2.2 GENE FLOW BETWEEN NEIGHBOURING CROPS

Trials of transgenic crops prior to the PGS application had (almost) all included the option of containment or isolation to prevent cross-pollination of neighbouring crops. The prospect of large-scale release removes this option. We have looked at the 1994/95 risk assessment specifically to see how this problem was addressed and whether the implications were explicitly recognised. Again, it is difficult to separate this issue completely from that of the spread of genes for herbicide-tolerance.

2.2.1 The PGS application

Consistent with its assertion that the biology of the hybrid oilseed rape is in almost every respect identical to conventional varieties, PGS did not acknowledge that gene flow between neighbouring crops may increase or may present a novel problem. From a review of the literature and from data obtained during the Biotechnology Action Program (BAP) (1985-90) and the Biotechnology Research for Innovation, Development and Growth in Europe (BRIDGE) Project (1990-93), the PGS application states variously that

"The restoration line and the restored hybrid do not differ in their pollination capacity and pollen acceptor capacity from the control plants."

"Fertile oilseed rape plants are essentially self-pollinated."

"The bulk of cross-pollination occurs over very short distances."

"Successful pollination declines exponentially with increasing distance between the pollen source and the nearest recipient plant."

"Oilseed rape pollen might occasionally traverse several hundred meters."

In discussing ways in which gene flow by cross-pollination can be reduced, or prevented, the application quotes isolation distances required for the production of pure seed stocks in Canada and Europe (as 400m and 1km, respectively) and refers to the effective use in earlier trials of borders of non-transgenic oilseed rape. It acknowledges, however, that

"Although these isolation safety measures have been shown to be effective at a small scale, pollen spread between the transgenic oilseed rape and its non-transgenic counterpart (or Brassica napus relatives) cannot be limited anymore via these techniques when the transgenic oilseed rape is grown at large scale."

Further studies of pollen flow were seen by PGS to be important in determining "to what degree the transgenic oilseed rape can be isolated from their non-transgenic relatives when it comes to pure seed production."

2.2.2 ACRE's view

The problem of scale was specifically raised at ACRE's meeting on 19 April 1994. Minute 3.20 states

"It was accepted that most releases to date had been on a small scale, and to some extent contained, and that they had not provided detailed information on environmental safety."

Faced with a known, but unquantified, risk of contamination of neighbouring crops, ACRE's discussions centred on two main issues.

First, since at that time clearance of the product for human food or animal feed had not been given, the Committee asked for and received assurances from the PGS representatives that the levels of containment and post-trial destruction would be maintained until the relevant authority had given clearance. (Human food approval was given on 6 January 1995, following deliberation by the Advisory Committee on Novel Foods and Processes, and MAFF gave PGS animal feed approval on 21 December 1995.)

Second, the Committee identified a potential "harm to man's property" if sufficient herbicide-tolerant volunteers (produced by cross-pollination between tolerant and non-tolerant crops) grew "in future years to interfere with the yield of a neighbour's crop". Other definitions of "harm" in Section 107 of Part VI of the Environmental Protection Act 1990 - harm to humans and other organisms, interference with ecological systems, and offence to man's senses - were considered and thought unlikely to occur from this release. It was concluded that harm to a neighbour's property could occur only in relation to the spread of herbicide-tolerant plants. ACRE's conclusion, that this was a low risk and one that could be managed, is consistent with its general view on herbicide-tolerant crops at that time (see Section 3).

2.2.3 Other views

The point that an application to place a genetically modified crop on the market should address the risks of large-scale, potentially uncontrolled, release was raised by several commentators in 1994, either explicitly (Greenpeace in an open letter to the Chairman of ACRE on 9 April 1994 - "Marketing applications must be viewed in a very different light than field trial applications"), or as implied by concerns with wider environmental impacts. Typical of the latter are the comments of the Danish Competent Authority (The Danish Environmental Protection Agency), which states that the notification should not be approved as it does not take into consideration the "product's effect on the use of herbicides and its long-term environmental consequences" (letter from DEPA to DETR on 10 August 1994).

The issue of gene flow between neighbouring crops per se does not appear to have been raised, although English Nature's concern with the risk of spread of viable seed by farm machinery once the crop is in commercial use expresses a similar problem (letter to DETR on 26 August 1993, re an earlier application for consent to release glufosinate-tolerant oilseed rape).

2.2.4 Recent developments

As ACRE acknowledged in its deliberations, the commercial-scale release of a genetically modified crop could present a novel problem because of "contamination" of neighbouring crops by cross-pollination. This is not a new issue as far as oilseed rape crops are concerned. Growing "double-low" oilseed rape crops (with low erucic acid and low glucosinolate content) to produce edible oils alongside cultivars with high erucic acid content grown for industrial lubricating oils, presents an analogous problem.

