SEARCH

SEARCH BY CITATION

Keywords:

  • autogamous species;
  • hybridization;
  • inheritance;
  • introgression;
  • invasive species;
  • plants with new traits;
  • selection

Summary

  1. Top of page
  2. Summary
  3. Introduction
  4. Methods
  5. Results
  6. Discussion
  7. Acknowledgements
  8. References
  • 1
    Hybridization seems possible for many crop species after pollen transfer from crops to wild relatives in the surrounding vegetation. Subsequent introgression of crop-specific traits into wild relatives could lead to invasive introgressants. This process has become a public concern following the introduction of genetically modified (GM) crops. Until now, few studies have used demographic vital rates to compare the performance of hybrids with their wild relatives.
  • 2
    We created second-generation (S1 and BC1) hybrids between the non-transgenic crop Lactuca sativa and its entirely cross-fertile wild relative Lactuca serriola. Seeds of parents and hybrids were individually sown in field plots at three different locations. Next to germination and survival, we measured a range of single fitness components and morphological traits. We also compared observed phenotypes to phenotypes theoretically expected, according to different inheritance scenarios.
  • 3
    Phenotypes of both hybrid classes resembled L. serriola closely, and more than theoretically expected. However, demographic vital rates, i.e. germination and survival of hybrids were much higher than in L. serriola.
  • 4
    Our results indicate that hybrids between crop and wild Lactuca are phenotypically indistinguishable from the wild relative and thus will largely remain unnoticed when they occur. However, these hybrids could potentially become invasive because of substantial differences in vital rates and seeds returned per seed sown.
  • 5
    Synthesis and applications. A comparative study on single fitness components, such as seed production, alone would not have revealed the performance advantage of crop–wild hybrids in Lactuca. Therefore, studying demographic vital rates of hybrids and back-crosses to test for long-term consequences of hybridization should be part of any risk assessment of GM crops. Demographic vital rates are also important for the development of predictive modelling tools that can be employed to test the individual- and population-level consequences of new-to-add traits.

Introduction

  1. Top of page
  2. Summary
  3. Introduction
  4. Methods
  5. Results
  6. Discussion
  7. Acknowledgements
  8. References

Hybridization via pollen transfer between crops and wild relatives in the surrounding vegetation seems possible for many species (Beringer 2000; Snow 2002; Ellstrand 2003; den Nijs, Bartsch & Sweet 2004). Hence newly introduced (crop) genes could potentially escape into wild populations in agricultural areas, providing competitive advantage through insect, fungal and pest resistance and salt and drought tolerance (Snow 2002; Song et al. 2003; Stewart, Halfhill & Warwick 2003). A changed ecology could result, causing an increased invasiveness of introgressants (Snow & Palma 1997; Ellstrand & Schierenbeck 2000). In recent years, hybridization has therefore become a public concern following the introduction of genetically modified (GM) crops (Hails 2000; Gray 2004).

The first step in the introgression process, gene flow from crops to sympatric wild relatives, has already been demonstrated for most important crop species world-wide (Ellstrand 2003). In addition, the evidence of actual transgene escape is growing (Messeguer et al. 2001; Lu, Kato & Kakihara 2002; Massinga et al. 2003; Watrud et al. 2004). However, we still have limited understanding of the long-term consequences of hybridization between crops and wild relatives. The main question herein is: can introgressants be potentially invasive in non-agricultural habitats in the long term? Answers to this question are needed to assess the actual risks and develop predictive tools (Hails 2000; Gray 2004).

To address this knowledge deficit, a variety of fitness comparisons between crop–wild relative hybrids and their wild parental species have been conducted over the last decade. Comprehensive reviews are found in Ellstrand (2003) and Stewart, Halfhill & Warwick (2003), while Snow et al. (2003), Fuchs et al. (2004) and Song et al. (2004) are the most recent studies using actual transgenic crops. Results across species, however, are often conflicting, depending on initial conditions (Hauser, Damgaard & Jørgensen 2003), costs of new trait expression (Bergelson 1994a; Beringer 2000), and the ploidy level (van Tienderen 2004). Risk assessment of future-added traits into crops will therefore still need to apply a case-by-case scenario (Beringer 2000). In this study, we have demonstrated a promising applied approach to test hybrid performance and generated data suitable for a subsequent risk assessment of plants with novel traits.

