info4action archive


GE - Farmscale evaluations,herbicide tolerance

  1st commentary and article from NLP, second from Jean - many thanks, sorry
they have been shortened
1 NLP intro
2 Farmers weekly article
3 brief intro into DETR report
4 DETR report
As the UK moves forward into its field scale trials on GM oilseed rape crops 
a number of points are worth pointing out arising from the Farmers Weekly 
article 15 January 1999 below. This article looks at what is already 
happening in Canada with these crops that have reached the commercial stage.
Given this experience with GM herbicide resistant oilseed rape in Canada the 
question arises as to why we need oilseed rape fieldscale trials in the 
first place in the UK as it is already clear that the crop is going to cause 
severe problems to farmers let alone the environment, including:
1) Herbicide resistant volunteers are going to end up on neighbouring (as 
well as source) farms who will have to widen the range of chemicals they use 
to control them.
2) Some of the chemicals they will be forced to use to do this will be more 
toxic than others they are using at the moment.
3) Farmer litigation is already taking place in Canada arising from fields 
infested with herbicide resistant rape volunteers.
In the story below reported by Farmers Weekly the farmer concerned was 
forced to resort to using 2,4-D to control unexpected GM herbicide tolerant 
volunteers, but was worried about the effects of 2,4-D residues on broad 
leaved crops and was unhappy about handling the more toxic chemical. He had 
been growing three types of canola (oilseed rape) each resistant to a 
different herbicide (glyphosphate, gluphosinate-ammonium and imazethapyr) 
forcing him to switch to a fourth herbicide to control volunteers. The 
problem arose even where a particular herbicide resistant variety had not 
been grown in the field concerned. Pollen brought the culprit volunteers in 
Both AgreEvo and Zeneca have already acknowledged publicly that farmers will 
need to use additional types of chemicals to deal with oilseed rape 
volunteers from herbicide resistant crops. Zeneca in Canada have openly 
promoted a new product (Gramoxone PDQ) to farmers for dealing with 

An update on the Canadian GM oilseed rape volunteer pollution problem was 
recently given in a Financial Times supplement. See .
If you would like to see a photograph of a field of herbicide resistant 
oilseed rape volunteers that have survived spraying with glyphosphate in 
Canada - visit:
First GM tolerance transfer is feared in Canadian OSR
Farmers Weekly 15 January 1999
By Stephen Leahy
A CANADIAN field is thought to 
be the site of the first accidental 
on-farm transfer of herbicide 
tolerance between genetically mod- 
ified and conventional crop plants.
Last summer oilseed rape grow- 
er Tony Huether was surprised to 
see volunteers thriving after two 
glyphosate (Roundup) sprayings 
in a 57ha (140 acre) field where no 
glyphosate-resistant rape had ever 
been sown.
He had planted a glyphosate- 
resistant variety in a nearby field 
in 1997 and believes the trait was 
transferred by pollen movement.
"Cross pollination was going to 
happen. It surprises me it didn't 
show up sooner," says Phil 
Thomas, Alberta Agriculture's 
provincial oilseed specialist.
UK studies have shown pollina- 
tion transfer up to 400m by wind 
or insects can result in out-crosses, 
says Mr Thomas. "Just because 
you're not growing herbicide-resis- 
tant canola [oilseed rape], it does 
not mean you can't get herbicide- 
resistant volunteers."
Mr Huether, who farms 860ha 
(2100 acres) of oilseed rape, peas 
and wheat at Sexsmith, northern 
Alberta, says no one told him GM 
traits could be transferred. 
"Monsanto never made farmers 
aware of the possibility of this hap- 
pening when we signed the TUA 
[Technical Use Agreement]."
Although 60% of Canada's 
oilseed rape is now herbicide-toler- 
ant, 1997 was the first time Mr 
Huether planted GM varieties. 
That spring he put the 57ha field 
into a Smart variety resistant to 
the herbicide imazethapyr and 8ha 
(20 acres) of Innovator, a variety 
tolerant to glufosinate-ammonium 
( Liberty).
Thirty metres away, on the 
other side of a dirt road, he plant- 
ed 16ha (40 acres) of Quest, a 
glyphosate-resistant variety.
Mr Huether says the fields were 
harvested at different times and 
rules out mechanical transfer. 
Before blaming cross pollination, 
government crop expert John 
Huffman investigated other possi- 
bilities. "We have taken this as a 
serious issue."
Researchers, including repre- 
sentatives from Monsanto, have 
since taken plants and seeds. 
Government scientists are growing 
the seeds on in greenhouses and 
doing DNA mapping. A compre- 
hensive report is expected soon.
Both Mr Thomas and Mr 
Huffman say the addition of any 
broadleaf herbicides, such as 2,4- 
D, can easily control unexpected 
GMHT volunteers. Both agree 
that keeping accurate records is 
also important. "It may be a bit of 
a headache to know what your 
neighbour grows and to plan your 
strategy accordingly," acknowl- 
edges Mr Thomas.
Mr Huether is less relaxed. 
"It's not that easy a problem to 
solve when direct seeding and 
planting a wide range of crops." He 
worries about the effects of 2.4-D 
residues on broad leaved crops and 
loathes handling the more toxic 
Farmers are not getting all the 
facts they need on GM crops, he 
The minutes of the last meeting of the Scientific Steering Group overseeing 
farmscale trials have now been posted on the DETR web site along with the 
interim report produced by the scientists involved.
No real surprises. Moans about hysteria in the press etc etc and a snipe at 
Greenpeace and FoE. "We were particular disappointed about statements made by 
Greenpeace claiming that the science was flawed, only days after an open 
seminar in London at which we presented the science case. Greenpeace and 
Friends of the Earth were amongst a packed audience that day and neither 
any defects in project design or methodology." 
The trials ARE flawed.............
