Livestock Research for Rural Development 30 (1) 2018 Guide for preparation of papers LRRD Newsletter

Citation of this paper

Using heart girth to estimate live weight of heifers (Bos indicus) in pastoral rangelands of northern Kenya

W O Odadi

Department of Natural Resources, Egerton University, P O Box 536-20115, Egerton, Kenya
woodadi@yahoo.com

Abstract

Accurate estimation of livestock live weight is necessary for purposes of making grazing management and animal husbandry decisions. Estimating livestock live weights using morphological measurements such as heart girth can be useful, especially in situations where weighing scales are of limited applicability like in many pastoral rangelands in Africa. Developing accurate live weight prediction models for livestock in such rangelands is thus paramount. This study evaluated the use of heart girth measurements to predict live weights of Zebu heifers in pastoral rangelands of northern Kenya. The study used heifers aged 1-3 years and spread across wide-ranging pastoral group ranches in this region.

Heart girth and live weight averaged 131 cm ± 1 (SE) and 176 kg ± 3. There was a strong linear positive relationship between heart girth and live weight (R 2 = 0.90, p < 0.001). The relationship was described by the equation LW = -265 + 3.37HG, where LW = live weight and HG = heart girth. The model prediction error (residual standard error) was 12.8 kg, which translates to 7.3% of the mean live weight of the heifers used in this study. A comparison of this model with other models previously developed for Zebu female cattle in East African rangelands showed that this model had the highest prediction accuracy (i.e. lowest prediction error). Unlike other models, the model developed in this study was found to be accurate enough to enable estimation of heifer live weights for various management purposes including assessment of production-related traits, detecting small seasonal live weight changes, setting dosage levels for administration of veterinary drugs and grazing management decision-making.

Key words: Boran Zebu, grazing management, live weight prediction model, morphological measurements, pastoral group ranches, small east African Zebu


Introduction

Assessing livestock live weights is necessary for ensuring informed management of both livestock and their natural environment. Specifically, this information is required for a number of purposes including determining appropriate feeding level and nutritional condition of animals, growth rate, sale prices, correct drug dosage, and responses to genetic selection (Young 1972, Machila et al 2008, Ozkaya and Bozkurt 2008, Lesosky et al 2013, Lukuyu et al 2016). Additionally, in range-based livestock production systems, assessing animal live weights is necessary for setting appropriate levels of forage allowance (amount of forage per unit body weight), grazing intensity and stocking rates. Therefore, in such livestock production systems, assessing animal live weight is critical in implementing appropriate grazing management practices for sustainable rangeland productivity.

The most direct way to estimate livestock weights is to use of a well calibrated and accurate weighing scale. However, this method is rarely applicable in many rural areas in Africa because weighing equipment are costly, need technical maintenance and are difficult to transport, especially in arid and semiarid pastoral rangelands (Young 1972, Machila et al 2008, Kashoma et al 2011). Consequently, pastoralists often rely on visual appraisal of animal body weight or condition, but this method suffers from subjectivity and is prone to substantial inaccuracies (Lesosky et al 2013, Tebug et al 2016). In the absence of a weighing scale, livestock live weights can be estimated indirectly through allometric relationships, where live weight is predicted from morphological measurements taken at specific locations of the body (see Kugonza et al 2011 for examples). Among the morphological measurements commonly used, heart (chest) girth is the most strongly correlated with live weight both in cattle (Ulutas et al 2002, Kashoma et al 2011, Kugonza et al 2011, Katongole et al 2013, Rashid et al 2013) and other livestock species (Ihuthia et al 2010, Mutua et al 2011)

The accuracy and applicability of allometric models for predicting cattle live weights can be influenced by sex, breed, environment, production system and animal husbandry practices (Young 1972). Heart girth-live weight models have been developed for indigenous cattle in many parts of sub-Saharan Africa, but majority of these models were developed in private ranches and research stations as opposed to traditional pastoral settings (Goe et al 2001). In Kenya, the few studies that have developed these models under pastoral production systems were conducted 30-45 years ago and were largely concentrated in southern and coastal rangelands (for example, see Young 1972, Semenye 1979, Sandford et al 1983). Models for predicting cattle live weights have rarely been developed for semiarid pastoral rangelands in northern Kenya despite the expansiveness and local and national economic importance of these rangelands. In addition, the few previous model development for cattle in this region largely involved male cattle or relatively old (> 5 year-old) female cattle (Young 1972). Accurate and timely models for predicting live weight of heifers in these pastoral rangelands are lacking.

