Do dairy cows have preferences for different concentrate feeds? Har malkekøer præferencer for forskellige kraftfodermidler?

Master Thesis (45 ECTS) by Lasse Primdal 20085402 Supervisor: Senior Scientist Martin Riis Weisbjerg Department of Animal Science, Faculty of Science and Technology Research Center Foulum, Aarhus University.

December 2013

Preface and acknowledgments This master thesis was made as the final project on the master program Agrobiology – Animal health and welfare at the faculty of Science and Technology, Aarhus University. The thesis has an equivalent of 45 ECTS. The project included a literature review and an experimental part. In the experimental work a part of the experimental planning and the conducting of the experimental work were done by me as a part of the thesis. The experimental work and process of writing were carried out at the Institute of Animal Science, section of Animal Nutrition and Environmental Impact at the Research Center Foulum, Aarhus University. The thesis was done in the period from April 2013 to December 2013. The purpose of this thesis was to study whether dairy cows have preferences for different concentrate feeds. The results addresses everyone with interest in knowledge about dairy cows preferences for different concentrate feeds. The popular scientific article included in this thesis addresses readers with a practical and not necessarily scientific approach to the topic. The experiment included in this thesis was done as a part of a larger project granted under the name “Funktionel kraftfoder”, which was already granted at the beginning of this thesis. After the end of this thesis the large project continues and may use some of the results found in this thesis. I would like to express my gratitude to all the people at the Research Center Foulum who have helped me during this thesis. In special I would like to thanks the technicians Torkild Nyholm Jakobsen, who has been very helpful during the practical parts of the experimental work and also did a part of the preference trials, and Ole Hartvig Olsen, who has been very helpful handling and reading the results into the statistical program SAS. Also thanks to my supervisor Senior Scientist Martin Riis Weisbjerg for supervising me in all parts of the project. My sincere gratitude goes to all my friends and family, who has supported me through the whole thesis. A special thanks to Susanne Frydendal Nielsen and Uffe Krogh Larsen for critical reading and giving comments on the written part of the thesis, it has helped me a lot. Finally I would like to thanks my wonderful girlfriend Marianne Johansen for moral and professional support in all parts of this thesis.

University of Aarhus Faculty of science and technology Department of Animal Science Foulum, December 2013

_____________________________ Lasse Primdal

1

Summary Automatic milking systems (AMS) are based on the concept that cows voluntarily enter the milking robot to be milked. For an optimal use of the robot a milking frequency of 2.5-3.5 milkings per cow per day is desired. The use of concentrate feeds as rewards in the milking robot is common used in AMS, but to achieve the desired milking frequency the use of highly preferred concentrates are needed. Therefore it was relevant to study whether dairy cows had preferences for different concentrate feeds. Previous studies have found positive effect on feed preferences by addition of molasses as a flavoring agent and found pelletized forms of concentrates to be preferred above fine forms. Therefore this study evaluated the effect physical form of concentrates and the addition of sugar beet molasses on the feed preferences. To study the preferences for different concentrate feeds a series of preference trials were done in a cafeteria setup. It was possible to test the preferences of four different concentrates at a time in one trial. To compare between trials the same control concentrate were used in all trials. Therefore nine separate preference trials were needed to study 26 different (and 2 replicates) concentrate feeds. Four cows were used in each preference trial designed as a 4 x 4 Latin square. Each trial were consisted of four registration days, and as it was not possible to do registrations on all the cows at the same time each cow fasted one, two, three or four hours per day. Within the four registration days each cow experienced all times of fasting. Placement of the different concentrates in the cafeteria was confounded with fasting hours, in that way the cows were presented to the four different concentrates at all four possible places in the cafeteria within a full trial. The cafeteria was presented to the cows for ten minutes. Within each registration day three replications were done on each cow. Feed leftovers after the ten registration minutes (leftovers) and the time after offering the cafeteria a box containing concentrate were considered empty (time) were the responses to be recorded. Behavioral registrations regarding the behaviors eat and sniff were also done. Preference ranking of the concentrates based on leftovers and time resulted in similar ranking orders between the concentrates. The behavioral registrations supported the ranking done by the responses leftovers and time. The results showed that the cows were able to discriminate between the offered concentrates and that they preferred some concentrates to others. Only one concentrate (soybean meal) was preferred higher than the control (P < 0.05). Few concentrates were preferred at the same level as the control. Most of the tested concentrates were preferred at a lower level than the control (P < 0.05).

A tendency that pelletized barley was preferred to rolled barley was found, but as the highest preferred concentrates included the forms of meal, pellets and rolled no clear evidence of relation between physical form of the concentrates and feed preferences was shown. The content of the highest preferred concentrates were different, thus it was difficult to make direct comparisons of the physical form of these concentrates. The addition of 25 % sugar beet molasses as a flavoring agent raised the preferences of four tested concentrates (wheat, dried grass pellets, peas and dried sugar beet pulp) as they all with the addition of sugar beet molasses were preferred higher than the pure form regarding both leftovers and time. Pure sugar beet molasses was preferred low in the experiment. In conclusion dairy cows have preferences for different concentrate feeds, and the preferences were affected by the addition of sugar beet molasses as a flavoring agent. The effects of physical form were not possible to show by this experiment.

2

Resumé Automatiske malkesystemer (AMS) er baseret på at køerne frivilligt opsøger malkerobotten for at blive malket. For at udnytte den fulde kapacitet på malkerobotten er en malkefrekvens på 2,5 - 3,5 malkninger pr. ko pr. dag ønskeligt. Brugen af kraftfoder som lokkemiddel i malkerobotten er almindelig praksis i AMS, hvorved der er behov for kraftfodermidler som køerne har høje præferencer for. Derfor var det relevant at undersøge om malkekøer havde præferencer for forskellige kraftfodermidler. Tidligere studier har vist positiv effekt på foderpræferencer ved tilsætning af melasse til et fodermiddel, samt fundet at køer har højere præferencer for pelleterede former af kraftfodermidler, sammenlignet med formalet former. Derfor indgik kraftfodermidler med forskellige fysiske former, samt tilsætning af sukkerroemelasse i denne undersøgelse. Der blev udført 9 præferenceforsøg i et cafeteria setup for at undersøge præferencerne for 26 forskellige kraftfodermidler (og 2 gentagelser). For at sammenligne mellem forsøgene blev det samme kontrol kraftfoder brugt i alle forsøgene. Fire køer blev brugt i hvert præferenceforsøg. Hvert præferenceforsøg blev udført over fire dage, og da det ikke var muligt at registrere på alle køer på en gang inkluderede forsøget også fire niveauer af faste (1-4 timer) per dag. Forsøget blev udført som et 4x4 romerkvadrat. Efter et præferenceforsøgs fire registreringsdage havde alle 4 køer således fastet alle mulige timer, samt fået præsenterede fodermidlerne på alle mulige pladser i cafeteriaet. Cafeteriaet var tilbudt til hver ko i 10 minutter, hvilket blev gentaget 3 gange. Foderrest efter 10 minutter (rester) og tiden fra tildelingen til det blev vurderet at kassen med et fodermidlet var tom (tid) blev registeret som responser. Adfærdsregistreringer på hvor lang tid koen brugte på at æde og snuse til hvert enkelt fodermiddel blev også foretaget. En præferencerangering af fodermidlerne baseret på enten rester eller tid viste den samme rangering. Adfærdsregistreringerne understøttede rangeringen. Resultaterne viste at køerne havde præferencer for de forskellige kraftfodermidler. Køerne havde kun signifikant højere præference for et af de testede kraftfodermiddel (sojaskrå) sammenlignet med kontrollen (P < 0,05). Køerne havde samme præferencer for få af de testede kraftfodermidler sammenlignet med kontrollen. Præferencerne var lavere for de fleste af de testede kraftfodermidler sammenlignet med kontrollen (P < 0,05). Der blev fundet en tendens til køernes præferencer for pelleteret byg var højere sammenlignet med valset byg. De fodermidler køerne havde højest præferencer for inkluderede piller, skrå og valset, hvorved en direkte betydning af fysisk form af kraftfodermidlet for køernes præferencer ikke kunne vises. Da indholdet i de kraftfodermidler køerne havde højest præferencer for var forskelligt var det dog vanskeligt at sammenligne dem direkte. Tilsætningen af 25 % sukkerroemelasse øgede køernes præferencer for de kraftfodermidler det blev testet på (hvede, grøntpiller, ærter og roepiller) både ved rangering ud fra rester og tid. Sukkerroemelasse i ren form blev rangeret forholds vis lavt i forhold til begge responser. Ud fra forsøget kan det konkluderes at malkekøer har præferencer for forskellige kraftfodermidler og at deres præferencer kunne påvirkes ved tilsætningen af sukkerroemelasse. En direkte effekt af fysiskform af kraftfodermidlet kunne ikke vises.

3

Content 1

2

Introduction................................................................................................................................. 5 1.1

Aim and project questions ..................................................................................................... 6

1.2

Method and objectives ........................................................................................................... 7

Literature review ........................................................................................................................ 8 2.1

Definitions of terms ............................................................................................................... 8

2.2

Eating behavior of cows ........................................................................................................ 9

2.3

Factors affecting feed preferences ...................................................................................... 11

2.4

Methods for measuring feed preferences and palatability .................................................. 14

2.5

Effect of voluntary visits in the milking robot on cow performance ................................... 17

3

Scientific paper.......................................................................................................................... 23

4

Popular scientific paper ........................................................................................................... 57

5

Conclusions................................................................................................................................ 63

6

Perspectives ............................................................................................................................... 64

7

References .................................................................................................................................. 65

4

1 Introduction The use of small amounts of highly palatable concentrate feeds to guide cows in a specific direction has always been used in different management situations. The introduction of automatic milking systems (AMS) have resulted in an increasing interest in the use of highly palatable concentrates to guide cows into the milking robot (Spörndly and Åsberg, 2006). AMS is a system that is fully automated by a robot unit which handles all the steps in the milking process of a cow. In a traditional milking parlor the work of milking is carried out by a mix of manual and machinery work, like the number of milking per day (milking frequency) is also controlled by humans, by bringing the cows to the milking parlor on a regularly basis, traditionally twice a day. In contrast the concept of AMS is based on a self-service approach and that the cows voluntarily enters the milking robot to be milked on a regularly basis. As the need of the cow to be milked is not always enough for the cow to visit the milking robot the use of concentrates in the milking robot is common practice to motivate the cows to visit the milking robot (Koning and Rodenburg, 2004; Weisbjerg and Munksgaard, 2008). In AMS the milking frequency is desirable between 2.5-3.5 milkings per cow per day. A milking frequency below 2 will decrease the yield, whereas a milking frequency higher than 3.5 will lower the milking capacity of the robot (Weisbjerg and Munksgaard, 2008). On farm level it is not unusual than about 10 % of the cows in an AMS herd has a milking frequency of less than two. Such cows will typically be fetched, and brought into the milking robot. The time of fetching cows in relation to milking frequency varies in relation to farm management, but the typically will be to fetch cows if the time since last milking exceed 12 hours (Koning and Rodenburg, 2004). In the last 15 years the use of AMS has increased as an alternative to traditional milking parlors. Especially in the Western Europe and in the Scandinavian countries milking robots has competed well with traditional milking systems. The main reasons are the modest herd sizes, which fits well for the concept of AMS, and the high cost of labor in these areas (Rodenburg, 2011). Under Danish conditions the distribution of milking robots has also increased since the late 1990ties, where the first milking robots came on the market, until today where the share of milking herds using AMS are about 25 % (Søgaard, 2012). Erdman and Varner (1995) calculated a positive effect on milk yield of 3.5 kg/d by increasing the milking frequency from two to three times per day, which could be another motivation for using AMS (Konning and Rodenburg, 2004). Despite of the reported potential for decreased need of labor to milking and increased milk production in AMS herds some producers in practice experience an increased need for labor to fetch cows that not voluntarily turn into the milking robot and a low milk yield because of an irregular and low milking frequency (Bach and Busto, 2005; Bach et al., 2007). However, it is expected that on cow level milking frequency during the lactation is to be more various in AMS compared to a traditional milking parlor (Bach and Busto, 2005). In Denmark some producers give up AMS due to the above mentioned problems, which were shown in particular in the year of 2011 where the number of unin-

5

Introduction

stalled milking robots was greater than the number of new robots installed. Many of the uninstalled milking robots were replaced with traditional milking parlors to overcome the mentioned problems (Worsøe, 2011). To use the fully positive potential in AMS it is important to make sure that the cows visit the milking robot on a regular basis. To assure sufficient visits on a regular basis more concepts to control the cows are possible. Most of the different concepts involve concentrate feeding in the milking robot (Spörndly and Åsberg 2006). The use of a high amount of concentrate in the robot within a short period of time increase the risk of metabolic diseases and will have a negative effect on feed costs (Halachmi et al., 2005). Therefore dairy cows’ preferences for different concentrate feeds, which could be used in small amounts as concentrates in the milking robot, are important to study. Other previous studies (Klopfer et al., 1981; Spörndly and Åsberg, 2006; Madsen et al., 2010) have found that cows had preferences for different concentrate feeds, tastes and physical forms. Migliorati et al., (2009) has found that addition of flavoring agents has the potential to higher the feed palatability and cows’ preferences. Baumont (1996) and Spörndly and Åsberg (2006) suggests molasses as a flavoring agent to higher cows’ preferences for a feed.

1.1 Aim and project questions The concept of AMS is based on the cows’ motivation to voluntarily visit the milking robot, to assure a sufficient milking frequency between 2.5-3.5 milkings per day. If the milking frequency is not sufficient the potential to a decreased need of labor and an increased milk yield in AMS is misused. The use of concentrate feeds to guide the cows into the milking robot is widely used to assure sufficient milking frequency. This presupposes the use of highly preferred concentrates that, if they were used as concentrate feeds in the milking robot, could be used in small amounts and still has the potential to increase the number of voluntary visits. Therefore the overall aim of this thesis was to study if dairy cows have preferences for some concentrates compared to others, answering the flowering questions:   

Do dairy cows have preferences for a specific concentrate, when it is offered together with others? Are the preferences depending on physical form of the concentrate? Is it possible to higher the preferences by addition of sugar beet molasses as a flavoring agent?

The amount and content of the tested concentrates offered should not compromise the intake of the ad libitum feed, fed at the feeding lane.

6

Introduction

1.2 Method and objectives The first part of the thesis includes a literature review which in further details describes cows’ eating behavior, factors affecting cows’ preferences for a feed and methods for measuring cows’ feed preferences. Finally the effect of voluntary visits to the milking robot on cow performance was reviewed. The objectives of the literature review offers background information on the topic feed preferences. Further the literature review provides inspiration for different concentrate feeds to test, and methods to use for measuring the feed preferences in the experimental part of this thesis. Also the literature review serves as background for the interpretation and discussion of the results found in the experimental work. The second part of the thesis includes a report of the experimental part of this thesis, conducted to answer the project questions. The experimental work will be presented in two forms in this thesis. First a manuscript for a fully scientific article according to the guidelines for Journal of Dairy science is presented. The objective of the scientific paper is to present the results of the experimental work to other researchers. Secondly a manuscript for a popular scientific article presenting a short review of the experimental work is included. The objective of the popular scientific paper is to present the results of the experimental work to people with interest in the topic, but without having a scientific approach to it.

7

2 Literature review The areas of special interest in this literature review was cows’ eating behavior, factors affecting cows’ preferences for a feed and methods for measuring cows’ feed preferences. Finally the effect of voluntary visits to the milking robot on cow performance was reviewed. This chapter was based on review of articles found available in the literature.

2.1 Definitions of terms The purpose of this section is to define the understanding of terms used in this thesis which may have multiple understandings. The term mixed ration (MR) is used to describe the daily ration that cows are fed, at ad libitum amounts. Another term for the same is total mixed ration (TMR). TMR was defined by Schroeder and Park (2010) as “the practice of weighing and blending all feedstuffs into a complete ration which provides adequate nourishment to meet the needs of dairy cows.” Furthermore they states that “each bite consumed contains the required level of nutrients (energy, protein, minerals and vitamins) needed by the cow”. Common praxis when using separate concentrate feeding is to feed some of the energy or protein separate as concentrates. If so, the MR contains a little less energy or protein than the requirements to avoid supply in excess. The term feed preferences was defined by Hodgson (1979) as the animals’ ability to discriminate between different feeds offered at the same time in free choice situations. This is also the understanding of feed preferences in this review. The term palatability is often used solely of the characteristic of taste (Forbes, 2010). When the literature on the subject was studied is was found that the interpretation of palatability was more than that. Palatability is linked to more factors concerning the feed and the animal in question (Baumont, 1996; Forbes, 2010; Meier et al., 2012). Therefore the use of the term palatability cannot be used only regarding taste as is a dynamic term that changes between situations. It is assumed that palatability and preferences are linked in the way that a high palatability of a feed will result in high feed preferences. Many terms are used to describe the concept of preference experiments. Terms as choice feeding experiments, choice experiments, preference trials, preference tests, preference study or palatability experiments are used to describe experiments where animals have to make a choice between feeds offered at the same time. A special type of preference trials is termed cafeteria trials or cafeteria experiments. These experiments describes situations where animals are offered more feeds at the same time and then should choose which one they prefers (Meier et al., 2012).

8

Literature review

2.2 Eating behavior of cows The purpose of this section is to give a summary of the normal feeding- and eating behavior of dairy cows. Since behavioral science and ethology is not the approach used in this thesis no detailed discussions on the topic will be included. The main concern for an animal is to ingest enough food to survive. Different animal species are morphological adapted to eat different types of feeds that are available at the natural habitats and feeds that have the characteristics to fulfill the animal’s nutritional requirements (Albright, 1993). Based on this, different animal classes can be defined. Cows and other herbivores is the class of animals, which consume only plant material, but the class can further be divided into subgroups according to the animal’s anatomy and choice of feeds. In that way grazers and browsers are two different subgroups of herbivores. Browsers are highly selective animals that composite their diet from many different types of feeds, whereas grazers are less selective and eat mostly monocultures, such as grass (Shipley 1999). Considering the eating behavior of dairy cows it is obvious to evaluate cows’ very natural eating behavior, as it was even before domestication. Albright (1993) stated that cows were “crepuscular animals”, which means that they mainly foraged around sunrise and sunset. Cows are herbivores and ruminants, meaning that their natural feeds will consist of plant material, which is digested in the rumen and the remaining part of the digestive tract (Tucker, 2009). The natural way for a cow to eat is to gather plant material (mainly grasses) into the mouth using their tongue. Cows do not have incisors on the upper plate, but instead they have a hard ridged dental pad, which is mainly used as a grinding surface during rumination. Rumination is the process of re-chewing of partly digested feed. This process helps breaking down the heavily digestible parts of the feed, mainly the fibers. The rumination time per day is related to the type of feed ingested, but is in general between 6-8 hours per day (Tucker, 2009). It has also been suggested that rumination may be an “anti predator strategy” from the undomesticated cows, making them able to gather a lot of food in the open areas around sunrise and sunset. During the day the cows then could lie in a more safe area in the woods and digest the food. This strategy minimized the time in the open land highly exposed to predators (Tucker, 2009). Within ruminant livestock both browsers and grazers are found. Cows and sheep are grazers, whereas goats and deer are browsers (Shipley 1999). Within each of the two classes of animals differences in the animals selectivity and ability to composite their own diet to fulfill their nutritional requirements are found. Fraser et al., (2009) found by comparing cows to sheep that sheep are more selective than cows. In the experiment reported by Fraser et al., (2009) the two species applicability for nature preservation by grazing on extensive grasslands were compared. It was found that the sheep were more selective in their diet composition both regarding the number of plant species they like to eat and the quality of the plants compared to the cows. In that way the sheep consumed a much more diverse diet, selecting more fresh and new leaves than the cows. This could also be related to the morphological differences between the two species, as sheep has a smaller mouth than cows, making it easier for them to be selective in their choice of feeds.