Most studies of hybridization between crops and wild relatives include single fitness components, such as fecundity, but lack any information on demographic vital rates, i.e. germination and survival (van Tienderen 2004). Yet such data are of high importance to determine the sensitivity of species to ecological changes by the incorporation of novel traits, and to develop predictive models to assess the fate of introgressed populations (Bullock 1999; Gray 2004). Such models make it possible to test whether, for example, increased fecundity or decreased predation will increase invasiveness on a longer time scale. Moreover, vital rates encompass a range of single components, including the net effect of interactions between them, and seem therefore a more reliable estimate of overall performance (Conner, Glare & Nap 2003).

Not all traits introgress and are expressed with the same likelihood (Shim & Jørgensen 2000; Stewart, Halfhill & Warwick 2003): genetic processes such as purifying selection, linkage disequilibrium and gene masking might cause deviation of single traits from (neutral) Mendelian expectations. For example, Fuchs et al. (2004) found that after one back-cross most hybrids regained most wild-type characteristics, becoming morphologically indistinguishable, but still expressed the added transgenic virus resistance. Clearly, information about inheritance of single fitness components and morphological characteristics is important to derive estimations of the number of generations after which hybrids are no longer identifiable in field surveys.

Autogamous (selfing) species have (by definition) low out-crossing rates and were therefore mostly excluded as model systems for crop–wild hybridization. However, recent studies on autogamous species, mostly transgenic rice, have shown that even out-crossing rates of 1% can lead to a substantial gene flow into wild relatives (Song et al. 2003; Chen et al. 2004).

In this paper we present results of fitness and demographic vital rates of second-generation hybrids between non-transgenic lettuce Lactuca sativa L. and its conspecific wild relative prickly lettuce Lactuca serriola L. (Asteraceae). We examined the morphological differences between parents and hybrids, as well as estimated Mendelian inheritance of single phenotypic fitness components.

Specifically, we asked the following questions, which we then discussed in the light of future risk assessment (RA) and the development of predictive RA tools. (i) Are there morphological differences between second-generation hybrids and their parental species L. serriola and L. sativa? (ii) Do second-generation hybrids differ in demographic vital rates from their parental species? (iii) Are there significant differences in single fitness components between second-generation hybrids and their parental species? (iv) Do inheritance patterns for single fitness components of second-generation hybrids follow Mendelian segregation?

Methods

  1. Top of page
  2. Summary
  3. Introduction
  4. Methods
  5. Results
  6. Discussion
  7. Acknowledgements
  8. References

species studied

Individual plants of the wild L. serriola (2n = 18) are between 0·80 and 1·20 m tall, with bluish green compassing leaves that contain white bitter latex. Two forms have been distinguished on the basis of leaf shape: f. integrifolia, with entire, and f. serriola, with serrate leaves. The stem base and downside leaf midribs have characteristic 2-mm long spines. In The Netherlands, L. serriola is a summer and winter annual that flowers in July–August. Capitula are small (1·5 cm) and contain 10–20 yellow florets, which develop into brown seeds. The involucral bracts are reflexed when the seeds are ripe (van der Meijden 1996). Plants are predominantly autogamous, but 1–5% out-crossing via insect pollen vectors is reported (Thompson et al. 1958; D. A. P. Hooftman et al., unpublished data). Lactuca serriola is mainly found in anthropogenically disturbed habitats, such as roadsides, railways and urban areas, throughout Europe, northern Africa and North America (Zohary 1991; de Vries, van der Meijden & Brandenburg 1992). Recently, the species has been reported spreading rapidly in The Netherlands (Frietema de Vries 1996).