The reports about the individual sites are somewhat interesting but lack 
significant detail.  I still think that the timing and frequency of the 
application of pesticides to the GM and non GM crops will make a big 
Department of the Environment, Transport and the Regions
>Genetically modified crop Farm-Scale Evaluations
>Scientific Steering Committee Minutes
>October 11th 1999
>Attending:Members:Professor Chris Pollock (chair) (IGER)
>Dr David Gibbons (RSPB)
>Professor Mick Crawley (Imperial College)
>Dr Nick Sotherton (GCT)
>Dr Nicholas Aebischer (GCT)
>Mr Jim Orson (Morley Research Centre)
>Dr Alastair Burn (EN)
>Assessors:Dr Brian Johnson (EN)
>Mr Trevor Denham (MAFF)
>Dr Emma Hennessey (MAFF)
>Dr Roger Turner (SCIMAC)
>Dr Toby Willison (SERAD)
>Dr Linda Smith (DETR)
>Dr Nick Brickle (DETR)
>Contractors:Dr Les Firbank (ITE)
>Professor Joe Perry (IACR)
>Dr Geoff Squire (SCRI) 
>Dr Peter Rothery (ITE)
>Dr Mark Hill (ITE)
>Dr Alan Dewar (IACR)
>Apologies for absence
>Minutes of last meeting 
>The minutes of the last meeting have already been approved and are publicly
>available on the farm-scale evaluations web-site.
>Contractors report of progress
>The contractors presented an interim report of progress, included as Annex
>Finalisation of sampling protocols
>The steering committee reviewed each protocol in turn. Members were generally
>happy with the protocols, considering them to be scientifically rigorous. A
>number of minor improvements were recommended: 
>Vascular plants
>The committee was generally content, but suggested that transect locations
>should be randomised and that a simple measurement of the standing seed
>crop or
>seed biomass be made in order to compare weed productivity between
>An assessment of standing seed crop (and its state of maturity) could be used
>as a measure of food availability to wildlife that feeds on weed seeds. The
>committee also recommended that a ‘pen-picture’ of weed distribution in each
>field be used to document any obvious patchiness.
>Seed bank
>Some reservations were voiced regarding the power to detect differences in
>seed bank between treatments, however, it was felt that a well designed
>analysis could still yield interesting results. The contractors were urged to
>continue developing the protocol and to consider the correct balance of
>sampling between fields and headlands. The committee agreed to consider this
>issue further at the next meeting.
>The committee considered the invertebrate sampling protocol and suggested
>several refinements. It was agreed that a measure of the abundance of
>invertebrates (including pest species) on crop plants should be incorporated.
>It was also felt that the within-field number of samples need to be more
>clearly justified and the extraction efficiency of vacuum samples to be
>The committee discussed whether pitfall trapping should be stratified within
>fields. Such an analysis was welcomed by the committee and the contractors
>agreed to consider further whether the proposed number of pitfall traps per
>field and their location was appropriate. 
>Discussion between the contractors and relevant committee members should take
>place as soon as possible to decide on any appropriate refinements.
>Bees & butterflies
>Concerns were raised about the low density/total count likely to be recorded.
>However, it was felt that a two tier analysis, using abundance data for
>commoner species and also diversity of all species, would be acceptable. 
>Slugs & Snails
>Some further validation was sought for the accuracy of the trapping method.
>The committee agreed with the contractors that earthworm sampling should
>not be
>undertaken as the agreed experimental design is inappropriate. Thus to
>them would be a poor use of resources.
>Split and paired fields and the number of fields
>Following the last steering committee meeting the contractors carefully
>considered the issues in co-operation with members of the steering
>committee. A
>detailed paper was presented to the committee that is summarised in the
>report (it is anticipated that the detailed paper will be published in a peer
>reviewed journal in due course). 
>On balance, the statistical and logistical arguments favoured adopting a
>field approach and so such an approach was agreed unanimously by the
>The detailed statistical analysis was also used to calculate the most
>appropriate number of fields. The approach was based upon the need to detect
>effects of an order of magnitude reasonably anticipated from previous
>equivalent ecological studies. It was unanimously agreed to seek 25 fields,
>crop, per year, for three years (75 fields per crop total). 
>It was also agreed that the size of fields should seek to reflect commercial
>reality. Fields should not be so large that they are unrepresentative, nor so
>small that edge effects dominate. As a rough guide the size of each half
>be in the region 5-10 ha, although this might be somewhat smaller in the case
>of maize.
>Site selection
>The committee discussed at length how the range of farming practices for a
>particular crop should be reflected in the sample of fields monitored. It was
>agreed that the farms selected needed to be representative of farming in the
>UK. Because those farmers that might readily volunteer to grow GM crops
may be
>biased to particular farming practices, those adopting other practices may
>to be sought proactively.
>The possibility of using some form of a stratified sample was considered
>sub-divided by type/intensity of usual management). It was discussed whether
>such a stratified sample would represent farm types in the proportion they
>occur in the UK, or be deliberately skewed in order, say, to have an equal
>number of each type. The committee advised that further consideration be
>to this issue by the contractors, in contact with relevant committee members,
>with a view to confirming the approach as soon as possible. 
>The committee agreed that it would be necessary to establish the range of
>farming practices for each crop to guide any decision on stratified sampling,
>this information would also be relevant to the interpretation of results.
>The contractors asked to be provided with considerably more potential farms
>than required. They would obtain further details, select farms and a field in
>each farm to fulfil the selection criteria, then randomise the allocation of
>treatments to experimental units, in consultation with the Steering Group.
>Units to be planted with GM crops would then be notified to SCIMAC and
>The committee supported this request and SCIMAC undertook to find potential
>SCIMAC fodder/sugar beet farm-scale evaluations
>The committee reiterated that it was only prepared to oversee this research
>programme if identical terms and conditions applied as to the government
>research, particularly regarding the publication of results. SCIMAC confirmed
>that this would be the case.
>Pesticide Advice
>The committee unanimously agreed the need for openness in the advice given to
>farmers on pesticide use. Members agreed that SCIMAC should be advising as
>little as necessary for the GM crop, and should play no part in advising for
>the non-GM crop. It was agreed that the contractors, DETR and SCIMAC would
>discuss further how a greater degree of openness could be achieved.
>Site security
>The contractors raised concerns about the damage to the research programme
>the concerns for the safety of their own staff as a result of illegal
>and vandalism at field sites.
>The committee agreed that measures to reduce the loss of sites by
>vandalism, or
>to increase the safety of field workers, would be welcome. Opinion differed
>over whether increased secrecy would achieve either of the two objectives.
>Discussion should continue between relevant parties.
>Public Relations
>The contractors raised the issue of public relations; wishing the
committee to
>note the difficulty in getting across accurate information on the research
>programme. The contractors would welcome further support to provide
>to balance the frequently inaccurate information presented in the media. The
>steering committee noted the concern and agreed that a further meeting of the
>relevant parties should discuss the issue.
