Bioprediction of Body Weight from Zoometrical Traits of Non-descript Goats using Linear and Non-Linear Models in North Central Nigeria

A Yakubu, A O Ladokun* and M M Adua

Animal Science Department, Faculty of Agriculture, Nasarawa State University,
Keffi, Shabu-Lafia Campus, P.M.B. 135, Lafia, Nigeria
abdul_mojeedy@yahoo.com
* Department of Animal Physiology, University of Agriculture, Abeokuta, Nigeria

Abstract

Data on body weight and four zoometrical measurements – rump height, withers height, heart girth and body length of 135 Non-descript goats were used to develop linear, quadratic and allometric equations to predict body weight. The animals were extensively managed in north central Nigeria. They were stratified into two groups on the basis of dentition. Goats that were between 1.0–14.3 months old (milk teeth age) were classified as kids while those between 19.4–30.6 months old (animals having between 2–8 permanent incisors) were categorized as adults. Animals were also classified on the basis of incidence of wattle into wattled and non–wattled goats.  

Highly significant (P<0.01) coefficients of determination (R2 = 0.38–0.95) were recorded, showing the inter-relationship between body weight and the linear body measurements investigated. However, body weight was better estimated in kids (pooled data for males and females) than adult (does) goats (R2 = 0.84–0.94; MSE = 0.01–1.55 versus R2 = 0.41–0.92; MSE =0.005–16.80). Withers height appeared as the highest single prediction variable in kids (R2 = 0.91–0.94; MSE = 0.01–0.88), while in adult goats, heart girth was found to account more for variation in the body weight (R2 = 0.91–0.92; MSE = 0.005–2.54). Among the prediction equations tested, the allometric model seemed to give a better fit, closely followed by the quadratic and linear models respectively. The wattle gene impacted mostly on adult parameters where prediction accuracy appeared to be better in wattled adults compared to their non-wattled counterparts. The present findings could be exploited in designing appropriate management and selection programmes.

Keywords: body measurements, body weight, goats, prediction models, wattle

Introduction

Goats are hollow-horned small ruminants, together with sheep are sometimes referred to as poor man’s cow. They are so referred because of their ability to provide adequate meat, milk and skin for the small-holder or subsistence farmers’ own use as well as a little extra for sale. In Nigeria, the population of goats is put at 34.5 million (RIM 1992). Out of these, traditionally reared stock is 99.97%, while 0.03% of the stock is commercially managed. Traditionally, animals are visually assessed, which is a subjective method of judgement (Abanikannda and Leigh 2002). Therefore, the development of objective means (linear body measurements) for describing and evaluating body size and conformation characteristics (Yakubu and Ibrahim 2011; Jimcy et al 2011) would overcome many of the problems associated with visual evaluation.

Body weight measurement is used the most to evaluate body development in animals (De Brito Ferreira et al 2000); but it is not easily measured in the field. This is due to the time and energy expended while determining it. Regression equations have been established to estimate body weight from body dimensions (Singh and Mishra 2004; Gorgulu et al 2005; Oke and Ogbonnaya 2011). These regression models allow a fast evaluation of the body weight of an animal; and are also used for the optimization of feeding, determination of optimum slaughtering age, and selection criteria. However, such measurements are not across breeds and climatic zones on the choice of model that gives the best fit (Islam et al 1991; Benyi 1997).

Non-descript goats constitute a large population of indigenous goats in Nasarawa State, north central Nigeria. However, studies on the relationship between body weight and biometric measurements of Non-descript goats in the State are virtually non-existent. Therefore, this investigation set out to obtain prediction equations for estimating body weight from linear body measurements of wattled and non-wattled goats in the field in the southern guinea savanna zone.

