Introduction
Over the last two decades several developed countries have approved the release of genetically modified (GM) varieties of more than 10 agricultural species. Today, more than 175 million hectares have been planted with GM crops worldwide (James 2013). Most commonly, GM crops have received genetically engineered sequences (transgenes) that confer them with tolerance to herbicides and resistance to herbivores and viruses (James 2013). The dispersion of transgenes into wild relatives from GM crops may represent an ecological risk, especially for relatives located on the sites of origin and diversification of cultivars (Gepts and Papa 2003, Snow et al. 2005, Ellstrand et al. 2013).
Mexico has approved the experimental planting of GM squash (Cucurbita pepo) with transgenes conferring resistance to three common viruses (public communication;
Gene flow between crops and their wild ancestors is a commonplace phenomenon that may allow the introgression (permanent incorporation) of transgenes from GM cultivars to their wild relatives (Stewart et al. 2003, Kwit et al. 2011, Ellstrand et al. 2013). The introgression of transgenes into wild populations involves gene flow from a transgenic crop to a wild relative through hybridization, backcrossing, and selection (Wolfenbarger and Phifer 2000, Stewart et al. 2003, Chapman and Burke 2006, Kwit et al. 2011). The initial escape of a transgene into a wild population may occur via pollen or seeds. Further spread and establishment of the transgene in the wild population depends on the fitness of the transgenic hybrids and backcross (BC) progeny relative to the wild‐type plants (Stewart et al. 2003, Hooftman et al. 2008). Additionally, successful transgene introgression requires that the transgene remain functional as it is inherited across generations (Dale 1994, Dietz‐Pfeilstetter and Kirchner 1998).
Gene flow is contingent upon the spatial and temporal overlap in flowering between the GM cultivar and the wild plants (Wolfenbarger and Phifer 2000, Stewart et al. 2003). Low hybrid fitness (hybrid breakdown) is a common barrier to gene flow among species (Arnold et al. 1999, Burke and Arnold 2001), and therefore, it could be the main impediment to the spread of transgenes. However, hybrids formed from the cross between varieties of the same species (or across species) could have equal or greater fitness (heterosis) than the parents (Arnold and Hodges 1995, Burke and Arnold 2001). Clearly, the persistence of an escaped transgene in a wild population depends on the fitness of the hybrid progeny (Hails and Morley 2005).
Disease‐resistance transgenes could confer hybrids with selective advantage over wild‐type plants, especially under conditions of high disease incidence. Such an adaptive advantage would increase the potential of introgressed plants to become weeds, which could be the case of the hybrid progeny resulting from crosses between the Liberator III transgenic zucchini, genetically engineered for resistance against three common viruses and wild squashes (Tricoll et al. 1995, Laughlin et al. 2009). After deregulating transgenic Cucurbita in the USA in 1996 (NRC 2002), the cultivation of this agricultural species has potentially allowed for the introgression of transgenes into wild relatives in the USA (NRC 2002) and in Mexico—center of origin, diversification, and domestication of Cucurbita (Whitaker and Bemis 1964, Hurd et al. 1971, Sanjur et al. 2002). Eleven taxa of wild Cucurbita occur within close physical proximity of agricultural fields throughout Mexico (Lira‐Saade 1995), and gene flow between wild and domesticated squashes has been documented already (Montes‐Hernández and Eguiarte 2002). Thus, the conditions for the escape of transgenes exist. Transgene escape may have negative consequences for conservation and biodiversity if hybridization between GM crops and their wild relatives results in the development of invasive plants (Wolfenbarger and Phifer 2000, Pilson and Prendeville 2004), or if the transgene increases the transmissibility or virulence of the pathogen (Tepfer 2002, Pilson and Prendeville 2004). On the other hand, transgene escape could be innocuous to the environment if transgenic hybrids are unable to persist in nature.
Here we present an experimental study of hybridization between the transgenic zucchini cultivar Liberator III, and a wild cucurbit that is not the closest relative of the cultivar, but is widely distributed in Mexico: Cucurbita argyrosperma ssp. sororia (Merrick 1990). Our main objective was to assess the transmission, via pollen, of a virus resistance transgene (VRT) from transgenic zucchini (Cucurbita pepo) to experimental populations of its wild relative (C. argyrosperma ssp. sororia) at the center of origin of the genus. We assessed transgene inheritance in F1 and F2 hybrids, and BC progeny, and evaluated performance in each generation under greenhouse and field conditions.
Materials and Methods
Parental species
In order to study the inheritance of the transgene across species, we generated F1 hybrids from crosses between cultivated Virus Resistant Transgenic (VRT) Cucurbita pepo L. as pollen donors and wild C. argyrosperma ssp. sororia as recipients because gene flow is more likely to occur via pollen flow into wild populations. Moreover, this cross is the only one that produces hybrid progeny in matings between C. argyrosperma and C. pepo (Merrick 1990). Permits needed for this study with GM zucchini in Mexico were granted by SAGARPA (0330, 0390, 02034).