>Further research 
>The possibility of further research being conducted on the farm-scale
>evaluation sites was discussed. DETR indicated that it is hoping to initiate
>further gene flow monitoring and research studies on birds and mammals. The
>committee continues to welcome further research proposals and provisionally
>agreed the following position:
>· In principle the committee is keen to maximise the value of the farm-scale
>evaluation site; 
>· the committee is prepared to view any proposal for research on the
>evaluations sites in order to determine the level of compatibility with the
>primary ecological research. 
>· In principle the scientific steering committee would be prepared to take a
>deeper interest in other biodiversity research that would add value to the
>farm-scale evaluations research. 
>Further research proposals, from any source, should be directed to DETR
>Interim report
>It was agreed that the interim report submitted by the contractors would be
>published on the internet. More detailed discussion of the protocols
>in an appendix to the report would not be put on the internet immediately, as
>the contractors are seeking to publish the data in refereed journals. The
>committee supported this intention.
>Next meeting
>March 10th 2000
>Published 23 December 1999 
>Department of the Environment, Transport and the Regions
>Genetically modified crop Farm-Scale Evaluations
>Interim Report - November 11th 1999
>Interim Report 
>11 November 1999
>Edited by
>Les Firbank1
>Written by
>John Bater2, Helaina Black1, Gill Champion2, Roger Daniels1, Alan Dewar2,
>Margaret Dixon1, Les Firbank1, Peter Greenslade1, Cathy Hawes3, Matt Heard1,
>Mark Hill1, Peter Lutman2, Martin Maudsley2, Mike May2, Juliet Osborne1, John
>Pidgeon2, Joe Perry2, Peter Rothery1, David Roy1, Rod Scott1, Geoff Squire3,
>John Watkins1, Ian Woiwod2
>Executive summary
>The main purpose of this project is to test the null hypothesis that there
>no significant differences between the biodiversity associated with the
>management of GM winter oilseed rape / spring oil seed rape / maize crops
>are tolerant to herbicides and comparable non-GM crops at the farm scale.
>we report on progress in the period April — September 1999. 
>The approach is to compare indicators of soils, vegetation and invertebrates
>for land used for GM crops with the same indicators in comparable non-GM
>Effects on following crops are also assessed. It is assumed that any
>effects on
>biodiversity will arise from changes in crop and field management, and not
>as a
>direct result of the use of GM technology. Our methodology is the same for
>three crops, as far as possible. The experiment does not address gene
flow, as
>this was assessed separately by the National Institute of Agricultural
>Virtually all of the work so far has been concerned with the development of
>methodology. Field trials and new analyses have been used to ensure that the
>methods are both rigorous and yet practicable. The field trials have included
>maize and spring rape sites, using both split and paired fields. The data
>these trials are proving to be an invaluable source of information for the
>conduct of these trials. Site selection will be on the basis of selecting
>a pool of possible sites selected by SCIMAC. 
>New analyses confirm for adequate replication, planting should proceed at
>25 farms per crop per year. The current weight of evidence suggests that a
>split field approach is likely to be more appropriate to the indicators we
>recording, and simpler to achieve in practice, than a paired field approach. 
>The collection of data on farm management has been trialed. Whilst we make no
>suggestion that the field management has been manipulated in any way, we
>need greater standardisation on the provision of agronomic advice to
>Methods of assessing biodiversity have been developed during the season. We
>confirm that we should focus on species groups that are indicative of longer
>term change, and of change higher up the food chain. 
>In particular, we will monitor:
>Vegetation in and around the field
>Soil seed bank
>Slugs and snails
>Arthropods on vegetation (with particular reference to plant bugs,
>and caterpillars of butterflies, moths and sawflies)
>Carabid beetles and ground-dwelling arthropods
>Bees and butterflies
>The methodologies still need some refinement, notably in terms of finalising
>appropriate sampling intensities and timings.
>The field experiments in 1999 comprised four sites with maize and three with
>spring oil seed rape. Of these sites, one was taken out of the experiment by
>the farmer, and there was extensive damage to two other sites by protestors,
>although we continued to monitor them. 
>The field studies for the winter rape trials have begun. There are two paired
>and one split field study sites. Seed bank sampling has already taken place.
>Work this autumn will concentrate on finalising sampling and data management
>protocols. We also need to plan the spring 2000 plantings as soon as
>The research project is of great public interest, and while we strongly
>wish to
>be pro-active and positive about our work within the public debate, this is
>difficult given that our priority must remain firmly focussed on the
>conduct of
>the experiment itself.
>1The Institute of Terrestrial Ecology
>2The Institute of Arable Crops Research
>3The Scottish Crop Research Institute
>Published 23 December 1999 
>Department of the Environment, Transport and the Regions
>Genetically modified crop Farm-Scale Evaluations
>Interim Report - November 11th 1999
>1.1 Objective
>A range of crops has been genetically modified for tolerance to
>herbicides. The proposed benefit of this technology is that it will reduce
>costs and the amount of herbicide applied, as weed management for the farmer
>will be easier. However, this same benefit has an important possible side
>effect, that weed control becomes sufficiently efficient to threaten farmland
>species that have already been in decline because of changes in farming
>practice. The purpose of this project is to assess the impacts of
>genetically-modified herbicide tolerant (GMHT) crops on biodiversity through
>the effects of the crop management. The research is also intended to
>to an assessment of the wider question of whether the commercial use of GM
>crops will change the management of farming systems and the agricultural
>landscape. The crops under the study funded by DETR / MAFF / SE are maize,
>winter rape and spring rape. 
>The main purpose is to be achieved by testing the null hypothesis that
>there are
>no significant differences between the biodiversity associated with the
>management of GM winter oilseed rape / spring oil seed rape / maize crops
>are tolerant to herbicides and comparable non-GM crops at the farm scale.
>Here we report on progress from the period April — September 1999.
>1.2 Mission statement
>Within the project management team, we are working to the following mission:
>To test the null hypotheses of no effects of GMHT maize, spring rape and
>rape on biodiversity
>To conduct high quality science on the biodiversity of agricultural systems,
>while testing the null hypotheses
>To ensure that deliverables (reports to the Steering Committee, database and
>scientific papers) are timely and to appropriate quality
>To ensure impartiality, quality and transparency of the project
>To ensure the health, safety and security of the staff involved in the
>To ensure the effective presentation of the project to participants and the
>wider community
>2 Testing the null hypothesis 
>2.1 Overall approach
>The approach is to compare indicators of soils, vegetation and
>invertebrates on
>land used for GM crops with the same indicators for comparable non-GM crops.