Materials and methods

The study was conducted in Nasarawa State which falls within the southern guinea savanna agro-ecological zone of north central Nigeria. It is found between latitude 7^o 52’N and 8^o 56’N and longitudes 7^o 25’E and 9^o 37’E respectively. The State is characterized by tropical sub-humid climate with two distinct seasons. The wet season starts from about the beginning of May and ends in October. The dry season is experienced between November and April. Annual rainfall figures range from 1100mm to about 2000mm, respectively. Temperatures are generally high during the day, particularly between the months of March and April. The mean monthly temperatures in the State range between 20^oC and 34^oC.

135 Non-descript goats comprising 58 kids (23 males and 35 females) and 77 adults (females only) were randomly sampled. The animals were extensively managed where they grazed during the day on natural pasture containing forages such as northern gamba grass (Andropogon gayanus), stylo (Stylosanthes gracilis) and goat weed (Ageratum conyzoides). Phenotypically, the goats conform to the classification descriptors of the West African Dwarf goats (Yakubu et al 2010). The evolutionary history of the sampled goats’ genome indicated that they might have migrated from the southern parts of the country where WAD goats are largely concentrated. Certainly, this could have been facilitated by the migratory system of grazing; the nomadic life of herdsmen especially the Fulanis of northern Nigeria; and the transportation of livestock to new lands by traders.  However, in as much as information was not available on selective breeding of the goats and they did not emanate from a closed population, which is a case of heterogeneity; they were classified as Non-descript animals. The goats were stratified into two groups on the basis of permanent incisors eruption as described by Matika et al (1992). Goats between 1.0–14.3 months old (milk teeth age; no permanent incisors) were classified as kids while those that were between 19.4–30.6 months old (having 2–8 permanent incisors) were categorized as adults. Animals were also classified on the basis of the incidence of wattle into two categories; the wattled goats (W_a^w) and the non-wattled goats (W_a⁺). Animals with either unilateral (one) or bilateral (two) wattles were treated the same since single-wattled individuals are allelically double-wattled (COGNOSAG 1987). Efforts were also made to ensure that only animals that appeared physically sound with no signs of illness were measured. Pregnant animals were not measured since the effect of pregnancy has been implicated in some morphometric parameters especially those of the thoracic and rump regions.

Figure 1: Map of Nigeria showing Nasarawa State (Sampling Location)

Body weight (BW) and four exterior traits were taken on each animal in the morning before they were released for grazing. The measured parts were withers height (WH), measured as the distance between the most dorsal point of the withers and the ground; rump height (RH), distance from Spina illiaca to the ground; body length (BL), measured from distance between the occipital protuberance and tail-drop; and heart girth (HG), circumference of the chest just behind the forelimbs. Body weight was measured in (kg) using a hanging scale. The height measurements (cm) were done using a graduated measuring stick while the length and circumference measurements (cm) were effected using a measuring tape. In order to avoid between-individual variations, all the measurements were carried out by the same person.

Each body parameter was regressed against body weight separately for wattled (n = 21) and non-wattled  (n = 37) kids, and wattled (n = 32) and non-wattled (n = 45) adults, respectively. Similarly, regression equations were obtained for kids (n = 58) and does (n = 77), respectively. The simple linear and non-linear programmes of GENSTAT (1995) and SPSS (2001) statistical packages were employed. The models used were:

            Linear: Y = a + bX

            Quadratic:       Y₁ = a + bX + cX²

                 Allometric:     Y₂ = aX^b

            where,

Y,Y1,Y2 =      dependent variables (body weights).

a =  intercept (the value of the dependent variable when the independent variable is zero).

b = regression coefficient associated with independent variable.

c = coefficient.

X = independent variables (rump height, withers height, heart girth, body length).

