WGM e.V.

„Analyse der Eutergesundheit auf einem Großbetrieb auf der Basis einer betriebsei-genen Erfassung des somatischen Zellgehaltes in der Milch“

C. Brückner

Inhaltsverzeichnis
Verwendete Abkürzungen 6
1. Einleitung 7
2. Literaturübersicht 9
2.1 Gesetzliche Regelungen zur Milchhygiene 9
2.1.1 Milchverordnung 9
2.1.2 Milchgüteverordnung 10
2.2 Physiologie der Milchzellen 12
2.2.1 Polymorphkernige Leukozyten 12
2.2.2 Lymphozyten 13
2.2.3 Makrophagen 13
2.2.4 Nicht differenzierbare Zellen und Zellteile 14
2.2.5 Spezielle Zellen 14
2.3 Verfahren der Zellzahlerfassung 14
2.3.1 Sedimentbeurteilung 14
2.3.2 Indirekte Methoden 15
2.3.3 Indirekte Methoden über Hilfsmerkmale 16
2.3.4 Direkte Methoden 17
2.4 Kategorien der Eutergesundheit 18
2.4.1 Grenzwerte für Zellzahlen 18
2.4.2 Definitionen von Grundbegriffen zur Eutergesundheit 19
2.5 Einflussfaktoren auf den somatischen Zellgehalt 21
2.5.1 Mastitisauslösende Faktoren 21
2.5.1.1 Infektionserreger 22
2.5.1.2 Toxine und Traumata 24
2.5.2 Andere Faktoren 25
2.6 Ökonomische Aspekte 32
2.7 Genetik und Zucht 34
3. Material und Methoden 37
3.1 Der Versuchsbetrieb 37
3.1.1 Haltungsbedingungen 37
3.1.2 Abkalbungen 39
3.1.3 Leistungsniveaus und Zellgehalte bei
Milchleistungsprüfungen 39
3.1.4 Ergebnisse einer bakteriologischen Untersuchung 39
3.2 Datenerfassung 40
5
3.3 Datenaufbereitung 43
4. Ergebnisse 45
4.1 Methodenvergleich 45
4.2 Zellgehalt 46
4.2.1 Euterviertel 47
4.2.2 Alter 49
4.2.2.1 Altersstruktur der Versuchstiere 50
4.2.2.2 Zellzahlen getrennt nach Eutervierteln
und Laktationen 51
4.2.3 Saisoneffekt 56
4.2.4 Leistungen der vorangegangen Laktation 56
4.2.5 Leistungen der aktuellen Laktation 57
4.2.6 Genetik 60
4.3 Statistische Modelle 65
4.3.1 Modell 65
4.3.2 Weitere Modelle 68
4.3.2.1 Modell für den maximalen Somatic Cell Score 69
4.3.2.2 Modell für die erste Laktation 72
5. Diskussion und Schlussfolgerungen 76
5.1 Einordnung des Betriebes aufgrund der Daten aus
Milchleistungsprüfungen 76
5.1.1 Einsatzleistung 76
5.1.2 Gehalt an somatischen Zellen innerhalb der Einsatzleistung 78
5.2 Versuchsdaten und Ergebnisse 80
5.2.1 Methodenvergleich 80
5.3 Einflussfaktoren 81
5.4 Statistische Modelle 84
5.4.1 Modell 84
5.4.2 Weitere Modelle 86
6. Zusammenfassung 89
7. Literaturverzeichnis 91
8. Tabellenverzeichnis 97
9. Abbildungsverzeichnis 100
6
Verwendete Abkürzungen
Abb. Abbildung
AID Auswertungs- und Informationsdienst für Ernährung, Landwirt-schaft und Forsten e. V.
cm Zentimeter
DNA Desoxyribonukleinsäure
DVG Deutsche Veterinärmedizinische Gesellschaft e. V.
e. V. eingetragener Verein
HL hinten links
HR hinten rechts
kg Kilogramm
KNS Koagulase-negative Staphylokokken
Max Maximum
Max_SCS Maximaler Somatic Cell Score
Mio. Millionen
Min Minimum
ml Milliliter
MLP Milchleistungsprüfung
r Korrelationskoeffizient
RZM Relativzuchtwert im Merkmal Milchleistung
RZS Relativzuchtwert im Merkmal Somatischer Zellgehalt der Milch
SCS Somatic Cell Score (= Logarithmierter Zellgehalt)
VIT Vereinigte Informationssysteme Tierhaltung
VL vorn links
VR vorn rechts
ZZ Zellzahl

1
Einleitung
Die Milch ist eines der wenigen landwirtschaftlichen Erzeugnisse aus der tierischen Produktion, welches im Grunde von Natur aus ohne weitere Be- und Verarbeitung als Lebensmittel zum direkten menschlichen Verzehr geeignet ist.
Etwa ein Fünftel der in Deutschland erzeugten Milch wird als Trinkmilch verbraucht. Der Großteil wird jedoch zu den verschiedensten Milchprodukten (wie zum Beispiel Butter, Käse und Joghurt) weiter verarbeitet. Der Verbraucher stellt hohe Ansprüche an die Milch und deren Erzeugnisse. Sie sollen frisch, natürlich und schmackhaft sein, hy-gienisch einwandfrei und haltbar.
Aufgrund ihrer biochemischen Zusammensetzung zählen jedoch gerade Milch und Milchprodukte zu den besonders leicht verderblichen Lebensmitteln .Dies verlangt zwangsläufig besonders hohe Anforderungen an die Sicherung der Hygiene und Quali-tät bei der Gewinnung, der Behandlung, der Be- und Verarbeitung sowie im Lebensmit-telhandel bis hin zum Verbraucher.
Ein bedeutender Qualitätsparameter ist der Gehalt an somatischen Zellen in der Milch. Unter somatisch versteht man in diesem Zusammenhang, dass es sich um körpereigene Zellen aus dem Eutergewebe und aus dem Blut handelt. Der Zellgehalt in der Milch lässt somit als Bestandteil des unspezifischen Abwehrsystems im tierischen Organismus Aussagen über die Eutergesundheit zu. Nur gesunde Kühe mit gesunden Eutern können den Leistungs- und Qualitätsanforderungen gerecht werden.
Der Gehalt an somatischen Zellen in der Milch ist als Gütemerkmal in der Verordnung über die Güteprüfung und Bezahlung der Anlieferungsmilch (Milchgüteverordnung vom 9. Juli 1980, in der Fassung der 6. Änderungsverordnung vom 30.10.2003) defi-niert. Der Grenzwert, bei dem der Landwirt den vollen Grundpreis für seine abgelieferte Milch erzielt, liegt derzeit bei 400.000 Zellen/ml Herdensammelmilch. Die Milch ist verkehrsfähig, wenn die Zellzahl unter diesem Grenzwert liegt. Eine Vorraussetzung für die hygienische Unbedenklichkeit und die Qualität der aus der Rohmilch erzeugten Pro-dukte wird somit erfüllt. In Thüringen zum Beispiel lag der durchschnittliche Zellgehalt 2003 und 2004 bei etwa 228.000 Zellen/ml Milch (TVL, 2004).
Ein wichtiger Aspekt aus tierschutzrechtlicher Sicht ist die Tiergesundheit. Aus diesem Grund werden niedrige somatische Zellgehalte als Kennzeichen eines gesunden Euters angestrebt.
Von den Eutergesundheitsstörungen zählen Entzündungen der Milchdrüse (Mastitiden) zu den am häufigsten auftretenden Erkrankungen bei Hochleistungskühen. Mastitiden
8
sind in der Regel mit hohen Behandlungskosten verbunden. Einnahmeverluste entstehen durch den Milchleistungsabfall und die Wartezeiten aufgrund von Ablieferungssperren für Milch von Tieren, die mit Medikamenten (vor allem Antibiotika) behandelt wurden. Aus diesem Grund haben die Milchproduzenten großes Interesse an einer frühzeitigen Erkennung von erhöhten Zellgehalten in der Milch.
Die Firma DeLaval hat ein tragbares Zellzahlmessgerät entwickelt, mit dessen Hilfe es für den Milcherzeuger möglich ist, den Zellgehalt der Herdensammelmilch, des Ge-samtgemelks eines Einzeltieres oder des Viertelgemelks selbst zu erfassen. Der Herstel-ler empfiehlt dieses Messgerät vor allem zum Auffinden subklinischer Mastitiden. Bei dieser Mastitisform sind noch keine Symptome in Form von Flocken, Rötung, Schmerzempfindlichkeit, Wärme oder Schwellung der Milchdrüse sichtbar und eine Infektion ist nur über einen erhöhten Gehalt an somatischen Zellen in der Milch nach-weisbar. Der so genannte „DeLaval Cell Counter DCC“ zeichnet sich durch eine einfa-che und schnelle Handhabung aus.
In einem Thüringer Milchviehbetrieb wurde das Messgerät fast ein Jahr lang im prakti-schen Einsatz getestet. Es diente der betriebseigenen Erfassung somatischer Zellgehalte in Viertelgemelksproben von Frischabkalbern. Die Tiere wurden in der Regel am dritten Laktationstag beprobt. Anhand der gemessenen Zellzahlen konnten auffällige Tiere, das heißt solche mit erhöhten Zellgehalten, erkannt und gleich behandelt werden.
Gegenstand dieses Projektes war die Auswertung der vom Versuchsbetrieb erfassten Daten. Die Besonderheit liegt hierbei in der sehr genauen, viertelspezifischen und früh-zeitigen Erfassung des somatischen Zellgehaltes. Bevor dieses Gerät im Versuchsbe-trieb eingesetzt wurde erfolgte die erste Zellzahlbestimmung in der Frühlaktation bei der ersten Milchleistungsprüfung. Jedoch nicht viertelspezifisch, sondern nur für das Gesamtgemelk eines jeden Tieres.
Ziel des Projektes war es zu untersuchen, inwiefern sich der „DeLaval Cell Counter DCC“ zur Zellzahlbestimmung in einem Milchviehbetrieb eignet. Außerdem wurde analysiert welche Tiere im Hinblick auf die Eutergesundheit Probleme bereiten und von welchen Faktoren der Gehalt an somatischen Zellen wenige Tage nach der Abkalbung signifikant beeinflusst wird. Der Versuchsbetrieb stellte aktuelle Daten seines Manage-mentprogramms (zum Beispiel Ergebnisse von Milchleistungsprüfungen) zur Verfü-gung. Außerdem waren für den Großteil der beprobten Tiere die Relativzuchtwerte ihrer Väter vom VIT erhältlich. Dadurch war es möglich genetische Einflüsse auf den soma-tischen Zellgehalt der Frischabkalber zu untersuchen.

Untersuchungen zur elektrischen
Leitfähigkeit von Schafmilch und ihrer
Nutzung zur Eutergesundheitskontrolle bei
Milchschafen

Elke Burow

Inhaltsverzeichnis
1 EINLEITUNG................................................................................................................1
2 LITERATURÜBERSICHT ............................................................................................2
2.1 DAS SCHAFEUTER......................................................................................................2
2.1.1 Anatomie und Melkbarkeit .................................................................................2
2.1.2 Physiologie – Milchbildung,- speicherung, -abgabe ..........................................3
2.1.3 Blut- Milch/ Milch- Blut- Schranke (BMS)...........................................................4
2.2 SCHAFMILCH ..............................................................................................................5
2.2.1 Milchinhaltsstoffe ...............................................................................................5
2.2.1.1 Mengenanteile............................................................................................5
2.2.2 Milchzellen.........................................................................................................8
2.2.2.1 Somatischer Zellgehalt ...............................................................................8
2.2.3 Milchleistung......................................................................................................9
2.3 MASTITIS....................................................................................................................9
2.3.1 Euterpathogene ...............................................................................................10
2.3.1.1 Staphylokokken ........................................................................................10
2.3.1.2 Streptokokken...........................................................................................11
2.3.1.3 Pasteurellen..............................................................................................11
2.3.2 Abwehrreaktionen............................................................................................12
2.3.3 Mastitisbedingte Änderung der Milchzusammensetzung ................................12
2.3.4 Definition des Gesundheitsstatus ....................................................................13
2.3.5 Mastitisformen .................................................................................................13
2.3.6 Prävalenz.........................................................................................................14
2.4 DIAGNOSTIK.............................................................................................................14
2.4.1 Bakteriologische Untersuchung (BU)...............................................................15
2.4.2 Zellgehalt .........................................................................................................15
2.4.2.1 Zytologische Untersuchung - Zellzählung ................................................15
2.4.2.2 Zytologische Orientierungsuntersuchung - Zellschätzung........................18
2.4.3 Elektrische Leitfähigkeit (LF) ...........................................................................20
2.4.3.1 Allgemein..................................................................................................20
2.4.3.2 Methode und Geräte.................................................................................22
2.4.3.3 Auswertungsmethoden.............................................................................22
2.4.3.4 Messbeeinflussende Faktoren..................................................................23
2.4.3.5 Vergleich mit anderen Diagnostikparametern ..........................................25
2.4.3.6 Grenzwerte/ Aussagekraft ........................................................................27
2.4.4 Eiweiß – Akutes Phase Protein .......................................................................27
2.4.5 Klinische Untersuchung...................................................................................28
2.4.6 Gesamtdiagnostik............................................................................................28
3 SCHLUSSFOLGERUNGEN AUS DER LITERATUR................................................29
4 EIGENE UNTERSUCHUNGEN .................................................................................31
4.1 MATERIAL UND METHODEN UNTERSUCHUNGSÜBERGREIFEND ...................................31
4.1.1 Allgemein.........................................................................................................31
4.1.2 Messtechnische Ausstattung...........................................................................31
4.2 HÄLFTENANFANGSGEMELKSUNTERSUCHUNGEN (HAUPTUNTERSUCHUNG).................32
4.2.1 Material und Methoden....................................................................................32
4.2.1.1 Allgemein..................................................................................................32
4.2.1.2 Probennahme...........................................................................................33
4.2.1.3 Laboruntersuchung...................................................................................33
4.2.1.4 Auswertung...............................................................................................34
4.2.1.5 Definitionen...............................................................................................35
4.2.2 Ergebnisse.......................................................................................................35
4.2.2.1 Angaben zu den Betrieben.......................................................................35
4.2.2.2 Gültige Probenzahlen...............................................................................38
4.2.2.3 Bakteriologische Untersuchung................................................................39
4.2.2.4 Somatischer Zellgehalt .............................................................................40
4.2.2.5 Schalm- Mastitis- Test ..............................................................................42
4.2.2.6 Elektrische Leitfähigkeit............................................................................45
4.2.2.7 Milchinhaltsstoffe......................................................................................48
4.2.2.8 Korrelationen ............................................................................................49
4.2.2.9 Multivariates Verfahren.............................................................................51
4.2.3 Diskussion .......................................................................................................53
4.2.3.1 Ergebnisdiskussion der einzelnen Parameter ..........................................53
4.2.3.2 Gegenüberstellungen der Ergebnisse der Parameter BU, ZZ und der
Milchinhaltsstoffe mit der LF ...................................................................................58
4.2.3.3 Kritik..........................................................................................................60
4.3 LEITFÄHIGKEITEN ÜBER GEMELKSFRAKTIONEN (NEBENUNTERSUCHUNG I) ................62
4.3.1 Material und Methoden....................................................................................62
4.3.1.1 Allgemein..................................................................................................62
4.3.1.2 Probennahme...........................................................................................62
4.3.1.3 Laboruntersuchung...................................................................................62
4.3.1.4 Auswertung...............................................................................................62
4.3.1.5 Definitionen...............................................................................................62
4.3.2 Ergebnisse.......................................................................................................62
4.3.3 Diskussion .......................................................................................................65
4.4 HÄLFTENANFANGSGEMELKE ZUR LEITFÄHIGKEITSUNTERSUCHUNG IN ABHÄNGIGKEIT
DER TEMPERATUR (NEBENUNTERSUCHUNG II) .................................................................67
4.4.1 Material und Methoden....................................................................................67
4.4.1.1 Allgemein..................................................................................................67
4.4.1.2 Probennahme...........................................................................................67
4.4.1.3 Laboruntersuchung...................................................................................67
4.4.1.4 Auswertung...............................................................................................68
4.4.1.5 Definitionen...............................................................................................68
4.4.2 Ergebnisse.......................................................................................................68
4.4.3 Diskussion .......................................................................................................69
4.5 SCHLUSSFOLGERUNGEN AUS DEN EIGENEN UNTERSUCHUNGEN UND AUSBLICK........71
5 ZUSAMMENFASSUNG.............................................................................................73
6 LITERATURVERZEICHNIS.......................................................................................75
7 ANHANG....................................................................................................................82
7.1 ANLAGE 1: FRAGEBOGEN MILCHSCHAFHALTUNG.........................................................I
7.2 ANLAGE 2: GESAMTDIAGNOSTIK............................................................................... VI
7.3 DANKSAGUNG......................................................................................................... VII
7.4 EIDESSTATTLICHE ERKLÄRUNG .............................................................................. VIII

1 Einleitung
In dem Zeitraum von 1992 bis 2002 ist in Deutschland der Bestand an Milchschafen laut
Herdburchzucht um 57 % gestiegen (VRIES 2003). Es existieren jedoch keine genauen
Angaben zur produzierten Menge von Schafmilch, da diese i. d. R. auf den
schafhaltenden Betrieben direkt verarbeitet wird und keine zentralen Milcherfassungen
bzw. herdendeckenden Milchleistungsprüfungen (MLP) durchgeführt werden. Im
Zusammenhang mit einer Intensivierung der Milchschafhaltung kommt es nach WINTER
(2003) zu einer Zunahme von Milchzellzahl- und Mastitisproblemen in den Herden.
Klinische Mastitisfälle nehmen zwar einen drastischeren und meist folgenschwereren
Verlauf, die subklinischen Mastitiden drohen aber aufgrund ihrer Unscheinbarkeit
unerkannt zu bleiben und machen den größeren Anteil aus. Nach DEUTZ und
OBRITZHAUSER (2003) ist jedes fünfte Schaf chronisch oder subklinisch euterkrank. In
Untersuchungen von HONEGGER (1994) wiesen sogar 60 %, in Untersuchungen von
GONZALO ET AL. (2002) immerhin 25 %, der untersuchten Euterhälften subklinische
Mastitiden auf. Die mit der subklinischen Mastitis einhergehende
Milchleistungsdepression beläuft sich nach BEHRENS ET AL. (2001) auf 30 bis 40 % pro
Euterhälfte.
Neben der Milchmengeneinbuße und dem erhöhten Arbeitsbedarf bei mastitiserkrankten
Schafen stellen veränderte Verarbeitungseigenschaften und ein Qualitätsverlust der Milch
jener Tiere (JAEGGI ET AL. 2003; BIANCHI ET AL. 2004) einen, für deren hochpreisige
Produkte, gravierenden Schadensfaktor dar. So wird die Mastitisbekämpfung generell
sogar als Teil der Lebensmittelhygiene (HAMANN und KNAPPSTEIN 2003) gesehen.
Da regelmäßige MLP bei Milchschafherden kein Standard sind, Tankmilchkontrollen
keinen Einzeltierbefund liefern und Mastitiden zudem für einen größtmöglichen
Behandlungserfolg ein frühstmögliches Erkennen zu unmittelbar einsetzender
Maßnahmen erfordern, besteht die Notwendigkeit einer zuverlässigen, vom Landwirt
selbst durchführbaren, Diagnostikmethode. Diese sollte vor dem Hintergrund der
hierzulande i.d.R. geringen Betriebsgröße verhältnismäßig kostengünstig, einfach zu
bedienen, zuverlässig in der Aussage, eindeutig und schnell durchführbar sein. Neben
dem Schalm- Mastitis- Test, der zur Eutergesundheitskontrolle im Milchviehbereich weite
Verbreitung gefunden hat, ist aus dem Rinderbereich die Leitfähigkeitsmessung zur
Mastitisdiagnostik bekannt und vielfach bewährt und eingesetzt, im Schafbereich jedoch
noch wenig erforscht und genutzt.
Ziel dieser Arbeit ist, das Verhalten der Leitfähigkeit von Schafmilch und ihre Nutzbarkeit
zur Eutergesundheitskontrolle bei Milchschafen festzustellen. Hierzu sollen nach einem
Literaturüberblick mit daraus entwickelter Literaturschlussfolgerung in eigenen
Untersuchungen an 150 Milchschafen deren Eutergesundheit mittels bakteriologischer
und Zellgehalts- Untersuchungen bestimmt und dem ebenfalls erhobenen Parameter
Leitfähigkeit gegenübergestellt werden. Der zusätzlich ermittelte Gehalt an
Milchinhaltsstoffen, sowie Tierdaten u. a. zum Laktationstag und zur Laktationsnummer
sollen neben weiteren Versuchen zur Leitfähigkeit Aufschluss über Zusammenhänge mit
möglichen Einflussfaktoren geben.

Untersuchungen zu
Milchejektionsstörungen bei
erstlaktierenden Kühen

Untersuchungen von Zusammenhängen zwischen
morphologischen Merkmalen des Euters, der Eutergesundheit und
melktechnischen Parametern bei Tieren der Rasse Deutsches Holstein

Katja Graff

1 EINLEITUNG UND AUFGABENSTELLUNG 6
2 LITERATUR 9
2.1 MORPHOLOGISCHE MERKMALE DES EUTERS UND DER ZITZEN 9
2.2 MORPHOLOGIE UND EUTERGESUNDHEIT 19
2.3 MORPHOLOGIE UND MILCHABGABE 24
2.4 MILCHABGABE UND EUTERGESUNDHEIT 29
2.5 MELKTECHNIK 31
3 MATERIAL UND METHODIK 38
3.1 MATERIAL 38
3.2 DATENERFASSUNG 41
3.2.1 TIERSPEZIFISCHE DATEN 42
3.2.2 MORPHOLOGISCHE MERKMALE 42
3.2.3 EUTERGESUNDHEIT 46
3.2.4 MILCHFLUSSPARAMETER 47
3.2.5 MELKTECHNIK 50
3.3 DATENAUSWERTUNG 52
4 ERGEBNISSE 57
4.1 DARSTELLUNG DER ERGEBNISSE ANHAND STATISTISCHER KENNZAHLEN 57
4.1.1 MORPHOLOGIE 57
4.1.2 EUTERGESUNDHEIT 59
4.1.3 MILCHFLUSSPARAMETER 60
4.2 VARIANZ DER MERKMALE 61
4.2.1 MORPHOLOGIE 62
4.2.2 EUTERGESUNDHEIT 63
4.2.3 MILCHFLUSS 64
Inhaltsverzeichnis
3
4.3 BEZIEHUNGEN ZWISCHEN DEN MERKMALEN 64
4.3.1 MORPHOLOGIE 65
4.3.2 MILCHFLUSSPARAMETER 67
4.3.3 KORRELATIONEN ZWISCHEN DEN MERKMALSKOMPLEXEN 68
4.4 EFFEKTSCHÄTZUNG 73
4.4.1 KALBEALTER • LAKTATION 73
4.4.2 LAKTATIONSTAG 75
4.4.3 BODENABSTAND 78
4.5 MELKTECHNIK 79
4.5.1 PULSVERHÄLTNISSE UND PHASENLÄNGEN 79
4.5.2 VAKUUMVERHÄLTNISSE WÄHREND DES MELKVORGANGES 80
4.5.3 BEZIEHUNGEN ZWISCHEN MILCHFLUSS, MORPHOLOGIE UND DRUCKEINWIRKUNG 81
5 DISKUSSION UND SCHLUSSFOLGERUNGEN 86
5.1 MORPHOLOGIE 86
5.2 MORPHOLOGIE UND EUTERGESUNDHEIT 93
5.3 MORPHOLOGIE UND MILCHABGABE 97
5.4 MILCHABGABE UND EUTERGESUNDHEIT 101
5.5 MELKTECHNIK 102
6 ZUSAMMENFASSUNG 104
7 SUMMARY 106
8 LITERATURVERZEICHNIS 108
9 TABELLENVERZEICHNIS 106
10 ABBILDUNGSVERZEICHNIS 118
11 ANHANG 119