The objective of this study was to develop a heart girth-live weight prediction model for heifers in pastoral cattle herds in semiarid rangelands of northern Kenya. Pastoralists in this region and similar landscapes are in need of such models to enhance livestock husbandry and grazing management practices. In particular, many pastoral group ranches in these and similar rangelands across East Africa are currently considering or are in the process of altering grazing management practices with a view to improving rangeland productivity (Skinner et al 2010, Odadi et al 2017). Successful implementation of the new grazing management practices requires development of grazing plans to ensure that forage availability is matched with consumption by livestock. Therefore, accurate estimation of livestock live weight is paramount to development of such grazing plans and monitoring livestock condition under altered grazing management regimen. This need is particularly great for female cattle, which usually form the bulk of pastoral cattle herds in these rangelands and other parts of sub-Saharan Africa (Food and Agriculture Organization [FAO] 2002).


Methods and materials

Study area

The research was conducted in eight communal pastoral properties (group ranches) in northern Kenyan rangelands, namely, Il Motiok, Koija, Il Ngwesi, Leparua, Il Polei, Monishoi, Murupusi and Musul (Figure 1). All sampled group ranches except Leparua are located in Laikipia County. Leparua is located in the neighbouring Isiolo County. All the sampled group ranches form part of the Northern Rangelands Trust (NRT) (Odadi et al 2017).

Figure 1. A map of the study area showing the location of the sampled pastoral group ranches
(Il Motiok, Koija, Il Ngwesi, Leparua, Il Polei, Monishoi, Morupusi and Musul)

Rainfall in the study area is generally low (annual mean ~ 430 mm) and highly variable both in space and time. In general, rainfall occurs bi-modally, with peaks in April-June (long rains) and October-November (short rains). The area is generally hot (mean annual temperature 16-33°C). The dominant vegetation is savanna grassland with varying densities of woody vegetation, primarily comprised by a mixture of several species of Acacia, Commiphora, Balanites, Boscia and Grewia. The understory comprises a diverse mixture of annual and perennial grasses and forbs (Odadi et al 2017). Pastoralism dominates among land use systems in the study area. Cattle (Bos indicus), sheep (Ovis aries) and goats (Capra aegagrus hircus) are the dominant livestock species in the area.

Study design and measurements

The study used a total of 160 (Bos indicus) heifers obtained from randomly selected pastoral herds from the sampled group ranches. Cattle in the pastoral group ranches in the study region are primarily unimproved Boran Zebu and its crosses with various strains of the Small East African Zebu. For Leparua and Il Ngwesi, 16-20 heifers owned by three to six different families were used in each group ranch. For each of the other sampled group ranches, 15-25 heifers were obtained from herds owned by five different families (approximately five heifers per family). The ages of the study heifers were approximated using dentition (Lawrence et al 2001). All study heifers were aged 1-3 years based on dentition; all had between 0 and 4 permanent incisors. Heifers with no permanent incisors were confirmed to be older than 1 year based on owner recalls.

For Leparua and Il Ngwesi, live weight and heart girth of study heifers were measured once in June 2014, while for all other sampled group ranches (Koija, Il Motiok, Il Ngwesi, Il Polei, Munishoi, Murupusi and Musul), measurements were made once in August 2015. Weather conditions during these measurements ranged from dry to moderately wet. Live weight was measured to the nearest 1 kg using a portable digital weighing scale; Tru-Test™ (Tru-Test Limited, Auckland, New Zealand). The weighing scale comprises a pair of heavy duty load bars, on which an aluminium platform is mounted. An electronic weigh scale indicator is connected to the load bars via a special cable. A portable crush was constructed for the weighing scale to help restrain animals during weighing. Heart girth was measured while heifers were standing just before being led to the weighing scale. To facilitate heart girth measurement, an experienced livestock handler from among the local community members applied traditional techniques to hold each heifer being measured still. An ordinary plastic tape measure was drawn around each heifer directly behind its front legs and the base of its hump. Heart girth was measured to the nearest 1cm. All measurements and readings were performed by an experienced crew. All measurements were performed at cattle bomas (night enclosures) early in the morning before herds left for grazing.