The Liberator III crookneck squash (Seminis, San Louis, Missouri, USA) cultivar is a commercially available transgenic zucchini cultivar (C. pepo ssp. pepo) with coat protein‐based resistance to three of the most common viruses on cultivated squash: zucchini yellow mosaic virus (ZYMV), watermelon mosaic virus (WMV), and cucumber mosaic virus (CMV). The VRT construct contains the promotor 35S of the cauliflower mosaic virus and the neomycin phosphotransferase gene (NPT II) as a marker; and it is inherited hemizygotically (Tricoll et al. 1995). All three viruses have a worldwide distribution and seem to have an important anthropogenic dispersal component (Lecoq et al. 1998, Simmons et al. 2008), and preliminary data show that at least ZYMV is present in wild populations of cucurbits associated to rural human settlements (R. Cruz‐Reyes; unpublished data) and may have detrimental effects on plant fitness (Sasu et al. 2009).
Cucurbita argyrosperma ssp. sororia (L. H. Bailey), one of several widespread wild taxa that could harbor transgenes escaped from GM zucchini, occurs in disturbed areas along seasonal streams and near agricultural lands in Mexico and Central America (Merrick 1990). For this study, we used seeds from nine natural populations obtained from the Chamela region in Jalisco, Mexico (Appendix A: Table A1).
The two parental species are part of a complex of wild‐domesticated species within which there is ample gene flow (Wilson 1990). C. argyrosperma ssp. sororia is the most genetically distant species of the clade that includes C. pepo ssp. pepo, C. argyrosperma ssp. argyrosperma, and C. moschata. Both species are diploid (n = 20), monoecious, protandric, self‐compatible annuals. C. pepo ssp. pepo has short internodes, and therefore a more compact habit compared to C. argyrosperma, which is a vine (Merrick 1990, Singh 1990).
Inheritance of the transgene
During the summer of 2007, we conducted manual pollinations on 110 wild plants grown in an experimental plot at the Instituto de Investigaciones en Ecosistemas y Sustentabilidad UNAM, in Morelia, Michoacán. Transgenic plants were grown inside a greenhouse to impede any uncontrolled flow of transgenic pollen. One day before anthesis, we twist‐tied the corollas of staminate and pistillate flowers to prevent pollinator visitation. Upon anthesis, we removed the pollen of a minimum of five transgenic pollen donors, and mixed it up in a plastic container. We repeated the above procedure using pollen from five wild donors. Pollen from these containers was then used to conduct the experimental pollinations to produce F1 progeny.
We conducted the following hand pollination treatments: (1) saturation of the stigmata of wild plants using only transgenic pollen (SAT), (2) mixed pollen loads in either a 15:1, 7:1, or a 3:1 transgenic:wild pollen ratio (Appendix A: Table A2). We chose these ratios because crosses between these two species invariably favor siring by conspecific pollen (Quesada et al. 1993, 1995, 1996). Because of hemizygotic inheritance, only half of the pollen produced by transgenic plants is expected to carry the transgene. Pollen mixtures delivered roughly 300 pollen grains in total (285 ± 52 [mean ± SE] transgenic and 18 ± 6 wild pollen grains for the 15:1 mixture, 261 ± 42 transgenic and 36 ± 11 wild grains for the 7:1 mixture, and 239 ± 23 transgenic and 75 ± 11 wild grains for the 3:1 mixture). We used the flat tips of metal rods of different diameters in order to deliver precise amounts of pollen for all mixtures. Ovaries of recipients contained 262 ± 34 ovules, and therefore the loads we used were sufficient to fertilize most ovules in the ovary (Quesada et al. 1993). Recipient flowers were labeled and seeds (F1) from the resulting fruits were extracted and counted. In 2008, we grew a sample of 30 F1 progeny from each recipient to assess performance of families, screen them for the presence of the transgene, and use them to obtain F2 and BC progeny (Appendix A: Table A2; Appendix B). F1 progeny that carried the transgene are hereafter referred to as F1 VRT.
In order to generate F2 progeny, we conducted two hand pollination treatments: (1) F1 non‐VRT as pollen recipients and F1 VRT plants as pollen donors (F1 non‐VRT × F1 VRT); (2) F1 VRT plants as both pollen donors and recipients (F1 VRT × F1 VRT). Since we were also interested in the persistence of transgenes through backcrossing, we generated two kinds of backcrosses: (1) F1 VRT as pollen recipients and wild plants as donors (F1 VRT × W); (2) wild recipients and F1 VRT hybrids as donors (W × F1 VRT). For all these pollinations, we saturated the stigmata of recipients with the corresponding pollen (no mixtures were used, Appendix A: Table A2). We assessed the performance of F2 progeny in an experimental plot.
