25.12.2013 Views

novel approaches to expression and detection of oestrus in dairy cows

novel approaches to expression and detection of oestrus in dairy cows

novel approaches to expression and detection of oestrus in dairy cows

SHOW MORE
SHOW LESS

Create successful ePaper yourself

Turn your PDF publications into a flip-book with our unique Google optimized e-Paper software.

ate decreased dramatically <strong>to</strong> an average <strong>of</strong> 53.5%, without any <strong>in</strong>crease<br />

<strong>in</strong> script accuracy; therefore the optimised script resulted <strong>in</strong> the best<br />

oestrous <strong>detection</strong>.<br />

From the daily analysis <strong>of</strong> the reports <strong>of</strong> <strong>oestrus</strong>, mount<strong>in</strong>g <strong>and</strong> st<strong>and</strong><strong>in</strong>g<br />

behaviour, it was clear that the <strong>in</strong>cidence <strong>of</strong> erroneously declared mount<strong>in</strong>g<br />

was more prevalent on days when <strong>cows</strong> were actually <strong>in</strong> <strong>oestrus</strong> (Tables<br />

5.5 <strong>and</strong> 5.6). It was also clear that more error was attributed <strong>to</strong> <strong>cows</strong> that<br />

came <strong>in</strong><strong>to</strong> <strong>oestrus</strong> dur<strong>in</strong>g the trial compared <strong>to</strong> control <strong>cows</strong> not <strong>in</strong> <strong>oestrus</strong>.<br />

The <strong>in</strong>creased error dur<strong>in</strong>g <strong>oestrus</strong> could be due <strong>to</strong> the number <strong>of</strong> <strong>cows</strong> <strong>in</strong><br />

<strong>oestrus</strong> at once; for example <strong>in</strong>creased number <strong>of</strong> <strong>cows</strong> <strong>in</strong> <strong>oestrus</strong><br />

<strong>in</strong>creases oestrous <strong>expression</strong> (Hurnik et al., 1975;Van Vliet <strong>and</strong> Van<br />

Eerdenburg, 1996), which may mean that the error is a result <strong>of</strong> the<br />

experimental design because 6 <strong>cows</strong> were synchronised <strong>to</strong> come <strong>in</strong><strong>to</strong><br />

<strong>oestrus</strong> simultaneously. In a commercial situation the number <strong>of</strong> <strong>cows</strong> <strong>in</strong><br />

<strong>oestrus</strong> <strong>to</strong>gether may be lower which could result <strong>in</strong> a more accurate<br />

<strong>detection</strong> rate.<br />

Spikes <strong>in</strong> the data affect the accuracy <strong>of</strong> mount<strong>in</strong>g <strong>detection</strong> because <strong>in</strong> the<br />

UWB data error spikes are recorded <strong>in</strong> the same format as actual mounts<br />

<strong>and</strong> so could be mistaken. Falsely declared mounts were <strong>in</strong>vestigated by<br />

look<strong>in</strong>g at the MU position on CCTV cameras. This showed that there was<br />

no pattern associated with the error spikes. Potential reasons could be<br />

attributed <strong>to</strong> a <strong>cows</strong>’ position <strong>in</strong> an area <strong>of</strong> poor geometry <strong>in</strong> the <strong>dairy</strong> barn<br />

(Figure 4.9), or due <strong>to</strong> non-l<strong>in</strong>e <strong>of</strong> sight from the master unit (Harmer et<br />

al., 2008). Because the <strong>dairy</strong> farm is a complex environment with various<br />

obstructions, l<strong>in</strong>e <strong>of</strong> sight may become obstructed <strong>and</strong> thus cause<br />

reflections <strong>in</strong> the signal which give false or less accurate 3D positions.<br />

Although, filter<strong>in</strong>g by script analysis can help <strong>to</strong> elim<strong>in</strong>ate these spikes.<br />

Upon further analysis <strong>of</strong> the data from POC 2 <strong>and</strong> POC 3 it appears that<br />

data quality <strong>in</strong> POC 2 was better than POC 3, therefore data from POC 3<br />

required further filter<strong>in</strong>g. It is obvious that 2 MUs <strong>in</strong> POC 3 had higher<br />

<strong>in</strong>cidence <strong>of</strong> error <strong>and</strong> more false positives were declared by script analysis<br />

(Cow 38 <strong>and</strong> 623; control cow <strong>and</strong> <strong>oestrus</strong> cow). These particular MUs<br />

were tested aga<strong>in</strong>st a normally function<strong>in</strong>g MU <strong>in</strong> the same BU network <strong>in</strong><br />

the same positions <strong>of</strong> good geometry; results revealed no difference <strong>in</strong><br />

position or error, lead<strong>in</strong>g <strong>to</strong> the conclusion that error could be attributed <strong>to</strong><br />

a cow’s favoured position with<strong>in</strong> the location, which could be an area <strong>of</strong><br />

poorer geometry <strong>and</strong> more obstruction. Furthermore it is possible that<br />

140

Hooray! Your file is uploaded and ready to be published.

Saved successfully!

Ooh no, something went wrong!