Abstract:
Proc. Aust. Soc. Anim. Prod. Vol. 19 THE RELATIONSHIP OF SOME NOVEL MEASURED CHARACTERS WITH VISUAL SCORES MADE BY SHEEP CLASSERS G. M. HEAD and L. G. BUTLER Tasmanian Dept of Primary Industry, PO Box 180, Kings Meadows, Tas. 7249. SUMMARY Three experienced sheep classers visually graded 47 sheep for staple ` length' fleece ` , density' ` ' , skin and overall ` productivity' These grades were converted to numeric scores and compared by regression . with a range of fleece and histological measurements and with a WOOLPLAN index. In general, the classers' visual grades accounted for less than 50% of the variation in the measured parameters, suggesting that their grading for any one character actually took account of a range of measured characters. Keywords: sheep classer grades, measured parameters, relationship, visual scores. INTRODUCTION Although Kinghorn (1984) makes the point that it is difficult to make a direct comparison between visual classing and measurement, a number of reports have examined the ability of classers either to rank animals in order of measured merit, or in terms of their efficiency of selection compared with using measured parameters. Napier and Jones (1979) concluded that, although in the 1950s most assessments showed classing to be 30-40% efficient in selection for clean fleece weight (CFW), efficiency could be as high as 85%. McGuirk and Rose (1979) suggested that the average classer achieved approximately 60% of the possible selection differential for fleece weight. The efficiency of stud sire selection in 2 studs (McGuirk et al. 1982) was in excess of 70%. Recently Ponzoni et aZ. (1991) reported remarkably good agreement between visual grading into 4 grades by classers and the WOOLPLAN rankings. In their work 42% of all rams were graded the same and 93% were graded within 1 grade difference. However, production by sheep, and in particular fleece weight, is a complex trait. Classers generally acknowledge that their assessment of individuals for a particular ` character' usually involves a complex of component characters, and are concerned with characters in addition to those directly related to production. In particular, sheep classers have, over the last decade, become interested in assessing the characteristic of ` productive skin' (Coy 1982). This paper reports the degree of agreement of classer visual scores for 4 new characteristics with a selection of measured parameters which are different from those that have been previously reported. MATERIALS AND METHODS A total of 47 wethers of 4 genotypes (Merino, Polwarth and their reciprocal crosses) were housed indoors for 12 months and had 16 months wool growth. They had been fed at 1 of 2 feed levels such that the high group were 13% heavier at assessment. Fleeces were weighed at shearing and efficiency of wool production calculated as the greasy fleece weight per kg of feed consumed. Other parameters (Table 1) were measured by standard methods. Three experienced Tasmanian sheep classers (2 Stock Agent Wool Managers and a private consultant) were each asked to grade every animal for staple ` length' fleece ` , density' (2 classers only), ` ' and overall ` skin productivity' Each animal was presented . to the classer in turn and all 4 characters assessed at the 1 occasion. The classer ` productivity' score is essentially an assessment of the animal' expected profitability s and assumed to be approximately equivalent to a WOOLPLAN index (option 1) in which all traits are free to move in the direction that will increase profit. ` Skin' is assessed by a variety of criteria including apparent thickness, colour, looseness and lifting characteristics when the wool is parted. No measurement of skin thickness, the only obvious equivalent measurement, was available. Staple ` length' and fleece ` density' scores were assumed to be equivalent to measured staple length and follicle density respectively. The data were analysed by simple regression analysis. The classer grades were converted to numerical scores (range l-9) and regressed against various measured parameters (Table 1). In addition the classers' scores for each sheep were averaged and regressed against the same measured parameters. To provide a basis on which to compare the likely maximum degree of association that could be achieved for any regression, the measured value assumed equivalent to the classer graded character was also regressed against the other measured values. 