Results

Equations predicting body weight from rump height, withers height, heart girth and body length in wattled and non-wattled kids are presented in Table 1. Highly significant (P<0.01) coefficient of determination (R²) values (0.76–0.95) and low residual mean square (MSE) (0.01–2.85) were recorded using linear, quadratic and allometric equations. However, the allometric model appeared to give slightly higher regression estimates. This was closely followed by the quadratic and linear models respectively. Prediction of body weight from withers height seemed to be higher in both wattled and non-wattled kids compared to other barymetric measurements. The reverse, however, was the case in heart girth, whose estimation was higher in non-wattled than wattled kids (R² = 0.90–0.93; MSE = 0.01–0.93 versus R²= 0.76–0.77; MSE = 0.04–2.85) in all the three animal models investigated.

In wattled and non-wattled adults (Table 2), regression models were also found to be highly significant (P<0.01). However, higher coefficient of determination values were recorded in wattled adults compared to their non-wattled counterparts (R² = 0.57– 0.95 versus 0.38–0.91). The use of the allometric equation also appeared to be more reliable in estimating body weight from rump height, withers height and body length in wattled adults (R² = 0.81, 0.67 and 0.08; MSE = 0.03, 0.05 and 0.03 respectively), and body length in non-wattled adults (R² = 0.70; MSE = 0.01). The variation of body weight due to rump height and withers height in non-wattled adults seemed to be best explained by the quadratic model (R² = 0.43 and 0.55). All the prediction equations gave good estimates for heart girth (R² = 0.90 – 0.95; MSE = 0.004 – 2.76) in both wattled and non-wattled adult goats. In kids (pooled data), the allometric model appeared to produce better goodness of fit with slightly higher coefficients of determination (R² = 0.85 – 0.94) and lower residual mean square (MSE = 0.01 – 0.02). This was chronologically followed by the quadratic (R² = 0.84 – 0.93; MSE = 0.64 – 1.55) and the linear (R² = 0.84 – 0.91; MSE = 0.88 – 1.53) fitted functions respectively (Table 3). Withers height was more reliable in predicting body weight (R² = 0.91 – 0.94; MSE = 0.01 -0.88) than other body parameters estimated. In adults (pooled data), the quadratic models gave higher R² values of 0.55 and 0.45 for rump height and withers height respectively. However, MSE estimates of 13.12 and 15.86 were high compared to 0.03 recorded for the allometric function (R² = 0.48; 0.41). Prediction accuracy was better using heart girth (R² = 0.91 – 0.92; MSE = 0.005 – 0.03).

Discussion 

The moderate and high coefficients of determination obtained in the present study are indications that linear body measurements succeeded in describing more variation in body weight. Similarly, Gorgulu et al (2005) reported highest R² values from equations with heart girth, hip height and withers height among others. Prediction of body weight from linear body measurements seemed to be better in kids than adult goats. The present result showed that as age advanced, coefficients of determination decreased while residual mean square increased. This is in consonance with the findings of Thiruvenkadan (2005) in Kanni Adu kids under farmers’ management in southern part of India. Height at withers and rump height have been reported to be limited in their values as indicators of weight. However, in this study, they have been found to best explain variation in body weight in kids than body length and heart girth. The present findings are consistent with the submission for Aziz and Sharaby (1993) that height at withers could be considered as a better predictor for body weight in lambs. Similarly, Boggs and Merkel (1984) noted that height at withers is the most accurate and repeatable measurement for frame size. This trend was reversed in adult Non-descript goats where the model including heart girth was found to be better in estimating body weight. The superiority of heart girth over other body measurements have been reported by several other workers (Topal et al 2003; Thiruvenkadan 2005; Yakubu 2010a and b). This is not unexpected considering the high environmental sensitivity of heart girth. Salako (2004) reported that heart girth and body weight, grow in response to environmental components such as feed and management; as such, they are used to assess environmental impact on breeds. The higher association of body weight with chest girth could also be attributed to the relatively larger contribution in body weight by chest girth, which consists of bones, muscle and viscera.