Einleitung und Aufgabenstellung
Die Wirtschaftlichkeit der Milcherzeugung wird aktuell u. a. durch hohe Bestandesergänzungen
von durchschnittlich 40,9 % (Mittel der Jahre 1999 – 2003 in Sachsen)
gemindert. Die Selektion aufgrund von Eutergesundheit und Melkbarkeit lag in den Jahren
von 2000 bis 2003 bei durchschnittlich 24 % (23,5 % - 24,8 %) und nimmt damit die größte
Selektionsposition ein (LKV Sachsen 2003). Diese Problematik wird durch VIT Verden
bestätigt, wonach im Jahr 2003 in Abhängigkeit von der Milchleistung zwischen 18 % und
21 % der Tiere aufgrund von Euter- und Melkbarkeitsproblemen selektiert wurden.
Bei der Betrachtung der Eutergesundheit über das Hilfsmerkmal „Gehalt an somatischen
Zellen je ml Milch“ ergibt sich nach Auskunft des Landeskontrollverbandes Schleswig-
Holstein eine stetige Verschlechterung innerhalb der Anlieferungsmilch (1999:
198.000 somatische Zellen je ml Milch – 2003: 220.000 somatische Zellen je ml Milch)
(zit. n. MAHLKOW-NERGE 2004). Ein vergleichbarer Trend wurde vom Sächsischen
Landeskontrollverband (2003) festgestellt. Der LKV Sachsen differenzierte nach Milchmenge
je Tag (in kg) und nach Anzahl der Laktationen und erhielt in jeder Aufstellung eine
Steigerung der Zellzahlmittelwerte im Vergleich der letzten drei Jahre. In 2003 lag die
durchschnittliche somatische Zellzahl aller Proben bei 234.000 somatische Zellen je ml Milch
(LKV Sachsen 2003).
Für die Aufrechterhaltung und Verbesserung der Eutergesundheit und die Ermöglichung
einer zügigen, gewebeschonenden Milchabgabe ist eine optimale Ausgestaltung der
melktechnischen Ausrüstung unerlässlich. Dieser Aspekt tritt besonders bei hohen bis sehr
hohen Milchleistungen und bei der Erhöhung der Melkhäufigkeit in den Vordergrund. Durch
die ständige Weiterentwicklung der Melktechnik in Bezug auf die Forderungen des
Tierschutzes zur Euterschonung und der Erhaltung der Wirtschaftlichkeit der
Milchgewinnung (Melkdurchsatz) werden immer öfter die technischen Grenzen der
Umsetzbarkeit erreicht (RUDOVSKY ET AL. 1984).
Als problematisch stellt sich seit Beginn des Maschinenmelkens u. a. die Haftung der
Melkzeuge bei Tieren mit sehr kurzen Zitzen (< 4 cm) sowie die Positionierung des
Melkzeuges bei sehr geringen (< 6 cm) und sehr großen (> 25 cm) Abständen der Zitzen
untereinander heraus (WILKE 1960, KOHLSCHMIDT ET AL. 1979, RUDOVSKY ET AL. 1984,
WENDT ET AL. 1994). Weiterhin tritt bei extrem ausgeprägten Maßen hinsichtlich des
Zitzendurchmessers und der Zitzenlänge eine suboptimale Positionierung der Zitze im
Zitzengummi auf, wodurch ein idealer Milchentzug nicht mehr gewährleistet werden kann
(HAMANN ET AL. 1994b). Die nichtoptimale Zitzenpositionierung im Zitzengummi führt zu einer
ungenügenden Massage des Zitzengewebes.
Einleitung und Aufgabenstellung
7
Neben der Position der Zitze im Zitzengummi spielt der Milchfluss eine bedeutende Rolle für
die Beanspruchung des Zitzengewebes während des Milchentzuges. Negative
Auswirkungen auf die Eutergesundheit werden dem Blindmelken und der uneingeschränkten
Vakuumeinwirkung auf das Zitzengewebe zugeschrieben (IDF 1994b).
Aussagen über die Auswirkungen von extremen morphologischen Ausprägungen hinsichtlich
der Zitzenlänge, des Zitzendurchmessers, der Zitzenabstände und Bodenabstände sowie
der Formmerkmale der Zitze und Kuppe auf die Eutergesundheit und Melkbarkeit können
derzeit in Qualität und Quantität nicht mit ausreichender Sicherheit getroffen werden, da
keine aktuellen Untersuchungen vorliegen.
Erst mit der Kenntnis über die in der Praxis vorzufindenden Euter- und Zitzenmorphologie in
Form absoluter Messwerte und deren Häufigkeitsverteilung sowie der bestehenden
Zusammenhänge zu den Merkmalen der Eutergesundheit und der Melkbarkeit kann die
technische Gestaltung von Zitzengummis und Melkmaschine weiter an die tierischen
Bedürfnisse angepasst werden.
Bei der aktuellen Zuchtbewertung des Sächsischen Rinderzuchtverbandes (SRV), in dessen
Zuchtgebiet die nachfolgenden Untersuchungen durchgeführt wurden, erfolgt die
Einflechtung von Zitzenlänge, Eutertiefe und Strichplatzierung durch die subjektive
Beurteilung in den Zuchtwert „Exterieur“. Da sich die Bewertung an der aktuellen
Merkmalsausprägung innerhalb der zugrundeliegenden Population orientiert und keine
metrische Bewertung der Merkmale erfolgt, sind Aussagen über Veränderungen hinsichtlich
der Mittelwerte der Merkmale nicht möglich.
Aus diesem Grund wird es als notwendig erachtet, die aktuell vorhandenen Ausprägungen
von Zitzen- und Eutermerkmalen an einer ausgewählten Teilpopulation zu erheben und auszuwerten.
Dabei sind die Verteilungshäufigkeiten, Mittelwerte und Extremwerte von großem
Interesse. Durch Wiederholungsmessungen und die Abstammungsdaten der Tiere innerhalb
der Teilpopulation lassen sich die grundlegenden Einflüsse auf die phänotypische Varianz
der untersuchten Merkmale ermitteln. Anhand der bestehenden Varianzen der Einzelmerkmalen
und Kovarianzen zwischen den Merkmalen lassen sich komplexe Beziehungen
in Art und Größe ermitteln.
Von Interesse sind weiterhin der Einfluss von fixen und variablen Effekten auf die Merkmalsausprägungen
von Morphologie, Eutergesundheit und Melkbarkeit.
Einleitung und Aufgabenstellung
8
Im aktuellen Zuchtgeschehen einer Vielzahl von Praxisbetrieben liegt die Priorität in der
Steigerung der Milchleistung. Einige, häufig eingesetzte Bullen mit hohen positiven
Zuchtwerten im Merkmalskomplex Milchleistung weisen negative Teilzuchtwerte im
Euterexterieur (z. B. Zitzenlänge) auf. Diesen wird aufgrund der geringen prozentualen
Bewertung innerhalb des Exterieurzuchtwertes nur eine geringe Beachtung zugeschrieben.
Die Auswirkungen der auftretenden Veränderungen der Zitzen- und Eutermorphologie auf
die Eutergesundheit und Melkbarkeit sind weitgehend unbekannt und werden aktuell nicht
untersucht, bedürfen jedoch wegen der wirtschaftlichen Bedeutung einer gesteigerten
Beachtung.

Untersuchung mechanischer Belastungen am Euter
bei verschiedenen Melksystemen

S.Rose

Inhaltsverzeichnis
Seite
Abkürzungs- und Formelverzeichnis IV
Tabellenverzeichnis V
Abbildungsverzeichnis X
1 Einleitung 1
2 Stand des Wissens 3
2.1 Melkanlage - Aufbau und Funktion 3
2.1.2 Pulsation und Vakuum 5
2.2 Melksysteme 9
2.2.1 Konventionelle Melksysteme 9
2.2.2 Automatische Melksysteme 16
2.3 Melktechnik und Eutergesundheit 20
2.3.1 Vakuumschwankungen 22
2.3.2 Melkzeugpositionierung 26
2.4 Messsysteme zur Überprüfung von Melkanlagen 29
3 Ziele und Aufgaben 34
4 Material und Methode 35
4.1 Kraftmessungen 35
4.1.1 Prüfstand zur Messung von Kräften am Euter 35
4.1.2 Untersuchte Merkmale und Bedingungen 37
4.1.3 Versuchsdurchführung 40
4.2 Vakuummessungen 43
4.2.1 Aufbau Versuchsmelkstand 43
4.2.2 Eingesetzte Messtechnik 44
4.2.3 Versuchsdurchführung 46
4.3 Ergebnisauswahl und -darstellung 48
5 Ergebnisse 50
5.1 Kraftmessungen in konventionellen Melksystemen 50
5.1.1 Vertikalkraft 50
5.1.1.1 Vergleich von Side-by-Side-Melkständen und FGM 33°/50°
bei verschiedenen Euterformen 50
5.1.1.2 Unterschiede zwischen den Melkstandseiten in einem FGM 33° 52
5.1.1.3 Analyse von Schlauchführungshilfen bei
verschiedenen Euterformen 53
5.1.1.4 Unterschiede zwischen den Melkplätzen bei verschiedenen
Melkstandformen und normaler Euterform 58
5.1.1.5 Einfluss des Melkers in zwei Fischgrätenmelkständen 33°
bei normaler und stufiger Euterform 61
5.1.1.6 Multifaktorielle Darstellung 63
Inhaltsverzeichnis
II
Seite
5.1.2 Drehkraft 63
5.1.2.1 Vergleich von Side-by-Side-Melkständen und FGM 33°/50°
bei verschiedenen Euterformen 63
5.1.2.2 Unterschiede zwischen den Melkstandseiten in einem FGM 33° 66
5.1.2.3 Analyse von Schlauchführungshilfen bei
verschiedenen Euterformen 67
5.1.2.4 Unterschiede zwischen den Melkplätzen bei verschiedenen
Melkstandformen und normaler Euterform 70
5.1.2.5 Einfluss des Melkers in zwei Fischgrätenmelkständen 33°
bei normaler und stufiger Euterform 73
5.1.2.6 Multifaktorielle Darstellung 75
5.1.3 Horizontalkräfte 75
5.1.3.1 Vergleich von Side-by-Side-Melkständen und FGM 33°/50°
bei verschiedenen Euterformen 75
5.1.3.2 Unterschiede zwischen den Melkstandseiten in einem FGM 33° 79
5.1.3.3 Analyse von Schlauchführungshilfen bei
verschiedenen Euterformen 80
5.1.3.4 Unterschiede zwischen den Melkplätzen bei verschiedenen
Melkstandformen und normaler Euterform 83
5.1.3.5 Einfluss des Melkers in zwei Fischgrätenmelkständen 33°
bei normaler und stufiger Euterform 89
5.1.3.6 Multifaktorielle Darstellung 92
5.2 Kraftmessungen in automatischen Melksystemen 93
5.2.1 Vertikalkraft 93
5.2.1.1 Unterschiede zwischen Melkmodulen und viertelindividueller
Schlauchführung bei verschiedenen Euterformen 93
5.2.1.2 Vergleich der Boxen innerhalb eines Fabrikates
bei normaler Euterform 95
5.2.2 Drehkräfte 96
5.2.2.1 Unterschiede zwischen Melkmodulen und viertelindividueller
Schlauchführung bei verschiedenen Euterformen 96
5.2.2.2 Vergleich der Boxen innerhalb eines Fabrikates
bei normaler Euterform 98
5.2.3 Horizontalkräfte 99
5.2.3.1 Unterschiede zwischen resultierenden Horizontalkräften
bei verschiedenen automatischen Melksystemen 99
5.2.3.2 Vergleich der Boxen innerhalb eines Fabrikats
bei normaler Euterform 101
5.3 Vergleich von konventionellen und automatischen Melksystemen102
5.3.1 Vertikalkraft 102
5.3.2 Drehkraft 104
5.3.3 Horizontale Längs- und Querkräfte 106
5.4 Zyklische Vakuumschwankungen bei konventionellen und
zentralenlosen Melkzeugen 109
5.4.1 Vergleich des zitzenendigen Vakuums bei
zunehmendem Milchfluss 109
5.4.2 Vakuumschwankungen an verschiedenen Punkten am Melkzeug 111
Inhaltsverzeichnis
III
Seite
6 Diskussion 116
6.1 Kräfte am Euter bei verschiedenen Melksystemen 116
6.1.1 Vertikalkräfte 116
6.1.2 Drehkräfte 121
6.1.3 Horizontalkräfte 123
6.2 Vergleich zyklischer Vakuumschwankungen bei
konventionellen und zentralenlosen Melksystemen 127
7 Schlussfolgerungen 129
8 Zusammenfassung (Abstrakt) 131
9 Literaturverzeichnis 132
10 Anhang 141

Einleitung
Über Jahrtausende hinweg haben die Menschen die Kühe von Hand gemolken. Im
18. Jahrhundert sind erste Versuche des mechanischen Milchentzugs zu datieren.
Für das Jahr 1851 sind Versuche belegt, das Säugen des Kalbes mechanisch nachzuempfinden
(HERRMANN, 1996). Damit versuchte man den Melkprozess zu mechanisieren
und so naturnah wie möglich durchzuführen. Bei allen Versuchen
blieb zunächst das Problem des konstanten Vakuums an den Zitzenspitzen ungelöst.
Ein großer Entwicklungsschritt wurde von Dr. Alexander Shields 1885 mit
der Erfindung einer Maschine mit integriertem Pulsator erzielt. Der zweite entscheidende
Schritt gelang 1903 Alexander Gillies mit der Konstruktion des Zweiraumbechers.
Damit war in Verbindung mit dem Pulsatoreinsatz der Durchbruch
zu einer funktionsfähigen Melktechnik erreicht, nach deren Grundprinzip bis heute
alle Melkmaschinen arbeiten. 1927 wurde unter der Markenbezeichnung "Westfalia"
auf der Dortmunder DLG-Ausstellung eine Eimermelkmaschine mit patentiertem
Kolbenpulsator und Zweiraum-Melkbechern präsentiert (TRÖGER, 2003).
Für die Entwicklung funktionierender Rohrmelkanlagen erwies sich die von Alfa
Laval 1925 konstruierte Milchschleuse als bedeutend. Der erste Melkstand (Releaser)
wurde 1939 in Weihenstephan in Betrieb genommen. Mit dieser Entwicklung
erfolgte eine räumliche Trennung des Melkens vom Haltungsbereich der
Kühe.
Mit Beginn der 50er Jahre konzentrierten sich technische Neuerungen auf die
Weiterentwicklung der Melkstände. Der Vorteil von Gruppenmelkständen (Fischgrätenmelkstand,
Side-by-Side-Melkstand) gegenüber Eimer- und Rohrmelkanlagen
wurde erkannt. In Gruppenmelkständen werden die Tiere gleichzeitig gemolken
und verlassen gemeinsam den Melkstand. Mitte der 80er Jahre begann die
weitere Automatisierung der Melktechnik. Dies fand zusammen mit der Entwicklung
von Mikroelektronik und Sensorik in den 90er Jahren mit den ersten Prototypen
der automatischen Melksysteme eine konsequente Fortsetzung.
Mit der Einführung des maschinellen Melkens kam es unabhängig von den technischen
Verbesserungen und Weiterentwicklungen zum verstärkten Auftreten von
Euterentzündungen. Dies führte in den 30er Jahren wiederholt zur Stagnation im
Absatz der Melkmaschinen (HERRMANN, 1996).
Ursachen für das noch heute bestehende Problem der schlechten Eutergesundheit
sind vielschichtig, jedoch auch in der Melktechnik zu suchen. In diesem Zusam-
1 Einleitung
2
menhang spielen die Anpassungsfähigkeit der Melkzeuge an verschiedene Euterformen,
die Zitzengummis und das Vakuumverhalten in Melkanlagen eine entscheidende
Rolle. Die stetige Steigerung der Milchleistung der Kühe stellt zusätzliche
Anforderungen an die Melktechnik.
Es bedarf somit weiterer Entwicklungsschritte, um zu einem euterschonenderen
Melken zu gelangen.

1 Einleitung 1
2 Literatur 2
2.1 Ursachen und Auswirkungen von Stress während des Melkens 2
2.1.1 Stressdefinition 2
2.1.2 mögliche Stressoren und betrachtete Stressreaktion 2
2.1.3 Bedeutung der Stressreaktion für die Milchabgabe 4
2.2 ausgewählte Parameter zur Stressermittlung 6
2.2.1 Überblick 6
2.2.2 Herzfrequenz 7
2.2.3 Blutdruck 8
2.2.4 Cortisolkonzentration in Körperflüssigkeiten, speziell im Speichel 10
3 Material und Methoden 14
3.1 Versuchsbetrieb 14
3.2 Tiere 15
3.3 Meßverfahren 16
3.3.1 Herzfrequenz 16
3.3.2 Blutdruck 17
3.3.3 Speichelcortisol 18
3.4 zeitlicher Ablauf der Versuche 19
3.5 statistische Methoden 22
4 Ergebnisse 23
4.1 Einflußfaktoren auf die Herzfrequenz 23
4.1.1 Melkvakuum 23
4.1.2 Strichverletzung 27
4.2 Betrachtung der indirekten Blutdruckmessung und Vergleich des
Blutdruckes mit der Herzfrequenz 29
4.2.1 spezielle Eigenschaften der indirekten Blutdruckmessung
nach dem oszillometrischen Prinzip 29
4.2.2 ungewohnte Situation 32
4.2.3 Melken 32
4.3 Darstellung des Verlaufs der Cortisolkonzentration im Speichel 35
4.3.1 Tagesrhythmik einer trockenstehenden Kuh 35
4.3.2 Routinemelken (44 kPa) 37
4.3.3 Melken mit erhöhtem Vakuum und Melken einer Kuh mit
Strichverletzung 38
5 Diskussion 39
6 Zusammenfassung 51
7 Literaturverzeichnis 53

Melktechnische Parameter zur Charakterisierung der Milchabgabe von
Kühen unter besonderer Berücksichtigung der Vor- und Hauptphase

Lutz Daßler

1. Einleitung und Aufgabenstellung 7
2. Stand der Erkenntnisse zur Milchbildung, Milchabgabe
und Milchgewinnung 9
2.1 Milchbildung und Milchgewinnung 9
2.1.1 Milchbildung, -speicherung und -abgabe 9
2.1.2 Physiologische Abläufe bei der Milchgewinnung 14
2.1.3 Maschinelle Milchgewinnung 19
2.1.3.1 Stimulationsverfahren 24
2.1.3.2 Pulsationsverfahren während der Hauptmelkphase 30
2.1.3.3 Nachmelkverfahren und automatisches Abschalten 35
2.2 Parameter der Milchflusskurve und deren Einflussgrößen 38
2.3 Zusammenhänge zwischen den Parametern der Milchflusskurve 50
2.4 Zellzahlen als Ausdruck der Eutergesundheit und deren
Zusammenhänge zu Milchflusskurven 54
3. Material und Methoden 62
3.1 Allgemeines 62
3.1.1 Erfassung der Milchflusskurven 62
3.1.2 Datenerfassung 66
3.1.3 Erfassung der Zellzahlen und Milchinhaltsstoffe 66
3.2 Untersuchungen von monatlich gewonnenen Milchflusskurven
im Prüfbetrieb mit Stimulations- und Nachmelktechnik 67
3.2.1 Angaben zum Prüfbetrieb 67
3.2.2 Datenaufbereitung und statistische Auswertung 69
3.2.3 Datenaufbereitung und statistische Auswertung der Rassen
Holstein und Angler 75
3.3 Vergleich der Milchflusskurven von Tieren eines Betriebes vor und
nach Wechsel der Melktechnik bzw. Änderung der melktechnischen
Einstellung 75
3.3.1 Datenaufbereitung und statistische Auswertung 76
3.3.2 Angaben zu den Prüfbetrieben 77
3
4. Ergebnisse 85
4.1 Ergebnisse von monatlich gewonnenen Milchflusskurven im
Prüfbetrieb mit Stimulations- und Nachmelktechnik 85
4.1.1 Effekt des Laktationsstadiums auf die Milchflusskurvenparameter 85
4.1.2 Effekt der Laktationsnummer auf die Milchflusskurvenparameter 88
4.1.3 Effekt der Saison auf die Milchflusskurvenparameter 91
4.1.4 Zusammenhang zwischen der Zellzahl und ausgewählten
Milchflusskurvenparametern 94
4.1.5 Zusammenhang zwischen ausgewählten Milchflusskurvenparametern
und der Zellzahl 97
4.2 Ergebnisse der Milchflusskurvenaufzeichnung von Tieren in den Prüfbetrieben
vor und nach Wechsel der Melktechnik bzw. Änderung
der melktechnischen Einstellung 102
5. Diskussion der Ergebnisse 117
5.1 Verlaufsuntersuchungen in einem Betrieb 117
5.1.1 Veränderung der Parameter der Milchflusskurve in den
Laktationsabschnitten 117
5.1.2 Veränderung der Parameter der Milchflusskurve über Laktationen 121
5.1.3 Veränderung der Parameter der Milchflusskurve in der Saison 124
5.1.4 Einfluss der Zellzahl auf die Milchflusskurve 125
5.1.5 Beziehungen zwischen Parametern der Milchflusskurve und
der Zellzahl 128
5.2 Einflüsse unterschiedlicher Melkverfahren 133
5.2.1 Einflüsse der Melktechnik unterschiedlicher Hersteller auf
die Parameter der Milchflusskurve 133
5.2.1.1 Einfluss des Wechsels eines Melkanlagenherstellers mit milchflussgesteuertem
Stimulationsverfahren zum Melkanlagenhersteller
mit zeitgesteuerter Vibrationsstimulation 133
5.2.1.2 Einfluss des Wechsels eines Melkanlagenherstellers mit Vibrationsstimulation
zum Melkanlagenhersteller ohne automatisches
Stimulationsverfahren 136
5.2.1.3 Einfluss des Wechsels eines Melkanlagenherstellers mit
Druckluftstimulation zum Melkanlagenhersteller mit milchflussgesteuertem
Stimulationsverfahren 139
5.2.2 Einfluss unterschiedlicher Einstellungen der Vibrationsstimulation
eines Melktechnikherstellers auf die Parameter der Milchflusskurve 140
4
5.2.3 Einfluss unterschiedlicher Einstellungen des Pulsationsverfahrens
eines Melktechnikherstellers auf die Parameter der Milchflusskurve 142
6. Zusammenfassung und Schlussfolgerungen 144
7. Summary and conclusions 147
8. Literaturverzeichnis 150
9. Tabellenverzeichnis 162
10. Abbildungsverzeichnis 167
11. Anhang

Einleitung und Aufgabenstellung
Sich verschärfende wirtschaftliche Rahmenbedingungen in der Milchproduktion führen zu
einem gestiegenen Interesse an der Mastitisvorsorge der Milchkühe. Dem Melken wird ein
Anteil von rund 50 % an der Entstehung von Euterkrankheiten zugeschrieben. Um diese
Risiken zu minimieren, sollte die Milchhergabe bei jedem Melken schonend, zügig und
weitestgehend vollständig sein (WORSTORFF, 2000). Dabei zwingen Fortschritte in der
Entwicklung der Melkanlagen und in der Rinderzucht dazu, die auftretenden Veränderungen
des Zusammenwirkens von Mensch, Tier und Maschine immer neu zu erfassen und zu
optimieren (TRÖGER, 1980). Obwohl diese Aussage schon 1980 getroffen wurde, ist sie
aktueller denn je. Bei der weitergehenden Technisierung und Automatisierung der Milchgewinnung
ist darauf Wert zu legen, dass zwischen der Technik und der Biologie, verkörpert
durch Mensch und Tier, aber auch umgekehrt, eine bessere Angleichung zu erreichen
ist. Im Vordergrund stehen dabei die physiologischen und hygienischen Anforderungen
beim Melkprozess (WENDT, 2000).
Auch POPP (1989) spricht von der Anpassung der Melktechnik und Melkroutine an eben
diese physiologischen Anforderungen. Hier sind die physiologischen Abläufe im Tier während
des Melkens unbedingt zu berücksichtigen. Eine mangelnde Melkbereitschaft zu
Melkbeginn ist zu einem späteren Zeitpunkt nicht mehr wett zu machen. Das Weglassen
von Routinearbeiten (Vormelken, Euterreinigung, Handstimulation) führt zur unmittelbaren
Beeinträchtigung der Milchabgabe und zu Einbußen in der Laktationsleistung.
Unterschiedlich lang anhaltende Milchabgabe der Einzelviertel führt zu Blindmelkzeiten
einzelner Viertel. Diese wiederum können Gewebeschädigungen bedingen, was ein ernst
zu nehmendes Infektionsrisiko darstellt. Außerdem hemmt Blindmelken die Milchabgabe
der noch melkenden Viertel und führt verstärkt zu Lufteinbrüchen. Diese Lufteinbrüche
können ein Übertragen euterpathogener Keime von Viertel zu Viertel eines Euters bedingen
(GÖFT, 1992).
Ausgehend von der Tatsache, dass künftig immer weniger Tiere mit hohem züchterischen
Wert immer intensiver geprüft werden müssen, sind Zusammenhänge zwischen Milchflussparametern
sowie der Melkbarkeit und Eutergesundheit im Hinblick auf Zellzahl und
Euterbehandlungen von Bedeutung. Im Zusammenhang mit der Leistungsprüfung von Tieren
der Rasse Deutsche Holstein wird dabei zentraler Punkt der Untersuchungen der Ein8
fluss der technischen Ausrüstung des Melkstandes (Melktechnikhersteller und deren melktechnischen
Besonderheiten, Automatiksysteme) auf die Milchflussparameter sein.
In der vorliegenden Arbeit werden neue und erweiterte Kenntnisse zum Verhalten mittels
LactoCorder gewonnener Milchflussparameter während der Laktation, über Laktationen
sowie saisonal dargelegt. Dabei wird im Besonderen auf die Unterschiede zwischen den
Ergebnissen mit und ohne Berücksichtigung der Gemelksmenge Wert gelegt. Neu ist dabei,
dass nicht über die gesamte Laktation auf Milchmenge korrigiert wurde, sondern nur
in den jeweils gebildeten Gruppen. Im Weiteren werden Erkenntnisse zu den Einflüssen im
Zusammenhang verschiedener Melktechnik und Arbeitsroutinen auf die Milchflussparameter
heraus gearbeitet, die über die bisherigen hinaus gehen. Ziel dabei ist es, den Kenntnisstand
nicht nur zu erweitern, sondern auch für künftige erweiterte Managementsysteme,
die ja Parameter der Milchflusskurve mehr oder weniger gut und in der Zukunft besser
erfassen können, zu ermitteln und welche gewinnbaren Daten aus den Milchmengenmessgeräten
zur Herdenbewirtschaftung genutzt werden können. Es sollen Aussagen getroffen
werden, wie stark der Einfluss der Milchmenge auf die Parameter der Milchflusskurve ist
und wann notwendige Korrekturen auf Milchmenge durchzuführen oder zu unterlassen
sind, damit klare Aussagen möglich und nicht verwischt werden oder sich ins Gegenteil
verkehren. Zudem soll aufgezeigt werden, welche Milchflussparameter zur Vorwarnung
von Eutererkrankungen verwendbar sind.
Dazu wird im ersten Teil der Arbeit untersucht, wie sich Milchflusskurven der Rassen
Deutsche Holstein und Angler, in einem Testbetrieb monatlich über 2,5 Jahre gewonnen,
in der Laktation und über Laktationen verändern.
In einem zweiten Teil sollen die Auswirkungen melktechnischer Änderungen auf die
Milchflusskurve dargestellt werden.
- Veränderung der Milchflusskurven der Kühe nach einem Wechsel des Melktechnikanbieters
- Einfluss der Melktechnik auf die Ausbildung der Milchflusskurve
Dabei werden in großen Praxisbetrieben vor und nach dem Umbau bzw. technischen Veränderungen
der Melktechnik Milchflusskurven aller melkenden Tiere erfasst und gegenübergestellt.
Resultierend daraus wird nachgewiesen, ob es Unterschiede bezüglich des Milchabgabeverhaltens
beim Melken von Kühen mit unterschiedlicher Melktechnik und bei Veränderungen
der melktechnischen Einstellungen gibt. Der Wert der Untersuchung besteht darin,
9
dass die Möglichkeit genutzt werden konnte, in mehreren Betrieben mit entsprechend großen
Tierzahlen vor und nach melktechnischen Veränderungen (Umbau, melktechnische
Einstellungen) unter weitestgehend gleichen Umwelteinflüssen in einer Herde, Milchflusskurven
aufzuzeichnen und gegenüber zu stellen.