Lactuca sativa (2n = 18) has been known as a crop species since 2500 bc and most probably originates from the eastern Mediterranean area (Frietema de Vries 1996; Ryder 1999). Lactuca sativa is considered to be conspecific with L. serriola because the two taxa are fully cross-fertile (Koopman, Zevenbergen & van den Berg 2001). Lactuca sativa differs from L. serriola in that it lacks the spines on leaves and stem, has whitish instead of brown seeds, has erect instead of reflexed involucral bracts and produces the typical butterhead. Lactuca sativa is commercially grown in greenhouses in the western part of The Netherlands, which provide limited possibilities for escape. However, most Dutch cities and villages have small-scale non-commercial vegetable gardening complexes at their edges, often situated along the roads and railway systems where L. serriola occurs. Simultaneously with L. serriola, unharvested L. sativa bolts and flowers regularly in these gardens, providing an unpredictable, spatially heterogeneous pollen source available for cross-pollination.

Both Lactuca species are diploid, as are their hybrids, sharing equal chromosome length and genetic make-up (Frietema de Vries 1996; Koopman, Zevenbergen & van den Berg 2001). Therefore expression of barriers to successful meiosis is unlikely (van Tienderen 2004).

hybridization

F1 progeny was created by using L. serriola as mother plants and L. sativa as pollen donors. Back-crossing (BC1) was done by using F1 hybrids maternally and L. serriola as pollen donors, mimicking the most likely occurring scenario. F1 hybrids proved to be extremely viable under greenhouse conditions, with low mortality and full seed set. In agreement with earlier studies on Lactuca (de Vries 1990; Zohary 1991; Bergelson 1994a), we did not find male or female sterility or irregularities in hybrid growth.

Progeny of the exact L. serriola seed lineages, which generated the F1 hybrids, was used for back-crossing. We let F1 hybrids self-pollinate to obtain a second-generation line of autogamous hybrid progeny (hereafter called S1). The crossing technique followed protocols by Nagata (1992) and Ryder (1999), including both full emasculation by removal of all stamens before elongating of the styles and subsequent removal of remnant pollen by washing with a fine mist of water to avoid spurious autogamy. This method results in a very high percentage of crossings, approaching 100% (Ryder 1999). To ascertain individual plant hybrid status, before producing BC1 and S1 progeny we conducted a pilot study (D. A. P. Hooftman et al., unpublished data), sampling 18 F1 parental combinations using Amplified Fragment Length Polymorphism (AFLP). In short, we followed the protocol of Vos et al. (1995) with minor modifications. Two primer combinations (EcoACA–MseCGG and EcoATG–MseCTC) (Biolegio BV, Malden, The Netherlands) were applied. The final product was loaded on a Li-cor 6·5% denaturing polyacrylamide gel (Li-cor Biosciences, Lincoln, Nebraska); subsequent banding patterns were scored manually. For hybrid identification we used 10 L. sativa and 16 L. serriola private alleles, which were dominantly present in the F1; confirmed hybrids contained all paternal alleles. Hybrid status was subsequently linked to F1 morphology, allowing identification of further spurious L. serriola autogamous progeny, which were subsequently removed from the crossing scheme.

The crossing scheme included two widely used L. sativa cultivars and two L. serriola populations. The L. sativa cultivars were (i) Dynamite (Nunhems zaden, 01-2002, 60826), a new commercial cultivar with high mildew and aphid resistance; and (ii) Meikoningin (May Queen) (Nunhems zaden, 2001, 816), an older ‘amateur’ cultivar with fewer known resistances.

In each of two populations of L. serriola we randomly collected 10 seed lineages in August 2001. The first population, Eys, is from the southern part of The Netherlands (N50°49′, E05°55′), occurring in a grass-dominated vegetation on a south-facing calcareous slope. This population has been present at the same location for several decades (J. C. M. den Nijs & J. G. B. Oostermeijer, personal observations). The second population is from Zarrentin, northern Germany (N53°32′, E10°54′), and is a typical transient population from an anthropogenically disturbed site around a sand quarry.