Although initial periods of living organisms are mostly linearly modeled (Kocaba et al 1997), more sophisticated methods could be used to model the later stages of growth. This explains the slight edge, the allometric and quadratic models had over the linear equation in the present study. The present findings are also in agreement with the reports of Benyi and Karbo (1998) that the geometric (allometric) equation estimated live weight with a high degree of reliability compared to the linear equation. This is an indication that allometric growth equation offered a quantitative description of meat conformation. This was corroborated by the reports of Thys and Hardouin (1991) and Osinowo et al (1992) that the relationship between body measurements and body weight are curvilinear and well defined by geometric regression equations. In another study, Smail and Da (2006) concluded that the quadratic model gave a better fit to the body parameters investigated by its higher R² and lower MSE (residual mean square) compared to the simple linear model. The impact of wattle gene was mostly exerted in adults, where prediction accuracy seemed to be better in wattled adults compared to their non-wattled counterparts. This seemingly marked difference could be exploited as a selection criterion for the improvement of indigenous herds. This is because the incidence of wattle could be an adaptive feature influencing better performance, since a possible role in thermo-regulation was suggested for wattles in the Nigerian forest goat (Odubote 1994). 

Conclusion 

• Body weight of Non-descript goats in Nasarawa State, north central Nigeria can be predicted from linear measurements such as rump height, withers height, heart girth and body length using different regression models.
  
  
• The allometric model seemed to produce a better goodness of fit, followed by the quadratic and linear models respectively.
  
  
• Body weight of kids was better predicted from withers height while heart girth was found to account more for variation in the body weight of the female adult goats.
  
  
• Effect of the incidence of wattles manifested in adult goats, where body weight appeared to be better estimated in wattled adults compared to their non-wattled counterparts.
  
  
• The present findings could aid management and selection decisions on Non-descript goats in the study area.

Acknowledgement

The authors wish to express their profound gratitude to Dr. Marcos De Donato of the Department of Animal Science, Cornell University, Ithaca, New York, USA for assisting with the editing of the map of Nigeria.

References

Abanikannda O T F, Leigh A O and Olutogun O 2002 Linear  measurements based discriminant classification of Zebu cattle in Lagos State. In: Fanimo,A.O. and J.A. Olanite (eds). Contributory role of animal production in national development. Proceedings of the 7^th Annual Conference  of Animal Science Association of Nigeria, held at the University of Agriculture, Abeokuta, September 16-19, 2002. pp 355-356

Aziz M A and Sharaby M A 1993 Collinearity as a problem in predicting body weight from body dimensions of Nadji sheep in Saudi Arabia. Small Ruminant Research 12: 117 – 124

Benyi K 1997 Estimation of live weight from chest girth in pure and crossbred West African goats. Tropical Animal Health and Production 29: 124-128

Benyi K and Karbo N 1998 Estimation of live weight from chest girth in pure and crossbred West African sheep. Tropical Agriculture (Trinidad) 75: 380-384

Boggs D L and Merkel R A 1984 Live animal carcass evaluation and selection manual. 2^nd Edition. Kendall/Hunt Publishing Comp., Dubuque, I.A. p.16.

COGNOSAG 1987 Committee on Genetic Nomenclature of Sheep and Goats, Workshop.

De Brito Ferreira M P, Ramos De Carvalho F F,  Nogueira Barros N and De Assis Mello A 2000 Relationship between body measurements and live weight in Saanen and Nubian goat breeds. Proceedings of the 7^th  International Conference on Goats, France, 15-21 May, 2000. p. 223.

GENSTAT 1995 General Statistics 5, Release 3.2 Lawes Agricultural Trust (Rothamsted Experimental Station), U.K.

