Detecting gene flow from GM crops to wild relatives

Mike Wilkinson

Crop Species Interspecific hybrids in field

Wheat Triticum aestivum,T.turgidum

+

Rice Oryza sativa,O. glaberrima

+

Maize Zea mays +

Soybean Glycine max +

Barley Hordeum vulgare + (m)

Cotton Gossypium hirsutumG. barbadense

+

Sorgham Sorgham bicolour +

Millet Eleusine coracana,Pennisetum glaucum

+ (m)

Beans Phaseolus vulgaris +

Oilseed rape/canola Brassica napus,B. rapa

+

Ground nut Arachis hypogaea -

Sunflower Helianthus annuus +

Sugar cane Saccharum officinarum +

Interspecific hybrids between crops and wild relatives, worldwide

Ellstrand et al. 1999

In the United Kingdom

Of 30 crops reviewed

7 have no cross-compatible wild relatives

11with the potential for hybrid formation

12 with a history of hybridization

From Raybould and Gray, 1993, amended by Wilkinson 2002

In the United Kingdom

First wave of GM crops

Maize No relatives

Oilseed rape Many relatives

Sugar beet 1 close relative

Potato 3 relatives, none compatible

The process of transgene recruitment and spread

1. Initial Hybridization2. Introgression3. Gene flow between populations4. Changed fitness leading to change in population

size, density or distribution5. Effects on other organisms

1. Bilateral interactions(e.g. bitrophic)2. Trilateral interactions (Tritrophic)

6. Changed community

Initial hybridization

Why quantify hybridization?

1. Hybrid frequency affects the likelihood of all subsequent consequences

GM crop

A specialist parasitoid

Example Impact

GM crop

F1 hybrid in region

Insect resistant transgene stabilises by introgression

Transgene spreads to most populations

Enhanced resistance to herbivore depresses herbivore numbers

Depressed herbivore numbers depress specialist parasitoid

abundance

2. Hybrid abundance dictate the value of measures to repress/prevent

hybrid formation1. Isolation distance2. Male sterility3. Integration site4. Chloroplast transformation5. Inducible promotors/ ‘terminator technology’6. Transgene excision

Estimates of hybrid abundance and other exposure terms have most

value for regulation when applied at national scale

How do you estimate F1 hybrid formation between crop and wild

relatives across the UK?

1. Identify recipients

2. Quantify ‘local’ gene flow rates

3. Estimate long-range gene flow

4. Combine 1-3 to estimate frequency and location of hybrids

Possible primary recipientsScheffler & Dale (1994) Transgenic Res. 3, 263-278

• Brassica rapa• B. oleracea• B. carinata• B. juncea• B. nigra• B. adpressa

• Raphanus raphanistrum

• Diplotaxis erucoides• D. muralis

Brassica rapa

• Casuals of disturbed land

• A weed of B. napus

• Stable wild populations of river banks

Weedy B. rapa in oilseed rape

Hybrid seed formation

• 9-93% of seeds from B. rapa are hybridsAverage 60%

Surely lots of hybrids will be everywhere

Jorgensen and Andersen (1994)

No, because

1. B. rapa is an infrequent weed

2. Crop rotation and weed control

In Yorkshire and the Humber region37.4% of OSR fields contain weedy rapa

In all other regions0.54% of OSR fields contain weedy rapa

Weedy B.rapa

Hybrid survival

Hybrid seeds show <10% dormancy

Weedy B. rapa shows 60-90% dormancy

So in next year, 90% of hybrids germinate but only 10-40% of B. rapa

WEED CONTROL IS EFECTIVE IN CEREALS

Linder (1998) Ecological Applications 8 (4): 1180-1195

Wild B. rapa

Local hybrids

Long-range hybrids

Screen seed offspring for hybrids

15341 seeds sown from two populations

8647 seeds germinated

46 hybrids [morphology, flow cytometry, chromosome counts, ISSR] (0.5%)

Hambledon (5m separation- 0.4%)

Culham (1m separation- 1.5%)

Wilkinson et al (2000)

How often do populations coincide?

(Wilkinson et al., 2000)

Sympatry Likelihood Map

Total hybrid number per annum

Weed and wild hybrid numbers are now being calculated with error estimates for the entire UK

Conclusions• The number and distribution of hybrids

determine the scale, speed and possibility of any subsequent ecological change

• Hybrid numbers in a country determines the feasibility of corrective measures

• Error estimates will be fairly large initially but can be improved

• Hybrid frequency estimates should be followed by measures of other parts of the pathway to change

Thoughts

Quantification of each stage in the pathway to change is possible but requires consideration of location, context and biology of both crop and recipient

Quantification requires effort to integrate data from different disciplines

Information generated is generic for the crop in the country concerned