Data analysis

Prior to analysis, the sampled group ranches were grouped into three clusters based on their spatial closeness to each other, namely Il Ngwesi cluster (comprising Il Ngwesi and Leparua), Il Motiok cluster (Il Motiok and Koija), and Il Polei cluster (Il Polei, Morupusi, Munishoi and Musul). The relationship between heart girth and live weight was determined with general linear models in the R environment (R 3.3.0; R Core Team 2016). First, a full multiple linear regression model was run with live weight as the dependent variable, and heart girth, group ranch cluster and heart girth by group ranch cluster interaction as the independent variables. Because heart girth by group ranch cluster interaction was not significant, a simple linear regression model with live weight as the dependent variable and heart girth as the independent variable was run. Both multiple and simple linear regression models were tested for normality, homoscedasticity and other requisite assumptions using the gvlma function in the gvlma package (Pena and Slate 2014), and were found to satisfy all the assumptions.

The developed simple linear regression model was evaluated by comparing it with several models developed for Zebu cattle in different parts of East Africa (Table 1). Model comparisons were made using the mean-square prediction error (MSPE; Yan et al 2009) calculated as MSPE = 1/n (Ʃ[P-A] 2), where P = predicted live weight, A = actual live weight and n = number of pairs of values being compared. The square root of MSPE (RMSPE), which equals the residual standard deviation, and pRMSPE, which equals RMSPE as a proportion of mean actual live weight, were used to describe the prediction accuracy (Yan et al 2009).

Table 1. Live weight prediction models developed previously for Zebu female
cattle of different ages across East African rangelands. The models were used to
evaluate the models developed in the current study

Breed or Strain

Age (years)

Model

R 2

Tanzanian short-hornA

All

LW = -525 + 6.24HG

0.87

Taita (indigenous)B

< 2

LW = -137 + 2.22HG

0.93

Taita (indigenous)B

≥ 2

LW = -289 + 3.53HG

0.70

Taita (improved)B

≥ 2

LW = -334 + 3.98HG

0.80

Boran (northern Kenya)B

> 4

LW = -426 + 4.67HG

0.71

Maasai (indigenous)B

< 2

LW = -180 + 2.72HG

0.91

Maasai (indigenous)B

2

LW = -301 + 3.73HG

0.77

Maasai (indigenous)B

3

LW = -132 + 2.60HG

0.34

Maasai (indigenous)B

4

LW = -242 + 3.39HG

0.63

Maasai (indigenous)B

≥ 5

LW = -204 + 3.21HG

0.56

Maasai (indigenous)B

2-4

LW = -262 + 3.49HG

0.64

Maasai (crossed with Sahiwal)B

< 2

LW = -233 + 3.25HG

0.93

Maasai (crossed with Sahiwal)B

2-4

LW = -447 + 4.75HG

0.79

Maasai (crossed with Sahiwal)B

≥ 5

LW = -510 + 5.27HG

0.82

Boran (southern Ethiopia?)C

>3

LW = -433 + 4.81HG

0.66

LW = live weight; HG = heart girth; A Kashoma et al 2011; BYoung 1972;
CNicholson and Sayers 1987


Results and discussion

Heart girth and live weight of the heifers used in this study were 131 cm ± 1 (SE) and 176 kg ± 3, respectively. These values compare well with those reported for indigenous Taita Zebu female cattle aged 2-3 years (girth 132 cm, live weight 177 kg), but differ markedly from those reported for improved Taita Zebu and indigenous and improved Maasai Zebu belonging to the same age range (Young 1972). The simple linear regression model developed in the present study is LW = -265 + 3.37 HG (Figure 2), where LW is live weight (kg), and HG is heart girth (cm). This equation implies that a 1 cm change in heart girth would result in a weight change of 3.37 kg. This model is comparable to models developed for indigenous Taita Zebu and indigenous Maasai Zebu female cattle aged 2-4 years (Table 1; Young 1972).

