Using plant wax markers to estimate diet ... - CSIRO Publishing

CSIRO PUBLISHING

www.publish.csiro.au/journals/ajar

Australian Journal of Agricultural Research, 2007, 58, 1215–1225

Using plant wax markers to estimate diet composition and intakes of mixed forages in sheep by feeding a known amount of alkane-labelled supplement E. CharmleyA,C and H. DoveB A

Agriculture and Agri-Food Canada, Crops and Livestock Research Centre, Nappan, Nova Scotia B0L 1C0, Canada. Current address: CSIRO Livestock Industries, PO Box 5545, Rockhampton Mail Centre, Qld 4702, Australia. B CSIRO Plant Industry, GPO Box 1600, Canberra, ACT 2601, Australia. C Corresponding author. Email: [email protected]

Abstract. The feeding of known amounts of supplements to grazing animals can be accomplished relatively easily. If the supplement and the other diet components have distinctive profiles of cuticular wax n-alkanes, then the supplement intake and the alkane profiles of the supplement, other dietary components and faeces can be used to estimate the proportions and hence intakes of several forages by the grazing animal. However, this method requires knowledge of recoveries of n-alkanes in faeces. Twenty four wethers were fed one of four diets comprising equal combinations of 1, 2, 3 or 4 forages. Forages used were subterranean clover, phalaris, annual ryegrass and wheat straw. Forages were chopped using a chaff cutter and fed with solvent-extracted cottonseed meal (CSM) labelled with beeswax and synthetic C28 alkane to provide a characteristic alkane profile. Faecal grab samples were taken from sheep from 14 to 23 days after administration of an intra-ruminal controlled-release device (CRD) containing 1 g of each of C32 and C36 alkane. Total faeces were collected from half the sheep on each treatment in order to measure alkane recoveries in individual sheep. Faecal concentrations of the n-alkanes C25 to C31 and C33 were corrected for recovery using the individual sheep value, the treatment mean or the grand mean for all four treatments. Dietary compositions were then estimated from corrected faecal concentrations of n-alkanes using a least-squares procedure and, together with the known supplement intakes, were used to estimate the intakes of all other diet components. Estimates from this ‘labelled supplement’ method were compared with the amounts fed or those estimated using the alkanes derived from the CRD. The labelled supplement method accurately and precisely estimated dietary component proportions and intake for all treatments when measured recoveries for individual sheep were used. Precision declined when recovery was based on estimated recoveries for treatment means or the grand mean. Estimates of intake based on dietary C33 and the measured release rates of C32 or C36 alkanes from the CRD did not differ from measured intakes. Estimates based on the C32 : C31 alkane pair over-estimated intake. Estimates of wholediet digestibility based on the various ways of estimating intake were all very close to the digestibilities calculated from directly-measured intakes and faecal outputs. It is concluded that the feeding of a known amount of supplement can be successfully used to estimate dietary proportions and hence intakes of diet components, in mixed diets with up to five ingredients, but this approach requires estimates of faecal alkane recovery. Additional keywords: beeswax, controlled-release device, cottonseed meal, phalaris, ryegrass, subterranean clover, wheat straw.

Introduction The n-alkanes of plant cuticular wax have been used as markers to estimate feed intake since the late 1980s (Mayes et al. 1986). In the conventional use of this approach, a known amount of an even-chain alkane is given to the animal either by daily dosing, by labelling a concentrate with C32 or C36 alkane (e.g. Unal and Garnsworthy 1999), or by administering an intra-ruminal controlled-release device (CRD; Dove et al. 2002a). Intake is estimated from the faecal ratio of the dosed even-chain alkane and an adjacent odd-chain alkane originating from the forage, the measured alkane concentrations of these two alkanes in the forage and the known dose of exogenous alkane (Mayes © CSIRO 2007

et al. 1986; Dove and Mayes 1991). As with all marker grazing techniques, a limitation is obtaining a representative sample of the forage consumed (Penning 2004). Nevertheless, the alkane method has been demonstrated to work in sheep (Mayes et al. 1986; Dove and Mayes 2005), but relatively less work has been done with cattle. It requires that the recovery in faeces of the alkane pair is equal; validation studies have shown that these recoveries are usually close (Dove and Mayes 2005). The estimate of intake is directly related to the release rate of evenchain alkane from the CRD, and thus can be compromised by irregular alkane release from the CRD (Charmley et al. 2003). In order to obtain absolute estimates of intake, the 10.1071/AR07187

