| Author | Message | ||
     Gareth Burns (gareth_burns) New member Username: gareth_burns Post Number: 3 Registered: 2-2011 |
Again many thanks for your comments Rob, I�ll definitely take on board all these comments. I just wrote some information on the discriminate analysis linking the faecal spectra to certain characteristics of animals but then I see you are already more than aware of these papers and have summed them up nicely in your review article. An excellent article may I add that I was not aware of until you brought it to my attention. I also see you are from Queensland and I hope you weren�t effected too much by the flooding. | ||
| Rob Dixon (robdixon) New member Username: robdixon Post Number: 2 Registered: 2-2011 |
Hello Gareth, In my comments I did not mean to imply that I thought the spectra from faeces are likely to be a better predictor of voluntary intake than spectra of the forage. We work with faeces because in the context of cattle grazing extensive rangelands in northern Australia we can sample faeces easily, but it is very difficult to obtain forage samples representing the diet actually selected by the grazing animal. Indeed in general I would expect better prediction of VI from the dietary forage than from faeces simply because the former should contain more spectral information about the diet. My concern about animal factors is that these potentially cause VI to vary. For example all else equal lactating cows have a higher VI than non-lactating cows, and VI is likely to be higher in high-producing cows than low-producing cows. Similarly factors such as maturity and compensatory growth of the animal will modify VI. But to my knowledge noone has been able to identify spectral characteristics in faeces which identify such respective classes of animals and (potentially) provide a means to adjust a VI prediction. And by definition we could not find such differences in the forage spectra. If the reference values for VI depend on n-alkanes marker measurements there may be substantial experimental error in these reference values. Regards Rob | ||
| Gareth Burns (gareth_burns) New member Username: gareth_burns Post Number: 2 Registered: 2-2011 |
Many thanks for the reply Rob, Apologies I must have been a little unclear in my description. The 15,000 spectra I refer to are from variety evaluation trials for ryegrass. c.2000 of these have wet chemistry values and the calibration developed from these. These were all dried at 80oC for 16hours. This trial was on a separate site from those trials below that have voluntary intake values. I would love to have 15,000 samples with intake potential, however I have a more modest 600. These are from a number of different trials; some samples have been dried at 40oC for 48 hours and others 80oC for 16hours. The objective would be to apply the intake potential equation to the 15,000 spectra from the variety evaluation trials. This could mean have an intake potential value for each grass cultivar as well as the chemical composition and DM digestibility. The intake potential values are generated using n-alkanes. The samples themselves are from March � November. Voluntary intake ranges from 9-29kg DM/cow/d. All samples are ad lib � no concentrate. You can probably pick up several issues with this, as the first trials are from variety evaluation schemes they have no animals on them and get a mechanical harvested yield (different management). However we have a trial that can be used as a validation; 15 cultivars sown with 2 treatments - mechanical harvesting and grazing, so will compare if the predicted intake potential from the mechanical harvesting matches with our intake values from the animals. I shall get your paper today and study it; I agree with your conclusion regarding the spectra and animal factors and that faeces can be a better predictor of voluntary intake. For the purposes of variety evaluation, faeces samples are impractical. However as you stated, forage samples are limited in their ability to predict something that is associated with animal factors. My suspicion is that what I am trying to do is asking a little too much, I may be able to derive an equation for voluntary intake from forage samples; however I don�t believe it will be to the degree of accuracy required that would allow discrimination between grass cultivars. | ||
| Rob Dixon (robdixon) New member Username: robdixon Post Number: 1 Registered: 2-2011 |
Hello Gareth, You describe a very interesting project. Could I suggest a couple of ideas. 1. There will be obvious links between the calibrations for chemical composition and DM digestibility as you indicate you have done, and developing a calibration for 'intake potential'. See for example the paper by Decruyenaere et al. 2009 Animal Feed Science and Technology 148, 138-156. There would be reason to scan the samples for the intake calibration using the same procedures (as far as possible) as used for the samples for the forage composition. For example if you redried and dessicated the grass samples before scanning for the composition calibration surely you should do the same for the samples for the 'intake' calibration. You do not mention how many samples were involved in the development of the calibration for forage composition and whether were they all from the same lab / study etc ? 2. A set of 15,000 samples for 'intake potential' is obviously enormous. Perhaps it would be wise to first scan a subset selected a represent a defined set of conditions and to represent a 'boxcar' range of the reference values and origins. Are the reference values derived from in vivo experiments, and what is the variability of the samples comprising the sample set? We have been working with calibrations to predict dietary forage composition and voluntary intake by cattle from the spectra of faeces. One of the major problems is that voluntary intake is influenced by many animal factors which are not likely to be, or cannot be, associated with the NIR spectra (see for example Dixon and Coates,2009, JNIRS 17, 1-31). I hope that these comments are useful, Rob | ||
| Gareth Burns (gareth_burns) New member Username: gareth_burns Post Number: 1 Registered: 2-2011 |
Dear all, First off I would like to say how amazed I am that such a resource exists. I am a Ph.D student doing research on grass samples from national variety evaluation schemes. I have designed a calibration to measure the quality (buffering capacity, crude protein, in vitro dry matter digestibility, water soluble carbohydrates) of dried, milled grass samples using FOSS XDS (all pretty standard so far). The next stage of my project is to design a calibration for intake potential � I have quite a large number of grass samples with ad lib intake potential measurements from historical trials (and also a nice validation trial comparing grazed trials with simulated grazing trials for intake potential) My problem arises as the samples are from historical trials they have undergone different drying procedures, some at 40oC for 48 hours and others at 80oC for 16 hours. My inclination would be to re-dry at say 40oC overnight to remove residual moisture, however I would like to apply the intake potential equation to the data-set used for the initial quality equation (c.15, 000 spectra) and this was dried at 80oC for 16hours and stored in air tight container. Should I re-dry the samples to remove the effects of residual moisture? If so, is there a way of re-drying a sub-set of the 15,000 and have some sort of bias adjustment for moisture, as re-drying 15,000 samples is unfeasible. Gareth |