We evaluated the performance of hybrids and BC progeny in comparison with that of the parental lines (wild and VRT) in terms of the following variables: proportion of seeds that germinated, proportion of plants that survived to reproductive age (census conducted at the middle of the reproductive season), days to emergence, leaf area at day 14 (after emergence), staminate and pistillate flower production, and pollen viability (as measured by its stainability with Alexander's green). For consistency with data from F2 progeny, we only assessed performance on the F1 generated by means of the stigma saturation pollination treatment.
Seeds for all generations were first germinated in a growth chamber and later transplanted to an agricultural plot (Appendix B). Performance was assessed the first year for the parental lines, the second year for the F1 and the third year for the F2 and BC progeny using the same experimental plot every year (see planting design details in Appendix B).
Detection of the transgene
All plants in our study (parentals, F1, F2, and backcrosses) were screened for the presence of the VRT using the expression of NPTII and the promoter 35S of the cauliflower mosaic virus (P‐35S CaMV) as indicators of VRT presence and functionality (Tricoll et al. 1995, Lipp et al. 1999; see molecular analyses in Appendix B).
Genetic and statistical analyses
We tested for Mendelian inheritance of the VRT overall, and among families (Appendix A: Table A3) within each type of cross by means of a G goodness‐of‐fit test, assuming the VRT to be hemizygotic (Sokal and Rohlf 1995, Tricoll et al. 1995). Expected ratios for F1 progeny were: 1:1 VRT: non‐VRT for the saturation treatment, and 15:17, 7:9, and 3:5 VRT: non‐VRT for the 15:1, 7:1, and 3:1 mixtures, respectively (Appendix A: Table A2). The asymmetry between pollen loads simulates a scenario where a large plot of transgenic plants is surrounded by a wild population, thus promoting a higher frequency of pollen grains from the domesticated crop than wild ones on stigmata of plants near the plot. Additionally, a previous study on these species with similar loads found conspecific fertilizations to be more successful than heterospecific ones (unpublished data). Since we only used the saturation treatment to produce the F2 progeny, the expected ratios for the F2 were 1:1 VRT: non‐VRT from the cross (F1 non‐VRT × F1 VRT), and 3:1 from the (F1 VRT × F1 VRT) cross. Lastly, expected ratios for both backcrosses were 1:1 VRT: non‐VRT.
We conducted a multivariate analysis of variance (MANOVA) to determine if the presence of the transgene in the F1 generation and the type of cross used to produce it (pollination treatments: SAT, 15:1, 7:1, 3:1) resulted in differential performance among VRT and non‐VRT plants. We used number and mass of seeds, days to germination, leaf area and pistillate and staminate flower production as indicators of performance. To meet assumptions of normality and homoscedasticity, the following variables were transformed as indicated: number of seeds (log10), staminate flower production (1/square root), mass of seeds (square root). Leaf area of seedlings was not included because no data were collected on this variable for progeny from mixed pollen load treatments.
Similarly, we used MANOVA to test for effects of the type of cross used to produce F2 progeny (F1 non‐VRT × F1 VRT or F1 VRT × F1 VRT) and of the presence/absence of the transgene on days to germination, leaf area, and staminate and pistillate flower production as indicators of performance. Since backcrosses produced fewer progeny, our sample sizes were too small to conduct a MANOVA. Instead we performed univariate analyses (GLM) to test for differences between transgenic and non‐transgenic progeny for both types of backcrosses (F1 VRT × W and W × F1 VRT) on days to germination, leaf area, and staminate and pistillate flower production.
Since we found no differences in vigor between transgenic and non‐transgenic F1, F2 and BC progeny, we used pooled data for each type of cross to compare performance among parental, F1, F2 and BC progeny by means of a general linear model that tested for the effects of generation (parental, F1, F2 and BC) and maternal family nested within generation followed by a priori Tukey multiple‐comparison tests. To achieve model assumptions, days to germination, foliar area, staminate and pistillate flower production, and seeds per fruit were log‐transformed, and seed mass was square‐root transformed.
We used a chi‐square test of independence to test for differences among progeny from parental, hybrid and backcrosses on seed germination, survival to reproduction, and pollen viability. We followed this analysis with an a posteriori Marascuilo multiple‐comparison test for differences between progeny types in the proportion of seeds that germinated, survived to reproduction, and also for the proportion of viable pollen.