0004-9409/07/121215

1216

Australian Journal of Agricultural Research

actual alkane release rate under the conditions of the current experiment must be known (Ferreira et al. 2004), although a comparison of estimated intakes between experimental treatments is still valid when based on the manufacturer’s stated release rates. Release rate can be determined by measuring the end-point of release by frequent faecal sampling or by measurement of the disappearance of CRD payload in the rumen (Dove et al. 2002a). Estimating release rate thus requires additional animal intervention, which can be difficult under certain conditions. Different plant species, and even cultivars within species, have different patterns of alkane concentration within their cuticular wax. These characteristic alkane patterns can be used to differentiate between forages (and other components) and to estimate their contribution to the diet from the aggregated faecal concentration of alkanes, using least-squares principles (Dove and Moore 1995; Dove et al. 2002b; Lewis et al. 2003). If diet composition can be estimated, it follows that if one feed component is given in known quantity, it is possible to estimate the intake of all other feed components. In grazing livestock, one component, the concentrate supplement, is often fed or could be fed in known amounts. For example, dairy cows receive a known concentrate allowance on a daily basis. If the supplement does not contain a distinctive alkane profile (e.g. cereals, protein meals), it can be labelled with an alkane mixture. Elwert and Dove (2005) used this approach to obtain accurate estimates of intake of a single forage (subterranean clover) fed in varying proportions with beeswaxlabelled cottonseed meal. This approach (‘labelled supplement’) of using the supplement as the means for estimating diet composition and thence component intakes eliminates the requirement for separate dosing with alkanes, thus removing the need to ensure and verify steady-state delivery rates either from pulse dosing (Sibbald et al. 2000) or CRDs (Dove et al. 2002a). In effect, it employs a mix of alkanes as ‘the dose’, but to be used requires estimates of faecal alkane recovery because the first step in the method is the estimation of diet composition. The purpose of this research was to extend earlier work with the labelled supplement approach and only a single source of forage (Dove et al. 2002b; Elwert and Dove 2005) to consider diets containing supplement and up to four forages. An alkane-enriched oilseed supplement was used for alkane delivery and emphasis was placed on determining the critical points which affected either the accuracy or precision of the method. The labelled supplement method was also compared with the conventional use of intra-ruminal CRD for alkane dosing. Materials and methods Experimental animals, design and diets Twenty four Merino-cross wethers (38 to 46 kg live weight (LW)) were used in a 37 day experiment with four diets (Table 1). Each diet comprised 660 g/day (air-dry) forage and 150 g/day (air-dry) cottonseed meal (CSM) supplement (dry matter digestibility (DMD) 0.662). Forages were chopped using a chaff cutter (Star Farm Machinery Manufacturing Co., Ltd,

E. Charmley and H. Dove

Table 1.

Component composition of the diet (g/kg DM offered) and total organic matter intake (g/day) by sheep

Beeswax-labelled cottonseed meal Subterranean clover Phalaris Annual ryegrass Wheat straw Organic matter intake

Diet 1

Diet 2

Diet 3

Diet 4

191

191

191

191

809 0 0 0 672

404 404 0 0 675

270 270 270 0 668

202 202 202 203 679

Japan). The forage used in diet 1 (one forage component), was subterranean clover (Trifolium subterraneum; DMD 0.743), diet 2 (two forage components) contained equal amounts of subterranean clover and poor-quality phalaris (Phalaris aquatica; DMD 0.376), diet 3 (three forage components) contained equal amounts of subterranean clover, phalaris and annual ryegrass (Lolium rigidum; DMD 0.636) and diet 4 (four forage components) contained equal amounts of subterranean clover, phalaris, annual ryegrass and wheat straw (Triticum aestivum; DMD 0.537). The supplement was solventextracted CSM labelled with beeswax and C28 alkane as alkane sources. The sheep were ranked according to LW and blocked into three groups of eight animals. Two wethers from each block were assigned at random to one of the four diets, with one from each pair designated for eventual housing in metabolism cages and the other for housing in individual pens. Immediately after assignment, two sheep from diet 1 were selected at random and transferred to diet 4. Thus, in total there were four sheep allocated to diet 1, six each to diets 2 and 3, and eight allocated to diet 4. This was done to account for the greater variance expected with more complex diets. All animals were initially housed in individual pens for 12 day’s adaptation to diets. Half the sheep were then transferred to metabolism cages to facilitate total collection of faeces. On day 14, controlled-release devices (CRD; Captec (NZ) Ltd, Auckland, New Zealand) containing 1 g of each of C32 and C36 were administered to all sheep. According to the manufacturer, the CRD had an expected release period of 20 days, giving an expected daily release rate of 50 mg for each alkane. Total collection of faeces from the 12 sheep in cages and rectal grab sampling of faeces from all sheep commenced on day 21 (7 days after CRD insertion, by which time alkane excretion would have reached equilibrium) and continued for 6 days. Thereafter, two sheep from each treatment remained on feed until 23 days after CRD administration (day 37 of the experiment) for determination of the duration of alkane release from the CRD. Diets (Table 1) were fed once daily in the mornings and water was available at all times. Lights were left on 24 h/day. Samples of the forages and CSM were taken daily. A subsample was used to determine daily DM concentration while the remainder was bulked over 6 days, before freeze-drying for subsequent alkane determination. Feed refused was collected daily, bulked over 6 days and stored frozen before freeze-drying for subsequent alkane analysis. Total faeces were collected daily over 6 days using a chute, fitted with a urine separation device,