>Effects on following crops grown on the land in subsequent seasons are also
>assessed. It is assumed that any effects on biodiversity will arise from
>changes in crop and field management, and not as a direct result of the
use of
>GM technology. Indeed, our approach would be identical for herbicide-tolerant
>crops resulting from more conventional breeding methods. Our methodology is
>same for all crops, as far as possible. The experiment does not address gene
>flow, as this was assessed separately by NIAB. 
>The basic approach is to measure differences in biodiversity between an area
>with a GM crop and one with a comparable non-GM crop, placed close
together on
>the same site. Sites are chosen to represent fully the range of variation in
>soil, climate, species occurrence and farm management that is likely to be
>found in the commercial growing of GM crops. The pairing of the GM and non-GM
>areas within each site ensures this variation is accounted for within the
>analyses, and does not affect the comparison between the crops. The form of
>comparison between the crops is direct, and so requires no information on
>changes in biodiversity before and after sowing; there is therefore little
>for baseline data. Within-site variation, that might cause a difference
>the pair of areas in their potential for biodiversity, will be allowed for by
>measurements of the soil seed bank and surrounding habitats.
>Both GM and non-GM farming systems are managed according to current
>practice, although within this constraint, management practices will be kept
>constant. The non-GM variety is matched to the GM variety in its
>development as
>far as possible. Thus the design looks at differences in biodiversity between
>crops, but also between following crops, including any due to differences in
>the rotation, provided that these differences are a consequence of the two
>varieties of crops being used.
>The study involves recording both how the crops are managed, and the
>effects of
>the crops and their management on a range of indicators of biodiversity. We
>concentrate on those animal and plant indicators that are likely to be
>sensitive to changes in farming system, that are indicative of other groups
>directly assessed, and can be considered to be indicative of slower and
>larger-scale processes not readily detected within the experiment directly. 
>A study undertaken as part of the Tender suggested that an appropriate
>level of
>replication would be achieved by planting at about 25 pairs of sites per crop
>per year, in years 2-4, with a smaller number in the first year to develop
>methodologies. The analysis will consider each biodiversity indicator in
>and also examine relationships between indicators, taking into account data
>from other relevant studies. 
>2.2 Developing the methodology 
>Virtually all of the work so far has been concerned with the development of
>methodology. Field trials and new analyses have been used to gain experience
>about how to test the null hypotheses using methods that are both rigorous,
>practicable. The field trials have been on maize and spring rape sites, using
>both split and paired field configurations. The data are proving to be an
>invaluable source of information for the conduct of these trials. 
>2.3 Farm and field selection 
>The sites for the spring planting were selected through agreement between
>individual farmers and AgrEvo, the suppliers of the seed, operating on behalf
>of SCIMAC. This process will continue. However, in itself, it does not
>guarantee that that the range of farms will be representative of those farms
>most likely to grow the crops under commercial conditions. Therefore, we wish
>to be provided by SCIMAC with a very wide choice of potential sites, from
>we would select samples to ensure representative ranges of size, intensity of
>management, geography, and so on. We will collect data about the national
>context of the farm from existing national databases, and data about the more
>immediate context of the fields, their margins, boundaries and surrounding
>countryside through mapping on site; these may all be covariates in the
>analyses. We will allocate GM and non-GM treatments to units at random
from an
>appropriate pair of units. The Consortium would therefore take responsibility
>for site and treatment selection, within the range of options provided by
>2.4 Consideration of sample size 
>The importance of determining sample size was emphasised by the Steering
>Committee at its first meeting. It has to be large enough to be able to
>the desired levels of change, but too large and resources are being
wasted. We
>therefore established a working group between statisticians on the project
>and on the Steering Committee. The method was to use simulated data to
>the power of different sample sizes to reveal genuine ecological differences
>between the GM and non-GM treatments. Broadly speaking, the model included
>between-farm differences, allowed for low counts and incorporated dependence
>between variance of counts and mean density. Our interpretation is that the
>results will broadly corroborate the previously calculated powers, suggesting
>that around 25 sites per crop should be planted per year, including some
>provision for wastage of sites as a result of farmers withdrawing from the
>experiment, or damage from activists.
>2.5 Consideration of split versus paired fields
>The design proposed in our Tender was a randomised block design with two
>treatments per block and approximately 25 blocks lasting one year (plus
>up recording); this design to be repeated, with fresh randomisation, in
>years. The experimental unit proposed in the Tender was a field of about 10
>The blocks would be represented by whole farms, and the within-block units by
>whole fields. However, the Steering Committee considered the possibility that
>inter-field variation may be too great, and that the two halves of the same
>fields should be used as the experimental units. In other words, that split
>fields may be preferable to paired fields. The overall design remains
>unchanged; the issue is one of size and location of the experimental units. 
>The Steering Committee and the Consortium both recognise that there are
>arguments in favour of both the paired and split field approaches. The split
>fields may control variability due to previous management and cropping
>but the larger field size requirement may introduce bias in farm selection to
>larger units, and there may be interference between treatments. 
>This issue was also considered by the working group of statisticians. The
>approach was to analyse vegetation data from some of our spring 1999 sites,
>along with existing data similar to those we will be collecting. 
>Data from the MAFF LINK project were considered, but proved unsuitable,
>effects could not be estimated independently of confounding year x treatment
>interactions, and also because there were very few instances of rotations
>included the crops with which we are concerned. Data from Loddington,
>by The Game Conservancy Trust, were also not suitable for the analysis of
>weeds, but were more suitable for invertebrates. These data show that within
>split-fields we might expect similar levels of variability to those that we
>have assumed in our power calculations. Also, that while for many fields the
>variability between paired fields would not be much greater, occasionally the
>difference could really be very great, and the paired-field would then be
>less efficient. However, in the Loddington data, fields had not been matched,
>and that might have inflated between-field variability. In addition, we would
>expect less variability in our trials because our within-unit sampling effort
>will be much greater. The weed data from our own sites give little additional
>information regarding the choice of design, because of the small number of
>sites available. The subcommittee needs to consider these results.
However, we
>suspect that they shed little new light on the debate, as the results were
>as expected on the basis of existing knowledge.