Additionally we determined reproductive rate for each type of cross from a simplified life table with data on survival to reproductive stage and fecundity through the female or male function. Fertility through the female function was calculated as seed production divided by the number of individuals that survived to reproductive stage; fertility through the male function was calculated as the product of the number of male flowers, pollen viability, and the mean number of pollen grains per male flower of C. pepo (28016 grains; Quesada et al. 1995) divided by the number of plants at reproductive stage. For female fertility of the F1 progeny derived from treatments with mixed pollen loads, we did not differentiate between the values of VRT and non‐VRT plants.
Results
Inheritance of the transgene
F1 progeny
All types of crosses produced enough seeds for an adequate analysis of the inheritance of the transgene. The inheritance of the VRT deviated from the expected Mendelian proportions for a hemizygotic gene in all four crosses used to generate F1 hybrids (G = 76.98, df = 39, P = 0.0003). In the stigma saturation treatment (only VRT pollen used), we expected a 1:1 transgenic:wild phenotype ratio. Similarly for the pollen mixture treatments, observed ratios differed significantly from theoretical ratios (Fig. 1A).
Observed frequencies of (A) F1 hybrids with (grey bar) or without (open bars) the VRT transgene, at four different loads of transgenic:wild pollen (saturation, 15:1, 7:1, 3:1); and (B) F2 and backcrossed progeny. To generate F2 progeny, we used F1 non‐VRT and F1 VRT plants as recipients and F1 VRT as donors. For the two kinds of backcrosses, F1 VRT progeny were used as recipients or donors in crosses with wild parents. Streaks on bars indicate progeny from families that did not follow the expected Mendelian frequencies. If these areas are not taken into account, the observed phenotype frequencies approach more the expected VRT:non‐VRT Mendelian frequencies. ** P < 0.01, *** P < 0.001.
Overall, we observed a deficit in the number of F1 VRT progeny (Appendix C: Table C1). Moreover, as the transgenic:wild pollen ratio decreased, the absolute numerical difference between transgenic and non‐transgenic progeny increased. However, significant deviations from the expected ratios did not occur in all the families (progeny derived from a single fruit). Transgene segregation was not significantly different from the expected ratios in 85% of the families in the saturation treatment, and 82%, 50%, and 35% of those in the 15:1, 7:1 and 3:1 pollen mixtures respectively (G = 34.25, df = 15, P = 0.004; G = 63.03, df = 11, P = 0.000001; and G = 97.34, df = 16, P = 0.00001; Fig. 1A).
F2 progeny
Overall, the inheritance of the VRT also deviated from the expected Mendelian ratios for F2 progeny, regardless of the type of cross (F1 non‐VRT × F1 VRT or F1 VRT × F1 VRT; G = 111.48, df = 25, P = 0.00001 and G = 64.8, df = 18, P = 0.00001, respectively, P < 0.01 for both cases; Fig. 1B). However, as in the F1 progeny, the observed frequencies of VRT and non‐transgenic F2 progeny did not deviate significantly from the expected 3:1 Mendelian ratios in 71% of the (F1 non‐VRT × F1 VRT) crosses and in 63% of the (F1 VRT × F1 VRT) crosses. More VRT hybrids were produced in the latter cross because both parents carried the VRT (Appendix C: Table C2).
Backcrosses
In contrast to the previous crosses, overall the VRT segregated according to the expected Mendelian ratios in the backcrosses. The VRT segregated according to the expected ratios in all the families regardless of the kind of cross (F1 VRT × W, G = 1.14, df = 2, P > 0.05, or W × F1 VRT, G = 3.5, df = 3, P > 0.05; Fig. 1B; Appendix C: Table C3).
Progeny performance
Effect of the transgene and pollen load
We found no significant differences in performance between hybrids that inherited the transgene and those that did not among the F1, F2 or BC progeny (Appendix C: Tables C4 and C5). No overall effect of pollination treatment (saturation, 15:1, 7:1 and 3:1) was found either. No significant effect of the type of cross was found among the F2 (F1 non‐VRT × F1 VRT vs. F1 VRT × F1 VRT) or BC (F1 VRT × W vs. W × F1 VRT) progeny (Appendix C: Tables C4 and C5).
Performance of progeny and parental lines
Having found no significant effect of having the transgene on the vegetative performance of F1, F2 or BC progeny, we pooled all progeny data to compare performance amongst them and also against the parental lines.
We found significant differences among the different generations in all variables examined: days to germinate (F6, 143 = 2.73, P < 0.02), leaf area at 14 days (F6,25 = 86.79, P < 0.0001), staminate (F6,75 = 61.48, P < 0.0001) and pistillate (F6,75 = 8.42, P < 0.0001) flower production per plant, number of seeds per fruit (F5,98 = 16.38, P < 0.0001) and seed mass (F6,34 = 105.33, P < 0.0001; Fig. 2A–F). The effect of the maternal family nested within generation was not significant except for seed mass (F6,34 = 1.64, P < 0.05).
