AU719197B2 - Method for determining feed quality - Google Patents

Description

WO 97/21091 PCT/AU96/00776 -1- Method for Determining Feed Quality This invention relates to a method for quantifying biomechanical properties of animal feed based on a correlation between the chemical and biomechanical properties of the feed, and to methods for objectively measuring the quality of animal feed, such as fodders including hay, pastures and forages.

Diet is the major determinant of productivity of an animal. In the livestock industry, animals are farmed for meat, wool and other valuable products. The diet of farmed livestock is largely dictated by man and, given the effect of diet on animal production, it is highly desirable to optimise the diet of livestock to gain maximum benefit from the natural resource.

Feed quality is one variable that has a major impact on animal productivity. In this respect, feed quality affects the amount of feed an animal will consume and the feeding value it gains from the feed consumed. In the case of cattle, sheep and other ruminants, feed quality depends on digestibility, chemical attributes (nutrient composition) and biomechanical attributes (namely how easy it is for an animal to chew the feed during ingestion and rumination).

It is generally accepted that there are constraints on the intake of feed by ruminant animals, that the amount of useful energy obtained by a ruminant animal may fall short of the amount that the animal can potentially use, and that this would result in reduced productivity. For example, the principal constraints to voluntary intake of fodders are resistance of fodder fibre to chewing and digestibility (provided that the intake is not otherwise constrained by low palatability, deleterious secondary compounds, or the inadequacy of essential nutrients). Differences between feeds, such as fodders, in their resistance to chewing are reflected in differences in biomechanical properties, including comminution energy, shear energy, compression energy, tensile strength, shear strength and intrinsic shear strength.

WO 97/21091 PCT/AU96/00776 -2- Hay is a common feed, and its quality is significantly affected by factors such as seasonal differences, haymaking practices and pasture composition. It has been shown in one recent survey that in some years as little as 11% of hay produced was good enough to promote liveweight gain in weaner sheep. This possibility of wide variation in measures of hay quality is a matter of increasing concern, and has given rise to a demand for a method of objective quality assessment.

A hay quality system adopted in the United States of America uses a measure known as relative feed value (RFV) to distinguish between hays of different quality. The RFV is calculated from the dry matter digestibility, which is predicted from acid detergent fibre (ADF) content, and from the dry matter intake, which is predicted from neutral detergent fibre (NDF) content.

The RFV based system suffers from a number of disadvantages. For example, the ADF and NDF contents of fodders are determined by chemical methods which take several days to complete, and thus are expensive in terms of resources.

While objective quality assessment and product specification has become an integral part of the production and marketing in domestic and export markets for the Australian grain, wool, meat and dairy industries, performance-based quality standards are not presently in place for feeds such as hays and other fodders.

Consequently; the feed buyer cannot be sure of getting value for money, and this is likely to become increasingly important in respect of export markets if other exporting countries are able to guarantee standards for their product; the feed producer cannot be sure of getting a higher price for a superior product; WO 97/21091 PCT/AU96/00776 -3livestock producers are unable to objectively formulate rations or supplementary feeding regimes to achieve animal production targets; and the market for animal feed tends to be unstable.

Whilst the relationship between biomechanical properties of feed and feed quality is now accepted, there is a need for a convenient, inexpensive and relatively accurate assay method for feed to determine its quality. An accurate determination of feed quality allows for optimisation of feeding regimes and improved animal production for obvious economic gains.

It is an object of this invention to overcome or at least partially alleviate the aforementioned problems and/or reduce the uncertainties and concomitant problems of the prior art systems for measuring the biomechanical properties of feed and hence determining feed quality.

Thus, the present invention provides a method for determining a biomechanical property of a feed, the method comprising the steps of; subjecting the feed to infrared radiation to obtain spectral data; and using the spectral data to determine the biomechanical property; whereby the biomechanical property of the feed is determined on the basis of the bond energies of the chemical constituents of the feed.

The spectral data may be used directly to determine the biomechanical property of the feed. Alternatively, the spectral data may be used to determine another property of the feed and the other property is used to determine the biomechanical property on the basis of a correlation between the other property and the biomechanical property.

WO 97/21091 PCT/AU96/00776 -4- When the biomechanical property is determined via another property, the other property is preferably a chemical property of the feed such as the ADF content or the NDF content or the lignin content.

There is a variety of biomechanical properties of the feed that may be determined. Preferably, the biomechanical properties are selected from the group comprising shear energy, compression energy, comminution energy, tensile strength, shear strength and intrinsic shear strength.

The spectral data may comprise a reflectance spectrum at a combination of wavelengths or over a predetermined range of wavelengths such as 700nm- 3000nm, or more preferably 1100nm-2500nm. Preferably, the data obtained for the spectral range of 1850nm-1970nm is disregarded, this being the range over which water reflects strongly.

The spectral data may be recorded at one or more wavelength intervals throughout the spectral range. When the spectral data is a reflectance spectrum over a predetermined range it is preferably measured at 2nm intervals over the range. Of course, if so desired the spectral data may be measured at intervals other than 2nm.

When the spectral data is used to directly determine a biomechanical property, the biomechanical property is preferably determined by comparison of the spectral data with a calibration equation that reflects the relationship between reflectance and the biomechanical property. Preferably, the calibration curve is determined on the basis of laboratory data establishing a correlation between reflectance and the biomechanical property.

Thus, the present invention also provides a method for determining a biomechanical property of a feed, the method comprising the steps of; subjecting the feed to infrared radiation to obtain spectral data; and WO 97/21091 PCT/AU96/00776 comparing the spectral data obtained in with a calibration equation to determine the biomechanical property; whereby the biomechanical property of the feed is determined on the basis of the bond energies of the chemical constituents of the feed.

The present invention also provides a method for determining feed quality, the method comprising the steps of; subjecting the feed to infrared radiation to obtain spectral data; using the spectral data to determine a biomechanical property of the feed; and using the value of the biomechanical property obtained in step (b) to determine feed quality; whereby the biomechanical property of the feed and thus the feed quality is determined on the basis of the bond energies of the chemical constituents of the feed.

In one particular form, the method described immediately above may further comprise the determination of an additional property of the feed. The additional property may vary and preferably is selected from the group comprising the digestibility of the feed in vivo or in vitro, the ADF content or the NDF content, or the lignin content.

The present invention is based on research establishing a strong correlation between the bond energies as they relate to the physical structure, and the biomechanical properties of feed. Once this correlation is established the bond energies of the chemical constituents, and in turn the biomechanical properties of the feed, can be determined using infrared spectroscopy. The biomechanical WO 97/21091 PCT/AU96/00776 -6properties quantified in this way are useful for accurately determining feed quality.

In this respect, research resulting in the present invention has shown that the biomechanical attributes of feeds such as cereal and legume hays, straws, and mature, dry subterranean clovers are much more strongly related to animal performance than are digestibility or chemical composition of the feeds.

Thus, comminution energy, the energy required to grind or comminute fodder material, has proved to be a very effective indicator of forage consumption constraint (FCC), which is the difference between the quantity of forage an animal should consume to satisfy its capacity to use energy (a theoretical maximum) and the actual voluntary dry matter intake achieved.

Shear energy, the energy required to shear fodder material, and compression energy, the energy required to compress fodder material, are two biomechanical feed characters of fodders that are closely related to comminution energy and which also are good predictors of FCC.

In this respect, feed quality can be assessed in a number of ways. The forage consumption constraint (FCC) is one convenient measure of feed quality and equates to the difference between the quantity of the fodder that the animal would be attempting to consume to satisfy its capacity to use energy (theoretical maximum intake) and the voluntary forage consumption (VFC).

Thus, the present invention also provides a method for determining feed quality, the method comprising the steps of; subjecting the feed to infrared radiation to obtain spectral data; using the spectral data to determine a biomechanical property of the feed; and WO 97/21091 PCT/AU96/00716 -7using the value of the biomechanical property obtained in step (b) to determine the forage consumption constraint (FCC) or voluntary feed consumption (VFC) as a measure of feed quality; whereby the biomechanical property of the feed and thus the feed quality is determined on the basis of the bond energies of the chemical constituents of the feed.

The present invention is based on the finding that variations in biomechanical properties such as shear energy, comminution energy and compression energy are reflected in NIR spectra of fodders. This finding, together with recognition of the value of biomechanical characters for the prediction of FCC (and, in turn, the prediction of voluntary feed consumption (VFC) makes it possible for quicker, less expensive, more convenient and more reliable prediction of feed quality than hitherto known and predicted.

Accordingly, this invention provides a method of assessing the suitability of a fodder, such as a forage, to meet a required animal performance; or (ii) predicting the VFC of a forage; or (iii) predicting the FCC of a forage, which method comprises subjecting a sample of the forage to NIR radiation and determining the reflectance at selected wavelengths.

It has been found that the biomechanical properties, such as shear and comminution energy values for a given fodder, correlate with the fodder's reflectance of infrared radiation. More specifically, the invention is based on research showing that: NIR wavelengths at which reflectance namely the second derivative of the logarithm of the inverse of R, correlates significantly with the variation in energy required to shear fodder materials are 1168nm, 1458nm, 1598nm, 1718nm, 1828nm and 2048nm. For the prediction of fodder shear energy (yl, kJ.m2) the following equation may be used: WO 97/21091 PCT/AU96/00776 -8yi 19.95 10239.46 R 1168 3623.49 R 1458 4255.61 Rss 98 5319.88 R 1718 5148.38 R 1828 2452.05 R 2048 NIR wavelengths at which the second derivative of the logarithm of the inverse of reflectance correlates significantly with the variation in energy required to comminute fodder materials are 1138nm, 2018nm, 2128nm and 2408nm.

For the prediction of fodder comminute energy (y2, kJ.kg DM 1 the following equation is proposed: Y2 231.42 18224.74 R 1138 4955.12 R 2018 3005.37 R 2128 4290.18 R2408 NIR wavelengths at which the second derivative of the logarithm of the inverse of reflectance correlates significantly with the variation in compression energy are 1268nm, 1588nm, 1728nm, 2278nm. For the prediction of compression energy (y 3 kJ.kgDM'") the following equation may be used: y3 -0.71 911.04 R 1 26 8 112.57 R 1588 79.48 R 1728 28.02 R22 78 NIR wavelengths at which the second derivative of the logarithm of the inverse of reflectance correlates significantly with variation in in vivo digestibility of dry matter (DMD) (y 4 is 1158nm, 1238nm, 1668nm, 1908nm, 1918nm, and 2248nm. For prediction of the DMD (y 4 of a fodder the following equation is proposed: y4 46.62 8162.72 R 11 58 8799.69 R 1238 1249.01 R 1668 519.46 R1s08 367.08 R 1918 161.84 R22 48 NIR wavelength at which the second derivative of the logarithm of the inverse of reflectance correlates significantly with variation in in vitro WO 97/21091 PCT/AU96/00776 -9digestibility of dry matter (IVDMD) is 1698nm, 1748nm, 1908nm, 1918nm and 2158nm. For prediction of the DMD (in vitro) of a fodder the following equation is proposed: ys 63.43 2186.89 R 1 698 1491.99 R 1748 981.30 R19o 8 556.01

R

1 918 2003.05 R 2158 Accordingly, in a preferred method according to this invention, the infrared wavelengths at which reflectance is measured comprise one or more of the following: 1168nm, 1458nm, 1598nm, 1718nm, 1828nm, 2048nm, 1138nm, 2018nm, 2128nm, 2408nm, 1268nm, 1588nm, 1728nm, 2278nm, 1158nm, 1238nm, 1668nm, 1908nm, 2248nm, 1698nm, 1748nm, 1918nm and 2158nm.

It will be understood that the foregoing are wavelengths at which the strongest correlations have been observed, and the possibility of useful correlations being observed at other wavelengths are highly likely.

Essentially, it can be shown that in the same way that a decrease in comminution energy is reflected by a decrease in forage consumption constraint, there is also a linear relationship between comminution energy or shear energy and the consumption constraint of a fodder. Thus, the use of NIR spectra, in conjunction with the equations detailed at paragraphs to above, permits estimation of the VFC of a fodder, which together with estimates of digestibility (conveniently obtained from NIR spectra) can be expected to provide a valuable basis for performance-based quality standards for fodders.

