AgBioView from www.agbioworld.org : October 27, 2005
- Christopher Preston
The subject of crop yields often comes up in the debate about GM
crops. Promoters of the technology often talk about how yields can be
increased. Just as often we hear from the other side that GM does not lead to higher yields and frequently that GM crops yield less. Recently, Julie Newman of the Network of Concerned Farmers in Australia has been quoted in the Australian press making statements that GM canola yields 20% less than conventional varieties in Australia. Having seen trials and crops of both Roundup Ready and InVigor canola, I found this assertion somewhat surprising and chose to investigate its basis in data.
Having sifted through the Network's web site, I found a document authored by Julie Newman called "Bayer Cropscience's GM Invigor Canola" (http://www.non-gm-farmers.com/news_details.asp?ID=1992 ),
which seems to be the basis of these claims. Reading through the document I found her claim is made based on a passage from the Office of the Gene Technology Regulator's risk assessment. To quote:
"Invigor hybrid canola displayed approximately 15% better vigour than a conventional open pollinated variety, but 20% less vigour than a conventional hybrid variety". From this it appears Julie Newman can confidently assert that "actual trials of GM canola have produced yields of up to 20% less than non-GM varieties" (as cited on the ABC 20th September 2005). It seems somewhat of a stretch to turn data about vigour differences when you are not sure of the basis of measurement between varieties into actual yield trials.
In addition, Julie Newman fails to quote the first part of the passage from the risk assessment. This part refers to actual yields:
"InVigor canola varieties have displayed yield increases of 10-20% over conventional open pollinated varieties in Australia and greater than 20% in Canada ". As well as ignoring that statement, Julie Newman further supports her stance on yields being less for InVigor canola by citing trial work from Western Australia where InVigor 40 had the same yield as a conventional TT variety (InVigor Hybrid, Canola, WA Crop Updates available from
What Julie Newman is not telling is that InVigor 40 is a mid-late season variety, the trial quoted was sown late and therefore, the early-mid season varieties had a distinct advantage. Where growing seasons are short, later flowering varieties are often at a disadvantage because of the higher temperatures and lack of moisture in late spring and early summer. Even worse, Julie Newman fails to mention two experimental InVigor varieties grown in the same trial, both of which are early-mid maturity and both significantly out-yield InVigor 40 and all conventional varieties. Likewise, a comparison of trials from eastern Australia shows InVigor 40 yields generally similar to those of Hyola 60 (a conventional hybrid) and far above those of a conventional and a TT variety. Yields of InVigor (109%) were below those of Hyola 60 (120%) in 2001, but greater in 2002 (122% compared with 112%). Further trial work in 2003 produced
identical yield results for these two varieties
Clearly, there is a fair bit of cherry picking of data to make the case that Julie Newman is confidently asserting.
Julie Newman makes other comments about yields of InVigor canola including an odd comment that GM hybrid vigour is less than non-GM varieties. To quote (it is in both bold and underlined in the document): "The vigour displayed by F2 progeny would still be at least 25-30% less than that in some non-GM canola varieties." This comment either shows a lack of understanding of the process of hybrid creation or is deliberately intended to mislead. Hybrid crop varieties are always sold as the F1. That is the seed of a cross between two divergent parents. The heterosis created from such crosses overcomes problems caused by inbreeding depression. As the parents are divergent (otherwise there is no benefit to creating hybrids) the F2 will segregate and lose the advantage of hybrid
vigour. The F2 will also segregate for all other traits that differ including disease tolerance and time to flowering and would be useless for commercial production. This is why growers of hybrid
crops need to buy new seed every year. To talk about what might happen with yields of the F2 is pointless, as growers will never be growing the F2.
I found a second document, also authored by Julie Newman, which touches on yields of Roundup Ready canola in addition to InVigor canola ("Will GM canola yield more in Australia?"
www.non-gm-farmers.com/news_details.asp?ID=914). Early on this document has the bold statement that " it appears there are yield penalties with glyphosate resistant crops." and directs the reader to a news report from the website of "The Scientist" as the authoritative source of information. The document further goes on to say: "Monsanto's Australian trials revealed the real performance of GM Roundup Ready compared with unnamed non-GM varieties (and perhaps
deliberately selected low yielding varieties). The best yields revealed on Monsanto's website are dismal by any farmers comparison (1.055tonne/ha) and actually equates to 17% less than the Australian average canola crop ". The last section is all in bold. The rest of the discussion of yields of Roundup Ready canola focuses on this 1.055 tonne/ha and compares it unfavourably with specific individual trials conducted elsewhere and promotional literature of trial results. In one case, a comparison is made with a trial conducted elsewhere and the statement is made "GM Roundup Ready trials yielded almost 5 times less than non-GM canola trials." As different trials
are conducted in different places with different sowing and harvesting dates and different rainfall patterns, it is pointless to compare individual results from different trials in this way. However, it is a good way to ensure such statements are likely to be taken out of context for use elsewhere.