>Other new information has come to light, however. Firstly, the more detailed
>consideration of the protocols for more mobile species has clarified that for
>these we will be considering the choice of use of the crops, rather than
>to estimate population sizes. A split field approach is actually more
>for butterflies and bees. Secondly, experience with farmers has shown that
>can rarely offer well-matched fields in practice, that there can be
>problems in
>randomising which field receives which treatment, and that different fields
>diverge in management to too great an extent in subsequent years. The
>statistics working group will consider this issue, and report to the Steering
>Committee. The Consortium itself has come to a view that we now recommend
>fields, with the caution that we must ensure that the sample units are
>so large that they exclude certain types of farm, nor so small that edge
>effects begin to predominate.
>2.6 Farm management 
>The project requires an assessment of how GMHT cropping will affect the
>management of farming systems and the agricultural landscape. We will address
>this requirement by allowing the GMHT and non-GMHT cropping systems to be
>managed along appropriate commercial lines. Individual farms will provide
>the GM and the paired non-GM sites.
>This means that the GMHT cropping systems will be managed under SCIMAC
>guidelines. Changes to the rotations, or to field margin management, will
>therefore be allowed. The only constraints we wish to impose are that the two
>varieties being considered are as similar as possible in traits other than
>herbicide tolerance, especially phenology and crop structure, and that where
>non-herbicide treatments are imposed on both GMHT and the local control, they
>should be applied at the same time. 
>The Steering Committee has recognised the clear need for robust validation of
>the range and nature of the herbicide and other management regimes used by
>farmers in the farm-scale evaluations. We have been piloting a protocol for
>collecting farm management records for the period prior to, during and after
>the test cropping year. The information will include the type of farm and its
>management, prior cropping of the trial fields, historical records of inputs
>for relevant crops, actual inputs and actions on the monitored fields and
>pesticide inputs and management changes resulting from the growing of the
>crops. The data will be used as covariates in analyses, to classify the farm
>management type (eg high input, ICM etc) and to identify any major changes
>previous years. At the end of the season, if a split field design is used,
>farmer will be asked whether, in a commercial situation, growing the GM crop
>might have led to a change in subsequent management of the different
halves of
>the field. These data will form the basis of any study on the
>representativeness of pesticide practices.
>We certainly found no evidence that farmers or their advisors were modifying
>advice and management to influence the results. However, the issue of
>appropriate sources of agronomic advice to the farmers is still less than
>perfectly resolved, as we do not yet have systems for ensuring that advice to
>farmers comes from a BASIS-trained advisor who is independent from SCIMAC
>working within their guidelines. 
>Crop growth stage is assessed at every survey visit. We do not currently
>undertake estimates of yield, other than by asking farmers for their own
>2.7 Indicators of biodiversity
>Biodiversity is defined as the range of variation in living creatures, at
>genetic, population and community scales, quantified using data on the
>abundance of members of each type collected at the appropriate scale. In
>practice, such variation cannot be recorded for all species in a system,
>and so
>simplifying assumptions must be made. The level of recording is vastly
>simplified if genetic variation is ignored — justifiable in this study.
>Secondly, the level of recording is simplified further if indicators of
>biodiversity are selected to represent particular species groups.
>indicators are required for processes that may lead to significant ecological
>shifts that cannot be detected directly within the limited time and spatial
>scales for the project. These ideas are recognised within the instructions to
>tender; priority is to be given to species groups that do not forage over
>areas or occupy higher trophic levels, concentrating on higher plants and
>invertebrates. Amphibia, birds and mammals will not be monitored directly,
>inferences will be made on possible effects on these groups.
>Our fundamental approach is to contrast indicators of biodiversity between
>and non-GMHT cropping systems. However, while reporting these effects, some
>be placed into the context of national recording schemes (notably plants and
>butterflies) that can help show the relationships between the biodiversity
>associated with the study sites and arable areas in general. Secondly, some
>data can feed directly into the BAP process, especially the presence of
>BAP-listed species.
>2.7.1 The choice of indicators 
>The plans for the project in the Tender addressed a range of plant and
>invertebrate groups. The Steering Committee had considered the plant and
>invertebrate monitoring protocols presented by the contractors to be
>sound. It was felt that the inclusion of a protocol for sampling Collembola
>(springtails), weed seed return and possibly additional soil organisms
>(including micro-organisms) should be seriously considered. Collembola at
>may be included using existing sampling methods.
>Protocols for virtually all of the proposed indicators have been tested in
>field, and the points raised by the Committee have been carefully considered.
>As a result some changes have been made, in terms of both the choice of
>indicators and the methods of field recording that are being used. We have
>tended to oversample, on the grounds that it will then be easier to establish
>the most efficient sampling regimes. This task is well on the way to
>for most protocols, but there still remains work to integrate the needs of
>individual protocols across the whole study. 
>Soil seed bank
>The actual weed flora present at any time is typically only 1% to 10% of the
>total seedbank, whose abundance ranges between 1,000 and 1,000,000 seeds in a
>square metre of field down to plough depth. The species that germinate from
>seedbank are the ones most stimulated by the environment during seedbed
>preparation and least harmed by current weed management. Seedbanks in arable
>land are therefore a more reliable indicator of potential diversity and
>weed problems than are the weeds present in any season. While the
>use of herbicide in GM and non-GM crops is likely to affect the seed cycle,
>crops such as oilseed rape (particularly) shed seed that can become part of
>seedbank and therefore alter its composition.
>The purpose of seedbank studies in these trials is -
>to detect differences caused by GM and non-GM treatments in the composition
>abundance of buried seed;
>to estimate the likely implications of any effects for the future biomass and
>habitat in the weed flora;
>to make a rapid assessment of the background seedbank population of a field,
>and thereby to identify sites that are rich or poor in plant diversity. 
>The final point above partly addresses one of the concerns of the Steering
>Committee on the need to include weedy sites that contribute
>to the total arable diversity: given early knowledge of such diverse
sites, it
>will be possible to direct attention to them as required. A further point
>raised by the Steering Committee was the need for measurements of seed rain,
>rather than, or in addition to, the seedbank. It was decided that (a)
>sufficient background was available on seedbanks to enable us to place the
>measurements in the trials in a wider context and (b) the measure of weed
>biomass just before harvest would give an appropriate indicator of seed
>(see below). 
>A general methodology exists for seedbank studies. The purpose of preliminary
>studies in the present experiment is to refine logistics and determine the
>number and location of measurements needed to define change in broad but
>functional attributes of the seedbank. Seedbank records have been collected
>from fields of spring oilseed rape, maize and winter oilseed rape. These,
>together with existing data on the UK seedbank, will be analysed in order to
>confirm a set of criteria that will adequately define the seedbank in these
>trials; define the minimum number and optimum location of samples that will
>reveal statistically significant effects through these criteria; and estimate
>the best time for re-sampling to observe long-term effects.