It is to be appreciated that the intention of this invention is to offer a quick, reliable and relatively inexpensive means of obtaining information from which the fodder producer and user, such as purchaser, might make informed judgements about the market value of a given fodder sample relative to alternatives, and of its suitability for a particular purpose.

WO 97/21091 PCT/AU96/00776 Conceivably, fodder quality predictions obtained by the method of this invention could be a useful component of, or used in conjunction with, for example, Decision Support Software (DSS) packages designed to assist livestock management.

It is further envisaged that by combining NIR measurements made by a remote sensing system, such as Landsat, with data from a Geographical Information System, the invention will provide a means of making reliable predictions of pasture quality. These predictions, together with predictions of feed intake and animal performance, should then provide a useful basis for strategies of supplementary feeding to improve performance in grazing ruminants.

The present invention also provides for a spectrometer configured to determine biomechanical properties and/or quality of feed according to the methods of the present invention. Preferably, the spectrometer includes a data processing means which enables the spectrometer to receive a feed sample and quantify either or both the biomechanical properties of the feed and the quality of the feed. In one particular form the data processing means includes a calibration equation to facilitate the determination of the feed quality or biomechanical property.

The invention will now be described with reference to the following examples. The description of the examples is in no way to limit the generality of the preceding paragraphs.

WO 97/21091 PCT/AU96/00776 -11

EXAMPLES

The energy of molecular vibrations correspond to the energy of the infrared spectrum of the electromagnetic spectrum, and these molecular vibrations may be detected and measured in the wavelength range of the infrared spectrum.

Functional groups in molecules have vibration frequencies that are characteristic of that functional group and that are within well-defined regions of the infrared spectrum.

For organic compounds the principal analytical features of the near infrared (NIR) spectrum are due to absorbance of radiant energy by bonds between hydrogen, carbon, nitrogen, oxygen or with sulphur, phosphorus and metal halides. When organic compounds are irradiated with infrared radiation at wavelengths between 700 and 3000nm part of the incident radiation is absorbed and the remainder is reflected, refracted or transmitted by the sample. Most quantitative reflectance analyses are made in the wavelength range of 1100 to 2600nm. The amount of energy absorbed or diffusely reflected at any given wavelength in this wavelength range is related to the chemical composition of the organic compound. NIR spectroscopy uses detectors to measure the amount of radiation that is diffusely reflected by the irradiated sample.

NIR spectroscopic analysis is an analytical procedure calibrated to a primary reference method. Calibration in NIR spectroscopy (NIRS) relies on similarities among the spectra, and analytical properties of interest in the reference samples. In this example the analytical properties of interest were the biomechanical characters of forages, and the procedure that was adopted in this example was as follows: a) prediction of biomechanical characters of a range of grasses using NIR spectroscopy was established by developing a calibration equation(s) from laboratory determined values of a set of reference samples.

r /AU 9 6 0 0 7 7 6 RECEIVED 0 6 OCT 1997 -12/1b) validation of the equation(s) either by using laboratory determined values of a separate set of samples, or by a cross-validation procedure using the laboratory determined values of the reference samples.

c) using the NIRS-predicted values for biomechanical characters of the forages and for digestibility of the forages, forage consumption constraint (FCC) was predicted, and in turn voluntary feed consumption (VFC) was predicted.

d) the predicted FCC and VFC were compared with actual data from groups of animals fed each of these forages.

Example A: Developing a calibration equation to predict biomechanical properties of herbage: The samples used in this example were a range of varieties of Panicum spp.

harvested at a range of plant maturities throughout the growing season (Table Each of the samples was dried and chaffed, and then fed to groups of sheep (8 sheep per group) which were penned individually, to determine in vivo dry matter digestibility (DMD), VFC and FCC. Samples of the hays were stored for laboratory analyses.

Biomechanical properties of the forages were determined using published methods; the energies required to shear or compress the forages according to Baker, Klein, de Boer and Purser (Genotypes of dry, mature subterranean clover differ in shear energy. Proceedings of the XVII International Grassland Congress 1993. pp 592-593.) and the energy required to comminute the forages according to Weston, R.H. and Davis, P (1991) 'The significance of four forage characters as constraints to voluntary intake' p.33 in 'Proceedings of The Third International Symposium on the Nutrition of Herbivores' (Compiled by: Wan Zahari, Amad Tajuddin, Abdullah, N. and Wong, H.K' Published by the Malaysian Society of Animal Production (MSAP). ISBN: 967-960-026-2). In vitro digestibility of dry matter (IVDMD) was determined by the pepsin-cellulase AMENDED

SHEET

IPEA/tUu PC/Au 6 U 0 7 6 RECEIVED 0 6 OCT 1997 -12/2technique as modified by Klein and Baker (Composition of the fractions of dry, mature subterranean clover digested in vivo and in vitro. Proceedings of the XVII International Grassland Congress 1993. pp593-595.).

AMENDED SHEET IPEA/r.- WO 97/21091 PCT/AU96/00776 -13- There are several ways to process samples for NIRS analysis, and in this example the samples were ground through a cyclone mill with a 1 mm screen and equilibrated at 250C for at least 24h before NIRS analysis. The samples were scanned by a monochromating near infrared reflectance spectrophotometer (Perstorp NIRS 6500) and the absorption spectra recorded for the range 1100 to 2500nm at 2nm intervals. The spectral range 1850 to 1970nm, where water absorbs strongly, was disregarded in further analysis of the spectral data.

For NIRS analysis the samples were divided into two groups: one group to be used as a 'calibration' set to establish a prediction equation, and a second group, the 'validation' set, to be used to validate the prediction equation. There are a number of ways to select the samples for each set. In this example the samples were ranked according to each of the characters that were to be predicted and every other sample was selected for the calibration set (33 samples) and the validation set (32 samples). Thus, for each character that was evaluated, a different selection was made from the 65 samples to establish the respective calibration and validation sample sets.

The ranges, mean, median and variation in the laboratory-determined values for each of the characters of interest in the calibration and validation sets are listed in Table 2.

The software for scanning, mathematical processing and statistical analysis were supplied with the spectrophotometer by the manufacturers. The spectral data were transformed by taking the second derivative of the logarithm of the inverse of the reflectance at each wavelength log (1 The similarities amongst the spectra (Figure 1) of the samples in the validation and calibration sets were determined using principal components scores to rank the spectra according to the Mahalanobis distance from the average of the spectra. The Mahalanobis distance values were standardized by dividing them by their average value, and were denoted 'global' H values (Table 3).

WO 97/21091 PCT/AU96/00776 -14- Calibration equations were developed using the calibration samples by regressing the data from the laboratory analyses of each biomechanical property against the corresponding transformed spectral data using the following mathematical methods: a) Stepwise linear regression b) Step-up linear regression c) Principal components regression (PCR) d) Partial least squares regression (PLS), and e) Modified partial least squares regression (MPLS).

Stepwise calibrations were developed for each calibration set of samples using the mathematical treatments of the spectral data 2,2,2; 2,5,5; 2,10,5; and 2,10,10; where the first number denotes that the second derivative was used, the second indicates that second derivatives of the spectral data (determined at 2nm intervals) were taken at intervals of 4, 10 or 20nm, and the third indicates that the function was smoothed using the 'boxcar' method over intervals of wavelength of 4, 10, or 20nm (Table 4a). Likewise step-up calibrations were developed for each calibration set with up to 6 terms in each calibration equation using mathematical treatments 2,2,2; 2,5,5; 2,10,5; and 2,10,10 (Table 4b).

Calibrations developed for each calibration set using principal components regression, partial least squares regression, or modified partial least squares regression each were developed using mathematical treatments 2,5,5 and 2,10,10 (Table 4c).

In developing the calibration equations in the stepwise and step-up regressions, only wavelengths with partial F-statistic of more than 8 were accepted for the models.

For each calibration using each calibration set the following calibration statistics were determined: WO 97/21091 PCT/AU96/00776 a) Squared multiple correlation coefficient an indication of the proportion of the variation in the calibration set that is adequately modelled by the calibration equation.

b) The standard error of calibration (SEC) together with its confidence interval CL), which is the standard deviation for the residuals due to difference between the laboratory determined (reference) and the NIR predicted values for samples within the calibration set Once the calibration equations were developed, each equation was validated by using it to predict the respective biomechanical property values for each sample in the validation sample set. For each calibration equation the following validation statistics were determined: a) Simple linear correlation coefficient (r 2 between the laboratory determined and NIR predicted values.

b) The bias (or systematic error) in the regression relationship between the laboratory determined (reference) and NIR predicted values.

c) The confidence limits of the bias in the regression relationship between the laboratory determined (reference) and NIR predicted values.

d) The standard error of prediction, corrected for bias which represents the unexplained error of the prediction, the deviation of the differences between laboratory determined and NIR predicted values.

e) The coefficient of determination, or slope and y-intercept of the linear regression relationship between the laboratory determined and NIR predicted values.

f) The residual standard deviation (RSD) of the linear regression relationship between the laboratory determined and NIR predicted values.

In addition, the calibration equations were validated using a procedures of crossvalidation. These are procedures where every sample in the calibration set was used once for prediction, and the standard error of validation corrected for bias (SEV(C), for stepwise and step-up regressions) and cross-validation (SECV, for multivariate regressions) can be determined.

WO 97/21091 PCT/AU96/00776 -16- Calibration equations for each biomechanical character were selected using the following criteria: a) Lowest partial F-ratio, highest R 2 lowest SEC and, for PCR, PLS and MPLS, lowest SEV(C) (or, for multivariate regressions, SECV) b) Highest r 2 lowest bias and I bias I bias confidence limit, lowest SEP(C), B closest to 1.0, a closest to 0, and lowest RSD. As well, SEP(C) was compared with the standard error of laboratory determined values amongst all 65 samples, listed in Table Calibration equations were similarly established to predict in vivo digestibility and in vitro digestibility. The coefficients for each wavelength in the selected calibration equations from stepwise or step-up regression analyses are listed in Table 6a, and those from multivariate analyses are listed in Table 6b.

Simple linear correlation coefficient (r 2 between the laboratory determined and NIR predicted values for each of the biomechanical characters (energies required to shear, comminute or compress) and digestibility of dry matter determined in vivo or in vitro of the samples in the validation set are shown in Figures 2a, 2b, and 2c. The NIR predicted values are predicted using calibration equations that best met the criteria listed above.

Example B: Prediction of FCC and VFC using NIR determinations of energy required to shear and in vivo digestibility: To demonstrate the prediction of voluntary feed consumption using NIR determined values for a biomechanical character and digestibility of forages, samples of Panicum spp. hay were selected which were common to both of the validation sample sets used to establish the NIR prediction equations for energy required to shear and in vivo digestibility. The hays represented the range of varieties in the sample set, and are listed in Table 7. The samples were scanned by the same spectrophotometer that was used to establish the WO 97/21091 PCT/AU96/00776 -17calibration equations, and the absorption spectra were recorded in the range 1100 to 2500nm at 2nm intervals. Values for energy required to shear and in vivo digestibility were predicted from calibration equations (Tables 4a, 4b and 4c) using the recorded spectral data.

These values then were used to estimate FCC from the relationship between biomechanical character(s) and FCC of the range of forages used by Weston and Davis (1991). Energy required to shear the forages used by Weston and Davis was determined according to Baker et al. (1993). The relationship between the energy required to shear these forages (kJ/m 2 and FCC (g organic matter (OM) d kg metabolic body weight (MBW)) was described by the relationship: Energy required to shear -26.13 5.53 (FCC where R 0.92; RSD 8.70; N 13; P 0.0001.

FCC from this relationship and in vivo digestibility predicted by NIR were then used to estimate VFC, as the difference between the animal's capacity to use energy (as defined by Weston and Davis, 1991) and FCC. These data are summarised in Table 8.

VFC predicted n ai this way explained most of the variation in actual VFC (R 0.87; RSD 5.04; P 0.023) (Figure 3).

Throughout the specification, unless the context requires otherwise, the word "comprise" or variations such as "comprises" or "comprising", will be understood to imply the inclusion of a stated integer or group of integers but not the exclusion of any other integer or group of integers.

