26 May 1995

NIABwrong with OSRlists, says seed firm…

By Charles Abel

SHORTCOMINGS in the winter oilseed rape recommended lists have been slammed by Zeneca Seeds.

It says 13 varieties are ranked incorrectly, with yield differences between some exaggerated by up to 3%.

Managing director Tony Hayter urges growers to treat NIABs analysis with "extreme care", particularly the response to fungicide table. "Were not knocking the trials protocols or the data. Its how it is analysed and interpreted thats at issue."

The company says errors stem from a statistical procedure used by NIABand the breeding industry, known as Fitcon. It analyses trends to produce an average result for years where data is limited. "It works well where up to a third of the data is missing. But where more than half is missing it is less reliable," explains technical manager Paul Hickman.

In that situation Fitcon places too much weight on years of numerous trials. If a variety performs particularly well one year and not so well another, that can create a misleading picture.

Last years results bore little resemblance to 1993s. Light leaf spot- resistant lines like Amber and Nickel did particularly well. With few results from earlier years, Fitcon drew heavily on 1994 results, placing the varieties top for yield in the response to fungicide tables.

Tables incorrect

Yet 70% of their data was missing. "Our work shows Amber and Nickel were actually 5% lower yielding than Apex, in 1992 and 1993. The NIAB tables are not correct," says Mr Hickman.

Fitcon could also affect a varietys chances of recommendation. "You have to remember 1% yield can be the difference between recommendation and failure." This years new recommended rapes all carry 8 and 9 ratings for light leaf spot resistance, he notes.

Although light leaf spot may have caused last years variations, other factors may be at play, which could influence results in future.

Mr Hayter urges NIAB to adopt an alternative statistical system, which would use Fitcon to come up with a varietys average performance each year and only then calculate an overall 3-year mean.

He stresses that the criticism is not on marketing grounds. "We just want to raise the debate. NIAB has accepted our point, and now indicates the number of years a variety has been in trials. But it says it cant change the system unless all the industry wants it."

Breeders, like growers, want consistent advice, points out Mr Hickman. "We dont want to start multiplying a variety up in the expectation that it is going to do well and then find that weighted analysis knocks it down. And the farmer doesnt want to choose a variety which looks better than it actually is.