Cereal production in the African rain forest is both hectic and highly capital
intensive as a result of the activities of pests. This is one reason that accounts
for the high cost of weaning formulas where cereals are the principal raw materials.
Most families are unable to properly wean their children thereby exposing them
to diet-related diseases and infant mortality.
Cereals forms the primary basis for most of weaning foods in West Africa. In
Nigeria, the usual first weaning food is called pap, akamu, ogi or koko and
is made from maize (Zea mays), millet (Pennisetum americanum),
or guinea corn (Sorghum sp.) (King and Ashworth, 1987;
Longhurst, 1984; Cherian, 1981;
Osuhor, 1980). From the nutritional stand point, protein
content of cereals is of poor quality, low in lysine and tryptophan which are
indispensable in growth of the young child (Oyenuga, 1968).
Since, a balanced protein preparation which can provide essential amino acids
can be formulated through mixing proteins from vegetable sources, a number of
alternatives have been developed for infant weaning foods (Plahar
and Hoyle, 1991; Takyi et al., 1991; Fashakin
et al., 1986; Ketiku and Ayoku, 1984; Fashakin
and Ogunsola, 1982). Flour from Dioscorea dumetorum have been developed
and nutrient bioavailability studies on laboratory animals and school age children
carried out coupled with amino acid composition of Dioscorea dumetorum
(Alozie et al., 2008, 2009;
Mbome et al., 1995; Mbome
and Treche, 1994; Treche et al., 1984). It
was observed that D. dumetorum flour has a high protein content which
was quite balanced in essential amino acid (with slight deficiency in lysine,
methionine and cystine) and comparable to cereals (maize and rice). The feeding
of D. dumetorum based diet was observed to result in high protein utilization
parameters and comparable to cereal based diet.
However, the procedure for determination of protein quality and amino acid
analysis which are important indices of food quality to elucidate authenticity
of the food product or raw material for food manufacture is cumbersome and capital
intensive. Ratio based nutritional indices as useful as they are, have probable
unrealistic assumption of an isometric relationship between denominator and
numerator variables (Raubeheimer and Simpson, 1992).
From the foregoing, this study sought to develop a statistical technique for
assessing protein quality of a given foodstuff thereby reducing cost of analysis.
The study sought to do this using data obtained from an experiment on the feeding
of groups of albino rats using maize, rice and Dioscorea dumetorum, steamed
flours conducted in an earlier study (Alozie et al.,
2008). Both the initial and final weights of all the rats were taken. These
weights constitute the data upon which the analysis of covariance was applied.
This comparative statistical technique is used to determine the quality of the
nutritional value of maize, rice and Dioscorea dumetorum. The validity
of the method draws from the existence of a linear relationship between the
original weights and the final weights.
Analysis of covariance is employed when concomitant (extraneous) variables
cannot be held fixed but can be measured (Johnson, 2004).
It is an experimental situation for which we cannot block the extraneous variables.
The linear statistical model for the analysis of covariance is given by:
for i = 1,2,
,k; j = 1,2,
where μ is the grand mean, αi is the effect of the ith treatment, β is the slope of the linear regression equation and the eijs are independent, normally distributed variables with zero means and common variance σ2. xij is the concomitant variable and it is eliminated by estimating β by least squares methods. An analysis of variance is then performed on the adjusted ys, namely:
Let S, T and E denote sums of squares, cross products for total treatments
and error respectively. Then (Montgomery, 1978):
The least squares estimator for β is:
The error sum of squares:
The experimental error variance is estimated by:
If there is no treatment effect (i.e., αi = 0) we obtain a reduced model whose error sum of squares is:
with a -2 degrees of freedom.
To test the hypothesis of no treatment effects
H0: Ti = 0, we compute
which is distributed as Fa-1, a(n-1)-1 if the null hypothesis is true.
The hypothesis of no linear relationship between the response variable and the concomitant variable stated as H0:β = 0 was tested by using the test statistic:
and comparing it with F1, a(n-1)-1
RESULTS AND DISCUSSION
The original weights of 21 weanling albino rats were taken before they were
randomly assigned to three of carbohydrate foodstuffs, namely rice, maize and
Dioscorea dumetorum (trifoliate yam). The carbohydrate types were steamed.
The data is captured in the Table 1.
||Body weights of albino Wistar rats fed cereals (maize and
rice) and Dioscorea dumetorum
Under the null hypothesis, H0: αi = 0; the result of the analysis of covariance performed on the data shows:
Thus, the null hypothesis that there is no significance difference between the three carbohydrate types is accepted.
Under the null hypothesis H0:β = 0:
This implies that there is a linear relationship between the original weights of the rats and the final weights and the adjustment provided by the analysis of covariance is necessary.
Results obtained in this study confirm the result earlier reported by Alozie
et al. (2008) on the comparative evaluation of protein quality of
Dioscorea dumetorum varieties and cereals using biological methods. The
findings showed the Dioscorea dumetorum varieties to be of similar protein
quality to the cereals and resulted to higher values for protein utilization
parameters in albino wistar rats. Using the analysis of co-variance in this
study, the Dioscorea dumetorum was not significantly (p<0.05) differed
from the cereal fed rats. The establishment of a significant linear relationship
between the original and final weights is an indication that the raw material
(Dioscorea dumetorum) is of good protein quality. This advantage of examining
the relationship between original and final weights is a major benefit of using
analysis of co-variance rather than analysis of variance. We conclude that D.
dumetorum is equally good as raw material for the production of infant formula
as rice and maize which are currently used by manufacturers and that analysis
of covariance of weight gain or loss of animals used in experimental feeding
trials may serve as an alternative method for determination of protein quality
of foodstuff thereby eliminating or reducing the cost and/or cumbersomeness
encountered by using the chemical and biological methods.