Figure 2. Linear regression model showing relationship between heart girth (HG) and
live weight (LW) of heifers in northern Kenyan pastoral rangelands.

The model developed in the present study showed a strong positive linear relationship between heart girth and live weight, with a coefficient of determination (R2) of 0.90 and a residual standard error of 12.8 (p < 0.001; Figure 2). Multiple linear regression showed that the relationship between heart girth and live weight did not differ among group ranch clusters (main effect of cluster p = 0.192, F = 1.67; cluster by heart girth interaction effect p = 0.959, F = 0.042; Figure 3). Because the group ranch clusters used here are wide ranging and differ markedly in soils, vegetation and climatic characteristics (Odadi et al 2017), the observed lack of significant influence of group ranch cluster on the relationship between live weight and heart girth suggests that the simple linear model developed here is robust enough to be used across different pastoral locations in the study region.

Figure 3. Relationship between heart girth and live weight of heifers across
different pastoral group clusters in northern Kenyan rangelands.

The prediction error (residual standard error) of 12.8 kg obtained from the model developed for the present study translates to approximately 7.3% of the mean live weight of study heifers. This error is much smaller than the acceptable error for accurate dosing of veterinary drugs (20%; Lesosky et al 2013), assessing production-related traits in individual animals (10%, Goopy et al 2017), and detecting small seasonal live weight changes (11-17%; Goopy et al 2017). When live weights predicted by models previously developed for Zebu female cattle in different parts of East Africa (see Table 1) were regressed against actual live weight measurements from the present study, the resulting coefficients of determination ( R2) were all approximately 0.9, similar to that obtained in the present study (Table 2). However, these previous models had higher prediction errors when compared to the present study; their mean-square prediction error (MSPE), square root of MSPE (RMSPE), and RMSPE as a proportion of actual mean live weight ranged from 218 to 14622 kg, 13.3-121 kg and 7.6-69%, respectively (Table 2).

Table 2. Prediction errors (predicted vs. actual live weight) of the present study and previous studies for Zebu female cattle in East African rangelands

Breed or Strain

Age (years)

MSPE

RMSPE

pRMSPE

Tanzanian short-hornA

All

14622

121

68.8

Taita (indigenous)B

< 2

789

28.1

16.0

Taita (indigenous)B

≥ 2

178

13.3

7.58

Taita (improved)B

≥ 2

329

18.1

10.3

Boran (northern Kenya)B

> 4

459

21.4

12.2

Maasai (indigenous)B

< 2

218

14.8

8.41

Maasai (indigenous)B

2

311

17.6

10.0

Maasai (indigenous)B

3

1244

35.3

20.1

Maasai (indigenous)B

4

850

29.2

16.6

Maasai (indigenous)B

≥ 5

8482

92.1

52.4

Maasai (indigenous)B

2-4

510

22.6

12.9

Maasai (crossed with Sahiwal)B

< 2

13386

116

65.9

Maasai (crossed with Sahiwal)B

2-4

412

20.3

11.6

Maasai (crossed with Sahiwal)B

≥ 5

656

25.6

14.6

Boran (southern Ethiopia?)C

Mature

869

29.5

16.8

BoranD

2-4

163

12.8

7.26

A Kashoma et al 2011; BYoung 1972; CNicholson and Sayers 1987;DPrsesent study; N.E.P = North Eastern Province, Kenya ; MSPE = mean-square prediction error; RMSPE = square root of MSPE; pRMSPE = RMSPE as a proportion of mean actual live weight; R2 = 0.90 for all cases