9

Literature review

Cows are social animals, meaning that one cow can be stimulated to a behavior (for example eating) by the impact of other cows in the group doing the same behavior (Albright, 1993). The domestication of dairy cows has led to adaptation to new feeds, feeding systems and housing systems, but taking the animals natural behaviors into account in strategies for feeding and housing cows may be beneficial and lead to increased feed intake, animal welfare and thereby an increased milk yield. Eating behavior of modern dairy cows may have changed, compared to the natural one. Eating behavior of dairy cows is influenced both by the distribution and type of feed offered (Tucker, 2009). Cows kept on pasture have an eating behavior most related to the natural one. This means that all cows (more or less) will have a synchronized behavior pattern which means that they also will eat at the same time, mostly in day times. Cows on pasture will spend up to 10 hours a day grazing (Tucker, 2009). Cows housed in free stall barns use less time (4-6 hours per day) to eat, and due to limit space around the feeding lane, not all cows are allowed to eat at the same time, forcing the cows further away from their natural eating behavior (Munksgaard et al., 2005; Tucker, 2009). In a competitive eating situation, cows of high rank in the group spend more time eating leading to a higher feed intake. This correlation between eating time/feed intake and social rank has been found to be more distinct the more limited the space at the feeding lane is per cow (Olofsson, 1999). However, the study by Munksgaard et al., (2005) showed that cows have some ability to compensate for a shorter eating time, by increasing the speed of feed intake. This, however, was shown by restricting the time available for eating, and not in a competitive eating situation. In general, the regulation of cow eating behavior is controlled by interactions between internal and external factors. Internal factors regulating the eating behavior could be age of the animal, condition of the teeth, the physiological stage, etc. but in general the internal regulation of the cows’ eating behavior are controlled by many complex, not well understood mechanisms, involving hormones, nerves, blood concentrations, etc. (Valros and Hänninen, 2009). Also external factors regulate the eating behavior of cows. In that way factors related to the feed and surroundings are also important for cows’ eating behavior (Valros and Hänninen, 2009). For example high temperatures in the surroundings depress feed intake and milk production (Albright, 1993). More details regarding the effect of feed related characteristics and the animals’ physiological stage on eating behavior and feed preferences are included later in this review.

2.2.1

Conclusions



Cows are grazers, making them relative low in selectivity of food.



Cows are social animals that eat many hours a day.



Competition for feed decrease eating time and feed intake especially in low ranking cows.



Many internal and external factors regulate eating behavior of cows.

10

Literature review

2.3 Factors affecting feed preferences The purpose of this section is to summarize the different factors that can influence a ruminant’s preferences for a feed. Since many of the experiments reported in the literature within the topic are done on smaller ruminants, also these are studied, but the main focus is dairy cows. The stated differences in selectivity and preferences between animal species should be kept in mind when results from experiments using different animal species are evaluated. Ruminants’ ability to choose between different feedstuffs has been studied for many years. One of the first experiments with dairy cows reported in the literature was Nevens, (1927) (cf. Meier et al., 2012). The interest in studying cows’ preferences comparing different feedstuffs can be multiple, either related to the experimental feeds or related to the animal’s behavior related to the experimental feeds (Meier et al., 2012).

2.3.1

Feed related factors

Generally, both physical characteristics and chemical characteristics of a feed affect an animal’s feed preferences. Morand-Fehr (2003) found that goats prefer dry pelletized feeds to the same dry feeds in the form of flour. Andersen and Lykkeaa (1966) stated that cows prefer a pellet form of concentrates to a grounded one. Meier et al., (2012) stated that cows’ preferences for a dusty or fine grounded feedstuff can be enhanced by moistening or pelletizing. Meier et al., (2012) stated that physical characteristics of a feedstuff, such as form, length and density, may influence the feed preferences. An example of the influence of physical form on feed preferences from Kenny and Black (1984) is shown in Figure 2.1. The figure shows how sheep’s preferences for a feedstuff (straw and hay) is affected by particle size as the preferences are increased, as the proportion of short particles in a mixture of short and long particles are raising (0:11:0).

Figure 2.1 Preferences for a mixture containing short and long particles in different ratios. Adapted from Kenny and Black (1984)

11

Literature review

The chemical composition of a feed is also important for feed preferences. Also the general quality of the feed is found to influence the palatability and feed preferences. Ginane et al., (2003) found that quality traits of a fiber rich feed for cows could be rumen fill or the passage rate, as they in an experiment on pasture with heifers found that the heifers preferred grass of good quality. The good quality included low rumen fill and a rapidly digestion to maintain a high total dry matter intake of the heifers. Other experiments reviewed by Meier et al., (2012) also found that the fiber content and properties of the fibers in a fiber rich feed can influence the quality of the feed and by so the feed intake rate and feed preferences. Also it seems possible that the ash content could affect the general quality, and thereby the palatability and preferences of a feed with high ash content associated with a low quality of the feed. Morand-Fehr (2003) reviewed a series of feed preference tests on goats. It was found that the preferences for concentrates and byproducts differed according to sugar content, fatty acid composition, oxidative level of fat and content of anti nutritional agents, such as glucosinolates in rapeseeds. The goats’ preferences for different mixed concentrates increasing the proportion of low palatable rapeseeds (high in content of glucosinolates) were studied. It was found that when the proportion of low palatability feeds in a mixed concentrate was increased then the preferences for the concentrate dropped. Larsen et al., (2012) found that cows reduced total dry matter intake when they were fed a MR containing increasing levels of oilseeds (rapeseed and linseed). However a denser energy level in the ration containing oilseeds compared to a control diet (without oilseeds) compensated for that, so energy intake was similar between the control and the oilseed rations. Decreased dry matter intake is common when unsaturated fat and oil is fed to cows properly due to metabolic regulation on feed intake, or a decreased NDF digestibility and a lower the passage rate (Larsen et al., 2012). Baumont (1996) reviewed experiments on modification of the taste of feeds by adding flavor agents to the feeds. It was found that a sweetening, by adding molasses to a manure affected area of a sward improved grazing. However, the adding of molasses to a normal palatable pellet of hay, did not significantly affect the intake of that pellet compared to a pellet of hay without addition of molasses. Spörndly and Åsberg (2006) tested the addition of molasses to a pellet of barley and found that heifers’ preferences for the molasses added barley pellet to be higher compared to a barley pellet without molasses addition.

2.3.2

Animal related factors

Separating the effect of taste and smell on feed preferences was attempted in the master thesis by Michalet-Doreau (1975) (cf. Baumont, 1996). The experiment tested the effect of taste and smell on silage intake by using anosmic (non-smelling) and agustatory (non-tasting) sheep compared to normal sheep. All animals were fed a good and a badly preserved silage and the feed intake was measured. Only the anosmic animals differed in feed intake compared to normal animals. This difference was largest when the badly preserved silage was used. The anosmic sheep had a higher feed intake

12

Literature review

of 33 % compared to normal sheep when the badly preserved silage was used. This trial emphasized that the effect of smell is of great importance for animals’ choice of a feed and feed preferences. Preferences for a feedstuff are also affected by other factors. First of all animal species will influence feedstuff choice and feed preferences like accounted for earlier. Therefore it is important to notice which animal species an experiment is conducted on when results from preference experiments are evaluated. Even though ruminant animals have many things in common, many things regarding feeding behavior differs between the ruminant species (Meier et al., 2012). Also hunger seems to be of importance for an animal’s feed preferences, as hungry animals tend to eat the first feed offered to them (Meier et al., 2012). Villalba and Provenza (1999) studied the effect of physiological status on feed preferences in lambs. They tested the lambs preferences for alfalfa vs. barley, when lamb were feed a MR either in nutritional balance, high in protein and low in energy, or low in protein and high in energy. They found that, if the lambs were fed the MR in nutritional balance they preferred barley over alfalfa (P < 0.001). When the lambs were feed the MR low in protein and high in energy, they ate more alfalfa (71 vs. 9 g) and less barley (142 vs. 232 g.) than the lambs fed the MR high in protein and low in energy. The experiment also included different structures of the feeds. The overall conclusion of the experiment was that the nutritional value of the feeds in relation to the physiological requirements of the lambs was more important for the feed preferences than the structure of the feeds. However, it was stated by Atwood et al., (2001) that even at the same physiological status individual animals have different requirements and also different feed preferences. Consequently, it is important to compensate for variation between animals preferences when a preference experiment is planned, with the Latin square as a way of minimizing the bias from individual animals (Meier et al., 2012). Previous experiences of the animal will also affect the preferences of a feed. Distel et al., (1994) stated that especially if an animal in early life (in sensitive periods) are exposed to a specific feed, it will affect the preferences for that feed if it is exposed to it again later in life. Also Villalba et al., (2004) showed that, in an experiment with weaned lambs. In the study by Villalba et al., (2004) experimental lambs were fed feeds containing high levels of anti nutritional factors at weaning. When doing a preference experiment on the lambs after 3 and 8 month the experimental lambs had higher preferences for feed containing anti nutritional factors than control lambs. This shows that early experience affect feed preferences in later life.

2.3.3

Conclusions



Ruminants’ preferences for a feed depend on characteristics linked to the feed and the animal in question.



The feed characteristics can either be linked to the physical or the chemical properties of the feed.

13

Literature review



Cows prefer a pelletized form of concentrate above a ground form of concentrate.



It is possible to improve feed preferences by modifying the palatability by adding flavor agents, like for example molasses.



An important animal characteristic is physiological status of the animal, but individual animals have individual preferences.



Previous feed experiences influence the animals feed preferences.

2.4 Methods for measuring feed preferences and palatability The purpose of this section is to summarize the different methods for measuring cows’ feed preferences. The focus in this section is on dairy cows but also experiments involving smaller ruminants are included. Like in the previous section the stated differences in selectivity and preferences between animal species should be kept in mind.

2.4.1

Measurements

It is difficult to measure the palatability of a feed and the animals’ preferences for the feed directly. Therefore mostly factors associated with the palatability and preferences can be measured. The most used indicators for the palatability and the preferences for a feed are the choice of feeds in a choice situation, the voluntary feed intake and behavioral measurements (Baumont, 1996; Meier et al., 2012). The choice of a feed and behavioral registrations measure the motivation to ingest a feed, whereas, the voluntary feed intake is the result of the motivation to ingest the feed (Baumont, 1996). The voluntary feed intake is often measured as feed intake rate, as the amount of feed eaten in a specific unit of time (Spörndly and Åsberg, 2006). Under certain conditions (e.g. grazing on pasture or loose housing) it is more reasonable only to measure the time spend eating and then associates that with the feed intake (Baumont, 1996). In the literature it has been discussed in which degree cows have the ability to choose feeds that will fulfill their physiological requirements (Forbes, 2010; Meier et al 2012). In this literature review it is shown that cows are not highly selective animals, making it a challenge for them to choose feeds that completely will fulfill their physiological requirements efficiently. Forbes (2010) however states that it is important to distinguish between short and long term feed intake as it will take some time for the animal to experience the physiological effect of the feed. Consequently it is important not to draw long term conclusions on short term experiments. Behavioral measurements can be done in different ways. Measurements can be done on animals with free access to more feeds offered at the same time, e.g. in a cafeteria system. Another type of 14

Literature review

behavior measurement is to record the willingness of the animals to do a certain amount of work to obtain access to the feed or to measure the animals willingness to get to access an elected feed (Baumont, 1996).

2.4.2

Factors of importance for measuring feed preferences

In experiments with feed preferences the housing system is of importance. The first thing to consider is whether it is a preserved or a fresh feed which is of interest. If it is a fresh feed it is very likely that the experiment is to be done outdoor on pasture. In the case of outdoor experiments on pasture the choice will mostly be between different plots of cultivated monocultures of e.g. grass or clover species (Meier et al., 2012). The measurements of feed intake will most likely be as a time recording of time spend grazing the different types of swards which then is associated with the feed intake and thereby the preferences (Baumont, 1996). When preserved feeds are of interest the experiment is most likely to be carried out under indoor housing conditions, but even so more opportunities are to be considered, most important are whether the animals are housed individual or in a group. Mostly the individual housing system are used for preference experiments with cows, as it is easier to measure feed intake and do behavioral registrations on individual housed animals (Meier et al., 2012). When the experiment is done on individual housed animals, the feeds are offered as two or more at the same time in containers with a suitable size in order to avoid feed losses. As individual animals can have an inborn preferences for one side (left or right) is it important to rotate the physical placement of the feed containers so that such a possible bias are eliminated within the experiment (Meier et al., 2012). Pervious feed experiences affect a cow’s preferences for a feed. Therefore it is suggested both by Meier et al., (2012) and Forbes (2010) that cows before a preference experiment are adapted to the feeds so that they will recognize them in the choice situation in the experiment. The optimal length of the adaptation period is depending on the purpose of the experiment. For example if total feed intake and digestion related topics are of interest the adaptation period need to be longer than in palatability and preference studies (Meier et al., 2012). Hodgson (1979) defined a preference ranking, as a possible ranking of the preferences between feedstuffs in a free choice experiment. However he addresses the fact that a complete free choice situation is hard to establish under experimental conditions as some constraints are already made when designing the experiment. Also external factors make it hard to offer complete freedom. Therefore he defines the term “diet selection” as a function of the preferences modified by the opportunity for selection, which is determined by constraints in the setup and environment. Furthermore many mammalian animals, also ruminants, are inborn with preferences towards specific flavors, often sweets, which will drive the animals towards specific feeds no matter of previous experiences (Forbes 2010).

15

Literature review

Many of the important factors to consider when planning a preference test are summarized in Figure 2.2. The figure shows the most important factors to consider when a choice feeding experiment is planned and carried out. Most of the factors are described more detailed in the above section.

Figure 2.2 Factors to consider when designing a choice feeding experiment. Adapted from Meier et al., (2012).

2.4.3

Conclusions



It is difficult to measure the palatability or feed preferences direct, but choice of a feed, the voluntary feed intake and behavioral measurements can be used as expressions for the palatability and feed preferences.



More factors have to be considered when performing experiments with palatability or feed preferences. Important factors that can affect the outcome can be related either to the o Animal  Genetics, physiological stage and previous experiences. o Feedstuff  Type of feed, structure and number of feeds. o Experimental setup and environment  Housing system, placement of feed containers and adaptation period.

16

Literature review

2.5 Effect of voluntary visits in the milking robot on cow performance The purpose of this section is to summarize and describe the effect of voluntary visits in the milking robot for cow performance. Furthermore possibilities to improve the number of voluntary visits in the milking robot are described. Different housing, feeding, and milking-systems that are used within milk production will be included, but the focus in this section will be on the concept of AMS. As mentioned the concept of AMS presuppose that the cows are willing to visits the milking robot voluntarily. Based on results from 19 studies reported in the literature Erdman and Varner (1995) calculated a positive effect on milk yield of 3.5 kg/d by increasing the milking frequency from two to three times per day. Wagner-Storch and Palmer (2003) concluded that an increased milk yield (26.4 vs. 25.8 kg/d) in an AMS compared to a traditional milking parlor was mostly correlated with an increased milking frequency (+0.4 times/d, 2.0 vs. 2.4) in the AMS. If cows in an AMS do not turn into the milking robot to be milked on a regularly basis the udders are forced to storage different amounts of milk depending on the regularity in the milking intervals, increasing the risk of udder related diseases like mastitis (Bach and Busto, 2005). Bach and Busto (2005) found that the negative effect of irregularly milking intervals became more severe as days in milk increased and also more severe for mulitparous cows compared to primiparous cows.

2.5.1

Traffic system

In productions with AMS different cow traffic systems are possible. The two most used systems are the “forced” and the “free” cow traffic system. In the forced cow traffic system, cows are forced to pass through the milking robot to get access to essential facilities like the feeding lane, water troughs etc. In the free cow traffic system it is assumed that cows voluntarily will pay sufficient visits to the milking robot (Halachmi et al., 2009). Within the concept of free cow traffic different concepts of attracting the cows into the robot are used (Halachmi et al., 2009). One concept is the “semi forced” cow traffic, where cows have to go through the robot to get access to semi essential facilities like for example concentrate feeders (Hermans et al., 2003). Another concept is to feed the cows a mixed ration containing a little less protein or energy than their physiological requirements. This will stimulate them to visit the robot to get concentrate to fulfill their physiological needs (Halachmi et al., 2005). The last widely used concept within a free cow traffic system is the “candy concept” where the cow is attracted to the robot by a tasty feed that they like to eat (Halachmi et al., 2006). The topic of cow traffic systems in AMS has been well studied in the literature. Hermans et al., (2003) compared two cow traffic systems, a forced and a semi forced system. In the semi forced system cows had to pass through the milking robot to get access to concentrate offered in concentrate feeders. They did behavioral registrations and recorded the milk yield and the number of visits to the robot, with or without milking. The visits without milking were defined as cows which not were allowed to enter the robot if time since last milking were less than 6 hours. Regarding the number of visits they found no significant differences between the two systems, but the number of 17

Literature review

visits without milking tended (P < 0.10) to be lower in the semi forced system. From the behavioral registrations they found that the cows in the forced system spend significant less time at the feeding lane, reducing the daily dry matter intake of the MR. They found no significant difference in the milk yield between the two systems. Overall they concluded that the semi forced system was better than the forced in the way that the cows in the semi forced system spend more time eating and that they could do it at the same time benefiting the natural eating behavior of the cows. It was mentioned that the tendency towards a lower number of visits without milking in the semi forced concept, could indicate that this system used the capacity of the robot better as less none-milking cows preoccupied the robot. A study by Munksgaard et al., (2011) compared forced and free cow traffic in a crossover experimental setup. In agreement with Hermans et al., (2003) they did not find any significant effect on milking frequency or yield between the two systems, however a tendency to a small increase in milk yield (0.6 kg/day) in the free cow traffic system were detected. Furthermore no differences between the eating behaviors of the cows in the two systems was found, which could indicate that the cows in the two systems spend the same time at the feeding lane in contrast to the findings by Hermans et al., (2003). From the findings Munksgaard et al., (2011) concluded that the traffic system did not significantly affect any of the parameters in question. However, they addresses the points that the milking frequency in general were high in the experiment (mean > 3 milking per cow per day) and that the number of cows per robot in the experiment were relative small (35 cows). Other studies (Rodenburg and Wheeler 2002, cf. Munksgaard et al., 2011) have showed that the milking frequency decreases with increasing number of cows per robot, especially with group sizes above 60 cows per milking robot. In contrast to the findings by Hermans et al., (2003) and Munksgaard et al., (2011) Bach et al., (2009) found a higher milking frequency and a lower number of cows to be fetched in a forced cow traffic system compared to a free cow traffic system. The experiment included a crossover setup with two periods and 2 treatments. Each period was 3 months with the two last months as being registration periods. MR and concentrates allowance were the same in the two traffic systems. Besides registrations on milking frequency and number of cows to be fetched, registrations also included individual cow eating behavior and feed consumption. The forced system had a higher total number of milking and voluntarily milking compared to the free cow traffic system. The forced traffic system also had a lower number of cows to be fetched and a lower milking interval. Despite the higher milking frequency in the forced cow traffic system no significant differences on milk yield were detected. This lack of positive yield response on increased milking frequency could be explained by an increased variation in weekly milking frequency, which was significantly higher in the forced traffic system. Bach and Busto (2005) stated that an uneven milking frequency, with a variation above 27 % will result in a decreased milk yield, with multiparous cows more sensitive than primiparous. Bach et al., (2009) showed an altered eating behavior between cows in the free or forced traffic systems. There was no significant difference between the total dry matter intake of either MR or concentrate between cows in the two traffic systems. However, cows in the forced traffic system had

18

Literature review

significant fewer meals per day than cows in the free traffic system (6.6 vs. 10.1), resulting in significantly more time spend per meal and a higher dry matter intake per meal.