Performance of parentals (C. pepo VRT, wild C. argyrosperma ssp. sororia) and hybrids (F1, F2) and backcrosses. Seedling stage: (A) days to germinate (n = 3948) and (B) leaf area 14 days after germination (n = 340), reproductive stage: number of (C) female (n = 928) and (D) male (n = 906) flowers per plant, (E) average seed mass (n = 1957) and (F) number of seeds per fruit (n = 148). All values are back‐transformed least‐squares means ± s.e.; means with different letters differ significantly from each other according to a Tukey‐Kramer multiple comparisons procedure. We obtained legal permits to exclusively cultivate VRT Cucurbita pepo enclosed in greenhouses and these plants did not produce seeds because of a lack of pollinators. This is the reason for the missing data point for VRT seed number.
According to Tukey multiple‐comparison tests among generations, F1 hybrids took significantly longer to emerge than either of the parental lines or F2 hybrids, and they tended to take as long to emerge as both BC progeny lines (Fig. 2A). Time to emergence of F2 hybrids and BC progeny was longer than that of the parental lines, but only significantly so from the wild parent. Leaf area was the greatest in the VRT parent and smallest in the wild parent, with all other generations intermediate and significantly different from both parents but not amongst themselves (Fig. 2B). Pistillate flower production was the highest among the VRT parental plants, and the lowest in the F1 progeny, with the wild parent and both F2 lines intermediate and significantly different from the F1 hybrids and the VRT parent. Both BC progenies were also intermediate, although given their slightly lower means, they did not differ significantly from the wild parent or the F1 hybrids (Fig. 2C).
A similar pattern, (but reversed for the parental lines) is seen in staminate flower production, with extreme high and low flower production for the wild and VRT parental lines respectively and F1 and F2 hybrids having also low, production, but significantly greater than VRT parentals. As for the BC progeny lines, they produced as many (F1 VRT × W) or more (W × F1 VRT) flowers as the hybrids. Notably, staminate flower production in the wild parent exceeded that of the other generations by almost one order of magnitude (Fig. 2D).
Similarly, wild parents produced almost four times more seeds per fruit than any of the other generations. No differences among progeny generations were detected for this variable (Fig. 2F). Average seed mass of hybrids was intermediate between (and significantly different from) both parental lines, although closer to that of the wild parent. Average seed mass differed significantly between the two lines of BC progeny: seeds of VRT mothers were larger than those of wild mothers. Interestingly, again, the backcross involving wild mothers produced seeds similar in mass to those of the wild parent (Fig. 2E).
Seed germination success, seedling survival and pollen viability
Overall, we detected significant differences among generations in the proportion of seeds that germinated (χ2 = 286.9, df = 6, P < 0.00001; Table 1). Germination was significantly greater in the parental lines than in the hybrid or BC progeny. Germination was 26% lower in F1 hybrids, and 8% (F1 non‐VRT × F1 VRT) and 11% (F1 VRT × F1 VRT) lower in F2 hybrids, compared to the wild parent. Germination in BC progeny was lower by 29% (F1 VRT × W) and 36% (W × F1 VRT) in relation to the wild parent. Germination rates of the hybrids and the BC progeny did not differ significantly from one another (Table 1).
Proportion of seeds germinated, proportion of plants that survived to reproduction, and proportion of viable pollen among parentals, hybrids (F1, F2) and backcrosses (BC).
Differences in survival to reproductive age among generations were less pronounced (χ2 = 93.6, df = 6, P < 0.00001). Survival of F1 progeny was 22% lower than that of the wild parent. Survival was also lower than that of the parental lines in F2 hybrids and BC progeny, but not significantly so. Among parental lines, survival was greater in wild than VRT plants (Table 1), but many of the latter were affected by a whitefly infestation in the greenhouse.
Pollen viability varied significantly among generations (χ2 = 4254, df = 4, P < 0.00001). Both parental lines had greater pollen viability than the hybrids and the BC progeny. The BC (W × F1 VRT) and the F2 (F1 VRT × F1 VRT) progeny had greater pollen viability than F1 hybrids (Table 1).
A potential effect of year is not supported by our analyses because only in two of the six performance variables were there differences between the F1 and the F2 or BC progeny (Table 2), despite the fact that F1 hybrids were grown in a different year than the others. Similarly, seed germination rates did not differ significantly between F1 hybrids and F2 or BC progeny. Also, pollen viability varied between F2 and BC progeny, despite being grown during the same year.
Reproductive rate (survival × fertility) of wild parents, hybrids (F1, F2), and backcrosses (BC progeny).