Table 1. Description of herbage used in this example.

Genus species I Variety Common name Part of Process undergone I Stage of maturity I Regrowth I plant Penicumn Panlcumn Ponicum Panicum Panleumn Ponlcum Penicum Panlcum Panlcum Panicumn Panlcumn Panicum Penicumn Panlcum Panlcum, Panicumn Penlcum Penicum Penlcumn Penicum Panicum Penicum Pornicum Panicumn- Panicum Penicum caloretum coloralum color. turn colorstum coloralum cotoralum coIorotum coloraluni coloralumn colorelUm colorslurn colorefurn color. him col"Olaum coloreturm Calorslum" var Makadlcadaonse colorstum var Makodke'fense cot orstum ver Makeifhajense coloralumn ver Makerikaef nse colorfuui ver Malcfkodkeense colorolum var Makanlaulens colorsuum var Mokerikedone coloratumn var Makarlkedene maxkmum Maximum maximm Bambatel Buniastal Bambatel Bambstal Bambetal Bambatel Kabuisbuia CPI 16796 Kabulabula CPI 1679 Kabtoibla CPI 16796 Katadabula CPI 16796 KAbtdabuja CPI 16796 Kabuabula CPI 16796 Burnett Burnett Burnett Burnett Burnett BDrnet Burnett Colono Cuoxba Makarikad gross Makarikud gross Makartkadl grass Makariked gross Makarikaut gross Makarikad grass Makerikail grass Makurikaut grass Makartkarl gross Makarikad grass Makarikad grass Makarikaul gross Moarikud grass Makarikad grass Makadhaed gross Makauikud grass Makarikad grass Makarikad gross Makarikad grass Makarlkad gross Makarikad grass Mokaulkarl grass Makarikart grass Guinea grass Gulnea grass Guinea gross aerial serial aeria serial Sorial aerial aerial serial aeIal serial serial Serial merial serial serial aerial serial dried and chaffed dried and chuffed dried and chaffed dried end chuffed dried and chaffed dried! and chaffed dried and chaffed.

dried and chuffed dried avid chaffed dried and chaffed dried and chaff ed dried and chaffed d&We and duoffed drie and chaffed dried end chaffed dried and chaffed dried and chaffed dried and chuffed dried and chuffed dried and chuffed dried anid chuffed dried and chaffed dried and chaffed dried and chaffed dried and chaffed dried anid chmffed late bloom (9 weoes' regrowth) late bloom (13 weeks' regrowth) late bloom (4 weeks' regrowth) mid bloom (I monthae regrowth) mid bloom (I month regrowth) mid bloom (10 weeks! regrowth) mld bloom (6 weeks' regrowth) vegetative regrowth (29 days') late bloom late bloom (4 weeks' regrowth) late bloom (19 weeks! regrowith) late bloomn (14 weeks' regrowith) mId bloom (9 weeks' regrowth) mid bloom (6 weeks' regrowith) vegetative regrowthi (28 days') early bloom (I month' regrowth) late bloom (14 weeks' regrowth) late bloom (4 weeks' regrowth) mid bloom (6 weeks' regrowth) mid bloom (10 weeks' regrowith) vegeative regrowth (31 days') mid bloom (1 month's regrowth) mid bloom (4 weeks' regrowth) latelbloom (4 weeks'regrowith) mid lbloom (13 weeks':rsgrowth) mid bloom (10 woe'regrawth) late bloom regrowth late bloom, regrowth late bloom regrowth mid bloom regrowth mid bloom regrwtlh mid bloom.- regrowth mid bloom regrowth veative regrowth late bloom late bloom regrowth lots bloom regrowth loaebloomn regrowth mid bloom regrowth milom- regrowth vegetative regrowth eairly bloom regrowith late bloom regrowth late bloom regrowthi mid bloom regrowith midd bloom regrowthi vegetative regrowthi mid bloom regrowth mid bloom regrowth lote bloomn regrowth mid blom regrwth mid bloom regrowth Table 1. Description of herbage used In this example.

Genus species Variety Common namea Part of Process undergone Stage of maturity Regrowth Panicumn maximum Colonloo Guinea grass serial dried and chaffed vegetative regrwth (4 weeks!) vegetatve regrowtlh Ponicum maximum Cokonio Guinea gross arial dried and chuffed vegetative regrowth (33 days) vegetative regrowth Panlcumn maximum Colonlac Gulnea grass aerial dried and chaffed vegetttve regrowth (21 mdnys) vegetativ regrowth Panicumn maxiMum Colaniso Guinea gross aeril dried and chatted vegetative regrowth (I month's) vegetative regrowth Panicum maximum Colonlac Guinea grass aerial dried and chaffed vegetative regrowth (6 moneks) vegetative regrowth Panicum maximum Handis Guirea grass aerial dried and chaffed vea lom(v mn regrot 8 w tW) vea bom regrowth Panicumn maximum Hand Gulnea grass srial dried and chaffed early bloom (10 weeks regrowth) early bloom regrowth Panicumn maximum Hamm Guinea grass Weild dried and chaffed latey bloo weekrgrwh) latey boom regrowth Panicum maximuwm "amd Guinea grass aeria dried and chaffed latd bloom (43 weeks regrowth) mids bloom. regrowth Panicumn maximum Han" Guinea gross l dried and chuffed 54 day.' r1egm Wlh rot rsgruwt-hrgrwt Panicumn maximnum Hand Guinea grass leaf dried and chuffed 5 days' regrowth regrowth Panicum maximum Hamil Guinea grase leaf dried and chaffed 75 days' regrowth regrowth Panicum maximum Hand Ginea. gross seral dried and chuffed vegetati'veg(6owtek rerwhegeothv Panicumn maximum Hand Ginea grass aeria dried and chuffed vegetative regweeksh (9gweets) Vegetativerer h Panicumn maximum Harril Guirva grass Srial dried and chaffed vegetativ regrowth (32 dayks!) vegetative regrowth Panicum maximum va.Idhgim Pti. Green Panics astial dried and chuffed leativeoom growtehs'32egryst) laatem- regrowth Panicum maximum var. lfchoglums, Petie Green Pando aeria dried and chaffed latd bloom (8 month' regrowh) lotd bloom regrowth Panicum maximuim var. tdchoglume Pe"l Green Panio Weil dried and dhafd mid bloom (10 weeks' regrowthu) mid bloo regrowth Panicum maximum" var. fichoplume Petine Green Panio aerial dried and chaffed mid bloom (14 weeks' regrowth) mid bloom regrowth Panicum maximuim var. hichogfure Petrie Grewn Panic aerial dried and duaffed: mid bloom (14 weks' regrwt) mid bloom regrowth Panicumn maximuim var. Ilchogfumf Petie Green Panic serial dried and dhuffed midd Maon (15 week! regrwth) mid blooma regrowth Panicum maximum ver. 1icogume Petie Green Pario aerial dried and chuffed mid bloom (95 weeks regrwh) mid bloom regrowth Panicum, maximm var. tidchoofume Pe"t Green Panic seril dried and chuffed midd bloom (1 month's)owth mid bloom regrowth Panicum maximum var. Idcfaogfurn. Pe"l Green Panic serial dried and chuffed mid bloOM (13 wee1s1 rerut mid bloom regrowth Punicum maximum var. ldchogrlwne Petrie Green Panic sewial dried and chuffed mid bloomn (43 weeks' regrwth) midbloom regrowth Panicumn maximum var. Idchogfume, P"tl Green Panic ser"a dried and chuffed mid bloom (41 weeks' regrowth) mid bloom regrowth Panicumn maximum var. Idchoglum. Petrie Green Pardo aerial dried and chaffed vegetative regrowth (28 days!) vegetative regrowib Panicumn maximumn var. tihoglue Petrie Green Pario aeria dried and chaffed vegetative regrow~th (32 dayW) vegetativ regrowth Panicum, maximum ver. Itichogtumse Petrie Green Panic aeil dried and chuffed vegetativ regrowth (4 weeks') vegetative regrowth Panicum 'maximum var. ticogum. Petris Green Panic ael dried and chaffed vegetative regrowth (4 weeks') Vegetative regrowth -20- WO 97/21091 PCT/AU96/00776 Table 2. Summary statistics for each calibration and validation set Energy required Energy required Energy required Digestibility of Digestibility of dry to shear to comminute to compress dry matter in viva matter in vitro (kJ/m 2 (kJ/kg DM) (kJ/kg DM) "Energy required to shear- Calibration set mean 15.48 134.9 3.70 55.7 53.3 median 15.17 133.8 3.65 56.0 55.1 maximum 20.95 216.5 4.39 64.0 63.0 minimum 10.80 72.5 3.25 43.0 39.8 standard deviation 2.572 37.50 0.265 5.73 6.97 Validation set mean 15.43 130.9 3.78 55.6 52.7 median 15.20 128.3 3.75 56.5 53.3 maximum 20.43 205.2 4.24 64.0 63.0 minimum 10.94 54.5 3.34 47.0 40.1 standard deviation 2.444 37.50 0.229 5.36 7.01 i i -Ene grequrequi tocommionute eL- Calibration set mean 15.01 133.1 3.69 55.7 52.8 median 14.76 129.5 3.70 57.0 54.7 maximum 19.97 216.5 4.18 64.0 63.0 minimum 10.80 54,5 3.25 43.0 39.8 standard deviation 2.444 38.82 0.227 5.64 7.06 Validation set mean 15.92 132.9 3.79 55.6 53.2 median 15.97 130.2 3.79 55.5 54.7 maximum 20.95 205.2 4.39 64.0 62.5 minimum 11.46 60.7 3.34 47.0 40.9 standard deviation 2.490 36.20 0.263 5.47 6.92 vf jiiyu Eiet rinli red t co mprea ls Calibration set mean 15.28 128.1 3.74 56.3 53.8 median 15.07 128.4 3.72 57.0 54.7 maximum 19.97 204.0 4.39 64.0 63.0 minimum 10.80 54.5 3.25 47.0 39.8 standard deviation 2.477 38.00 0.261 5.15 6.64 Validation set mean 15.64 138.0 3.74 55.0 52.2 median 15.42 132.5 3.72 54.5 54.5 maximum 20,95 216.5 4.24 64.0 62.0 minimum 11.46 60.7 3.34 43.0 40.1 standard deviation 2.530 36.39 0.240 5.87 7.26 :::.DigestIb la ityblf of dry matter in vo:: Calibration set mean median maximum minimum standard deviation Validation set mean median maximum minimum standard deviation 15.14 15.17 20.95 10.80 2.528 15.78 15.20 20.37 10.94 2.446 133.9 128.4 216.5 60.7 36.39 132.1 134.7 205.2 54.5 38.70 3.74 3.72 4.24 3.25 0.247 3.75 3.72 4.39 3.34 0.255 RECTIFIED SHEET (RULE 91) WO 97/21091 WO 9721091PCT/AU96/00776 Table 2 (cont'd).

Summary statistics for each calibration and validation Energy required to Energy required Energy required Digestibility of Digestibility of dry shear to comminute to compress dry matter in matter in vitro (kJ/kg DM) (kJ/kg DM) vivo Calibration set mean 14.70 131.3 3.75 55.6 53.0 median 14.34 129.3 3.71 56.0 54.7 maximum 19.36 216.5 4.39 64.0 63.0 minimum 10.80 54.5 3.25 43.0 40.1 standard 2.235 42.58 0.241 5.78 6.94 deviation Validation set mean 16.19 134.6 3.74 55.6 53.0 median 16.16 133.8 3.74 56.0 54.7 maximum 20.95 194.8 4.24 64.0 63.0 minimum 10.94 65.7 3.34 47.0 39.8 standard 2.538 31.84 0.260 5.33 7.05 deviation Table 3.