Previously developed models for Tanzanian short-horn (all age groups), indigenous Maasai Zebu females aged 5 years or older, and Maasai x Sahiwal females aged less than 2 years gave the largest prediction errors (>52%; Table 2). Therefore, application of these models in predicting live weight of heifers in the study region can lead to serious inaccuracies even for purposes of administration of veterinary drugs. Models for indigenous Taita Zebu (≥ 2 years), indigenous Maasai Zebu (< 2 and 2 years), and improved Taita Zebu (≥ 2 years) gave the smallest but slightly higher prediction errors than the error obtained from the present study (7.6-10.4% vs. 7.3%; Table 2), indicating that they can be useful in predicting live weight of heifers in the study region even for purposes of assessing production-related traits and detecting seasonal weight changes. All other assessed previous models gave prediction errors ranging from 11.6 to 20.1%, suggesting that they can be useful for administration of veterinary drugs or estimating seasonal weight changes but not for assessing production-related traits in heifers in the study region. The observed variations between the model developed in this study and several models developed by other studies could be due to the different genetic effects, age differences and management practices of animals involved in the studies.


Conclusions


Acknowledgements

This research was funded by The Nature Conservancy (TNC). I am grateful to Boniface Kirwa, Martin Shisanya, James Kipsoi and Joseph Sadrbabi for field assistance, several pastoral community members for study cattle, and several community conservancies and group ranches as well as the Northern Rangelands Trust (NRT) for helping make this research possible.


References

Food and Agriculture Organization (FAO) 2002 Cattle and small ruminant production systems in sub-Saharan Africa: A systematic review. FAO, Rome, Italy. Retrieved November 09, 2017, from http://www.fao.org/ag/againfo/resources/en/publications/agapubs/AGAL-Y4176E.pdf

Goe M R, Alldredge J R and Light D 2001 Use of heart girth to predict body weight of working oxen in the Ethiopian Highlands. Livestock Production Science, Volume 69, Issue 2001: 187-195.

Goopy J P, Pelster D E, Onyango A, Marshall K and Lukuyu M 2017 Simple and robust algorithms to estimate liveweight in African smallholder cattle. Animal Production Science, Available early online. Retrieved November 09, 2017, from http://www.publish.csiro.au/AN/AN16577

Ihuthia P M, Wahome R G, Wanyoike M M M 2010 Correlation of actual live weight and estimates of live weights of Camel calves (Camelus dromedarius) in Samburu District of Northern Kenya. Journal of Camelid Science, Volume 3: 26-32.

Kashoma I P B, Luziga C, Werema C W, Shirima G A and Ndossi D 2011 Predicting body weight of Tanzania shorthorn zebu cattle using heart girth measurements.Livestock Research for Rural Development. Volume 23, Article #94. Retrieved August 30, 2017, from http://www.lrrd.org/lrrd23/4/kash23094.htm

Katongole C B, Mpairwe D, Bareeba F B, Mukasa-Mugerwa E and Ebong C 2013 Predicting body weight from heart girth, height at withers and body condition score in Bos indicus cattle bulls of Uganda. Livestock Research for Rural Development. Volume 25, Article #46. Retrieved August 31, 2017, from http://www.lrrd.org/lrrd25/3/kato25046.htm

Kugonza D R, Nabasirye M, Mpairwe1 D, Hanotte O and Okeyo A M 2011 Productivity and morphology of Ankole cattle in three livestock production systems in Uganda. Animal Genetic Resources, Volume 48: 13–22.

Lawrence T E, Whatley, J D, Montgomery T M and Perino L J 2001 A comparison of the USDA ossification-based maturity system to a system based on dentition. Journal of Animal Science, Volume 79, Issue 7: 1683-1690.

Lesosky M, Dumas S, Conradie I, Handel I G, Jennings A,·Thumbi S,·Toye P and de Clare Bronsvoort B M 2013 A live weight–heart girth relationship for accurate dosing of east African shorthorn zebu cattle. Tropical Animal Health and Production, Volume 45, Issue 1: 311-316.