>Vegetation in the crop is sampled on three occasions in the first season —
>before spraying, after both GM and conventional crop have received herbicide
>and the weeds have wilted, and at crop maturity. The first two samples are
>counts of seedlings and young plants in a 0.5-m square, identified to
>The third sample is a biomass sample, with removal of plants in a 1-m square,
>sorting, identification, drying and weighing. These data will be used to
>provide estimates of seed return using existing experimental data (and new
>if required). In the season following harvest of the GM crop, seedlings
>will be
>counted again in the same locations. If significant differences are detected,
>seedlings will be counted in the following season. Vegetation sampling in the
>crop is based on a design
>2 treatments x
>12 transects from the crop edge x
>5 quadrats per transect (reduced to 2 relatively large quadrats per transect
>for biomass sampling).
>Vegetation sampling on the field margin is designed to fit in with
sampling of
>invertebrates. The field margin is defined as the non-cropped area between
>field boundary and the crop. All species present in a 10 x 1 m quadrat are
>recorded, together with an estimate of cover. The field boundary is
defined as
>a physical feature which is an interface between the margin and other land
>cover types (e.g. hedge, ditch). Where no such feature exists, a 10 x 1 m
>quadrat is taken in the adjacent land cover (e.g. cereal field). Margins and
>boundaries will be recorded on three occasions during the summer. Observers
>asked to comment on whether there are signs of plant necrosis due to spray
>drift and to record the species in flower. The following design will be used.
>2 treatments x
>3 positions on the edge of the field x
>2 positions relative to the crop (margin and boundary as defined above).
>Terrestrial Gastropods (Slugs and Snails) 
>Slugs and snails are generalist herbivores, which are known to be
>influenced by
>and to exert influence on the composition of plant communities. Slugs are
>to be affected by a range of farming practices. Several species of snails of
>conservation interest inhabit field margins and could be affected as a
>consequence of any changes in field margin plant communities resulting from
>herbicide drift. Because slugs and snails are relatively immobile, any
>at a field scale resulting from the growing of GM crops will not be
>obscured by
>movement of individuals, as with flying insects. 
>Monitoring will be done at each site once in spring and once in autumn each
>year. Sampling for slugs and snails within the crop and in field margins will
>be with unbaited refuge traps. In addition snails will be sampled in field
>margins using timed searches. All individuals will be identified to species.
>Arthropods on vegetation
>Insects and other arthropods are often the primary consumers of the aerial
>parts of crops and other plants within the field and the field margin. They
>will therefore reflect the ecological impact that GM crops may have on the
>wider environment. Associated with the primary consumers will be other, more
>mobile, arthropods that feed on them, which are in turn fed upon, either by
>larger insects or birds. However, the mobility of the organisms at the top of
>the food chain makes it difficult to assess any wider impact that GM crops
>have on their abundance, even at the field scale, hence the need to use
>indicator species at lower trophic levels.
>We have concluded that the most useful indicator groups of invertebrates for
>our purposes are plant bugs (Heteroptera), springtails (Collembola) and the
>caterpillars of butterflies, moths and sawflies. These groups will be
>identified to species where possible. All invertebrates will be collected
>each field or split-field using Vortis suction samplers in the field margin,
>headland and within the crop, placed adjacent to the vegetation sample
>locations. Other arthropods will also be extracted from the samples and
>into groups such as spiders, flies, beetles, wasps etc. Total invertebrate
>biomass will be estimated by weighing sub-samples and scaling up. Samples
>be collected on three or four occasions in spring and summer in each crop,
>depending on crop growth stage and harvest time.
>Carabid beetles and ground-dwelling arthropods
>Carabid beetles are an important component in agricultural ecosystems. Their
>value in ecological studies and as indicators in assessing environmental
>changes, especially in agricultural ecosystems, has been well documented
>and is
>currently an important component in the UK Environmental Change Network
>Programme. Their value as a food source, especially for farmland birds, has
>been studied by both the BTO and RSPB.
>The aim of this protocol is to assess the effects of GM crops and their
>management on numbers and species diversity of carabid beetles. Assessment
>be made by trapping beetles using standard pitfall traps set at predetermined
>points along transects within the crop. These same transects will be used to
>assess vegetation within the crop, the vegetation providing valuable cover
>food for carabids. Counts of individuals and their identification to species
>level will be made. Two other groups important as bird food items will be
>counted, namely spiders (Aranae) and weevils (Curculionidae).
>Bees and butterflies
>Butterflies and bees are valued biodiversity components of agricultural
>ecosystems and they are sensitive to the weed flora of crops and field
>Bees and butterflies are of conservation importance but also, through
>pollination, they play an essential role in the maintenance of plant
>biodiversity. The abundance and diversity of butterflies and foraging bees
>scored using the standard method of "transect" walks along the field margins
>and in the crops themselves, to test for differences between treatments.
>However, a transect walk in the tall maize crop (up to 3 m) proved impossible
>so timed observations over a fixed area will be performed instead.
>Earthworms have been widely used as indicators of change in the farm
>environment where they have a fundamental role in maintaining soil
>structure and
>nutrient cycling and an important constituent of vertebrate and invertebrate
>diets. Hence, they were considered a relevant group to monitor to address the
>null hypotheses in this project. However, the Steering Committee expressed
>doubts as to their suitability. 
>An earthworm protocol was developed and methods were tested across four
>in May 1999. This process highlighted serious problems regarding earthworm
>sampling within this project. The sampling intensity and timing required to
>detect changes in earthworm abundance, biomass or diversity is not manageable
>within this project. There were three main reasons for this:
>Appropriate earthworm sampling would not be possible for monitoring in winter
>oil seed rape and spring oil seed rape. These crops are either harvested or
>sown during the earthworms’ inactive period so preventing a complete crop
>assessment of any impacts. 
>The field sampling required to obtain reasonable estimates of earthworm
>abundance, biomass and diversity (>5 samples at pre-sow and pre-harvest) in
>maize is extremely resource-intensive for the quality of information that
>Changes in earthworm abundance, biomass and diversity are likely to occur
>the medium to long term i.e. over a time-scale greater than a single crop
>cycle. It is recognised that other invertebrate groups e.g. collembola are
>likely to show responses within an annual crop cycle. This has been taken on
>within the arthropod protocol. 