Mahalanobis distances Mean Median Range For full sample set: 0.655 0.623 0.203 1.983 For calibration sets for: Energy required to shear 0.588 0.549 0.171 1.646 Energy required to comminute 0.718 0.676 0.350 1.553 Energy required to compress 0.757 0.760 0.188 1.440 Digestibility of dry matter in vivo 0.673 0.634 0.389 1.547 Digestibility of dry matter in vitro 0.645 0.574 0.185 1.178 C VTP~r T" .R TT(RAP 261 WO 97/2 1091 22 Table 4a. Calibration and validation statistics Stepwise Regression 2,2,2 2,5,5 2,10,5 2,10,10 Lowest partial F-ratio 10.27 6.18 8.27 4.70 R20.798 0.787 0.795 0.780 SEC 1.155 1.188 1.166 1.207 SEC CL 1.493 1.535 1.507 1.560 SEV(C) 1.230 1.306 1.273 1.322 r20.368 0.625 0.520 0.495 Bias 0.690 0.710 0.700 0.720 Bias CL 1.484 1.527 1.498 1.551 SEP 1.500 1.540 1.520 1.570 Slope 0.604 0.617 0.598 0.758 I ntercept 6.340 5.440 5.640 3.710 R.S.D. 1.627 1.627 1.484 1.476 IBiasi Bias CL -0.794 -0.817 -0.798 -0.831 JBiasi Bias CL? Yes Yes Yes Yes number of terms 6 5 5 6 ENr 44 req i"ed O10 i com m fnute Stepwise Regression 2,2,2 2,5,5 2,10,5 2,10,10 Lowest partial F-ratio 5.54 4.45 16 .55 10.89

R

2 0.910 0.802 0.818 0.831 SEC 11.626 17.281 16.546 15.980 SEC CL 1.493 1.535 1.507 1.560 SEV(C) 13.103 18.040 17.587 17.100 r20.363 0.429 0.374 0.213 Bias 6.980 10.370 9.930 9.590 Bias CL 14.941 22.209 21.264 20.537 SEP 15.110 22.460 21 .510 20.770 Slope 0.530 0.575 0.607 0.417 Intercept 58.300 48.900 48.600 74.600 R.S.D. 28.900 27.360 28.650 32.1 IBiasj Bias CL -7.961 -11.839 -11.334 -10.947 Biasi Bias CL? Yes Yes Yes Yes number of terms 6 -3 4 4 .Enegy rtepws -Regoressio Lowest partial F-ratio 5.05 4.44 7.90 16.1 R20.784 0.500 0.525 0.534 SEC 0.121 0.209 0.204 0.202 SEC CL 1.493 1.535 1.507 1.560 SEV(C) 0.135 0.224 0.217 0.215 r20.069 0.113 0.008 0.067 Bias 0.070 0.130 0.120 0.120 Bias CL 0.156 0.269 0.262 0.260 SEP 0.160 0.270 0.270 0.260 Slope 0.180 -0.080 0.314 0.211 Intercept 3.060 4.030 2.580 2.960 R.S.D. 0.229 0.229 0.227 0.232 Biasi Bias CL -0.086 -0.139 -0.142 -0.140 1 Biasi Bias CL? Yes Yes Yes Yes Inumber of terms 6 4 4 4 PCT/AU96/00776 SUBSTITUTE SHEET (RULE 26) WO 97/2 1091 PCTIAU96/00776 23 Table 4a (cont'd) Lowest partial F-ratio 7.63 20.68 4.28 6.08 R 2 0.934 0.917 0.914 0.921 SEC 1.107 1.236 1.258 1.207 SEC CL 1.493 1.535 1.507 1.560 SEV(C) 1.215 1.368 1.341 1.284 r'0.654 0.881 0.878 0.876 Bias 1.070 0.890 0.910 0.900 Bias CL 0.156 0.269 0.262 0.260 SEP 2.320 1.940 1.980 1.960 Slope 0.705 0.878 0.840 0.827 Intercept 16.500 6.690 8.640 9.340 R.S.D. 3.153 1.852 1.873 1.888 jBiasj Bias CL 0.914 0.621 0.648 0.640 IBiasl Bias CL? No No No No number of terms 6 6 6 Stepwise Regression 2,2,2,1 2,5,5 2,10,5 2,10,10 Lowest partial F-ratio 7.68 11.84 4.33 6.31 R20.935 0.933 0.915 0.922 SEC 1.808 1.751 2.052 1.974 SEC CL 1.493 1.535 1.507 1.560 SEV(C) 1.984 1.981 2.186 2.100 02 .699 0.847 0.743 0.736 Bias 1.080 1.050 1.230 1.180 Bias CIL 2.324 2.250 2.637 2.537 SEP 2.340 2.280 2.670 2.570 Slope 0.839 0.962 0.775 0.763 Intercept 8.790 1.650 12.200 12.700 R.S.D. 3.805 3.794 3.805 2.719 IBiasi Bias CL -1.244 -1.200 -1.407 -1.357 1 Biasl Bias CL? Yes Yes Yes Yes Inumber of terms 6 -5 -6 5 1 SUBSTITUTE SHEET (RULE 26) Table 4b. Calibration and validation statistics (Step-up regression) Eergy equied toshee Step-up Regression 2,2.2 Step-up Regression 2.5.5 1 term 2 terms 3 terms 4 terms 5 terms 6 terms 1 term 2 terms 3 terms 4 terms 5 terms 6 terms LoetprilFrto 29.33 13.83 6.65 5.29 5.89 3.23 25.70 21.99 5.71 1.60 4.92 1.70 R2 0.470 0.625 0.684 0.725 0.766 0.784 0.436 0.663 0.709 0.7 15 0.778 0.792 -SEC 1.873 1.575 1.445 1.349 1.244 1.196 1.932 1.492 1.387 1.373 1.211 1.173 SEC CL 2.420 2.035 1.867 1.743 1.608 1.546 2.497 1.928 1.792 1.774 1.565 1.516 SEVOC 1.973 1.672 1.561 1.476 1.390 1.318 2.022 1.571 1.476 1.470 1.357 1.319 0.371 0.344 0.310 0.205 0.202 0.168 0.375 0.531 0.557 0.557 0.631 0.635 Bias 1.120 0.950 0.870 0.810 0.775 0.720 1.160 0.900 0.830 0.820 0.730 0.700 Bias CL 2.407 2.024 1.857 1.734 1.599 1.537 2.4,83 1.917 1.783 1.765 1.556 1.507i SEP 2.430 2.050 1.880 1.750 1.620 1.550 2.510 1.940 1.800 1.780 1.570 1.5201 Slope 1.000 0.795 0,784 0.598 0.549 0.498 0.643 0.606 0.616 0.633 0.644 0.633 Intercept -0.120 3.030 3.290 6.090 6.950 7.790 5.390 5.790 5.560 5.270 5.050 5.170 R.S.D. 1.896 1.803 1.773 1.769 1.732 2.058 1.891, 1.806 1.698 '1.656 2.317 1.945 IBiasi Bias CL -1.287 -1.074 -0.987 -0.924 -0.824 817 -1.323 -1.017 -0.953 -0.945 -0.826 -0.807 Iilasi <Bias CL? Yes Yes Yes Yes Yes Yes Yes Yes Yes Yes Yes Yes Step-up Regression 2.2.2 Step-up Regression 2.5.5 I term 2 terms 3 terms 4 terms 5 terms 6 terms 1 term 2 terms 3 terms 4 terms 5 terms 6 terms Lowes paiaF-ratio 81.33 12.82 9.62 6.10 6.71 2.92 67.30 8.96 2.73 4.91 5.06 4.26 R2 0.715 0.794 0.840 0.864 0.887 0.894 0.974 0.741 0.755 0.782 0.810 0.830 -SEC 20.719 .17.629 15.538 14.330 13.061 12.620 22.149 19.757 19.213 18.105 16.921 15.983 SEC CL 2.420 2.035 1.867 1.743 1.608 1.546 2.497 1.928 1.792 1.774 1.565 1.516 SEV(C 21 .511 18.353 16.378 15.230 13.967 13.633 22.769 20.547 20.096 19.092 18.262 17.484 0.322 0.424 0.421 0.411 0.371 0.373 0.183 0.199 0.148 0.099 0.114 0.098 -Bias 12.430 10.580 9.320 8.600 7.840 7.570 13.290 11.850 11.530 10.860 10.150 9.590 -Bias CL 26.627 22.656 19.969 18.416 16.785 16.21,9 28.465 25.391 24.692 23.268 21.746 20.54 1 -SEP 26.940 22.920 20.200 18.630 16.980 16.410 28.790 25.680 24.980 23.540 22.000 20.780 -Slope 0.605 0.623 0.577 0.560 0.524 0.521 0.491 0.518 0.441 0.346 0.365 0.317 -Intercept 47.100 43.900 .48.900 52.900 58.300 57.800 60.100 58.500 70.800 84.600 82.700 89.600 RH.S.D. 29.810 27.480 27.550 27.790 28.720 28.670 32.720 32.400 33.420 34.370 34.070 34.380 IBias Bias CL -14.197 -12.076 -10.649 -9.816 -8945 -8.649 -15.175 -13.541 -13.162 -12.408 -11.596 -10.951 IBiasi Bias CL? Yes Yes Yes Yes Yes Yes Yes Yes Yes Yes Yes Yes Table 4b. (cont'd) EneraY rquir-.-- c77777s Step-up Regression 2,2.2 Step-up Regression 1 term 2 terms 3 terms 4 terms 5 terms 6 terms I term 2 terms 3 terms 4 terms 5 terms 6 terms LoetarialFratia 7.28 6.24 3.22 2.65 3.10 0.0 8.07 4.23 4.08 2.21 5.36 1.87 R2 0,164 0.285 0.334 0.370 0.440 0.530 0.181 0.258 0.327 0.445 0.520 0.535 SEC 0.270 0.250 0.241J 0.235 0.221 0.203 0.268 0.255 0.243 0.220 0.205 0.202 SEC CL 2.420 2.035 1.867 1.743 1.608 1.546 2.497 1.928 1.792 1.774 1.565 1.516 SVQ0.277 0.259 0.252 0.248B 0.238 0.226 0.276 0.268 0.257 0.24 1 0.227 0.222 r 20.067 0.089 0.087 0.104 0.067 0.033 0.039 0.064 0.038 0.005 0.010 0.006 Bias 0.160 0.150 0.140 0.140 0.130 0.120 0.160 0.150 0.150 0.130 0.120 0.120 Bias CL 0.347 0.321 0.310 0.302 0.284 0.261 0.344 0.328 0.312 0.283 0.263 0.260 SEP 0.350 0.330 0.310 0.310 0.290 0.260 0.350 0.330 0.320 0.290 0.270 0.260 Slope 0.367 0.394 0.345 0.341 0.280 0.156 0.267 0.295 0.198 0.068 0.085 0.063 Intercept 2.380 2.270 2.460 2.470 2.690 3.160 2.750 2,640: 3.010 3.490 3.420 3.510 R.S.D. 0.235 0.232 0.235 0.239 0.239 0.239 0.236 0.233 0.230 -0.229 0.233 0.239 iBlas Bias CL -0.187 -0.171 -0.170 -0.162 -0.154 -0.1411 -0.184 -0.178 -0.162 -0.153 -0.143 -0.140 IBiasl Bias CL? Yes Yes Yes Yes Yes Yes Yes Yes Yes Yes Yes Yes.

