J.C.M. Trail and J. Durkin
Quantity and value of previous animal productivity research in Africa
Present ILCA resources for analysis of current animal productivity research
In planning for increased productivity of animals in tropical environments, breeders need information on the performance of major animal types in different ecological zones, production systems, management levels, etc. Logical decisions on selection between breeds require both (a) comparisons made in the same environment, and (b) information on a sufficient number of performance traits to allow an acceptable index of overall productivity to be constructed. An assessment of past research work which satisfies these criteria can save on future inputs. The first part of this paper comprises an examination, using cattle as an example, of the quantity of past research work and its value in determining the comparative performance of breeds in Africa. Secondly there is a description of the present ILCA resources available for analysis and interpretation of field data.
Previous research on the productivity of cattle breeds in Africa has been assessed and is described here. There has been no similar assessment for sheep and goats although the procedure would be parallel to that for cattle. In a bibliography (Trail, 1981) covering performance aspects of indigenous, exotic and crossbred cattle in Africa south of the Sahara, approximately 550 items are listed, of which 50 are review or descriptive papers. The remaining 500 contain objective original data on some aspect(s) of reproductive performance, growth, viability or milk production. This bibliography was built up over a three-year period as a by-product of the studies of production systems being carried out in various zones of Africa by the International Livestock Centre for Africa. It covers the 30-year period, 1949 to 1978. For the purpose of evaluation, each of the 500 items was classified by year of publication, type of publication, language, ecological zone, cattle group, and performance trait(s) measured.
The 500 references are distributed throughout the five ecological zones as shown in Table 5. There appears to be some positive correlation between cattle numbers and publications produced, although the very arid zone, where conditions are extremely harsh and the temperate highland zone where conditions are relatively favourable, are both the subject of few publications relative to their cattle populations and the tsetse infested zone has been the subject of a considerably larger proportion of publications.
The number of publications has increased sharply over the past three decades: 9% were produced from 1949 to 1958, 38% from 1959 to 1968 and 53% from 1969 to 1978. Fifty-eitht percent of the publications were produced in English, 38% in French, 2% in German and 2% in Portuguese.
Table 5. Percentages of cattle and bibliography references in the five ecological zones
|
Ecological zone (rainfall) |
Percentage of cattle population |
Percentage of items in bibliography |
|
Very arid (< 400 mm) |
6 |
2 |
|
Arid to semi-arid (400 - 600 mm) |
22 |
22 |
|
Semi-arid to humid (> 600 mm) |
44 |
51 |
|
Temperate highland |
22 |
9 |
|
Humid, tsetse infested |
6 |
16 |
With reference to types of publication, 52% of the items listed are articles in scientific journals, 17% appear in bulletins or university or FAO publications, 15% appear in research station reports, 9% are papers presented at regional meetings, 4% are theses, and 2% are papers presented at international meetings. (Annual reports covering the same topic over several years were counted only once).
Eighty-one percent of the items contain information on indigenous breeds, 30% on crossbreeds and 14% on exotic breeds. With reference to performance traits, 72% of the publications contain information on growth, 41% on reproductive performance, 33% on milk production and 18% on viability.
Among publications on indigenous, crossbred or exotic cattle types, the percentage of publications containing information on the four main performance traits are shown in Table 6. Reports on milk production thus represent a much larger proportion of all publications on exotic breeds than on indigenous breeds, with crossbreeds in an intermediate position. The distribution of reports on growth runs in the opposite direction, while reports on reproduction and viability are distributed evenly among all three cattle types.
Table 6. Distribution of performance traits by cattle types
|
Cattle types |
Percentage of items covering: |
|||
|
Milk |
Growth |
Reproduction |
Viability |
|
|
Indigenous |
27 |
78 |
41 |
19 |
|
Crossbred |
37 |
65 |
41 |
20 |
|
Exotic |
48 |
57 |
41 |
17 |
To sum up the results of the overall analysis of this bibliography, it shows that only about 20% of the references contain information on three or more performance characters sufficient to allow characterisation of breed types through a productivity index. For example one simple productivity index used is 'weight of calf plus liveweight equivalent of milk produced per unit weight of cow maintained per year', this index being extended to cover more traits if information is available. Additionally only 20% of the references contain comparative information on two or more breed types. When these two necessary attributes are put together, only 5% or 25 of the reports provide sufficient data to allow breed comparisons on the basis of a productivity index. This illustrates that the majority of research work does not give useful information about the comparative performance of cattle breeds in Africa. If a similar analysis was performed for small ruminants it is likely that a similar (or worse) picture would emerge: i. e. breed comparisons must depend on a small amount of relevant information.
In 1980, ILCA purchased a Hewlett-Packard 3000 series III computer system. The system consisted of the Computer Processor, two 300 line-per minute printers, two 120 million-character on-line disks, a nine-track magnetic tape unit, a 300 card-per minute reader, and 12 interactive terminals. An extra 5 terminals have now been added because the need for data entry and analysis has increased.
ILCA cannot afford to develop large computer software, that is computer packages comprising sophisticated programs. Programmers today are very expensive relative to the cost of computer hardware. As a result the ILCA Computer Unit has sought out and purchased software already written for the HP 3000 III. This software is then supported by larger organisations with many programmers and it is repeatedly being enhanced. For Research, ILCA has acquired SPSS (Statistical Package for the Social Sciences) and BMDP (Biomedical Computer Programs) from McMaster University, Hamilton, Canada; LINDO (Linear Programming Package) from Chicago and "Harvey's Least Squares" from Ohio State University.
While the design of such complete packages are beyond ILCA's resources, the programming staff are frequently writing smaller programs to edit, verify and construct data files for use by the larger software packages.
The computer system is used predominantly for analysis of scientific research data, mainly agricultural production statistics. SPSS and BMDP are used extensively in ILCA and they allow the scientist to tabulate and statistically analyse his data quickly and efficiently. SPSS developed by Nie, Hull, Jenkins, Steinbrenner and Bent (1975), has been updated to its present version SPSS 9.1. It allows many difficult, time-consuming and generally unrewarding tasks to be undertaken using a variety of single purpose computer programs, each with it own idiosyncratic control-card syntaxes and input data formats. The researcher thus spends less time as a data-preparation clerk and more time as a scientist analysing substantive results.
One very simple SPSS procedure is tabulation of data to show frequencies so that errors can be isolated very quickly and corrections can be made prior to the larger more expensive Least Squares Analysis. In SPSS each variable and values in certain variables are labelled so that data verification can be performed very easily and the 'breakdown' procedure gives a provisional estimate of the means for different traits. LINDO is used by researchers to analyse resource allocations within the constraints of various farming systems. Harvey's Least Squares (Harvey, 1977) is used to analyse animal productivity data and is especially valuable if genetic parameters have to be estimated.
The unit provides statistical analysis for staff from all ILCA field programmes and has started to analyse data from national research institutes. In the past twelve months, work has been completed for the Institute of Agricultural Research in Ethiopia and the Centre de Recherche Zootechnique in Kolda, Senegal. It is hoped that over the next twelve months assistance with data analysis will be given to more research institutes.
In an effort to streamline data collection from ILCA's field programmes, microcomputers will be installed within programmes. These computers will enable scientists to more easily verify their data and undertake initial analysis in the field. Eventually it is hoped that a complete data collection network can be set up using the micro-computers. In all interaction with computers, the cooperation between scientist and programmer is of utmost importance. The scientist must ensure the biological correctness of his results, and the programmer must perform analyses by the simplest, fastest techniques.