Institute of Agricultural Engineering
University of Hohenheim
Field Livestock Systems Engineering
Prof. Dr. Thomas Jungbluth
Swiss Federal Veterinary Office
Centre for proper housing of ruminants and pigs
Prof. Dr. Beat Wechsler
Influence of automatic milking systems on
behaviour and health of dairy cows
Dissertation
Submitted in fulfilment of the requirements for the degree
„Doktor der Agrarwissenschaften“
(Dr.sc.agr. in Agricultural Sciences)
to the
Faculty of Agricultural Sciences
presented by
Dipl.-Ing. sc. agr. Isabelle Neuffer
Bad Homburg
2006
This thesis was accepted as a doctoral dissertation in fulfilment of the requirements for the degree
„Doktor der Agrarwissenschaften“ by the Faculty of Agricultural Sciences at University of Hohenheim
on 1st September 2006.
Date of oral examination: 6th October 2006
Examination Committee
Supervisor and Reviewer Prof. Dr. Beat Wechsler
Co-Reviewer Prof. Dr. Thomas Jungbluth
Additional Reviewer Prof. Dr. Michael Grashorn
Vice-Dean and Head of the Committee Prof. Dr. Werner Bessei
Copyright 2006
Im Selbstverlag: Isabelle Neuffer
Bezugsquelle: Universität Hohenheim
Institut für Agrartechnik -440-
Garbenstr. 9
70599 Stuttgart
Germany
Alle Rechte, auch der Übersetzung und des Nachdruckes sowie jede Art der photomechanischen
Weitergabe, auch auszugsweise, bleiben vorbehalten.
i
Table of contents
List of figures ........................................................................................iii
List of tables ......................................................................................... v
1 General introduction.......................................................................... 1
1.1 Composition of this thesis .................................................................................. 4
1.2 Methods to assess cow welfare........................................................................... 5
1.2.1 Behaviour during milking ................................................................................. 5
1.2.2 Heart rate and heart-rate variability.................................................................. 6
1.2.3 Teat-cup attachment and time need for milking processes ................................. 7
1.2.4 Udder health .................................................................................................. 7
1.2.5 Milk cortisol .................................................................................................... 9
1.2.6 Milking frequency............................................................................................ 9
2 Animals, materials and methods....................................................... 10
2.1 Milking systems, animals and farms .................................................................. 10
2.2 Measurements ................................................................................................ 12
2.2.1 Behaviour during milking ............................................................................... 12
2.2.2 Heart rate and heart-rate variability................................................................ 13
2.2.3 Teat-cup attachment accuracy and time need for different phases .................... 14
2.2.4 Somatic cell count and milk cortisol ................................................................ 15
2.3 Statistical evaluation........................................................................................ 15
3 Results........................................................................................... 19
3.1 Restless behaviour, heart rate and heart-rate variability ..................................... 19
3.1.1 Behaviour during milking ............................................................................... 19
3.1.2 Heart rate and heart-rate variability................................................................ 23
3.1.3 Influences of explanatory variables ................................................................ 25
3.2 Operational reliability and time need for milking processes.................................. 25
3.2.1 Teat-cup attachment success......................................................................... 25
3.2.2 Time need for milking processes .................................................................... 26
3.2.3 Milking interval / Milking frequency ................................................................ 27
3.2.4 Somatic cell count......................................................................................... 29
3.2.5 Influences of explanatory variables ................................................................ 29
3.3 Milk cortisol .................................................................................................... 30
4 Discussion ...................................................................................... 32
4.1 Restless behaviour, heart rate and heart-rate variability ..................................... 32
4.1.1 Restless behaviour ........................................................................................ 32
4.1.2 Heart rate and heart-rate variability................................................................ 33
4.2 Operational reliability and time need for milking processes.................................. 34
4.2.1 Teat-cup attachment success......................................................................... 34
4.2.2 Time need for milking processes .................................................................... 35
4.2.3 Somatic cell count......................................................................................... 36
4.2.4 Milking interval / Milking frequency ................................................................ 37
4.2.5 Influences of explanatory variables ................................................................ 38
ii
4.3 Milk cortisol .................................................................................................... 38
4.3.1 General differences between the two types of AMS and ATM............................ 38
4.3.2 Influences other than milking system ............................................................. 39
4.3.3 Differences between the two AMS models....................................................... 40
5 General discussion .......................................................................... 42
5.1 Parameters studied in this investigation............................................................. 42
5.1.1 Restless behaviour ........................................................................................ 42
5.1.2 Heart rate and heart-rate variability................................................................ 42
5.1.3 Teat-cup attachment..................................................................................... 43
5.1.4 Udder health ................................................................................................ 44
5.1.5 Milk cortisol .................................................................................................. 44
5.2 Other important issues in AMS.......................................................................... 45
5.2.1 Management ................................................................................................ 45
5.2.1.1 Herd size................................................................................................... 45
5.2.1.2 Low-ranked cows....................................................................................... 45
5.2.1.3 Cow traffic ................................................................................................ 46
5.2.1.4 Human-animal interaction........................................................................... 48
5.2.2 Animal health ............................................................................................... 49
5.2.2.1 Lameness.................................................................................................. 49
5.2.2.2 Reproduction............................................................................................. 49
5.2.3 Grazing........................................................................................................ 50
5.2.4 Technical aspects.......................................................................................... 50
5.2.4.1 Noise and vibration .................................................................................... 50
5.2.4.2 Forces applied to the teats ......................................................................... 51
5.3 Conclusions .................................................................................................... 51
6 Regulations .................................................................................... 52
6.1 Specific requirements concerning the use of AMS in Switzerland.......................... 52
6.2 Annotations to the requirements concerning the use of AMS in Switzerland (in German)
........................................................................................................... 53
Summary ............................................................................................ 56
Zusammenfassung............................................................................... 58
References .......................................................................................... 60
Acknowledgements .............................................................................. 75
Curriculum Vitae .................................................................................. 76
Lebenslauf .......................................................................................... 77
Erklärung ............................................................................................ 78
iii
List of figures
Figure 1: Swiss authorisation procedure for mass-produced farm animal housing systems. ............2
Figure 2: Design of AMS, exemplified by Lely Astronaut one-box AMS. 1: Entry and exit gates,
2: Concentrates dispenser, 3: Sensor to determine the cow’s position, 4: Robot arm
with: 4a: Laser sensor, 4b: Rotating brushes, 4c: Teat cups, 5: Computer-aided process
control. Other components (not shown): Milking machine, cleaning system, computer
with management software (Schön et al., 2000). ................................................................3
Figure 3: Stepping rates (per minute, Mean±StdErr) of individual cows during the entire milking
process in two different automatic milking systems (AMS-1, AMS-2) and autotandem
milking parlours (ATM) on four farms each. Horizontal bars and dashed lines
show mean values and standard errors per milking system, respectively.............................19
Figure 4: Foot-lifting rates (per minute, Mean±StdErr) of individual cows during the entire
milking process in two different automatic milking systems (AMS-1, AMS-2) and autotandem
milking parlours (ATM) on four farms each. Horizontal bars and dashed lines
show mean values and standard errors per milking system, respectively.............................20
Figure 5: Mean stepping rates per minute of individual cows in two different automatic milking
systems (AMS-1, AMS-2) and auto-tandem milking parlours (ATM) on four farms
each during the milking preparation phase (individual cows: black diamonds; milking
system: black line) and the actual milking phase (individual cows: open diamonds; milking
system: dotted line). ................................................................................................21
Figure 6: Mean (±StdErr) stepping rates per minute in the course of milkings lasting 7 minutes
of cows milked in two different automatic milking systems (AMS-1: diamonds, AMS-
2: circles) and auto-tandem milking parlours (ATM: triangles) on four farms each. ..............22
Figure 7: Mean (±StdErr) foot-lifting rates per minute in the course of milkings lasting 7 minutes
of cows milked in two different automatic milking systems (AMS-1: diamonds,
AMS-2: circles) and auto-tandem milking parlours (ATM: triangles) on four farms each. ......22
Figure 8: Mean heart rates (HR; per minute) of individual cows during resting (open symbols)
and milking (filled symbols) in two different automatic milking systems (AMS-1, AMS-2)
and auto-tandem milking parlours on 3 or 4 farms each. Thick bars show mean HR during
milking (thin bars: StdErr), dashed bars show mean HR during resting (thin dashed
bars: StdErr). ..................................................................................................................23
Figure 9: Mean SDRR (ms) of individual cows during resting (open symbols) and milking (filled
symbols) in two different automatic milking systems (AMS-1, AMS-2) and autotandem
milking parlours on 3 or 4 farms each. Thick bars show mean HR during milking
(thin bars: StdErr), dashed bars show mean HR during resting (thin dashed bars:
StdErr)...........................................................................................................................24
Figure 10: Mean rMSSD (ms) of individual cows during resting (open symbols) and milking
(filled symbols) in two different automatic milking systems (AMS-1, AMS-2) and autotandem
milking parlours on 3 or 4 farms each. Thick bars show mean HR during milking
(thin bars: StdErr), dashed bars show mean HR during resting (thin dashed bars:
StdErr)...........................................................................................................................24
Figure 11: Percentages of successful milkings per farm in two different automatic milking systems
(AMS-1, AMS-2) on four farms each. ........................................................................26
Figure 12: Percentage of finished preparation phases depending on the duration of the milking
preparation in two different automatic milking systems (AMS-1: diamonds, AMS-2:
circles) and in auto-tandem milking parlours (ATM: triangles; n=1550 milkings) on four
farms each......................................................................................................................27
Figure 13: Duration of milking intervals (in hours) in two different automatic milking systems
(AMS-1, AMS-2) and auto-tandem milking parlours (ATM) on four farms each
(Mean±StdErr of individual focal cows).............................................................................28
iv
Figure 14: Milking frequency (Mean±StdErr) in relation to days in lactation in two different
automatic milking systems (AMS-1, AMS-2) on four farms each..........................................28
Figure 15: Median somatic cell counts (in thousand, cells per ml) per cow in two different automatic
milking systems (AMS-1, AMS-2) and auto-tandem milking parlours (ATM) on
four farms each...............................................................................................................29
Figure 16: Milk cortisol concentration versus time of day (m = morning, a = afternoon) in two
different automatic milking systems (AMS-1, AMS-2) and auto-tandem milking parlours
(ATM) on four farms each (rows, resulting in a total of 12 farms). Data points represent
the individual milkings. AMS: black bars on time axis = time of morning and evening
milkings with inlayed box-plots based on the data in that period; thick black lines
= predictions of the model on daily periodicity without the behavioural variables;
thin black lines = loess smoother (local polynomial regression) based on all individual
milkings. .........................................................................................................................31
v
List of tables
Table 1: Characteristics of the studied farms.............................................................................11
Table 2: Definitions of investigated phases of the milking process ..............................................14
Table 3: Behaviour (frequencies per minute) during the different phases of the milking process
in different milking systems (Mean±StdErr) ...............................................................20
Table 4: Duration of different phases of the milking process (Mean ± StdErr, in sec) in two
different automatic milking systems (AMS-1, AMS-2) and auto-tandem milking parlours
(ATM)............................................................................................................................26
Table 5: Influences of the explanatory variables days of lactation, parity and daily milk yield
on the investigated parameters ........................................................................................30
1
1 General introduction
The introduction of an automatic milking system (AMS) on a farm substantially changes the
daily routines and relieves the farmer of the fixed milking hours. AMS also reduce the workload of
the dairy farmer (Sonck, 1995). In addition to economic aspects such as milk yield and productivity,
AMS aim at the improvement of animal welfare, animal health and udder health (Hamann,
2001). An increase in milk yield due to higher milking frequencies was aspired on AMS farms, as
changes from two to three, four or even five times daily milking in conventional milking parlours
lead to substantial increases in milk yield (Hillerton and Winter, 1992; Ipema and Benders, 1992;
Kruip et al., 2002; Royle et al., 1992). Under the conditions of fully automated milking, the cow is
offered the possibility to choose whether and when to be milked, which is more comparable to calf
suckling than being milked twice daily at fixed hours (Hamann, 1999). Being milked in AMS enables
the animals to control their daily rhythm themselves and corresponds to their physiological regulatory
mechanisms of three to five milkings and about seven eating bouts per day (Schön et al.,
2000). All AMS are equipped with concentrate dispensers, which motivate the animals to visit the
milking unit regularly (Prescott et al., 1998a; Ketelaar-de Lauwere et al., 1999a). The frequency of
visits to the milking unit may be enhanced by certain types of cow-traffic, e.g. by forcing the cows
to pass the AMS or a selection gate on the way from the lying to the feeding area. In the case of
forced cow traffic, access to the feeding area might be restricted, therefore disturbing the cows’
feeding behaviour (Ketelaar-de Lauwere et al., 2000).
AMS were subjected to the Swiss authorisation procedure regarding animal welfare for
mass-produced housing systems (Wechsler and Oester, 1998), as the cows are in contact with the
AMS without constant supervision by the farmer or another person. Stressful situations might occur
during the milking process, without the possibility for instant intervention by the farmer as given in
conventional milking parlours. As stress during milking not only impairs the welfare of the cows but
also has negative influences on milk ejection and subsequently increases residual milk (Zschöck et
al., 1998), it should be avoided, whenever possible.
The authorisation procedure, which was introduced in Switzerland in 1981 as part of the
Swiss Animal Welfare Act, is mandatory for all mass-produced housing systems and equipment for
farm animals. When asking for an authorisation, detailed documentation concerning the housing
system must be sent by the manufacturer or the importer to the Federal Veterinary Office. Authorisations
can only be granted if the housing system meets the requirements of the Swiss animal welfare
legislation. In case of experience with similar equipment or available literature, these are used
as a basis for the decision. However, practical testing may be required in some cases, which may
include veterinary, physiological and behavioural measurements.
Authorisations are given by the Federal Veterinary Office, which may consult an advisory
board consisting of experts in animal husbandry, animal housing construction, and animal protection
(Figure 1). Detailed information about given authorisations can be found at
http://www.bvet.admin.ch/stallliste (Wechsler, 2005). Temporary authorisations were granted in
2
2000 for two AMS models: Lely Astronaut and DeLaval Voluntary Milking System. These preliminary
authorisations allowed manufacturers and importers to sell and farmers to install the housing
system, thus enabling the Federal Veterinary Office to investigate AMS on commercial farms in
Switzerland. After the investigations, the preliminary and limited authorisations were replaced by
final authorisations for Lely Astronaut and DeLaval VMS in 2005.
Figure 1: Swiss authorisation procedure for mass-produced farm animal housing systems.
This work presents and discusses the results of a research project (2.01.06 BVET), which
were the basis for the decision taken by the Swiss Federal Veterinary Office whether and with
which specific requirements AMS may be used in Switzerland. The specific requirements, formulated
by Hauser and Wechsler (2005), form part of the aforementioned final authorisations and are
listed in chapter 6.1 (in English) and 6.2 (in German, with annotations). In the course of the Swiss
authorisation procedure described above, the welfare of cows milked in AMS was investigated. Ear3
lier studies raised the question whether being milked in an AMS is more or less stressful than being
milked in a conventional milking parlour. Various parameters, e.g. teat-cup attachment success,
aversive behaviour, and heart-rate variability were used to answer this question.
Technically, AMS are comparable to tandem milking stalls, with one milking stall (single-box)
or several milking stalls installed in a row (multi-box AMS), depending on the AMS manufacturer
and the AMS model. Teat location and consequential automatic teat-cup attachment which are crucial
for AMS functioning were made possible with the development of sophisticated sensors, leading
to further development of the first experimental models into marketable products. AMS combine
teat cleaning, teat-cup attachment, the actual milking and teat-cup detachment in one automated
process and do not require the constant presence of the farmer / caretaker. To enhance a
high number of cow visits to the milking box, all AMS are equipped with concentrate dispensers
(Prescott et al., 1998a). The principal components of a one-box AMS, i.e. milking stall, robot arm
and concentrate dispenser are shown in Figure 2.
Figure 2: Design of AMS, exemplified by Lely Astronaut one-box AMS. 1: Entry and exit gates, 2: Concentrates
dispenser, 3: Sensor to determine the cow’s position, 4: Robot arm with: 4a: Laser sensor, 4b: Rotating
brushes, 4c: Teat cups, 5: Computer-aided process control. Other components (not shown): Milking machine,
cleaning system, computer with management software (Schön et al., 2000).
4
For detailed descriptions of AMS models and AMS functioning, the following publications
might be consulted: Schön (2000, in German), KTBL (2005, in German) and Rossing and Hogewerf
(1997). AMS are becoming a widespread and accepted technology, and it is estimated that they
are in use on more than 2200 farms in over 20 countries worldwide. More than 80% of AMS are
located in Northern Europe (De Koning and Rodenburg, 2004). In 2005, the market leaders announced
more than 3000 (Lely) and 1000 (DeLaval) AMS to be sold worldwide (Lely, 2005; De-
Laval, 2005).
AMS were investigated in numerous studies, mostly covering technical and economical aspects.
As the research project presented here focused on the effects of AMS on animal health and
animal behaviour during milking, only topics that are directly connected with animal welfare during
the milking process are discussed in the following chapters.
To assess welfare, non-invasive methods were chosen due to the comparatively large number
of animals and farms and due to the fact that the animals were privately owned. In line with
other experimental approaches (Hagen et al., 2004; Hopster et al., 2002; Wenzel et al., 2003), a
combination of behavioural and physiological parameters was used. For each of the two AMS models
tested (Lely Astronaut, DeLaval VMS), four practical farms were visited. The control group consisted
of four farms with auto-tandem milking parlours. On each farm, the behaviour of the cows
as well as physiological indicators of stress, milk quality and accuracy of teat-cup attachment were
investigated for three consecutive days. Data was obtained using video observations of the milkings,
continuous heart-rate recordings, milk samples and additional information from management
programs of the AMS.
The experiments described in this thesis were designed in accordance with the ethical principles
and guidelines for scientific experiments on animals („Ethische Grundsätze und Richtlinien
für wissenschaftliche Tierversuche“) of the Swiss Academy of Medical Sciences (Schweizerische
Akademie der Medizinischen Wissenschaften, SAMW) and the Swiss Academy of Sciences
(Schweizerische Akademie der Naturwissenschaften, SANW) and approved by the Swiss Federal
Veterinary Office.
1.1 Composition of this thesis
In the context of the authorisation procedure for mass-produced farm animal housing systems,
the assessment of the welfare of cows milked in AMS was based on behavioural and physiological
parameters (chapter 1.3). Results (chapter 3) and discussions (chapter 4) are grouped in
three main sections each: the first section describes the behaviour of the cows in the different
milking systems and the implications of the different systems on heart rate and heart-rate variability
during milking and resting. Teat-cup attachment success and durations of the different phases
of the milking process are described in the second section, supplemented by data concerning differences
in somatic cell count in milk between the milking systems. Influences on milk cortisol content
are investigated in the third section. An overview comprising the results of the present inves5
tigation as well as other important aspects of AMS is given in chapter 5, supplemented by regulations
that must be applied by farmers using AMS in Switzerland (chapter 6).
1.2 Methods to assess cow welfare
In addition to the observations of the behaviour of the cows during milking, it was checked
whether milking in an AMS is associated with stress using physiological parameters. Different
methods have been used in stress research to gain information about the stress level of animals.
As cows react sensitively to changes in their environment, methods with as little influence on the
behaviour of the animals as possible had to be chosen for the present study.
1.2.1 Behaviour during milking
Especially before and during the milking process, excitation must be avoided, as stress hormones
inhibit complete milk ejection and udder emptying. Frequent defecation, stepping during
milking and kicking against the milking cluster are signs of excitation. During milking, a relaxed
state of the cow is wanted. Fear, pain and stress inhibit the milk removal (Zschöck et al., 1998).
Willis (1983) considered stepping, kicking and constant movement during milking to be indicative
of a stressful situation. Stepping increased, kicking and foot-lifting decreased during milking
in an unfamiliar room, compared to milkings in a familiar parlour (Rushen et al., 2001). Kicking
against the milking cluster was judged as a clearly aversive behaviour of the cows by Nuber
(1989), who used this parameter as an indicator of well-being during milking. The combination of
low milk flow and fully applied vacuum induced more frequent kicking, leading Nuber (1989) to the
interpretation that kicking was a reaction to pain. A reduction in flinching of the stomach muscles,
stepping and kicking during milking was achieved by handling the cows intensively around their
first calving (experimenter approached the cow regularly and allowed the cow to sniff his hands)
(Hemsworth et al., 1987; Hemsworth et al., 1989). Weight shiftings with the hind legs exceeding
40kg and leg lifting during milking were mostly induced by the AMS touching the teats and collisions
between the AMS and the legs, respectively (Ordolff, 1989). The latter was also reported by
Prescott et al. (1997). Additionally, kicking during milking could be elicited by flies, while general
unrest during milking is often observed in cows with impaired claw health (Hassall et al., 1993).
Kicking can lead to injuries of the cow, damage of the AMS and unsuccessful teat-cup attachment
(Bohlsen, 2000; Prescott et al., 1997). A reduction of unrest and aversive behaviour during milking
can be achieved by habituation to the AMS (Metz-Stefanowska et al., 1992; Kremer, 1993).
While concentrates are offered in all AMS during milking, this is not the case in most milking
parlours. Ceballos and Weary (2002) found that feeding concentrates in the parlour did not lead to
changes in elimination during milking. In a study conducted by Prescott et al. (1998b), no differences
in behaviour were found if food was offered in the AMS or not. Kicking was observed only
