4. Minimising inclusion of GM Material in Bee Products

4.1 GM Crop Plants That Could be Sources of Bee Forage in New Zealand

Table 2 lists crop species grown over sizeable areas in New Zealand and for which GM varieties are either available commercially from other countries or are being developed for research purposes in New Zealand. The relevance of each crop to honey bees, in terms of providing nutrients for the bees or pollination services for the plants, is noted. The plant's role in providing raw materials (nectar, pollen etc.) for a bee product for human consumption is also recorded.

Of the plants listed, clover is the most important for honey production, although canola and eucalypts have potential as honey sources as well. Even though forage and vegetable brassicas have the potential to produce useful nectar for honey, only if grown for seed are they permitted to flower before harvest (Christey and Woodfield, 2001) and so they probably do not contribute greatly to New Zealand bee products. Kiwifruit and apple pollen may make up a significant component of bee-collected pollen products and may also be present in some honeys. Maize (or sweetcorn) pollen may be an unintentional ingredient in honey or harvested pollen, since bees have been observed to collect this pollen (Emberlin et al., 1999). Potato, tamarillo, onion and leek pollen may also represent sources of unintentional pollen in honey, but probably only on a very small scale. Grass pollen is regularly found at low levels in New Zealand honey and some grasses are apparently worked by bees for their pollen (Moar, 1985). Pine pollen is produced in vast quantities and is well-dispersed by wind throughout the environment. As such it may end up in bee hives, but there are no records of bees actively collecting pollen from pine trees. Pine pollen has been detected at an extremely low level (less than one percent of total pollen grains) in a clover honey sample (Moar, 1985).

The following discussion of means for separating GM and non-GM varieties will focus on major New Zealand crops which may be genetically modified and may affect bee products such as honey, pollen or propolis.

4.2 Relevance of Cross-pollination Studies with GM Plants

Cross-pollination between GM and non-GM crops and between GM crops and their wild or weedy relatives is a topic for which there is a large and growing body of scientific literature. These data are used to define separation or isolation distances, buffer zones, border crops and other strategies to minimise unwanted cross-pollination. Such strategies can also use knowledge gained over many years in the development of seed certification systems for conventionally bred crop plants (Christey and Woodfield, 2001). The dispersal of pollen is a critical element in cross-pollination studies and often the role of bees is discussed. Obviously such studies can sometimes provide useful information for determining bee foraging distances and bee behaviour in relation to a crop. However, they are not always entirely relevant in determining if bees will gather GM material and if it will be incorporated into a bee product. For example, the ability of the plants concerned to produce viable hybrid offspring is irrelevant, as is the viability of the pollen (since it will still represent GM material if it occurs in honey). Important questions not generally considered in such studies include:

Will bees visit and gather pollen and/or nectar from the crop plant?
How far will they fly to do this?
How long will the gathered material remain in the hive?
How much of the material gathered from the crop plant will be present in the products of the hive?

There are some data available to answer some of these questions and these will be covered in the following sections.

Since there are no relevant available data on bees foraging for plant resins or honeydew, this discussion will focus on nectar and pollen collection.

4.3. Strategies to Minimise Inclusion of GM Material in Bee Products

Some possible strategies to minimise accidental inclusion of GM material in bee products include:

separating GM and non-GM crops via planting distances or flowering times
screening the crop to exclude bees
using bee management techniques that maximise foraging on a particular crop (including attractants and repellents)
using biotechnological solutions
using post-harvest honey treatments to remove pollen.

4.3.1 Separation of flowering GM crops and hives

One method for ensuring that bee products derived from GM and non-GM plants are kept separate would be to plant the different crops far enough apart to ensure that bees from a single hive could not visit both. This physical separation of the hive from the undesired flower source is the rationale used in the production of organic honey. Hives must be placed at least 3, 5 or 7 km (depending on the standard used) from non-organically grown crops (see Section 3.2 above for more detail). However, these organic rules recognise that, with this method, some contamination is inevitable and so maximum pesticide tolerance limits are set. It has been suggested that similar limits will need to be set for GM material in organic products (Moyes and Dale, 1999; Christey and Woodfield, 2001), but it appears that a "zero-tolerance" policy is in place thus far.

The required distance for GM/non-GM crop separation would depend on the maximum distance that a bee will travel to forage on that crop and it may vary depending on the relative attractiveness of the crop compared to other flowering plants in the same area. Information on bee foraging distances is summarised below.