The four crossing lineages per hybrid class (S1, BC1) provided by this crossing scheme, the two wild populations and two cultivars, will be referred to as ‘main lines’. Seed lineages within main lines are subsequently named ‘sublines’. The cultivars were considered to be genetically monomorphic.

We did not include F1 hybrids in this study because of the high likelihood of strong heterosis, which would provide little evidence for increased hybrid fitness (Klinger & Ellstrand 1994). Instead, we used subsequent (back-cross) generations, in which such heterosis effects are likely to be reduced (Arnold & Hodges 1995).

experimental set-up

Experimental field plots were created at three locations: Amsterdam, The Netherlands (N52°21′, E04°58′), Sijbekarspel, The Netherlands (N52°42′, E05°00′) and Aachen, Germany (N50°46′, E06°03′). At every location, two 100-m2 plots were made and ploughed to create soil disturbance before the experiment. We allowed for competition from the spontaneously emerging pioneer vegetation. In April 2003, 750 seeds were sown individually in each plot in a 30 × 30 cm grid (total 4500 seeds), orthogonally divided over the main lines. The grid positions of seeds within plots were fully randomized.

We assessed the following morphological characteristics, based on differences between both parental species: (i) the presence of spines on the leaf and/or (ii) stem base; (iii) the type of leaf margin (serrate, rounded, or intact); (iv) the forma (f. integrifolia or f. serriola); (v) the position of the involucral bracts upon fruit ripeness (erect or reflexed); (vi) the inflorescence shape (pyramid or spike); and (vii) the seed colour (white or brown). We measured demographic vital rates in the form of germination (presence of a seedling 1 month after sowing) and survival (a seedling developing into a reproducing adult). Individual plants that had not produced seeds by December 2003 were recorded as not having survived. As single fitness components we first measured the number of leaves plant−1 at the start of the bolting phase. Subsequently, at the first stage of seed set, we determined various reproductive components: the number of reproductive basal shoots, the number of branches and the length of the main shoot. Seeds from 15 randomly chosen capitula plant−1 were sampled to assess seed production capitulum−1 and average seed mass. For the latter variable, a batch of 50 randomly sampled seeds plant−1 was weighed on a microbalance.

The number of capitula was counted on 315 harvested individuals, orthogonally distributed over the main lines. We performed linear regression of the number of capitula on the number of branches and basal shoots (R2= 0·51, P < 0·001), which yielded the following equation:

  • image( eqn 1 )

This equation was subsequently used to estimate the amount of capitula for all surviving plants (including the ones that were initially completely counted). The total seed output was then calculated by multiplying the estimated amount of capitula plant−1 with the average number of seeds capitulum−1 plant−1. Moreover, we included a combination of the vital rate and single component approach by multiplying the vital rates plant−1 with the total seed output plant−1. This gave an estimation of the number of seeds returned per seed sown.

expected inheritance values

Deviations from Mendelian inheritance were tested by comparing the observed phenotype of hybrids (pooled data of main line × plot) against the expected phenotype for single components (Lynch & Walsh 1997; Keller, Kollmann & Edwards 2001). Components measured on S1 hybrids were tested against the expected mean of both parental lines:

  • image( eqn 2 )

in which P1 = L. serriola and P2 = L. sativa

For the BC1 we used a hierarchical scheme of three different models: (i) neutral inheritance (cf. Lynch & Walsh 1997), (ii) including wild-type genetic dominance and (iii) including 50% heterosis (cf. Keller, Kollmann & Edwards 2001):