rarely. Prescott et al. (1998b) concluded that milking in AMS did not overly arouse the animals.
6
Half of all registered kicking behaviour was shown by a single cow, demonstrating the individuality
of the animals. Consistency in the responses of individual cows were found in behavioural responses
to udder preparation, i.e. kicking and stepping (Van Reenen et al., 2002), stepping during
milking (De Rosa et al., 2003), feeding and drinking patterns (Melin et al., 2005) as well as in reactions
to a novel environment test and a handling test (Hofmann, 2001; Hopster, 1998). Home-pen
activity and the reactivity towards a stressful situation were consistent and formed individual characteristic
traits in 35 Holstein-Friesian cows (Müller and Schrader, 2004; Müller and Schrader,
2005). A relatively high consistency in the responsiveness of individual heifers to a range of feareliciting
situations was also reported by Boissy and Bouissou (1995). Large percentages of the investigated
bulls and steers showed consistent behavioural reactions during restraint on different
days (Grandin, 1993).
1.2.2 Heart rate and heart-rate variability
In animal research, heart rate, which is relatively easy to measure, is often used to assess
stress (Hansen, 1999). In many studies, factors influencing heart rate were investigated. HR has
been widely used to judge the welfare of dairy cows (e.g. Hagen et al., 2005; Hopster et al., 2002;
Rushen et al., 2001; Wenzel et al., 2003). For example, higher heart rates were measured in dairy
cows milked by an aversive handler than in cows milked by a gentle handler (Rushen et al., 1999).
Compared with heart rate, heart-rate variability (HRV) allows a more detailed assessment of
animal stress. HRV is defined as the oscillation in the interval between consecutive heart beats
(Task Force, 1996). With certain parameters of HRV, the rapid oscillations in heart rate reflecting
vagal activity can be assessed noninvasively and be used as a physiological measurement indicative
for stress (Hansen, 1999; Porges, 1995). HRV is assumed to be a valuable physiological indicator
for stress load in animals (Mohr et al., 2002). For example, exposing rats to continuous noise
stress led to changes in HRV (Takeuchi et al., 2001). Reduced vagal activity was found in horses
subjected to waiting prior to concentrate feeding, compared to horses fed without waiting, indicating
a stressful situation (Hohmann et al., 2005). Simulating grooming in pigs led to an increase in
parasympathetic activity, indicating that grooming could contribute to the well-being of pigs (Hansen,
1999). In horses, mental stress during exercise was assessed using HRV measurements
(Rietmann et al., 2004). Other studies investigated changes in HRV due to stressful situations in
rats (Sgoifo et al., 1997), horses (Bachmann et al., 2003; Visser et al., 2002), pigs (De Jong et al.,
2000a; Geverink et al., 2002), dwarf goats (Langbein et al., 2003) and chicks (Korte et al., 1999).
The feasibility to use HRV as a physiological indicator for the assessment of stress in cattle
was demonstrated by Mohr et al. (2002) who observed lower HRV parameters in calves subjected
to heat stress, compared to calves with no obvious stress load. Assuming that cows in AMS experience
higher stress than cows in ATM, higher HR and lower HRV values, especially a reduction of
the parasympathetic activity, would be expected in the data of cows milked in an AMS.
7
1.2.3 Teat-cup attachment and time need for milking processes
The main technical difference between modern milking parlours and AMS is the automatic
teat-cup attachment. Depending on the AMS model, ultrasonic, laser or optical sensors are applied
for teat localisation (Artmann, 1997). The success of the teat-cup attachment process depends on
the reliability, accuracy and speed of the teat detection (Artmann, 2000), as well as on animal behaviour,
udder shape, hair coat of the udder and cleanliness of the teat detection system (Huschke
and Klimetschek, 2000). Besides technical and animal aspects, the importance of the exact programming
of the teat coordinates by the farmer during the first visits of a cow to the system can
not be underestimated.
Stefanowska et al. (2000) and Winblad von Walter et al. (1999) showed that omitted milkings
were associated with signs of unrest in the animals. Inconsistent milking processes with unpredictable
outcomes (milked / not milked) might disturb the cow’s preparedness to milking, as
cows do prefer consistent timing and can rather precisely anticipate forthcoming events. This
might result in sub-optimal milk let-down (Hurnik, 1994). A period of incomplete milk removal or of
extended milking intervals resulted not only in a temporary lower milk yield, but also in changes in
the composition of the secreted milk (Wheelock et al., 1965). In goats milked incompletely in one
gland and normally in the other, no effect on enzyme activities was found after two weeks of incomplete
milkings, but after 24 weeks the activities of several key enzymes were significantly lower
in this gland (Wilde et al., 1989). These results indicate a partial involution of secretory cells,
caused by a local chemical feedback in the gland and leading to reduced milk yield.
Failed milkings should be avoided not only for welfare but also for technical reasons, as reliable
teat-cup attachment is one of the factors determining the milking system capacity of an AMS
(Kaufmann et al., 2001). Other important influences on the milking system capacity are the durations
of different phases of the milking process. For example, an increase of the average handling
time (walking in and out of the system, identification, udder preparation and attachment of teat
cups) by 0.5 min lowers the milking capacity of the AMS by 5-8% (De Koning and Ouweltjes,
2000). Cows lingering in the AMS after milking decrease the milking capacity of the AMS (Oostra
and Sällvik, 2000a). Kremer (1993) and Nuber (1989) noted during their experiments with automatic
milking that cows did not leave the milking stall voluntarily, thus blocking the milking stall
until they were forced to leave. Besides variation in duration of milkings caused by the cow, systematic
differences in time need for teat cleaning and teat-cup attachment between Lely Astronaut
AMS and DeLaval VMS were found on two farms (Hvaale et al., 2002).
1.2.4 Udder health
Due to the lack of direct contact between animal and farmer during the milkings in AMS, udder
health is monitored by means of e.g. milk conductivity sensors, measurement of milk temperature,
detection of deviations in milk yield and colour sensors detecting blood in milk. Compared to
8
conventional milking systems, a delay in the detection of clinical and especially subclinical mastitis
may occur, leading to a delayed start of therapy and possibly a smaller proportion of healed animals.
A prolonged time span between the beginning of the infection and the diagnosis leads to
prolonged periods of pain and suffering and impairs the welfare of the cow (Hamann, 1999).
Hamann (2001) listed contagion by teat cleaning mechanisms and increased risk of contamination
by frequent usage of the same milking stall as health risks for cows milked in AMS. As
most of the routine contamination of the milk and the equipment originates from the teats of the
cows (Mottram, 1997), efficient teat cleaning is indispensable. Investigations concerning udder
health in automatic milking systems concentrated on various aspects, e.g.:
- Teat condition, which improved after changing to the AMS (Neijenhuis et al., 2004)
- Milk leakage, which can increase the risk for udder infections and mastitis, was observed more
often in AMS compared with a conventional milking system. In 39% of AMS-cows, milk leakage
was observed at least once, in contrast to 11% of cows in a conventional milking system
(Persson Waller et al., 2003).
- No differences in residual milk were found between AMS and conventional milking systems
(Hopster et al., 2002; Marnet et al., 2001).
- In all tested AMS (Lely, Fullwood, DeLaval, Westfalia), the pre-milking udder cleaning devices
adequately induced oxytocin release and milk ejection as a prerequisite for optimal udder emptying
during milking (Bruckmaier and Meyer, 2004).
Deficits of AMS that may lead to udder infections were presented by Petermann et al. (2000):
- Insufficient cleaning of udders and teats. Especially the teat tip is not cleaned sufficiently. No
adaptation of the cleaning intensity to the degree of contamination.
- Wet cleaning. Teats may be wet before the onset of teat-cup attachment.
- Teat cleaning is carried out before premilking. Cisternal milk rich in pathogens might mix with
alveolar milk.
- Recontamination of already cleaned teats might occur in some AMS models (as shown by Melin
et al., 2002).
- Teat cups that fall down or are kicked down are reattached without intermediate cleaning.
To support the mastitis detection devices of the different AMS models, all cows should be
tested for chronic subclinal mastitis on each quarter before installation of the AMS. This procedure
should be carried out as well for newly introduced heifers and newly calved cows. The continuous
control of the parameters offered by the management program, combined with visual and tactile
examination of the udders, is also very important for a good level of mastitis detection (Sällvik and
Oostra, 2002). Promising results concerning a so-called mastitis indicator, aiming at the development
of an online sensor array for an early detection system of udder infections in AMS, were presented
by Culina et al. (2005).
The importance of effective prophylactic measures to safeguard udder health in AMS was
shown by Pallas and Wendt (2001) who achieved a reduction in infected quarters by a steady disinfection
of the cleaning brushes and the teat cups in a Lely Astronaut AMS, assisted by a more
9
intensive cleaning of the cubicles. Udder health monitoring programmes have been installed in
Denmark (Rasmussen et al., 2002) and Germany (Brade, 2001a; Harms and Wendl, 2005a;
Knappstein et al., 2000; Labohm, 2001), the latter being compulsory for AMS farms.
1.2.5 Milk cortisol
The secretion of cortisol is an accepted indicator of stress, even though clear reactions occur
predominantly in acutely stressful situations. Under stress, the secretion of corticosteriods is increased,
mirroring the stress-induced activity of the hypothalamic-pituitary-adrenal axis (HPA axis).
In cattle, the most important corticosteroid is cortisol, both in plasma and in milk (Butler and Des-
Bordes, 1980).
If animals are chronically stressed, for example by a straining housing system, clear results
on cortisol responses may be difficult to obtain because the difference to the normal pattern of
cortisol secretion might be very subtle. Additionally, individual differences in cortisol responses, the
diurnal rhythm of cortisol secretion and the influence of the reproductive state of the female animal
must be taken into account (Thun et al., 1985). Despite these difficulties, changes in cortisol
secretion could be shown in animals subjected to chronic stress (Stuhec et al., 1984; Borell and
Ladewig, 1986; Ladewig, 1984).
Although the results of cortisol measurements in situations of chronic stress are not always
easy to interpret, this hormone is the most common and, if measured in milk samples, the most
convenient parameter to assess the activity of the HPA axis in cows.
1.2.6 Milking frequency
Barn layout seems to be a very important factor influencing the frequency in which cows
visit the AMS (e.g. Harms et al., 2001; Ketelaar-de Lauwere et al., 1998; Ketelaar-de Lauwere et
al., 2000). Free cow traffic, where no restrictions concerning the access of the cows to either AMS,
feeding alley or lying area are applied, might result in lower milking frequencies (Thune et al.,
2002). Guided or forced cow traffic, where cows are forced to pass the AMS on their way from the
lying to the feeding area lead to higher milking frequencies (Thune et al., 2002), but shorter eating
time, compared to free cow traffic (Hermans et al., 2003). In a study comparing free cow traffic
and semi-forced cow traffic with selection gates, the number of feeding visits was higher with free
cow traffic, but the total time spent eating did not change (Haverkamp et al., 2004). In conclusion,
the efficiency of a cow traffic system influences the total number of milkings and therefore system
capacity (Melin et al., 2006).
10
2 Animals, materials and methods
2.1 Milking systems, animals and farms
For the investigations, 12 farms were selected in Switzerland, each equipped with one of
two AMS models or an auto-tandem milking parlour (4 farms each). Lely Astronaut® (Lely Industries
N.V., Maassluis, The Netherlands), further called AMS-1, is a one-box system with a service
arm onto which the teat cups and the teat cleaning brushes are mounted. During the whole milking
process, the service arm remains under the cow. DeLaval Voluntary Milking System VMS®
(DeLaval International AB, Tumba, Sweden), further called AMS-2, is also a one-box system, but
here, the multipurpose arm fetches the teat cleaning cup and the teat cups in succession from a
mounting at the side of the milking stall. The teats are localised with a combination of laser and
optical systems, mounted on the service arm. During milking, the arm acts only as a support for
the milk tubes, staying beside the cow. To represent common modern milking technology, autotandem
milking parlours (ATM) were chosen because of their similarities to the AMS, e.g. individual
milking stalls, individual entry and leaving of the milking parlour. Three farms with an ATM were
equipped with a 2x2, one farm with a 2x3-milking parlour.
In both AMS models, teat cups are removed per quarter, whereas in the milking parlours,
the whole milking cluster is removed at the same time. In addition to the technical differences,
systematic differences between the milking systems were found in material and structure of the
floor of the milking stalls (AMS-1: metal flooring, AMS-2: rubber mat) and in cow traffic. As recommended
by the manufacturers, unrestricted access to feeding and lying areas (free cow traffic)
was granted to the cows in AMS-1, while in AMS-2 cows had to pass selection gates from the lying
to the feeding area. If a cow had exceeded a certain time limit, access to the feeding area was
only possible via the AMS (guided cow traffic with selection gates). Within the milking systems,
cow traffic and flooring were identical.
Farms with an AMS had been suggested by the manufacturers. On all farms, the milking system
had to be in use since at least six months prior to the investigations. All animals were housed
in free-stall barns equipped with at least as many cubicles as dairy cows. Due to the nature of the
investigation, various feeding and grazing regimes were found. However, there were no systematic
differences in feeding and grazing regimes between the milking systems.
20 lactating cows were chosen as focal animals from each herd with the assistance of the
farmer it was decided that as negative influences on the measured parameters attributable to extremely
nervous or sick cows are to be avoided. The goal was to select a cross-section of the herd,
representing cows of different ages and stages of lactation that were considered trouble-free during
milking and of good health. On two farms, only 18 and 16 cows fulfilled these criteria. Milking
frequency of focal cows was 2.5±0.1 milkings/day in AMS-1, 2.4±0.0 in AMS-2 and 2.00±0.0 in
ATM (Mean±StdErr). Of the original 234 focal cows, 9 had to be excluded from the statistical
evaluation due to lacking somatic cell count (SCC) data (needed in the statistical model) or due to
11
the fact that all milkings of these cows were either assisted by the farmer or failed (unsuccessful
teat-cup attachment). Altogether, behavioural data was collected of 1550 milkings of 225 focal
cows. For each cow, mean daily milk yield was calculated as sum of all milkings of the individual
cows divided by the exact duration each group was observed (Table 1).
Table 1: Characteristics of the studied farms
Cows used for behavioural observations
Milking systems
and farms
Number
of cows
Parity1 Days of
lactation1
Daily milk
yield1
Number of
milkings with behavioural
data2
Herd size
(lactating
cows)
AMS-1 A 20 3.8 ± 0.4 140 ± 22 17.5 ± 1.4 5 (1-7) 39
B 20 4.0 ± 0.5 174 ± 12 22.4 ± 1.8 9 (5-13) 49
C 20 2.7 ± 0.3 130 ± 16 30.7 ± 2.1 9 (6-14) 49
D 20 2.7 ± 0.3 149 ± 20 25.7 ± 2.4 9.5 (3-14) 47
AMS-2 E 16 2.7 ± 0.4 165 ± 26 23.5 ± 2.1 3 (2-4) 50
F 19 2.4 ± 0.4 184 ± 17 23.0 ± 1.3 7 (2-10) 47
G 20 2.9 ± 0.4 173 ± 23 23.4 ± 1.6 8 (4-11) 56
H 17 2.2 ± 0.3 177 ± 19 15.3 ± 1.3 8 (6-11) 30
ATM I 20 3.3 ± 0.4 91 ± 15 24.4 ± 2.1 6 (3-6) 26
J 15 2.4 ± 0.4 180 ± 21 24.1 ± 1.0 6 (5-6) 21
K 20 2.6 ± 0.3 204 ± 21 21.5 ± 1.6 6 (6-6) 43
L 18 2.4 ± 0.4 126 ± 13 23.7 ± 1.7 5 (3-6) 22
Cows used for HRV data collection
Milking systems
and farms
Number
of cows
Parity1 Days of
lactation1
Daily milk
yield1
Number of
milkings with
HRV data2
Number of
resting bouts
with HRV data2
AMS-1 A 0
B 6 4.3 ± 0.8 133 ± 13 29.3 ± 2.5 3 (2-5) 3 (3-3)
C 5 3.2 ± 0.5 105 ± 20 33.4 ± 4.2 3 (2-6) 3 (3-3)
D 10 3.0 ± 0.4 146 ± 10 25.9 ± 3.4 5 (1-12) 3 (3-4)
AMS-2 E 4 1.8 ± 0.3 92 ± 45 28.5 ± 9.5 2.5 (2-3) 3 (3-3)
F 4 2.0 ± 0.4 135 ± 39 27.2 ± 4.7 3.5 (2-5) 3 (3-3)
G 10 2.9 ± 0.4 169 ± 33 23.5 ± 2.4 3 (1-4) 3 (3-3)
H 8 2.0 ± 0.3 136 ± 29 17.8 ± 2.2 5.5 (3-9) 3 (3-3)
ATM I 7 4.3 ± 0.7 105 ± 24 24.6 ± 3.0 5 (3-6) 3 (3-3)
J 5 2.0 ± 0.6 162 ± 23 25.0 ± 2.0 3 (2-4) 3 (3-3)
K 8 2.1 ± 0.4 245 ± 24 19.0 ± 1.9 3 (2-4) 3 (3-3)
L 6 2.7 ± 0.7 165 ± 16 21.0 ± 1.2 2.5 (2-3) 3 (2)
1Mean±StdErr, 2Median (Min-Max)
For heart rate measurements 10 of the focal cows were chosen in each herd. Technical
problems, e.g. damaged belts or dried-out electrodes, led to an important loss of data, resulting in
a data set containing measurements from 73 cows kept on 11 farms (only 3 farms with AMS-1,
Table 1). On 11 of the 12 farms, only breeds selected for high milk yield (Holstein, Red Holstein,
Brown Swiss) were present. One farm with an ATM kept Swiss Simmental cows, representing a
dual-purpose breed.
12
2.2 Measurements
The investigation was carried out in autumn/winter 2001/2002, spring 2002 and
autumn/winter 2002/2003. Each farm was visited once during one of these periods. On each farm,
data was recorded on three successive days, thus attempting to collect data of at least six milkings
per cow assuming a minimum milking frequency of two milkings per day. Video cameras were installed
for the duration of the investigation. One camera was installed directly at the AMS, such
that udder and hind legs were visible. On the control farms, up to four cameras were installed in
the milking parlours, thus allowing simultaneous observation of all milking stalls. A program combining
the time signal from the video recorder with entered codes for the observed behavioural
elements and for the beginning and end of the different phases of the milking process was used
for analysis of video observations (ETHO, R. Weber, Agroscope FAT, Taenikon).
2.2.1 Behaviour during milking
Behavioural data consisted of the number of stepping, foot-lifting and kicking events occurring
during the period from the first tactile contact between the cow and the milking machine or
the milker, i.e. the beginning of udder cleaning, to the retrieval of the last teat cup. Movements
before and after this time span were ignored. Stepping was defined as shifting weight from one
hind foot to the other, lifting the foot less than 10 cm, in contrast to foot-lifting, where one hind
foot was lifted more than 10 cm. Kicking was defined as a faster and more purposeful movement
as foot-lifting, often directed at items around the cow. All behavioural variables were calculated as
rates (e.g. steps per minute) to allow comparisons between milkings of different lengths. No statistical
analysis of the parameter kicking was possible, as kicking occurred too rarely. Due to numerical
problems, foot-lifting was only evaluated for the entire milking. Three different approaches
were used to investigate the behaviour during milking:
- ENTIRE milking: The frequencies of all behavioural elements, i.e. total numbers of stepping,
foot-lifting and kicking, were divided by the total duration of a given milking process, consisting
of teat cleaning, teat-cup attachment and the actual milking phase.
- PHASES of milking: The rates of the recorded behaviours were analysed separately for the two
phases of the milking, i.e. the preparation phase (teat cleaning and teat-cup attachment) and
actual milking phase.
- MINUTES into milking: The rates of the behaviours were calculated for each minute, beginning
with the start of milking preparations. Analyses were carried out separately for milkings of a
length of 6 (n=312 milkings), 7 (n=338), 8 (n=245) and 9 minutes (n=182), as these represented
the majority of all milkings (AMS-1 64%, AMS-2 71%, ATM 76%). Incomplete last minutes
were discarded, so that in the analysis of milkings of 6 min length, the minutes 1 to 5
were included.
13
2.2.2 Heart rate and heart-rate variability
For HRV recordings, the chosen focal animals on each farm were equipped with a custom
made belt with three electrodes. Animals carried their belts for at least 12 hours before measurements
were recorded. Representative resting phases of the focal animals that were needed as a
baseline in the comparison with HRV values during milking were found by a time sampling method
(every 5 min during daylight hours, i.e. 8:00 h to 18:00 h). In these observations it was recorded
whether the focal animals were lying quietly in a cubicle.
The pulse signal as registered by the electrodes was processed by an amplifier and led to a
voltage-controlled oscillator, generating a frequency-proportional signal in the telephony spectrum.
Using low power device radio units, each sending on an individual frequency, the signals were
transmitted from the animals to a close-by vehicle. This allowed the simultaneous processing of
the data of the ten cows. With a phase locked loop circuit, the signal was reprocessed into its
original form of a voltage signal. Digitalisation of the signal was achieved with a standard interface
(Daq-Book 112, IOTech Inc., Cleveland, Ohio, USA) and at a sampling rate of 1 kHz (as recommended
by Hejjel and Roth, 2004). Direct current drifts originating in the hardware that may influence
the analysis were corrected with a low-pass filter 4th grade and a threshold frequency of 0.1
Hz. On the basis of DasyLab 5.5 (IOTech Inc., Cleveland, Ohio, USA), a Windows-based graphical
data acquisition software package, software to measure intervals between successive heartbeats
was developed. As the amplitude of the signal depended on factors such as movement of the cow
and wetness of the electrodes, a threshold value was continuously calculated. This threshold value
was calculated as the sliding median of the maximum amplitude over the last 60 heartbeats and
multiplied by 0.8. The time intervals between points when the threshold value was exceeded were
stored to the closest ms and used as estimates for beat-to-beat intervals.
Milking and resting intervals were selected based on the data of milking times and the time
sampling of lying, respectively. Only intervals exceeding 5 min in length were considered for further
analysis because data should be obtained from recordings of the same duration for the comparison
of time-domain measures (Task Force, 1996). Thus, intervals shorter then 5 min were discarded.
For milkings, the first 5 min of the milking were used (from the first contact between
milker / robot arm and udder onwards). Using Polar Precision Performance 2.0 (Polar Electro Oy,
Kempele, Finland), those intervals were chosen that had less than 5% errors in the single beat-tobeat
measurements and corrected using the algorithm provided by that software (validated by
Hopster and Blokhuis, 1994 and Marchant-Forde et al., 2004). Based on these restrictions, 1-12
milking intervals and 3-4 resting intervals remained per cow (Table 1).
Two HRV parameters were chosen for the analysis: the square root of the mean squared differences
of successive NN intervals (rMSSD) and the standard deviation of the NN interval (SDNN).
RMSSD takes into account short-term, high frequency components of HRV (Task Force, 1996) and
strongly reflects vagal tone (Kleiger et al., 1992; confirmed in calves by Després et al., 2002). In
14
contrast, SDNN reflects the overall HRV (Task Force, 1996). Besides rMSSD and SDNN, mean heart
rate (HR) was stored for evaluation.
2.2.3 Teat-cup attachment accuracy and time need for different phases
Definitions of the investigated phases of the milking process are presented in Table 2. Teat
cleaning and teat-cup attachment could not be recorded separately in the control system (ATM), so
that the two durations were summed up as milking preparations.
The analysis of the accuracy of the teat-cup attachment was based on the same video material
that was used for the analysis of the durations of phases of the milking process and the analysis
of the behaviour during milking. Each milking was judged as successful or failed. A milking was
declared as failed if the attachment process was not successful for all four teats, so that either
none or less than all quarters were milked. A milking was also declared as failed if all teat cups
were successfully attached in the beginning, but were kicked off or fell off directly afterwards, followed
by futile attempts of teat-cup reattachment. Failed milkings were separated in two classes:
failed due to technical problems, e.g. failure to detect the teats, and failed due to agitated behaviour
of the cow, which rendered teat-cup attachment impossible. During the investigation, the visits
of cows to the milking stall and milk yields were recorded automatically by the management
systems in the AMS. Data of the management programme concerning the success of milking procedures,
i.e. milkings marked with “X” in AMS-1 and so-called “incomplete milkings” in AMS-2,
were used to calculate percentages of failed milkings for the focal cows. In the ATM, milk yield was
also recorded automatically and stored in the management system. Mean milking frequencies were
calculated by dividing the total number of successful milking visits by the duration of the investigation
on a given farm.
Table 2: Definitions of investigated phases of the milking process
Phase Definition
Admission time
Time lag between the closing of the entrance gate and the onset of
milking preparations (first tactile contact between animal and milking
system or milker)
Milking preparation time First tactile contact between animal and milking system or milker
until successful attachment of all teat cups
Milking time End of teat-cup attachment until removal of the last teat cup
Leaving time Removal of last teat cup until all four legs are outside the milking
stall
Entire milking time Entering + Preparation + Milking + Leaving
15
2.2.4 Somatic cell count and milk cortisol
On the farms with AMS, milk samples were collected with automatic milk sampling units on
at least one day with the intention of collecting at least two samples per cow. Where conditions
allowed, milk samples were taken on up to three consecutive days. In ATM, milk samples were
generally collected on one day and the farmer’s usual routine for taking milk samples during milk
recordings was followed. Median somatic cell counts (medSCC) were calculated over all samples
for each individual cow. The Swiss Brown Cattle Breeders’ Federation, 6300 Zug, Switzerland, conducted
SCC analysis. A total of 1457 milk samples of uninterrupted milkings (median per cow: 3,
range: 1 to 7) of 468 cows (20 to 54 cows per farm) were collected for analyses. Of the focal
cows, 642 milk samples (median per cow: 3, range 1-7) were analysed for SCC.
To measure total milk cortisol (MCC, free and bonded cortisol), the cortisol in 0.5 ml
skimmed milk was extracted with 2 ml dichloromethane. 20 μl of this extraction was used in the
LIA kit for the cortisol measurement (LIA kit no. RE 620 11, IBL-Hamburg) which was read by an
MPL-1 luminometer. The inter-assay variation was 10.1%, the intra-assay variation 5.3%. For the
statistical analysis, measurements below the detection limit (0.44 nmol/l) were set arbitrarily at 0.4
nmol/l (77/1457 milk samples, 5.3%). The response variable was the measured CC corrected
based on one sample of known concentration that was run in each assay.
It has been shown that CC measurements in milk and blood correlate closely (Shutt and Fell,
1985; Verkerk et al., 1998). If milked for short intervals, MCC instantaneously reflects changes in
BCC (Termeulen et al., 1981). Whereas Shutt and Fell (1985) found a 1:1 relation for absolute levels