Plant wax markers to estimate diet composition


1217

suspended under each metabolism cage. Daily faecal output was recorded 1 day in arrears of feed sampling. Faeces were weighed and a 10% sub-sample taken and frozen for subsequent freeze-drying and analysis for alkane concentration. Faecal DM was determined daily in duplicate for each sheep. Rectal grab samples were taken from all sheep each morning before feeding. Samples taken on days 21 to 26 (corresponding to the total faeces collection period) were bulked for alkane analysis. Subsequently, in order to determine CRD release rate from the extinction of exogenous alkanes in faeces, samples from eight sheep were collected and frozen on a daily basis for a further 12 days, before freeze-drying for alkane analysis. All procedures in this experiment were granted written approval from the Animal Ethics Committee, CSIRO Divisions of Plant Industry and Entomology.

This method requires separate dosing of the animals with a source of exogenous alkanes, and its accuracy is directly related to the accuracy of the release rate of C32 and C36 alkanes from the CRD. Moreover, implicit in Eqn 1 is the assumption that if the two alkanes in the ratio Fj /Fi have equal recoveries, no correction for faecal alkane recovery is required. 2. The labelled supplement method is based on knowing the amount of supplement consumed, then estimating the proportions of supplement and the different forages in the total diet and finally calculating the individual intakes of forages using Eqn 2, in which Is is the intake of supplement, Ps is the proportion of supplement in the diet and Pf is the proportion of a given forage.

Preparation of the labelled supplement Beeswax was added to cottonseed meal as described by Dove et al. (2002b) and Elwert and Dove (2005). Briefly, 45 kg CSM was labelled with 550 g finely grated beeswax which had been dissolved in 2.5 L of n-heptane, with gentle heating. Synthetic C28 alkane (Sigma Aldrich Australia, Castle Hill, NSW) was also added to the solution to provide a final concentration of between 250 and 300 mg/kg organic matter (OM). These additions ensured an alkane profile of the CSM supplement that was markedly different from all other dietary components. The beeswax/C28 solution was sprayed onto CSM as it was mixed in a horizontal mixer. The heptane solvent was then allowed to evaporate from the CSM at ambient temperature overnight.

This approach does not require separate dosing with alkanes, but because it is based on the estimation of diet composition, it requires the correction of faecal alkane concentrations for incomplete faecal alkane recovery. Since cereal and oilseed supplements usually contain only small quantities of cuticular wax alkanes, it was necessary to label our supplement with alkanes; in this experiment, beeswax and C28 alkane were used. The labelled supplement then had a unique alkane composition (Table 2). The proportion of supplement in the diet was estimated using the computer package EatWhat (Dove and Moore 1995) which employs a nonnegative, least-squares procedure to estimate the combination of alkane patterns of diet components which best matches the pattern of recovery-corrected alkane concentrations observed in faeces. The alkanes used in this analysis were C25 to

Analytical procedures The DM content of feed offered, feed refused and faeces was determined by drying to constant weight in a forced air oven at 70◦ C. OM was determined as the weight lost upon ashing feeds and faeces at 550◦ C in a muffle furnace for 24 h. Alkanes were extracted from samples using a direct saponification procedure and quantified by gas chromatography using a modification (Salt et al. 1992) of the method of Mayes et al. (1986); tetracosane (C24 alkane) and tetratriacontane (C34 alkane) were used as internal standards. Calculation of intake and digestibility Intake was estimated by two methods. 1. Estimates of OM intake using the alkane CRD (hereafter ‘CRD method’) were made using alkane pairs and a simplification of the equation given by Mayes et al. (1986): OM intake = Dj /[(Fj /Fi ) × Hi − Hj ]

(1)

where Hi and Fi are the concentrations of the odd-chain alkane in the feeds and faeces, respectively. Hj and Fj are the equivalent concentrations of the even-chain alkane in feed and faeces and Dj is the daily release rate of the evenchain alkanes (C32 and C36) from the CRD. Values for Hi and Hj were based on known diet proportions. Separate estimates of intake were derived for the C32 : C31, C32 : C33 and C36 : C33 alkane pairs.