2.5.2

Amount of concentrate

Feeding dairy cows in a system with AMS is an interaction between the feeding of the MR on the feeding lane, and feeding of the concentrate in the robot (Madsen et al., 2010). Weisbjerg and Munksgaard (2009) conducted an experiment including three levels of energy in the MR (low, medium, and high) in combination with two levels of concentrate allowances (3 vs. 6 kg) offered in the milking robot. This resulted in four treatments, low energy MR + 6 kg concentrate, medium energy MR + 6 kg concentrate, medium energy MR + 3 kg concentrate, and high energy MR + 3 kg concentrate. They found that the milking frequency was respectively 3.13, 2.64, 2.52 and 2.66 (P < 0.04), the number of visits without milking permission was respectively 2.44, 1.30, 0.57 and 0.91 (P < 0.01) and a milk yield (ECM) of respectively 31.4, 31.8, 30.2, and 34.9 kg/day (P < 0.01). These results showed that at low energy level in the MR it was beneficial to offer a high amount of concentrate in the milking robot, as this resulted in an increased milking frequency compared to the other treatments. However, the number of cows visiting the robot without milking permission was also increased, thus the capacity of the milking robot was reduced as stated by Hermans et al., (2003). Bach et al., (2007) tested whether a high amount of the total concentrate offered in the robot would affect the milking frequency, eating behavior or milk yield of 115 cows in a 90 days crossover experiment. The experimental treatment was high amount of concentrate offered in the milking robot (up to 8 kg per cow per day) and low amount of concentrate in the milking robot (up to 3 kg per cow per day). All the cows got the same net energy and crude protein despite experimental treatment, but with different amounts of concentrate offered in the robot. They found no significant effect on any of the parameters in question between the two treatments concluding that a high amount of concentrate offered in the robot could not simulate the cows to visit the robot. However, they also analyzed the data excluding data from the cows that had to be fetched into the milking robot. This analysis showed a significant difference on milking frequency (2.4 vs. 2.7) between low and high amount of concentrate offered in the robot. From this it could be concluded that offering a high amount of concentrate in the robot do not stimulate visits of the cows that do not voluntarily want to visit the robot, whereas cows that already visit the robot voluntarily will be stimulated to visit it even more. Halachmi et al., (2005) found no significant effect on milking frequency between a high concentrate allowance in the milking robot (up to 7 kg per cow per day) and low concentrate allowance in the milking robot (up to 1.2 kg per cow per day). The experiment included 100 cows divided into two groups, one on each treatment (high or low concentrate allowance in the milking robot). Both treatment groups were offered ad libitum access to the MR. From the experiment it was concluded that 1.2 kg concentrate per cow per day was sufficient to attract the cows to the milking robot. Throughout the experiment the cows on the high concentrate allowance treatment increased the

19

Literature review

milk yield significant, which the cows on the low allowance treatment did not. This could indicate that the cows on the low allowance treatment not were able to compensate by eating more of the MR. Despite the positive effect of high concentrate allowance on milk yield, they addressed the importance that a high amount of concentrate offered in the robot in a short period of time potentially can cause metabolic diseases and definitely will have a negative effect on feed costs.

2.5.3

Taste and palatability of the mixed ration and the concentrate

Weisbjerg et al., (2013) found a numerically higher milking frequency for cows fed a MR containing saturated fat compared to a ration containing unsaturated fat. The concentrate fed in the milking robot was the same in the two treatments. The MR in the treatment including saturated fat was considered to have a low palatability. Logically, if the palatability of the mixed ration offered at the feeding lane is high relative to the palatability of the concentrate offered in the milking robot visiting frequency should decrease (Weisbjerg et al., 2013), and by so oppositely increase if the relative palatability changes. This suggests that the milking frequency is depending not only on the palatability of the concentrate used in the robot, but also depending on the interaction between the palatability and taste of the mixed ration fed at the feeding lane relative to the palatability and taste of the concentrate fed in the milking robot. Migliorati et al., (2009) reported an effect of adding flavoring and appetizing substances to the concentrate used in the robot. The experimental design was a crossover design with 52 cows divided on two treatments (control: concentrate without addition of substance. Test: the same concentrate as control, but with addition of substance). Milking frequency, total intake of concentrate, visits without milking permission (time since last milking less than 5 hours), milking interval and milk yield were recorded. The cows ate all of both concentrates, thus no difference in total intake of the two concentrates was detected. The cows on the test concentrate had a significantly higher milking frequency and number of visits without milking, a shorter milking interval and a higher milk yield. In agreement with other studies the higher milk yield in this experiment may be explained by the significant higher milking frequency causing a shorter milking interval. Hermans et al., (2003) suggested that a high number of none milking visits may lower the capacity of the milking robot. If so, the positive effect of adding the substances may have been lowered in the experiment by Migliorati et al., (2009) due to a misuse of the robot capacity.

2.5.4

Composition of concentrate

Madsen et al., (2010) tested six different concentrates against a control concentrate rich in fiber. All of the six test concentrates were based on different components (barley, wheat, barley/oat, maize, artificially dried grass and high fat). The experimental setup was six separate crossover trials with 30 cows (15 on each treatment) tested for 14 days before the crossover. Allowance of each concentrate was up to 5 kg/cow/day, but released to the specific cow in relation to time of the day and total time spend in the robot per day. The reported responses were concentrate intake, concentrate lefto-

20

Literature review

vers, number of visits with or without milking, milk yield and number of cows to be fetched. The visits without milking were defined as cows that were not allowed to enter the robot if time since last milking were less than 6 hours. Cows were fetched when time since last milking exceeded 14 hours until 90 days in milk and 16 hours after 90 days in milk. The results showed differences between concentrate compositions within all the reported responses. The barley/oat containing concentrate was better than the control, with a higher concentrate intake, less leftovers, higher number of visits (P < 0.05) (both with or without milking), higher milk yield and lower number of cows to be fetched. The wheat containing concentrate also showed positive (not significantly) effects on the measured parameters. The artificially dried grass had a negative (significantly) effect on the measured parameters compared to the control concentrate, except the number of visits without milking and cows to be fetched. Also the fat containing concentrate showed negative (significantly) effects on the measured parameters. The results on the measured parameters using the pure barley and the maize containing concentrates corresponded to the results by using the control concentrate. The authors concluded that it is possible to manipulate the number of voluntarily visits into the milking robot by the composition of the concentrate offered in the robot. Spörndly and Åsberg (2006) made a series of preference tests to screen the preferences for different concentrate feeds that could be used in a milking robot. The experiment included a series of preferences tests on dairy heifers and did not include any lactating cows or an AMS. In the experiment different concentrate feeds of different physical form were tested against rolled barley as a control. Higher preferences for pelletized concentrates compared to rolled barley were found. A ranking of the tested concentrates showed that the heifers preferred some concentrates to others and also that some concentrates were not eaten at all. Together with results from study by Madsen et al., (2010) the results from the study by Spörndly and Åsberg (2006) indicates that the composition of a concentrate will influence the animal’s preferences and possible affect the number of voluntary visits to the milking robot.

2.5.5

Management

This present review on the effect of voluntary visits into the milking robot on cow performance, and possibilities to affect the number of voluntary visits into the milking robot, has shown various results between experiments reported in the literature. Other researchers reviewing the same area (e.g. Jacobs and Siegford 2012) also found inconsistent conclusions between articles published. Therefore Jacobs and Siegford (2012) suggest that differences in management and farm-level variables may be more important to AMS efficiency and milk production than features of the milking system itself. The findings in this present review somehow support this understanding.

21

Literature review

2.5.6

Conclusions



The number of voluntary visits are important to utilize the fully potential in AMS



Milking frequency is positive correlated to milk yield



More variables are found to affect the milking frequency, including traffic system, amount of concentrate offered, the relative palatability between the mixed ration and concentrate, and concentrate composition.



Traffic system has shown various results regarding the number of voluntary visits



Amount of concentrate offered in the robot has shown various results regarding milking frequency. A positive effect of offering a high amount of concentrate was found, especially for cows that already pay a high number of voluntarily visits to the robot.



Composition and palatability of the concentrate offered in the robot has shown to affect cows’ preferences for the concentrate and the milking frequency when used in a milking robot.

22

3 Scientific paper

Do dairy cows have preferences for different concentrate feeds? L. Primdal Department of Animal Science, Faculty of Science and Technology, Research Center Foulum, Aarhus University. Denmark.

ABSTRACT To study dairy cows’ preferences for different concentrate feeds a series of preferences trials were done in a cafeteria setup. It was possible to test the preferences of four different concentrate feeds at a time. To compare between trials the same control concentrate were used in all trials. Therefore 9 separate preference trials were needed to study the 26 different (and 2 replicates) concentrates. Four cows were used in the preferences trials, in that way the experiment was done in a 4 x 4 Latin square. One trial were consisting of four registration days, and as it was not possible to do registrations on all the cows at the same time each cow fasted between one and four hours per day. Within the four registration days each cow fasted all possible hours. Placement of the different concentrates in the cafeteria were confounded to fasting hours, in that way within a trial the cows were presented to the four different concentrates at all four possible places in the cafeteria. The cafeteria was presented to the cows for ten minutes. Within each registration day three repetitions were done on each cow. Feed leftovers after the ten registration minutes (leftovers) and time, after offering the cafeteria a box containing concentrate were considered empty (time) were the responses to be recorded. Behavioral registrations regarding the behaviors eat and sniff were also done. It was found that the two responses leftovers and time both were sensitive showing almost the same preferences ranking of the concentrates. The behavioral registrations supported the ranking done by the responses leftovers and time. The results showed that the cows were able to discriminate between the offered concentrates and that they preferred some concentrates to others. Only one concentrate (soybean meal) was preferred higher than the control (P < 0.05). Few concentrates were preferred at the same level as the control. Most concentrates were preferred lower than the control (P < 0.05). A tendency that pelletized barley was preferred to rolled barley was found, but as the highest preferred concentrates included the forms of meal, pellets and rolled no clear evidence of relation between physical form of the concentrates and feed preferences were shown. However the content of the highest preferred concentrates were different making them difficult to compare.

23

Scientific paper

The addition of 25 % sugar beet molasses as a flavoring agent raised the preferences of four tested concentrates (wheat, dried sugar beet pulp, dried grass pellets and peas) as they all with the addition of sugar beet molasses were preferred higher than the pure form regarding both leftovers and time. Pure sugar beet molasses were preferred low in the experiment regarding both responses.

INTRODUCTION The use of automatic milking systems (AMS) has won forward the recent years, resulting in an increased interest in the use of small amounts of concentrate to guide cows into the milking robot (Spörndly and Åsberg, 2006). The benefits from AMS compared to traditional milking parlors are a decreased work load for milking and cost of labor (Rodenburg, 2011). Furthermore an increase in the milking frequency from the traditional two times a day in traditional milking parlors to more than two when using AMS has the potential to higher the milk yield (Erdman and Varner, 1995). The utilization of the fully potential in AMS is depending on the cows’ voluntary visits to the milking robot. Otherwise a lot of labor is needed to fetch cows on a regularly basis and guide them into the milking robot and the potential increased milking frequency and milk yield is not obtained (Rodenburg, 2011). Two traffic systems are widely used to guide the cows into the milking robot, the “free” and “forced” cow traffic systems. In the forced cow traffic system, cows are forced to pass through the milking unit to get access to essential facilities like the feeding lane or resting areas (Halachmi et al., 2009). In the free cow traffic system it is assumed that cows voluntarily will pay sufficient visits to the milking robot (Halachmi et al., 2009). Forced cow traffic is found to reduce eating- and resting time, which could compromise the milk yield and welfare of the cows (Hermans et al., 2003). Another method to attract the cows to the milking robot is to lower the energy or protein level in the mixed ration offered at the feeding lane combined by the offer of a high amount of energy or protein rich concentrate in the robot. This will stimulate the cows to visit the milking robot to get the concentrate they need to fulfill their physiological requirements (Weisbjerg and Munksgaard, 2009). A high amount of concentrate offered in the robot within a short period of time can potentially cause metabolic diseases and will have a negative effect on feed costs (Halachmi et al., 2005). These possible negative impacts of forced cow traffic and high amounts of concentrate eaten in a short period of time makes it of special interest to find a highly preferred concentrate feed that used in a small amounts can attract the cows voluntarily into the milking robot often and regularly. Other previous studies (Klopfer et al., 1981; Spörndly and Åsberg, 2006; Madsen et al., 2010) have found that cows had preferences for different concentrate feeds, tastes and physical forms. Migliorati et al., (2009) has found that addition of flavoring agents has the potential to higher the feed palatability and cows’ preferences. Baumont (1996) and Spörndly and Åsberg (2006) suggests molasses as a flavoring agent to higher cows’ preferences for a feed. Based on these findings this experiment aims at comparing concentrate feeds regarding cows’ preferences, to find concentrate feeds that has the

24

Scientific paper

potential to attract cows to the milking robot often and regularly, even if they are used in small amounts. The objective of this study was 1) to investigate if cows have preferences for some concentrate feeds, when different concentrates were offered together. 2) To clarify if the cows’ preferences were depending on physical form of the concentrate, and 3) if it was possible to higher the preferences by addition of sugar beet molasses as a flavoring agent.

MATERIALS AND METHODS This current experiment was conducted in the stable of Research Center Foulum, Aarhus University, Denmark. The experiment was done as 9 separate trials. Each trial was conducted as a preference test with three test concentrates tested against a control concentrate in a cafeteria system. The control concentrate was the standard concentrate used at the Danish Cattle Research Centre (KFC) as concentrate in the milking robot. The control concentrate was the same in all 9 trials. Cows and housing Two sets each of four cows was used for the 9 trials (set one, and set two). Each set was consisting of four second parity Danish Holstein cows 40-85 days after calving at the beginning of the experiment. The cows were blocked in the two sets according to milk yield prior to the experiment, to obtain similar groups regarding milk yield. In that way each set were consisting of one relative high yielding, two relative middle yielding and one relative low yielding cow. The cows were housed in a tie stall on rubber mats and sawdust as bedding. They were stabled with one empty tie box between each cow. Set one was used in trial 1, 2, 3, 6, 7 and 9 whereas set two was used in trial 4, 5 and 8. Diet, feeding, and milking The cows were fed one time a day (at 8 AM) with a basic forage rich mixed ration (MR). On dry matter (DM) basis the MR was composed of 36.6 % grass clover silage, 30.3 % maize silage, 22.2 % crushed barley, 9.6 % rapeseed cake, and 1.3 % minerals and vitamins. The MR had an energy content of 6.6 MJ NEL/kg DM (estimated by NorFor, 2013). Chemical composition of the MR (Table 3.1) were based on table values for concentrates and for forages bore samples analyzed by NIR by Steins Eurofins (Holstebro, Denmark). The bore samples of the forages were taken prior to the experiment. The feeding of the MR was adjusted on a daily basis assuring that each cow was fed ad libitum (adjusted after daily leftovers). The cows were milked two fixed times a day (at 5 AM and 4 PM). At the milkings cows were fed concentrate in separate concentrate mangers.

25

Scientific paper

Table 3.1 Chemical composition of the mixed ration (MR). The results are based on table values for concentrates and bore samples analyzed by NIR for forages.

MR [%] Dry matter (DM)

36

Contents of DM Ash Crude protein Crude fat Crude fiber NDF1 Starch

7.2 14.6 3.9 17.1 35.8 21.4

1

NDF: Neutral detergent fiber.

The 9 trials were conducted over nine weeks, with one trial each week. The experimental days were from Thursdays to Fridays and experimental time from 9 AM to 12:30 PM. On experimental days the cows were not fed the mixed ration until after experiment registrations, nor fed concentrate at the morning milking. Adaptation period Before each new trial week, the cows were adapted to the following weeks test concentrates by feeding them a mixture consisting 750 g of each new concentrate plus 750 g of the control concentrate. The mixture was given in the separate concentrate mangers. From Saturday - Monday the mixture was given at the morning milking, but at Friday (the last day of the current weeks trial), the mixture was given after registrations at 12:30 PM. During the entire experiment (experimental- and adaptation- days) the cows were fed 1 kg of soybean meal at evening milking to assure protein supply. Feed samples and chemical analysis Feed samples of the MR and the single ingredients of the MR were taken weekly. Also a sample of each test concentrate and the control concentrate in the current week’s trial were taken. All the samples were kept at -20°C until analysis. Only the concentrates were chemically analyzed for; dry matter at 103°C, ash content by combustion at 525 °C for 6 h (method 923.03, AOAC 1990), total nitrogen content was determined by the Dumas method (Hansen, 1989) and crude protein (CP) was calculated as N x 6.25. The neutral detergent fiber (NDF) content was determined by a neutral detergent extraction according to Mertens (2002) with a FibertecTM M6 system (Foss Analytical, Hillerød, Denmark) using heat stable amylase and sodium sulfite and corrected for ash. Crude fat (CF) was extracted with petroleum ether (Soxtec 2050, Foss Analytical, Hillerød, Denmark) after hydrolyzing with HCl (Stoldt, 1952). In vitro organic matter digestibility (IVOMD) was determined as described by Weisbjerg and Hvelplund (1993). The control concentrate samples were pooled

26

Scientific paper

before analysis. The results of the chemical analysis, together with the physical form and the combination of the concentrates in the 9 trial are shown in Table 3.2. Table 3.2 Combination in trials, physical form and content of dry matter (DM), ash, crude protein (CP), neutral detergent fiber (NDF), crude fat (CF) and In Vitro organic matter digestibility (IVOMD) of the tested concentrates. The control used in all trials was a standard robot concentrate. % of DM Trial All 1

2

3

4

5

6

Concentrate

Physical form1

DM %

Ash

CP

NDF

CF

IVOMD

Control2

Pellets, 4 mm

87.2

6.5

19.9

24.7

3.7

90.1

Barley

Rolled

88.9

2.2

9.8

20.7

3.2

90.8

Maize

Pellets, 3mm

88.1

1.4

10.3

8.4

4.6

98.1

Barley

Pellets, 3mm

87.5

2.0

10.5

16.9

3.2

92.5

Wheat, NaOH

Whole

80.5

5.9

11.1

7.8

2.1

98.7

Wheat

Rolled

87.4

1.5

11.6

12.5

2.6

97.7

Oat

Rolled

88.3

2.5

10.5

28.8

5.7

78.4

Dried sugar beet pulp

Pellets, 8mm

90.6

4.3

10.5

40.8

1.3

95.1

Citrus pulp

Pellets, crushed

91.2

8.2

7.8

24.1

2.8

96.8

Dried sugar beet pulp

Rolled

91.3

4.3

10.3

40.0

0.6

94.9

DDGS3 I

Pellets, 5mm

89.9

5.8

35.1

24.3

6.6

92.1

"Kalvevalse I”

Rolled+ pellets, 4mm

86.3

7.9

22.9

17.0

3.4

94.7

Sugar beet molasses

Liquid

68.0

10.2

11.4

-

0.2

-

Rapeseed meal

Meal

88.2

7.3

39.9

30.0

3.6

82.2

Sunflower meal

Meal

90.2

6.7

40.7

32.7

3.0

77.8

Soybean meal

Meal

88.5

7.2

54.3

8.0

2.9

99.6

DDGS II

Pellets, 5 mm

90.4

5.6

34.9

25.9

6.8

90.9

Pellets, 4 mm

88.9

6.5

28.7

24.1

7.0

90.9

Rolled

86.5

1.9

8.1

17.1

1.9

97.3

Rolled

85.7

3.2

23.9

14.0

2.1

99.6

"EO"

4

Rye 7

Pea Pea + SBM

8

9

5

Pellets, 3mm

82.1

4.8

20.8

10.6

1.8

99.1

"Kalvevalse II"

Rolled+ pellets, 4mm

86.2

8.7

24.6

17.2

3.6

93.5

Dried sugar beet pulp + SBM

Pellets, 3mm

89.8

6.2

11.1

33.3

0.6

94.2

Wheat + SBM

Pellets, 3mm

83.7

3.5

11.8

9.9

2.0

97.3

Dried grass + SBM

Pellets, 3mm

88.7

10.3

17.3

40.1

2.6

80.7

Barley, Co2

Rolled

80.9

2.1

10.9

20.7

2.9

92.2

Dried grass

Pellets, 5 mm

90.7

10.2

18.6

48.4

3.2

76.0

Wheat, Co2

Rolled

81.9

2.1

13.7

13.5

2.6

95.8

1

For pellets, diameters are given. 2 Standard robot concentrate. 3 DDGS = Dried Distillers Grains with Solubles. 4 ”EO” = Essential oils. 5 SBM = Sugar beet molasses.