of free cortisol in blood and milk, they and others (Bremel and Gangwer, 1978; Termeulen et
al., 1981; Verkerk et al., 1998) found total MCC of roughly 1/6 of the values in blood.
2.3 Statistical evaluation
In all analyses except the analyses concerning failed milkings, only successful milkings (i.e.
with successful teat-cup attachment) that were completed without human assistance, e.g. manual
teat-cup attachment, were included for farms equipped with AMS. For all analyses, generalised
linear mixed-effects models were used (Pinheiro et al., 2005; Pinheiro and Bates, 2000; Venables
and Ripley, 2002) in R 2.1.1 (R Development Core Team, 2005).
To account for the repeated milkings of the cows and the potential dependency among cows
on each farm, milking system, farm and individual cow were included as hierarchically nested random
effects in all models. Using stepwise backward regression, variables with significant influences
on the discussed parameters were identified. For some response and explanatory variables transformations
(log: stepping, SDRR, milking preparation time, milking time, leaving time, medSCC,
MCC; sqrt: admission time) were necessary to meet statistical assumptions on the distribution of
the residuals. These assumptions (normality and homoscedasticity) were checked using graphical
displays of the residuals. Foot-lifting and kicking were such rare events that a generalized mixed16
effects model with a binomial distribution was used (function glmmPQL; Venables and Ripley,
2002). The results of these models provide information on the changes of the probability of a certain
event (e.g. the occurrence of foot-lifting during the milking). All mean values of behavioural or
HRV measures presented for the milking systems are based on farm means which in are in turn
based on mean values of the individual cows. These mean values are thus corrected for differences
in number of measurements.
Behaviour. Stepping was investigated for the three approaches described above (ENTIRE,
PHASES, MINUTES) and foot-lifting only for the entire milking. Data of 225 cows during 1550 milkings
was analysed (Table 1). The following explanatory variables were included: milking system
(AMS-1, AMS-2, ATM), parity, days of lactation, mean daily milk yield, duration of the milking
phase and medSCC. In the PHASES and MINUTES analysis, the phase and the minute, respectively,
as well as the interaction between milking system and phase or minute were added to the
statistical model. In the MINUTES analysis, 1077 milkings lasting between 6 and 9 minutes were
included.
HRV. Due to the great inter-individual variability in the absolute levels of HRV, data of a
given cow obtained during the visits to the AMS or the milking parlour were compared to data of
the same cow while resting. Thus, an additional factor ('situation') for the situations milking and
resting was included in the statistical model. In the statistical analyses, HR, rMSSD and SDRR
served as response variables in three different models. The milking system, the situation (milking /
resting) and the interaction of milking system and situation were included in the models as main
explanatory variables. Number of lactations, days of lactation, mean daily milk yield during the investigation
and medSCC were added as possible confounding effects.
Milking phases. Milking system, parity, days of lactation and mean daily milk yield during
the investigation were included as explanatory variables. 1550 milkings (AMS-1: 658, AMS-2: 487,
ATM: 405) of 225 cows were analysed (Median: 6 milkings / cow).
Milking intervals. The analysis was restricted to AMS farms, as milking interval was invariable
on ATM farms. In total, 1081 milkings (AMS-1: 610, AMS-2: 471) of 152 cows of 8 farms were
analysed (Median 7 milkings / cow). The following explanatory variables were included: milking
system (AMS-1, AMS-2), parity, days of lactation, mean daily milk yield, duration of the milking
phase and medSCC.
Somatic cell count. The investigation was based on milkings of which milk samples were
analysed (n=642 milkings of focal cows; AMS-1: 251, AMS-2: 214, ATM: 177). Number of lactations,
days of lactation and mean daily milk yield during the investigation were added as possible
confounding effects. With the Mann-Whitney-U-Test, differences between milking systems in several
questions concerning failed milkings and SCC were investigated.
17
Milk cortisol. Three main models were calculated that differed in their sets of explanatory
variables. The structure of all the models can be written as subsets of:
yijkl = μ + bi + bij + αk + { βl
βl⋅X1
} + β2⋅X2 + β3⋅X3 + β4⋅X4 + β5⋅X5 + β6⋅X6 + … + εijkl,
with μ, the intercept, the fixed effects αk, type of milking system (factor with k = 3 levels
and treatment contrasts: ATM, AMS-1, AMS-2), time of day (either βl, factor with l = 2 levels and
treatment contrast: morning and afternoon, or X1, continuous: seconds since midnight), X2, mean
daily milk yield (continuous), X3, median SCC (continuous), X4, median frequency of stepping (continuous),
X5, median frequency of leg lifting (continuous), X6, median frequency of kicking (continuous)
and …, all possible two- and three-way interactions, and the nested random effects bi,
farm and bij, the individual cow. Statistical assumptions are that εijkl ~ N (0, σ2) iid, bi ~ N (0, σ1
2)
iid and bij ~ N (0, σ2
2) iid (iid = independently identically distributed).
Milk cortisol - Simple approach. A simple model with only the type of milking system as
explanatory variable was calculated, because no clear daily pattern in MCC was obvious and thus
the question of a general difference between the systems could be addressed. 1457 milkings were
included in this analysis.
Milk cortisol - Complex approach. A model including milking system, time of day (morning
versus afternoon), daily milk yield, median SCC and their interactions was calculated including
all milkings in the ATM (which started between 5:00 and 6:00 in the morning and between 16:30
and 18:00 in the evening and lasted about one hour) and those milkings in the AMS that were observed
at the same time of day, i.e. between 4:00 and 8:00 in the morning and 16:00 and 20:00 in
the evening. 559 milkings (1 to 4 per cow) of 321 cows (14 to 40 cows per farm) on all 12 farms
were included in this analysis. In a further model, which also included the median frequencies of
steps, leg lifts and kicks as explanatory variables, 297 milkings (1 to 4 per cow) of 159 cows (8 to
20 of the focal cows per farm) fulfilled the temporal criteria.
Milk cortisol - Daily Periodicity. A model including milking system, daily milk yield, somatic
cell count, time of day (seconds since midnight) and their interactions was evaluated based
on all the data for the AMS farms to investigate the daily periodicity in more detail. 1008 milkings
(1 to 7 per cow) of 345 cows (26 to 54 cows per farm) on the 8 AMS farms were included in this
analysis. A similar model that also included the behavioural variables was restricted to 449 milkings
(1 to 7 per cow) of 153 of the focal animals (17 to 20 cows per farm).
To allow an a priori unrestricted smooth shape of the daily periodicity, natural splines were
used for the effect of time (Venables and Ripley, 2002, p. 228). The number of knots in the splines
necessary to represent the daily pattern was found by continuously increasing this number as long
as the increase resulted in a statistically significant improvement of the model. In the model including
all the data from the AMS farms, 9 knots were necessary. In the model including the behavioural
variables, it was found that a greatly increased number of knots would lead to a better
18
model fit (statistically significant improvement of the model). The fitted curve was very rugged,
though, and this was taken as an indication of over-fitting. Thus, the same number of knots was
used as in the model with all the data.
In some of the models heteroscedasticity was observed among the different milking systems.
Thus, a term that accounted for different variability per milking system was included in all
the models for consistency reasons, though it did not lead to a statistically significant improvement
in each model. Model selection started out with the models as described above. Interactions were
only included up to three-way interactions due to interpretability and the number of degrees of
freedom that were available with the sample size. It was then tested whether random terms for
fixed effects in addition to the intercept were needed (this is the same as saying whether interaction
occurred among the fixed and random effects). Then, the fixed effects were reduced until only
statistically significant effects and main effects that were still included in interactions remained in
the model.
In the reduced models (simple approach, complex approach without and with behavioural
variables, daily periodicity without and with behavioural variables), it was again checked whether it
was necessary to include random terms for fixed effects. Such terms were not necessary in any of
the models, and thus only a random term for the intercept was included in each of the models presented
here. In the models including the behavioural variables, it was not possible to include all
three-way interactions due to the available number of degrees of freedom. Thus, the same interactions
as in the model without the behavioural variables were allowed plus the main effects of the
behavioural variables. After the selection steps described above, it was checked whether the behavioural
variables showed any two-way interactions with each other or the other explanatory variables
or a significant three-way interaction among themselves.
19
3 Results
3.1 Restless behaviour, heart rate and heart-rate variability
3.1.1 Behaviour during milking
ENTIRE milking. Stepping during the entire milking process was observed in more than
95% of all milkings (AMS-1 95%, AMS-2 98%, ATM 96%). In AMS-2, the stepping rate was elevated
compared to the ATM, and ATM had slightly increased values in comparison to AMS-1
(F2,9=7.03, p=0.014, Figure 3). Foot-lifting was observed in 33% of all milkings (AMS-1 31%, AMS-
2 47%, ATM 18%). The probability of foot-lifting was highest in AMS-2, followed by AMS-1 and
ATM (F2,9=7.90, p=0.011, Figure 4). Mean kicking rate per minute was nearly zero in all milking
systems (Mean±StdErr: AMS-1 0.01±0.0, AMS-2 0.06±0.03, ATM 0.02±0.02). One or more kicks
were observed in only 9% of all milkings. In AMS-1, kicking was seen in 4% of all milkings,
whereas in AMS-2 and ATM, higher percentages were registered (13% and 10%, respectively).
Figure 3: Stepping rates (per minute, Mean±StdErr) of individual cows during the entire milking process in
two different automatic milking systems (AMS-1, AMS-2) and auto-tandem milking parlours (ATM) on four
farms each. Horizontal bars and dashed lines show mean values and standard errors per milking system,
respectively.
20
Figure 4: Foot-lifting rates (per minute, Mean±StdErr) of individual cows during the entire milking process in
two different automatic milking systems (AMS-1, AMS-2) and auto-tandem milking parlours (ATM) on four
farms each. Horizontal bars and dashed lines show mean values and standard errors per milking system, respectively.
PHASES of milking. An interaction between phase of the milking process and milking system
influencing the stepping rates was found (F2,1542=24.55, p<0.0001, Table 3, Figure 5). While
in AMS-1, only a slight increase in stepping rates was found from preparation phase to the actual
milking phase, a pronounced increase was observed in ATM. In AMS-2, generally higher values for
stepping were found in both phases, with a slight decrease from preparation to milking. Foot-lifting
and kicking during preparation and milking is also presented in Table 3, with slightly higher footlifting
frequencies both during preparation and milking in both AMS, compared to ATM. Kicking
frequencies were very low in both phases and in all milking systems.
Table 3: Behaviour (frequencies per minute) during the different phases of the milking process in different
milking systems (Mean±StdErr)
Milking system Stepping Foot-lifting Kicking
Preparation Milking Preparation Milking Preparation Milking
AMS-1 0.52±0.06 0.67±0.10 0.20±0.08 0.15±0.03 0.01±0.00 0.01±0.00
AMS-2 0.98±0.10 0.96±0.25 0.26±0.08 0.22±0.07 0.08±0.06 0.05±0.02
ATM 0.32±0.18 0.75±0.13 0.05±0.02 0.05±0.03 0.02±0.02 0.02±0.02
21
Figure 5: Mean stepping rates per minute of individual cows in two different automatic milking systems (AMS-
1, AMS-2) and auto-tandem milking parlours (ATM) on four farms each during the milking preparation phase
(individual cows: black diamonds; milking system: black line) and the actual milking phase (individual cows:
open diamonds; milking system: dotted line).
MINUTES into milking. The influence of the time into milking on stepping rate differed
between the milking systems. This was shown for milkings with a total duration of 6 min (Interaction
Minute:Milking system, F8,1236=2.03, p=0.0398), 7 min (Interaction Minute:Milking system,
F10,1675=3.41, p=0.0002) and 8 min (Interaction Minute:Milking system, F12,1452=2.15, p=0.0121).
Similarly looking time courses of the stepping rate were found for milkings of 6, 7 and 8 minutes
duration, so only milkings of 7 min length are shown in detail (Figure 6). In AMS-1, a rapid decline
in stepping rate during the first two minutes was observed, compared to an increase in stepping
rate in AMS-2 and ATM. In all three milking systems, the general tendency was a declining of the
stepping rate from the beginning to the end of the milking. In AMS-2, a higher level of stepping
was found during all but the first minute compared to AMS-1 and ATM, the latter two having a
similar time course during the second half of the milking. The rate of stepping in milkings of 9 min
length was significantly affected by the time into milking (F7,1267=2.39, p=0.0195). It increased
until the 6th minute and then declined. No significant effect of the milking system (F2,9=1.05,
p=0.39) and no significant interaction between the time into milking and the milking system were
found (Interaction Minute:Milking system, F14,1253=1.50, p=0.10). The time course of the rate of
foot-lifting during milkings of 7 minutes length differed between the milking systems (Figure 7),
with higher foot-lifting rates in both AMS during all minutes except the last, compared to ATM.
22
Figure 6: Mean (±StdErr) stepping rates per minute in the course of milkings lasting 7 minutes of cows milked
in two different automatic milking systems (AMS-1: diamonds, AMS-2: circles) and auto-tandem milking parlours
(ATM: triangles) on four farms each.
Figure 7: Mean (±StdErr) foot-lifting rates per minute in the course of milkings lasting 7 minutes of cows
milked in two different automatic milking systems (AMS-1: diamonds, AMS-2: circles) and auto-tandem milking
parlours (ATM: triangles) on four farms each.
23
3.1.2 Heart rate and heart-rate variability
In all three milking systems, higher HR were found when comparing resting to milking, with
lower HR both during resting and milking on farms with ATM. While AMS-1 and AMS-2 had comparable
values during resting, the difference to the values measured during milking was more pronounced
in AMS-2 (Interaction Milking system: Situation, F2,418=6.39, p=0.0018, Figure 8). SDRR
was lower in cows during resting compared to milking in all milking systems (Interaction Milking
system: Situation, F2,418=4.370, p=0.0132, Figure 9) with the largest difference between milking
and resting in AMS-1. In AMS-2 and ATM, the difference in SDRR between milking and resting was
comparable. Similar to the pattern observed in SDRR, rMSSD levels during resting and milking in
cows on farms with AMS-1 deviated in from those of cows kept on farms with the other milking
systems. While both AMS-2 and ATM were characterized by a markedly higher rMSSD levels when
comparing resting to milking, with the values of the measurements in both situations being on a
higher level in ATM, this difference was less pronounced in AMS-1 (Interaction Milking system:
Situation F2,418=9.61, p=0.0001, Figure 10).
Figure 8: Mean heart rates (HR; per minute) of individual cows during resting (open symbols) and milking
(filled symbols) in two different automatic milking systems (AMS-1, AMS-2) and auto-tandem milking parlours
on 3 or 4 farms each. Thick bars show mean HR during milking (thin bars: StdErr), dashed bars show mean
HR during resting (thin dashed bars: StdErr).
24
Figure 9: Mean SDRR (ms) of individual cows during resting (open symbols) and milking (filled symbols) in
two different automatic milking systems (AMS-1, AMS-2) and auto-tandem milking parlours on 3 or 4 farms
each. Thick bars show mean HR during milking (thin bars: StdErr), dashed bars show mean HR during resting
(thin dashed bars: StdErr).
Figure 10: Mean rMSSD (ms) of individual cows during resting (open symbols) and milking (filled symbols) in
two different automatic milking systems (AMS-1, AMS-2) and auto-tandem milking parlours on 3 or 4 farms
each. Thick bars show mean HR during milking (thin bars: StdErr), dashed bars show mean HR during resting
(thin dashed bars: StdErr).
25
3.1.3 Influences of explanatory variables
Influences of the explanatory variables days of lactation, parity, daily milk yield and SCC on
behavioural and HR/HRV response variables were scarce. Stepping during the entire milking increased
with higher parity (F1,211=4.85, p=0.029) and higher SCC (F1,211=3.90, p=0.049). For both
milkings of 6 and 7 min length, stepping frequency was higher in cows with an increased number
of days in lactation (6 min: F1,101=10.31, p=0.032; 7 min: F1,125=6.40, p=0.026). Both rMSSD and
SDRR were lower in cows with a higher daily milk yield (rMSSD: F1,61=6.92, p<0.0001; SDRR:
F1,61=8.29, p=0.0055).
3.2 Operational reliability and time need for milking processes
3.2.1 Teat-cup attachment success
A significant difference in the percentages of successful milkings was found between the
milking systems. In AMS-1, 97.8% of milkings were successful, compared to 93.5% in AMS-2. Excluding
milkings that failed due to the behaviour of the cow, 98.4% of milkings in AMS-1 and
94.3% of milkings in AMS-2 were successful (Figure 11; Mann-Whitney-U test: U=16, p=0.028 for
all milkings, U=16, p=0.029 without milkings failed due to agitated behaviour). At least one failed
milking process was observed in 11 % of the focal animals in AMS-1 and 31% in AMS-2 during the
investigation (Mann-Whitney-U test: U=2.5, p=0.14). In accordance with the results presented
above, which were based on the analysis of video recordings, the percentage of successful milkings
based on information stemming from the management programmes of the AMS differed significantly
between the milking systems (AMS-1 97.5%, AMS-2 89.7%, Mann-Whitney-U test: U=0,
p=0.028). From the time of entry until the time of exit of the cows from the milking stall, durations
of failed milkings ranged from 208 sec to 1276 sec (Median 377 sec) in AMS-1 (n=15) and from 47
sec to 1530 sec (Median 469 sec) in AMS-2 (n=36; Mann-Whitney-U test: U=10, p=0.68).
26
Figure 11: Percentages of successful milkings per farm in two different automatic milking systems (AMS-1,
AMS-2) on four farms each.
3.2.2 Time need for milking processes
Between the milking systems, no difference in the duration of the entire milking process was
found (F2,9=1.13, p=0.36). In AMS-2, the admission phase was longest, followed by ATM and AMS-
1 (F2,9=5.85, p=0.023, Table 4). Milking preparations were shortest in ATM, compared to AMS-2
and AMS-1 (F2,9=130.74, p<.0001, Figure 12). In 90% of all observed milkings, teat cleaning and
teat-cup attachment was completed after 121 sec (AMS-1), 198 sec (AMS-2) and 52 sec (ATM). In
extreme cases, milking preparations lasted up to 617 sec (AMS-1), 363 sec (AMS-2) and 144 sec
(ATM). The longest milking phases were recorded in ATM, whereas milkings in AMS-1 and AMS-2
had similar durations (F2,9=5.84, p=0.024, Table 4). No significant difference in the time needed to
leave the milking stall was found between the milking systems (F2,9=0.24, p=0.79, Table 4). In
90% of all milkings in both AMS models the cows left the milking stall within 30 sec, but extreme
durations of up to 219 sec (AMS-1) and 279 sec (AMS-2) were observed. In the ATM, 90% of the
cows left the milking stall within 61 sec, with a maximum duration of 285 sec.
Table 4: Duration of different phases of the milking process (Mean ± StdErr, in sec) in two different automatic
milking systems (AMS-1, AMS-2) and auto-tandem milking parlours (ATM).
Milking system Admission Milking
preparation
Milking Leaving time Entire milking
AMS-1 18 ± 0 89 ± 2 333 ± 17 21 ± 1 461 ± 16
AMS-2 22 ± 1 152 ± 3 304 ± 13 23 ± 1 501 ± 13
ATM 20 ± 2 32 ± 1 396 ± 12 28 ± 2 477 ± 12
27
Figure 12: Percentage of finished preparation phases depending on the duration of the milking preparation in
two different automatic milking systems (AMS-1: diamonds, AMS-2: circles) and in auto-tandem milking parlours
(ATM: triangles; n=1550 milkings) on four farms each.
3.2.3 Milking interval / Milking frequency
The cows in ATM were milked twice daily, whereas milking frequencies of 2.5 (AMS-1) and
2.4 (AMS-2) were observed in cows on AMS farms (Figure 13). No significant difference in the duration
of milking intervals was found between the two AMS models (F1,6=0.64, p=0.45, Figure 13).
On AMS farms, very short milkings intervals (<3h) were observed only rarely (1.6% of all milkings).
Mostly, these were caused by cows revisiting the AMS shortly after a failed milking (17 out
of 22 milkings with a milking interval <3h). The majority of milking intervals lasted between 6 and
12 hours (66% in AMS-1, 69% in AMS-2). While the proportion of milking intervals shorter than 6
hours was quite small (14% in AMS-1, 8% in AMS-2), a considerable portion of milking intervals
lasted longer than 12 hours (20% in AMS-1, 23% in AMS-2). Only in 4 animals, milking intervals
longer than 20h were observed. With one exception, these individuals were in late lactation. Cows
in the early phase of lactation (first 100 days) were milked more often in AMS-1 than in AMS-2,
whereas this pattern was inverse for cows in late lactation (Figure 14).
28
2
4
6
8
10
12
14
16
18
20
22
A B C D E F G H I J K L
AMS-1 AMS-2 ATM
Milking interval
Figure 13: Duration of milking intervals (in hours) in two different automatic milking systems (AMS-1, AMS-2)
and auto-tandem milking parlours (ATM) on four farms each (Mean±StdErr of individual focal cows).
Figure 14: Milking frequency (Mean±StdErr) in relation to days in lactation in two different automatic milking
systems (AMS-1, AMS-2) on four farms each.
29
3.2.4 Somatic cell count
In AMS-1 and in AMS-2, a tendency for higher SCC was found compared to ATM (F2,9=3.81,
p=0.06, Figure 15). In the different milking systems, average values for SCC were 200±97 (AMS-
1), 192±51 (AMS-2) and 96±41 (ATM, Mean±StdErr, in thousand). Higher percentages of cows
with a median SCC > 250.000 were found on farms with an AMS compared to farms with ATM.
This was confirmed using data of the focal cows (Kruskal-Wallis chi-squared=6.1729, df=2,
p=0.046) as well as for the entire herds (n=463 cows, Kruskal-Wallis chi-squared=6.1681, df=2,
p=0.046). Regarding cows with a median SCC > 100.000, such a difference between the milking
systems was found for the focal cows (Kruskal-Wallis chi-squared=6.0385, df=2, p=0.049), but
not on herd level (Kruskal-Wallis chi-squared=3.5, df=2, p=0.17).
Figure 15: Median somatic cell counts (in thousand, cells per ml) per cow in two different automatic milking
systems (AMS-1, AMS-2) and auto-tandem milking parlours (ATM) on four farms each.
3.2.5 Influences of explanatory variables
Influences of the explanatory variables days of lactation, parity, daily milk yield and SCC on
behavioural and HR/HRV response variables were scarce. Stepping during the entire milking increased
with higher parity (F1,211=4.85, p=0.029) and higher SCC (F1,211=3.90, p=0.049). For both
milkings of 6 and 7 min length, stepping frequency was higher in cows with an increased number
of days in lactation (6 min: F1,101=10.31, p=0.032; 7 min: F1,125=6.40, p=0.026). Both rMSSD and
SDRR were lower in cows with a higher daily milk yield (rMSSD: F1,61=6.92, p<0.0001; SDRR:
F1,61=8.29, p=0.0055).
30
Time need for neither admission, nor milking preparation nor leaving was influenced by days
of lactation, parity or daily milk yield (Table 5). Actual milking as well as entire milking duration
decreased with increasing number of days in lactation and were higher in cows with higher parity
and higher daily milk yield. SCC was lower in cows with a lower number of days of lactation and
increased with higher daily milk yield. In AMS, milking intervals increased with advancing lactation.
Shorter milking intervals were found in cows with higher daily milk yield.
Table 5: Influences of the explanatory variables days of lactation, parity and daily milk yield on the investigated
parameters
Days of lactation Parity Daily milk yield
Admission time n.s. n.s. n.s.
Milking preparation time n.s. n.s. n.s.
Milking time F1,210=11.38*** F1,210=13.38*** F1,210=7.78**
Leaving time n.s. n.s. n.s.
Entire milking time F1,210=9.20** F1,210=10.65*** F1,210=6.97**
Somatic cell count F1,214=17.60*** F1,214=6.64* F1,214=22.97***
Milking interval (AMS farms
only)
F1,142=32.11*** n.s. F1,142=22.15***
/: Parameter increases / decreases with increased days of lactation, parity or daily milk yield. Statistical
significance: n.s.: p>0.05, *p<0.05, **p<0.01, ***p<0.001.
3.3 Milk cortisol
Simple approach. Cows milked in the two types of AMS had slightly elevated estimated
MCC in comparison to cows milked in the ATM based on the statistical model (by a factor of 1.09
and 1.04 for AMS-2 and AMS-1, respectively), but this difference was far from significant
(F2,9=0.11, p=0.90).
Complex approach. If only milkings in the morning and evening were considered, i.e.
those between 4:00 and 8:00 and between 16:00 and 20:00, a decrease in MCC from the morning
to the evening milkings by a factor of 0.84 was found (F1,237=9.54, p=0.002, Figure 16). In the
ATM there was a slight tendency for cows with higher SCC to have lower MCC, whereas this relation
was negligible in cows milked in AMS-2 and positive in cows milked in AMS-1 (interaction:
F2,306=6.22, p=0.002).
In the model that included the behavioural variables, the reduction of MCC in the evening
compared to the morning milking was by a factor of 0.73, 0.46 and 0.95 for the ATM, AMS-2 and
AMS-1, respectively (interaction: F2,135=4.36, p=0.015). A similar interaction of milking system and
SCC was found as in the model without behavioural variables, i.e. in ATM, cows with an elevated
SCC had reduced MCC whereas these were increased in cows in AMS-2 and even higher in cows in
AMS-1 (interaction: F2,141=6.38, p=0.002). In addition, an interaction of the frequency of steps and
foot-lifting was found (F1,141=12.10, p=0.001), i.e. MCC was increased in cows that stepped or
31
lifted their feet more often but did not increase additively if both values were high in the same
cow.
Daily periodicity. Daily periodicity of MCC differed between the two types of AMS (interaction:
F9,645=3.25, p=0.001). In AMS-2, there was a clear early morning peak in MCC and a broader
late afternoon elevation, whereas MCC had several small peaks in AMS-1 with the highest MCC at
around noon (Figure 16). No other explanatory variable showed a significant co-variation with MCC
in this data set. The same interaction of time and milking system was found in the model that included
the behavioural variables (interaction: F9,278=2.67, p=0.005). The behavioural variables
themselves had no statistically significant influence on MCC. The estimated effect of time of day
showed a qualitatively similar pattern (not shown) as in the model without behavioural variables.
Figure 16: Milk cortisol concentration versus time of day (m = morning, a = afternoon) in two different automatic
milking systems (AMS-1, AMS-2) and auto-tandem milking parlours (ATM) on four farms each (rows,
resulting in a total of 12 farms). Data points represent the individual milkings. AMS: black bars on time axis
= time of morning and evening milkings with inlayed box-plots based on the data in that period; thick
black lines = predictions of the model on daily periodicity without the behavioural variables; thin black lines
= loess smoother (local polynomial regression) based on all individual milkings.
32
4 Discussion
4.1 Restless behaviour, heart rate and heart-rate variability
4.1.1 Restless behaviour
Between the milking systems, differences in the occurrence of stepping and foot-lifting were
found. Highest rates for stepping and foot-lifting during the entire milking were found in AMS-2,
