02.01.2015 Views

RA 00110.pdf - OAR@ICRISAT

RA 00110.pdf - OAR@ICRISAT

RA 00110.pdf - OAR@ICRISAT

SHOW MORE
SHOW LESS

Create successful ePaper yourself

Turn your PDF publications into a flip-book with our unique Google optimized e-Paper software.

Pearl Millet Regional Trials by CILSS, 1981-1984<br />

(Essais regionaux du mil conduits par le CILSS 1981-1984)<br />

G. Loynet, A. Kere, and O. Sidibe<br />

Institut du Sahel, CILSS, Bamako, Mali<br />

Superior pearl millet varieties grown in the Sahelian region were evaluated from 1981 to 1984 as part of the<br />

Crop Improvement Project for pearl millet, sorghum, cowpea, and maize, directed by the Institut du Sahel.<br />

The best varieties were identified for different climatic situations extending from Chad in the east to Senegal<br />

and the subdesert, Sahelian, and Sudano-Sahelian Zones in the west. The varieties were grouped geographically<br />

by dividing the region into three agroecologically representative zones. Areas with rainfall between 200<br />

and 600 mm are most suited to growing pearl millet; disease risks increase in high-rainfall regions.<br />

Analysis of the International Pearl Millet Adaptation Trial (IPMAT)<br />

(Analyse de l'Essai international du mil—objet adaptation (IPMAT))<br />

J.R. Witcombe<br />

Principal Plant Breeder, Pearl Millet Group, ICRISAT, Patancheru, Andhra Pradesh 502 324, India<br />

The International Pearl Millet Adaptation Trial ( I P M A T ) , which includes both hybrids and varieties as<br />

entries, has been grown multilocationally in India and Pakistan. The grain yield data over 5 years have been<br />

analysed in a number of ways.<br />

A regression analysis indicated that the breeder's procedure of selecting from among the highest-yielding<br />

entries across environments is satisfactory, as it also selects entries that perform well in poor environments.<br />

Selecting entries on predicted performance in the lowest-yielding environment appears to be a less reliable<br />

procedure.<br />

The hybrids are generally higher-yielding than the varieties, but are less stable. The most important source<br />

of genotype x environment interaction in the regression analysis was due to the deviation from the regressions<br />

(S 2 d values), rather than variation between the regressions. The varieties were superior to the hybrids in this<br />

respect, with lower than average S 2 d values.<br />

A mean-standard deviation analysis showed that the highest-yielding genotype would always be preferred<br />

by the average, risk-averse farmer.<br />

Discussion<br />

Because composites may have high levels of variability, the suggestion was made to derive varieties by random<br />

mating a few selected progenies from the composite rather than using the composite itself as the finished<br />

product. It was pointed out that a few selection cycles in a diverse composite may result in reduced variability,<br />

so that the composite can be tested as the finished product.<br />

It was suggested that analysis of variance and t-test may be better methods than regression analysis for<br />

stability analyses. Nearest neighbor analysis (NNA) and lattice designs were also suggested. In reply it was<br />

pointed out that mean-standard deviation analysis was proposed as a better alternative to conventional<br />

regression analysis, and that N N A and lattice designs were impractical for across-location analyses.<br />

The comment was made that stability can be a property of individual genotypes. Varieties appeared to be<br />

more stable than hybrids for s 2 d values but this could be the result of a statistical artifact of the regression<br />

analysis.<br />

293

Hooray! Your file is uploaded and ready to be published.

Saved successfully!

Ooh no, something went wrong!