Reproductive rate
Reproductive rates of the F1 and F2 hybrids were lower than those of the wild parent (Table 2). Compared to the wild parent, reproductive rates through the female function were only 2–6% in the F1, 1% in the F2, and 9% in BC progeny (W × F1 VRT). Through the male function, reproductive rates of the F1, F2 (only progeny from F1 VRT × F1 VRT tested) and BC (W × F1 VRT) progeny were, respectively 6%, 0.9%, and 20% compared to the wild parent. Notably, reproductive rates of W × F1 VRT BC progeny, were ca. 2.5 and 51 times greater (female and male function, respectively) than those of the F1 generation. Overall, reproductive rates of hybrids (F1, F2) and BC progeny were greater than one, except in the F2 generation, which means that the number of hybrids and BC progeny in populations should increase.
Discussion
Our study showed that GM Cucurbita pepo can produce fertile hybrids with its wild relative, C. argyrosperma ssp. sororia. To our knowledge, we are the first to show that F1 progeny that inherited the transgene are capable of transmitting it to their progeny through crosses with other hybrids or in backcrosses. The transgene followed Mendelian inheritance as a dominant trait in a majority of families, and its frequency in the hybrid progeny increased as the transgenic:wild pollen ratio increased in pollen loads.
Inter‐specific transmissibility of the transgene
In crosses that used pure transgenic pollen on wild maternal plants, 44% of the progeny was F1 VRT and deviated from the expected Mendelian ratio, but such deviation was caused by the progeny produced in 6 out of 40 plant families. Mendelian segregation of the VRT occurred in 85% of the families, which is consistent with reports from intraspecific crosses in Beta vulgaris and C. pepo (Dietz‐Pfeilstetter and Kirchner 1998, Sasu et al. 2009).
Pollen of GM plants competed successfully with wild pollen in wild pistils as seeds were sired by both type of donors in pollen mixtures treatments. The pooled F1 progeny obtained from the other three pollen mixture treatments significantly deviated from Mendelian inheritance of the transgene. However, as the proportion of VRT pollen increased in mixed pollen loads on wild recipients, the transfer of the transgene to the progeny was more likely to follow Mendelian inheritance as a dominant trait in more families. Similarly, in these pollen mixture experiments, as the proportion of transgenic pollen in the load increased, the proportion of transgenic hybrid progeny increased too. These results suggest a density‐dependent effect over fertilization success of transgenic pollen in competition with the wild species. Although wild pollen in mixed loads reduced the frequency of hybrid VRT, microgametophytes of the modified crop were able to compete successfully in the style and fertilize wild ovules.
Our study with Cucurbita is an advancement with respect to other studies because we were able to identify transgenic and non‐transgenic F1 donors and recipients before conducting the experimental crosses. This allowed us to clearly establish the inheritance of the transgenes in F2 hybrids and BC progeny. In the pooled F2, we observed deviations from Mendelian inheritance in the transgene frequency. However, the F2 progeny of 71% of the families derived from crosses of F1 VRT pollen on F1 non‐VRT recipients followed Mendelian inheritance. Similarly, the F2 progeny of 63% of the families derived from F1 VRT donor and recipient crosses followed Mendelian inheritance. Despite deviations from Mendelian inheritance of the transgene in interspecific hybridizations (possibly due to hybrid inter‐specific cross‐compatibility issues related to fertilization, seed abortion, or differential seed germination), we demonstrated that transgenes can be transferred from C. pepo VRT to the wild relative C. argyrosperma ssp. sororia with stability.
Inheritance of the VRT among BC progeny followed Mendelian expectations regardless of whether the F1 VRT acted as donors or recipients in crosses with the wild relative. The transgenic virus resistance construct is inherited as a single dominant trait in an interspecific hybridization system and can be incorporated in the gene pool of the wild relative through backcrossing. A consistent segregation of the transgene across several generations suggests that the transgenic construct is stable despite the possibility of genetic recombination during meiosis (Stewart et al. 2003, Bicar 2009).