27

Scientific paper

Test concentrates Most of the used test concentrates were single component concentrates. Exceptions, however, was the control concentrate and test concentrates “Kalvevalse (I and II)” and “EO”. These were all artificial mixed concentrates combined by more ingredients. “Kavlevalse” were composed of 50 % concentrate pellets and 50 % steam rolled barley. The “EO” concentrate differed from the other concentrates as it had a content of essential oils (EO) of 0.02 %. The composition of control, “EO” and an average consignment of “Kalvevalse” can be seen in Table 3.3.

Table 3.3 Composition of the control and the test concentrates, "Kalvevalse" and "EO". The values represent content in percentage. Control Sugar beet waste (dry)

16.9

Rapeseed meal

16.9

Wheat

14.6

Barley

14.6

Soybean meal (toasted)

9

Citrus pulp

7

Sunflower meal

7

Lucerne

5

Wheat bran

5

Molasses (sugar cane)

3

Salt, Vitamins, and Minerals

1

“Kalvevalse”1

“EO” 15

4.2

10.6

15

19.5

4.8

0.58

Barley, steam rolled

50

Linseed cake

10

Sugar beet molasses

7

Maize

5

Dried grass pellets

3

8

Palm fat

1

1.8

4.5

Rapeseed cake

35

Pea

5

Essential oils

0.02

1

The composition of “Kalvevalse” is based on information on an average consignment.

The concentrates tested with addition of molasses (tested in trial 7 and 8) consisted of one feedstuff mixed with molasses. These all were added 25 % sugar beet molasses. The molasses added concentrates and the concentrates “maize, pellet”, and “barley, pellet” were made at the feedstuff mill of Research Center Foulum, Aarhus University. The pellets were made without the use of steam and

28

Scientific paper

were considered to be of good quality regarding pellet strength. The concentrates “Kalvevalse” and “DDGS” (Dried Distillers Grains with Solubles) were tested two times each (I and II), but from different consignments. Furthermore consignment I and II for both concentrates were tested with a different set of cows, meaning that I and II for each concentrate represents a real repetition both regarding feed and cows. The term “CO2“for barley and wheat in trial 9 means that the grains were harvested with a water content about 20 % and stored in silos filled with CO2 for anaerobic preservation. Preference test The experiment was performed as 9 single preference tests. In the preference tests the four concentrates in each trial (three test concentrates and the control) was offered in a cafeteria system, with the concentrates offered in plastic boxes lined in a row. Each box contained 150 g (± 2 g) of one concentrate. The cows in rotation fasted one, two, three, or four hours each day before they were offered the cafeteria. The concentrates were offered for ten minutes with three after each other following replications. The placement of the concentrate containing boxes in the cafeteria was confounded to time of fasting, and the placement were the same within the three replications. The rotation of the boxes in relation to time of fasting was done in a complete Latin square (Table 3.4), as well as the rotation of the cows within hours of fasting were done in a complete Latin square (Table 3.5). The Latin squares were similar every experimental week. In that way each cow fasted all hours (1 – 4) within a full trial week, and thereby also all the four different concentrates were presented to one cow at all four possible placements (see Table 3.4 and Table 3.5). Plastic boxes within one week of trial were used for only one type feed. Between trial weeks plastic boxes was washed carefully. Table 3.4 Experimental design within cows. Placement (Box A, B, C, D) of test concentrates (CON, 1 CON always control) in relation to hours of fasting.

Box Fasting (Hours)

Replication

A

B

C

D

1

1 2 3

1CON 1CON 1CON

2CON 2CON 2CON

3CON 3CON 3CON

4CON 4CON 4CON

2

1 2 3

4CON 4CON 4CON

1CON 1CON 1CON

2CON 2CON 2CON

3CON 3CON 3CON

3

1 2 3

2CON 2CON 2CON

3CON 3CON 3CON

4CON 4CON 4CON

1CON 1CON 1CON

4

1 2 3

3CON 3CON 3CON

4CON 4CON 4CON

1CON 1CON 1CON

2CON 2CON 2CON

29

Scientific paper

Table 3.5 Experimental design among cows. Fasting (hours, 1-4) divided by days.

Cow/fasting hours Experimental day 1 2 3 4

Cow 1

Cow 2

Cow 3

Cow 4

1 4 2 3

2 1 3 4

3 2 4 1

4 3 1 2

Measurements and registrations The main responses to be recorded included a weight of the feed leftovers from each box after the ten minutes presentation to the cows (leftovers) and a subjective estimate made by the observing person, of the time after offering the cafeteria a box was considered empty, if it was before the ten minutes had pass (time). The observing person was linked to a specific set of cows, so one observing person did the registrations in trial 1, 2, 3, 6, 7 and 9 whereas another observing person did the registrations in trial 4, 5 and 8. Prior to the trials the observing persons were trained together to synchronize registrations. Furthermore, some behavioral registrations were done with a computer program special designed for the purpose. The program included two set of buttons to be activated by the observing person. One button controlled three predefined behaviors (sniff, eat or none). The other button controlled which one of the four boxes (or none) the cow did one of the behaviors in. The program did the registrations on a time scale in relation to a countdown from the starting point at 600 sec. (10 minutes). For every time one of the two buttons in the program was activated by the observing person, a line was generated in a computer text file containing information on cow, behavior, box and time. The behaviors were defined as eat when the cows had visible feed intake. Sniff when the cows were near one of the concentrate boxes doing visible sniffing without feed intake. None were declared when the cows did not sniff or eat. In that way all ten registration minutes were defined by one of the behaviors. Daily feed intake of MR, weekly milk yield and milk composition, every second week’s live weight of the cows and concentrate leftovers in the adaptation period were registered as describing measurements for documenting the status and performance of the cows. Also the leftovers of MR were used to adjust amount of MR fed to the individual cow to reach ad libitum level (leftovers 2- 4 kg). The milk composition was analyzed by Steins, Eurofins laboratories (Holstebro, Denmark) and included milk fat, milk protein, and milk lactose.

30

Scientific paper

Statistical analysis Experimental data from the responses leftovers and time were analyzed with the MIXED procedures of the Statistical Analysis Systems Institute (SAS® version 9.3). The model included cow, day, concentrate, repetition, and fasting hours, including their double and triple interactions. Repeated measures were handled assuming cow x day random and including repetitions as repeated measures with an unstructured (UN) variance structure. In trial three the variance structure was handled by the autoregressive (AR 1) to obtain convergence. The degrees of freedom were estimated by the “Satterthwaite” function. To achieve variance homogeneity the data were transformed by the natural logarithm (LN) (SAS, 2010). Results reported in tables show only main effects, as no interactions were significant. Comparisons of feeds within experiments were done as multiple comparison from SAS using the treatment least squares means and standard errors of mean (SEM). Significant difference was declared when P < 0.05. The data from the behavioral registrations was not used for statistics analysis by modeling. Instead a sum of time in relation to each test concentrate and behavior was made. Also a mean time within each time a cow ate or sniffed to one of the concentrates was made. This was done in the Statistical Analysis Systems Institute (SAS® version 9.3), by the use of the “Proc Means” function (SAS, 2010). Results from the behavioral registrations include only data within the first two minutes of each trial. The two minutes was chosen as it was the time where the cows potential could empty one of the boxes. The results of the behavioral registrations were made as a sum of all four cows, all three replications and all four experimental days in each trial.

RESULTS Describing measurements The average daily milk yield and standard deviation (SD) throughout the 9 trials for all cows (cows from set one included for six weeks and cows from set two included for three weeks) was 41.1 kg (SD 3.9). The average content of milk fat was 3.7% (SD 0.3), milk protein 3.1% (SD 0.1), and milk lactose 4.9 % (SD 0.08). The average daily feed intake of the MR on DM basis and the SD are shown in Table 3.6. The results are an average of the four cows in the trials. The feed intake was divided into two periods, three days prior to the trial and the four days in the trial. For trial 1 the prior period include only one day prior to the trial.

31

Scientific paper

Table 3.6 Average feed intake of mixed ration (kg DM/day) of the cows three days (for trial 1 only one day) prior to and the four days in the trial for all 9 trials. SD in parentheses.

Trial

Cow set

Prior

Trial

1

1

23.3 (0.25)

22.0 (5.8)

2

1

21.1 (5.9)

22.0 (5.8)

3

1

21.6 (7.5)

22.0 (6.5)

4

2

22.6 (5.6)

22.9 (5.3)

5

2

23.1 (7.1)

22.6 (5.7)

6

1

23.9 (5.3)

24.4 (5.8)

7

1

20.8 (3.6)

22.0 (5.1)

8

2

22.4 (7.0)

23.5 (5.5)

9

1

21.3 (5.6)

21.3 (4.9)

As the feed intake of MR was not a primary response in the experiment, and the cows used in the different trials were not the same, it is not possible to compare between trials in Table 3.6. No major difference in feed intake between the periods prior and the trial was observed. The average live weight of the cows (average of all the eight cows, in set one and set two) at the start of the experiment was 586 kg (SD 48). After the end of trial 9 the average live weight of the eight cows had not changed as it was 589 kg (SD 44). In the adaptation period no leftovers of the adaptation mix of new concentrates were observed from any of the cows. Individual cow preferences, best and worst Table 3.7 shows the total number of times out of possible 48 times (4 feeds x 4 cows x 3 repetitions) each concentrate in a trial were chosen as the best or worst. Also the table shows the number of times out of 12 possible (4 feeds x 3 repetitions), each cow in each trial chooses a concentrate as best or worst. The best or worst was defined from leftovers with the best as the one with the least leftovers and the worst as the one with the most leftovers. Cows in the experiment showed individual differences within best or worst, for a part of the concentrates, but for others the responses of the cows were similar.

32

Table 3.7 Number of times (out of 48) each concentrate were chosen as the best or worst in a trial. Number of times (out of 12) each cow chose the concentrate as best or worst. The best or worst were defined from least and most leftovers respectively. Best or Worst out of 12

Total Best or Worst out of 48 Trial

Concentrate

1

Cow

Total Best

Total Worst

Best

21 13 5 9

4 22 13 9

5 2 1 4

1

Cow

Worst

Best

0 7 4 1

1 4 4 3

1-1 Control2 Barley, rolled Maize, pellets Barley, pellets

2 34 0 13 1

0 27 4 17

11 0 1 0

Control Dried sugar beet pulp, pellets Citrus pulp Dried sugar beet pulp, rolled

39 1 0 8

0 20 19 9

7 1 0 4

3

0 11 1 0

6 0 6 0

0 3 6 3

12 0 0 0

14 16 13 5

9 24 0 24

3 0 6 3

Control Rapeseed meal Sunflower meal Soybean meal

9 15 0 24

0 6 42 0

0 0 0 12

5

4 6 1 1

9 3 0 0

5 4 3 0

0 2 10 0

6 6 0 0

1-2

Worst

Best

0 5 4 3

6 4 0 2

0 0 1 11

9 0 3 0

0 5 6 1

12 0 0 0

0 4 2 6

8 0 3 1

0 7 2 3

8 0 0 4

0 0 12 0

0 0 0 12

2-2

0 12 0 0 4-1

3-2 0 12 0 0

0 4 4 4 4-1

3-1

9 3 0 9

Worst 4-1

3-1

2-2 0 11 0 1

Cow

3-1

2-1

1-2 Control DDGS3 I "Kalvevalse I" Sugar beet molasses

Best

2-1

1-1

4

Worst 2-1

1-1 Control Wheat, NaOH Wheat, rolled Oat, rolled

Cow

0 5 5 2 4-2

0 0 0 12

6 0 4 2

0 2 10 0

3 9 0 0

3-2

0 10 0 2 4-2 0 2 10 0

Table 3.7 continued Best or Worst out of 12

Total Best or Worst out of 48 Trial

Cow

Concentrate

Total Best

Total Worst

Best

Control DDGS II "EO"4 Rye, rolled

10 10 20 8

7 19 10 12

3 0 4 5

Control Peas, rolled Peas + SBM5, pellets "Kalvevalse II"

18 0 10 20

8 24 10 6

4 0 4 4

6

Worst

Best

1 6 3 2

1 4 7 0

4 1 3 4

6 0 0 6

1-1

7

11 3 8 26

11 26 8 3

2 0 2 8

Control Barley, Co2, rolled Dried grass pellets Wheat, Co2, rolled

13 9 15 11

8 13 16 11

4 4 4 0

9

Worst

Best

3 6 0 3

6 0 5 1

0 5 7 0

2 0 4 6

6 0 0 6

2 3 4 3

7 1 3 1

1-1

Worst

Best

0 4 3 5

0 6 4 2

1 11 0 0

6 0 2 4

0 1 2 9

1 4 4 3

2 0 0 10

2-1

3 3 4 2 4-1

3-2 0 8 4 0

Worst 4-1

3-1

2-2 0 12 0 0

Cow

3-1

2-1

1-2 Control Dried sugar beet pulp + SBM, pellets Wheat + SBM, pellets Dried grass pellets + SBM, pellets

Cow

2-1

1-1

8

1

Cow

3 7 0 2 4-2

7 2 2 1

3 2 4 3

1 3 8 0

0 4 8 0

3-1

4 4 2 2 4-1 4 3 0 5

Cow 1-1: Cow 1 in set 1, Cow 2-1: Cow 2 in set 1, Cow 3-1: Cow 3 in set 1, Cow 4-1: Cow 4 in set 1, Cow 1-2: Cow 1 in set 2, Cow 2-2: Cow 2 in set 2, Cow 3-2: Cow 3 in set 2, Cow 4-2: Cow 4 in set 2. 2 Standard robot concentrate. 3 DDGS = Dried Distillers Grains with Solubles. 4 ”EO” = Essential oils. 5 SBM = Sugar beet molasses.

Table 3.8 Means and standard error of the means (SEM) for the two recorded responses leftovers (g) and time (sec). P-values for effect of concentrate (Con.), cow, day, repetition (Rep.) and fasting hours (Fast). Leftovers Trial

Concentrate

Mean

SEM

1 1

Control Barley, rolled Maize, pellets Barley, pellets

17.0 79.6 64.8 65.9

P- values Con.

Cow

Day

Rep.

Fast.

< 0.001

< 0.001

0.002

0.33

0.18

2.6 10.0 9.6 10.2

2 13.2 114.6 45.6 86.6

0.02

0.46

0.04

208 566 359 485 < 0.001

Control Dried sugar beet pulp, pellets Citrus pulp Dried sugar beet pulp, rolled

16.3 118.9 146 89.8

3.5 8.0 2.7 10

Control DDGS2 I "Kalvevalse I" Sugar beet molasses

14.4 82.2 10.4 72.8

1.9 9.3 1.0 9.5

Control Rapeseed meal Sunflower meal Soybean meal

18.7 87.2 140.1 2.4

4.4 10.0 4.8 0.3

4

< 0.001

5

< 0.001

< 0.001

< 0.001

0.006

0.53

0.24

0.96

0.31

0.82

0.003

SEM

P- values Con.

Cow

Day

Rep.

Fast

< 0.001

< 0.001

0.005

0.90

0.22

< 0.001

0.14

0.57

0.08

0.02

< 0.001

0.004

0.51

0.004

0.40

< 0.001

0.01

0.14

0.02

0.53

< 0.001

< 0.001

0.94

0.23

0.09

21.0 27.4 24.4 29.4

0.35

2.5 8.2 8.5 9.7

3

Mean

228 443 430 412 < 0.001

Control Wheat, NaOH Wheat, rolled Oat, rolled

Time

21.0 12.9 30.2 24.4

0.34 222 577 600 457

28.8 8.6 0 27.3

163 483 196 497

23.5 23.3 19.0 21.9

197 482 591 133

22.0 23.6 6.2 8.3

0.37

0.48

Table 3.8 continued Leftovers Trial

Concentrate

Mean

SEM

Control DDGS II "EO"3 Rye, rolled

9.7 40.4 25.0 26.4

2.0 8.0 6.2 7.1

Control Peas, rolled Pea + SBM4, pellets "Kalvevalse II"

10.6 78.6 41.9 4.7

3.4 10.8 9.3 1.2

6

7

P-values Con.

Cow

Day

Rep.

Fast

0.06

< 0.001

0.63

0.01

0.89

< 0.001

8

< 0.001 Control Dried sugar beet pulp + SBM, pellets Wheat + SBM, pellets Dried grass pellets + SBM, pellets

8.7 42.7 10.5 3.8

1

Standard robot concentrate. DDGS = Dried Distillers Grains with Solubles. 3 ”EO” = Essential oils. 4 SBM = Sugar beet molasses 2

0.002

0.63

0.73

0.02

0.91

1.5 2.8 2.2 0.5

Mean

SEM

186 397 268 296

16.9 23.2 26.6 26.0

137 399 261 106

15.6 31.1 29.4 11.2

0.98

0.79 146 336 160 167

0.37 6.9 9.0 10 3.9

< 0.001

2.1 9.0 4.3 0.7

9 Control Barley, Co2, rolled Dried grass pellets Wheat, Co2, rolled

Time

0.01

0.14

0.11

Con.

Cow

Day

Rep.

Fast

< 0.001

0.002

0.06

0.17

0.80

< 0.001

< 0.001

0.05

0.50

0.56

< 0.001

< 0.001

0.09

0.95

0.19

< 0.001

0.01

0.03

0.31

0.03

11.8 25.3 13.9 12.7

0.47 125 186 225 176

P-values

10.5 18.5 15.8 9.4

Scientific paper

Concentrate leftovers and time of empty boxes Table 3.8 shows the mean and SEM for the two recorded response feed leftovers (gram) and time (seconds). Also the P value for the effect of concentrate (con), cow, day, repetition (rep.), and fasting hours (fast) in the statistical model are shown. The statistical analysis showed almost the same results for the model using leftovers as response and the one using time as response. Both responses showed effect of concentrate (except the model using leftovers as response in trial 6 and 9) meaning that within a trial the cows had different preferences for at least two of the offered concentrates. Also the models showed effect of cow (except the model having time as response in trial 2) meaning that cows within the trial responded different but as no interactions between cow and concentrate were found the cows had the same preferences. Generally (with few exceptions), no effect of day, repetition or fasting hours was detected. Ranking of the concentrates Table 3.9 and Table 3.10 show the ranking of the concentrates in relation to the two main responses feed leftovers and time. The ranking was made as difference to the control within a trial.

37

Scientific paper

Table 3.9 Ranking of the concentrates in relation to feed leftovers (gram). The values are differences to the control within a trial. Ranking 1* 2 3 4 5 6 7 8 9 10* 11* 12* 13* 14* 15* 16* 17* 18* 19* 20* 21* 22* 23* 24* 25* 26* 27* 28*

Concentrate Soybean meal “Kalvevalse II” Dried grass pellets + SBM4, pellets “Kalvevalse I” Wheat, Co2, rolled Control1 Wheat + SBM, pellets Barley, Co2, rolled Dried grass pellets “EO”3 Rye, rolled DDGS2 II Pea + SBM, pellets Wheat, rolled Dried sugar beet pulp + SBM, pellets Maize, pellets Barley, pellets Sugar beet molasses Barley, rolled DDGS I Peas, rolled Rapeseed meal Oat, rolled Dried sugar beet pulp, rolled Wheat, NaOH Dried sugar beet pulp, pellets Sunflower meal Citrus pulp

Difference to control within trial -16.3 -5.9 -4.9 -4.0 -3.0 1.8 2.1 3.1 15.3 16.7 30.7 31.3 32.4 34.0 47.8 48.9 58.4 62.6 67.8 68.0 68.5 73.4 73.5 101.4 102.6 121.4 129.7

Trial 5 7 8 4 9 All 8 9 9 6 6 6 7 2 8 1 1 4 1 4 7 5 2 3 2 3 5 3

*

Indicate significant difference to the control (P < 0.05) Standard robot concentrate. 2 DDGS = Dried Distillers Grains with Solubles. 3 ”EO” = Essential oils. 4 SBM = Sugar beet molasses. 1

Only soybean meal was significantly (P < 0.05) preferred higher than the control concentrate expressed as less leftover. “Kalvevalse I and II”, dried grass pellets + SBM and wheat, Co2, rolled showed numerical less leftover than the control. The concentrates in ranking 2-9 showed no significant difference in leftovers compared to the control concentrate. The rest of the test concentrates had more leftovers than the control.