Intake of forage f = Is × (Pf /Ps )

(2)

Table 2. Alkane concentrations and proportions of the four forages and the beeswax/C28-labelled cottonseed meal used in estimation of diet species composition Subterranean clover

C25 C26 C27 C28 C29 C30 C31 C33

4.9 4.7 52.3 18.1 361 10.9 80.8 6.0

C25 C26 C27 C28 C29 C30 C31 C33

9.0 8.7 97.2 33.5 670 20.2 150 11.2

Phalaris

Annual ryegrass

Wheat straw

Concentration (mg/kg organic matter) 10.7 17.2 5.7 4.2 7.6 5.0 8.4 51.0 29.0 3.8 17.7 6.4 14.2 254 165 3.8 22.9 6.2 22.2 411 198 7.6 77.4 27.9 Proportion (g/kg) 143 20 55.8 8.9 123 59.3 50.8 20.7 190 296 50.0 26.7 296 478 102 90.2

12.9 11.2 65.2 14.5 373 14.0 447 62.9

Labelled cottonseed meal 122 16.1 496 283 338 10.6 262 41.0 77.8 10.3 316 180 215 6.7 167 26.2

1218


C31 and C33, and recovery corrections were made in one of three ways: a) using alkane recovery data for the individual sheep in metabolism cages, based on their alkane intakes (adjusted for the alkane content of feed refused) and faecal alkane excretions; b) using treatment mean recovery data for these sheep; or c) using the grand mean values for faecal alkane recovery in these sheep. The direct measurement of faecal output and intake, plus their estimation in several different ways, permitted the calculation of the OM digestibility of diets 1–4 using three different approaches: 1. Measured intakes were combined with measured faecal outputs or with faecal outputs estimated using the C36 alkane from the CRD as a faecal output marker. Since the faecal recovery of C36 is incomplete, faecal C36 concentrations were corrected before making the latter calculations using either the grand mean C36 recovery (0.967) or an assumed recovery of 0.95 (Dove and Mayes 1996). For a given diet, this approach provides three estimates of digestibility. 2. The above estimates of faecal output were combined with intakes estimated using the C32 : C31, C32 : C33 or C36 : C33 alkane pairs, in which the dietary concentrations of the odd-chain alkanes were derived from known diet compositions, or from diet compositions based on individual, treatment mean or grand mean estimates of alkane recovery. Similarly, digestibilities were estimated using intakes based on the supplement method derived from individual, treatment mean or grand mean estimates of alkane recovery. For a given diet, this resulted in 45 separate estimates of OM digestibility. 3. As Dove and Moore (1995) have discussed, the estimate of diet composition provided by EatWhat is in fact a normalised estimate of the amounts of each dietary component which, taken together, would result in a hypothetical 1 kg of faeces with the observed alkane concentrations. It follows that for a diet consisting of components (a, b, c, . . . , n), an estimate of digestibility is ((a + b + c + · · · + n) – 1)/(a + b + c + · · · + n). This approach permits a further three estimates of digestibility derived from diet compositions based on individual, treatment mean or grand mean estimates of alkane recovery. Statistical procedures The trial was a randomised complete block design, except that sheep from two blocks (chosen at random) allocated to the 1-forage treatment (diet 1) were re-assigned to the 4-forage treatment (diet 4). This was done in anticipation of higher standard error of means surrounding the more complex dietary treatment. The effect of treatment on estimates of intake was tested for significance using a general linear model in GENSTAT (2005). Precision of intake estimates was determined with the mean prediction error (MPE) or discrepancy (Dillon 1993). This was done by summing the squares of differences between