Temporal separation of GM and non-GM crops (taking advantage of different flowering times) may also provide a possible means of separation.

Beekeepers have extensive experience in ensuring that their bees preferentially visit a particular plant in flower, since they must do this to produce unifloral honeys. This usually involves siting hives where the desired plant predominates at the appropriate time to capture its "nectar flow". While these methods are adequate for producing honeys of sufficient floral purity to satisfy consumer demand, they do not allow for the exclusion of pollen from a range of plants from the honey. To be called unifloral, a honey must have at least 45% of its total pollen content from the nominal plant species (Molan, 1998). Thus pollen from other plants commonly occurs in "unifloral" honeys. Concurrent foraging visits to other flowering plants account for only some of this pollen. Contamination may also occur when honey is extracted by crushing combs or using a loosening device (as with thixotropic honeys), since this can release stored pollen from nearby cells in the comb. Re-using comb from which honey was extracted during the previous season can also lead to contamination with "old" pollen.

4.3.1.1 Bee foraging distances

According to Winston (1987), most honey bees in agricultural areas forage within a few hundred metres of their hives, although significant populations have been found at 3.7 km. In forested regions, they forage at a median radius of 1.7 km from the hive and most can be found within 6 km. He points out that bees can be recruited to feeding stations up to 10 km from a hive if there are no competing food sources. Williams (2001) confirms this 10 km maximum flight distance.

According to Gary (1992), bees have a strong tendency to forage at the nearest source for each floral species in an area. He also mentions "distant flight" behaviour in agricultural areas where attractive crops are planted in widely dispersed fields, such that significant bee populations may be found at least 6.5 km from an apiary. Gary notes that bees in a desert will fly up to 13.7 km to a food source if there are no other food sources closer to the hive. This is the maximum bee foraging distance mentioned in the literature.

Moyes and Dale (1999) note mean foraging distances of 1.66 km and 557 m for bees foraging on flowering carrots and onions, respectively, and maximum distances for these crops on 6.17 and 4.25 km. Ramsey et al., (1999) notes bees flying 5 km to reach an oilseed rape field.

In New Zealand, a 5 km distance is generally recognised as being the minimum when shifting hives to ensure that they will not return to their old hive site (Matheson, 1997).

Friends of the Earth recently commissioned a study of pollen dispersal by honey bees from a herbicide-tolerant GM oilseed rape farm-scale field trial site (Emberlin and Brooks, 2001). Pollen traps were placed on six hives, two at each of three apiary sites, up to 4.5 km from the flowering GM oilseed rape crop. There were apparently no other flowering crops in the vicinity. Forty samples of pollen were taken from the traps and examined for oilseed rape pollen pellets (identified by colour and shape). Of these, six samples (presumably one from each hive) with numerous oilseed rape pollen grains were selected and sent for DNA analysis to the Austrian Federal Environment Agency Laboratory. PCR tests for the nos terminator and bar gene (see Section 2.2 above) gave positive results for each of the six pollen samples, suggesting that even bees from the furthest hive had gathered pollen from the trial site. Once again, there was no analysis of control samples of oilseed rape pollen collected from non-GM crops, so that the possibility of microbial contamination giving positive readings for nos or bar DNA cannot be discounted.

The literature thus suggests that a distance of more than 13.7 km from hive to GM crop would give a 100% guarantee that bees would not forage on the crop. However, this figure is derived from an experiment with bees in a desert with no other sources of forage, which is not a realistic agricultural situation. A better approach perhaps would be to define realistic foraging distances for different cropping situations and to assign probabilities that a least one copy of transgene DNA will occur in a pollen or honey sample as a function of distance. This approach supposes a zero tolerance limit for such DNA in these bee products. A less stringent limit would produce a different set of probability values.

4.3.1.2 Accidental inclusion of wind-borne pollen in bee products

A second possible source of GM pollen in hives and bee products could be that from GM crops which produce significant quantities of wind-borne pollen, e.g. ryegrass or pine. The discovery of a tiny amount of pine pollen in a sample of New Zealand clover honey (Moar, 1985) suggests that such an occurrence may not be completely improbable. However, there are no published data on how close a honey-producing hive would have to be to a pine plantation or ryegrass pasture for such pollen to occur in the honey or pollen harvested from that hive.

