Spontaneous hybridization has, for instance, been reported in weedy populations of B. rapa growing in agricultural crops and in natural populations of B. rapa occurring near waterways. Second, flowering time has been extensively studied in B. rapa, and temporal clines in phenotypic traits have been observed. For example, time to first flowering has been shown to be positively correlated with stem height and stem diameter. Third, transgenic lines of B. napus containing a green fluorescent protein gene associated with the Bt transgene have been constructed. The presence of the Bt transgene in the offspring of weedy plants can therefore be inferred by exposing the plants to UV light.The aim of our experiment was to assess the impact of interspecific hybridization between weedy B. rapa and transgenic B. napus on the evolution of the weedy phenotype. This was done by identifying the phenotypic traits increasing hybridization opportunities for weedy individuals, searching for associations between these phenotypic traits and the transgenic trait in the offspring of weedy mothers and evaluating the relative fitness of hybridizing weeds. Our results show that weedy individuals that flowered later and for longer periods were more likely to receive transgenic pollen from crops and weed6crop hybrids. Because stem diameter is correlated with flowering time, plants with wider stems were also more likely to be pollinated by transgenic plants. Our results suggest that the transgene and maternal genes promoting late flowering, long flowering periods and stem thickening may be preferentially associated in the offspring of weedy mothers. However, growing trays although time to first flower is a heritable trait in B. rapa, our experiment did not confirm the gametic association between the transgene and genes promoting late-flowering in the offspring of hybridized weedyplants.
Indeed, given the very small numbers of Bt-GFP+ seedlings recovered from the experimental populations, we could not study the association between the transgenic trait and other phenotypic traits in weed plant offspring. We also found that the weedy plants with the highest probability of hybridization produced fewer seeds, despite producing larger numbers of flowers. The most straightforward interpretation of this result is that fecundity was reduced by hybrid crosses. Controlled crosses between the weedy and transgenic plants used in the experiment and several previous studies have indeed shown that crops and weed6crop hybrids have lower siring success than do weeds. Therefore, our experiment suggests that maternal weeds that flowered late and for long periods are less fit, because they have a higher probability of hybridizing with GM crop plants or hybrids. This may result in counter-selection against this subset of weed phenotypes, and a shorter earlier flowering period. It is noteworthy that this potential evolution in flowering time does not depend on the presence of the Bt transgene in the crop, and may even be counter-balanced by positive selection acting on the transgene if the latter was positively associated with maternal genes promoting late flowering and long flowering periods. Recent experiments indeed indicate that the Bt transgene does not induce any fitness costs in hybrids between transgenic B. napus and weedy relatives. It may therefore convey a selective advantage under insect herbivore pressure. In conclusion, our analyses show that phenological differences between weedy birdseed rape and transgenic rapeseed are likely to alter the phenotypic structure of weed populations, by promoting interspecific hybridization in only a subset of weedy plants with specific phenotypes and by altering the fitness of hybridizing weeds. Unfortunately, we could not verify the non-random association between the transgenic trait and other phenotypic traits in the offspring of weedy populations because of the very low rate of transgene introgression.
Nine populations, each composed of 15 Brassica rapa plants and 15 of one of three types of transgenic plants were sown as seeds and then grown from germination until death in a glasshouse at the University of California, Irvine. The nine populations were divided into three blocks, with each transgenic type replicated once per block. Plants were grown in individual Conetainer H pots filled with a 75/25 mixture of potting soil and sand. Before planting, seeds were vernalized on wet filter paper at 4uC for 5 days. Pots were spaced 7.6 cm apart and were watered every day until 90% stopped producing flowers. An equal amount of 10:10:10 NKP liquid fertilizer was applied to each pot on the sowing date. The three types of transgenic plants were: Bt-transgenic B. napus crop plants, Bt-transgenic B. napus 6B. rapa F1 hybrids, and first generation backcrosses . Over 20 unique seed and 20 unique pollen parents were used to produce each of the three types. B. rapa plants served as seed parents for the F1 and backcross types. B. napus were all homozygous for the Bt-GFP insertion, whereas the F1 plants were all hemizygous. The backcross generation was expected to consist of an equal mixture of hemizygotes and null homozygotes for the insertion. B. rapa seeds were obtained from over 400 mature plants in a population at Back Bay, near Irvine, California. Transgenic B. napus plants were derived from spring rapeseed lines . In addition to the Btcry1Ac gene from Bacillus thuringiensis , these lines contained a green fluorescent protein gene under the control of the cauliflower mosaic virus 35S promoter and a nopaline synthase terminator cassette. The fate of the Bt transgene could therefore be inferred by exposing the offspring to UV light. Flowering schedules were constructed for each individual plant by recording the time to first flower and the number of opened flowers on every fourth day until the end of the flowering period. The lifetime of a flower is about three days , so this procedure made it possible to estimate the total number of flowers produced by each plant over the flowering period.
