View source for Conformation Recording: Guidelines on usage of data from AMS systems to derive udder conformation traits


== Introduction ==
To collect information on quality of conformation of udders, the udders of dairy cows are scored by professional classifiers. In the ICAR guidelines several linear udder conformation traits are defined for Dairy Cattle ([https://www.icar.org/Guidelines/05-Conformation-recording-Appendix-1.pdf Appendix 1]) and Dual Purpose Cattle ([https://www.icar.org/Guidelines/05-Conformation-recording-Appendix-2.pdf Appendix 2]).

High genetic correlations have been reported in literature between linear scored traits and the same traits based on data from automatic milking systems (AMS) (Byskov et al., 2012<ref>Byskov, K., L. H. Buch, and G. P. Aamand. 2012. Possibilities of implementing measures from automatic milking systems in routine evaluations of udder conformation and milking speed. Interbull Bull. 46:28–32.</ref>; Poppe et al, 2019<ref>Poppe, M., H.A. Mulder, B.J. Ducro, and G. de Jong, 2019. Genetic analysis of udder conformation traits derived from automatic milking system recording in dairy cows. J. Dairy Sci. 102:1386–1396.</ref>). The advantage of using data from AMS is that information is collected continuously during the whole lactation and also over lactations. In a regular classification system a cow is scored only once during her life, at one moment in the first lactation. Further the collection of data via an AMS system is cost effective.

== Description of AMS data about udder traits ==
In AMS information is stored about the positions of teats to be able to attach the milk cups in a short time. The storage or information of the teat positions can be different across manufacturers, but at the end the positions of all four teats are determined by an AMS and are stored.

Based on teat positions, five different udder traits can be derived. Each milking record contained Cartesian coordinates (x, y, z) of each of the 4 teats, indicating the 3-dimensional location of the teat tips. The z-coordinate is a measure of the distance from the teat tip to the floor, the y-coordinate is a measure of the position of the teat on the axis parallel to the long side of the robot, and the x-coordinate is a measure of the position of the teat on the axis perpendicular to the long side of the robot. The Cartesian teat coordinates in the AMS data set can be used to calculate the udder conformation traits: rear teat distance (RTD), front teat distance (FTD), udder depth (UD), distance between front and rear teats (DRF) and udder balance (UB, the difference in depth between front and rear udder). A specification of the udder conformation traits and equations for calculation are given in Table 3.

The trait rear teat distance is similar to rear teat placement, and front teat distance is similar to the trait front teat placement. Rear teat placement, front teat placement and udder depth are currently in the ICAR list of conformation traits for dairy and dual purpose cattle.


''Table 3. Calculation of udder conformation traits based on teat positions measured by AMS.''
{| class="wikitable"
|'''Udder trait'''
|'''Description'''
|'''Calculation'''<sup>1</sup>
|-
|UD
|Average  distance of teat ends to the floor in mm
|(Z<sub>lf</sub> + Z<sub>rf</sub> + Z<sub>lr</sub> +  Z<sub>rr</sub>)/4
|-
|FTD
|Distance  between left and right front teat ends in mm
|X<sub>lf</sub> - X<sub>rf</sub>
|-
|RTD
|Distance  between left and right rear teat ends in mm
|X<sub>lr</sub> - X<sub>rr</sub>
|-
|UB
|Average  difference in distance to the floor between the front and rear teat ends in  mm
|(Z<sub>lr</sub> + Z<sub>rr</sub>)/2 - (Z<sub>lf</sub>  + Z<sub>rf</sub>)/2
|-
|DRF
|Average  distance between the front and rear teat ends in mm
|(Y<sub>lf</sub> +  Y<sub>rr</sub>)/2 - (Y<sub>lf</sub> + Y<sub>rf</sub>)/2
|}
<sup>1</sup> Z = Z-coordinate, Y = Y-coordinate, X = X-coordinate, lf = left front, rf = right front, lr = left rear, rr = right rear.


== 1.3         Edits screening AMS data ==
When AMS data is used for analysis, checks need to be applied. Recommended edits are: 
<ol style="list-style-type:number">
<li> Milking is not a failure or not refused; </li>
<li> Milk yield is larger than 0 kg; </li>
<li> Teat positions are known; </li>
<li> Measures of z-coordinates (distance teat - floor) larger than 0 mm; </li>
<li> Distances between left front and rear or right front and rear larger than 0 mm; </li>
<li> Distance between left and right rear teats larger than -30 mm<sup>1</sup>; </li>
<li> Distance between left and right front teats larger than 0 mm; </li>
<li> Observation deviates less than 4 standard deviations from the expected observation of an animal; </li>
<li> At least 5 hours interval with previous milking; </li>

</ol>

10.    Measurements are available on all four teats. So, data of cows with less than four milking quarters has to be removed.

<sup>1</sup> A negative distance is possible in case of crossing teats. The distance depends on the order of attaching the cups, if for example the left teat is stored as left teat or as right teat.


Fields required using information form AMS:

a.      Animal id;

b.     Herd id;

c.      Date and time of miking;

d.     x, y, z coordinates;

e.      Milk yield;

f.       Information on success of milking.

== 1.4         Analysis of data in genetic evaluation ==
From the AMS, information of every separate milking more than 700 measurements are made available during a lactation. To analyze the data in genetic evaluations several options are possible:

-         Usage of all measurements.

-         Usage of every x<sup>th</sup> measurement.

-         Usage of average of n-measurements (e.g. 10) or of a certain period (e.g. week).

== 1.5         Considerations ==
-         Determined teat positions are actually the position of teat ends. In case trait udder depth is derived from teat positions, one also should or could consider teat length to get a more exact measure of udder depth. This correction only is possible in case the AMS system is also determining teat length.

-         Further it should be noticed that stature of the cow affects the distance from floor to udder (z coordinate). It also should be noticed that scoring udder depth by classifiers, udder depth is scored in relation to the hock (as reference point). This also results in the situation that udder depth scored by classifiers is also affected by stature.

-         When using udder depth data from AMS for genetic evaluation one always should check the genetic correlation between udder depth scored by classifiers and udder depth based on AMS data. A way to improve the genetic correlation is to add the breeding value for stature and teat length as extra effects to the model for udder depth based on AMS data.

-         In the future, data from AMS might also become available from udder traits like teat length, teat direction or teat thickness.