.iet~bt v m..e Step-up Regression 2.2.2 Step-up Regression 2,5,5 1 term 2 terms 3 terms 4 termfs 5 terms 6 terms I term 2 terms 3 terms 4 terms 5 terms 6 terms -Loes pta F-ratio 72.42 5.79 8.71 16.80 13.18 2.86 80.75 12.35 8.41 7.34 6.03 3.72 _R 20.616 0.714 0.830 0.862 0.883 0.906 0.679 0.826 0.897 0.909 0.919 0.924 SEC 3.555 -3.069 2.365 2.127 1.962 1.755 3.248 2.394 1.840 1.728 1.635 1.586 SEC CL 2.420 2.035 1.867 1.743 1.608 1.546 2.497 1.928 1.792 1.774 1.565 1.516 SVQ3.666 3.250 2.557 2.312 2.152 1.956 3.328 2.457 1.972 1.884 1.828 1.782 _(20.785 0.712 0.664 0.787 0.884 0.768 0.755 0.740 0.884 0.893 0.876 0.869 -Bias 2.130 1.840 1.420 1.280 1.180 1.050 1.950 .1.440 1.100 1.040 0.980 0.950 Bias CL 0.347 0.321 0.310 0.302 0.284 0.261 0.344 038 0.3 12 0.283 0.263 0.260 -SEP 4.620 3.990 3.070 2.770 2.550 2.280 4.220 3.110 2.390 2.250 2.130 2.060 -Slope 1.050 0.805 0.792 0.826 0.777 0.731 1.090 1.050 0.889 0.880 0.866 0.885 Intercept -2.180 10.300 11.000 9.040 11.900 14.900 -5.290 -3.300 5.850 6.430 7.210 6.08 R.S.D. 2.484 2.877 2.584 2.652 2.734 3.108 2.476 1.825 1.825 1.750 1.884 1.940 IBiast Bias CL 1.783 1.519 1.110 0.978 0.896 0.789 1.606 1.112 0.788 0.757 0.717 0.690 jBiasj Bias CL? No No No No No NO 140 no FMO 140 1 Table 4b. (cont'd) Step-up Regression 2.2.2 Step-up Regression 2.5.5 1 term 2 terms 3 terms 4 terms 5 terms 6 terms 1 term 2 terms 3 terms 4 terms 5 terms 6 terms LoetarialF-raio 73.30 5.73 8.71 17.00 13.23 2.99 81.21 12.38 8.48 7.23 6.07 3.72 Az 0.692 0.733 0.788 0.863 0.905 0.915 0.715 0.791 0.833 0.863 0.884 0.894 SEC 3.913 3.645 2.251 2.610 2.177 2.058 3.768 3.222 2.883 2.616 2.407 2.294 SEC CL 2.420 2.035 1.867 1.743 1.608 1.546 2.497 1.928 1.792 1.774 1.565 1.516 SE()4.020 3.186 3.411 2.781 2.324 2.203 3.855 3.360 3,063 2.809 2.615 2.490 rz0.731 0.694 0.687 0.644 0.685 0.671 0.735 0.856 0.845 0.849 0.801 0.800 Bias 2.350 2.190 1.950 1.570 1.310 1.230 2.260 1.930 1.730 1.570 1.440 1.380 Bias CL 5.029 4.684 2.893 3.354 2.798 2.645 4.842 4.141 3.705 3.362 3.093 2.94 8 SEP 5.090 4.740 4.230 3.390 2.830 2.680 4.900 4.190 3.750 3.400 3.130 2.980 -Slope 0.946 0.868 0.861 0.877 0.860 0.830 1.080 0.994 1.020 0.976 0.975 0.914 Intercept 2.000 5.890 6.550 6.140 7.240 9.150 -4.700 -0.410 -1.970 0.240 0.240 4.040 R.S.D. 3.565 3.601 3.842 3.882 4.143 3.895 3.576 2.637 2.733 2.694 3.097 3.1,03 IBlasi Bias CL -2.679 -2.494 -0.943 -1.784 -1.488 -1.415 -2.582 -2.211 -1.975 -1.792 -1.653 -1.568 IBlasj Bias CL? Yes Yes Yes Yes Yes Yes Yes Yes Yes Yes Yes Yes t~ e Step-up Regression 2,10,5 Step-up Regression 2,10,10 1 term 2 terms 3 terms 4 terms 5 terms 6 terms 1 term 2 terms 3 terms 4 terms 5 terms 6 terms LowtatialF-raio 25.11 14.92 5.87 4.42 7.61 1.89 23.54 15.23 4.01 4.30 4.35 4.66 R2 0.430 0.606 0.661 0.697 0.755 0.763 0.413 0.598 0.641 0.678 0.721 0.755 SEC 1.942 -1.613 1.496 1.415 1.273 1.252 1.970 1.631 1.541 1.460 1.358 1.274 SEC CL 2.510 2.084 1.933 1.829, 1.645 1.618 2.546 2.108 1.991 1.887 1.755 1.646_ SE()2.020 1.674 1.611 1.541 1.403 1.392 2.047 1.689 1.612 1.569 1.494 1.411 0.273 0.398 0.456, 0.473 0.476 0.498 0.291 0.333 0.401 0.454 0.517 0.541 Bias 1.170 0.970 0.900 0.850 0.760 0.750 1.180 0.980 0.920 0.880 0.1810 0.760 Bias CL 2.496 2.073 1.923 1.818 1.636 1.609 2.532 2.096 1.980 1.876 1.745 1.637 SEP 2.520 2.100 1.950 1.840 1.650 1.630 2.560 2.120 2.000 1.900 1.760 1.660 Slope -0.706 0.723 0.717 0.707 0.610 0.616 0.737 0.581 0.639 0.653 0.715 0.709 Intercept 4.150 3.950 4.320 4.260 5.620 5.440 3.720 5.850 5.040 5.040 4.130 4.160 R.S.D. 2.170 2. 193 2.343 2.367 2.524 2.375 2.179 1.607 1.666 1.644 1.889 1.893 IBiasi Bias CL -1.326 -1.103 -1.023 -0.968 -0.876 -0.859 -1.352 -1.116 -1.060 -0.996 -0.935 -0.877 IBiasl Bias CL? Yes Yes Yes Yes Yes Yes Yes Yes Yes Yes Yes Yes Table 4b. (cont'd) I ~navrnI tanmiIraat ftn rnmwninrt*.