38

Scientific paper

Table 3.10 Ranking of the concentrates in relation to time (sec). The values are differences to the control within a trial. Ranking 1* 2* 3 4 5 6 7* 8* 9* 10* 11* 12* 13* 14* 15* 16* 17* 18* 19* 20* 21* 22* 23* 24* 25* 26* 27* 28*

Concentrate Soybean meal “Kalvevalse I” “Kalvevalse II” Control1 Wheat + SBM4, pellets Dried grass pellets + SBM, pellets Wheat, Co2, rolled Barley, Co2, rolled “EO”3 Dried grass pellets Rye, rolled Pea + SBM, pellets Wheat, rolled Barley, pellets Dried sugar beet pulp + SBM, pellets Maize, pellets DDGS2 II Barley, rolled Dried sugar beet pulp, rolled Peas, rolled Oat, rolled Rapeseed meal DDGS I Sugar beet molasses Dried sugar beet pulp, pellets Wheat, NaOH Citrus pulp Sunflower meal

Difference to control within trial -64 -57 -31 14 21 51 61 82 100 110 124 151 184 190 202 211 215 235 262 277 285 320 334 355 358 378 394

Trial 5 4 7 All 8 8 9 9 6 9 6 7 2 1 8 1 6 1 3 7 2 5 4 4 3 2 3 5

*

Indicate significant difference to the control (P < 0.05) Standard robot concentrate. 2 DDGS = Dried Distillers Grains with Solubles. 3 ”EO” = Essential oils. 4 SBM = Sugar beet molasses. 1

Table 3.10 showed almost the same ranking of the concentrates as Table 3.9. Soybean meal and “Kalvevalse I” were preferred significantly higher than the control regarding time. The number of concentrates that was not different from the control was respectively seven and three when feed leftovers and time was evaluated. For the rest of the concentrates the time from offering the concentrates to the box were considered empty were longer than for the control.

39

Scientific paper

Behavioral registrations Figure 3.1 shows the time within the first two minutes of a trial the cows spent eating the test concentrates relative to the time they spent eating the control concentrate (control = 100). The test concentrates was related to the control within each trial.

180

Eating time relative to control

160 140 120 100 80 60 40 20 Citrus pulp

S.B. pulp, pellets

Wheat, NaOH

Sunflower meal

S.B. pulp, rolled

Rapeseed meal

DDGS I

Sugar beet molasses

Oat, rolled

Peas, rolled

DDGS II

Maize, pellets

Wheat, rolled

S.B.pulp + SBM

Pea + SBM

Barley, pellets

Dried grass pellets Wheat, Co2, rolled Barley, rolled Rye, rolled Barley, Co2, rolled

"EO"

Wheat + SBM

Grass pellets + SBM

"Kalvevalse I"

Control

"Kalvevalse II"

Soybean meal

0

Figure 3.1 Sum of time within the first two minutes cows spent eating each of the test concentrates relative to the control (control = 100). Control = standard robot concentrate. SBM = sugar beet molasses. “EO” = Essential oils. DDGS = dried distillers grains with solubles. S.B. pulp = Sugar beet pulp.

The cows spent relative more time eating soybean meal and “Kalvevalse II” than the control. They spent the same time eating “Kalvevalse I” as the control. Relatively less time was spent eating the rest of the test concentrates compared to the control. Figure 3.2 shows the time within the first two minutes of a trial the cows spent sniffing the test concentrates relative to the time they spent sniffing the control concentrate (control = 100). The test concentrates was related to the control within each trial.

40

Scientific paper

1600

Sniffing time ralative to control

1400 1200 1000 800 600 400 200

Wheat, rolled Wheat, NaOH Oat, rolled Sunflower meal Rapeseed meal Sugar beet molasses DDGS I S.B.pulp + SBM S.B. pulp, rolled Peas, rolled Dried grass pellets Citrus pulp Pea + SBM S.B. pulp, pellets Wheat, Co2, rolled Wheat+SBM Barley, rolled Grass pellets + SBM Barley, Co2, rolled DDGS II Soybean meal Control Rye, rolled "EO" "Kalvevalse II" Maize, pellets Barley, pellets "Kalvevalse I"

0

Figure 3.2 Sum of time within the first two minutes cows spent sniffing each of the test concentrates relative to the control (control = 100). Control = standard robot concentrate. SBM = sugar beet molasses. “EO” = Essential oils. DDGS = dried distillers grains with solubles. S.B. pulp = Sugar beet pulp.

Comparing sniff to eat the tendency were that the concentrates that the cows spent relative little time eating (Figure 3.1), they spent more time sniffing (Figure 3.2), and vice versa. Further Figure 3.2 showed that the cows were “active” in all the trials, as they sniffed to all the concentrates, within the first two minutes. Table 3.11 shows the mean eating and sniffing time per time a cow were active at a concentrate. Further the number of times a cow were active at one of the concentrates are shown (N). Soybean meal and “Kalvvevalse II” gave rise to the highest mean eating time in trial 5 and 7 respectively. In trial 4 the mean eating time did not differ between “Kalvevalse I” and the control. The control concentrate gave rise to the highest mean eating time in the remaining trials. The number of times cow were active (N) showed a relative high number in the column eat for the concentrates that were ranked high, and a relative high number in the column sniff for the concentrates that were ranked low, supporting the understanding that the cows were active in the experiment.

41

Scientific paper

Table 3.11 Mean eating and sniffing times (s) of each concentrate per time (N) a cow were active at a concentrate. The table includes data from the first two minutes of each replication in each trial Eat Trial

1

Concentrate

Sniff

Mean

N

Mean

N

1

Control1 Barley, rolled Maize, pellets Barley, pellets

46 30 21 26

39 21 18 21

31 11 4 5

3 19 13 7

2

Control Wheat, NaOH Wheat, rolled Oat, rolled

62 5 40 58

38 1 17 7

2 4 7 3

3 17 13 18

3

Control Dried sugar beet pulp, pellets Citrus pulp Dried sugar beet pulp, rolled

77 2 0 26

27 2 0 5

12 3 3 4

2 19 22 19

4

Control DDGS2 I “Kalvevalse I” Sugar beet molasses

42 15 42 16

42 20 42 11

4 5 3 7

4 24 2 20

5

Control Rapeseed meal Sunflower meal Soybean meal

37 26 14 56

40 5 1 45

2 5 4 2

3 14 16 3

6

Control DDGS II “EO”3 Rye, rolled

43 15 34 23

43 28 39 27

24 4 6 4

2 18 6 12

7

Control Peas, rolled Pea + SBM4, pellets “Kalvevalse II”

40 21 23 44

36 12 19 40

3 3 3 2

6 19 13 5

8

Control Dried sugar beet pulp + SBM, pellets Wheat + SBM pellets Dried grass pellets + SBM, pellets

39 20 30 32

41 24 38 43

2 3 5 3

1 2 2 6

9

Control Barley, Co2, rolled Dried grass pellets Wheat, Co2, rolled

48 21 33 25

45 34 30 35

2 3 3 3

5 5 10 7

Standard robot concentrate. DDGS = Dried Distillers Grains with Solubles. 3 ”EO” = Essential oils. 4 SBM = Sugar beet molasses. 2

42

Scientific paper

DISCUSSION Main responses A general effect of concentrate was seen on both recorded responses leftover and time. This showed that the cows were able to discriminate between the concentrates within a trial, meaning that the results of the trials were useful answering the project questions. No effect of concentrate on leftovers was observed in trial 9. However having time as response the effect of concentrate in trial 9 were significant, meaning that even though the cows in trial 9 ate all the concentrates offered, they ate them in a different order. Furthermore this indicate that time as response provide extra information compared to leftovers alone, especially in cases where the cows ate all the concentrates offered. Soybean meal was preferred at a higher level than the control by the cows in this experiment. The reason for that might be explained by the chemical properties of soybean meal compared to the control. The dry matter and ash content in soybean meal and the control concentrate was almost similar. Meier et al., (2012) stated that quality marks like NDF content could possibly affect feed preferences. This statement might explain why soybean meal was preferred higher than the control as the NDF content were lower in soybean meal compared to the control. It has been suggested that ruminants has some ability to regulate their feed intake in relation to their physiological requirements and by that affecting their feed preferences (Bach et al., 2012). If this is the case it could be expected that high yielding dairy cows will prefer highly digestible concentrates, which was also the case comparing soybean meal to the control as the IVOMD was higher for soybean meal. The NDF content together with IVOMD could, however, not be the only explanation why soybean meal was preferred higher than the control, as some of the other test concentrates that were ranked lower than the control (etc. maize, pellets) also had a lower NDF content and a higher IVOMD compared to the control. The crude protein content was higher in soybean meal compared to the control. More studies in the literature (Vilalba and Provenza, 1999; Bach et al., 2012) have found that the nutritional values of feeds in relation to the nutritional requirement of the animals are of great importance for feed preferences for ruminants. The cows in this current experiment were given 1 kg soybean meal per day to assure sufficient protein supply. Consequently lack of protein should not be the reason that the cows preferred soybean meal higher than other of the test concentrates. The fact that soybean meal was used as protein supply in this current experiment may have made the cows more familiar with that concentrate compared to the others. This could have affected the cows preferences as it has been shown that previous experiences with a feed has a positive effect on feed preferences later on (Distel et al.,1994; Villalba et al., 2004). This, however, should also benefit the control as this concentrate was used in all 9 trials. This, however, seemed not to be the case as the effect would have been most distinct in trial 9 because of long experience with the control, which it was not as the all concentrates tested trial 9 ranked high compared to the control. In relation to the comparison of the chemical properties of soybean meal and the control concentrate, it must also be considered that the taste of soybean meal also affected the cows’ preferences.

43

Scientific paper

The control concentrate had a content of soybean meal of 9 %. This may have affected the preferences for the control if it was the taste of soybean meal that the cows preferred, resulting in the high preferences for the control found in this experiment. In an experiment comparable to this, done on dairy heifers, Spörndly and Åsberg (2006) found soybean meal to be preferred low compared to other concentrates. The amount of concentrate offered in the experiment by Spörndly and Åsberg (2006) however was 1 kg compared to only 150 g in this experiment, which may have affected the results. Ranking The ranking of the concentrates showed that evaluating the response leftover only one concentrate (soybean meal), or evaluating the response time, two concentrates (soybean meal and “Kalvevalse I”), was found significantly preferred higher than the control. Some concentrates were found to be preferred at the same level as the control and most of the tested concentrates were found to be preferred at a lower level than the control. By the choice of the control it was not intended to choose a control that was preferred highly by the cows, instead it was expected that the control should have been of middle preferences, making the ranking more even distributed. This, however, was not the case as the control was preferred highly by the cows in this experiment. The behavioral registrations supports the ranking of the concentrates as the concentrates the cows preferred highest regarding leftovers and time was also the concentrates they choose to eat first as the behavioral registrations shows the sum of activity only after the first two minutes of each trial. The behavioral registrations regarding the behavior eat also showed that soybean meal was preferred above the control and the other test concentrates. Besides soybean meal “Kalvevalse I” was the only concentrate to be preferred significantly higher than the control when ranking in relation to the response time. However “Kalvevalse I” also were ranked numerically higher than the control in relation to leftovers as response. “Kalvevalse II” (which was a different consignment of the same product as “Kalvevalse I”) also ranked numerically higher than the control in the rankings regarding both leftovers and time. Overall it must be concluded that “Kalvevalse” were highly preferred in this experiment. “Kalvevalse” is a special designed product, suitable for small calves. In the herd of Research Center Foulum, where all the cows in this experiment were raised “Kalvevalse” is used as feed for all the small calves. This means that all the cows in this experiment had previous experience with this concentrate feed from their early life. This could explain why “Kalvevalse” was preferred high in this experiment as more findings in the literature (Distel et al., 1994; Villalba et al., 2004) has shown, that if an animal in early life get experiences with a feed it will have a high preference for that feed if it meets it again later in life. Contrary to the expectations, pure sugar beet molasses in this experiment were preferred in the lower part of the ranking for both the recorded responses. Murphy (1999) showed a positive effect on feed intake by increasing levels of molasses inclusion in a grass-molasses mixture. Baumont (1996) reported positive effects on feed intake by adding molasses to low palatable feeds, resulting in an increased intake of the molasses added feeds compared to the same feeds without molasses addi-

44

Scientific paper

tion. However Klopfer et al., (1981) made a ranking (comparable to the rankings presented in this current experiment) of 20 feeds, including concentrates, byproducts, roughages and minerals and also found pure molasses to be ranked low. The four concentrates with addition of 25 % sugar beet molasses was all ranked above pure molasses and above the pure form of the other ingredients than molasses for both feed leftovers and time. Especially wheat and dried grass pellets with the addition of molasses were preferred high, as they were both ranked equal to the control within both recorded responses. Pure dried grass pellets were in the upper part of the preference ranking. Madsen et al., (2010) compared the effect of different concentrate compositions on number of visits in a milking robot and found a significantly negative effect on milking frequency when dried grass pellets were used. It was concluded that dried grass pellets was not suitable as concentrate in a milking robot, as the cows did not like to eat it, shown by a low milking frequency. The findings in this current experiment, however, showed that the cows in this experiment had relative high preferences for the dried grass pellets tested. Rolled peas were preferred low in this experiment. Peas are known to have a high content of protein and starch (Vander et al., 2008). More studies in the literature have shown no effect on total dry matter intake by replacing soybean meal and corn grain in a MR with peas in the levels of 15 % (Vander et al., 2008) and 20 % (Petit et al., 1997). Also Klopfer et al., (1981) ranked peas in the upper part in a ranking like the one made in this experiment. The addition of sugar beet molasses and pelletizing of peas in this current experiment raised the preferences for peas, as pelletized peas with addition of sugar beet molasses ranked above rolled peas. Dried sugar beet pulp was tested in the forms of whole pellets, rolled, and in pellets with addition of sugar beet molasses. In the case of dried sugar beet pulp the addition of sugar beet molasses raised the preferences compared to the two other forms of dried sugar beet pulp tested. As the form of whole and rolled dried sugar beet pulp both was in the lower end of the ranking dried sugar beet pulp could be considered to be less palatable than some of the other concentrates tested by which the addition of sugar beet molasses raises the preferences in accordance to the earlier findings reported by Baumont (1996). In conclusion on sugar beet molasses and the addition to other concentrate feeds the finding in this experiment showed that the addition of 25 % sugar beet molasses to another concentrate feed ranked higher than pure sugar beet molasses. Findings from this experiment and the experiment by Klopfer et al., (1981), suggest that the taste of pure sugar beet molasses is too sweet for the cows to like. Mixed with other concentrate feeds it is possible to have a positive effect on preferences by addition of sugar beet molasses as all the four concentrates with addition of sugar beet molasses ranked above the same four ingredients (wheat, peas, grass pellets and dried sugar beet pulp) without addition of sugar beet molasses. Spörndly and Åsberg (2006) also found positive effect on eating rate by the addition of 30 % molasses to a barley pelletized concentrate. Besides molasses as a sweetening additive they found no positive effect on feed preferences by addition of 15-20 % of others sweetening additives. This indicates that sugar beet molasses as a sweetening additive in the

45

Scientific paper

ratio of 25-30 % has the potential to increase the preferences for a concentrate, but also there seems to be a limit for how sweet the cows prefer the feed. Rolled barley ranked in the lower part which was not expected, since it is common known that cows like to eat rolled barley. In the ranking by Klopfer et al., (1981) and the study by Spörndly and Åsberg (2006) rolled barley ranked high. In the study by Madsen et al., (2010) they found no positive effect on milking frequency when a concentrate pellet with 53 % barley was used compared to standard robot concentrate pellet, like the control in this experiment. However a positive effect on milking frequency of a concentrate pellet with a combination of 26.3 % barley and 26.3 % oat was found. In this current experiment pure rolled oat, like pure rolled barley ranked quite low, suggesting that in the findings by Madsen et al., (2010) the combination of barley and oat or maybe the pelletizing resulted in a taste that cows liked to eat. From the behavioral registrations it was observed that rolled oat in this experiment had a high mean eating time. This suggests that few (or one) of the cows liked to eat rolled oat, as they spent long time eating rolled oat when they had started to eat it. This support the findings by Atwood (2001) that different animals no matter of same obvious similarities can have different feed preferences, making it a challenge to find a concentrate feed that are preferred by a whole herd of animals. The anaerobic preservation of barley and wheat in CO2 resulted in higher feed preferences as barley CO2 and wheat CO2 ranked above “normal” rolled barley and wheat. NaOH treatment of wheat decreased the preferences compared to “normal” rolled wheat, however, the physical form of the two was different, as NaOH wheat was whole, and “normal” wheat was rolled (effect of physical form are going to be discussed later on). De Campeneere et al., (2006) found that NaOH wheat in a MR fed ad libitum, increased feed intake and milk production compared to “normal” rolled wheat in Holstein dairy cows. In the experiment by De Campeneere et al., (2006) the two forms of wheat were used in a MR, whereas in this current experiment they were tested as single feeds. The lower preferences of NaOH wheat in this experiment could be due to the NaOH treatment which was considered to add a more salty and alkaline taste to the wheat. Salt was found to be the lowest ranking feed by Klopfer et al., (1981). Also the alkaline taste could be too strong for the cows, resulting in a low preference. The behavioral registrations on the behavior sniff showed that the three concentrates rolled wheat NaOH, wheat and rolled oat had a high value relative to the control compared to the other test concentrates. Rolled wheat, NaOH wheat and rolled oat were all in the same trial (trial 2). In that trial the value for the control for sniff was low, meaning that the cows within the first two minutes of trial 2 spend little time sniffing the control resulting in a high relative value of the three tests concentrates in trial 2. Rolled rye was both in the case with leftovers and time as response in the upper part of the ranking. Limited information is available on the feeding value and feed preferences of rye as feed for lactating dairy cows (Sharma et al., 1983). Generally rye is not fed to dairy cows as it is known that rye can have a low palatability probably due to a possible high content of ergots (Sharma et al., 1981). Ergots is caused by fungus affecting cereal grains, especially rye is sensitive to infection by ergots.