  • image( eqn 3 )
  • image( eqn 4 )
  • image( eqn 5 )

statistical analyses

To test for morphological differences between parents and hybrids, the morphological variables were ordinated with discriminant analysis using four classes (L. serriola and L. sativa, and S1 and BC1 hybrids) and two canonical functions (Jongman, ter Braak & van Tongeren 1987). All quantitative variables were analysed with a hierarchical anova with sequential sum of squares (procedure Generalized Linear Models (GLM) in the statistical package SPSS; SPSS Inc., Chicago, IL) and using the F distribution. By using a hierarchical design we were able to account for the variation explained by differences among locations and plots prior to testing the effects of hybridization, which was of primary interest. Interaction terms that were not significant were removed from the analyses. Subsequently, Tukey's post-hoc Honest Significant Difference (HSD) comparison of means was performed among classes. Prior to analysis, all quantitative variables were log-transformed to improve normality. Expected inheritance of single fitness components, as described above, were tested in a similar way against the observed values

For germination and survival, we conducted an analysis of deviance (genstat; Payne et al. 1995) with the complementary log-log link (Candy 1986). We calculated ratios of the mean changes in deviance, which approximately follow the F-distribution (Payne et al. 1995).

Results

  1. Top of page
  2. Summary
  3. Introduction
  4. Methods
  5. Results
  6. Discussion
  7. Acknowledgements
  8. References

morphology

The first canonical function explained 90% of all morphological variance (Fig. 1). Correlated to this function were the involucral bract position (correlation coefficient 0·626), the presence of spines on the leaf midrib (0·609) and stem base (0·443), and the shape of the inflorescence (0·388). The type of leaf margin correlated with the second canonical function (correlation coefficient 0·591), explaining 10% of all variation. Seed colour and forma were not significantly correlated with either of the two functions, explaining very low amounts of variation (0·1%).

image

Figure 1. Discriminant analysis on four different plant groups, L. serriola, L. sativa, S1 and BC1, based on seven morphological characteristics. Only the first two canonical functions were used; the first function explains 90·3% (λ = 3·2), the second function explains an additional 9·6% (λ = 0·34) of all morphological variation.

Download figure to PowerPoint

As expected, L. sativa was clearly distinct from the other classes (Fig. 1), showing little morphological variance. Centroids of L. serriola and BC1 hybrids were not distinguishable from each other, although BC1 hybrids had a higher variance (Fig. 1). The position of the S1 centroid was roughly intermediate between the parents, and was clearly separated from both BC1 hybrids and L. serriola. The S1 showed the highest morphological variance of all classes, largely overlapping the other classes (Fig. 1).

demographic vital rates

Germination was significantly lower for L. serriola compared with L. sativa and both hybrid classes (P < 0·001; Table 1 and Fig. 2a). No differences were found between L. sativa and hybrid classes nor among the hybrid classes. The survival of hybrids was significantly higher than that of L. serriola (250%) and L. sativa (170%, P < 0·001; Table 1 and Fig. 2b). Survival was not different among the two hybrid classes (P > 0·05; Table 1). Significant differences among the main lines were caused by different performances among the parental lines within classes. Within L. serriola, Eys had higher survival than Zarrentin (P < 0·01, Tukey's HSD); within L. sativa, Meikoningin had higher survival than Dynamite (P < 0·001, Tukey's HSD).

Table 1.  anova for fitness components and analysis of deviance for demographic vital rates, testing for among-class differences. Sequential sum of squares was employed. Class includes L. serriola, L. sativa, S1 and BC1. For F calculation: location was tested against the among-plot variation; class was tested against plot (location) × main line interaction; mainline was tested against the among-subline variation. For Tukey's HSD post-hoc results see Figs 2 and 3
 d.f.Demographic vital ratesFitness components
Germination n = 4500Survival n v= 1201Capitula n = 555Leaves n = 1046Basal shoots n = 724Height main shoot n = 724Seeds capitulum−1 n = 655Seed output n = 520Seed mass n = 648Seeds returned n = 4500
  • *

    P < 0·05;

  • **

    P < 0·01;

  • ***

    P < 0·001.

  • Analysis of deviance.

  • These components could not be measured in Aachen plot 2 for L. serriola: none survived, consequently the degrees of freedom is less.