>We therefore consider that earthworms should be removed from the current
>project. This does not mean they are unimportant, rather that there are more
>suitable experimental approaches which would address the issue of whether
>management of GM-modified herbicide resistant crops affects earthworm
>and soil functioning in general.
>2.7.2 Further work required on field sampling protocols 
>The protocols have evolved throughout summer 1999, and there remain issues to
>be resolved before spring 2000. The first is to arrive at optimum sampling
>intensities. This will be undertaken using 1999 field data, by considering
>coefficient of variation for each biodiversity indicator that we have
>This has been done for vegetation, confirming the methods in use.
>protocols will also be considered in this way in the coming months. The
>issue is to test the revised protocols whereby many indicators are sampled in
>close spatial proximity. 
>2.8 Managing the visits to sites
>Managing the whole process of field surveys has proved complex, because of
>need to fit in with the needs of the individual farmer (including site
>considerations), the timetable of availability of survey staff, and the
>shifting phenologies of crops, management and species according to
weather. We
>have established protocols for the co-ordination of site visits through the
>crop co-ordinator and the farmer, and while some changes are required, the
>essential approach is sound.
>Crop co-ordinators are involved with sites from their adoption as a test
>location. They meet with farmers and liaise on the allocation of GM — non-GM
>treatment to units. The crop co-ordinators act as the focus for links between
>farmers and the sampling teams from the consortium, ensuring growers are
>of impending visits. Farmers notify the crop co-ordinators when they
intend to
>spray so that teams visiting the site can be warned and precautions taken as
>necessary. Crop co-ordinators are responsible for ensuring that all sampling
>protocols are carried out on the test sites and that all farm management data
>is collected. Following site visits they collect the data, check and code it
>before sending it for data entry and archiving.
>We still plan to use field teams from local sites to undertake the field
>rather than specialised teams for each protocol (although such teams have
>used this summer, for the purposes of protocol development).
>2.9 Management of the data
>The transfer of data from field survey and eventual analysis will be managed
>via a system of standard data collection forms, security safeguards and data
>validation checks. The data managers have been involved at all stages of the
>systems development to ensure that the resulting data set fulfil data
>and data analysis requirements. The data created from the surveys, along with
>any supporting data, will be integrated in a central database via the data
>entry and validation process. The database is implemented from a relational
>design that specifies rules for referential integrity and types checking.
>the design is implemented these rules are embedded into the structure of the
>database. This means that all data must conform to the validation rules to
>reside in the database. All analyses that contribute to the final results
>only take place using data that have been released from this central,
>2.10 Data analysis 
>The basic model is one that tests the null hypothesis of no effects of the
>treatment on each indicator individually. Broadly speaking, the model
>between-farm differences, allows for low counts and incorporates dependence
>between variance of counts and mean density. It is similar to but more
>realistic than the model given in the tender document, and has been tested
>using simulated data as part of the re-assessment of power calculations. We
>have a novel method for analysis that should be more efficient than previous
>techniques. Other methods and analyses will be adopted as the project
>2.11 A review of the field studies, April — September 1999
>2.11.1 Objectives of summer 1999 work
>As already stated, the summer field studies of 1999 were intended to provide
>information and experience to enable us to develop and test the protocols for
>the study. 
>The field studies for maize and spring rape comprised farms growing GM crops
>and comparable non-GM crops. The sites were selected by SCIMAC, and the
>allocation of GM crop treatment was in consultation with the farmer. 
>This was a paired field site, sown on 4 April 1999. The weed flora was
>on 7 May only. Shortly after this, the farm’s trustees ordered that the GM
>field was to be sprayed with a herbicid and the trial ended.
>This split-field site was sown on 29 April 1999. The Liberty herbicide was
>applied to the GM half (10 ha) of the field about one month after sowing. No
>pre-emergence herbicide was given to the non-GM half (also 10 ha) at sowing,
>but herbicide-resistant blackgrass in the conventional crop was treated by
>herbicide applications. The site was invaded during a demonstration and
>received substantive damage (see below). However, the sequence of
>continued as planned. The GM and non-GM crops had a similar rate of
>development. At flowering, for instance, the GM was two or three days later
>than the conventional variety (Hyola 401). Desiccant was applied to the
>crop in
>late August and the field was harvested on 2 September, 125 days after
>This was also a split-field site, sown on 28 April. Again the GM variety (8.2
>ha) was slightly later than the conventional (Hyola 401, 9.0 ha). The Liberty
>herbicide was applied to the GM half of the field six weeks after sowing,
>than in Oxfordshire because of unsuitable weather A pre-emergence herbicide
>applied to the soil of the non-GM half at sowing; additional local spraying
>carried out to remove patches of thistles and grass. The crop was swathed in
>late August, then harvested on 1 September.
>A split field site, sown 12th May 1999. The field is approximately 6.4ha in
>area, divided equally between GM and conventional Maize. The eastern
margin of
>the field is the river Ver. Because of this, approximately 4 m into the GM
>from this edge could not be sprayed. A broad-spectrum Glyphosate herbicide
>applied 11 days prior to sowing. The non-GM crop was sprayed with Atrazine
>herbicide about one month after sowing, Liberty was applied to the GM half of
>the field a week later. The crop was harvested on 22nd (GM) and 23rd (non-GM)
>of September. The field has now been planted with barley.
>A split field site, sown 10th May 1999. The field is approximately 19.9 ha in
>area, with approximately 5 ha of GM crop. The non-GM part of the field was
>treated with Atrazine herbicide in early May prior to weed emergence. The GM
>crop was sprayed almost a month after sowing and again two weeks later. The
>crop is due to be harvested in early October and planted with winter wheat.
>A paired field site, the earliest sown on 7th May 1999. The GM field is
2.8 ha
>and the conventional is 2.4 ha. Both fields have been reclaimed from gravel
>pits and covered with topsoil. The non-GM field was sprayed with Atrazine in
>early June. The GM field was sprayed with a first application of Liberty
>approximately a week later than the non-GM (a month after sowing), the second
>application was at the end of June. This site was damaged by protestors on
>July and received substantial damage (see assessment below). However,
>from this site continued as planned. The crop is due to be harvested in early
>October and probably seeded with grass.