46

Scientific paper

Ergots are shown to lower the palatability and also to induce toxic effects like reduced feed intake, and bad breeding results (Dinnusson et al., 1971; Sharma et al., 1981). A study by Sharma et al., (1981) showed no effect on feed intake in lactating dairy cows when up to 75 % of barley was replaced with rye in a mix of rolled cereals. Traditionally rye has not been grown in Denmark for feed purposes but for flour used for bread. Therefore the rye used in feeds has mostly been grains that have been discharged for bread (Mortensen and Madsen, 1984). The rye used in this experiment was considered to be of high quality. The results from Sharma et al., (1981) and result from this experiment suggest that rye is a feed that cows like to eat, when grains of a good quality are used. DDGS I and DDGS II were different consignments of the same product and they both ranked low. Even so difference between DDGS I and DDGS II was seen in the ranking regarding both leftovers and time. Azarfar et al., (2012) stated that DDGS often differs in nutrient content between different consignments. This, however, could not explain the different ranking between the two consignments in this experiment as the nutritional value of the two consignments were found to be similar on the analyzed parameters done in this experiment. Many studies found in the literature focus on DDGS ability as substitute for protein sources in a MR, due to the high protein content (e.g. Franke et al., 2009; Abdelqader and Oba, 2012). Also studies focusing on DDGS ability as substitute for a part of the forage in a MR, due to the high fiber content (e.g. Zhang et al., 2010) have been found. Finally, studies focusing on the ability of DDGS to substitute grains as energy source in a MR are found (e.g. Yang et al., 2012). However no studies evaluating DDGS fed as a single concentrate were found in the literature. The studies testing DDGS used in a MR however has found no (Franke et al., 2009), or positive (Zhang et al., 2010; Abdelqader and Oba, 2012) effect on dry matter intake by increasing amounts of DDGS in the ration, which do not indicate that the palatability of DDGS is low. In this current experiment DDGS were ranked low, which do not indicate a high preference for DDGS when it is fed as a single concentrate to dairy cows. The exact composition of the essential oils in the multiple ingredient concentrate “EO” is not known, thus it is difficult to compare to other findings in the literature. The fact that the concentrate “EO” was preferred lower than the control indicates that the content of essential oils in this concentrate did not improve the feed preferences compared to another commercial mixed concentrate like the control. Spörndly and Åsberg (2006) found that rapeseed meal was preferred above soybean meal in an experiment comparable to this experiment. Emanuelson (1989) (cf. Spörndly and Åsberg, 2006) found that feeding rapeseed from double low cultivars had no negative effect on feed intake at the use of 2.5 kg of total dry matter. The double low refers to a low content of erucic acid and glucosinolates witch are antinutritional factors known from rapeseeds and in high amounts are known to reduce the palatability (Kelstrup and Sepstrup, 2009). The content of eruca acid and glucosinolates of the rapeseed meal used in this experiment are not known. Brzóska (2008) found similar feed intake of concentrates containing soybean meal or rapeseed meal, making no evidence that the palatability of soybean meal should be higher than rapeseed meal, though soybean meal in this current experiment were ranked significantly higher than rapeseed meal. The lower preferences for sunflower meal compared to soybean meal and rapeseed meal may be explained by the higher NDF content in sun-

47

Scientific paper

flower meal as Meier et al., (2012) stated that quality marks like NDF content could possibly affect feed preferences. Citrus pulp was ranked low in this experiment. Other studies reviewed by Bampidis and Robinson (2006) has found that a inclusion of citrus pulp in a concentrate pellet as substitute for corn and barley grain did not affect the concentrate intake, when fed to dairy cows. Belibasakis and Tsirgogianni (1996) found no significant effect on feed intake between two mixed rations containing 20 % dried citrus pulp or none citrus pulp. However Bath et al., (1980) suggest that citrus pulp is introduced gradually into a ration for cattle to accustom to the distinctive smell and taste. Also differences in taste and smell between different citrus species are shown, with pulp originating from lemons being more accepted that pulp from oranges and grapefruits (Bath et al., 1980). It is not known which citrus species the pulp in this experiment originated from, which may have influenced the palatability and preferences. In the adaptation period prior to the trial no leftovers from citrus pulp were recorded. This is in accordance with the findings reviewed by Bampidis and Robinson (2006) showing that in a mixed concentrate pellet inclusion of citrus pulp does not affect the feed intake. However the fact that as a single concentrate citrus pulp was preferred low suggests that the taste and smell of citrus pulp is too heavy and perfumed for cows to like. Also, it is possible that the adaptation period in this experiment regarding citrus pulp was too short for the cows to accustom to a feed as distinctive as citrus pulp. The results from this experiment showed that the feed preferences for some concentrates (e.g.DDGS, NaOH treated wheat, peas, and citrus pulp) were low, though they in other studies have found that the addition of these feedstuffs to a mixed feed had no or positive effect on the feed intake. As literature on the topic feed preferences for different concentrate feeds was found to be limited a lot of the literature available has considered the concentrate feeds as ingredients in a mixed feed. The contrary findings between results from the literature and this current experiment emphasis that using a concentrate feed as ingredients in a mixed feed compared to the use in pure form may not be directly comparable like also stated by Spörndly and Åsberg, (2006). Physical form For some of the concentrates in this experiment more than one physical form was tested. Both barley and dried sugar beet pulp were tested as rolled and as pellets. Rolled dried sugar beet pulp was preferred higher compared to whole dried sugar beet pulp which was in contrast to the findings by others who have found pelletized form of concentrates to prefer to other forms (Andersen and Lykkeaa, 1962; Rodenburg et al., 2004; Spörndly and Åsberg, 2006; Meier et al., 2012). The whole dried sugar beet pulp pellets, however were large (8 mm.) and very hard, which may have influenced the cows’ preferences. The preferences for pelletized barley tended to be higher compared to the rolled form which is in agreement with the findings by others (Andersen and Lykkeaa 1962; Rodenburg et al., 2004; Spörndly and Åsberg 2006; Meier et al., 2012). The results of this experiment indicate that the positive effects on feed preferences by pelletizing of concentrates are highest when the other form than the pelletized form is fine, like rolled barley. This is in agreement with

48

Scientific paper

Meier et al., (2012) who stated that the preferences for fine and dusty feed could be improved by pelletizing. Wheat were tested in the physical forms of whole (NaOH treated), rolled and pellets (with addition of sugar beet molasses). Peas were tested in the forms of rolled and pellets (with addition of sugar beet molasses). Dried grass pellets were tested with two different pellet diameters 5 mm and 3 mm (3 mm with addition of sugar beet molasses). Also dried sugar beet pulp was tested with a reduced pellet diameter from 8 mm of whole to 3 mm (3mm with addition of sugar beet molasses). In the cases where one of the physical forms included a 3 mm pellet with addition of 25 % sugar beet molasses a positive effect on feed preferences were seen by comparing the 3 mm pellet with addition of molasses to whole or rolled form. In this experimental setup it is not possible to conclude whether the positive effect on feed preferences were caused by the addition of sugar beet molasses or by the pelletizing in 3 mm pellets, or an interaction between pelletizing and sugar beet molasses addition. Also for wheat it is not possible to conclude whether higher preferences for rolled compared to whole were linked to the physical form or the NaOH treatment of the whole form. Evaluating the concentrates that was ranked above or similar to the control regarding both the responses leftovers and time no clear relation between preferences and physical form was observed. The physical form of the highest ranked concentrates counts for meal, pellets and rolled. Rodenburg et al., (2004) found that the strength of concentrate pellets was of importance for the number of visits in the milking robot, with weak and crumble pellets associated with a fewer visits. Also Spörndly and Åsberg (2006) found that pellets were preferred above ground concentrates. This however, could not be shown from the results of this current experiment. However, it is noticed that the content of the highest ranked concentrates were different, thus it is difficult to compare them directly. The different physical form and pellet diameter of the concentrates used in this experiment may have affected the response feed leftovers. Spörndly and Åsberg (2006) stated that a larger particle size may raise the eating rate. This could also be the case in this current experiment, as it is possible that more of a concentrate with larger particle size could have been eaten within one mouthful. This possible effect of particle size was expected to have been most pronounced for concentrates with a low preference, which the cows eat only few mouthfuls from, as it were expected that the cows continued to eat the concentrates that they preferred high. From the physical properties of the concentrates in this experiment it was not possible to conclude which physical form that determined the cows feed preferences, as the contents of the highest ranked concentrates were different. Though, it must be considered that the property of taste maybe more important for feed preferences than the physical form of the concentrate. Individual cow preferences Reported as best or worst the four cows in each trial showed some individual preferences for the tested concentrates. The individual cow preferences were more distinct for some concentrates, whereas for other concentrates, the four cows in the trial showed the same preferences. One exam-

49

Scientific paper

ple with cows expressing different individual preferences for one concentrate is DDGS I in trial 4. DDGS I was in total chosen by the four cows as being the best 16 times and worst 24 times out of 48 possible. Considering the individual cows DDGS I was chosen by one cow as the best 12 times and worst 0 times, whereas for two other cows DDGS I was chosen as the worst respectively 10 and 11 times, and the best 0 times for both the two. Also Spörndly and Åsberg (2006) found individual cow preferences and eating rates. They found also different preferences and eating rates within the same animal depending on whether the feeds were offered in the morning or in the evening. This experiment was planned to handle the known factors of importance for planning a preference experiment stated by Meier et al., (2012), including similar environmental and housing conditions. The cows in the experiment were of the same breed, and more or less at the same physiological state, factors stated by Villalba and Provenza (1999) to be of importance for animals’ feed preferences. The cows in the experiment had the same adaptation period prior to the experiment, where they were given a mixture of test concentrates to get experienced with the smell and taste of the different test concentrates. In the adaptation period no leftovers of the test concentrate mix were reported for any of the cows in any of the 9 trials. This means that all cows had eaten all test concentrates in the adaptation period indicating that the cows were familiar with the feeds when the trial started. Other things stated by Meier et al., (2012) to affect the outcome of a preference experiment is the time of exposure (fasting hours) and placement of the feed (placement of boxes in the cafeteria). These factors were handled in the experiment by the rotation of the feeds in the cafeteria and the cows in relation to fasting hours in a Latin square. In a full trial all cows starved all hours (1-4) and were presented for each concentrate at all four possible places in the cafeteria. Even though the known major factors to affect an animal’s preferences in a preference experiment were taken into account in the experimental setup still some individual cow preferences were observed. Other experiments (Atwood et al., 2001) have shown that at uniform experimental conditions and by using animals at the same physiological state individual animals have different requirements for nutrients. As feed preferences are linked to the animal’s requirements for nutrients the different requirements will result in different feed preferences. That animals’ feed preferences depend on their nutrient requirements are well known (Villalba and Provenza, 1999; Atwood et al., 2001). This is also observed from wild living herbivores which select their diet from an array of available plants that differ in physical and chemical characteristics (Shipley 1999). This has resulted in the hypothesis by Villalba and Provenza (1999) that feed preferences “Originates from the interrelationship between taste, and postingestive feedback”. This hypothesis could also explain why some of the test concentrates in the experiment (for example citrus pulp) consequently never was characterized as the best simply if the cows did not like the taste of it. In this experiment the animals’ previous experiences with the concentrate feeds earlier in life was not controlled. Distel et al., (1994) and Villalba et al., (2004) found that if animals got experiences with specific feedstuffs in early life, especially around sensitive periods, they would recognize this feedstuff later in life. The cows in this experiment came from the production herd at the Research Center Foulum, Aarhus University. They were all born in the herd, within a period of one year. It is not known if they all have had the same experiences with feedstuffs in life before the experiment,

50

Scientific paper

but the fact that they were all born and raised in the same herd within a period of one year minimizes the bias from earlier experiences. Evaluating the individual feed preferences from the factor of best or worst discriminate very sharply between the concentrates. The best and worst were made evaluating the differences of feed leftovers, which means that some of the leftovers were separated by only 0.1 gram. This small difference could also be because of experimental error instead of real difference in leftovers. However, this experiment together with other experiments (Atwood et al., 2001; Spörndly and Åsberg, 2006) has shown that individual cows may have individual feed preferences, which should be considered when results of feed preferences using the mean of more animals are evaluated. On the other hand if the outcome of this experiment should be scaled up to production level these individual cow differences will also occur in a production herd. By so it seems reasonable to use the mean of more animals’ feed preferences to rank the feeds. Finally it should be considered that the number of cows in each trial in this experiment was only four. Meier et al., (2012) summarized that the number of animals used in preference experiments is often low (5-8), due to the high resource demand in such experiments. It must though be taken into account that with some individual animal preferences a number of four animals are low to conclude globally. Method The describing measurements milk yield, milk composition, feed intake of MR, and live weight of cows seemed not to be affected by the trials and the different offer of concentrates. Also the SD of the describing measurements showed no differences between individual cows. Even though the describing measurements not were evaluated with statistic calculations, as they were not parameters in question, the reported results on the parameters suggests that the cows in the experiment were not affected on their physiological status or performance. The feed intake of MR was the same prior to and in the experimental days which show that cows had sufficient time to eat the amount of MR they needed in experimental days, even though they did not get the MR until after the daily registrations ended at 12:30 PM. An effect of test concentrates intake on cow performance and MR intake was not intended as the purpose of the experiment was to find concentrates that cows preferred in relation to taste and palatability, when a MR of high energy and good quality was offered at ad libitum amounts. The project questions can therefore be answered without any reservation regarding physiological status or performance of the cows, as these were not affected. The overall effect of the tested parameters in the models having leftovers and time as responses was the same, meaning that both responses were sensitive to the experimental setup. When considering the interrelationship between the two responses it seems reasonable that if a box was considered empty after a short period of time after offering, then the feed leftovers was also low. Opposite if the feed leftovers were high, then the box should not have been considered empty. Since time was subjective estimated by the observing person, the feed leftovers could be considered as a more valid response, as this were determined by weighing the residues after the ten minutes. The analyses of the models using the two responses however showed that the response time in this experiment were a response just as valid as the response leftovers. If the leftovers after ten minutes were low for all

51

Scientific paper

test concentrates the response time may provide more information than the leftovers. That two different observing persons were involved in the experiment did not seem to affect the outcome as time and feed leftovers both were sensitive in all the trials no matter of which of the two observing persons that were involved. This indicates that the training of the observing persons prior to the experiment were sufficient to synchronize the registrations. Spörndly and Åsberg (2006) used eating rate to determine preferences for different concentrate feeds. They used the eating rate during the first three minutes and total eating rate until the cows had eaten all the feed or ten registrations minutes had passed. The eating rate during the first three minutes was measured by weighing the feed left after three minutes, whereas the total eating rate were calculated as feed left in relation to the time that had passed. This means that if the cows had eaten all the feed before the ten registration minutes had passed, the eating rate was made on basis of an estimate (of the observing person), of when the feed was eaten. Spörndly and Åsberg (2006) found that eating rate during the first three minutes by weighing and total eating rate by estimate of the observing person were correlated, making no difference which of the two that were used. No general significant effects of day, repetition, or fasting hours were found in the trials in this experiment. In relation to the experimental setup it was not indented to see an effect of these parameters. That there was no effect of day indicates that the adaptation period in the experiment were long enough for the cows to get familiar with the concentrates. Otherwise it could be that the preferences for one concentrate increased in relation to days in the trial as the cows got more familiar with it. The length of adaptation period used in other preference experiments has been found varying from one week to more weeks in relation to feeds tested and the purpose of experiment (Meier et al., 2012). It was expected to see effect of fasting hours guiding the cows to be less selective, as fasting hours increased. A fasting time of four hours seems high for a high yielding dairy cow that eats 4-6 hours per day (Munksgaard et al., 2005). The behavioral registrations showed that the cows in the experiment were “active” as they spend time sniffing all the concentrates, also those they did not eat. Also the behavioral registrations showed that the cows quickly found the concentrates they liked to eat. This all shows that the experimental setup worked as the cows “checked” all the boxes and were able to distinguish between the offered concentrates and choose those they liked to eat. The general interpretation of the result in this experiment was that the experimental setup worked in relation to answer the project question. Especially that there were significant effect of concentrate, means that the cows had different preferences for the different concentrates offered.

CONCLUSIONS The results showed that soybean meal was the only of the tested concentrates to be preferred significantly higher than the control concentrate regarding both of the recorded responses leftovers and time. Few concentrates were preferred at the same level as the control concentrate whereas most of the tested concentrates were preferred at a lower level than the control concentrate. The preferences

52

Scientific paper

for the control concentrate in the experiment was high making the ranking of the test concentrates in relation to the control uneven distributed. From this experiment it was not possible to find a general correlation between the physical form of the concentrate and the preferences, as the concentrates the cows preferred highest counted both meals, pellets and rolled. However, comparing a pellet form of barley to a rolled form of barley a tendency was found that the pelletized form was preferred above the rolled form. The addition of 25 % sugar beet molasses as a flavoring agent raised the preferences for wheat, peas, dried grass pellets and dried sugar beet pulp compared to the pure form of these. The combination of molasses with another concentrate was also preferred higher than pure sugar beet molasses. The combination was fed as 3 mm pellets, by which the experiment does not make it possible to determine whether the higher preferences were because of the physical form, or the addition of sugar beet molasses, or an interaction between physical form and sugar beet molasses addition. The results however indicate that it is possible to increase the preference for a concentrate feed by addition of sugar beet molasses as a flavoring agent. From the results achieved in the experiment it is concluded that cows are able to distinguish between different concentrate feeds offered at the same time and that they have preference for some concentrate feeds compared to others.

REFERENCES Abdelqader, M.M., Oba, M. 2012. Lactation performance of dairy cows fed increasing concentrations of wheat dried distillers grains with solubles. J. Dairy Sci. 95: 3894-3904. Andersen, P.E., Lykkeaa, J. 1962. Almindeligt kraftfoder contra kraftfoderpiller. Bilag til landøkonomisk forsøgslaboratorium efterårsmåde 37-41. AOAC 1990. Official Methods of Analysis, 15th edition, AOAC International, Gaithersburg, MC, USA. Atwood, S.B., Provenza, F.D., Wiedmeier, R.D., Banner, R.,E. 2001. Influence of free-choice vs. mixedration on food intake and performance of fattening calves. J. Anim. Sci. 79: 3034-3040. Azarfar, A., Jonker, A., Hettiarachchi, I.k., Yu, P. 2012. Nutrient profile and availability of co-products from bioethanol processing. J. of Animal Physiology and Animal Nutrition. 96: 450-458. Bach, A., Villalba, J.J, Ipharraguerre, I.R. 2012. Interactions between mild nutritional imbalance and taste preferences in young ruminants. J. Anim. Sci. 90: 1015-1025. Bampidis, V.A., Robinson, P.H. 2006. Citrus by-products as ruminant feeds. Animal Feed Science Technology. 128: 175-217. Bath, D.L., Dunbar, J.R., King, J.M., Berry, S.L., Leonard, R.O., Olbrich, S.E. 1980. By-products and unusual feedstuffs in livestock rations. Western Regional Extension Publication No. 39 USDA-ARS, Washington, DC, USA.

53

Scientific paper

Baumont, R. 1996. Palatability and feeding behavior in ruminants. Ann Zootech. 45: 385-400. Belibasakis, N.G., Tsirgogianni, D. 1996. Effect of dried citrus pulp on milk yield, milk composition and blood components of dairy cows. Animal Feed Science Technology. 60: 87-92. Brzóska, F. 2008. Milk Production and Composition as Influenced by Soybean Meal, Rapeseed Meal, or Rapeseed Cake in Concentrates for Dairy Cows. Ann. Anim. Sci. Vol. 8 No. 2: 133-143. De Campeneere, S., De Boever, J.L., De Brabander, D.L. 2006. Comparison of rolled, NaOH treated and ensiled wheat grain in dairy cattle diets. Livestock Science. 99: 267-276. Dinnusson, W.E., Haugse, C.N., Knutson, R.D. 1971. Ergots in Rations for Fattening Cattle Farm Research November-December: 20-22. Distel, R.,A., Villalba, J.,J., Laborde, H.,E. 1994. Effects of early experience on voluntary intake of lowquality roughage by sheep. J. Anim. Sci. 72: 1191-1195. Emanuelson, M. 1989. Rapeseed products of doublet low cultivars to dairy cows. Effects of long term feeding and studies on rumen metabolism PhD Thesis, Report 189, Dept. Anim. Nur. Management, Swedish Univ. Agric. Sci., Uppsala, Sweden. Erdman, A.A., Varner, M. 1995. Fixed yield response to increased milking frequency. J. Dairy Sci. 78: 11991203. Franke, K., Meyer, U., Flachowsky, G. 2009. Distillers dried grains with soluble compared with rapeseed meal in ration for dairy cows. J. of Animal and Feed Science. 18: 601-612. Halachmi, I., Ofir,S., Miron, J. 2005. Comparing two concentrate allowances in an automatic milking system. J. Anim. Sci. 80: 339-343. Halachmi, I., Shoshani, E., Solomon, R., Maltz, E., Miron, J. 2009. Feeding soy hulls to high-yielding dairy cows increased milk production, but not milking frequency in an automatic milking system. J. Dairy Sci. 92: 2317-2325. Hansen, B. 1989. Determination of nitrogen as elementary N, an alternative to Kjeldahl. Acta Agric Scand. 39: 113-118. Hermans, G.G.N., Ipema, A.H., Stefanowska, J., Mertz, J.H.M. 2003. The effect of two traffic situations on the behavior and performance of cows in an automatic milking system. J. Dairy Sci. 86: 1997-2003. Kelstrup, L., Sepstrup, P. 2009. Korn, raps og ært 2th edition 2009 Landbrugsforlaget:101-128. Klopfer, F.D., Kilgour, R., Matthews, L.R. 1981. Paired comparison analysis of palatabilities of twenty foods to dairy cows. Proceedings of the N.Z. Society of Animal Production. 41: 242-247. Madsen, J., Weisbjerg, M., R., Hvelplund, T. 2010. Concentrate composition for automatic milking systems – Effect on milking frequency. Livestock Sci. 127: 45-50.