Step-up Regression 2.10,5 Step-up Regression 2, 10,.10 1 term 2 terms 3 terms 4 terms 5 terms 6 terms I term 2 terms 3 terms 4 terms 5 terms 6 terms Lowet aiaFrio 76.72 5.31 2.76 2.18 1.68 1.01 74.85 5.39 2.13 2.96 4.38 1.49 0.703 0.739 0.754 0.763 0.772 0.800 0.698 0.735 0.745 0.761 0.787 0.79 1 SEC 21.158 19.825 19.267 18.887 18.518 17.344 21.345 19.977 19.611 18.982 17.929 17.768 SEC CL 2.510 2.084 1.933 1.829 1.645 1.618 2.546 2.108 1.991 1.887 1.755 1.646 SEV(C) 21.803 20.707 20.279 19.690 19.499 18.691 22.033 20.904 20.985 19.911 18.777 18.634 ?0.460 0.468 0.414 0.394 0.330 0.215 0.434 0.450 0.408 0.397 0.357 0.387 Bias 12.690 11.890 11.560 11.330 11.110 10.410 12.810 11.990 11.700 11.390 10.760 10.660 Bias CL 27.191 25.478 24.761 24.273 23.799 22.290 27.432 25.674 25.203 24.395 23.042 22.835 SEP 27.510 25.770 25.050 24.550 24.070 22.550 27.750 25.970 25.490 24.680 23.310 23.100 Slope 0.793 0.737 0.688 0.649 0.598 0.468 0.776 0.729 0.694 0.645 0.622 0.633 Intercept 18.300 24.500 31.800 39.600 48.100 67.100 20.200 25.600 30.800 39.000 43.600 42.200 R.S.D. 26.610 26.420 27.720 28.170 29.640 32.080 27.230 26.850 27.860 *28.100 29.030 28.350 iBiasi -Bias CL -14.501 -13.588 -13.201 -12.943 -12.689 -11.880 -14.622 -13.684 .13.503 -13.005 -12.282 -12.175 !Biasi Bias CL? Yes Yes Yes Yes Yes Yes Yes Yes Yes Yes Yes, Yes Step-up Regression 2.10.5 Step-up Regression 2,10,10 1 term 2 terms 3 terms 4 terms 5 terms 6 terms 1 term 2 terms 3 terms 4 terms 5 terms 6 terms Loetarial F-ratio 8.16 2.24 8.06 4.06 1.97 4.98 6.50 4.94 4.91 1.75 3.54 3.83 R2 0.183 0.214 0.364 0.457 0.532 0.592 0.147 0.243 0.397 '0.412 0.461 0.512 SEC 0.267 '0.262 0.236 0.218 0.202 0.189 0.273 0.257 0.230 0.227 0.217 0.207 -SEC CL 2.510 2.084 1.933 1.829 1.645 1.618 2.546 2.108 1.991 1.887 1.755 1.646 SE()0.278 0.275 0.252 0.235 0.218 0.210 0.283 0.273 0.250 0.250 0.247 0.235 0.010 0.028 0.052 0.076 0.086 0.053 0.006 0.057 0.045 0.052 0.035 0.029 Bias 0.160 0.160 0.140 0.130 0.120 0.110 0.160 0.150 0.140 0.140 0.130 0.1207 Bias CL 0.343 0.337 0.303 0.280 0.260 0.243 0.35 1 '0.330 0.296 0.292 0.279 0.266 SEP 0.350 0,340 0.3 10 0.280 0.260 0.250 0.360 0.330 0.300 0.290 0.280 0.270 Slope 0.127 0.212 0.239 0.252 0.218 0.142 0.102 024 0.216 0.212 0.149 0.149 Intercept 3.270 2.960 2.850 2.800 2.930 3.210 3.360 2.640 2.940 2.950 3.180 3.190 R.S.D. 0.234 0.233 0.235 0.236 1.942 1.495 1.736 1.938 1.980 2.030 2.179 2.183_ (Biasi Bias CL -0.183 -0.177 -0.163 -0.150 -0.140 -0.133 -0.191 -0.180 -0.156 -0.152 -0.149 -0.146 s V IBiasi Bias CL? Table 4b. (cont'd) Step-up Regression 2,10.5 Step-up Regression 2. 10, 1 term 2 terms 3 terms 4 terms 5 terms 6 terms I term 2 terms 3 terms 4 terms 5 terms 6 terms -LoestprialIF-raio 91.59 9.46 9.77 7.10 2.58 4.12 93.60 6.52 16.07 4.36 4.68 4.94 RZ 0.700 0.867 0.902 0.916 0.927 0.935 0.675 0.615 0.898 0.912 0.922 0.927 -SEC 3.139 2.095 1.794 1.660 1.545 1.457 3.271 2.467 1.828 1.698 1.598 1.546 -SEC CL 2.510 2.084 1.933 1.829 1.645 1.618 2.546 2.108 1.991 1.887 1.755 1.646 SEV(C) 3.332 2.282 1.997 1.830 1.694 1.607 3.357 2.572 2.016 1.905 1.840 1.787 r20.777 0.856 0.888 0.871 0.887 0.881 0.828 0.831 0.809 0.836 0.877 0.892 Bias 1.880 1.260 1.080 1.000 0.930 0.870 1.960 1.480 1.100 1.020 0.960 0.930 Bias CL 0.343 0.337 0.303 0.280 0.260 0.243 0.35 1 0.330 0.296 0.292 0.279 0.266 -SEP 4.080 2.720 2.330 2.160 2.010 1.890 4.250 3.210 2.380 2.210 2.080 2.010 -Slope 0.892 0.880 0.924 0.870 0.851 0.837 1.130 0.991 0.812 0.829 0.840 0.865 -intercept 7.380 6.750 3.930 6.850 8.320 8.960 -6.490 0.210 9.960 9.030 8.7 10 7.270 2.531 2.034 1.791 1.927 1.799 1.846 2.222 2.201 2.344 2.172 1.876 1.761 IBiasl Bias CL 1.537 0.923 0.777 0.720 0.670 0.627 1.609 1.150 0.804 0.728 0.681 0.664 iBiasi Bias CL? No No No No No No No No No No No No D~~ae~t~~bIIIW t~ I vt Step-up Regression 2,10,5 Step-up Regression 2,10,10 1 term 2 terms 3 terms 4 terms 5 terms 6 terms 1 term 2 terms 3 terms 4 terms 5 terms 6 terms Lowest partial F-ratio 94.12 6.62 16.01 4.48 4.61 5.01 92.14 9.60 9.70 10.55 2.67 4.41 0.744 0.784 0.856 0.87 1 0.886 0.901 0.740 0.797 0.842 0.81 0.888 0.900 -SEC 3.568 .3.283 2.680 2.532 2.384 2.224 3.596 3.182 2.802 2.430 2.361 2.235 -SEC CL 2.510 2.084 1.933 1.829 1.645 1.618 2.546 2.108 1.991 1.887 1.755 1.646 SE()3.633 3.371 2.785 2.667 2.563 2.364 3.655 3.280 2.892 2.618 2.525 2.416 r 0.828 0.816 0.813 0.802 0.819 0.851 0.823 0.807 0.810 0.844 0.818 0.823 Bias 2.140 1.970 1.610 1.520 1.430 1.330 2.160 1.910 1.680 1.460 1.420 1.340 Bias CL 4.585 4.219 3.444 3.254 3.064 2.858 4.621 4.089 3.601. 3.123 3.034 2.872 -SEP 4.640 4.270 3.480 3.290 3.100 2.890 4.680 4.140 3.640 3.160 3.070 2.910 Slope 0.960 0.971 0.906 0.862 0.867 0.864 0.937 0.927 0.882 0.935 0.881 0.841 Intercept 2.120 1.280 4.530 7.230 7.380 7.610 3.490 3.660 5.790 3.260 6.140 8.660 R.S.D. 2.978 3.002 3.088 2.952 2.681 2.922 3.023 2.742 2.959 2.846 0.231 0.239 IBiasi Bias CL -2.445 -2.249 -1.834 -1.734 -1.634 -1.528 -2.461 -2.179 -1.921! -1.663 -1.614 -1.532 IBlasi Bias CL? Yes Yes Yes Yes Yes Yes Yes Yes Yes Yes Yes Yes WO 97/21091 PCT/AU96/00776 29 Table 4c. Calibration and validation statistics (multivariate regressions) Energy r~equied t sea PCR PLS MPLS 2,5,5 2,10,10 2,5,5 2,10,10 2,5,5 2,10,10 R7 0.847 0.752 0.639 0.601 0.601 0.582 SEC 1.036 1.290 1.545 1.624 1.550 1.586 SEC CL 1.199 1.493 1.788 1.879 1.793 1.835 SECV 1.750 1.592 1.788 1 .933 1.600 1.583 r20.5441 0.4876 0.4938 0.41 57 0.3080 0.3563 Bias 0.620 0.770 0.930 0.970 0.930 0.950 Bias CL 1.331 1.658 1.986 2.087 1.992 2.038 SEP 1.350 1.680 2.010 2.110 2.020 2.060 Slope 0.6540 0.6850 0.7390 0.6270 0.5220 0.6000 Intercept 5.2900 4.8100 3.7500 5.4500 7.3100 6.0200 R.S.D. 1.671 1.776 1.761 1 .892 2.065 1.992 jBiasi Bias CL -0.711 -0.888 -1.056 -1.117 -1.062 -1.088 JBiasi Bias CL? Yes Yes Yes Yes Yes Yes PCR PLS ML 2,5,5 2,10,10 -2.5,5 2,10,10 2,5,5 2,10,10 RF 0.584 0.574 0.605 0.595 0.556 0.558 SEC 23.378 23.682 22.788 23.075 24.164 24.101 SEC CL 27.048 27.400 26.366 26.698 27.958 27.885 SECV 26.030 26.121 25.548 25.683 26.409 26.252 r20.349 0.337 0.332 0.325 0.33 0.328 Bias 14.030 14.210 13.670 13.840 14.500 14.460 Bias CL 30.044 30.435 29.286 29.655 31.055 30.974 SEP 30.390 30.790 29.620 30.000 31.410 31.330 Slope 0.649 0.636 0.644 0.632 0.657 0.638 Intercept 37.3 39.2 38.1 39.7 36.6 39.1 R.S.D. 28.246 28.676 28.714 28.884 28.651 28.900 IBiasi Bias CIL -16.014 -16.225 -15.616 -15.815 -16.555 -16.514 IBiasi Bias CL? IYes Yes Yes Yes Yes Yes PCR PLS MPLS 2,5,5 -2,10,10 2,5,5 2,10,10 2,5,5 2,10.10 R20.251 0.160 0.231 0.208 0.038 0.040 SEC 0.225 0.241 0.265 0.269 0.260 0.260 SEC CL 0.260 0.279 0.307 0.311 0.301 0.301 SECV 0.299 0.277 0.301 0.307 0.301 0.299 r20.0220 0.0120 0.0130 0.0090 0.0060 0.0080 Bias 0.140 0.140 0.160 0.160 0.160 0.160 Bias CL 0.289 0.310 0.341 0.346 0.334 0.334 SEP 0.290 0.310 0.340 0.350 0.340 0.340 Slope 0.2290 0.2270 0.1530 0.1330 0.2290 0.2590 Intercept 2.8900 2.9000 3.1700 3.2500 2.8900 2.7700 R.S.D. 0.235 0.236 0.236 0.237 0.237 0.237 IBiasi Bias CL -0.149 -0.170 -0.181 -0.186 -0.174 -0.174 1 Biasl Bias CL? IYes Yes IYes Yes IYes Yes I SUBSTITUTE SHEET (RULE 26) WO 97/21091 PCT/AU96/00776 30 Table 4c (cont'd) Calibration and validation statistics (multivariate regressions) tvo diym terh v PCR PLS MPLS 2,5,5 2,10,10 2,5,5 2,10,10 2,5,5 2,10,10 R 2 0.909 0.900 0.958 0.937 0.571 0.892 SEC 1.638 1.711 1.109 1.356 3.756 1.911 SEC CL 1.895 1.980 1.283 1.569 4.346 2.211 SECV 2.159 2.075 1.957 1.776 3.797 2.180 r 2 0.9022 0.8865 0.8447 0.8457 0.6963 0.8671 Bias 0.980 1.030 0.670 0.810 2.250 1.150 Bias CL 2.105 2.199 1.425 1.743 4.827 2.456 SEP 2.130 2.220 1.440 1.760 4.880 2.480 Slope 0.848 0.807 0.839 0.822 0.981 0.745 Intercept 8.77 11.1 8.65 9.41 1.99 14.6 R.S.D. 1.704 1.834 2.143 2.139 2.914 1.984 jBias! Bias CL -1.125 -1.169 -0.755 -0.933 -2.577 -1.306 JBiasl Bias CL? Yes Yes Yes Yes Yes Yes POR PLS MPLS 2,5,5 2,10,10 2,5,5 2,10,10 2,5.5 2,10,10 R0.820 0.790 0.780 0.760 0.420 0.490 SEC 2.880 3.100 3.330 3.470 5.380 4.850 SEC CL 3.332 3.587 3.853 4.015 6.225 5.611 SECV 3.170 3.560 3.830 3.900 5.690 4.780 r20.8120 0.7730 0.8530 0.8040 0.6910 0.6690 Bias 1.730 1.860 2.000 2.080 3.230 2.910 Bias CL 3.701 3.984 4.280 4.459 6.914 6.233 SEP 3.740 4.030 4.330 4.510 6.990 6.310 Slope 0.9180 0.9840 0.9530 0.6120 1.1200 0.8650 Intercept 3.4700 -0.3600 2.3100 4.6300 -7.5100 6.1800 R.S.D. 3.053 3.363 2.836 3.089 3.911 4.002 Bias! Bias CL -1.971 -2.124 -2.280 -2.379 -3.684 -3.323 Bias! Bias CL? Yes Yes Yes Yes Yes Yes SUBSTITUTE SHEET (RULE 26) WO 97/21091 PCT/AU96/00776 -31- Table 5. Standard error of laboratory determination (SEL) Minu SEL =5 0.116 5.780 0.078 notavauhbie 0.314 SEL CL (using mean SEL) 1.03 7.319 0.101 not A~abj 0.408 SEL CL (using median SEL) 1.024 7.140 0.111 noat avaiable 0.351 SUBSTITUTE SHEET (RULE 26) WO 97/21091 WO 9721091PCT/AU96/00776 -32- Table Sa. Components of possible prediction equations from stepwise and step-up regression analyses.

COeffiaiwt Waawmd Coeffidmi Wwent FjM~v to shear Rgresdn uiysswp ta~~ Mathematcal besnui .515 (2 trMs) Ism9 28.09 2482.05 2048 103877 2048 .425M6 1686N 70012 198 3=.49 1488 _19.88 1718 5148.38 w 1=.46 116s Eneamy~ure to caoms Rersso -nlf S "Stepv Mathanatical trafnntd 2.1010(4W=rn) 105 terns) 4X71 2.49 -2802 228-1.05 1728 112.57 1888 -108.89 184a .73.48 1728 -405M9 1268 4111.04 1266 to cornlaiute -erso -myi SWPW StOP4JP Mathenatical benmn -Z10.5 (4 tarw) 2.10.5(I term) 231.42 -89.08 43005w3 2128 .1521.33 286 4290.19 2408 -4M6512 2018 1=4.74 1138 Omqbft of. maUfer In vo Mathmatical beabhien 2.5.56 tuns .5 ra trm) 46.62 48.16 .367.00 Is18 -612.43 16o8 -879.6 1238 252.82 1418 8162.72 1188 -943.77 1618 1249.01 168 519.46 1906 -161.84 2248 of In ue r Wo -emso -rd~ _""stpU Matherratcal breamn 25.5 (5 Zen) -210,10 (4 temaI) 62.43 54.2 -556.0 1918 .1171.70 1698 981.30 1Me 311.12 1418 -2186.89 166G -2W57.69 1618 2050 2158 -2319.61 122B 4901.99 1748 SUBSTITUTE SHEET (RULE 26) Table 6b. Components of possible prediction equations from multivariate regression analyses.

Energy required to shear I Energy requlred to compress Energy required to comminute PCR jPCR (2.I.61 PCR P15 (2.6,6) Coefficient Wavelength Coefficient Wavelength Coetficlent Wavelength Coefficient Wavelength -3.33 18.1 1.76 -0.2 -0.84 1.15 -0.85 0.13 0.64 -0.04 -0.84 -1.1 -1.71 -1.73 -0.85 0.1:2 -0.9 -1.25 -0.25 0.2 -0.1 -0.68 1.25 2.34 -2.37 -10.62 .9 -1.68 8.65 23.88 13.87 -12.53 1108 1118 1128 1138 1148 1158 1168 1178 1188, 1198 1208 1218 1228 1238 1248 1258 1268 1278 1288 1298 1308 1318 1328 1338 1348 1358 1368 1378 1388 1398 1408 1418 1428 3.35 0.17 0.02 -0.01 -0.01 0.01 -0.01 0 0.03 0.04 0 -0.03 -0.05 -0.04 -0.01 0 0 -0.01 0 0.01 0.02 0.03 0.04 0.06 0.07 0.03 -0.06 -0.15 -0.01 0.16 0.3 0.49 0.2 -0.37 1108 1118 1128 1138 1148 1158 118 1178 1185 1198 1208 1218 1228 1238 1248 1258 1268 1278 1288 1298 1308 1318 1328 1338 1348 1358 1368 1378 1388 1398 1408 1418 1428 -22.8 93.07 8.*5 -7.27 -0.79 3.98 -10.39 -4.8 13.75 19.07 5.08 -8.92 -20.83 -17.9 3.26 05 -2.55 -4.67 1.29 5.24 5.9 9.22 14.26 22.7 27.63 8.84 -26.23 -66.38 -1.83 67.77 125.67 188.24 67.19 -153.54 1108 1118 1128 1138 1148 1158 1188 1178 11es 1198 1208 1218 1228 1238 1248 1258 1288 1278 1288 1298 1308 1318 1328 1338 INS8 1358 1388 1378 1388 1398 1408 1418 1428 -16.44 91.01 7.1'9 -5.83 -0.13 5.58 -8.55 -4.72 13.27 17.38 0.44 -13.96 -23.8 -15.15 -0.82 1.6 -0.82 3.84 1.1 4.93 7.12 10.4 18.14 23.58 27.85 10.27 -24.35 -60.7 2.2 67.91 115.99 174.79 82.89 -139.99 1108 I118 1128 1138 1148 115 1168 1178 I1se 1198 1208 1218 1228 1238 1248 1258 1268 1278 1288 1298 1308 1318 1328 1338 1348 1358 138 1378 1388 1398 1408 1418 1428 Table 6b. Components of possible prediction equations from multivariate regression analyses.