4.3.1.3 Feasibility of the crop/bee separation approach

Making sure that bees are sufficiently distant from any GM crop site to prevent visits to the crop or accidental occurrence of pollen from that crop in hives during any honey-making season could ensure that no or minimal GM pollen is accidentally introduced into the honey made that season. This would require careful, planned deployment of GM crops, especially clover or oilseed rape. Excellent communication between land-users and beekeepers would also be required. Since such communication is already an important factor in the success of beekeeping businesses, this should be achievable. However, some beekeepers have reported that they have been deterred from shifting to organic honey production by increased complications in dealing with landowners (Bourn et al., 1999), even without a consideration of the possibility of GM crops.

The carry-over from year to year of pollen in frames of empty comb and perhaps other hive equipment may mean that an equipment labelling and "quarantine" system, similar to that already employed in New Zealand for American foulbrood control (Goodwin and Van Eaton, 1999), would need to be implemented to segregate "GM" and "non-GM" hives.

The stringency of the techniques that will need to be used will, of course, be dictated by the tolerance limits for unintentional presence of GM material in bee products set by the countries where the honey will be sold.

4.3.2 Screening the crop to exclude bees

Screening a crop with bee-proof mesh would be practical only for small-scale field trial plots. It may be feasible for commercial crops of extremely valuable GM plants, for example those grown to produce very valuable proteins for extraction and purification ("biopharming"), where the areas planted may be relatively small.

4.3.3 Bee management techniques to direct bees to visit particular crops

A number of bee management techniques have been developed to enhance bee visits to particular crops, usually in order to increase pollination and/or fruit set. The most obvious method is to simply place hives near the crop and away from other flowering plants. This method's usefulness in relation to the presence of GM material in bee products has been covered in Section 4.3.1. Most of the other methods are based on the application of a bee attractant to the crop. They are summarised in Section 4.3.3.1 below.

A number of chemicals have also been identified as bee repellents. Some are insecticides and their bee-repellent qualities have been noted as a side-effect, whereas others have been used deliberately to keep bees away from potentially harmful insecticides or poisons intended for control of pests. These are summarised in Section 4.3.3.2 below.

Both strategies could be of use where there some GM material can be tolerated, e.g. for honey which at present does not require a GM label in New Zealand, Australian, EU, North American and Asian markets. They would not be suitable for situations where there is "zero tolerance" for GM material in bee products.

4.3.3.1 Bee attractants and other methods to maximise foraging on a crop

Spraying crops with sugar syrup in order to increase bee visits, pollination and fruit set has produced mixed results. Goodwin (1997) reviewed this use of sugar syrup and concluded that it was an unreliable method but potentially useful if further research could improve reliability.

A number of commercial products based on sugar syrup (e.g. Beeline, Bee-Q and BeeLure) are sold for spraying on crops to improve pollination. These also have mixed success and there are many reports of their failure to increase bee visits or seed set (e.g. Burgett and Fisher, 1979; Belletti and Zani, 1981; Rajotte and Fell, 1982; Margalith et al., 1984; Singh and Sinha, 1996; Ambrose et al., 1995).

Other commercial products for this purpose are based bee pheromones (e.g. BeeHere, Bee Scent and QMP). Success has been reported with their use on raspberries (Neira et al., 1997), strawberries (Butts, 1991), cranberry (MacKenzie and Averill, 1992), apple, cherry, pear and plum (Mayer et al., 1989). However, failure has been reported with some crops, such as apricots (McLaren et al., 1992), kiwifruit (Tsirakoglou et al., 1997), watermelon and cucumber (Ambrose et al., 1995).

A second use of sugar syrup is to feed syrup scented with the flowers of the target crops to bees, with a view to recruiting more bees to forage on the crop. Goodwin (1997) also reviewed this technique. He reported that it had mixed success and had received little attention over the last 30 years.

A third approach to improving crop pollination is to increase the number of pollen gatherers in a colony by feeding unscented sugar syrup within the hive (Goodwin, 1997). Increases in pollen collection using this method have been reported for a number of crop plants, although actual improvements in their pollination have not been assessed. Syrup feeding may also increase pollen collection from plants other than the target crop, so this method may have limited use in directing bees to forage only on a particular crop within their flight range.

Finally, breeding crops with increased levels of honey bee attractive floral volatiles (e.g. linalool) has been suggested as a method for increasing bee visits to crops such as alfalfa (Henning et al., 1992).