54

Scientific paper

Meier, J.S., Kreuzer, M., Marquardt, S. 2012. Design and methodology of choice feeding experiments with ruminant livestock. Applied animal behavior science. 149: 105-120. Mertens, D. R. 2002. Gravimetric determination of amylase-treated neutral detergent fiber in feeds with refluxing in breakers or crucibles: Collaborative study. L. AOAC Int. 85: 1217-1240. Migliorati, L., Speroni, M., Stelletta, C., Pirlo, G. 2009. Influence of feeding flavoring-appetizing substances on activity of cows in an automatic milking system. Ital. J. Anim. Sci. 8: 417-419. Mortensen, H.P., Madsen, A. 1984. Stigende mængder rug til slagtesvin. Statens husdyrbrugsforsøg, meddelelse nr. 534. Munksgaard, L., Jensen, M.B., Pedersen, L.J., Hansen, S.W., Matthews, L. 2005. Quantifying behavioral priorities-effects of time constraints on behavior of dairy cows, Bos Taurus. Applied Animal Behavior Science. 92: 3-14. Murphy, J.J. 1999. The effect of increasing the proportion of molasses in the diet of milking dairy cows on milk production and composition. Animal Feed Science Technology. 78: 189-198. Norfor, 2013. Nordic feed Evaluation System. Petit, H.V., Riox, R., Ouellet, D.R. 1997. Milk Production and Intake of Lactating Cows Fed Raw or Extruded Peas. . J. Dairy Sci. 80: 3377-3385. Rodenburg, J. 2011. Designing Feeding Systems for Robotic Milking Tri-State Dairy Nutrition Conference (19-20 April), 2011: 127-138. Rodenburg, J., Focker, E., Hand , K. 2004. Effect of the composition of concentrate fed in the milking box, on milking frequency and voluntary attendance in automatic milking systems. In: Meijering, et al., (Ed.), Automatic Milking - A Better Understanding. Wagening Academic Press, pp. 511–512. SAS Institute. 2010. Statistical Analysis System. User’s guide. SAS Institute Inc., Cary, NC. Sharma, H.R., Ingalls, J.R., Mckirdy, J.A., Sanford, L.M. 1981. Evaluation of Rye Grain in the Diets of Young Holstein Calves and Lactating Dairy Cows. J. Dairy Sci. 64: 441-448. Sharma, H.R., Ingalls, J.R., Mckirdy, J.A. 1983. Feeding value of Alkali-Treated whole Rye Grain for Lactating Cows and its Digestibility for Sheep. Animal Feed Science and Technology. 10: 77-82. Shipley, L.A., 1999. Grazers and Browsers: How Digestive Morphology Affects Diet Selection. In Launchbaugh, K.L., Sanders, K.D., Moesley, J.C. (Ed). Grazing Behavior and Wildlife: 20-27. Spörndly, E., Åsberg, T. 2006. Eating Rate and Preference of Different Concentrate Components for Cattle. J. Dairy Sci. 89: 2188-2199. Stoldt, W 1952. Vorschlag zur Vereinheitlichung der Fettbestimmung in Lebensmitteln. Fette Seifen 54: 206-207. Vander, M., Pol, A.N., Zaman, S., Delano, N. 2008. Peas Can Replace Soybean Meal and Corn Grain in Dairy Cows Diets. J. Dairy Sci. 91: 698-703.

55

Scientific paper

Villalba, J.J., Provenza, F.D. 1999. Effects of food structure and nutritional quality and animal nutritional state on intake behavior and food preferences of sheep. Applied animal behavior science. 63: 145-163. Villalba, J.,J., Provenza, F.,D., Han, G. 2004. Experience influences diet mixing by herbivores: implications for plant biochemical diversity. OIKOS 107: 100-104. Weisbjerg, M.R., Hvelplund, T., 1993. Estimation of net energy content (FU) in feeds for cattle. In: National Institute of Animal Science, Report No 3/1993, 39 (in Danish). Weisbjerg, M.R., Munksgaard, L. 2009. Effect of concentrate allowance in an automatic milking system. In: Book of Abstracts no. 15 of the 60th Annual Meeting of the European Association for Animal Production (EAAP), Barcelona, Spain, p. 349. Yang, W.Z., Li, Y.L., McAllister, T.A., McKinnon, J.J., Beauchemin, K.A. 2012. Wheat distillers grain in feedlot cattle diets: Feeding behavior, growth performance, carcass characteristics and blood metabolites. J. Anim. Sci. 90: 1301-1310. Zhang, S.Z., Penner, G.B., Yang, W.Z., Oba, M. 2010. Effects of partly replacing barley silage or barley grain with distillers grains with soluble on rumen fermentation and milk production on lactating dairy cows. J. Dairy Sci. 93: 3231-3242.

56

4 Popular scientific paper Malkekøer har præferencer for forskellige kraftfodermidler Af Lasse Primdal Specialestuderende ved Institut for Husdyrvidenskab, Faculty of Science and Technology, Aarhus Universitet

Nyt forsøg fra Forskningscenter Foulum, Aarhus Universitet, viser at malkekøer foretrækker sojaskrå blandt 26 forskellige testede kraftfodermidler. Indledning For at opnå optimal mælkeydelse og udnyttelses af robotkapaciteten er den ønskede malkefrekvens i AMS på 2,5-3,5 malkninger pr. ko pr. dag. Mange producenter opnår dog ikke den ønskede malkefrekvens, da køerne ikke frivilligt opsøger malkerobotten regelmæssigt, hvilket betyder et øget arbejdsforbrug til at hente køer samt lav malkefrekvens og lav mælkeydelse. Malketrangen alene er ofte ikke nok for at opnå den ønskede malkefrekvens, hvorved brugen af kraftfoder som lokkemiddel i malkerobotten er almindelig praksis. En udbredt metode til at lokke køerne ind i malkerobotten er at tilbyde en høj andel af det daglige energi- eller proteinbehov som kraftfoder i malkerobotten. Dog øges risikoen for udvikling af metaboliske sygdomme ved at tildele en stor mængde kraftfoder på kort tid. Desuden vil et højt kraftfoderforbrug have en negativ indvirkning på foderomkostningerne. Derfor er der behov for en øget viden omkring malkekøers præferencer for forskellige kraftfodermidler med henblik på at finde kraftfodermidler, som er så attraktive for malkekøer, at de kan anvendes som lokkemiddel i malkerobotten i små mængder. Da tidligere studier har vist, at kraftfoderets fysiske form og tilsætning af melasse som sødemiddel påvirker køers præferencer, indgik forskellige fysiske former af kraftfodermidler samt kraftfodermidler med tilsætning af sukkerroemelasse i denne undersøgelse. Formålet med dette forsøg var således at undersøge, hvilke kraftfodermidler malkekøer foretrækker, og om kraftfoderets fysiske form samt tilsætning af sukkerroemelasse kan påvirke køernes præferencer. Forsøgsbeskrivelse Forsøget blev udført som en række intensive præferenceforsøg, hvor fire malkekøer skulle vælge mellem fire forskellige kraftfodermidler, som blev tilbudt samtidig. Det ene af disse kraftfodermidler var en kontrol, som indgik i alle delforsøgene, for at kunne sammenligne de testede kraftfodermidler. Kontrolkraftfoderet var et standard robotkraftfoder, som til daglig anvendes i malkerobotten på Kvægbrugets Forsøgscenter (KFC). I alt blev præferencen for 26 forskellige kraftfodermidler undersøgt, hvilket gjorde, at det var nødvendigt at foretage 9 præferenceforsøg. Der blev brugt to 57

Popular scientific paper

grupper af køer, som i forsøget var opstaldet i bindestald på Forskningscenter Foulum, hvor forsøget blev udført i foråret 2013. Køerne blev tilbudt 150 g af hvert af fire kraftfodermidler (tre test kraftfodermidler og kontrollen) i kasser og kunne frit vælge, hvilke fodermidler de foretrak (se Figur 4.1). Hver af de fire køer fik i hvert præferenceforsøg dagligt tilbudt fodermidlerne i 10 min, tre gange umiddelbart efter hinanden. Et helt præferenceforsøg bestod af fire dages registreringer, hvor placeringen af de enkelte fodermidler foran koen var forskellig fra dag til dag. Bås

Foderkasse A B C D

Figur 4.1 Forsøgsopsætning, hvor de fire kasser (A-D) indeholdt et af de fire fodermidler som blev undersøgt i hvert præferenceforsøg. Placeringen af kasserne var forskellig mellem dagene i forsøget.

Køerne blev fodret ad libitum med fuldfoder med et energiindhold på 6,6 MJ NEL/kg tørstof (bestemt ved NorFor) og malket to gange dagligt. Ved aftenmalkningen alle ugens dage (både de 4 registreringsdage og de tre mellemliggende dage) blev køerne tildelt 1 kg sojaskrå for at sikre tilstrækkelig proteinforsyning. I de tre dage mellem hvert præferenceforsøg blev køerne tilvænnet den kommende uges kraftfodermidler, ved at de blev tildelt en blanding bestående af de tre nye kraftfodermidler samt kontrolkraftfodret. Alle kraftfodermidler blev analyseret for indhold af tørstof, aske, råprotein, NDF, råfedt og enzymatisk fordøjelighed af organisk stof (EFOS). Resultaterne af analyserne samt rækkefølgen af præferenceforsøgene kan ses i Tabel 4.1. Mængden af foderester efter 10 minutter (foderrest) og tiden fra tildeling til alt foderet var ædt (tid), hvis dette skete inden de 10 minutter var gået, blev brugt til at vise hvilke fodermidler, der var mest foretrukket. Derudover blev der registeret hvor lang tid den enkelte ko brugte på at æde og snuse til hvert enkelt kraftfodermiddel (adfærdsregistreringer). Det bemærkes, at kraftfodermidlerne Kalvevalse og bærme begge blev testet to gange. Dette inkluderede fodermidler fra forskellige partier, samt at præferencerne blev undersøgt på to forskellige grupper af køer, hvorfor I og II for begge var sande gentagelser. Fodermidlet ”ÆO” var et kommercielt kraftfodermiddel med et indhold af æteriske olier (ÆO) på 0,02 %. Fodermidlerne med sukkerroemelasse var tilsat 25 % sukkerroemelasse og var pelleteret uden brug af damp. Byg og majs var også pelleteret uden brug af damp. CO2 i delforsøg 9 betyder, at kornet er høstet med et vandindhold på ca. 20 % og lageret under anaerobe forhold i gastætte siloer. Resultater og diskussion Kun udvalgte resultater fra det samlede forsøg vises i denne artikel. Beskrivende registreringer omfattende mælkeydelse og mælkens sammensætning, foderoptag og levende vægt af køerne viste, at køernes produktion og fysiologiske status ikke blev påvirket, da disse beskrivende registreringer stort set var uændrede gennem hele forsøget. Den gennemsnitlige daglige mælkeydelse gennem forsøget for de involverede køer var 41,1 kg mælk.

58

Popular scientific paper

Tabel 4.1 Oversigt over delforsøg, fodermidler og kemisk indhold af tørstof (TS), aske, råprotein (RP), NDF, råfedt (RF) og enzymatisk fordøjelighed af organisk stof (EFOS) i fodermidlerne. % af TS Fysisk form1

TS %

Aske

RP

NDF

RF

EFOS

Kontrol2

Piller, 4 mm

87,2

6,5

19,9

24,7

3,7

90,1

Byg

Valset

88,9

2,2

9,8

20,7

3,2

90,8

Majs

Piller, 3mm

88,1

1,4

10,3

8,4

4,6

98,1

Byg

Piller, 3mm

87,5

2,0

10,5

16,9

3,2

92,5

Hvede, NaOH

Hel

80,5

5,9

11,1

7,8

2,1

98,7

Hvede

Valset

87,4

1,5

11,6

12,5

2,6

97,7

Havre

Valset

88,3

2,5

10,5

28,8

5,7

78,4

Roepiller

Piller, 8mm

90,6

4,3

10,5

40,8

1,3

95,1

Citruskvas

Piller, knust

91,2

8,2

7,8

24,1

2,8

96,8

Roepiller

Valset

91,3

4,3

10,3

40,0

0,6

94,9

Bærme I

Piller, 5mm

89,9

5,8

35,1

24,3

6,6

92,1

Kalvevalse I

Valset + piller, 4mm

86,3

7,9

22,9

17,0

3,4

94,7

Sukkerroemelasse

Flydende

68,0

10,2

11,4

-

0,2

-

Rapsskrå

Skrå

88,2

7,3

39,9

30,0

3,6

82,2

Solsikkeskrå

Skrå

90,2

6,7

40,7

32,7

3,0

77,8

Sojaskrå

Skrå

88,5

7,2

54,3

8,0

2,9

99,6

Bærme II

Piller, 5 mm

90,4

5,6

34,9

25,9

6,8

90,9

Piller, 4 mm

88,9

6,5

28,7

24,1

7,0

90,9

Valset

86,5

1,9

8,1

17,1

1,9

97,3

Valset

85,7

3,2

23,9

14,0

2,1

99,6

Piller, 3mm

82,1

4,8

20,8

10,6

1,8

99,1

Kalvevalse II

Valset + piller, 4mm

86,2

8,7

24,6

17,2

3,6

93,5

Roepiller + SRM

Piller, 3mm

89,8

6,2

11,1

33,3

0,6

94,2

Hvede + SRM

Piller, 3mm

83,7

3,5

11,8

9,9

2,0

97,3

Grøntpiller + SRM

Piller, 3mm

88,7

10,3

17,3

40,1

2,6

80,7

Byg, Co2

Valset

80,9

2,1

10,9

20,7

2,9

92,2

Grøntpiller

Piller, 5 mm

90,7

10,2

18,6

48,4

3,2

76,0

Hvede, Co2

Valset

81,9

2,1

13,7

13,5

2,6

95,8

Delforsøg Fodermiddel Alle 1

2

3

4

5

6

"ÆO"

3

Rug 7

Ærter Ærter + SRM

8

9

4

1

For piller er diameteren givet. Standard robotkraftfoder. 3 ”ÆO” = æteriske olier. 4 SRM = sukkerroemelasse. 2

59

Popular scientific paper

En statistisk behandling af resultaterne viste en effekt (P < 0,05) af kraftfoder i hvert delforsøg, hvilket betød, at køerne havde forskellig præference for de forskellige kraftfodermidler. Registreringer af foderrest og tid blev brugt til at rangere fodermidlerne i forhold til, hvordan de blev foretrukket relativt til kontrollen. Begge responser viste stort set den samme rangering, hvorfor kun rangeringen ud fra foderrest er vist i denne artikel (Tabel 4.2). Adfærdsregistreringerne understøttede rangeringen i Tabel 4.2.

Tabel 4.2 Rangering af fodermidlerne ud fra foderrest (gram). Rangeringen er lavet som difference til kontrollen inden for forsøget. Rangering

Fodermiddel

1* Sojaskrå 2 Kalvevalse II 3 Grøntpiller + SRM1, piller 4 Kalvevalse I 5 Hvede, Co2, valset 6 Kontrol2 7 Hvede + SRM, piller 8 Byg, Co2, valset 9 Grøntpiller 10* “ÆO”3 11* Rug, valset 12* Bærme II 13* Ærter + SRM, piller 14* Hvede, valset 15* Roepiller + SRM, piller 16* Majs, piller 17* Byg, piller 18* Sukkerroemelasse 19* Byg, valset 20* Bærme I 21* Ærter, valset 22* Rapsskrå 23* Havre, valset 24* Roepiller, valset 25* Hvede, NaOH 26* Roepiller, piller 27* Solsikkeskrå 28* Citruskvas * Viser signifikant forskel fra kontrollen (P < 0.05). 1 SRM = sukkerroemelasse. 2 Standard robotkraftfoder. 3 ”ÆO” = æteriske olier.

Forskel til kontrol indenfor delforsøg -16.3 -5.9 -4.9 -4.0 -3.0 1.8 2.1 3.1 15.3 16.7 30.7 31.3 32.4 34.0 47.8 48.9 58.4 62.6 67.8 68.0 68.5 73.4 73.5 101.4 102.6 121.4 129.7

Delforsøg 5 7 8 4 9 Alle 8 9 9 6 6 6 7 2 8 1 1 4 1 4 7 5 2 3 2 3 5 3

60

Popular scientific paper

Rangeringen viste, at køerne kun havde signifikant højere præferencer for sojaskrå sammenlignet med kontrollen. Køerne havde samme præferencer for syv fodermidler som for kontrollen. For de resterende fodremidler var køernes præferencer lavere end for kontrollen. Dette resultat var overraskende, da det var forventet, at flere af de testede fodermidler ville rangere højt sammenlignet med kontrollen. En lavere præference for kontrollen ville have fordelt test fodermidlerne mere jævnt omkring kontrollen. At sojaskrå var det mest foretrukne af de testede fodermidler kan skyldes flere faktorer bl.a. det lavere NDF indhold, højere proteinindhold og den højere fordøjelighed sammenlignet med kontrollen. Det er dog også muligt, at det var smagen af sojaskrå, som tiltalte køerne. Kalvevalse rangerede forholdsvis højt i dette forsøg, hvilket kan skyldes, at køerne i dette forsøg som kalve har været fodret med Kalvevalse. Flere tidligere studier har fundet, at dyr, der tidligt i livet tildeles et fodermiddel, vil genkende dette fodermiddel, hvis de præsenteres for det senere i livet, og have en høj præference for fodermiddelet. Hvis dette er tilfældet, kan landmænd med AMS ved fodring af deres småkalve påvirke præferencerne for det kommende robotkraftfoder og derved muligvis påvirke malkefrekvensen. Yderligere undersøgelser er dog nødvendige for at fastslå denne effekt. Grøntpiller rangerede højt i dette forsøg, hvilket viser, at køerne havde høje præferencer for de her afprøvede grøntpiller. Dette er i modsætning til konklusionerne fundet af et tidligere forsøg, som viste, at grøntpiller var uegnet som kraftfodermiddel i en malkerobot, vist ved en lav malkefrekvens. Ud fra rangeringen ses det, at lagringen af hvede og byg i gastæt silo øgede præferencerne for hvede og byg, da valset hvede CO2 og valset byg CO2 begge rangerede over almindeligt valset hvede og byg. Dette skyldes givet den syrlige lugt og smag, samt det højere vandindhold i det CO2 lagrede hvede og byg. Tidligere studier har vist, at køer foretrækker pelleteret kraftfodermidler frem for kraftfodermidler i formalet form, samt at præferencerne for et fint og støvet fodermiddel kan forbedres ved pelletering. Ud fra præferencerangeringen var det ikke muligt at bestemme, hvilken betydning fysisk form havde for præferencerne, da de fodermidler, som blev rangeret over eller på samme niveau som kontrollen, inkluderede de fysiske former skrå, piller (forskellige diametre) og valset, hvilket antyder, at smagen er af større betydning for køers præferencer end den fysiske form. Tilsætningen af 25 % sukkerroemelasse øgede præferencen for alle de fodermidler, det blev testet på (hvede, grøntpiller, ærter og roepiller). Det må dog bemærkes, at de melassetilsatte fodermidler var pelleteret i 3 mm piller, hvilket for byg viste en tendens til at øge præferencerne sammenlignet med en valset form. Det er derfor ikke muligt at konkludere, om den øget præference for de melassetilsatte fodermidler skyldes pelletering, melassetilsætning eller en vekselvirkning mellem disse. Dog har et tidligere studie ligeledes fundet, at tilsætningen af 30 % melasse til valset byg øgede præferencerne sammenlignet med valset byg uden melassetilsætning, hvilket er i overensstemmelse med resultaterne fra dette forsøg. Det skal desuden bemærkes at ren sukkerroemelasse rangerede forholdsvist lavt, hvilket antyder, at der er en grænse for hvor sødt, køer foretrækker deres fodermidler.