Lukuyu M N, Gibson J P, Savage D B, Duncan A J, Mujibi F D N and Okeyo A M 2016. Use of body linear measurements to estimate liveweight of crossbred dairy cattle in smallholder farms in Kenya. Springerplus, Volume 5, Article #63. Retrieved November 09, 2017, from https://www.ncbi.nlm.nih.gov/pmc/articles/PMC4722050/

Machila N, Fèvre E M, Maudlin I and Eisler M C 2008 Farmer estimation of live body weight of cattle: implications for veterinary drug dosing in East Africa. Preventive Veterinary Medicine Volume 87, Issues 3–4: 394-403.

Mutua F K, Dewey CE, Arimi S M, Schelling E, Ogara W O 2011 Prediction of live body weight using length and girth measurements for pigs in rural Western Kenya. Journal of Swine Health and Production, Volume 19, Issue 1: 26–33. Retrieved November 09, 2017, from https://cgspace.cgiar.org/bitstream/handle/10568/3087/mutua%20et%20al%202011.pdf?sequence=1&isAllowed=y

Nicholson M J and Sayers A R 1987 Relationships between body weights, condition score and heart girth changes in Boran cattle. Tropical Animal Health and Production, Volume 19, Issue 2: 115-120.

Odadi W O, Fargione J and Rubenstein D I 2017 Vegetation, wildlife, and livestock responses to planned grazing management in an African pastoral landscape. Land Degradation and Development, Volume 28, Issue 7: 2030–2038.

Ozkaya S and Bozkurt Y 2009 The accuracy of prediction of body weight from body measurements in beef cattle. Archiv Tierzucht, Volume 52, Issue 4: 371-377.

Pena E A and Slate E H 2014 gvlma: Global Validation of Linear Models Assumptions. R package version 1.0.0.2. Retrieved November 09, 2017, from https://cran.r-project.org/web/packages/gvlma/gvlma.pdf

R Core Team 2016 R: A language and environment for statistical computing. R Foundation for Statistical Computing, Vienna, Austria. Retrieved August 20, 2016, from https://www.R-project.org/

Rashid M, Hoque A, Huque S K, Talukder A. I and Bhuiyan A K F H 2015 Morphometric characterization of Brahman crossbred cattle and prediction of live weight using linear body measurements. Asian Journal of Medical and Biological Research, Volume 1, Issue 3: 569-577. Retrieved November 09, 2017, from https://www.banglajol.info/index.php/AJMBR/article/view/26480/17772

Semenye P 1979 The Estimation of Liveweight of Zebu Cattle From Heart Girth Measurements. Working Document. ILCA Nairobi.

Sandford J, Durkin J and Trail J C M 1983 Evaluation of cattle weights by band at Muhaka, Kenya. Working Document. ILCA, Nairobi.

Skinner D 2010 Rangeland Management for Improved Pastoralist Livelihoods: The Borana of Southern Ethiopia (MA thesis), Oxford Brookes University, Oxford, UK. Retrieved November 09, 2017, from http://architecture.brookes.ac.uk/research/cendep/dissertations/DjihanSkinner.pdf

Tebug S F, Missohou A, Sabi S, Juga J, Poole E J, Tapio M and Marshall K 2016 Using body measurements to estimate live weight of dairy cattle in low-input systems in Senegal, Journal of Applied Animal Research, DOI: 10.1080/09712119.2016.1262265. Retrieved November 09, 2017, from http://www.tandfonline.com/doi/pdf/10.1080/09712119.2016.1262265?needAccess=true

Ulutas Z, Saatci M and Özlütürk A 2002 Prediction of body weights from body measurements in East Anatolian Red calves. Indian Journal of Animal Sciences, Volume 72, Issue 10: 878–881.

Yan T, Mayne C S, Patterson D C and Agnew R E 2009 Prediction of body weight and empty body composition using body size measurements in lactating dairy cows. Livestock Science, Volume 124, Issues 1-3: 233–241.

Young D L 1972 The estimation of liveweight from heart girth within specified age/sex groups of Kenyan range cattle. East African Agricultural and Forestry Journal, Volume 38, Issue 2: 193–200.


Received 9 November 2017; Accepted 5 December 2017; Published 1 January 2018

Go to top