Research published since 1994 on gene flow from trial plots of oilseed rape largely supports the general statement in the PGS application that rates of cross-pollination fall rapidly with distance from the plant, or plot. Using glufosinate-tolerance as a marker, Scheffler et al. (1995) found hybridisation rates of 0.016% at 200m and 0.0038% at 400m from a central 400m² plot of transgenic plants - rates well below the limits of contamination accepted for breeders' basic seed and certified seed in the UK (0.1% and 0.3%, respectively - NIAB 1989). These levels differed from those in earlier trials (Scheffler et al. 1993) using the same marker (1.5% at 1m, 0.4% at 3m, and 0.0003% at 47m) - differences which the authors related to the effect of the size of the non-transgenic plot on bee foraging behaviour. Paul et al. (1995) recovered low numbers of hybrids (3.0%) in a central plot of mixed transgenic (Asulam-resistant) and unmodified oilseed rape plants, and were able to correlate the frequency of hybridisation in unmodified plants with the numbers of modified plants immediately surrounding them. They suggested that limited pollen carry-over by bees was a major factor. (In Cresswell's (1994) experiment with oilseed rape, more than 90% of gene transfer events occur in the first four flowers visited after the genetically modified plant.)

The dual rôle of insects and wind in oilseed rape pollination (Williams 1978, 1984) is undoubtedly one of several reasons why, despite generally leptokurtic distributions, the measured rates of hybridisation as a function of distance from a pollen source vary between different experiments. Other factors include the shape and size of the source and recipient plots, the density of plants, the degree of flowering synchrony, the timing of pollen release, the weather and the cultivar used. Attempts to standardise experimental methods and models (Lavigne et al. 1998, Crawford et al., in press) accounting for this variation, and to standardise statistical methods of analysing gene flow data (Gliddon et al., in press), offer the prospect of improving predictions about the levels of contamination, or containment, to be expected over a range of conditions.

Arguably the most relevant research for current considerations has been that of Timmons et al. (1995, 1996, in press), who provide clear evidence of long-range gene flow from fields of oilseed rape in their study areas in Tayside, southern Scotland. Seed were set on emasculated bait plants at distances of between 1.5 and 2.5km from the nearest oilseed rape field. From the pattern of distribution of fields and feral populations within their study area, and the greater capacity for long-range pollen dispersal from fields than from small plots, they conclude that hybridisation between transgenic oilseed rape and feral populations is inevitable following commercial release (Timmons et al. 1995). Interestingly, the frequency of gene flow between adjacent fields remained relatively low. The percentage of (inter-cultivar) hybrids declined with distance from 0.4% at 2m to 0.04% at 100m from the source field margin. This pattern of decline with distance, not dissimilar to that reported from smaller plots, was thought to reflect the effect of abundant local airborne pollen in the field (Timmons et al., in press).

Further research from the group working in Tayside (G.R. Squire, pers. comm.) has confirmed the greater potential genetic connectedness at the landscape scale, in both space and time (the latter via the persistence of feral populations), of oilseed rape crops and populations than predicted from the relatively small-scale experiments described earlier.

As discussed above, gene flow between neighbouring crops of oilseed rape may create a problem of contamination where high and low erucic acid crops are involved. In experiments described by Bilsborrow et al. (1998), levels of contamination were generally low, and less than 4% of subsamples had erucic acid concentrations above 2% (the EU standard for human consumption). Under the Arable Area Payment Scheme, MAFF has imposed minimum separation distances of 50m between double high and double low varieties, to reduce the chances that hybridisation could compromise oil quality. Whilst there are some parallels, contamination by a dominant gene for herbicide-tolerance will have a different set of consequences, especially for farm management.

2.2.5 Conclusions

It is evident that, from the application onwards, there was an awareness by all involved that, first, the environmental risks of growing the hybrid oilseed rape on a commercial scale raised additional questions to those posed by smaller-scale R&D trials of that crop, and, second, the data obtained from these small-scale trials might be of only limited help in answering those questions. In particular, the magnitude of gene flow between neighbouring crops and from transgenic crops to feral populations and wild relatives (dealt with in 2.3) was recognised very early in discussions to be scale- and time-dependent.

Research since then on gene flow at the appropriate field scale has been limited, and that at smaller scales not always appropriate for the development of generic predictive models. Variation between experiments in experimental protocol and in a range of key factors such as plot size and weather conditions makes it difficult to quantify risks based on existing data.

Nevertheless, this research confirms that cross-pollinations in oilseed rape are a variable, but often small, percentage of all pollinations, that most cross-pollinations occur over very short distances of less than one or two metres, that rates of cross-pollination fall rapidly beyond that to very low levels, and that rare long-distance pollinations occur, especially where the pollen source is large (a field) and the recipient population relatively small (e.g. a feral population).

This general statement (and recent studies at the landscape scale) are sufficient to indicate that complete genetic isolation of oilseed rape, were it required, would have to be on a regional (tens of kilometres) scale, and that, under current farm practices, local contamination between crops (probably at a low, but also variable, level) is inevitable. The level of contamination which is acceptable will depend on the trait. In the case of herbicide-tolerance, the occurrence of herbicide-tolerant oilseed rape plants in neighbouring fields as a result of gene flow is substantially equivalent to the occurrence of volunteer herbicide-tolerant plants, and can be managed in a similar way; except, as ACRE noted, where the neighbouring field is the property of another farmer. In these cases, the question of harm to man's property may arise.