>A paired field site, the latest sown on 18th May. This is the largest trial
>with approximately 9 ha GM and conventional fields. The GM field has a small
>copse within it. The non-GM field was sprayed with Atrazine herbicide at the
>end of May, the GM field was sprayed with Liberty a month later. The crop is
>due to be harvested in early October and planted with wheat.
>2.11.2 Work undertaken and still outstanding 
>The field studies themselves have been completed, except for follow-up
>of autumn vegetation. However, there remains work on the sorting and
>identification of samples, expected to be complete by November. The data from
>these will be used in three ways:
>to test the actual process of management of samples; for example, for some
>groups, the proposed method of freezing the samples simply does not work; 
>to help finalise issues of timing of sampling during the season 
>to undertake analyses of coefficients of variation to provide data on sample
>to rehearse the process of data management and the construction of the
>2.11.3 The effects of site loss and damage 
>The field studies have been conducted under the glare of publicity, and have
>suffered from site loss and damage.
>The only site that was actually lost was near Swindon, where a paired field
>trial was withdrawn because of concerns over the organic status of the farm.
>The Oxfordshire site suffered from several unauthorised incursions, not least
>from the media and from the National Pollen Research Unit at University
>Worcester. Actual damage occurred at two sites, at a split-field spring rape
>farm in Oxford, and a paired maize site in Norfolk. In both cases, the damage
>was extensive, but monitoring continued on undamaged areas. There was also
>damage to a field in Lincoln, but the demonstrators damaged a field that was
>not part of the experiment. 
>The effects on the science undertaken this summer appear to have been
>not least because data collected this summer are essentially for protocol
>development. In neither case was damage so extensive that monitoring had to
>stop. However, it should be noted that the damage to the spring rape site has
>resulted in a much larger population of volunteers than would have been
>expected, changing the post-harvest management regime.
>The summer has raised issues of the security of sites, farmers and the staff.
>Site security is not the responsibility of the science team, but we are
>concerned with the loss of valuable data and effort. We are more concerned at
>the effects on farmers, who are putting a great deal of time and effort into
>the research programme for what must seem little reward. Indeed, the most
>likely damaging effect of this summer’s protests is to reduce the number of
>farmers willing to take part in the study. Opinion within the consortium is
>split on the issue of secrecy vs openness of the sites.
>2.11.4 The winter oil seed rape plantings, 1999-2000 
>The next phase of the field survey is a pilot study of winter oil seed rape.
>The emphasis of this study is on finalising field protocols (notably those
>have evolved during the summer) and on the timing of surveys in the winter
>crop. Three sites were selected and seed bank sampling took place at all of
>these before the crop was sown.
>2.12 Overall recommendations for autumn / winter 1999 — 2000 
>The focus of the work of the project for the next few months is to complete
>development of protocols and systems that will be put into practice next
>Planning for the spring 2000 field work is already starting; for this to take
>place efficiently, it is imperative that we are provided with the list of
>possible sites to choose from as soon as possible.
>3. Scientific outputs and collaboration
>The Consortium has strongly welcomed the decision of the Steering Committee
>that results should be peer-reviewed by scientific journals. A letter from
>Consortium has been published by Nature that outlines the trials, and other
>papers on the methodology are in draft or are being planned. We have been in
>positive discussions with the editor of a major Journal for publishing the
>results of the project in a special issue.
>Members of staff are also involved in the BRIGHT project, and contacts are
>being established with large scale GM crop studies overseas. NIAB have also
>been conducting studies of gene flow of the farm-scale evaluation sites with
>our full co-operation.
>4. Consortium management 
>Most of the key staff required for the project are now in place. The major
>roles are:
>Project co-ordinator — the primary point of contact with the Steering
>and DETR, responsible for the integration of the study
>Project administrator — responsible for maintaining the developing
>of the project
>Crop managers — the scientists responsible for each of the contracts — one
>Crop co-ordinators — scientists dedicated to the project, responsible for its
>day-to-day conduct, especially of the field work, and for the delivery of
>validated data sets to the data managers, and for undertaking many of the
>analyses; again, one per crop
>Data managers - responsible for the integrity of the data and for guidance on
>overall quality assurance within the project
>Statisticians — responsible for experimental design and for supervising the
>eventual analyses
>Protocol leaders — responsible for the quality of the protocols and to
>assist in
>the interpretation of the results
>Surveyors, survey managers — responsible for data collection
>5. Health, safety and security 
>All staff are bound to the health and safety policies of their employers. In
>addition, all protocols address safety issues to guide risk assessments. In
>general, there are no risks over above those typically encountered on
>agri-environmental research on lowland farms. There remains a potential
>security issue for our staff, but no problems have been encountered so far. 
>6. Public Perception of the Project
>This research programme has become one of the most controversial ever
>in the UK. We have frequently, willingly and openly been involved in local
>meetings and worked with the media. However, over the summer, the scientific
>issues behind the research has been obscured by misinformed and
>disproportionately heavy media coverage of the risks associated with the
>trials. The point has now been reached where public understanding of the
>project has been compromised. 
>We take part in frequent interviews and briefings that do not lead to
>statements about the scientific issues. We are concerned that we have little
>chance to defend attacks on the integrity and quality of our science. We were
>particular disappointed about statements made by Greenpeace claiming that the
>science was flawed, only days after an open seminar in London at which we
>presented the science case. Greenpeace and Friends of the Earth were
amongst a
>packed audience that day and neither voiced any defects in project design or
>These farm scale evaluations have had the result of raising issues that go
>beyond the environmental safety of GM crops. Such issues concern food safety,
>countryside quality and the relationships between consumers, corporations and
>Government. We strongly wish to be pro-active and positive about our work
>within this public debate, but this is difficult given that our priority must
>be to remain firmly focussed on the conduct of the experiment itself. We are
>concerned that we have been given inadequate resources to present the science
>behind the trials at the level that the public expects, especially at the
>scale for the large number of sites to be used next year. 
>7. Acknowledgements
>We are hugely grateful to the farmers involved in the study, for all of the
>time and effort that have contributed, both to the management of the crops,
>to the dialogue with both pressure groups and members of the general
>public. We
>also much appreciate the work of Paul Rylott and Judith Jordan from AgrEvo in
>finding the maize and rape sites.
>8. Major outputs to date
>Firbank, L.G., Dewar, A.M., Hill, M.O., May, M.J., Perry, J.N., Rothery, P.,
>Squire, G. & Woiwod, I.P. (1999). Farm scale evaluation of GM crops
>Nature, 399, 727-728.