4.3.3.2 Bee repellents and other methods to prevent bee visits to a crop

Trap crops (non-GM borders grown around a GM crop) are used to minimise pollen flow via insects and wind from a GM crop. These take advantage of the fact that pollen dispersal has a highly leptokurtic distribution (i.e. pollen levels decrease dramatically within metres of the crop and then remain at very low levels over a far greater distance) (Williams, 2001). While this may help to reduce concerns about cross-pollination, its impact on GM pollen presence in bee products is not known. Williams (2001) also suggests that planting a surrounding trap crop of preferred bee forage may have potential as a means of reducing bee visits to a GM crop.

Pollen traps fitted to hives have been tried, but found to be unreliable, as a means of reducing the amounts of insecticide-treated pollen entering a hive from sprayed crops nearby (Erickson et al., 1997). Such traps may be of some use as a means of excluding GM pollen, but this has not yet been investigated.

A number of chemicals have been identified as honey bee repellents. Atkins et al. (1975) reported 42% and 69% bee repellence from flowering crops sprayed with ethyl hexanediol and decylamine, respectively.

Some pesticides appear to have bee repellent properties. For example, some pyrethroid insecticides have been shown to repel bees and this is thought to explain why this chemical causes less mortality in the field than would be expected from laboratory-based toxicity tests (Rieth and Levin, 1987; 1988; de Wael and van Laere, 1987). Fries (1985) noted that cypermethrin reduced oilseed rape pollen collection by honey bees. Orthene sprayed on pre-flowering raspberries resulted in a failure of pollination by honey bees (M. Goodwin, pers. comm.). The fungicide captan, although not toxic to honey bees, repels them if applied to flowering plants (van Praagh and von der Ohe, 1982).

Other compounds have been tested for their ability to repel bees, but not other insects, from insecticides or other poisons sprayed on crops or used in baits. The honeybee pheromone 2-heptanone was tested for this purpose but was found to be impractical and not sufficiently reliable (Reith et al., 1986). Goodwin and Ten Houten (1991) had better success with blackstrap molasses added to 1080/jam baits used to kill possums. They identified oxalic acid as the bee-repellent component of the molasses. However, the use of oxalic acid on flowering crops has not been tested and the possibility of phytotoxicity has not been discounted (M. Goodwin, pers. comm.).

4.3.4 Biotechnological solutions

There are a number of biotechnological approaches which may help to reduce GM material in bee products. Most are being developed in order to minimise cross-pollination (and thus gene flow from GM crops) and some are aimed at improving crop yield or reducing pollen allergenicity problems by eliminating flowering.

Promoters that direct transgene expression to tissues other than pollen or the nectaries could be used to minimise the presence of novel proteins in pollen and nectar. For example, leaf- or root-specific promoters are being developed (e.g. Santamaria et al., 2001; Imura et al., 2001), especially for transgenes encoding insecticidal proteins, so that the proteins occur where pest insects feed and not where beneficial insects such as bees do. However, with this method the transgene will still be present in pollen and will thus continue to represent a potential source of GM material for bee products. The following discussion will focus on methods which may eliminate transgene DNA from pollen.

4.3.4.1 Modification of chloroplast DNA

Commercially available GM plants have been modified via the insertion of a transgene into the plant's nuclear genome, so that every plant cell with a nucleus will contain the new DNA. However, chloroplasts, like some other organelles, contain their own DNA, separate from that contained within the nucleus. This is known as the chloroplast genome. It is possible to insert a transgene into the chloroplast genome so that only plant tissues composed of chloroplast-containing cells will carry the transgene and have the ability to express the novel protein it encodes (e.g. Daniell et al., 1998). Since the leaves, shoots and stems of plants are often the desired sites for expression of new traits (e.g. pest or disease resistance or altered nutritional properties), this method has potential for conferring such traits while avoiding the difficulties that the transgene's presence in pollen may pose (e.g. Lutz et al., 2001; DeGray et al., 2001).

In most flowering plants the chloroplast genome is absent from pollen. Because of this, chloroplast DNA sequences are used to study maternal inheritance in many plants (e.g. Balfourier et al., 2000). However, the conifers are a well-known exception to this and chloroplast genome sequences have been used for paternity analysis in Pinus radiata in New Zealand (Kent and Richardson, 1997). Paternal transmission of chloroplasts is also known in carrots (Moyes and Dale, 1999), and on some occasions in some other angiosperms, e.g. runner beans, peas, potatoes, meadow grass (Moyes and Dale, 1999), tobacco, lucerne (Stewart and Prakash, 1998; Daniell et al., 1998) and pelargonium (James et al., 2001). Thus the effectiveness of this method for eliminating GM material from the pollen of GM plants will depend on the plant concerned.