Gorgulu O, Sar A, Keskin M, Bicer O and Gul S 2005 Some prediction equations of live weight from different body measurements in Shami (Damascus) goats. Journal of Animal and Veterinary Advances 4: 532-534

Islam M R, Saadullah M, Howlider A R and Huq M A 1991 Estimation of live weight and dressed carcass weight from different body measurements in goats. Indian Journal of Animal Science 61: 460-461

Jimcy J, Raghavan KC and Sujatha KS 2011 Diversity of local goats in Kerala, India based on morpho-biometric traits. Livestock Research for Rural Development. Volume 23, Article #119. http://www.lrrd.org/lrrd23/5/jimc23119.htm

Kocaba Z,  Kesici T and Elicin A 1997 Growth curve in Akkaraman, Awassi x Akkaraman and Malya x Akkaraman lambs. Turkish Journal of Veterinary and Animal Sciences 27: 267-275

Matika O, Sibanda R and Beffa M L 1992 Eruption of permanent incisors in indigenous goats and sheep. In: Rey, B., S.H.B. Lebbie, and L. Reynolds (eds). Small Ruminant Research and Development in Africa Proceedings of the first Biennial Conference of the African Small Ruminant Research Network, ILRAD, Nairobi, Kenya, 10-14 December, 1990.pp. 499-504.

Odubote I K 1994 Characterization of West African Dwarf goat for certain qualitative traits. Nigerian Journal of Animal Production 22: 37-41.

Oke U K and Ogbonnaya E O 2011 Application of physical body traits in the assessment of breed and performance of WAD sheep in a humid tropical environment. Livestock Research for Rural Development. Volume 23, Article #024. http://www.lrrd.org/lrrd23/2/oke23024.htm

Osinowo O A,  Olorunju, S A S,  Otchere E O and Arigi L A 1992 Relationship between chest girth and live weight in Yankasa sheep and Red Sokoto goats. Journal of Animal Production Research 12: 69-72.

RIM 1992 Nigerian Livestock Resources: National Synthesis. Resource Inventory and Management Ltd., Jersey, UK.

Salako A E 2004 Maturity rate of some morphometric traits in the West African Dwarf sheep of Nigeria. Tropical Journal of Animal Science 7: 51-55

Singh P N and  Mishra A K 2004 Prediction of body weight using conformation traits in Barbari goats. Indian Journal of Small Ruminants 10: 173.

Smail K and Da B 2006 Comparison of the linear and quadratic models for describing the growth of live weight and body measurements in Anatolian Merino male lambs in fattening period. Journal of Animal and Veterinary Advances 5: 81-84.

SPSS 2001 Statistical package for the Social Sciences. SPSS Inc., New York.

Thiruvenkadan A K 2005 Determination of best-fitted regression model for estimation of body weight in Kanni Adu kids under farmer’s management system. Livestock Research for Rural Development. Volume 17, Article #085. http://www.lrrd.org/lrrd17/7/thir17085.htm

Thys E and Hardouin J 1991 Prediction of sheep body weight in markets in the far north Cameroon. Livestock Research for Rural Development. Volume 3 (1). http://www.lrrd.org/lrrd3/1/hardouin.htm

Topal M, Yildiz N, Esenbuga N, Aksakal V, Macit M and Ozdemir M 2003 Determination of best fitted regression model for estimation of body weight in Awassi sheep. Journal of Applied Animal Research 23: 201-208. 

Yakubu A 2010a Path coefficient and path analysis of body weight and biometric traits in Yankasa lambs. Slovak Journal of Animal Science 43: 17-25

Yakubu A 2010b Fixing multicollinearity instability in the prediction of body weight from morphometric traits of White Fulani cows. Journal of Central European Agriculture 11: 487-492

Yakubu A, Salako AE and Imumorin IG 2010 Multivariate analysis of spatial patterns of morphological traits in West African Dwarf goats in  three agro-ecological zones of Nigeria. Journal of Applied Animal Research 38: 257-260

Yakubu A and Ibrahim I A 2011 Multivariate analysis of morphostructural characteristics in Nigerian indigenous sheep. Italian Journal of Animal Science 10 (2): 83-86

Received 25 April 2011; Accepted 27 May 2011; Published 19 June 2011

Go to top