with AMS-1 and ATM on a comparable level. On an absolute scale, the difference in stepping rate
was about 0.5 steps per minute, hardly indicating a severe impairment in animal welfare. No differences
in the total number of steps during milking between an AMS (Lely Astronaut) and an ATM
were reported by Hopster et al. (2002), whereas Wenzel et al. (2003) observed stepping behaviour
more often in an AMS (Lely Astronaut) than in an ATM. In contrast to the aforementioned studies,
Hagen et al. (2004) observed a comparatively high stepping rate of the cows in a herringbone parlour
(Steps per minute: 2.22±1.60, Mean±StdDev), in comparison to the stepping rate in ATM in
the present study. Hagen et al. (2004) supposed that cows in herringbone parlours might experience
some stress, as they stand uncomfortably close to each other during milking. With regard to
the stepping rate in the AMS (Lely Astronaut), however, the cows in the study of Hagen et al.
(2004) showed comparable values (Steps per minute: 1.01±1.14, Mean±StdDev) to the cows in
AMS-1 in the present study.
Compared to the herringbone parlour, less stepping during teat-cup attachment was observed
in the AMS in the study of Hagen et al. (2004). This is in contrast to the results of the present
study, where stepping rates hardly changed between the preparation phase and the actual
milking phase in both AMS models.
As in the present study kicking was rare (Wenzel et al., 2003) or not observed at all (Hopster
et al., 2002) in other comparative studies, so that no differentiation between the milking systems
was possible. In contrast to these results, higher kicking rates were observed in a herringbone
parlour, compared to an AMS (Hagen et al., 2004). Unfortunately, to my knowledge no studies
covering stepping or kicking rates in DeLaval VMS have been published.
When comparing this study with others, differences in definitions must be taken into account:
While foot-lifting was combined with kicking in two studies (Hagen et al., 2004; Wenzel et
al., 2003), it was counted as stepping in the other (Hopster et al., 2002). However, as foot-lifting
and especially kicking were rare events, no changes in results could be anticipated, should footlifting
be combined with either stepping or kicking.
Compared to the analysis of the behaviour during the entire milking, the detailed analysis of
the behaviour in the different phases of the milking process and in the adjacent minutes of the
milking did not lead to new conclusions, so that the additional effort in data analysis might be
spared in future investigations. Automation of the image analysis of the behaviour during the milking
process would be desirable, e.g. as attempted by Ipema et al. (2004). Differentiation between
33
the behaviours stepping, foot-lifting and kicking during milking did not contribute notably to the
conclusions either, as especially kicking occurred too rarely to be analysed statistically. However,
kicking remains an important parameter, as it clearly shows an aversive reaction of the cow.
4.1.2 Heart rate and heart-rate variability
Significant differences in HR between resting and milking, and between the milking systems
were found in this investigation. These differences depended on the milking system and were largest
in AMS-2, but still within the same range as data published by Hopster et al. (2002), who described
an increase of HR of roughly 10% after entrance into the respective milking system, compared
to premilking values. It must be noted, though, that premilking values in the studies of Hopster
et al. (1998, 2002) were recorded in cows standing in front of the milking parlour and the
AMS, respectively. In a study investigating side preferences of dairy cows in milking parlours, mean
HR increased 8.2±1.0% during entry into the milking parlour compared to the average HR during
the reference period (Hopster et al., 1998). As HR during standing is higher than during lying
(Hagen et al., 2005), the difference between resting and milking of about 15 to 20% as found in
this study seems to be comparable to the other results. Consistent with all aforementioned studies,
absolute HR values measured in this study represent normal reactions of cows being milked.
Previous studies comparing HR during milking in AMS and in milking parlours came to controversial
results. While Hopster et al. (2002) consistently measured lower HR in cows milked in
AMS compared to cows milked in an ATM, Wenzel et al. (2003) observed higher HR during milking
in an AMS than in an ATM. No differences between a herringbone parlour and an AMS-group in HR
during milking were found by Hagen et al. (2005). In the present study, differences in HR between
resting and milking varied depending on the milking system, with lowest values in both situations
in ATM. The differences in HR between AMS and ATM were possibly not caused by the milking system
as such, as Weiss et al. (2004a) investigated the changeover from a conventional parlour to
an AMS and found that HR above baseline during AMS milking differed only slightly from results
obtained in the parlour. Normalizing of the HR was observed within 10 visits to the AMS, indicating
a successful coping of the cows with the AMS. The milking frequency (2x or 5x a day) did not influence
the HR response during milking (Royle et al., 1992), either.
In the present study, consistent differences in HR and HRV parameters were found not only
between AMS and milking parlours, but also between the two different AMS models, indicating different
situations for cows kept on farms with AMS-1 and AMS-2. One reason might be the barn
layout, as differences in cow traffic were found not only between barns with ATM and AMS, but
also between barns with different AMS models. While cows in AMS-1 were kept in barns with free
cow traffic, selection gates were installed between the lying and the feeding area in barns with
AMS-2. The results of the present study suggest that free cow traffic might subject the cows to a
higher degree of strain as indicated by the observed differences in HR and HRV. Unfortunately, all
studies mentioned above were conducted with Lely Astronaut AMS only; studies concerning HR or
34
HRV of cows in DeLaval VMS are not published yet. Further studies are required to identify the
causes of the suggested differences in stress load, as the reaction of the cows to the milking process
itself does not seem to be different in the investigated milking systems.
In accordance with Hagen et al. (2005), rMSSD was lower during milking compared to resting,
and cows milked in a parlour had higher rMSSD during lying than cows milked in an AMS. Following
the arguments of Hagen et al. (2005), this difference does not originate from the milking
process as such, but from the increased metabolic activity while standing compared to lying, underlining
the need to consider the level of activity in HRV studies.
In contrast to the results regarding HR and rMSSD, the interpretation of SDRR measures
remains challenging. This is especially true for the results originating from cows AMS-1, where
higher SDRR was measured during milking, compared to resting. As shown by Hagen et al. (2005),
standing as well as milking leads to an activation of the cardiovascular system, accompanied by a
decrease in HRV measures. In the present study, differences in SDRR between resting and milking
were very small in AMS-2 and ATM, in contrast to AMS-1. The general level of SDRR was, however,
comparable to that observed in the study of Hagen et al. (2005), who found only a small difference
between standing and lying in SDRR and concluded that SDRR might not be a very useful parameter
for measuring autonomic activity. As other factors than the milking system may influence HRV
measures during resting, e.g. cow traffic and management, differentiating between theses potential
sources of stress is hardly possible. Further studies are needed to identify the exact sources of
stress in AMS barns to allow optimisation of the system and reduction of stress.
HR and HRV measurements, albeit flawed by a high proportion of data loss, demonstrated
the need to further investigate the effects of environmental variables on cows held in AMS barns.
Further studies, concentrating less on the AMS as such but more on the situation of the cow during
resting, could help to identify possible causes for the differences observed in these parameters
between the to AMS models studied and thus lead to improvements in barn management.
To summarise, cows in AMS-2 showed more stepping and foot-lifting during milking than
cows in AMS-1 and ATM. On the other hand, cows in AMS-1 had a reduced difference in HRV between
milking and resting, hinting at strain during resting. As the absolute differences measured
between the milking systems were rather small for all parameters, it is concluded that, although
statistically significant, these differences do not seem to indicate an impairment of the welfare of
cows milked in AMS.
4.2 Operational reliability and time need for milking processes
4.2.1 Teat-cup attachment success
Failed teat-cup attachments constitute a loss of time, which would otherwise be available for
milking, as the cow concerned has to leave and revisit the milking stall in order to get milked. The
percentage of failed teat-cup attachments differed between the two investigated AMS models, with
35
significantly higher percentages of failed teat-cup attachments in AMS-2. Percentages of successful
milkings in AMS-1 (Lely Astronaut) match the results of other studies (95-98%, Förster et al.,
1997; Huschke and Klimetschek, 2000; Wendl et al., 1997). Fübbeker and Kowalewsky (2005) estimated
the duration of failed milkings to be half the duration of a regular milking. In contrast to
this supposition, failed milkings in both AMS models took nearly as long as successful milkings in
this study, reducing the capacity of the AMS.
Teat-cup attachment success differed significantly between the two investigated AMS models.
Consistently higher success rates in AMS-1 demonstrated the technical feasibility to realise a
highly reliable teat-cup attachment process.
4.2.2 Time need for milking processes
Admission. The duration of the admission phase in Lely Astronaut was nearly identical in
different studies: 17±15 sec (Mean±SD, Hagen et al., 2004), 17±1.4 sec (Mean±SEM, Hopster et
al., 2002) and comparable to the results of this study. Similar results were measured in DeLaval
VMS (28 sec, Sällvik and Sällvik, 2002). While in this study the duration of admission phases were
nearly identical in all three milking systems, longer admission phases have been observed in an
auto-tandem milking parlour (60±8.9 sec, mean±SEM, Hopster et al., 2002) and a herringbone
parlour (77±54 sec, mean±SD, Hagen et al., 2004). These differences underline the influences of
the working routing of the milker as well as of the size and type of the milking parlour on the duration
of the admission phase.
Milking preparation. Milking preparation in DeLaval VMS took about 60 sec longer than in
Lely Astronaut. This large difference between the two AMS models was caused by substantial differences
in the technical approaches of teat cleaning and teat-cup attachment. Similar to our results,
Hvaale et al. (2002) observed a considerable difference of about 48 sec in time need for
milking preparations between Lely Astronaut and DeLaval VMS. In their study, teat-cup attachment
took on average twice as long in DeLaval VMS, compared to Lely Astronaut (50 sec vs. 23 sec).
The absolute durations of milking preparations measured in the present study were comparable to
the durations observed by others, both for Lely Astronaut (Huschke and Klimetschek, 2000: 81-85
sec; Hvaale et al., 2002: 86 sec) and DeLaval VMS (Hvaale et al., 2002: 134 sec; Sällvik and
Sällvik, 2002: 161 sec). With 108 sec (Mean±SD) and 111 sec (Mean±SEM), Hagen et al. (2004)
and Hopster et al. (2002) respectively, measured longer milking preparations in Lely Astronaut
AMS, compared to this study.
Compared to AMS, milking preparations in ATM are significantly shorter. Hopster et al.
(2002) measured similar durations for teat cleaning in an ATM and an AMS (31±2.9 sec vs. 45±2.9
sec, mean±SEM), but teat-cup attachment was much faster in the ATM than in the AMS (5±0.2
sec vs. 66±17.3 sec, mean±SEM).
36
Whereas Kremer and Ordolff (1992) observed a decrease in milk yield with increasing time
required for teat-cup attachment, Macuhova et al. (2004) found no negative effects of the duration
of an attachment delay and of long-lasting teat-cup attachments on milk yield, milk production
rate, average milk flow and milking time. However, a total interruption of the teat-cup attachment
process for either 2 (Bruckmaier et al., 2001) or 3 min (Rasmussen et al., 1992) led to increased
residual milk and reduced milk yield, respectively.
Milking phase. For the higher duration of the milking phase in ATM, several reasons can be considered.
The machine on time depends largely on the yield and the milk flow rate of the individual
cow (De Koning and Ouweltjes, 2000). Although Hagen et al. (2004) matched cows in the AMS
and the parlour group for milkability prior to the investigation, lower milk flow was recorded in the
parlour. This was interpreted as a negative reaction to the herringbone parlour with a prototype
brisketbar for exiting. Feeding during milking as practised in both AMS models, but not in ATM,
results in decreased milking time, increased udder evacuation and milk flow (Svennersten and
Samuelsson, 1992).
Leaving time. After about 90% of the milkings cows left both AMS models within a time span of
about 30 sec after detachment of the last teat cup, compared to about 60 sec in the ATM. No differences
in leaving time could be found in this investigation between AMS-1, which had no movement
inductor, and AMS-2 with the pressurized air jet. Other studies reported comparable leaving
times of 12±0.4 sec (Mean±SEM) for Lely Astronaut (Hopster et al., 2002) and 31 sec for DeLaval
VMS (Sällvik and Sällvik, 2002). Hagen et al. (2004) observed that the cows took 43±65 sec
(Mean±SD) from the opening of the exit gate until the cow had left the AMS. Valuable AMS capacity
is lost due to cows remaining in the AMS after the end of milking (Oostra and Sällvik, 2000a).
With a combination of an acoustic signal and a moving tube, thus a combination of conditioned
stimuli and light negative reinforcement, Oostra (2005) achieved a reduction in leaving time for
cows milked in an AMS. As Oostra (2005) emphasizes, negative experiences of cows in the AMS,
such as electrical shocks, might invoke negative associations and induce reluctance to enter the
AMS.
4.2.3 Somatic cell count
Significant differences in SCC between the two AMS models and the ATM were found in the
present study, confirming the results of Poelarends et al. (2004) and van der Vorst and de Koning
(2002) who found higher SCC in AMS on a large number of farms in the Netherlands, Denmark and
Germany, compared to farms with conventional milking parlours. Opposed to the aforementioned
studies, two other studies covering 12 United States dairy farms and 46 French farms found no
differences in SCC between AMS and conventional farms (Billon, 2002; Helgren and Reinemann,
2003). Elevated SCC observed in AMS farms in the months after transition disappeared in the
37
course of the first year with AMS (Billon, 2002; Wirtz et al., 2002). Kruip et al. (2002) reported significant
increases in SCC after changing from 2- or 3-times daily milking to automatic milking on 93
farms, while changing the milking frequency from 2- to 3-times daily alone did not induce a change
in SCC. Variable milking intervals as such did not influence SCC, either (Weiss et al., 2002).
4.2.4 Milking interval / Milking frequency
Although cow traffic design differed between the two AMS models, with free cow traffic in
AMS-1 and selectively guided cow traffic in AMS-2, mean milking intervals and thus the milking
frequencies did not differ significantly. This seems to contradict the results of other studies, which
compared different types of cow traffic and found higher milking frequencies in barns with selectively
guided cow traffic, compared to free cow traffic (Harms et al., 2002; Ketelaar-de Lauwere et
al., 1998; Ketelaar-de Lauwere et al., 2000; Thune et al., 2002). Harms et al. (2002) concluded
that selectively guided cow traffic combined most advantages of free and guided cow traffic, such
as high milking frequency, few visits without milking and few cows to fetch.
Comparing the milking frequency of cows in different states of lactation indicated that especially
cows in later lactation realised higher milking frequencies in barns with selectively guided
cow traffic, compared to free cow traffic. In accordance with these results, Haverkamp et al.
(2004) found a reduction of variation in milking frequency between cows when the cow traffic was
changed from free to selectively guided cow traffic, due to the regulation effects of the selection
gates.
On AMS farms with Lely Astronaut in the present study as well as in other investigations, a
considerable proportion of milking intervals lasted longer than 12 hours, e.g. 18% (Hogeveen et
al., 2001), 17.7% (De Koning and Ouweltjes, 2000) and 13% (Spolders, 2002). Even if cows were
visited three times daily by a herdsman who fetched cows with a long interval, almost 10% of the
cows had a milking frequency of 2 or lower in the study of de Koning and Ouweltjes (2000). Milking
intervals exceeding 12 h also accounted for a considerable proportion of all milkings (14%) on
a DeLaval VMS farm with selectively guided cow traffic (Melin et al., 2006).
Both too short and too long milking intervals might have negative influences on udder health
and milk yield. Long-term negative consequences on milk yield were found if cows were milked
three times in two days (Rémond et al., 1992) or once daily (Rémond et al., 1999) for several
weeks in early lactation, as these effects persisted when two times daily milking was resumed.
While very long milking intervals, as obtained by once-daily milking, lead to increased SCC compared
to twice daily milking (Stelwagen and Lacy-Hulbert, cited in Hogeveen et al., 2001), short
milking intervals might lead to deterioration of the teat tissue, thus increasing the risk of mastitis
(Ipema and Benders, 1992; Neijenhuis and Hogeveen, 2001).
38
4.2.5 Influences of explanatory variables
As duration of admission and milking preparations is mainly influenced by technical settings
of the milking system, influences of the explanatory variables days of lactation, parity, and daily
milk yield were not expected. The effects of the explanatory variables on the other response variables
milking time, leaving time, entire milking time, SCC and milking interval matched the results
of other studies. For example, the negative influences of increased SCC (Bartlett et al., 1990) and
low milking frequency (Rémond et al., 1999) on daily milk yield have been observed previously. As
in this study Wendl et al. (1997) observed no influence of parity on the milking frequency on an
AMS farm.
4.3 Milk cortisol
4.3.1 General differences between the two types of AMS and ATM
No general difference in MCC was found between farms with ATM and farms with AMS and
the general MCC level was low (mostly in the range of 0.25 to 4 nmol/l ! 1.5 to 24 nmol/l in
blood ! 0.5 to 8.5 ng/ml; conversion factor of 6 from milk to blood; conversion factor of 0.36
from nmol/l to ng/ml, Graham et al., 2002). This is in line with the findings of Weiss et al. (2004a),
who did not find changes in faecal cortisol metabolites during a change from a milking parlour to
an AMS, but contrasts with those of Hopster et al. (2002), who found systematically (but nonsignificantly)
elevated acute BCC during milking in a Lely Astronaut AMS (average values of up to
about 12 ng/ml in blood), as well as Wenzel et al. (2003), Hagen et al. (2004) and Abeni et al.
(2005), who demonstrated significantly elevated MCC in a Lely Astronaut AMS (average values of
up to 3 nmol/l in milk and 8 ng/ml in blood). Thus, the general differences in CC seem to be small
and thus most likely negligible regarding their biological relevance.
No general differences were found among the milking systems in how much MCC values decreased
from morning to afternoon milkings, either, which implies that there are no major differences
in the general MCC change throughout the day. Though Hagen et al. (2004) did not find
differences between morning and afternoon milkings, the data presented by Wenzel et al. (2003)
seem to indicate that differences between MCC in morning and evening milkings were smaller in
the ATM than in the AMS. This corresponds to the results of the present study, if behavioural variables
of restlessness are included in the model. In the respective model, the decrease in MCC was
greater in AMS-2 than in the ATM, while AMS-1 had similar values to the ATM. Thus, AMS-2 seems
to have a more pronounced change in the MCC throughout the day than can be observed in the
ATM and in AMS-1.
Most of the observed MCC (in the range of 0.25 to 4 nmol/l ! 0.5 to 8.5 ng/ml in blood)
are in a similar range or even lower than basal CC measured in blood and reported in earlier studies:
5 to 10 ng/ml (e.g. Thun, 1987; Lefcourt et al., 1993; van Reenen et al., 2002; Rushen et al.,
39
2001). Thus, there was no general indication of severely elevated CC on any of the dairy farms
investigated in this study.
Being milked seems to be a strain in itself, in that milking increases CC whereas being suckled
by a calf does not (Lupoli et al., 2001). In accordance with this, cows are willing to forgo a reward
if they can avoid being milked. Visits to a feeding automat drop to one third if cows are
milked there while being fed (Grimm et al., 1980). A possible aversive effect of milking is also reflected
in the increase of BCC during milking to about 10 to 25 ng/ml (van Reenen et al., 2002;
Rushen et al., 2001; Hopster et al., 2002), though machine milking results in lower values than
hand milking (Gorwit et al., 1992).
In the present study, MCC measurements (corresponding to values of 0.5 to 8.5 ng/ml in
blood) are well below the BCC reported for acute stress responses: e.g. 10 ng/ml after transportation
(Bremel and Gangwer, 1978), 20 ng/ml due to 2 h of immobilisation (Kaufmann and Thun,
1998), 20 to 25 ng/ml during milking in a novel environment (Rushen et al., 2001), 25 ng/ml in
low ranking breeds after regrouping (Mench et al., 1990), 40 ng/ml during peer separation (Boissy
and Le Neindre, 1997), 20 to 35 ng/ml during sham, freeze and hot iron branding (Lay et al.,
1992a,b), 15 ng/ml after negative handling and being faced with a tester (Breuer et al., 2003). The
CC measured in this study thus seem to be at a level that can be considered easily tolerable.
Measures of basal CC levels under standard housing conditions could be complemented by measurements
of the cortisol reaction to an acute stressor in future studies to detect changes in stress
reactivity (as e.g. in Bachmann et al., 2003).
4.3.2 Influences other than milking system
Hagen et al. (2004) investigated whether breed, heart rate, day of lactation and day of
pregnancy co-varied with MCC, but did not find statistical support for such relationships. In this
study, breeds were not well balanced between farms, but number of lactations and days of lactation
were similar between the different milking systems. Thus, an in-detail evaluation of these
variables was not attempted. Breed, parity and days of lactation were reflected in the random effects
of the individual cows.
Cows with higher MCC also tended to show more behavioural patterns, which are thought to
reflect unrest (stepping, leg-lifting, kicking). Thus, behavioural unrest during milking coincides with
elevated basal MCC levels. The simplest explanation might be that these are cows that are easily
excited in general, though for a conclusive relationship, the stress reactivity would be more meaningful
than the basal CC levels.
Cows with higher SCC had lower MCC in the ATM whereas the MCC of such cows was elevated
in the two AMS. Thus, it seems that MCC is increased with an additional strain such as SCC,
or that the animals that had elevated MCC in the AMS situation were more likely to develop high
SCC compared to cows in ATM. This would parallel the findings of Rushen et al. (1999, 2001), who
stressed cows during milking and observed increased residual milk yield which may in turn lead to
40
increases in SCC (Harmon, 1994). As average daily milk yield dropped with all the statistical models,
there was no indication that high-performing dairy cows reacted more strongly to the different
milking systems.
4.3.3 Differences between the two AMS models
Comparing the two AMS, the daily periodicity in the fitted model showed a clear early morning
peak in MCC and a broader late afternoon elevation in AMS-2 but did not show a clear pattern
in AMS-1. The elevations found in the daily periodicity of the cows milked in AMS-2 coincide well
with the elevations found during a night-morning and mid-day period in cycling heifers by Thun
(1987), though Lefcourt et al. (1993) only found one peak at about 6 a.m. in lactating cows. In
general, light seems to be a time marker sufficient to induce circadian patterns in cows (e.g. Thun,
1987), pigs (Minton et al., 1989; Gromadzka-Ostrowska et al., 1999) and horses (Irvine and Alexander,
1994). Thus, the absence of a clear daily periodicity in AMS-1 could be considered a deviation
from an expected circadian pattern and could be seen as an indication for stressful conditions
that disrupt the 'normal' circadian pattern (e.g. in cattle: Zähner, 2001; in pigs: Janssens et al.,
1995; de Jong et al., 2000b).
On the other hand, the circadian patterns of CC found in earlier studies with cattle were
rather weak. In studies that did find circadian patterns, these were detected by splitting time of
day into broad windows, thus comparing wide ranges of time, and they were mainly based on very
short and high peaks in CC that may have strongly influenced mean CC in these time windows
(e.g. Hays et al., 1975; Thun, 1987; Lefcourt et al. 1993). Even though CC is thought to be integrated
in milk over at least short periods of time (Verkerk et al., 1998), such peak concentrations,
where present, were likely to be damped in the data of this study.
It is hard to judge how far the circadian patterns of CC found in earlier studies with cattle
(Hays et al., 1975; Thun, 1987; Lefcourt et al., 1993) could have been induced by external time
markers additional to daylight, such as fixed milking or feeding times (Wilkinson et al., 1979; Saito
et al., 1989) or even the regularly repeated process of blood sampling itself. Thus, it might be hypothesised
that the absence of a strong daily periodicity as seen in cows in AMS-1 could reflect a
'natural' state given minimal restrictions of behaviour over time. An absence of circadian patterns
was also found in other studies with cattle (Hudson et al., 1975; Abebe and Scott, 1992) and goats
(Alila-Johansson et al., 2003) in which animals were reported to be kept relatively undisturbed.
Nevertheless, the common assumption currently seems to be that there exists a circadian
pattern in undisturbed cattle (e.g. Thun, 1987). Thus, it might be concluded that the dairy cows in
AMS-1 were subjected to a situation that was more stressful on the basis of the absence of a clear
daily periodicity. These findings in MCC are supported by the observation that heart rate at rest
was also more elevated in AMS-1. It is unclear what might have caused this stress, especially because
in the comparison between morning and afternoon milkings, which are often taken to reflect
daily periodicity, the focal cows kept in ATM also lacked a strong decrease in CC similar to the
41
cows in AMS-1. It can be hypothesised that the situation in the waiting area was more similar and
more stressful in some respect(s) for the cows kept in the ATM and AMS-1 in contrast to those
kept in AMS-2. This could potentially be due to the fact that in both ATM and AMS-1, cows are
herded together before milking (either the complete herd or the cows that are fetched as they
have not been milked for a long period of time) and that such enforced closeness leads to an increase
in stress.
Though Hopster et al. (2002) argued that a comparison of an AMS to an ATM is sufficient for
a judgement regarding animal welfare because the latter is 'widely accepted as meeting current
animal welfare standards', the question of which is the natural circadian pattern of cortisol remains,
because hormonal changes in cows in an ATM may be caused by strong time markers and
may not reflect an original state. This question could either be addressed by measuring the circadian
pattern in e.g. suckler cows or cows under (semi-)feral conditions or by experimentally disrupting