Aside from being pointless comparisons, the yield data selected by Julie Newman are all higher than state average crops for those areas. For example, the CSIRO trial conducted in NSW had yields of 3.6 to 5.2 tonne/ha in 2001. The NSW state average canola yield in 2001 was in fact 1.5 tonne/ha according to the Australian Oilseed Federation
Likewise, the promotional brochure from Pioneer Hi-Bred has yieldsfor 2003 of its varieties at various sites in Western Australia ranging from 2.16 to 2.96 tonne/ha. The WA state average for 2003 was 1.53 tonne/ha. Clearly many farmers will be unable to achieve the yields quoted and so comparisons with such trials are meaningless.
In addition to all the other problems of comparing trial results to Australia's average canola yield, there is one additional particular problem for GM canola trials. The regulatory system in Australia has been such that approvals for planting trials in the past half a dozen years have come quite late in the growing season, meaning that many trials are not planted at the optimal time. This greatly reduces the yield potential of the trial. Trials sown at the appropriate time and grown in better conditions will yield better. Therefore, unfortunately there are few trials available from Australia where one could test Julie Newman's claims of lower yields for Roundup Ready canola.
There is one set of trial results published for Australia for Roundup Ready canola. These are in "The Benefits of Roundup Ready canola for Western Australian Farming Systems" from WA Crop Updates (available from
They show yields of a conventional variety, a triazine tolerant variety and a Roundup Ready variety averaged across six trial sites in 4 states. The yields are about 1.56 tonne/ha for conventional
canola, 1.86 tonne/ha for Roundup Ready canola and 1.49 tonne/ha for triazine tolerant canola. No statistical analysis is provided, so it is impossible to determine whether any differences are significant. These trials, where the same varieties were grown on the same site at the same time, clearly do not support Julie Newman's claims.
Other sources of information to test are the data sets on trial results from elsewhere in the world. Both Canada and the US have grown GM canola crops for some years. In both countries variety
trials include both GM and non-GM varieties. The results of the trails are published each year. As these trials are extensive, involve a number of different varieties with different traits, you
can compare within trials to get an idea of whether the claim that Roundup Ready canola crops have built in yield penalties.
In Canada, the trials are now conducted as the Prairie Canola Variety Trials. These have been conducted for the past 2 years and data are available from Provincial Seed Variety publications and from the Canola Council of Canada (www.canola-council.org/growing_trials.html). The trials are split across the three zones with different season lengths on the Canadian Prairies and within each zone the same varieties are grown at each site. In 2003 there were 7 reporting stations in the short season zone, 12 in the mid-season zone and 11 in the long season zone. In 2004 there were 8 reporting stations in the short season zone, 16 in the mid-season zone and 12 in the long season zone. Within years and zones, the same varieties were grown, but these did differ between years and zones. All data is reported as % yield of a single variety 46A65. The varieties are separated by herbicide resistance type:
none (= conventional varieties), Clearfield (resistance to imidazolinone herbicides), Roundup Ready and Liberty Link (resistance to glufosinate). The Clearfield types are herbicide resistant, but non-GM. The distribution of varieties was 6 conventional, 3 Clearfield, 4 Liberty Link and 30 Roundup Ready. In 2004, the distribution was 2 conventional, 11 Clearfield, 4 Liberty Link and 29
A comparison of different herbicide resistant types yields the following results. In 2003, the average of conventional varieties was 104, 105 and 105% respectively of the yield of 46A45 for the short, mid and long-season zones. In contrast, the yield of Clearfield types was 93, 95 and 93% of the yield of 46A45 for the same zones. The average yield of Roundup Ready types was 98, 100 and 101% of the yield of 46A45. Lastly, the yield of Liberty Link types was 115, 127 and 128% of the yield of 46A45 for each of the zones. The LSD for the trials varied from 4% to 21% depending on site. Therefore, one could conclude that the differences between Roundup Ready and conventional are unlikely to be statistically significant.In contrast, the Liberty Link varieties were likely to have significantly higher yield than the conventional, Clearfield or Roundup Ready types in these environments.