The length of the flowering period was defined as the number of scoring days on which the plant had opened flowers. Every fourth day, all open flowers on all plants were hand pollinated in each of the nine experimental populations . Each experimental population was composed of 30 plants which were numbered from 1 to 30. On each pollination day, a random sequence of 30 numbers was generated for each population. For a given population, a pollination session consisted of brushing all the flowers of the first plant in the sequence, and then brushing all of the flowers of the next plant. This was continued until the brush from the 30th plant was used to transfer pollen to the first plant. Each plant was brushed up and down several times to deposit the pollen from the previous plant in the sequence and collect the maximum amount of pollen. A given plant was only brushed if it was alive and had one or more open flowers. Otherwise the next plant in the sequence was considered. Each of the nine populations had its own brush, and new brushes were used for each pollination session. This hand-pollination procedure was chosen to approximate the behaviour of a bumble bee in a patch of oilseed rape. Bumblebees tend to visit many plants successively and rarely revisit the plants. They deposit most of the pollen from a source plant on immediate neighbours.We performed all statistical analyses with SAS/STATH software. Plants that died during the experiment were excluded from the analysis and the final data set contained 117 weedy plants. We first investigated how phenological traits affected the chances of interspecific hybridization between Bt-trangenic plants and weeds. We used a mixed linear model , with transgenic type as the fixed treatment effect, phenological traits of weeds as covariates, grow tray and block and treatment6block interaction as random effects. The response variable was the proportion of flowers receiving pollen from Bt-transgenic plants . The response variable was log-transformed to increase its normality . If a factor was not significant as a single effect or in interaction with other factors, it was eliminated from the model and the analysis was rerun. We continued until there was no further improvement in residual maximum likelihood. We then investigated how morphological traits affected the chances of hybridization. A mixed linear approach was then used to determine whether the morphological traits changing with time to first flower had a significant effect on PPR. As above, transgenic type was treated as a fixed treatment effect, morphological traits were covariates and block and treatment6block interaction were treated as random effects.
We used the mixed linear approach with block and treatment x block interactions as random effects, to investigate whether the phenological and morphological traits which were found to favour hybridization of weedy mothers were transmitted to their offspring. In this model, transgenic type was treated as a fixed effect, the maternal trait as a covariate and the average offspring phenotypic trait as the response variable. The normality of the response variables was checked , and data was transformed as necessary. Finally we investigated the relationship between opportunities for hybridization and fecundity in weeds. We used the mixed linear approach with transgenic type as the fixed treatment effect, PPR as the covariate and block and treatment6 block interaction as random effects. The response variable was the total number of filled seeds. Its normality was checked with a Kolmogorov-Smirnov goodness-of-fit test . We then checked that the mother plants with the highest expected probability of receiving transgenic pollen also had the highest proportion of Bt-GFP+ seedlings. The proportion of Bt-GFP+ seedlings did not follow a normal distribution and could not be transformed. We therefore checked the effects of transgenic type, PPR and block separately, in non parametric oneway ANOVAs . The correlation between PPR and the proportion of Bt-GFP+ seedlings was assessed using Spearman’s rank correlation test .Weeds are great challenges for crop production, particularly those that are the same biological species as the crop they infest. For example, weed beets infest sugar beet fields, and weedy rice infests cultivated rice fields. The phenotypic similarity of such conspecific weeds to the related crops often frustrates visually based hand weeding. Also, genetic similarity means that the crop and the weed are so physiologically similar that herbicide must be applied on the weed with great precision to prevent application on the crop, again requiring visual discrimination . Because of their close evolutionary relationship, conspecific weeds are typically cross-compatible with the related crop species . Thus, conspecific weeds represent a unique challenge for their control in crop production because gene flow can deliver useful genes/alleles to weed populations from both their domesticated relatives as well as nearby non-weedy wild relatives . This infusion of genetic diversity can provide a substrate for rapid adaptive evolution . Crop-toweed gene flow has played a significant role in the adaptive evolution of weeds, such as weed beet , weedy rice , and California wild radish . The foregoing examples are a subset of unmanaged populations with introgressed domesticated alleles that have evolved increased weediness or invasiveness . In addition, crop-to-wild gene flow may also affect the evolutionary dynamics of wild populations, causing weed problems .The advent of genetically engineered crop species has stimulated discussion about whether crop-to-weed and crop-to-wild transgene flow might have an ecological or environmental impact . Like any other genes, transgenes should move from a GE crop to its non-GE counterparts and to wild/weedy relatives via pollen-mediated gene flow . If a weed/wild population has acquired a transgene that confers a strong selective advantage, and is exposed to a relevant selective pressure , it is likely to exhibit enhanced fitness and evolutionary potential under the natural selection, that may result in unwanted environmental consequences such as increased weediness or invasiveness . An introgressed transgene with neutral fitness impacts is expected to persist in the population; whereas a transgene with negative fitness impacts is expected to be purged from the population unless is a replenished by subsequent gene flow . Thus, a better understanding of correlates of crop transgenes in wild/weedy populations facilitates biosafety assessment of impacts caused by transgene flow. Consequently, the fitness and phenotypic correlates of crop transgene presence under field conditions have been studied in many systems; e.g. squash – wild gourd, maize – teosinte, cultivated sunflower – wild sunflower. For the world’s most important transgenic crops that have been commercialized or are nearing commercialization, the cultivated rice – weedy rice system is perhaps the best studied in that context. In China, a large number of GE rice lines with various transgenes have been developed, and some of them are under bio-safety assessment or on their way for commercialization .