observed and estimated values within a treatment and dividing by replicates in the treatment. √ (3) MPE = 1/n × (estimated − measured)2 The various estimates of digestibility were compared with measured digestibility using t-tests for paired comparisons rather than analysis of variance because several of the different estimates had common elements and thus were not statistically independent from one another. Results Alkane profiles of the forages and CSM The alkane concentrations and proportions of the four forage species are in Table 2. Subterranean clover was characterised by a high concentration of C29 alkane, with other alkanes being present at below 100 mg/kg OM. The alkane profile of ryegrass differed from that of subterranean clover by having a high concentration of not only C29, but also C31 alkane. Wheat straw had a similar alkane profile to ryegrass although concentrations of all alkanes were less than half those in ryegrass. Phalaris had very low concentrations of all alkanes (250 mg/kg OM) but similar concentrations of C27, C28, C29 and C31. Proportionally, C29 accounted for 67% of alkanes in subterranean clover and C31 accounted for almost 50% of alkane (by weight) in annual ryegrass and wheat straw. In phalaris, no single alkane comprised more than 30% of the total. Using these alkane profiles, preliminary principal component analyses were conducted to assess, a priori, the extent to which the profiles could be discriminated. These showed that when all alkanes were included in the analyses, the first three principal component axes explained 99.98% of the variance in alkane profiles between dietary components. When the synthetic C28 alkane was excluded from the analyses, the result was almost identical (99.97%), indicating that the added C28 was contributing little to the discrimination between dietary components and could probably have been excluded from the label mixture. Nevertheless, it has been retained in the estimates of diet composition discussed below. Recovery of alkanes in faeces There was a significant effect of treatment on alkane recovery in faeces, with the curve for the 4-forage diet exhibiting a higher asymptote than the recovery curves for the other diets (Fig. 1). Fitting one curve to diets 1 to 3 and a separate one to diet 4 significantly reduced (P < 0.005) the overall error mean square compared with fitting a single curve for all four diets. Significant treatment effects on individual alkane recoveries were limited to C26 (P < 0.05). The relationship between alkane concentration in total faeces and a single, daily faecal grab sample was linear and was described by an equation which did not differ from the line of equality (Fig. 2), implying that the faecal alkane concentration in a single, morning grab sample was representative of the total faecal output.


100 90


Table 3. Observed means and differences and mean predictive error (MPE) between observed and estimated organic matter (OM) intake (g OM/day) using the CRD method in 24 sheep, based on faecal grab samples

Diet 4: % recovery = 103.5 – (12286 x (0.8087X)) R 2 = 0.904

80

% Recovery

70 60

Observed mean

50

Diets 1 to 3: % recovery = 84.3 – (95140 x (0.7373X)) R 2 = 0.971

40 30

10 0 C26

C27

C28

C29

C30

C31

C32

C33

C36

Fig. 1. Relationships between faecal recovery and alkane chain length in sheep fed diet 1 (䉬), diet 2 (䊏), diet 3 (N ) and diet 4 (×). Equations represent the best fit for diets 1 to 3 combined and diet 4.

Diet 1

Diet 2

Diet 3

Diet 4

674

675

672

s.e.m.A

675

4.48

3.17

Mean Difference MPE

Estimate based on C32 : C31 pairB 724 768 786 730 50.6 92.8 114 54.5 45.3 68.1 54.6 35.2

70.4 71.4

49.7 50.5

Mean Difference MPE

Estimate based on C32 : C33 pairB 630 707 710 731 −44.1 32.1 37.9 55.9 61.8 44.1 36.2 40.3

65.8 66.9

46.5 47.3

Mean Difference MPE

Estimate based on C36 : C33 pairB 627 652 752 731 −47.0 −22.6 80.0 55.7 57.7 29.7 52.1 42.0

64.6 64.4

45.6 45.6

20

C25

1219

Alkane concentration in faecal grab samples (mg/kg OM)

A Larger

1400

y=x Filled and open diamonds: y = 0.97x + 0.195 R 2 = 0.966

1200 1000

estimated and observed intake, and the MPE were smaller than with C32 : C31. Estimates based on the C36 : C33 pair were similar to those based on C32 : C33, with differences between observed and estimated values ranging between –47 and 80 g OM/day.

Filled diamonds: y = 0.95x + 3.22 R 2 = 0.952

800 600

standard error of the mean (s.e.m.) should be used in contrasts including diet 1 (n = 2); smaller s.e.m. in contrasts not including diet 1 (n > 2). B Based on a mean 21.5-day alkane payout period.

400 200

Estimates based on labelled supplement method

0 0

200

400

600

800

1000

1200

1400

Alkane concentration in total faeces (mg/kg OM) Fig. 2. Concentrations of C25–C33 and C36 alkanes in total faeces and faecal grab samples of sheep. Equations represent relationships with and without the inclusion of C29 alkane concentrations for sheep consuming diet 1.