Location  239·8 6·7321·297·226·7 8·837·0419·4 359*107
Plot  1 0·2010·2**18·9*** 9·35**10·3**14·1***4·48* 6·27* 0·03  0·02
Class  3 8·50***25·6***80·9***27·7***28·3***87·7***2·6726·0***25·6*** 11·9***
Main line  8 0·41 3·49** 2·25* 1·68 1·38 3·19**6·32*** 1·04 0·65  2·50*
Subline113 1·25 1·28 1·06 1·60*** 1·87*** 0·901·13 0·88 1·08  1·52***
Plot (location) × main line 57 4·52*** 2·05*** 1·66** 3·42*** 4·15*** 1·80***1·63** 1·06 0·88  4·34**
Error n–184          
image

Figure 2. Mean ± SEM of demographic vital rates for L. serriola, L. sativa and the S1 and BC1 hybrid classes. The letters in each bar show the results of Tukey's HSD post-hoc comparison of means; if two bars have no letter in common, the means differ significantly (P < 0·05). (a) Germination rate of individual seeds; (b) survival rate until seed set of individual plants, after they had germinated.

Download figure to PowerPoint

single fitness components

Most single fitness components showed significant variation among plots (Table 1), caused by significantly lower performance for all components in Aachen (P < 0·01, Tukey's HSD). Significant among-class differences were found for all measured components except number of seeds capitulum−1 (P < 0·001; Table 1). In all cases, both hybrid classes differed significantly from L. sativa (P < 0·05) but not from L. serriola for number of shoots, shoot height and estimated seed output (P > 0·05; Fig. 3). S1 hybrids differed significantly from L. serriola in seed mass (P < 0·001) but BC1 hybrids did not (P > 0·05; Fig. 3). Both hybrid classes returned significantly higher numbers of seeds per seed sown than L. serriola, with no significant difference among hybrid classes (Fig. 3). Components not shown in Fig. 3 showed similar patterns to the seed output, except for the numbers of seeds capitulum−1.

image

Figure 3. Mean ± SEM for four different fitness components for L. serriola, L. sativa and the S1 and BC1 hybrid classes. The letters in each bar show the results of Tukey's HSD post-hoc comparison of means; if two bars have no letter in common, the means differ significantly (P < 0·05). (a) Number of leaves plant−1 at the start of the bolting phase; (b) average mass of a batch of 50 seeds plant−1; (c) estimated total seed-output plant−1; (d) number of seeds returned seed sown−1.

Download figure to PowerPoint

expected inheritance

For the S1 class, observed values for four fitness components differed significantly from the expected values (P < 0·05; Table 2). In those cases, S1 hybrids were phenotypically closer to L. serriola than expected. The germination rate was not significantly different from the average among parents. In contrast, the survival (Fig. 2) was significantly higher than expected (P < 0·001; Table 2).

Table 2.  anova for fitness components and demographic vital rates, testing observed data against expected inheritance patterns. Sequential sum of squares was employed. For F-calculation: the contrast expected vs. observed data and the among-expectation model variation (expected type) were tested against plot(location) × expected type interaction. Expected type for BC1 includes three hierarchic models: model 1, <0·75 × P1 + 0·25 × P2> model 2, <0·5 × P1 + 0·5 × S1 > model 3, <0·375 × P1 + 0·125 × P2 + 0·5 × S1> expected inheritance for S1, <0·5 × P1 + 0·5 × P2>
 d.f.Demographic vital ratesFitness components
GerminationSurvivalCapitulaLeavesBasal shootsHeight main shootSeeds capitulum−1Seed outputSeed massSeeds returned
  • *

    P < 0·05;

  • **

    P < 0·01;

  • ***

    P < 0·001.

  • These components could not be measured in Aachen plot 2 for L. serriola: none survived, consequently degrees of freedom is 4 less.