In the crop/wild interface landscape, cultivars of C. pepo and wild populations of C. argyrosperma ssp. sororia coincide in different regions of Mexico (Arriaga et al. 2006). Both cultivated C. pepo and the wild gourds, share pollinators and bloom during similar periods throughout the year (Hurd et al. 1971, Lira‐Saade 1995). Xenoglossa and Peponapis bees have maintained a mutualistic evolutionary relationship with
Performance of hybrid and backcross progeny
Low hybrid viability could hinder the escape of transgenes to wild populations. In our study, hybrid and BC progeny generally showed both reduced vegetative and reproductive performance compared to the wild parent, and thus, we may conclude that introgressed progeny are not good competitors. However, we must take into account the variability in these measures. Transgenic and non‐transgenic hybrid performance is variable (Arnold and Hodges 1995, Hails and Morley 2005), and it may be equal to that of the parents (Arnold and Hodges 1995, Pertl et al. 2002, Ammitzbøll et al. 2005). Thus, an F1 hybrid generation with low fitness does not necessarily prevent transgene introgression, it merely slows down the process (Hails and Morley 2005). Although F1 hybrids of C. pepo VRT and C. argyrosperma ssp. sororia had lower fitness than the wild parent, they were fertile and therefore could allow the flow of genes (including transgenes) to future generations. Interestingly, experimentally generated hybrids of free‐living C. pepo ssp. texana and a transgenic cultivar of C. pepo ssp. pepo resistant to two viruses also had lower fitness than the wild parent (hybrid fitness relative to wild parent of 15–53%; Spencer and Snow 2001). One reason why the decrease in fitness reported in that study was not as severe as the one we found could be that the parental lines of those hybrids were more closely related (same species) than the parental lines we used. In either case, the F1 hybrids constitute a, at most, only a partial barrier to gene flow among transgenic crop and wild Cucurbita.
The vigor of F2 hybrids did not differ based on whether one or both parents were F1 VRT, but the overall fitness of the F2 generation was lower compared to the wild parents. Hybrid breakdown has also been shown to occur in the F2 Brassica napus and in herbicide‐resistant GM rice (Oryza sativa; Hauser et al. 1998, Zhang et al. 2003). Despite their low fitness, F2 hybrids remain a viable VRT reservoir from which further transgene introgression could occur.
The BC progeny of F1 hybrid donors and wild recipients had greater vegetative and reproductive performance than F1 hybrids. This suggests that crosses between F1 VRT and wild plants can generate progeny with improved fitness compared to the F1 hybrids. Similarly, in other GM hybrid systems, the backcross generations had greater fitness than the F1 and F2 generations (GM Brassica napus × B. rapa, and Raphanus raphanistrum × B. napus GM; Chèvre et al. 1997, Metz et al. 1997, Ammitzbøll et al. 2005). It is noteworthy that the inheritance of the transgene to BC progeny showed no deviations from Mendelian inheritance and that BC progeny had greater fitness than the F1 hybrids, when the maternal parents of the BC progeny were wild. This BC progeny could have a greater probability of surviving, reproducing, and transmitting the VRT than either of the hybrid generations.
Other factors may affect the fate of transgenes in wild populations after hybridization. VRT hybrids in environments with high incidence of viruses would have selective advantages over susceptible plants, and consequently should increase in frequency (Laughlin et al. 2009, Sasu et al. 2009). In contrast, transgene expression may have negative effects on the fitness of VRT plants, and thus, could decrease their frequency in a population. For example, indirect costs in VRT BC squash occurred because these plants attracted more herbivores (chrysomelid beetles) than non‐VRT BC susceptible to the virus but suffered greater infection by the lethal bacterium Erwinia as it was carried by the herbivores (Sasu et al. 2009). Studies are required to determine if similar costs occur in VRT hybrids in the tropics where ecological interactions could be more complex.
Concluding remarks
In conclusion, our study demonstrated that a virus‐resistance transgene in the cultivated squash, Cucurbita pepo, can be transferred to its distant wild relative Cucurbita argyrosperma ssp. sororia, and may result in introgression of transgenic sequences in natural populations in the region of origin and domestication of Cucurbita. The transgene is likely to follow Mendelian inheritance as the proportion of VRT pollen increases and under different pollen competition scenarios. The results of our study should be considered in risk assessment analyses for the introduction of GM squash cultivars at the site of origin of the genus. Future studies are needed to assess whether viral‐resistance transgenes provide these hybrids with a competitive advantage in wild conditions, and to determine if there are any direct or indirect costs of transgene expression in hybrids and BC progeny.
Acknowledgments
Permits to use GMO zucchini plants for this study in Mexico were granted by SAGARPA (0330, 0390, 02034). The authors thank all support staff at Laboratorio de Ecología Evolutiva y Conservación de Bosques Tropicales, IIES, and Escuela Nacional de Estudios Superiores, Unidad Morelia, UNAM for help in the lab, greenhouse, and experimental plot, and for providing the space for the experimental plot and the use of their facilities. Financial support for this study was provided by Consejo Nacional de Ciencia y Tecnología de México (174071/131008), Proyecto Laboratorio Nacional de Análisis y Síntesis Ecológica para la Conservación de Recursos Genéticos 2015‐2‐250996 and Programa de Apoyo a Proyectos de Innovación Tecnológica‐Universidad Nacional Autónoma de México (IN212714) to M. Quesada, and from Programa de Estancias de Investigación y Docencia, UNAM to G. Ávila‐Sakar. This study was performed in partial fulfillment of the requirements for the Ph.D. degree of R. Cruz‐Reyes at Posgrado en Ciencias Biológicas, UNAM.