a clear circadian pattern with increased stress.
42
5 General discussion
5.1 Parameters studied in this investigation
5.1.1 Restless behaviour
Comparisons of the behaviour of cows during milking in different milking systems as indicators
of stress were done with contradicting results. No differences between an AMS and a milking
parlour in behavioural signs of unrest were found by Hopster et al. (2002). While Hagen et al.
(2004) detected more restless behaviour (stepping, kicking) in the milking parlour than in the AMS,
the present study led to oppositional results, confirming the results of Wenzel et al. (2003). In both
AMS of the present study, notably in DeLaval VMS, the cows stepped more, compared to the conventional
milking system. Equally, higher probabilities of foot-lifting were shown for DeLaval VMS,
compared to the other milking systems. Kicking was rare in all systems.
Although differences in behavioural parameters have been found between the systems,
these should not be overestimated. Especially on an absolute scale, the differences are small, indicating
perhaps tendencies, but not major differences in the welfare of cows during milking.
5.1.2 Heart rate and heart-rate variability
While Wenzel et al. (2003) observed a more pronounced increase in heart rate in AMS than
in the milking parlour, higher heart rates in the milking parlour than in the AMS were measured by
Hopster et al. (2002). No difference in heart rate during milking between automatic and conventional
milking was found by Hagen et al. (2004), in contrast to the results of the study presented
here, where higher heart rates during milking and resting were observed in both AMS, compared
to the milking parlours. Examining HRV parameters (rMSSD, SDRR) by comparing milking and resting
on individual basis also showed marked differences between the milking systems. Both in De-
Laval VMS and in the milking parlours, clear differences between milking and resting were apparent
in the HRV parameters as expected. However, in Lely Astronaut, this difference was comparatively
small, possibly indicating increased strain on the cows during resting. Hagen et al. (2005)
compared HRV parameters in cows milked in an AMS and a milking parlour, respectively. Indications
for a higher strain were found in the AMS-cows in the data recorded during lying and standing,
but not during milking, leading Hagen et al. (2005) to the hypothesis that this difference might
rather be a long-term effect of the entire system than a reaction to the milking process as such.
The results presented here also seem to indicate that cows in barns with an AMS could be subjected
to a more stressful situation, compared to cows in conventional barns with milking parlours.
Possibly, the combination of a limited resource (AMS) and limited space allowance is experienced
as a strain, especially by low-ranking cows.
43
5.1.3 Teat-cup attachment
Differences in proportions of successful teat-cup attachment were found between the two
AMS models, but also between the farms of the same model. Lely Astronaut generally had higher
percentages of successful attachment attempts than DeLaval VMS (Lely Astronaut: 97.8% (individual
farms: 96.2-99.5%), DeLaval VMS: 93.5% (individual farms: 91.5-95.2%); calculated on the
basis of individual milkings), confirming other results (Lely Astronaut: 96-97% (Förster et al.,
1997), 95-98% (Wendl et al., 1999), 95% (Hügle et al., 1999), 95-97% (Huschke and Klimetschek,
2000), 99% (Huschke, 2002); DeLaval VMS: 90-95% (Olofsson et al., 2001), 87% (Halm, 2003)).
It must be emphasized that in the present study only 20 animals per herd were studied who were
chosen in accordance with the farmer. By doing so, a cross-section of the herd consisting of
healthy animals with calm behaviour in the AMS was selected. It is reasonable to assume that the
percentage of failed milkings would have been higher than the numbers presented here if all cows
of a herd had been considered.
In summary, the technical functioning of the teat-cup attachment process of the tested AMS
models can be called satisfactory, but improvements are still needed. Although teat-cup attachment
has reached success rates greater than 90% on all farms included in this study, the results
imply that on certain farms almost 10% of all observed milkings failed and cows had to leave the
milking stall partially milked or without being milked.
Recently, the AMS manufacturers have presented improvements concerning hardware as
well as software. Therefore, higher proportions of successful milkings would probably be measured
if the study were conducted again. Generally, the proportions found in the present study match the
results of other studies. Nevertheless, other factors should not be neglected: the udder shape must
be compliant to the standards defined by the AMS manufactures. These standards describe the
technical possibilities of the AMS, e.g. the minimum height needed between udder and floor, so
that the robot arm can attach the teat cups correctly. In a survey comprising 18 farms equipped
with Lely Astronaut AMS, 0-16% (mean 6.8%) of the herd could not be milked in the AMS, mostly
because of very low udders (Fübbeker and Kowalewsky, 2005). Rather compact udder shapes as
needed for automatic milking might be called more natural (Förster et al., 1997). Udder conformations
change with elapsed time since the last milking (Miller et al., 1995), during lactation and with
parity, so that the suitability of a cow to robotic milking can only be judged for the given moment
(Bohlsen, 2000). In addition to udder shape, the exact programming of the teat coordinates into
the management system of the AMS is crucial for a satisfactory functioning of the teat-cup attachment
process.
44
5.1.4 Udder health
In the present investigation, somatic cell count was used as indicator of udder health. Although
a tendency to higher median somatic cell counts was found in the AMS compared to control,
no important differences were found between the milking systems.
The results of other studies on udder health of cows milked in AMS are contradictory. While
in some studies, higher SCC were observed in AMS (Knappstein et al., 2002; Kruip et al., 2002;
Schwarzer, 2000), others reported no major differences between cows milked in AMS and in milking
parlours (Ekman et al., 2004; Davis and Reinemann, 2003; Helgren and Reinemann, 2003;
Klungel et al., 2000; van der Vorst and de Koning, 2002; Wirtz et al., 2002; Zecconi et al., 2004) or
even a reduction of SCC in quarter milk samples in herds milked with AMS (Berglund et al., 2002).
Fübbeker and Kowalewsky (2005) investigated the yearly average SCC on 16 AMS farms: on 3
farms, the SCC was below 150.000, on 10 farms, the SCC was between 150.000-250.000, and on 3
farms, the yearly average SCC exceeded 250.000 cells/ml milk. Comparing the SCC before and after
installation of AMS on these farms showed a wide variation, ranging from a reduction in SCC of
47% to an increase in SCC of 82%. Fübbeker and Kowalewsky (2005) supposed that the influence
of the management is more important than the influence of the milking technology, as farms with
low SCC before the introduction of AMS succeeded in maintaining this status.
Because of the multiple differences between conventional milking systems and AMS besides
the milking frequency, factors such as the regularity of milking, the way of cleaning the teats (wet
or dry) and the health of the milk canal must be considered when comparing udder health in AMS
and milking parlours (Kruip et al., 2002). No negative influence of AMS on SCC is to be expected, if
initial cow health status and overall herd management is good (Zecconi et al., 2004). Improving
the herd management, especially the regular observation of cows in the barn, followed by udder
examinations if necessary, leads to a reduction in SCC and the occurrence of clinical mastitis
(Schwarzer, 2000). Compared to conventional milking systems, milk quality did not differ in AMS
barns if the animals were kept clean and the AMS were installed and adjusted correctly (Harms et
al., 2005).
5.1.5 Milk cortisol
Higher milk cortisol was measured in AMS cows, compared to cows milked in a milking parlour
in several studies (Hagen et al., 2004; Spolders, 2002; Abeni et al., 2005; Wenzel et al.,
2003). Hopster et al. (2002) found no difference in faecal cortisol metabolites between AMS and
milking parlour, but short-term increases in plasma cortisol during milking were more pronounced
in AMS.
In the present study, no consistent difference in milk cortisol concentrations between ATM
and the two AMS models was found. Based on these cortisol measurements, it is concluded that
milking cows in ATM or AMS did not result in biologically relevant stress differences in the cows. It
45
seemed that the daily periodicity was less pronounced in AMS-1 than in AMS-2, which may indicate
strain. The relationship between such a change in the circadian pattern of cortisol concentrations
and parameters of animal welfare needs further investigation.
5.2 Other important issues in AMS
5.2.1 Management
5.2.1.1 Herd size
The capacity of an AMS depends on the amount of time needed for different technical processes,
milk yield, mean milk flow, visits without milking allowance and system cleanings. Depending
on these variables, 50-70 cows can be milked in a one-box AMS (Schick and Moriz, 2003). The
higher the milk flow, the more milkings per day can be realised by a one-box AMS, ranging from
143 to 203 milkings per day (de Koning and Ouweltjes, 2000). Using a behaviour-based simulation
model, Halachmi et al. (2002) calculated an optimal herd size of 71 cows for a specific farm with a
one-box AMS. A reduction of the mean milking frequency in one-box AMS for herd sizes exceeding
42 cows was estimated by Artmann (2001). Assuming a mean duration of a milking visit of 8.5 to
11 min, about 5.5 to 7 milkings per hour are feasible which leads to about 150 milkings per day.
With a mean milking frequency of 2.6 milkings per day, 56 cows could be milked by one AMS
(Huschke and Klimetschek, 2000). For one-box systems, a maximum number of 50-55 lactating
cows per AMS is recommended (Wenzel et al., 2000). Following Brade (2001b), 55-65 animals may
be milked by a one-box system. Fübbeker and Kowalewsky (2005) found in a survey comprising 18
farms that about 50 (min 39 - max 65) lactating cows were milked in Lely Astronaut AMS. Maximising
the capacity could lead to herd sizes of 60 to 80 lactating cows per Lely Astronaut, depending
on the milking frequency (Fübbeker and Kowalewsky, 2005). Overloading the capacity of the AMS
induces increased workload and a reduced milking frequency (Bohlsen, 2000).
5.2.1.2 Low-ranked cows
Hamann (2001) expressed a concern that low-ranked cows might be disturbed on their way
to milking in AMS. Harms and Wendl (2005b) analyzed displacements at the feed gate to calculate
dominance values. At preferred times and in situations with guided cow traffic, lower-ranked cows
accessed the feeding area and the AMS less than higher-ranked cows. In an experiment by Ketelaar-
de Lauwere et al. (1996) involving a herd of 30 cows, the timing of visits to the AMS was related
to the dominance value. In two phases of the experiment, cows with higher dominance values
paid more visits to the AMS between 12:00 and 18:00h, while low-ranking animals seemed to
pay more visits between 0:00 and 6:00h. Hopster et al. (2002) confirmed that low-ranking animals
might, even when not stressed by queuing, still have difficulties in gaining access to the AMS. Re46
stricted access to the feed gate might result if the animals are kept in a system with forced cow
traffic. If the herdsman does not closely supervise these animals, serious welfare problems might
arise. In comparison to free cow traffic, a larger number of cows waiting in front of the AMS were
observed in the same barn with forced cow traffic (Hogeveen et al., 1998). As aggressive social
interactions are more likely to occur during waiting, Hogeveen et al. (1998) feared that longer
waiting may affect animal welfare and concluded that waiting should be avoided as much as possible.
Hopster et al. (2002) recommended the installation of selection gates, allowing the cows to
pass from the lying to the feeding area without passing the AMS.
Waiting time in front of a Lely AMS with free cow traffic was 10-15 min in a herd of 56 dairy
cows, which means that an average cow had to wait for 2 to 3 cows before she was milked herself.
Compared to the conventional milking parlour with waiting times up to 1.5 hours, AMS substantially
decreases the waiting time (Oostra and Sällvik, 2000b). The number of cows waiting to be
milked was influenced by the feeding regime (fresh food at the feeding alley and opening of the
gate to the pasture increased the number of waiting cows) and cleaning of the AMS (Bohlsen,
2000). Also, the number of cows in the milking queue and therefore the waiting time in AMS depended
on the type of cow traffic (Thune, 2000, cited in Wiktorsson et al., 2003, p.10; Hogeveen
et al., 1998). With an identical number of animals, lowest milking queues were observed in free
cow traffic (2.2 waiting cows), followed by forced traffic with preselection (4.1) and forced cow
traffic (5.1 cows). Similar increases in milking queues with increasing levels of restrictions were
observed by Harms et al. (2001). The social rank of the cow might have an influence on the waiting
times of the individual cow (Oostra and Sällvik, 2000b), as shown by Ketelaar-de Lauwere et al.
(1996) and Forsberg et al. (2002, cited in Wiktorsson et al., 2003, p.10). They demonstrated that
cows with a higher dominance value could enter the AMS more often without waiting. If they had
to wait, they did spend less time in the waiting area. Ketelaar-de Lauwere et al. (1996) suggested
that if a waiting area in front of an AMS is designed, a kind of safety zone should be included to
enable low ranking animals to withdraw from aggression or to leave the area to avoid long waiting
times.
5.2.1.3 Cow traffic
In AMS barns, different types of cow traffic can be implemented. Cow traffic is called free if
the cows can move from the lying area to the feeding area without having to pass the AMS. Although
free cow traffic allows the cows a high degree of freedom, there are problems with insufficient
attendance to the AMS, leading to a high number of animals that need to be fetched and
brought to the AMS (Harms et al., 2001). In barns with guided or forced cow traffic, a passage
through the AMS is compulsive to reach the feeding area from the lying area. To return to the lying
area, one-way gates are passed. Although this reduces the number of cows to be fetched, queues
in front of the AMS with long waiting times and a reduced number of daily passages to the feeding
47
area might occur, impairing the welfare of the animals by restricting the access to the feeding
area. The addition of selection gates (guided / forced cow traffic with preselection), which allow
the passage of cows without going through the AMS, eases the bottleneck represented by the
AMS. With increasing level of restriction in the cow traffic (free, forced with preselection, forced)
system, the number of milkings per cow increased and the number of feeding visits decreased
(Thune, 2000 and Forsberg et al., 2002, both cited in Wiktorsson et al., 2003, p.10; Harms et al.,
2001; Haverkamp et al., 2004). In a comparison of forced with free cow traffic in two herds of only
35 cows each, no differences in milking frequency were found (Munksgaard et al., 2002). The importance
of the herd size, which was low in the last-mentioned study compared to commercial
conditions, must not be underestimated. Ketelaar-de Lauwere (1992) concluded that passive selection
(free cow traffic) is preferable to active selection (forced cow traffic), as the cows spent less
time at the feeding gate during active selection. Forced cow traffic may be questionable, as it restricts
the cows’ behaviour (Ketelaar-de Lauwere, 1998). Guided cow traffic with selection gates far
from the AMS combines the advantages of both free and forced cow traffic (Harms et al., 2001).
The utilisation of the milking unit is influenced by the design of the barn, the selection gates
placement and function and the duration of the milkings. The access to the feeding area and to the
milking system must be clear and distinct to the cows. If guided cow traffic is implemented, the
selection gate has a central role in the flow of cows through the system. These gates should be
easily accessible and an obvious way to choose for the animals (Nordin, 2003), enabling cows
without milking allowance to pass directly to the feeding area (Benninger et al., 2000). Differences
were found in the acceptance of active and passive selection gates, the former being easier to understand
as they open if the cow is allowed to pass, whereas the latter must be pushed open
(Harms and Wendl, 2005c). To ensure a frequent use of a selection gate, training of the cows is
necessary. As was shown by Olofsson et al. (2001), less than 40% of the cows on 4 farms used
the selections gates regularly, in contrast to 60% on the fifth farm where the cows had been
trained. Differences in the acceptance of selection gates on 4 individual farms with the same AMS
(DeLaval VMS) model were also observed in this study: the percentages of cows which used the
gates at least once during the investigation presented here were 50%, 73%, 100% and 100%.
Approach routes and entrances to the milking unit should be clearly visible and distinguishable
from their surroundings in order to stimulate voluntary entry. Additionally, walkways, holding
units and the operation of the milking device itself must be comfortable and non-aversive to the
cow, e.g. no discomfort must be caused by the structure of the floor under the hind legs intended
to direct the hind legs into positions favorable for teat-cup attachment (Hurnik, 1992; Hurnik,
1994). Waiting areas in front of the AMS should be constructed with an exit to the lying area, allowing
cows to leave if they feel threatened for any reason or the waiting room is full. Otherwise,
low-ranked cows might be trapped for hours in a closed waiting area. Wiktorsson et al. (2003) additionally
expressed their concern that the risk of injuries of legs and hooves could rapidly increase,
if cows in heat enter a waiting area without an exit.
48
Complicated access to the feeding area might result in a low number of passages and queuing
of animals (Nordin, 2003). Lindström (2000) recommended that all cows should be provided
with sufficiently long eating times, preferably by constant access to roughage. Constant unhampered
access to feed, lying areas, water, minerals and cow comfort devices, e.g. rotating brushes,
was also recommended by Wenzel et al. (2000). Wiktorsson et al. (2003) stated that the combination
of high milk yield after calving and prevention from free feed intake was not ideal for managing
cows the first weeks after calving and included the risk of developing acetonemia.
5.2.1.4 Human-animal interaction
The installation of an AMS releases the farmer of the fixed daily interactions with his cows
during the milking hours, allowing the farmer new possibilities in the observation of the individual
cows (Ordolff, 1989). Less contact with humans might lead to fear of humans, which may limit the
productivity of livestock and even lead to injuries, if fearful animals try to avoid humans during
routine inspections and handling (Hemsworth, 2003). Reduced contact was stated by Spolders
(2002) to be the cause of the more difficult handling of the cows during blood sampling in the AMS
group, compared to the parlour group. An increase in confidingness of cows towards the farmer
was observed by 89% (16 of 18 farmers) after installation of an AMS (Fübbeker and Kowalewsky,
2005). Hemsworth et al. (1993) concluded that if animals are fearful of humans and thus the attitude
and behaviour of the stockperson towards the animals may be negative, the stockperson’s
commitment to the surveillance of and the attendance to welfare issues can be questioned. Breuer
et al. (2000) examined the behaviour of the stockpersons towards the cows, the behavioural response
to humans and the milk production on 31 commercial dairy farms. Their results suggest
that on farms with lower milk yield cows had higher avoidance distances than on farms where the
milk yield was higher, confirming the results of Seabrook (1972). Waiblinger et al. (2002) also
showed the influence of the attitudes and personal characteristics of stockpersons on stockpersoncow
interactions, e.g. avoidance distance, and milk yield. Positive behaviour towards veal calves
increased the production results (Lensink et al., 2000).
During the investigation presented here, much variance in flight distance and ease of handling
was noticed between herds (qualitative observations). Herds with animals that were mostly
calm and herds where most animals showed large flight distances were observed in both AMS
models. As the avoidance distance of cows towards an unfamiliar person in the stable reflects the
human-animal relationship on the farm (Waiblinger et al., 2003), and the behaviour did not seem
to be different between the two investigated AMS models, it can be assumed that the installation
of AMS as such does not necessarily lead to problems in the human-animal interaction. Good consistent
interaction with the animals, defined as a combination of psychological traits (e.g. low aggression,
emotionally stable) and actions of the stockpeople (e.g. use of voice in a kind tone), was
49
called the key to effective stockmanship by Seabrook (1994). Good stockmanship is necessary especially
in AMS barns to compensate the lack of regular human contact during the milking.
5.2.2 Animal health
5.2.2.1 Lameness
Lameness is a major welfare problem for dairy cows and related to the frequency with which
dairy cows visit an AMS (Borderas et al., 2004; Grove et al., 2004; Hamann, 2001; Van Lenteren
and Korsten, 2002), as lame cows reduce their daily activity levels (O’Callaghan et al., 2003). Prolonged
milking intervals and a reduced number of milkings per day were observed in lame cows in
8 AMS herds (Klaas et al., 2003). Nonskidding floors reduce claw injuries impairing regular visits to
the AMS (Benninger et al., 2000). To detect lameness automatically during milking, alternations in
stance by the cows trying to accommodate discomfort might be used. Neveux et al. (2004) used a
platform with individual load cells to detect weight distribution. A similar approach was done by
Ahokas et al. (2004), who installed 4 strain gauge scales in an AMS.
5.2.2.2 Reproduction
A higher milk production, induced by a higher milking frequency, might influence the energy
balance of the cows negatively and be followed by negative effects on fertility. However, only very
little negative effects on reproduction were found under experimental conditions, if the milking frequency
was not higher than 3 milkings per day (Kruip et al., 2000). A comparison of data of 87
AMS-farms and 440 conventional farms milking either two times or three times daily showed no
differences in the non-return rate at 56 days post insemination, but an increase in the number of
days to first service in AMS-farms (Kruip et al., 2002). As no difference in the number of days to
first service was found between the different milking frequencies on conventional farms, higher
milk yield can be excluded as a cause, leading Kruip et al. (2002) to suppose that differences in
time spent for oestrus detection were responsible for the result. Neither a delay in resumption of
ovarian cycle postpartum nor an increase in days open was observed in cows milked in AMS compared
to cows milked twice daily (Weiss et al., 2004b). Both delayed postpartum resumption of
ovarian cycles and lower conception rates were found in cows milked six times a day for six weeks
after parturition compared to cows milked three times a day (Bar-Peled et al., 1992). In the treatment
group with cows milked three times a day, a higher milk yield was measured, which was
partly realized by a higher dry matter intake and partly by a drain on body reserves. Similar results
were found by Ipema and Benders (1992), who compared 2, 3 and 4 times daily milking in terms
of milk production and body condition, amongst others. Lower body scores were found in the cows
milked 4 times daily, indicating that the increase in milk yield (+15%, compared to two times daily
milking) was partly realized by utilizing body reserves. In an opposed experimental design,
66
Janssens, C.J.J.G., Helmond, F.A., Wiegant, V.M., 1995. The effect of chronic stress on plasma cortisol
concentrations in cyclic female pigs depends on the time of day. Domest. Anim. Endocrinol.
12, 167-177.
Kaufmann, C., Thun, R., 1998. Einfluss von akutem Stress auf die Sekretion von Cortisol und Progesteron
beim Rind. Tierärztl. Umschau 53, 403-409.
Kaufmann, R., Ammann, H., Hilty, R., 2001. Wirtschaftlichkeit von AMS in Abhängigkeit der technischen
Kapazität und der baulichen Integration. In: Institut für Agrartechnik der Universität
Hohenheim (Ed.), Bau, Technik und Umwelt in der landwirtschaftlichen Nutztierhaltung. Universität
Hohenheim, Stuttgart, 6.-7.03.2001, pp. 248-253.
Ketelaar-de Lauwere, C.C., 1992. The use of a selection unit for automatic milking: consequences
for cow behaviour and welfare. In: Ipema, A.H., Lippus, A.C., Metz, J.H.M., Rossing, W.
(Eds.), Prospects for automatic milking. Proceedings of the International Symposium,
Wageningen, The Netherlands, 23.-25.11.1992, Pudoc Scientific Publishers, Wageningen,
pp. 270-277.
Ketelaar-de Lauwere, C.C., Devir, S., Metz, J.H.M., 1996. The influence of social hierarchy on the
time budget of cows and their visits to an automated milking system. Appl. Anim. Behav.
Sci. 49, 199-211.
Ketelaar-de Lauwere, C.C., Hendriks, M.M.W.B., Metz, J.H.M., Schouten, W.G.P., 1998. Behaviour
of dairy cows under free or forced cow traffic in a simulated automatic milking system environment.
Appl. Anim. Behav. Sci. 56, 13-28.
Ketelaar-de Lauwere, C.C., Hendriks, M.M.W.B., Zondag, J., Ipema, A.H., Metz, J.H.M.,
Noordhuizen, J.P.T.M., 2000. Influence of routing treatments on cows’ visits to an automatic
milking system, their time budget and other behaviour. Acta Agric. Scand., Sect. A, Animal
Sci. 50, 174-183.
Ketelaar-de Lauwere, C.C., Ipema, A.H., Metz, J.H.M., Noordhuizen, J.P.T.M., Schouten, W.G.P.,
1999a. The influence of the accessibility of concentrate on the behaviour of cows milked in
an automatic milking system. Netherlands J. Agric. Sci. 47, 1-16.
Ketelaar-de Lauwere, C.C., Ipema, A.H., van Ouwerkerk, E.N.J., Hendriks, M.M.W.B., Metz, J.H.M.,
Noordhuizen, J.P.T.M., Schouten, W.G.P., 1999b. Voluntary automatic milking in combination
with grazing of dairy cows. Milking frequency and effects on behaviour. Appl. Anim. Behav.
Sci. 64, 91-109.
Klaas, I., Rousing, T., Fossing, C., Hindhede, J., Sorensen, J.T., 2003. Is lameness a welfare problem
in dairy farms with automatic milking systems? Anim. Welfare 12, 599-603.
Kleiger, R.E., Stein, P.K., Bosner, M.S., Rottman, J.N., 1992. Time domain measurements of heart
rate variability. Ambulatory Electrocardiography 10, 487-498.
Klungel, G.H., Slaghuis, B.A., Hogeveen, H., 2000. The effect of the introduction of automatic milking
systems on milk quality. J. Dairy Sci. 83, 1998-2003.
Knappstein, K., Bräunig, J., Reichmuth, J., 2000. Rechtliche Rahmenbedigungen zur Milchhygiene
beim Einsatz von automatischen Melksystemen. In: Schön, H. (Ed.), Automatische Melksysteme.
KTBL-Schrift 395. Landwirtschaftsverlag, Münster, pp. 75-80.
Knappstein, K., Suhren, G., Clawin-Rädecker, I., Reichmuth, J., 2002. Erprobung automatischer
Melksysteme (AMS) unter den Gesichtspunkten ökonomischer, hygienischer und ökologischer
Aspekte, des Tierschutzes sowie struktureller Auswirkungen auf die Rinderhaltung. Teil
II: Aspekte der Eutergesundheit und Milchhygiene. Ber. Landwirtsch. 80, 281-303.
67
Korte, S.M., Ruesink, W., Blokhuis, H.J., 1999. Heart rate variability during manual restraint in
chicks from high- and low-feather pecking lines of laying hens. Physiol. Behav. 65, 649-652.
Kremer, J.H., 1993. Untersuchungen zum automatischen Milchentzug. Dissertation, Christian-
Albrechts-Universität Kiel.
Kremer, J.H., Ordolff, D., 1992. Experiences with continuous robot milking with regard to milk yield,
milk composition and behaviour of cows. In: Ipema, A.H., Lippus, A.C., Metz, J.H.M.,
Rossing, W. (Eds.), Prospects for automatic milking. Proceedings of the International Symposium,
Wageningen, The Netherlands, 23.-25.11.1992, Pudoc Scientific Publishers, Wageningen,
pp. 253-260.
Kruip, T.A.M., Morice, H., Robert, M., Ouweltjes, W., 2002. Robotic milking and its effect on fertility
and cell counts. J. Dairy Sci. 85, 2576-2581.