153 Proc. Aust. Sot. Anim. Prod. Vol. 19 RESULTS Table 1 gives the correlation coeffkients (r values) of classer scores, the average of the classer scores, and the measured value equivalent to the character graded, when regressed against the measured parameters. Only significant values are presented. The significant r values range from 0.19 to 0.68 for the regressions of classer scores and from 0.24 to 0.64 for the regressions of the measured values. Table 1. The correlation coeffkients for classer scores regressed against actual measurement 154 Proc. Aust. Sot. Anim. Prod. Vol. 19 DISCUSSION The sheep were presented in a manner in which the sheep classers were unaccustomed; they had 16 months of fleece growth and had been shedded in individual pens for the previous 12 months on a ration which allowed only a small liveweight gain. Further the characters assessed by the classers do not necessarily relate directly to the measurements assumed to be equivalent. For example the idea of a ` productive skin' although it may have developed from the concept of a thick skin, may only partially , be an assessment of skin thickness. It is well established that the methods of measurement of wool characteristics are accurate within narrow limits. However there may be problems in accurately locating the midside site when sampling in the field and although the midside has been shown to be the site most representative of the fleece, there may be considerable individual animal differences in the relation between midside and average fleece data. Therefore the measured values are not necessarily completely reliable standards by which to judge the accuracy of assessment. The range of r values (Table 1) for the individual classers are similar to those of Mullaney and Sanderson (1970) who reported simple correlation coefficients ranging from 0.28 to 0.5 1. Generally the r values of the average classer score were slightly higher than the individual classer score. However there was some suggestion that individual classers were stronger in different areas, which is not surprising. The regressions showed that measured staple length in this group of animals was not highly related to any of the other measured characters with the exception of fibre diameter. There is reasonable agreement between the classer ` length' scores and measured staple length but poorer agreement with other parameters. Staple length score accounted for little variation in efficiency, follicle density, S/P ratio, liveweight or WOOLPLAN index. Similarly follicle density in these sheep was poorly related to the other measured characters except fibre diameter. However the 2 classer scores for ` density' were better related to greasy fleece weight, suggesting that their perception of fleece weight appearance may have influenced their density score. The high correlation of classers score for ` ' with fleece weight suggests that their assessment skin may have been strongly influenced by their perception of greasy fleece weight. Liveweight does not appear to be important in the classers' assessment of the characters in this study, despite the general industry preference for large sheep. Similarly efficiency of wool growth appeared to receive some consideration by the classers (they tended to downgrade the more efficient animals) but was not related to the measured parameters. The classers assessment of ` productivity' was more related to fleece weight than to WOOLPLAN index and less related to fibre diameter and efficiency. It is interesting that the classer scores were better related to measured fleece weight than were the measured characters equivalent to those scored by the classer. This confirms that the classers were incorporating additional characters into their score which is not a simple assessment of one measured character. These data therefore suggest that visual grades of the extended range of characters examined in this work usually appear to incorporate a number of characters and are not a simple trait relating to a single measurable character as the term applied to the character may imply. ACKNOWLEDGMENTS We are grateful to Bob Barrenger, Terry Heggaton and Rob Wallace for their time and expertise in grading the animals. The production measurements were made possible by a grant from the Australian Wool Corporation. REFERENCES COY, J. (1982). Aust. Assoc. Anim. Breed. Genet. 3: 218-21. KINGHORN, B. I? (1984). Aust. Assoc. Anim Breed. Genet. 4: 32-4. McGUIRK, B. J. and ROSE, M. (1979). Aust. Assoc. Anim. Breed. Genet. 1: 392-93. McGUIRK, B. J., ROSE, M. and SCOTT, R. (1982). Aust. J. Exp. Agric. Anim. Husb. 22: 274-80. MULLANEY, P. D. and SANDERSON, I. D. (1970). Aust. Assoc. Anim. Prod. 8: 159-65. NAPIER, K. M. and JONES, L. P. (1979). Aust. Assoc. Anim. Breed. Genet. 1: 391-92. PONZONI, R. W., ANCEL, P .M. C. and GIFFORD, D. R. (1991). Aust. Assoc. Anim. Breed. Genet. 9: 368-71. 155