Supplemental Material
Ecological Archives
Appendices A–C are available online:
You have requested "on-the-fly" machine translation of selected content from our databases. This functionality is provided solely for your convenience and is in no way intended to replace human translation. Show full disclaimer
Neither ProQuest nor its licensors make any representations or warranties with respect to the translations. The translations are automatically generated "AS IS" and "AS AVAILABLE" and are not retained in our systems. PROQUEST AND ITS LICENSORS SPECIFICALLY DISCLAIM ANY AND ALL EXPRESS OR IMPLIED WARRANTIES, INCLUDING WITHOUT LIMITATION, ANY WARRANTIES FOR AVAILABILITY, ACCURACY, TIMELINESS, COMPLETENESS, NON-INFRINGMENT, MERCHANTABILITY OR FITNESS FOR A PARTICULAR PURPOSE. Your use of the translations is subject to all use restrictions contained in your Electronic Products License Agreement and by using the translation functionality you agree to forgo any and all claims against ProQuest or its licensors for your use of the translation functionality and any output derived there from. Hide full disclaimer
© 2015. This work is published under http://creativecommons.org/licenses/by/3.0/ (the “License”). Notwithstanding the ProQuest Terms and Conditions, you may use this content in accordance with the terms of the License.
Abstract
The introgression of modified genetic sequences (transgenes) into wild populations hinges on the generation of viable and fertile hybrids from crosses between transgenic cultivars and their wild relatives. We assessed the risk of escape of a gene construct that confers resistance to three viruses (ZYMV, WMV, CMV) from a transgenic zucchini (Cucurbita pepo) to Cucurbita argyrosperma ssp. sororia, a wild relative at the center of origin of the genus in Mexico. We experimentally generated first and second generation hybrids, as well as backcross progeny (BC), and evaluated their performance. The virus‐resistance transgene was successfully inherited into both hybrid generations and also to BC progeny from the crosses of hybrids with the wild relative. The transgene generally followed Mendelian inheritance as a dominant trait. Both hybrid generations and the BC progeny had lower reproductive output compared to the wild parent. Given that the hybrid and BC progeny were viable and fertile, the escape and persistence of the transgene is possible via wild populations of C. argyrosperma ssp. sororia. This information is essential for biosafety policy in Mexico—center of origin and diversification of several crops—where the liberation of genetically modified plants is currently under approval, and risk assessment is necessary.
You have requested "on-the-fly" machine translation of selected content from our databases. This functionality is provided solely for your convenience and is in no way intended to replace human translation. Show full disclaimer
Neither ProQuest nor its licensors make any representations or warranties with respect to the translations. The translations are automatically generated "AS IS" and "AS AVAILABLE" and are not retained in our systems. PROQUEST AND ITS LICENSORS SPECIFICALLY DISCLAIM ANY AND ALL EXPRESS OR IMPLIED WARRANTIES, INCLUDING WITHOUT LIMITATION, ANY WARRANTIES FOR AVAILABILITY, ACCURACY, TIMELINESS, COMPLETENESS, NON-INFRINGMENT, MERCHANTABILITY OR FITNESS FOR A PARTICULAR PURPOSE. Your use of the translations is subject to all use restrictions contained in your Electronic Products License Agreement and by using the translation functionality you agree to forgo any and all claims against ProQuest or its licensors for your use of the translation functionality and any output derived there from. Hide full disclaimer
Details
1 Unidad Académica en Desarrollo Sustentable, Universidad Autónoma de Guerrero, Tecpan, Guerrero 40900 México; Laboratorio Nacional de Análisis y Síntesis Ecológica para la Conservación de los Recursos Genéticos, Escuela Nacional de Estudios Superiores Unidad Morelia, and Instituto de Investigaciones de Ecosistemas y Sustentabilidad, Universidad Nacional Autónoma de México, Morelia Michoacán 58190 México
2 Laboratorio Nacional de Análisis y Síntesis Ecológica para la Conservación de los Recursos Genéticos, Escuela Nacional de Estudios Superiores Unidad Morelia, and Instituto de Investigaciones de Ecosistemas y Sustentabilidad, Universidad Nacional Autónoma de México, Morelia Michoacán 58190 México; Department of Biology, The University of Winnipeg, Winnipeg, Manitoba R3B 2E9 Canada
3 Laboratorio Nacional de Análisis y Síntesis Ecológica para la Conservación de los Recursos Genéticos, Escuela Nacional de Estudios Superiores Unidad Morelia, and Instituto de Investigaciones de Ecosistemas y Sustentabilidad, Universidad Nacional Autónoma de México, Morelia Michoacán 58190 México