Energy required to shear Energy required to compress Energy required to comminute PCR PCR PCR PLS (2.6.6) Coefficient Wavelength Coefficient Wavelength Coefficient Wavelength Coefficient Wavelength -10.01 1438 -0.39 1438 145.28 1438 -137.82 1438 -1.04 1448 -0.21 1448 -66.74 1448 -75.27 1448 2.09 1458 -0.13 1458 -37.31 1458 -47.38 1458 -2.69 1468 -0.11 1468 -49.09 1468 -48.54 1468 -0.05 1478 -0.14 1478 -70.4 1478 -59.94 1478 -1.2 1488 -0.06 1488 -26.64 1488 -28.06 1488 -2.08 1498 -0.01 1498 -1.87 1498 -3.2 1498 -1.69 1508 0.01 1508 4.61 1508 3.33 1508 0.35 1518 0.07 1518 28.89 1518 28.91 1518 2.8 1528 0.12 1528 49.84 1528 48.03 1528 -0.01 1538 0.08 1538 34.38 1538 33.02 1538 -1.83 1548 0 1548 -3.32 1548 0.49 1548 -3.7 1558 -0.04 1558 -21.79 1558 -17.62 1558 -0.66 1568 -0.01 1568 -4.61 1568 -3.35 1568 -1.99 1578 -0.04 1578 -13.08 1578 -15.39 1578 S-3.67 1588 -0.09 1588 -43.91 1588 -40.21 1588 _i -5.26 1598 -0.07 1598 -30.2 1598 -32.77 1598 0.06 1608 -0.01 1608 -7.28 1608 -5.93 1608 2.89 1618 0.04 1618 17.4 1616 15 1018 C: 0.22 1628 0.09 1628 34.68 1628 33.13 1628 S-0.98 1638 0.13 1638 49.47 1638 48.83 1638 16.23 1648 0.1 1648 53.35 1648 44.32 1648 10.67 1658 0 1656 -3.16 1658 -2.24 1658 6.52 1668 -0.22 1668 -82.09 1668 .80.94 1668 -20.53 1678 -0.07 1678 -60.38 1678 -36.22 1678 -6.15 1688 0.15 1688 55.65 1688 60.49 1688 7.4 1668 0.06 1698 48.43 1698 42.6 1698 4.76 1708 0.08 1708 34.19 1708 27.79 1708 -19.73 1718 -0.09 1718 -54.88 1718 -46.38 1718 -5.98 1728 -0.13 1728 -54.69 1728 -69.44 1728 19.24 1738 0.15 1738 78.27 1738 63.67 1738 6.42 1748 0.19 1748 90.94 1748 91.13 1748 -3.1 1758 0.06 1758 21.41 1758 20.27 1758 -4.03 1768 -0.1 1768 -47.2 1768 -48.58 1768 -1.47 1778 -0.11 1778 -42.48 1778 -39.05 1778 -0.44 1788 -0.09 1788 -36.22 1788 -33.03 1788 1.72 1798 -0.01 1798 -2.33 1798 -3.76 1798 0 8 o '0 '0

I?

I

0 0% -4 94 Table 6b. Components of possible prediction equations from multivariate regression analyses.

Energy required to shear Energy required to compress Energy required to comminute PCR PCR PCR PLS (2.6,6) Coefficient Wavelength Coefficent Wavelength Coefficient Wavelength Coefficient Wavelength 2.76 1808 0.05 1808 20.97 1808 18.41 1808 .3.79 18 -0.01 1818 56 1818 -5.17 1818 -4.32 1828 -0.07 1828 -30.81 1828 -26.5 1828 -2.97 1838 -0.05 1838 -17.29 1838 -17.64 1838 1.97 1848 0.03 1848 17.65 1848 15.59 1848 -2.48 1858 0.06 1858 28.06 1858 28.11 1858 -6.48 1868 0.08 1668 42.2 1868 40.46 1868 -10.22 1878 0.23 1878 115.52 1878 108.98 1878 -1.64 1888 0.44 1888 219.15 1888 208.1 1888 31.11 1898 0.2 1898 98.82 1898 99.42 1898 2.3 1908 -0.35 1908 -179.9 1908 -157.11 1908 -25.69 1918 -0.47 1918 -251.3 1918 -220.58 1918 -12.22 1928 -0.34 1928 -171.93 1928 -172.09 1928 15.11 1938 -0.14 1938 -55.38 1938 -77.27 1938 26.89 1948 -0.03 1948 -6.17 1948 -23.21 1948 27.72 1958 -0.04 1958 -16.65 1958 -19.03 1958 6.93 1968 0.05 1968 12.02 1968 22.46 1968 -15.33 1978 0.23 1978 66.03 1978 86.86 1978 -9.25 1988 0.29 1988 85.84 1988 103.8 1988 -3.33 1998 0.27 1998 78.88 1998 988.03 1998 1.28 2008 0.32 2008 97.53 2008 115.14 2008 20.22 2018 0.25 2018 104.79 2018 99.25 2018 18.34 2028 0.08 2028 42.06 2028 33.39 2028 9.56 2038 -0.01 2038 23.98 2038 11.1 2038 5.59 2048 0.13 2048 101.85 2048 77.38 2048 -8.54 2058 0.21 2058 126.79 2058 109.6 2058 -16.75 2068 -0.2 2068 -75.99 2068 -71.48 2068 -18.33 2078 -0.40 2078 -206.62 2078 -188.9 2078 -10.13 2068 -0.38 2088 -151.16 2088 -143.85 2088 -5.78 2098 -0.26 2098 -107.24 2098 -104.89 2098 -8.44 2108 -0.26 2108 -108.06 2108 -107.49 2108 6.28 2118 -0.23 2118 -91.06 2118 -91.23 2118 -4.49 2128 -0.23 2128 -97.55 2128 -93.37 2128 -16.58 2138 -0.2 2138 -101.35 2138 -85.08 2138 -9.08 2148 -0.01 2148 -13.78 2148 -9.31 2148 -2.68 2158 0.08 2158 37.5 2158 38.99 2158 3.19 2168 0.26 2166 118.43 2168 110.03 2168 Table 6b. Components of possible prediction equations from multivarlate regression analyses.

Energy required to shear Energy required to compress Energy required to commlnute PCR PCR PCR PLS (2.6.6) Coefficient 15Wavelength Coefficient 2.Wavelengith Coefficient (.Wavelength CoefficientL Wavelength 4.27 2178 0.28 2178 122.68 2178 109.87 2178 -6.3 2188 0.16 2188 54.73 2188 58.98 2188 .13.33 2198 0.15 2198 38.08 2198 53.78 2198 -2.74 2208 0.35 2208 129.58 2208 147.19 2208 35.77 2218 0.38 2218 171.77 2218 149.55 2218 30.36 2228 0.42 2228 182.8 2228 168.64 2228 22.91 2238 0.22 2238 73.19 2238 86.39 2238 98.61 2248 -0.67 2248 -152.9 2248 -250.89 2248 -15.77 2258 -0.47 2258 -184.5 2258 -189.12 2258 -85.22 2268 -0.2 2268 -167.23 2268 -101.11 2268 -35.1 2278 -0.45 2278 -214.87 2278 -180.14 2278 27.27 2288 0.22 2288 148 2288 122.32 2288 4.27 2298 0.84 2298 352.88 2298 341.16 2298 -13.08 2308 0.29 2308 78.88 2308 63.66 2308 0.87 2318 -0.48 2318 -208.27 2318 -207.81 2318 -13.34 2328 -0.47 2328 -167.95 2328 -150.99 2328 -23.3 2338 -0.2 2338 -79.26 2338 -66.98 2338 0.66 2348 0.15 2348 62.44 2348 40.83 2348 7.96 2358 -0.07 2358 -23.09 2358 -30.25 2358 -15.62 2368 0.08 2368 25.53 2368 41.9 2368 -16.39 2378 0.05 2378 -3.86 2378 12.67 2378 4.44 2388 -0.05 2388 -33.15 2388 -26.65 2388 21.16 2398 0.2 2398 98.53 2398 79.92 2398 49.9 2408 0.22 2408 130.87 2408 89.16 2408 22.34 2418 0.29 2418 120.05 2418 107.71 2418 -1.47 2428 0.22 2428 72.57 2428 84.88 2428 17.19 2438 0.03 2438 8.72 2438 3.78 2438 15.21 2448 0.11 2448 55.93 2448 45.74 2448 -14.12 2458 0.16 2458 59.89 2458 71.29 2458 -24.15 2468 -0.04 2468 -31.79 2468 -13.92 2468 Table 6b. Components of possible prediction equations from multivarlate regression analyses.

Digestibility of dry matter In vitro Digestibility of dry matter In vive PLS (2.5,61) PCR (2,5,61) PLS (2.5.6) Coeficient Wavelength Coefflclen Wavelength Coefilcent Wavelength 59.77 -96.28 7.55 12.25 6.65 6.74 11.89 4.45 -10.06 -29.8 -20.43 -20.05 -15.55 2.41 6.62 8.6 7.47 -1.32 -7.39 -0.79 3.48 5.1 6.23 8.48 17.78 24.61 -0.07 -23.89 .29.88 -16.97 23.92 60.7 55.51 -0.3 1108 1118 t1128 1138 1148 11658 1168 1178 1188 1198 1208 1218 1228 1238 1248 1258 1268 1278 1288 1298 1308 1318 1328 1338 1348 1358 1368 1378 1388 1398 1408 1418 1428 40.56 -161.6 12.91 22.04 19.51 8.19 5.29 -0.92 -11.1 -37.4 -26.13 -38.03 -35.64 -13.39 0.89 14.13 17.48 -7.44 .22.67 -1.05 9.23 13.17 15.6 25.91 40.11 51.12 4.6 -88.33 -68.78 5.08 32.14 76.51 90.64 -1.47 1108 1118 1128 1138 1148 1158 1168 1178 1188 11968 1208 1218 1228 1238 1248 1258 1268 1278 1268 1298 1308 1318 1328 1338 1348 1358 1388 1378 1388 1398 1408 1418 1428 63.84 -78.7 4 8.68 3.94 7.08 6.24 0.43 -5.6 -15.1 -10.61 -15.98 -13.186 2.93 6.37 7.08 5.99 -0.25 -4.08 0.15 3.72 4.38 6.03 9.03 13 13.51 0.260 -13.81 -8.13 -1.4 16.06 32.08 11.94 -14.76 1108 1118 1128 1138 1148 1158 1168 1178 1188 1196 1208 1218 1228 1238 1248 1258 1268 1278 1288 1298 1308 1318 1328 1338 1348 1358 1368 1378 1388 1398 1408 1418 1428 Table 6b. Components of possible prediction equations from multivarlate regression analyses.

Digestibility of dry matter In vitro Digestibility of dry matter In vive PLS PCR PL (2.5.6) Coefficient Wavelength Coefficient Wavelength Coefficient Wavelength -28.21 1438 -11.15 1438 -19.94 1438 -32.57 1448 10.5 1448 -19.24 1448 -28.08 1458 18.31 1458 -16.15 1458 1.48 1468 -3.82 1468 3.7 1468 18.21 1478 -16.48 1478 10.65 1478 .0.65 1488 -25.14 1488 -0.75 1488 -22.99 1498 -74.43 1498 -7.65 1498 -23.44 1508 -75.49 1508 -9.41 1508 -15.74 1518 -58.78 1518 -8.08 1518 -9.09 1528 -25.77 1528 -5.21 1528 -2.8 1538 -7.83 1538 -3.63 1538 10.62 1548 17.83 1548 4.58 1548 28.78 1558 52.96 1558 15.5 1558 6.02 1568 21.09 1588 2.31 1568 -6.8 1578 -27.42 1578 -3.09 1578 -16.09 1568 -42.59 1588 -5.77 1588 -9.20 1598 -63.69 1598 -8.48 1598 4.88 1808 -3.06 1608 -0.65 1608 -3.82 1618 -1.62 1618 -0.87 118 -3.50 1626 -2.81 1828 0.24 1828 0.55 1838 -9.16 1638 3.42 1638 1648 13.81 1648 0.86 1848 23.87 1658 61.39 1658 13.12 1658 54.92 1668 159.6 168 28.91 1868 78.84 1678 101.05 1678 44.01 1878 -13.9 1688 -79.38 1688 -8.19 1688 -74.54 1698 -77.73 1698 -34 1698 -38.63 1708 -43.87 1708 -18.38 1708 24.48 1718 21.34 1718 2.17 1718 -17.91 1728 -25.8 1728 -31.77 1728 -43.01 17368 -40.51 1738 -21.21 1738 -29.25 1748 -28.11 1748 -5.24 1748 -18.33 1758 -4.55 1758 -11.67 1758 0.43 1768 2.68 1768 -0.49 1768 28.21 1778 20.91 1778 21.37 1778 18.92 1768 13.59 1788 13.25 1788 0.54 1798 9.35 1798 -0.93 1798 0 fr 0 a.' Table 6b. Components of possible prediction equations from mulivarlate regression analyses.