Similar effects were observed in 2004. The yield of conventional types (albeit based on only two varieties) was 109, 107 and 106% of the yield of 46A45 for the short-, mid- and long-season zones respectively. The yield of Clearfield types was 99, 96 and 102% of the yield of 46A45 for the same regions. The average yield of Roundup Ready types was 106, 102 and 100% of the yield of 46A45. Lastly, the yield of Liberty Link types was 133, 122 and 120% of the yield of 46A45 for each of the zones. The LSD for the trials varied from 8 to 14% across zones. This means, the differences between the conventional, Clearfield and Roundup Ready types are unlikely to be significant. However, the Liberty Link varieties were likely to have significantly higher yield than the other varieties in these environments.
North Dakota and Minnesota are the two main canola-growing states in the US. Universities in both states conduct canola variety trials at several locations within the states. In North Dakota variety trials are available for several years
(www.ag.ndsu.nodak.edu/aginfo/variety/2005/index.html and links therein). I have looked at the last two years, 2003 and 2004. In each year there were trials at 4 sites. These trials are organised and reported differently to those in Canada. The varieties are sorted and reported by Roundup Ready and "conventional"; however the "conventional" group contains Clearfield and Liberty Link types as well as non-HT types. At most sites, the trials were sown and harvested on the same day and appear to have been grown in the same soil type with the same nutrient analysis. Therefore, I believe comparisons are acceptable. At one site, the Roundup Ready and "conventional" trials were planted on different dates, so I excluded that site from analysis. There were 8 conventional varieties, 48 Roundup Ready varieties, 6 Clearfield and 6 Liberty Link varieties sown in 2003. Except for Liberty Link, these were a mix of open pollinated and hybrid types for all varieties. In 2004, there were 10 conventional, 49 Roundup Ready varieties, 4 Clearfield and 12 Liberty Link varieties sown across the trials. In 2004, these were a mix of open pollinated and hybrid varieties for all types. The varieties sown varied between sites and years and actual yields are reported rather than comparison to a check variety.
The yields of all conventional (non-HT) varieties average 2.62 t/ha at Langdon, 2.98 t/ha at Minot, and 1.53 t/ha at Williston in 2003. Roundup Ready types yielded at 112, 103 and 110% of the average yield of conventional types at Langdon, Minot and Williston respectively.
Clearfield types yielded at 84, 90 and 85% of conventional types at the three sites. Liberty Link types were not grown at Williston, but at the other two sites yielded 127 and 109% of the conventional types. In 2004, the average yield of conventional varieties averaged 2.96 t/ha at Langdon, 1.66 t/ha at Minot and 1.60 t/ha at Williston. Roundup Ready types yielded at 105, 130 and 120% of the average yield of conventional types at Langdon, Minot and Williston respectively.
Clearfield types yielded at 109, 90 and 101% of conventional types at the three sites. Again, Liberty Link types were not grown at Williston, but at the other two sites yielded 122 and 124% of the conventional types. The coefficient of variation across the trials ranged from 7 to 10.7% in 2003 and from 8 to 26.4% in 2004. The high coefficient of variation in 2004 means it is difficult to conclude any type was significantly better that the others without going back and analysing the original data. However, there are consistently high yields for the Liberty Link types at all sites in all years.
In Minnesota, trial data were organised and reported as for North Dakota. I accessed data from 2002 and 2003 as 2004 data was not available from the website
Non-Roundup Ready varieties were not grown at all sites, so those sites were excluded from the analysis. In addition, one site in 2002 was excessively damaged by hail leaving only a small number of non-Roundup Ready varieties, so I excluded that site as well. This left two sites in each of two years for comparison. In 2002 there were 5 conventional varieties, 34 Roundup Ready varieties, 6 Clearfield varieties and 5 Liberty Link varieties. The Roundup Ready and Liberty Link varieties were a mix of open pollinated and hybrids, the conventional varieties were all hybrids and the Clearfield varieties were all open pollinated. In 2003, there were 2 conventional varieties, 23 Roundup Ready varieties, 3 Clearfield varieties and 6 Liberty Link varieties. In 2003 the Roundup Ready and Clearfield varieties were a mix of open pollinated and hybrids, whereas the conventional and Liberty Link varieties were all hybrids.
Average yields of conventional varieties were 1.35 t/ha at Kennedy and 1.70 t/ha at Grygla in 2002. Average yields of Roundup Ready types were 120 and 111% of the average yield of conventional types in the two locations. Average yields of Clearfield types were 86 and
90% of conventional types, whereas average yields of Liberty Link types were 103 and 105% of conventional types. In 2003, average yields of conventional varieties were 2.69 t/ha at Roseau and 2.38 t/ha at Kennedy. Average yields of Roundup Ready types were 107 and 106% of the average yield of conventional types in the two locations.