Estimates based on CRD method Release rate of alkanes from the CRD: The release rate of alkanes from the CRDs was determined in eight sheep by measuring daily faecal alkane concentration and determining the extinction day for the individual CRDs by plotting faecal concentrations of the dosed alkanes v. time. The average duration of release was 21.5 days (± s.d. 1.07) which was 1.5 days or 7.5% longer than the 20 days suggested by the manufacturer, resulting in a correspondingly lower estimated mean release rate of 46.5 mg/day for each alkane. Since individual daily samples were not analysed separately, we were unable to account for possible daily variation in release. Estimated intakes: The CRD method, when based on the C32 : C31 pair, appeared to consistently overestimate intake of all four diets by ∼50 to 100 g OM/day, although this difference was not statistically significant (Table 3). Intakes involving C33 alkane did not differ significantly from measured intakes. When the C32 : C33 pair was used, the differences between

Estimation of diet composition: The labelled supplement technique commenced with the estimation of the proportions of CSM in the diet. These were then used to estimate the intakes of individual forage components and thus total intake (Eqn 2). Dietary proportions of supplement were well estimated when the faecal alkane recoveries used in the least-squares procedure were those for the 12 individual sheep on total faecal collection (s.e.m. 0.82–1.16; Table 4). However, when the treatment mean for faecal recovery was used in these same sheep, the s.e.m. increased markedly (19.7–27.8) as did the MPE. Despite this, the arithmetic differences between observed and estimated proportions were not large. When the treatment mean recovery was used to estimate CSM proportion in faeces in 24 sheep, the CSM proportion in diet 2 was significantly higher than for other diets (P < 0.05). When the grand mean of faecal alkane recovery for all treatments was used, CSM proportions were similar to observed values when data were restricted to the 12 sheep on total faecal collection; s.e.m. and MPE values were similar to those found using the treatment mean recoveries. However, when all 24 sheep were considered, once again there was a significant treatment effect on CSM proportions. The value for diet 2 was 25% higher than for other diets. The proportions of forages in the multiple-forage diets (Fig. 3) were satisfactorily estimated when the faecal recovery of alkanes from individual animals were used. For diets 1, 3 and 4, reliance upon the treatment or grand mean faecal recoveries

1220



Table 4. Observed and estimated proportions, differences and mean predictive error (MPE) of cottonseed meal proportion (g organic matter/kg) in the diet of sheep. Estimates based on using eight alkanes in a least-squares analysis in 12 or 24 sheep Within rows, values followed by different letters are significantly different at P < 0.05 Diet 2

Diet 3

Diet 4

197 197

196 196

197 196

195 197

1.72 0.578

1.44 0.409

193 −2.91 3.69

1.16 1.49

0.818 1.05

Observed proportion n = 12 n = 24

Estimated proportion and MPE in 12 sheep 196 195 195 −0.57 −0.532 −1.93 0.60 0.91 2.86

Individual sheep, proportion (n = 12) Difference MPE Treatment mean, proportion (n = 12) Difference MPE

196 −0.57 5.20

219 22.8 61.1

198 0.74 9.51

197 1.84 24.7

27.8 27.8

19.7 19.7

199 2.43 5.71

188 −7.53 44.9

195 −7.59 12.1

193 16.8 30.8

23.5 23.5

16.6 16.6

211a 14.8a 13.90

23.8 23.6

16.8 16.7

199a 2.8b 30.84

27.0 26.8

19.1 19.0

Grand mean, proportion (n = 12) Difference MPE

Estimated proportion and MPE in 24 sheep 203a 288b 210a 6.0a 92.6b 13.8a 7.63 106 36.09

Treatment mean, proportion (n = 24) Difference MPE Grand mean, proportion (n = 24) Difference MPE

s.e.m.A

Diet 1

208ab 11.0ab 5.71

253b 57.0a 44.91

191a −5.6b 12.12

standard error of the mean (s.e.m.) should be used in contrasts including diet 1 (n = 2); smaller s.e.m. in contrasts not including diet 1 (n > 2).