Digestibility of dry matter In vitro IDIgestibllty of dry matter In vivo PLS 1PCR PL5 (2.6.6) Coeffileid Wavelengh Coefficiert Wavelengh Coefficien Wavelengh -2.44 -2.72 -5.84 -4.37 -8.79 -7.72 -29.93 -98.18 -116.18 117.59 185 33.91 -35.31 -44.59 -9.25 35.73 28.58 10.68 10.98 65.12 83.13 7.23 0.99 -10.85 -84.46 -122.91 -35.85 34.94 28.83 18.03 5.09 -9.58 9.79 23.04 410.93 -16.07 -41.78 1808 1818 1828 1838 1848 l8o8 1868 1878 1888 1698 1908 1918 1928 1938 1948 1958 1968 1978 1986 1998 2008 2018 2028 2038 2048 2058 2068 2078 208 2098 2108 2118 2128 2138 2148 2158 2168 7.17 -15.36 -29.4 -16.21 2.3 -0.87 -21.29 -82.33 -102.15 211.27 204.51 -3.12 16.14 35.39 -8.24 -11.32 -37.1:5 -44.28 -81.81 -72.2 -63.31 37.37 183.21 156.66 -2.09 -178.03 -104.9 -7.7 62.28 54.28 14.14 -40.31 2.34 28.94 -31.58 3.05 -66.48 1818 1828 1848 1858 1868 1878 1888 1898 1908 1918 1928 1938 1948 1958 1968 1918 19M 1998 2008 2018 2028 2038 2048 2058 2088 2078 2088 2098 2108 2118S 2128 2138 2148 2158 2168 -2.3 -3.37 -0.05 -2.23 -3.79 -9.81 434.56 -52.89 38.2 68.78 28.1 -3.79 -19.45 -6.41 15.95 13.49 2.03 0.57 31.07 38.57 -1.21 -9.38 -14.51 -54.72 -68.29 -1.69 37.55 27.17 16.01 15.52 7.88 13.25 M649 -8.42 -9.12 -27.35 1828 1835 1848 1858 1888 1878 1888 1898 1Iw8 1918 1928 1938 I94 1958 1968 1978 1988 1998 2008 2018 2028 2038 2048 2058 208 2078 2088 2098 2108 2118 21,28 2138 2148 2158 218 Table 6b. Components of possible prediction equations from multivariate regression analyses.

Digestibility of dry matter In vitro Digestibility of dry matter In vivo PLS PCR (2,6.56) PLS Coefficient Wavelength Coefflclent Wavelength Coefficenl Wavelength -48.69 2178 -107.52 2178 -32.56 2178 -14.5 2188 -54.54 2188 -12.56 2188 -0.14 2198 -11.17 2198 4.5 2198 -7.15 2208 -2.14 2208 4.87 2208 -48.95 2218 -43.68 2218 -30.89 2218 -18.22 2228 -0.01 2228 -19.94 2228 68.33 2238 100.18 2238 14.89 2238 -24.11 2248 53.32 2248 -53.79 2248 55.99 2258 81.52 2258 18.98 2258 110.06 2268 48.93 2268 90.27 2268 52.16 2278 -9.18 2278 63.96 2278 -89.38 2288 25.1 2288 -35.59 2288 -109.99 2298 -47.83 2298 -77.96 2298 -54.11 2308 -23.3 2308 -81.38 2308 17.63 2318 -73.92 2318 22.34 2318 23.71 2328 -23.74 2328 46.6 2328 52.19 2338 13.64 2338 37.70 2338 -58.18 2348 -21.67 2348 -48.18 2348 -21.28 2358 -77.29 2358 -7.88 2358 4.01 2368 -88.75 2368 16.34 2368 32.53 2378 26.42 2378 17.8 2378 35.58 2388 68.01 2388 1.95 2388 -21.25 2398 22.17 2398 -27.54 2398 -70.01 2408 -26.96 2408 -50.39 2408 -16.86 2418 6.02 2418 -18.09 2418 61.66 2426 75.99 2428 17.75 2428 14.94 2438 56.83 2438 -3.66 2438 -11.88 2448 6.46 2448 -11.21 2448 5.35 2458 -4.33 2458 3.97 2458 10.49 2468 -33.25 2468 11.29 2468 Table 7. Table 7. Descriptions of forages used In Table 8.

SampleVait rnonam In Genus Species Vait ouo ae Part of Process undergone Stage of maturity Regrowth Table 8 0_ plant I Panlcum coloralum Kabulabula CPI 16796 Makarikad gross aerial dried and chaffed mid bloomn (9 weeks' regrowth) mid bloomn regrowth 2 Penlcum maximum Coloniso Guinea grass aerial dried and chaffed vegetative regrowth (4 weeks') vegetative regrowth 3 Panlcum coloratum Barnbatul Makarlcar grass aerial dried and chaffed mid bloomn (I mnths regrowth) mid bloomn regrowth 4 Panlcum maximum Hami Guinea grass aerial dried and chaffed earty bloom (I montl~e regrowth) early bloomn tegrowth Panleum coloratum vor Makarikasense Burnett Makarikai grass salW dried and chaffed mid bloomn (6 weeks' regrowth) mid bloom regrowth a Pasicumn masxkum ver. bichogfurm. Pett Green Panic aerial dried and chaffed midd bloom (4 wee". regrowth) mid bloomn regrowth Table 8. Examples of energy required to shear, digestibility of dry matner In vivo, forage consumption constraint (FCC), and voluntary feed consumption (VFC) Sample Energy required to Digestibility of dry In shear. matter In vivo, Predicted FCC*a Predicted VFC6 Actual VFC Actual VFC Table 7 predicted using NIR' predicted using WeR (g OM/dIMBW)' (g OM/dIMBW) (g OM/MBW) (g OM/d) (kJ/m 2

M%

1 20.61 51.29 88.85 32.52 30.77 534 2 16.70 54.73 66.92 44.95 39.47 Gas 3 13.75 56.69 51.49 58.51 48.79 848 4 13.16 59.59 48.41 54.34 53.58 931 6 17.52 55.18 71.21 39.74 43.66 759 6 18.63 55.88 66.55 43.01 45.68 793 Predicted using the calibration equation from stepwise regression analysis (Table Be).

2 Predicted using the calibration equation from stepwise regression analysis (Table 6a).

3 Predicted using predicted energy required to shear. and the relationship between energy required to shear and FCC.

4'Calculated from predicted FCC and predicted digestibility of dry matter In vlvo. 7 Abbreviations used: organic matter metabolic body weight (MBW) BIN" 1

Claims (19)

1. A method for determining a biomechanical property of a feed, the method comprising the steps of: subjecting the feed to infrared radiation to obtain spectral data; and using the spectral data to determine the biomechanical property; whereby, the biomechanical property of the feed is determined on the basis of the bond energies of the chemical constituents of the feed.

2. A method according to claim 1 wherein the biomechanical property is selected from the group comprising; shear energy, compression energy, communition energy, tensile strength, shear strength and intrinsic shear strength.

3. A method according to claim 1 or 2 wherein the biomechanical property of the feed is determined directly from the spectral data.

4. A method according to claim 1 or 2 wherein the spectral data is used to determine another property of the feed and the other property is used to determine the biomechanical property on the basis of a correlation between the other property and the biomechanical property.

A method according to claim 4 wherein the other property is ADF content, NDF content or lignin content.

6. A method according to any one of the preceding claims wherein the spectral data is a reflectance spectrum over a predetermined range of wavelengths.

7. A method according to claim 6 wherein the predetermined range is approximately 700nm to 3000nm. AMENDED SHEET IPEA/FAU -43-

8. A method according to claim 6 wherein the predetermined range is approximately 1100nm to 2500nm.

9. A method according to any one of claims 6 to 8 wherein the data obtained for the spectral range of approximately 1850nm to 1970nm is disregarded.

10. A method according to any one of claims 6 to 9 wherein the spectral data is recorded at 2nm intervals over the predetermined range.

11. A method according to claim 6 wherein the reflectance reading is taken at a combination of wavelengths.

12. A method according to claim 11 wherein the combination of wavelengths is selected from the group comprising: 1168nm, 1458nm, 1598nm, 1718nm, 1828nm, 2048nm, 1138nm, 2018nm, 2128nm, 2408nm, 1268nm, 1588nm, 1728nm, 2278nm, 1158nm, 1238nm, 1668nm, 1908nm, 2248nm, 1698nm, 1748nm, 1918nm and 2158nm.

13. A method according to claim 11 wherein the combination of wavelengths is le: 15 1168nm, 1458nm, 1598nm, 1718nm, 1828nm and 2048nm and the biomechanical property is shear energy.

14. A method according to claim 11 wherein the combination of wavelengths is 1268nm, 1588nm, 1728nm and 2278nm and the biomechanical property is compression energy.

15. A method according to claim 11 wherein the combination of wavelengths is 1138nm. 2018nm, 2128nm and 2408nm and the biomechanical property is comminution energy.

16. A method for determining a biomechanical property of a feed, the method comprising the steps of: subjecting the feed to infrared radiation to obtain spectral data; and PCT/AU9 6 0 0 7 7 6 -44 comparing the spectral data obtained in with a calibration equation to determine the biomechanical property; whereby, the biomechanical property of the feed is determined on the basis of the bond energies of the chemical constituents of the feed.

17. A method according to claim 16 wherein the calibration equation is yi 19.95

10239.46 R 1 6s 3623.49 R 1 45 4255.61 Rise 5319.88 R 1 71 8 5148.38 R1828 2452.05 R204 and the biomechanical property is shear energy(yi).

18. A method according to claim 16 wherein the calibration equation is y2 231.42

18224.74 R113 4955.12 R 2018 3005.37 R 2128 4290.18 R 2 408 and the biomechanical property is comminution energy (y 2

19. A method according to claim 16 wherein the calibration equation is y3 -0.71 911.04 Ri268 112.57 Rises 79.48 R 1728 28.02 R 2 2 78 and the biomechanical property is compression energy (y 3 A method according to any one of claims 16 to 19 wherein the calibration equation is determined from laboratory data establishing a correlation between reflectance and the biomechanical property. 21. A method according to any one of claims 1 to 20 wherein an additional property of the feed is also determined. 22. A method according to claim 21 wherein the additional property of the feed is digestibility of dry matter in vivo or in vitro. 23. A method for determining feed quality, the method comprising the steps of: subjecting the feed to infrared radiation to obtain spectral data; using the spectral data to determine a biomechanical property of the feed; and AMENDED SHEET IPFA/AU ?CT/AU 96 0 e, g0 using the biomechanical property obtained in step to determine feed quality; whereby, the biomechanical property of the feed and thus feed quality is determined on the basis of the bond energies of the chemical constituents of the feed. 24. A method according to claim 23 wherein the feed quality is determined as a measure of voluntary feed consumption (VFC). A method according to claim 23 wherein the feed quality is determined as a measure of forage consumption constraint (FCC). 26. A method substantially as herein described with reference to the description of the examples. 27. A spectrometer configured to carry out the method according to any one of claims 1 to 22 wherein the spectrometer is adapted to receive a sample of feed and determine a biomechanical property of the feed. 28. A spectrometer configured to carry out the method according to any one of claims 23 to 25 wherein the spectrometer is adapted to receive a sample of feed and determine the quality of the feed. 29. A spectrometer according to claim 27 or 28 further comprising a data processing means for determining the biomechanical property or the quality of the feed. AMENDED SHEET IPEA/AU