Gr. Iason et al., ULTRASONIC-SCANNING FOR DETERMINATION OF STAGE OF PREGNANCY IN THE LLAMA (LAMA-GLAMA) - A CRITICAL COMPARISON OF CALIBRATION TECHNIQUES, Journal of Agricultural Science, 120, 1993, pp. 371-377
Measurements of foetal head diameter (HD) and trunk diameter (TD) were
made using ultrasonic scanning of 11 pregnant llamas (Lama glama). Ea
ch llama was scanned fortnightly to obtain measurements of HD and TD f
rom c. 84-271 and 65-168 days since mating respectively. There were ap
proximate linear relationships between TD and days after mating and be
tween HD and the logarithm of days after mating. Calibration equations
for predicting the number of days after mating (d) from foetal measur
ements constructed using (i) the inverse method in which d is regresse
d on HD or TD and (ii) the classical method in which HD or TD are regr
essed on d. These calibration methods were assessed by cross-validatio
n, treating each animal in turn as the individual for which prediction
s were required. Analysis of the prediction errors showed bias in the
classical method, which consistently underestimated d at low values. A
components of variance analysis indicated substantial variation betwe
en individuals which must be taken into account in calculating standar
d errors of prediction (S.E.P.) and confidence intervals. S.E.P. of d
from TD can be reduced from 12.5 to 10.4 days by increasing the number
of observations on an individual from one to four at fortnightly inte
rvals. For prediction from HD, the equivalent figures are size depende
nt: examples are from 8.3 to 5.6 days, and from 26 to 18 days, for HDs
of 2 and 6 cm respectively. The effect of small positive correlations
between residuals of successive fortnightly measurements on the same
llamas had a negligible effect on S.E.P.s, increasing them by c. 0.2 d
ays. Ultrasonic scanning is suitable for determination of stage of pre
gnancy of llamas providing S.E.P.s which are small in relation to thei
r long gestation period (335-360 days).