A Larger

1.0

0.8

Straw

0.6

Ryegrass Phalaris Sub. clover

0.4

CSM

0.2

Diet 1

Diet 3

t nd ra

G

v

ea

bs

di

Tr

In

O

t

nd ra

G

v di

ea Tr

bs

Diet 2

In

O

t nd ra

G

v di

bs

ea Tr

In

O

t

nd ra

G

v

ea Tr

di

O

In

bs

0.0

Diet 4

Fig. 3. Observed and estimated proportions of the forages and cottonseed meal in four diets. Estimates derived using concentrations of eight alkanes in a least-squares analysis, with faecal alkane recovery corrections based on the individual sheep (‘Indiv’, n = 12), the treatment mean (‘Treat’, n = 24) or the grand mean (‘Grand’, n = 24).

resulted in some loss of accuracy and precision. However, for diet 2, estimates were not well estimated when treatment or grand mean faecal recoveries were used. In both cases, the proportion of phalaris was underestimated, being only 39 and 64% of observed proportions, respectively. The proportions of

the other dietary component, subterranean clover, were thus correspondingly over-estimated. Estimated intakes: Three estimates of intake were compared with the observed intake and differences or MPEs between the observed and estimated values were also compared (Table 5).



Table 5. Observed means and differences and mean predictive error (MPE) between observed and estimated organic matter intake using the labelled supplement method in 12 sheep, based on faecal alkane recoveries of the individual sheep, the treatment mean or the grand mean Diet 1 Observed mean

672

Diet 2 675

Diet 3 668

Diet 4 679

s.e.m.A 2.77

1.96

Table 6. Observed means and differences and mean predictive error (MPE) between observed and estimated organic matter intake using the labelled supplement method in 24 sheep, based on faecal alkane recoveries for the treatment mean or the grand mean Within rows, values followed by different letters are significantly different at P < 0.05

Observed mean

Estimate based on individual animal recovery Mean 674 676 675 688 6.28 Difference 1.97 1.84 6.70 10.33 5.25 MPE 1.45 1.82 5.74 6.55

4.44 3.71

Mean Difference MPE

Estimate based on treatment mean recovery 675 658 667 681 99.7 2.42 –16.0 –1.13 3.24 98.9 12.7 120 19.0 37.6

70.5 69.9

Mean Difference MPE

Mean Difference MPE Mean Difference MPE

Estimate based on grand mean recovery 664 752 697 633 –7.77 77.1 28.3 –45.3 13.4 130.1 26.2 41.4

100.0 99.4

70.8 70.3

A Larger


These were based on diet composition data obtained by correcting the alkane faecal concentrations with three different sets of recovery; recovery in individual sheep, the mean recovery for each treatment or the grand mean recovery for all four treatments. Estimates of intake were best when the individual animal recoveries were used, as evidenced by small differences between estimated and observed values (2.0–10.3 g OM/day) and low MPE (1.5–6.6 g OM/day). When the treatment mean or particularly the grand mean for faecal alkane recovery were used, estimates were less precise. Differences, although not significant, were apparently much larger when the grand mean was used as opposed to the treatment mean. However, MPEs were similar when either the treatment or grand mean were used. Treatment or grand mean recoveries determined from total faecal collections in sheep in metabolism cages were used to correct alkane concentrations in grab samples taken from all 24 sheep (Table 6). Differences between observed and estimated intakes were only significant for the 2-forage diet, with mean intakes for the 1-, 3- and 4-forage diets being less than 30 g different from the observed values. However, the MPE was small when the single forage diet was used, highest when phalaris was introduced and intermediate for the 3- and 4-forage diets (Table 6). Digestibility There was a clear diet effect on OM digestibility (P < 0.001) which was highest for diet 1, intermediate for diets 2 and 3 and lowest for diet 4 (Table 7), reflecting the known DMD of the diet components. Estimates of faecal output based on the use of C36 as a faecal output marker were not significantly different from known faecal output (P > 0.05; data not shown) and as a result, diet digestibility based on measured intakes and

1221

Diet 1

Diet 2

Diet 3

Diet 4

674

675

672

675

s.e.m.A 4.48

3.17

Estimate based on treatment mean recovery 654b 475a 642b 659b 78.1 –19.5b –199a –29.5b –16.5b 77.7 12.3 90.3 40.9 53.2

55.2 55.0

Estimate based on grand mean recovery 638 543 745 739 –35.3 –131 72.7 64.3 18.8 70.2 93.3 88.8