61

Popular scientific paper

Konklusion Forsøget viste, at køer har forskellige præferencer for forskellige kraftfodermidler, når flere kraftfodermidler tilbydes på samme tid. I dette forsøg var præferencen for kontrollen meget høj, og kun sojaskrå blev rangeret højere end kontrollen. Til trods for den høje præference for kontrollen var der syv fodermidler, som blev rangerede lige så højt som kontrollen, hvilket viser, at det er muligt at finde fodermidler som malkekøer har høj præference for sammenlignet med andre. Det var ikke muligt at vise nogen direkte effekt af fysisk form på køernes præferencer, da de højest rangerede fodermidler inkluderede forskellige fysiske former. Udover valget af fodermiddel kan køernes præferencer muligvis øges ved tilsætningen af 25 % sukkerroemelasse, hvilket var tilfældet for hvede, grøntpiller, ærter og roepiller. Dette forsøg var udført som et intensivt præferenceforsøg på køer i bindestald, hvorved en direkte sammenligning til AMS er med forbehold. Derfor er yderligere undersøgelser nødvendige for at fastslå de testede fodermidlers egentlige potentiale som alternative kraftfodermidler i AMS.

62

5 Conclusions In the literature review the effect of voluntary visits to the milking robot was found to be important to utilize the fully potential in the automatic milking system. It was found that the relative palatability between the MR offered ad libitum and the concentrate offered in the robot are important for the number of voluntary visits to the milking robot. Previous studies has found that cows has different preferences for different concentrate feeds and that it is possible to affect the number of voluntary visits to the milking robot by the composition of the concentrate offered in the milking robot. Further in the literature review it was found that the preferences for a feed is difficult to measure and only factors associated with feed preferences can be measured. The most used factors associated with feed preferences are choice of feeds, total feed intake, rate of feed intake and behavioral measurements. More factors are found to determine ruminants’ preferences for a feed. Most important animal related factors are the animal species, physiological stage and previous experiences. Most important feed related factors are the physical and chemical properties of the feed. The experiment showed that the cows had different preferences for the different concentrate feeds tested making it possible to make a ranking of the concentrates in relation to feed preferences. Leftovers and time were used as responses to describe the feed preferences. Behavioral measurements were done to support the other two responses. The two responses and the behavioral measurements showed the same tendencies in relation to ranking of the concentrates. It is concluded that having all the responses recorded provide extra information to the interpretation of the results, than only one of the recorded responses. Only soybean meal was found to be preferred higher than the control concentrate. A few concentrates were found to be preferred at the same level and most concentrates were found to be preferred at a lower level than the control concentrate. The preferences for the control concentrate were high in the experiment affecting the preference ranking of the test concentrates in relation to the control. Regarding physical form of the concentrates no clear correlation between physical form and feed preferences were found. A tendency was found that a pelletized form of barley was preferred above a rolled form. The addition of 25 % sugar beet molasses increased the preferences for wheat, peas, dried grass pellets and dried sugar beet pulp compared to the pure form of these. However, the addition of sugar beet molasses also included a pelletizing of the concentrates in 3 mm pellets making it difficult to determine whether the higher preferences were because of the physical form, or the addition of molasses, or an interaction between pelletizing and molasses addition.

63

6 Perspectives The experiment done in this thesis aimed to find concentrates that were highly preferred by dairy cows, with a potential to improve the number of voluntary visits to the milking robot, if they were used as concentrate feed in small amounts in the milking robot. From the results of this experiment one concentrate were found to be preferred at a higher level than the control. A range of concentrates was found to be preferred at the same level as the control and most concentrates tested were preferred at a lower level than the control. For the interpretation of the results it would have been more desirable if the preferences for the control had been in the middle of the preference ranking. Evaluating the results from a production point of view it seems reasonable to use a standard concentrate used in the milking robot in a production herd as the control. That the feed preferences for the control were high in the experiment acknowledge the production of commercial concentrate feeds. This further makes it a challenge to find alternative concentrate feeds that cows will prefer higher than commercial concentrate feeds, as the results from this study shows that the preferences of the commercial concentrates available on the market is already high. The concentrate “Kalvevalse” were tested two times from a different consignment and on a different set of cows. In both tests “Kalvevalse” were preferred high. Distel et al., (1994) and Villalba et al., (2004) has shown that if animals in early life, especially in sensitive periods experience a feed it will have high preferences for that feed if it is presented to it again later in the life. This seemed also to be the case in this experiment as the preferences for “Kalvevalse” were high and as “Kalvevalse” are used as feed for the calves in the herd where the cows in this experiment were raised. These findings suggest that it might be possible to higher the preferences for the concentrate which is going to be used in the milking robot, if the cows are presented to the concentrate from early life. This may be difficult if animals are bought into a herd, but if the animals are born within the herd, it seems possible to implement in practice. However, more studies are needed to determine this possible effect. This experiment did not involve an AMS but were done on cows housed in a tie stall. As the literature review has shown regarding cows’ behavior in an AMS many factors are involved. Jacobs and Siegford (2012) suggested that differences in management and farm-level variables may be more important to AMS efficiency and milk production than features of the milking system itself. Therefore the results found in this experiment need to be tested in a production trial where some of the concentrates can be tested as concentrate feeds in a milking robot. As the experiment in this thesis were a part of a larger appropriation this were done in a series of trials running at the Danish Cattle Research Centre (KFC), where different concentrate feeds are tested against the same control as the one used in this experiment. By the end of this thesis the series of trials were not completed and data from the trials that had finished were not analyzed. The results from these production trials will be interesting to study in relation to the findings in this experiment.

64

7 References Abdelqader, M.M., Oba, M. 2012. Lactation performance of dairy cows fed increasing concentrations of wheat dried distillers grains with solubles. J. Dairy Sci. 95: 3894-3904. Albright, J., L. 1993. Feeding behavior of dairy cattle. J. Dairy Sci. 76, 485-498. Andersen, P.E., Lykkeaa, J. 1962 Almindeligt kraftfoder contra kraftfoderpiller. Bilag til landøkonomisk forsøgslaboratorium efterårsmåde 37-41. AOAC 1990. Official Methods of Analysis, 15th edition, AOAC International, Gaithersburg, MC, USA. Atwood, S.B., Provenza, F.D., Wiedmeier, R.D., Banner, R.,E. 2001. Influence of free-choice vs. mixed-ration on food intake and performance of fattening calves. J. Anim. Sci. 79: 30343040. Azarfar, A., Jonker, A., Hettiarachchi, I.k., Yu, P. 2012. Nutrient profile and availability of coproducts from bioethanol processing. J. of Animal Physiology and Animal Nutrition. 96: 450-458. Bach, A., Busto, I. 2005. Effects on milk yield of milking interval regularity and teat cup attachment failures with robotic milking systems. J. Dairy Research. 72 101-106. Bach, A., Iglesias, C., Calsamiglia, S., Devant, M. 2007. Effect of amount of concentrate offered in automatic milking systems on milking frequency, feeding behavior, and milk production of dairy cattle consuming high amounts of corn silage. J. Dairy Sci. 90, 5049-5044. Bach, A., Devant, M., Igleasias, C., Ferrer, A. 2009 Forced traffic in automatic milking systems effectively reduces the need to get cows, but alters eating behavior and does not improve milk yield of dairy cattle. J. Dairy Sci. 92, 1272-1280. Bach, A., Villalba, J.J, Ipharraguerre, I.R. 2012. Interactions between mild nutritional imbalance and taste preferences in young ruminants. J. Anim. Sci. 90: 1015-1025. Bampidis, V.A., Robinson, P.H. 2006. Citrus by-products as ruminant feeds. Animal Feed Science Technology. 128: 175-217. Bath, D.L., Dunbar, J.R., King, J.M., Berry, S.L., Leonard, R.O., Olbrich, S.E. 1980. By-products and unusual feedstuffs in livestock rations. Western Regional Extension Publication No. 39 USDA-ARS, Washington, DC, USA.

65

References

Baumont, R. 1996. Palatability and feeding behavior in ruminants. Ann Zootech 45, 385-400. Belibasakis, N.G., Tsirgogianni, D. 1996. Effect of dried citrus pulp on milk yield, milk composition and blood components of dairy cows. Animal Feed Science Technology. 60: 87-92. Brzóska, F. 2008. Milk Production and Composition as Influenced by Soybean Meal, Rapeseed Meal, or Rapeseed Cake in Concentrates for Dairy Cows. Ann. Anim. Sci. Vol. 8 No. 2: 133-143. De Campeneere, S., De Boever, J.L., De Brabander, D.L. 2006. Comparison of rolled, NaOH treated and ensiled wheat grain in dairy cattle diets. Livestock Science. 99: 267-276. Dinnusson, W.E., Haugse, C.N., Knutson, R.D. 1971. Ergots in Rations for Fattening Cattle Farm Research November-December: 20-22. Distel, R.,A., Villalba, J.,J., Laborde, H.,E. 1994. Effects of early experience on voluntary intake of low-quality roughage by sheep. J. Anim. Sci. 72, 1191-1195. Emanuelson, M. 1989. Rapeseed products of doublet low cultivars to dairy cows. Effects of long term feeding and studies on rumen metabolism PhD Thesis, Report 189, Dept. Anim. Nur. Management, Swedish Univ. Agric. Sci., Uppsala, Sweden. Erdman, A.A., Varner, M. 1995 Fixed yield response to increased milking frequency. J. Dairy Sci. 78, 1199-1203. Franke, K., Meyer, U., Flachowsky, G. 2009. Distillers dried grains with soluble compared with rapeseed meal in ration for dairy cows. J. of Animal and Feed Science. 18: 601-612. Fraser, M.D., Theobald, V.J., Griffiths, J.B:, Morris, S.M., Moorby, J.M. 2009. Comparative diet selection by cattle and sheep grazing two contrasting heathland communities. Agricultural, Ecosystems and Environment. 129, 182-192. Forbes, J.,M. 2010. Palatability: principles, methodology and practice for farm animals. In Hemming, D. (Ed). Animal science reviews 2010. CAB Reviews pp 229-243. Ginane, C., Petit, M., D´Hour, P. 2003. How do grazing heifers choose between maturing reproductive and tall or short vegetative swards? Applied animal behavior science 83, 15-27. Halachmi, I., Ofir,S., Miron, J. 2005. Comparing two concentrate allowances in an automatic milking system. J. Anim. Sci. 80, 339-343.

66

References

Halachmi, I., Shoshani, E., Solomon, R., Maltz, E., Miron, J. 2006. Feeding of pellets rich in digestible neutral detergent fiber to lactating cows in an automatic milking system. J. Dairy Sci. 89, 3241-3249. Halachmi, I., Shoshani, E., Solomon, R., Maltz, E., Miron, J. 2009. Feeding soy hulls to highyielding dairy cows increased milk production, but not milking frequency in an automatic milking system. J. Dairy Sci. 92, 2317-2325. Hansen, B. 1989. Determination of nitrogen as elementary N, an alternative to Kjeldahl. Acta Agric Scand. 39: 113-118. Hermans, G.G.N., Ipema, A.H., Stefanowska, J., Mertz, J.H.M. 2003. The effect of two traffic situations on the behavior and performance of cows in an automatic milking system. J. Dairy Sci. 86, 1997-2003. Hodgson, J. 1979. Nomenclature and definitions in grazing studies. Grass and forage science 34, 11-18. Jacobs, J., A., Siegford, J.,M. 2012. The impact of automatic milking systems on dairy cow management, behavior, health and welfare. J. Dairy Sci. 95, 2227-2247. Kelstrup, L., Sepstrup, P. 2009. Korn, raps og ært 2th edition 2009 Landbrugsforlaget:101-128. Kenny, P.A., Black, J.L. 1984. Factors affecting diet selection by sheep. I Potential intake rate and acceptability of feeds. Aust. J. Agric. Res. 35, 831-838. Klopfer, F.D., Kilgour, R., Matthews, L.R. 1981. Paired comparison analysis of palatabilities of twenty foods to dairy cows. Proceedings of the N.Z. Society of Animal Production. 41: 242-247. Konning, K, Rodenburg, J. 2004. Automatic milking: State of the art in Europe and North America. In Mejringen et al., (Ed). Automatic milking – a better understanding. Wagening academic press pp. 27-37. Larsen, M.K., Hymøller, L., Brask-Pedersen, D.B., Weisbjerg, M.R. 2012. Milk fatty acid composition and production performance of Danish Holsteins and Danish Jersey cows fed different amounts of linseed and rapeseed. J. Dairy Sci. 95, 3569-3578. Madsen, J., Weisbjerg, M., R., Hvelplund, T. 2010. Concentrate composition for automatic milking systems – Effect on milking frequency. Livestock Sci. 127, 45-50.

67

References

Meier, J.,S., Kreuzer, M., Marquardt, S. 2012. Design and methodology of choice feeding experiments with ruminant livestock. Applied Animal Behavior Science.149, 105-120. Mertens, D. R. 2002. Gravimetric determination of amylase-treated neutral detergent fiber in feeds with refluxing in breakers or crucibles: Collaborative study. L. AOAC Int. 85: 1217-1240. Michalet-Doreau, B. 1975. Recherches sur les causes des variation des quantités dénsilage d´herbe ingérées par les ruminants. Thése Docteur-Ingénieur Université de Nancy, France, 99 p. Migliorati, L., Speroni, M., Stelletta, C., Pirlo, G. 2009. Influence of feeding flavouring-appetizing substances on activity of cows in an automatic milking system. Ital. J. Anim. Sci. 8, 417419. Morand-Fehr, P. 2003. Dietary choices of goats at the trough. Small ruminant research 49, 231-239. Mortensen, H.P., Madsen, A. 1984. Stigende mængder rug til slagtesvin. Statens husdyrbrugsforsøg, meddelelse nr. 534. Munksgaard, L., Jensen, M.B., Pedersen, L.J., Hansen, S.W., Matthews, L. 2005 Quantifying behavioral priorities-effects of time constraints on behavior of dairy cows, Bos Taurus. Applied Animal Behavior Science. 92, 3-14. Munksgaard, L., Rushen, J., Passillé, A., M., Krohn, C.,C. 2011. Forced versus free traffic in an automated milking system. Livestock science 138, 244-250. Murphy, J.J. 1999. The effect of increasing the proportion of molasses in the diet of milking dairy cows on milk production and composition. Animal Feed Science Technology. 78: 189-198. Nevens, W.B. 1927. Experiments in the Self-feeding of Dairy Cows. Bulletin no. 289 Agricultural Experiment Station, University of Illinois Urbana, Illinois, USA. Norfor, 2013. Nordic feed Evaluation System. Olofsson, J. 1999. Competition for Total Mixed Diets Fed for Ad Libitum Intake Using One or Four Cows per Feeding Station. J. Dairy Sci. 82, 69-79. Petit, H.V., Riox, R., Ouellet, D.R. 1997. Milk Production and Intake of Lactating Cows Fed Raw or Extruded Peas. . J. Dairy Sci. 80: 3377-3385. Rodenburg, J., Wheeler, B. 2002. Strategies for incorporating robotic milking into North American herd management. Proceedings of the first American Conference on Robotic Milking, March 20-22, Toronto, Canada.

68

References

Rodenburg, J., Focker, E., Hand , K. 2004. Effect of the composition of concentrate fed in the milking box, on milking frequency and voluntary attendance in automatic milking systems. In: Meijering, et al., (Ed.), Automatic Milking - A Better Understanding. Wagening Academic Press, pp. 511–512. Rodenburg, J., 2011. Designing feeding systems for robotic milking. Tri-state dairy nutrition conference April 19-20 2011, 127-138. SAS Institute. 2010. Statistical Analysis System. User’s guide. SAS Institute Inc., Cary, NC. Schroeder, J.W., Park, C.S. 2010. Using a Total Mixed Ration for Dairy Cows NDSU, Extension Service, North Dakota State University, July 2010. Shipley, L.A., 1999. Grazers and Browsers: How Digestive Morphology Affects Diet Selection. In Launchbaugh, K.L., Sanders, K.D., Moesley, J.C. (Ed). Grazing Behavior and Wildlife 2027. Sharma, H.R., Ingalls, J.R., Mckirdy, J.A., Sanford, L.M. 1981. Evaluation of Rye Grain in the Diets of Young Holstein Calves and Lactating Dairy Cows. J. Dairy Sci. 64: 441-448. Sharma, H.R., Ingalls, J.R., Mckirdy, J.A. 1983. Feeding value of Alkali-Treated whole Rye Grain for Lactating Cows and its Digestibility for Sheep. Animal Feed Science and Technology. 10: 77-82. Spörndly, E., Åsberg, T. 2006. Eating Rate and Preference of Different Concentrate Components for Cattle. J. Dairy Sci. 89: 2188-2199. Stoldt, W 1952. Vorschlag zur Vereinheitlichung der Fettbestimmung in Lebensmitteln. Fette Seifen 54: 206-207. Søegaard, L.S., 2012. Status på malkerobotter. Kvægnyt 6, 2012. Used the 8/8 2013 from http://www.vfl.dk/NR/rdonlyres/C5467A1A-89FF-4BCA-988A-7462588094B8/0/KvaegNYT0612.pdf

Tucker, C., B. 2009. Behavior of cattle. In Jensen, P. (Ed). CAB international The ethology of domestic animals 2nd edition pp. 151-160. Valros, A., Hänninen, L. 2009. Behavior and physiology. In Jensen, P. (Ed). CAB international The ethology of domestic animals 2nd edition pp. 25-38. Vander, M., Pol, A.N., Zaman, S., Delano, N. 2008. Peas Can Replace Soybean Meal and Corn Grain in Dairy Cows Diets. J. Dairy Sci. 91: 698-703.

69

References

Villalba, J.,J., Provenza, F.,D. 1999. Effects of food structure and nutritional quality and animal nutritional state on intake behavior and food preferences of sheep. Applied animal behavior science 63, 145-163. Villalba, J.,J., Provenza, F.,D., Han, G. 2004. Experience influences diet mixing by herbivores: implications for plant biochemical diversity. OIKOS 107, 100-104. Wagner-Storch, A.M., Palmer, R.W., 2003. Feeding behavior, milking behavior, and milk yields of cows milked in a parlor versus an automatic milking system. J. Dairy Sci. 86, 1494-1502. Weisbjerg, M.R., Hvelplund, T., 1993. Estimation of net energy content (FU) in feeds for cattle. In: National Institute of Animal Science, Report No 3/1993, 39. Weisbjerg, M.R., Munksgaard, L. 2008. Kraftfoderstrategier i et AMS system. Malkekoens fodring. ”Fodringstrategier” og ”Aktuel forskning” Temamøde 10 April 2008 Forskningscenter Foulum. Weisbjerg, M.R., Munksgaard, L. 2009. Effect of concentrate allowance in an automatic milking system. In: Book of Abstracts no. 15 of the 60th Annual Meeting of the European Association for Animal Production (EAAP), Barcelona, Spain, p. 349. Weisbjerg, M.R., Larsen, M.K., Hymøller, L., Thorhauge, M., Kidmose, U., Nielsen, J.H. Andersen, J. 2013. Milk production and composition in Danish Holstein, Danish Red, Danish Jersey cows supplemented with saturated or unsaturated fat. Livestock Science 155, 60-70. Worsøe, T., 2011. Antallet af malkerobotter på vej ned. Landbrugsavisen 11/11 2011. Used the 8/8 2013 from; http://www.landbrugsavisen.dk/Landbrugsavisen/2011/11/11/Antalletafmalkerobotterpaavejned.htm

Yang, W.Z., Li, Y.L., McAllister, T.A., McKinnon, J.J., Beauchemin, K.A. 2012. Wheat distillers grain in feedlot cattle diets: Feeding behavior, growth performance, carcass characteristics and blood metabolites. J. Anim. Sci. 90: 1301-1310. Zhang, S.Z., Penner, G.B., Yang, W.Z., Oba, M. 2010. Effects of partly replacing barley silage or barley grain with distillers grains with soluble on rumen fermentation and milk production on lactating dairy cows. J. Dairy Sci. 93: 3231-3242.

70