Kruip, T.A.M., Stefanowska, J., Ouweltjes, W., 2000. Robot milking and effect on reproduction in
dairy cows: a preliminary study. Anim. Reprod. Sci. 60-61, 443-447.
KTBL (Ed.), 2005. Automatisches Melken in modernen Milchviehställen. KTBL-Schrift 430. Landwirtschaftsverlag,
Münster.
Labohm, R., 2001. Automatische Melkverfahren in der Kritik. Grosstierpraxis 2:09, pp. 12-16.
Ladewig, J., 1984. Stresshormone als Indikatoren für Belastungssituationen. Aktuelle Arbeiten zur
artgemässen Tierhaltung. KTBL-Schrift 307. Landwirtschaftsverlag, Münster, pp. 200-205.
Langbein, J., Nürnberg, G., Manteuffel, G. 2003. Telemetric measurement of heart rate and heart
rate variability for evaluating psychological stress induced by visual discrimination learning in
dwarf goats. In: Cox, S. (Ed.), Precision Livestock Farming. Wageningen Academic Publishers,
Wageningen, pp. 105-111.
Lay Jr., D.C., Friend, T.H., Randel, R.D., Bowers, C.L., Grissom, K.K., Jenkins, O.C., 1992a. Behavioral
and physiological effects of freeze or hot-iron branding on crossbred cattle. J. Anim.
Sci. 70, 330-336.
Lay, D.C., Friend, T.H., Bowers, C.L., Grissom, K.K., Jenkins. O.C., 1992b. A comparative physiological
and behavioural study of freeze and hot-iron branding using dairy cows. J. Anim. Sci.
70, 1121-1125.
Lefcourt, A.M., Bitman, J., Kahl, S., Wood, D.L., 1993. Circadian and ultradian rhythms of peripheral
cortisol concentrations in lactating dairy cows. J. Dairy Sci. 76, 2607-2612.
Lely, 2005. Astronaut robotic milking system. www.lely.com. Accessed 01.01.2006
Lensink, J., Boissy, A., Veissier, I., 2000. The relationship between farmer’s attitude and behaviour
towards calves, and productivity of veal units. Ann. Zootech. 49, 313-327.
Lindström, T., 2000. Feeding behaviour in dairy cows. Motivational aspects. PhD-Thesis, Swedish
University of Agricultural Sciences, Uppsala, Sweden.
Lupoli, B., Johansson, B., Uvnäs-Moberg, K., Svennersten-Sjaunja, K., 2001. Effect of suckling on
release of oxytocin, prolactin, cortisol, gastrin, cholescytokinin, somatostatin and insulin in
dairy cows and their calves. J. Dairy Res. 68, 175-187.
Macuhova, J., Tancin, V., Bruckmaier, R.M., 2004. Oxytocin release and milk removal after delayed
or long-lasting teat cup attachment during automatic milking. Livest. Prod. Sci. 87, 237-244.
68
Marchant-Forde, R.M., Marlin, D.J., Marchant-Forde, J.N., 2004. Validation of a cardiac monitor for
measuring heart rate variability in adult female pigs: accuracy, artefacts and editing. Physiol.
Behav. 80, 449-458.
Marnet, P.G., Aupetit, H., Combaud, J.F., Portanguen, J., Aubry, J.M., 2001. Comparative study of
physiological responses of cows to classical versus automatic system of milking. In: Rosati,
A., Mihina, S., Mosconi, C. (Eds.), Physiological and technical aspects of machine milking.
ICAR Technical Series No. 7. Nitra, Slovak Republic, 26.-27.06.2001, pp. 49-54.
Melin, M., Hermans, G.G.N., Pettersson, G., Wiktorsson, H., 2006. Cow traffic in relation to social
rank and motivation of cows in an automatic milking system with control gates and an open
waiting area. Appl. Anim. Behav. Sci. 96, 201-214.
Melin, M., Wiktorsson, H., Christiansson, A., 2002. Teat cleaning efficiency before milking in De-
Laval VMS™ versus conventional manual cleaning, using Clostridium tyrobutyricum spores as
marker. In: McLean, J., Sinclair, M., West, B. (Eds.), Proceedings from The First North
American Conference on Robotic Milking, Toronto, Canada, 20.-22.03.2002, pp. II60-II63.
Melin, M., Wiktorsson, H., Norell, L., 2005. Analysis of feeding and drinking patterns of dairy cows
in two cow traffic situations in automatic milking systems. J. Dairy Sci. 88, 71-85.
Mench, J.A., Swanson, J.C., Stricklin, W.R., 1990. Social stress and dominance among group members
after mixing beef cows. Can. J. Anim. Sci. 70, 345-354.
Metz-Stefanowska, J., Huijsmans, P.J.M., Hogewerf, P.H., Ipema, A.H., Keen, A., 1992. Behaviour
of cows before, during and after milking with an automatic milking system. In: Ipema, A.H.,
Lippus, A.C., Metz, J.H.M., Rossing, W. (Eds.), Prospects for automatic milking. Proceedings
of the International Symposium, Wageningen, The Netherlands, 23.-25.11.1992, Pudoc Scientific
Publishers, Wageningen, pp. 278-288.
Miller, R.H., Fulton, L.A., Erez, B., Williams, W.F., Pearson, R.E., 1995. Variation in distances
among teats of Holstein cows: Implications for automated milking. J. Dairy Sci. 78, 1456-
1462.
Minton, J.E., Davis, D.L., Stevenson, J.S., 1989. Contribution of the photoperiod to circadian variations
in serum cortisol and melatonin in boars. Domest. Anim. Endocrinol. 6, 177-181.
Mohr, E., Langbein, J., Nürnberg, G., 2002. Heart rate variability. A noninvasive approach to measure
stress in calves and cows. Physiol. Behav. 75, 251-259.
Mottram, T., 1997. Requirements for teat inspection and cleaning in automatic milking systems.
Comput. Electron. Agr. 17, 63-77.
Müller, C., Herrmann, H-J., Knierim, U., Hesse, D., 2000. Bewertung der Tiergerechtheit von automatischen
Melksystemen. In: Schön, H. (Ed.), Automatische Melksysteme. KTBL-Schrift 395.
Landwirtschaftsverlag, Münster, pp. 132-136.
Müller, R., Schrader, L., 2004. Long-term consistency of home pen activity and separation stress in
individual dairy cows. In: Hänninen, L., Valros, A. (Eds.), Proceedings of the 38th International
Congress of the ISAE, Helsinki, Finland, p. 86.
Müller, R., Schrader, L., 2005. Individual consistency of dairy cow’s activity in their home pen. J.
Dairy Sci. 88, 171-175.
Munksgaard, L., Rushen, J., de Passillé, A.M., Krohn, C.C., 2002. Forced versus free traffic in an
automated milking system. In: McLean, J., Sinclair, M., West, B. (Eds.), Proceedings from
The First North American Conference on Robotic Milking, Toronto, Canada, 20.-22.03.2002,
pp. II51-II55.
69
Neijenhuis, F., Bos, K., Sampimom, O.C., Poelarends, J., Hillerton, E., Fossing, C., Dearing, J.,
2004. Changes in teat condition in Dutch herds converting from conventional to automated
milking. In: Meijering, A., Hogeveen, H., de Koning, C.J.A.M. (Eds.), Automatic Milking.
Wageningen Academic Publishers, Wageningen, pp. 141-147.
Neijenhuis, F., Hogeveen, H., 2001. Milking intervals and teat recovery. In: A. Rosati, Mihina, S.,
Mosconi, C. (Eds.), Physiological and technical aspects of machine milking. ICAR Technical
Series No. 7, Nitra, Slovak Republic, 26.-27.06.2001, pp. 211-212.
Neveux, S., De Passillé, A.M., Rushen, J., Weary, D.M., 2004. Changes in weight distribution to
detect hoof pain in dairy cows. In: Hänninen, L., Valros, A. (Eds.), Proceedings of the 38th
International Congress of the ISAE, Helsinki, Finland, p. 138.
Nordin, J., 2003. Utnyttjandemått i VMS – en nyckeltalsjämförelse av två grupphållningsstrategier.
(Utilization of VMS – A study comparing two housing systems with estimated key indicators).
PhD-Thesis, JTI – Institutet för jordbruks- och miljöteknik, Uppsala, Sweden.
Nosal, D., Rutishauser, R., Bilgery, E., Oertle, A., 2004. Lärm und Vibrationen als Stressfaktoren
beim Melken. Verursacher, Auswirkungen und Lösungsmöglichkeiten. FAT-Bericht Nr. 625,
Agroscope FAT Tänikon, Switzerland.
Nuber, B., 1989. Simulationsstudien zum automatischen Melken. Dissertation, Universität Hohenheim.
O'Callaghan, K.A., Cripps, P.J., Downham, D.Y., Murray, R.D., 2003. Subjective and objective assessment
of pain and discomfort due to lameness in dairy cattle. Anim. Welfare 12, 605-610.
Olofsson, J., Pettersson, G., Wiktorsson, H., Ekman, T., Sundberg, M., 2001. Teknisk provning av
DeLavals automatiska mjölkningssystem (VMS™). Rapport 249. Swedish University of Agricultural
Sciences, Uppsala, Sweden.
Oostra, H., Sällvik, K., 2000a. Exit times for cows milked in an AMS: Preliminary results. In: Hogeveen,
H., Meijering, A. (Eds.), Robotic Milking. Proceedings of the international symposium
held in Lelystad, The Netherlands, 17.-19.08.2000, Wageningen Pers, Wageningen, p. 65.
Oostra, H.H., Sällvik, K., 2000b. Waiting time for cows in a fully automated milking system. AgEng
2000, Warwick, Paper No. 00-AP-013.
Oostra, H.H., 2005. Technical and Management Tools in Dairy Production. Doctoral Thesis. Swedish
University of Agricultural Sciences, Alnarp.
Ordolff, D., 1989. Voraussetzungen und Grundlagen automatischer Milchgewinnung. Habilitation,
Universität Hohenheim.
Pallas, S., Wendt, K., 2001. The development of udder health of a dairy cow herd in an automated
milking system. In: Rosati, A., Mihina, S., Mosconi, C. (Eds.), Physiological and technical aspects
of machine milking. ICAR Technical Series No. 7. Nitra, Slovak Republic, 26.-
27.06.2001, pp. 175-179.
Persson Waller, K., Westermark, T., Ekman, T., Svennersten-Sjaunja, K., 2003. Milk leakage - an
increased risk in automatic milking systems. J. Dairy Sci. 86, 3488-3497.
Petermann, M., Wolter, W., Kloppert, B., Seufert, H., 2000. Aspekte der Eutergesundheit beim automatischen
Melken. In: Schön, H. (Ed.), Automatische Melksysteme. KTBL-Schrift 395.
Landwirtschaftsverlag, Münster, pp. 58-63.
70
Pinheiro, J., Bates, D., DebRoy, S., Sarkar, D., 2005. nlme: Linear and nonlinear mixed effects models.
R package version 3.1-60.
Pinheiro, J.C., Bates, D.M., 2000. Mixed-effects models in S and S-Plus. Springer, New York.
Poelarends, J.J., Sampimom, O.C., Neijenhuis, F., Miltenburg, J.D.H.M., Hillerton, J.E., Dearing, J.,
Fossing, C., 2004. Cow factors related to the increase of somatic cell count after introduction
of automatic milking. In: Meijering, A., Hogeveen, H., de Koning, C.J.A.M. (Eds.), Automatic
Milking. Wageningen Academic Publishers, Wageningen, pp. 148-154.
Porges, S.W., 1995. Cardiac vagal tone: A physiological index of stress. Neurosci. Biobehav. Rev.
19, 225-233.
Prescott, N.B., Mottram, T.T., Webster, A.J.F., 1997. Experiments studying the interaction between
dairy cow behaviour and automatic milking. In: ASAE (Ed.), Proceedings of the Fifth International
Symposium on Livestock Environment, Bloomington, Minnesota, 29.-31.05.1997, pp.
1090-1097.
Prescott, N.B., Mottram, T.T., Webster, A.J.F., 1998a. Relative motivations of dairy cows to be
milked or fed in a Y-maze and an automatic milking system. Appl. Anim. Behav. Sci. 57, 23-
33.
Prescott, N.B., Mottram, T.T., Webster, A.J.F., 1998b. Effect of food type and location on the attendance
to an automatic milking system by dairy cows and the effect of feeding during
milking on their behaviour and milking characteristics. Anim. Sci. 67, 183-193.
R Development Core Team, 2005. R: A language and environment for statistical computing. R
Foundation for Statistical Computing, Vienna, Austria. Available at: http://www.Rproject.
org.
Rasmussen, M.D., Bjerring, M., Justesen, P., Jepsen, L., 2002. Milk quality on Danish farms with
automatic milking systems. J. Dairy Sci. 85, 2869-2878.
Rasmussen, M.D., Frimer, E.S., Galton, D.M., Petersson, L.G., 1992. The influence of premilking
teat preparation and attachment delay on milk yield and milking performance. J. Dairy Sci.
75, 2131-2141.
Rémond, B., Coulon, J-B., Nicloux, M., Levieux, D., 1999. Effect of temporary once-daily milking in
early lactation on milk yield and nutritional status of dairy cows. Ann. Zootech. 48, 341-352.
Rémond, B., Petit, M., Ollier, A., 1992. Milking primiparous dairy cows three times every two days
in early lactation: milk secretion and nutritional status. In: Ipema, A.H., Lippus, A.C., Metz,
J.H.M., Rossing, W. (Eds.), Prospects for automatic milking. Proceedings of the International
Symposium, Wageningen, The Netherlands, 23.-25.11.1992, Pudoc Scientific Publishers,
Wageningen, pp. 227-232.
Rietmann, T.R., Stuart, A.E.A., Bernasconi, P., Stauffacher, M., Auer, J.A., Weishaupt, M.A., 2004.
Assessment of mental stress in warmblood horses: heart rate variability in comparison to
heart rate and selected behavioural parameters. Appl. Anim. Behav. Sci. 88, 121-136.
Rose, S., Klimetschek, H-J., Huschke, W., Brunsch, R., 2005. Entstehung von Kräften am Euter
durch unterschiedliche Schlauchführungssysteme bei Automatischen Melksystemen. In: Kuratorium
für Technik und Bauwesen in der Landwirtschaft (KTBL) (Ed.), 7. Tagung Bau,
Technik und Umwelt in der landwirtschaftlichen Nutztierhaltung, Braunschweig, 1.-
3.03.2005, pp. 181-186.
Rossing, W., Hogewerf, P.H., 1997. State of the art of automatic milking systems. Comput. Electron.
Agr. 17, 1-17.
71
Royle, C., Garnsworthy, P.C., McArthur, A.J., Mepham, T.B., 1992. Effects of frequent milking on
heart rate and other physiological variables in dairy cows. In: Ipema, A.H., Lippus, A.C.,
Metz, J.H.M., Rossing, W. (Eds.), Prospects for automatic milking. Proceedings of the International
Symposium, Wageningen, The Netherlands, 23.-25.11.1992, Pudoc Scientific Publishers,
Wageningen, pp. 237-243.
Rushen, J., de Passillé, A.M., Munksgaard, L., 1999. Fear of people by cows and effects on milk
yield, behavior and heart rate at milking. J. Dairy Sci. 82, 720-727.
Rushen, J., Munksgaard, L., Marnet, P.G., de Passillé, A.M., 2001. Human contact and the effects
of acute stress on cows at milking. Appl. Anim. Behav. Sci. 73, 1-14.
Saito, M., Nishimura, K., Kato, H., 1989. Modifications of circadian cortisol rhythm by cyclic and
continuous total enteral nutrition. J. Nutrit. Sci. Vitaminol. 35, 639-647.
Sällvik, K., Oostra, H., 2002. Detection of mastitis in different AMS systems. Bornimer Agrartechnische
Berichte 29, 63-79.
Sällvik, K., Sällvik, A., 2002. Comparison of the costs to milk cows in automatic (AMS) and manual
systems. In: Ronningen, O. (Ed.), NJF Seminar No. 337. Technology for milking and housing
of dairy cows. Hamar, Denmark, 11.-13.02.2002, pp. 92-97.
Schick, M., Moriz, C., 2003. Management von automatischen Melksystemen. Agrarforschung 10,
132-137.
Schön, H. (Ed.), 2000. Automatische Melksysteme. KTBL-Schrift 395. Landwirtschaftsverlag, Münster.
Schön, H., Wendl, G., Pirkelmann, H., 2000. Automatisches Melken – Bedeutung, Stand, Entwicklungstendenzen.
In: Schön, H. (Ed.), Automatische Melksysteme. KTBL-Schrift 395, Landwirtschaftsverlag,
Münster, pp. 9-17.
Schwarzer, K., 2000. Auswirkungen eines Melkroboters auf die Eutergesundheit und die Milchhygiene.
Dissertation, Ludwig-Maximilians-Universität München.
Seabrook, M.F., 1972. A study to determine the influence of the herdsman’s personality on milk
yield. J. Agric. Labour Sci. 1, 45-59.
Seabrook, M.F., 1994. Psychological interaction between the milker and the dairy cow. In: Bucklin,
R. (Ed.), Dairy Systems for the 21st Century. Proceedings of the Third International Dairy
Housing Conference. Orlando, Florida, USA, 2.-5.02.1994, pp. 49-58.
Sgoifo, A., De Boer, S.E., Westenbroek, C., Maes, F.W., Beldhuis, H., Suzuki, T., Koolhaas, J.M.,
1997. Incidence of arrhythmias and heart rate variability in wild-type rats exposed to social
stress. Am. J. Physiol. 273, H1754-H1760.
Shutt, D.A., Fell, L.R., 1985. Comparison of total and free cortisol in bovine serum and milk or colostrums.
J. Dairy Sci. 68,1832-1843.
Sonck, B.R., 1995. Labour research on automatic milking with a human-controlled cow traffic.
Neth. J. Agr. Sci. 43, 261-285.
Spolders, M., 2002. Effekte eines automatischen Systems des Milchentzugs (“Melkroboter”) auf
Futteraufnahmemenge, -rhythmik, Kau- und Wiederkauaktivität sowie stoffwechsel- und
leistungsbiologische Zusammenhänge bei Hochleistungskühen im Vergleich zum herkömmlichen
Melksystem. Dissertation, Tierärztliche Hochschule Hannover.
72
Spörndly, E., Wredle, E., 2004. Automatic milking and grazing – effects of distance to pasture and
level of supplements on milk yield and cow behaviour. J. Dairy Sci. 87, 1702-1712.
Stefanowska, J., Ipema, A.H., 2000. Effect of access to a field on the visiting pattern of cows to
the automatic milking system and other behaviour. In: Hogeveen, H., Meijering, A. (Eds.),
Robotic Milking. Proceedings of the international symposium held in Lelystad, The Netherlands,
17.-19.08.2000, Wageningen Pers, Wageningen, p. 194-195.
Stefanowska, J., Plavsic, M., Ipema, A.H., Hendriks, M.M.W.B., 2000. The effect of omitted milking
on the behaviour of cows in the context of cluster attachment failure during automatic milking.
Appl. Anim. Behav. Sci. 67, 277-291.
Stuhec, I., Schlichting, M.C., Smidt, D., Unshelm, J., 1984. The effect of different housing systems
on behaviour and plasma cortisol concentrations in gilts. In: Unshelm, J., van Putten, G.,
Zeeb, K. (Eds.), Proceedings of the International Congress on Applied Ethology of Farm
Animals. Kuratorium fur Technik und Bauwesen in der Landwirtschaft (KTBL), Darmstadt,
pp. 139-143.
Svennersten, K., Samuelsson, B., 1992. The effect of feeding during milking on milk production
and milk flow. In: Ipema, A.H., Lippus, A.C., Metz, J.H.M., Rossing, W. (Eds.), Prospects for
automatic milking. Proceedings of the International Symposium, Wageningen, The Netherlands,
23.-25.11.1992, Pudoc Scientific Publishers, Wageningen, pp. 233-236.
Takeuchi, H., Enzo, A., Minamitani, H., 2001. Circadian rhythm changes in heart rate variability
during chronic sound stress. Med. Biol. Eng. Comput. 39, 113-117.
Task Force of the European Society of Cardiology and the North American Society of Pacing and
Electrophysiology, 1996. Heart rate variability: Standards of measurement, physiological interpretation,
and clinical use. Eur. Heart J. 17, 354-381.
Termeulen, S.B., Butler, W.R., Natzke, R.P., 1981. Rapidity of cortisol transfer between blood and
milk following adrenocorticotropin injection. J. Dairy Sci. 64, 2197-2200.
Thun, R., 1987. Untersuchungen zur Tagesrhythmik von Cortisol beim Rind. Enke-Verlag, Stuttgart,
Germany.
Thun, R., Eggenberger, E., Zerobin, K., 1985. Twenty-four-hour secretory patterns of cortisol, progesterone
and estradiol in heifers during the follicular and luteal phases of the ovarian cycle.
Anim. Reprod. Sci. 9, 341-356.
Thune, R.Ø., Berggren, A.M., Gravas, L., Wiktorsson, H., 2002. Barn layout and cow traffic to optimise
the capacity of an automatic milking system. In: McLean, J., Sinclair, M., West, B.
(Eds.), Proceedings from The First North American Conference on Robotic Milking, Toronto,
Canada, 20.-22.03.2002, pp. II45-II50.
Van der Vorst, Y., de Koning, K., 2002. Automatic milking systems and milk quality in three european
countries. In: McLean, J., Sinclair, M., West, B. (Eds.), Proceedings from The First
North American Conference on Robotic Milking, Toronto, Canada, 20.-22.03.2002, pp. V1-
V12.
Van Lenteren, A.C., Korsten, G., 2002. Sub-optimal cow and barn condition and its effects on the
visiting frequency at the milking robot. In: McLean, J., Sinclair, M., West, B. (Eds.), Proceedings
from The First North American Conference on Robotic Milking, Toronto, Canada, 20.-
22.03.2002, pp. III64-III69.
Van Reenen, C.G., van der Werf, J.T.N., Bruckmaier, R.M., Hopster, H., Engel, B., Noordhuizen,
J.P.T.M., Blokhuis, H.J., 2002. Individual differences in behavioural and physiological responsiveness
of primiparous dairy cows to machine milking. J. Dairy Sci. 85, 2551-2561.
73
Venables, W.N., Ripley, B.D., 2002. Modern Applied Statistics with S. Fourth Edition. Springer, New
York.
Verkerk, G.A., Phipps, A.M., Carragher, J.F., Matthews, L.R., Stelwagen, K., 1998. Characterisation
of milk cortisol concentrations as a measure of short-term stress responses in lactating dairy
cows. Anim. Welfare 7, 77-86.
Visser, E.K., Van Reenen, C.G., van der Werf, J.T.N., Schilder, M.B.H., Knaap, J.H., Barneveld, A.,
Blokhuis, H.J., 2002. Heart rate and heart rate variability during a novel object test and a
handling test in young horses. Physiol. Behav. 76, 289-296.
Waiblinger, S., Menke, C., Fölsch, D.W., 2003. Influences on the avoidance and approach behaviour
of dairy cows towards humans on 35 farms. Appl. Anim. Behav. Sci. 84, 23-39.
Waiblinger, S., Menke, C., Coleman, G., 2002. The relationship between attitudes, personal characteristics
and behaviour of stockpeople and subsequent behaviour and production of dairy
cows. Appl. Anim. Behav. Sci. 79, 195-219.
Wechsler, B., 2005. An authorisation procedure for mass-produced farm animal housing systems
with regard to animal welfare. Livest. Prod. Sci. 94, 71-79.
Wechsler, B., Oester, H., 1998. Das Prüf- und Bewilligungsverfahren für Stalleinrichtungen. Agrarforschung
5, 321-324.
Weiss, D., Helmreich, S., Möstl, E., Dzidic, A., Bruckmaier, R.M., 2004a. Coping capacity of dairy
cows during the change from conventional to automatic milking. J. Anim. Sci. 82, 563-570.
Weiss, D., Hilger, M., Meyer, H.H.D., Bruckmaier, R.M., 2002. Variable milking intervals and milk
composition. Milchwissenschaft 57, 246-249.
Weiss, D., Reist, M., Bruckmaier, R.M., 2004b. The acyclic period postpartum in automatic and
conventional milking. J. Vet. Med. A 51, 268-272.
Wendl, G., Liebler, J., Sieber, O., Sieber, S., 1997. Erste Erfahrungen mit dem automatischen
Melksystem der Firma LELY auf einem Praxisbetrieb. In: Wendl, G. (Ed.), Aussenklimaställe
und automatische Melksysteme in der Milchviehhaltung. Landtechnik-Schrift Nr. 7, ALB Bayern,
pp. 105-114.
Wendl, G., Schön, H., Harms, J., Purucker, S., Pirkelmann, H., 1999. Tier- und Melkverhalten bei
automatischen Melksystemen – Ergebnisse von Einsatzuntersuchungen mit Einboxenanlagen.
In: Institut für Landtechnik der TU München-Weihenstephan (Ed.), Tagung: Bau, Technik
und Umwelt 1999 in der landwirtschaftlichen Nutztierhaltung. pp. 179-184.
Wenzel, C., Schönreiter, S., Unshelm, J., 2000. Empfehlungen zur Optimierung des Managements
von Milchrindern beim Melken in einem automatischen Melksystem aufgrund ethologischer
Kriterien. Congress of the German Veterinary Medical Society (DVG e.V.), Tagung der Fachgruppen
„Tierschutzrecht“ und „Tierzucht, Erbpathologie und Haustiergenetik“ der Deutschen
Veterinärmedizinischen Gesellschaft e.V., Nürtingen, 24.-25.02.2000, pp. 123-128.
Wenzel, C., Schönreiter-Fischer, S., Unshelm, J., 2003. Studies on step-kick behavior and stress of
cows during milking in an automatic milking system. Livest. Prod. Sci. 83, 237-246.
Wheelock, J.V., Rook, J.A.F., Dodd, F.H., 1965. The effect of incomplete milking or of an extended
milking interval on the milk yield and composition of cow’s milk. J. Dairy Res. 32, 237-248.
74
Wiktorsson, H., Spörndly, E., 2002. Grazing: An animal welfare issue for automatic milking farms.
In: McLean, J., Sinclair, M., West, B. (Eds.), Proceedings from The First North American Conference
on Robotic Milking, Toronto, Canada, 20.-22.03.2002, pp. VI32-VI43.
Wiktorsson, H., Pettersson, G., Olofsson, J., Svennersten-Sjaunja, K., Melin, M., 2003. Welfare
status of dairy cows in barns with automatic milking. Report D24. Produced within EUProject
„Implications of the introduction of automatic milking on dairy farms (QLK5-2000-
31006)”. www.automaticmilking.nl
Wilde, C.J., Blatchford, D.R., Knight, C.H., Peaker, M., 1989. Metabolic adaptations in goat mammary
tissue during long-term incomplete milking. J. Dairy Res. 56, 7-15.
Wilkinson, C.W., Shinsako, J., Dallman, M.F., 1979. Daily rhythms in adrenal responsiveness to
adrenocorticotropin are determined primarily by the time of feeding in the rat. Endocrinol.
104, 350-359.
Willis, G.L., 1983. A possible relationship between the flinch, step and kick response and milk yield
in lactating cows. Appl. Anim. Ethol. 10, 287-290.
Winblad von Walter, L., Olsson, K., Rundgren, M., 1999. Effects of omitting three milking occasions
in dairy goats. In: Boe, K.E., Bakken, M., Braastad, B.O. (Eds.), Proceedings of the 33rd International
Congress of the International Society for Applied Ethology. Lillehammer, Norway,
17.-21.08.1999, p. 118.
Wirtz, N., Oechtering, K., Tholen, E., Trappmann, W., 2002. Comparison of an automatic milking
system to a conventional milking parlour. In: McLean, J., Sinclair, M., West, B. (Eds.), Proceedings
from The First North American Conference on Robotic Milking, Toronto, Canada,
20.-22.03.2002, pp. III50-III55.
Zähner, M., 2001. Beurteilung von Minimalställen für Milchvieh anhand ethologischer und physiologischer
Parameter. Dissertation, ETH, Zürich, No. 14193.
Zecconi, A., Piccinini, R., Casirani, G., Binda, E., Migliorati, L., 2004. Introduction of AMS in Italian
dairy herds: Effects on teat tissues, intramammary infection risk, and spread of contagious
pathogens. In: Meijering, A., Hogeveen, H., de Koning, C.J.A.M., (Eds.), Automatic Milking.
Wageningen Academic Publishers, Wageningen, pp. 161-167.
Zschöck, M., Kloppert, B., Wolter, W., Seufert, H., Schwarz, H.P., Kötting, C., 1998. Zellzahlen der
Milch in grossen Beständen. Sonderdruck aus der Kartei für Rationalisierung 4.2.1.1, Rationalisierungskuratorium
für Landwirtschaft, Rendsburg-Osterrönfeld.
75
Acknowledgements
This study was performed at the Centre for proper housing of ruminants and pigs, Agroscope
FAT Tänikon, Switzerland, and funded by grants from the Swiss Federal Veterinary Office
No. 2.01.06 and No. 2.01.08. I would like to thank all the people who have influenced my work,
one way or another, and thus contributed to this thesis:
- Prof. Dr. Beat Wechsler, Head of the Centre for proper housing of ruminants and pigs, Agroscope
FAT Tänikon, for initiating the work, his support and encouragement.
- Prof. Dr. Thomas Jungbluth, Head of the Institute of Agricultural Engineering, University of Hohenheim,
for his collaboration.
- Dr. Rudolf Hauser for his ideas and help.
- Dr. Lorenz Gygax for statistical advice.
- Dr. Christine Kaufmann for sharing her experiences.
- Dr. Hans Kündig, Hubert Bollhalder, Roland Weber and Robert Meier for their technical and
computer assistance.
- Gallus Jöhl, Reto Rutishauser, Marion Riegel and Manuela Locher for their help and cooperation
during the experiments.
- Anthony Moses for the laboratory analysis of milk cortisol and Mirta Arnold for in-depth investigations
regarding details of the milk cortisol analysis.
- All involved farmers and their families for their cooperativeness and their hospitality.
- My husband Dominik for his patience and loving support during this process and my child Clara,
without whom this work would only have taken half of the time, but might have left me with
less joy in life.
76
Curriculum Vitae
Personal details
Name Isabelle Neuffer, née Gonidec
Date and place of birth 11.03.1975 in Bad Homburg (Germany)
Father Jean-Yves Gonidec
Mother Anke Gonidec-Kern, née Kern
Marital status Married, one daughter
Education
1980 – 1985 Hölderlinschule (Primary school), Bad Homburg
1985 – 1994 Humboldtschule (Higher school of education), Bad Homburg
10.1994 – 09.1996 Agricultural sciences, Justus Liebig University Giessen (Germany)
10.1997 – 01.2000 Agricultural sciences, University of Hohenheim (Germany)
Diploma in Agricultural Sciences
Work experience
07.1994 – 10.1994
and 03.1995 – 04.1995
N. Jäger (Dairy farm), Neu-Anspach (Germany)
11.1996 – 01.1997 P. Plouzennec (Pig breeding and pig fattening), Pluguffan (France)
02.1997 – 02.1997 Magasin Vert (Specialised dealer in agricultural supplies), Quimper
(France)
03.1997 – 04.1997 H. Le Corre (Laying hens), Ergué-Gabéric (France)
06.1997 – 08.1997 H. Le Grand (Dairy farm), Plogonnec (France)
11.1999 – 04.2000 Intern. German Agricultural Society, Groß-Umstadt (Germany)
05.2000 – 12.2000 Project engineer (Salescoordinator), Big Dutchman International GmbH,
Vechta (Germany)
02.2001 – 03.2006 PhD-Student, Swiss Federal Veterinary Office, Centre for proper housing
of ruminants and pigs, Agroscope FAT Tänikon, Ettenhausen (Switzerland)
77
Lebenslauf
Personalien
Name Isabelle Neuffer, geb. Gonidec
Geburtsdatum 11.03.1975 in Bad Homburg
Vater Jean-Yves Gonidec
Mutter Anke Gonidec-Kern, geb. Kern
Familienstand Verheiratet, eine Tochter
Ausbildung
1980 – 1985 Hölderlinschule, Bad Homburg
1985 – 1994 Humboldtgymnasium, Bad Homburg
10.1994 – 09.1996 Studium der Allgemeinen Agrarwissenschaften, Justus-Liebig-
Universität, Giessen
10.1997 – 01.2000 Studium der Allgemeinen Agrarwissenschaften mit Fachrichtung Tierproduktion,
Universität Hohenheim, Stuttgart
Berufliche Tätigkeiten
07.1994 – 10.1994
und 03.1995 – 04.1995
N. Jäger, Neu-Anspach, Gemischtbetrieb
11.1996 – 01.1997 P. Plouzennec, Pluguffan (F), Schweinezucht und –mast
02.1997 – 02.1997 Magasin Vert, Quimper (F), Landwirtschaftsbedarf
03.1997 – 04.1997 H. Le Corre, Ergué-Gabéric (F), Legehennenhaltung
06.1997 – 08.1997 H. Le Grand, Plogonnec (F), Milchviehhaltung
11.1999 – 04.2000 Praktikantin, Deutsche Landwirtschafts-Gesellschaft e.V.,
Fachbereich Landtechnik, DLG-Prüfstelle Groß-Umstadt
05.2000 – 12.2000 Projektingenieurin (Salescoordinatorin),
Big Dutchman International GmbH, Vechta
02.2001 – 03.2006 Doktorandin, Projekt „Einfluss von Automatischen Melksystemen auf
Verhalten und Gesundheit von Milchkühen“, Bundesamt für Veterinärwesen,
Zentrum für tiergerechte Haltung: Wiederkäuer und Schweine,
Agroscope FAT Tänikon, Ettenhausen (CH)
78
Erklärung
Hiermit erkläre ich, die vorliegende Arbeit selbständig angefertigt zu haben, nur die angegebenen
Quellen und Hilfsmittel benutzt und wörtlich oder inhaltlich übernommene Stellen als solche gekennzeichnet
zu haben.