114 102

81.1 72.4

A Larger


C36-based faecal outputs did not differ significantly (P > 0.05) from those based on measured intakes and faecal outputs (Table 7). Similarly, the various estimates of digestibility of individual diets did not differ significantly from measured digestibility, with one exception (see below). Given the large number of possible estimates (n = 48), for the sake of clarity, only the data for mean digestibility across diets are shown in the rest of Table 7. In addition, where intake was estimated using the CRD method, only data derived from C32 : C33 are shown, since results from C32 : C31 and C36 : C33 were very similar. As can be seen, regardless of the method of calculation, the only estimates which differed significantly from measured digestibilities were those based on measured faecal output and intake from the supplement method when based on individual alkane recoveries (P < 0.01). Even in this case, differences from measured digestibilities were very small (diet 1 = 0; diet 2 = 0.003; diet 3 = 0.003; diet 4 = 0.011) and could not be regarded as biologically significant. Discussion Our results extend earlier work (Dove et al. 2002b; Elwert and Dove 2005) by demonstrating that the labelled supplement approach can be used to estimate the intake of at least four roughage components in the diet and is an alternative approach to the estimation of intake if dosing with the alkane CRD presents problems for animal management. In evaluating the two approaches, two points should be borne in mind. 1. In using the alkane CRD to estimate intake, no corrections for faecal alkane recovery are required, since adjacent alkanes usually have similar faecal recoveries. However, the animals must separately be dosed with even-chain alkanes and for comparisons with known intakes, an accurate estimate of alkane dose rate is needed. Moreover, by itself, the CRD method does not provide an estimate of diet composition.

1222


Table 7.


Estimates of digestibility of diets 1–4 and their comparison with directly measured digestibility NA, not applicable; *, estimate differs from measured digestibility (P < 0.01)

Basis of intake estimate

Herbage alkane concentrations based on:

Diet

Measured

NA

1 2 3 4

0.758 0.642 0.654 0.576

0.720 0.610 0.597 0.595

0.725 0.617 0.604 0.602

Mean

0.642

0.620

0.627

Alkanes in diet as Diet composition using individual recoveries Diet composition using treatment mean recoveries Diet composition using grand mean recoveries

0.643B 0.644 0.636 0.634

0.621 0.621 0.613 0.614

0.628 0.628 0.620 0.620

Supplement method

Diet composition using individual recoveries Diet composition using treatment mean recoveries Diet composition using grand mean recoveries

0.648* 0.635 0.641

0.625 0.610 0.618

0.632 0.617 0.625

Digestibility from EatWhat (Dove and Moore 1995)

Diet composition using individual recoveries Diet composition using treatment mean recoveries Diet composition using grand mean recoveries

0.643 0.636 0.642

C32 : C33

A Source B Values

fedA

Basis of faecal out put (FO) estimate Measured C36, measured C36, assumed FO recovery (0.967) recovery (0.95)

of dietary C33 concentration used to estimate intake by controlled-release device (CRD) method. are means across diets for estimates.

2. By contrast, the labelled supplement approach provides an estimate of diet composition and thus the intake of all dietary components, and the supplement fed as part of normal management functions as the ‘alkane dose’. However, the accuracy of the approach is very much a function of the faecal alkane recovery estimates used as part of the diet composition estimates. Alkane profiles of the forages and CSM Relative proportions of different alkanes within a species were similar to those reported in the literature, but quantitatively, there was wide variation in individual alkane concentration compared to other reports (Dove and Mayes 1996; Chen et al. 1999; Lewis et al. 2003) Wide variation in individual alkane concentration within a species is typical and can be attributed to climatic and agronomic conditions as well as stage of growth (Dove and Mayes 1996; Boadi et al. 2002) and leaf to stem proportions (Smith et al. 2001). The low alkane concentration of phalaris is typical of this species (Dove 1992) and the inclusion of this species was expected to make discrimination among forages based on faecal alkane concentration more challenging. Recovery of alkanes in faeces Our estimates of the recovery of alkanes in total faeces are typical of the published data with sheep (Mayes et al. 1986; Dove and Mayes 1996). Occasional differences in faecal recovery associated with a given forage species have also been reported (e.g. Ferreira et al. 2005; Elwert et al. 2008; Lin et al. 2007) but are neither consistent nor easy to explain. In the present study, the higher recoveries observed with diet 4 cannot be associated with the CSM content of the diet, as observed by Elwert and Dove (2005), because the CSM proportion was the same across

diets. The higher asymptote for recovery (103%) for diet 4 compared to the other diets (84%) may be related to the lower digestibility of the diet containing the wheat straw, as Ferreira et al. (2005) demonstrated a negative relationship between recovery and digestibility. Using their relationship between C32 recovery and digestibility and the observed digestibilities in the current trial, estimated C32 recoveries were 93% for diet 4 and 79% for the other diets combined. These values were close to the C32 recoveries of 90% and 79%, respectively, which can be computed from the relationships in Fig. 1. Significant treatment effects on individual alkane recoveries were limited to C26, but estimates of the recovery of C26 may have been compromised by the very low concentrations involved (