Average yields of Clearfield types were 102 and 93% of conventional types, whereas average yields of Liberty Link types were 115 and 113% of conventional types. For the Minnesota trials, the coefficient of variation ranged from 9.7 to 13.6% in 2002 and from 4.5 to 6.9% in 2003. Therefore, it is probably unlikely that yields of Roundup Ready or Liberty Link varieties were significantly greater than the conventional varieties in 2002, but Liberty Link may have had significantly greater yield in 2003 - despite all conventional varieties being hybrids. Clearfield varieties likely had lower yield in 2002, but were unlikely to be significantly different in 2003.
In conclusion, from the available variety trials one can conclude there is no detrimental effect on yield of Roundup Ready canola compared to conventional varieties. Liberty Link types generally had higher yields than conventional types, even when all the varieties being compared were hybrids. This effect was obvious in the trials from Canada and North Dakota, where at the vast majority of sites and years the highest yielding variety was a Liberty Link type. Liberty Link types were also always in the top 3 yielding varieties in the Minnesota trials.
For Australia, there is too little data available to come to any definite conclusion about yields of GM versus non-GM varieties. However, the limited trial data that is available generally points to the GM varieties having the same, or possibly higher, yields than comparable non-GM varieties. This means there is no good evidence for a decrease in yield associated with the GM varieties. The claim that "actual trials of GM canola have produced yields of up to 20% less than non-GM varieties" is not generally true of Australia,Canada or the USA.
FIRST PRESTON POST ENDS
Julie Newman Responds to Chris Preston
AgBioView www.agbioworld.org : Nov 1, 2005
- Network of Concerned Farmers, Australia; julie-at-non-gm-farmers.com
Comments: I would like to respond to Chris Prestons article regarding yields of GM canola. The following statement is false: "Julie Newman can confidently assert that "actual trials of GM canola have produced yields of up to 20% less than non-GM varieties" (as cited on the ABC 20th September 2005). "
If Mr Preston cared to read the ABC transcript he would find I stated that GM Invigor canola had 20% less vigour than non-GM hybrids which is a direct quote from the OGTR document. I find it interesting that vigour is given as the reason for higher yields but when used in relation to lower yields than non-GM hybrids, it is disputed. Mr Preston seems to miss the point that the Network of Concerned Farmers support independent contained trials to collect accurate performance data comparing existing common varieties. What we find disappointing is that Bayer Cropscience and Monsanto are refusing to participate. What are they so afraid of?
PRESTON Response to Julie Newman's Response
- Christopher Preston, Australia
AgBioView www.agbioworld.org : November 2, 2005
Re Julie Newman's response. I would like to respond with an assertion that the quotation I attributed to Julie Newman was correctly quoted. In addition, I would like to provide documentary evidence for my case.
My first exhibit is a Press Release from the Network of Concerned Farmers from 18th September 2005 (available from: http://www.non-gm-farmers.com/news_details.asp?ID=2434 ). I quote: "…This is simply not true. GM gives lower yields, higher costs and market risk to a range of commodities and in no way represents a benefit to Australian farmers," said Julie Newman, National Spokesperson for the Network of Concerned Farmers. Mrs Newman explained that the GM debate was centred around GM canola with two different traits. Bayer Cropscience is offering a hybrid canola variety resistant to the chemical glufosinate ammonium and Monsanto is offering a variety that is resistant to glyphosate. "Even Bayer admits their variety yields 20% less than non-GM hybrids….”
Second exhibit, an ABC News report from Mount Gambier 20th September 2005 (available from: http://www.abc.net.au/news/australia/sa/gambier/200509/s1464042.htm). I quote: “…Julie Newman from the Network of Concerned Farmers says ABARE's claim that Australia's failure to adopt GM technology could cost $3 billion is misleading.
Ms Newman says the modelling is based on the assumption that GM crops will yield 10 per cent more, but actual trials of GM canola have produced yields of up to 20 per cent less than non-GM varieties.”
Third exhibit, identical news reports from ABC in Renmark, Alice Springs, Longreach, Bunbury, Bundaberg, Broken Hill, Sunshine Coast, Coffs Harbour, Launceston, Albury/Wadonga, Kimberly, Geralton, Newcastle, Townsville, Lismore and Central Queensland. I could go on, but I think there is ample evidence that Julie Newman made the quotation I ascribed to her and it is even in her own press release!
As to the NCF wanting trials, I think Julie Newman is being a little disingenuous. A quick look through the Network of Concerned Farmers website, which Julie manages, indicates an organisation totally opposed to GM crops. They even have attacks on Bt cotton on their website, which we know uses less pesticides to grow than conventional cotton. If InVigor canola were to be shown in trials to outyield conventional hybrids as it has in Canada, would the Network of Concerned Farmers stop their opposition? I doubt it.
Return to GMO Pundit