Flowering GM crop species visited by honey bees may be suitable candidates for this method. Scott and Wilkinson (1999) studied rates of maternal inheritance of chloroplast DNA in oilseed rape and concluded that there will be no or negligible pollen-mediated chloroplast-transgene dispersal from this crop. McKinnon et al. (2001) drew a similar conclusion from a study of eucalyptus.

4.3.4.2 GM plants without pollen

Male sterility is used in conventional hybrid plant breeding to control pollination and many crops have natural male sterility systems that can be exploited (Christey and Woodfield, 2001). It can also be introduced into crop plants via genetic modification.

Several different strategies are being investigated, but one of the best known is the barnase/barstar system. With this, a bacterial gene encoding a cytotoxic enzyme, barnase (Bacillus amyloliquefaciens RNase) is placed on a tapetum- or pollen-specific promoter so that it is expressed only in the anthers during pollen grain formation. Because it is cytotoxic, barnase disrupts this process so that the plants produce either no pollen, deformed or inviable pollen. Plants can also be modified to carry another gene from the same bacterium called barstar. This encodes a protein which inactivates barnase and blocks its cytotoxic effect. If the barstar gene is driven by an inducible promoter (i.e. one that works only when triggered by the application of a particular chemical), then it becomes possible to switch pollen production back on when desired by spraying with the inducing chemical. A number of GM male-sterile crop plants have now been successfully developed (but apparently not yet commercialised) with the barnase system, e.g. oilseed mustard (Arun et al., 2001), cabbage (Zhu et al., 2001), alfalfa (Rosellini et al., 2001), tobacco (Li et al., 1997), wheat (de Block et al., 1997), soyabean (Guo et al., 1997), poplar (Li et al., 2000) and rice (Zhang et al., 1998).

Obviously male-sterile GM plants completely lacking pollen would not be a source of GM material for honey bees. However, it is not certain whether deformed or inviable pollen would be rejected by foraging honey bees. Bees are known to exhibit preferences among pollen types when presented with a choice, apparently choosing on the basis of odour and physical configuration of the pollen grains (Winston, 1987).

4.3.4.3 GM plants without flowers

GM techniques may also be used to retard or prevent flowering, thus preventing undesirable gene flow from pollen dispersal. There may also be benefits in eliminating flowering from some crops in order to encourage vegetative growth (e.g. forage plant production) and to reduce the production of allergenic pollen (e.g. ryegrass). There are very obvious detrimental implications for honey bees in having non-flowering plants, especially with crops that are important for honey production, such as clover. The loss of flowers even from species that are chiefly wind-pollinated, such as maize, could impact negatively on honey bees that may rely on these plants as a supplementary pollen source. However, if the demand for GM-free bee products is sufficiently high, then the option of non-flowering GM plants may become attractive.

One strategy to prevent flower formation uses the barnase (cytotoxic) gene attached to genes that are expressed only in inflorescences, such as a MADS gene (Lemmetyinen et al., 2001).

There are no reports of the impacts of such plants on honey bees or bee products.

4.3.5 Post-harvest honey treatments

If GM pollen could be removed from honey after harvest, the likelihood of GM material (DNA or protein) occurrence would be greatly reduced or eliminated.

Honey is generally filtered after harvest to remove wax and debris before packaging (comb honey is an obvious exception to this). It is sometimes stated that the filtering of commercial honey reduces the level of pollen to 0.1% or less (Anon P, 2001; Anon Q, undated). This figure accords with most of the reports quantifying pollen content of honey (see Section 2.2), but not all. For example, Eady et al. (1995) reported 100,000 grains per ml in a UK commercial honey (equivalent to about 0.3%) and Molan (1985) gave a maximum pollen concentration of 5 million grains per 10 g of honey (about 1.5%).

In New Zealand, a relatively coarse nylon fabric filter is usually used to filter honey (Matheson, 1997) and this is unlikely to remove all pollen grains, although more sophisticated filtration units that use mesh filters may remove significant quantities of pollen (Bryant, 1987). High pressure filters using a series of paper filters, sometimes with diatomaceous earth added, are available and used in the United States (Tew, 1992). Molan (1998) reported that honey that has been filtered with diatomaceous earth has no pollen left in it.

Contact for Enquiries

Dr Sharon Adamson
Manager, Innovation Policy
Ministry of Agriculture and Forestry
PO Box 2526
Wellington
NEW ZEALAND

Phone: +64 4 894 0618
Fax: +64 4 4 894 0741
Contact this person