF-rich feeds. Both diets resulted in similar rates of voluntary milkings (3.31 vs. 3.39 visits/cow per d). Average yields of milk and percentages of milk protein were also similar in the 2 groups. The results suggest that an alternative pellet composition can be allocated in the AMS in conjunction with basic mixture in the feeding lane, without any negative effect on appetite, milk yield, milk composition, or milking frequency of the cows. It also opens the opportunity to increase yields of milk and milk solids by increasing the amount of pelleted concentrates that can be allocated to selected high-yielding cows via the AMS, because this can be done while maintaining a high frequency of voluntary milkings.

Effects of replacing roughage with soy hulls on feeding behavior and milk production of dairy cows under hot weather conditions

Two total mixed rations (TMR) containing different proportions of roughage neutral detergent fiber (NDF) were fed to lactating cows under Israeli summer conditions, and the effects on feeding behavior and milk production were measured. Forty-two lactating cows were divided into 2 groups fed ad libitum a TMR containing either 18% NDF of roughage origin (control group) or only 12% roughage NDF, in which the corn silage component (16.5% of dry matter [DM]) was replaced with soy hulls (experiment group). This and additional adjustments in TMR were reflected in higher net energy for lactation and in vitro digestibility of the experimental TMR. Cow behavior was investigated at the feeding lane during June 2001; about 11,000 cow visits were analyzed. Feed intake per meal and average meal duration were significantly higher in the experiment group (1.51 kg of DM per meal and 12.1 min per meal, respectively) as compared with the control group (1.22 kg of DM per meal and 9.47 min per meal, respectively). However, number of meals per day recorded in the feeding lane was significantly higher in the control group (21.0 vs. 16.6 meals/d per cow). Distribution of meals and feed intake along the day depended more on management practices, such as milking and feed dispensing times, than on feed composition or weather conditions. These differences between groups were expressed in similar daily eating duration (approximately 200 min), and because the rate of feed consumption was similar for both treatments (approximately 127 g DM/min), the daily average DM intake was also similar (25.0 to 25.7 kg). However, NDF intake was higher in the experiment group. Consequently, the average milk yield was higher in the experimental group, and production of milk fat, 4% fat-corrected milk, and economically corrected milk were significantly higher in the experiment group than in the control group. Data imply that the experimental TMR containing only 12% NDF of roughage origin is more suitable for cows under hot climate conditions compared with the control TMR.