BC1 Location 23·0749·0***51·0***62·5***52·2***17·0***21·2***35·4***13·8**12·0***
Expected vs. observed 13·9350·3*** 8·47* 2·81 5·64*19·0**11·6*15·1**25·9***18·8***
Expected type 21·2310·5*** 0·60 1·11 0·50 1·27 1·23 0·33 6·84** 0·64
Plot(location) × expected type187·84*** 1·85*** 4·33*** 8·97*** 5·94*** 2·70** 1·70 3·61** 1·26 2·32**
Error54          
S1 Location 20·2417·7** 9·85*22·3***17·4** 6·34* 4·44 8·35* 0·50 6·99*
Expected vs. observed 13·2037·9*** 3·70 6·80* 3·11 7·16* 7·66* 5·22 6·82* 5·00
Plot(location) × expected type 89·59*** 2·88* 5·11**18·4*** 7·73*** 3·99** 0·90 8·60*** 2·24 3·57**
Error48          

The phenotype of the BC1 hybrids resembled that of L. serriola much more than expected: except for the number of leaves, all observed fitness components differed significantly from the values predicted by all three hierarchically ordered models (P < 0·01; Fig. 3). The observed germination rate was not significantly different from the expected values; in contrast, the observed survival and number of seeds returned per seed sown of BC1 hybrids was higher than expected for all models (P < 0·001; Table 2).

Discussion

  1. Top of page
  2. Summary
  3. Introduction
  4. Methods
  5. Results
  6. Discussion
  7. Acknowledgements
  8. References

Individual plants of second-generation hybrids between the crop L. sativa and its wild relative L. serriola are as fit as the wild species. Hardly any significant differences were detected based on single fitness components. Furthermore, back-crossed hybrids were morphologically indistinguishable from L. serriola. Second-generation hybrids derived by selfing of the F1 had a larger variation in morphological traits than the other classes, but still largely overlapped the BC and L. serriola class. Furthermore, the phenotype of hybrids resembled L. serriola much more than expected following inheritance estimates based on single fitness components.

In contrast, the demographic vital rates yielded different results: the germination rates of both second-generation hybrid classes were equal to L. sativa but two times higher than L. serriola. Even more strikingly, the probability of survival and consequently the number of seeds returned seed sown−1 of the hybrids was at least two times higher than both parental species.

Consequently, if hybridization does occur, this could lead to better performing and thus potentially more invasive (hybrid) genotypes (Snow & Palma 1997), whereas these hybrids are not detectable based on their phenotype, a similar situation as found in Helianthus (Snow et al. 1998; Fauré, Serieys & Berville 2002) and Cucurbita (Fuchs et al. 2004). However, better performing genotypes do not automatically result in higher invasiveness. There might be high fitness costs associated with the expression of added traits (Bergelson 1994a; Beringer 2000) or hybrids might show stronger density dependence (Hauser, Damgaard & Jørgensen 2003) compared with the pure wild relative. Our study represents a predominantly density-independent situation regarding Lactuca, but was performed at sites in which the soil situation and the spontaneous occurrence of surrounding weeds presented a natural selective habitat. Our results can therefore be considered a realistic scenario for the colonization of not yet or sparsely Lactuca-occupied anthropogenically disturbed habitat.

In contrast to Brassica, for example (Hauser, Jørgensen & Østergard 1998), second-generation hybrids between the crop and its wild relative did not provide a barrier to introgression, agreeing with initial expectations for Lactuca (Frietema de Vries 1996). We think that the observed higher survival might be a strong heterosis, exceeding the 50% tested for here, an observation similar to the high heterosis found in crisphead hybrids within L. sativa (Langton, Smith & Edmondson 1990). An explanation could be the fixation of parental lines for sets of alleles that have additive effects and are located on positively selected chromosomal blocks (Rieseberg, Archer & Wayne 1999). Alternatively, negative selection on mildly deleterious alleles could cause similar phenotypic patterns.

The inheritance discrepancies we observed could be caused by linkage disequilibrium in crop–wild hybrids (Stewart, Halfhill & Warwick 2003), i.e. the integration of crop-specific, neutral or positively selected, alleles in negatively selected chromosomal blocks. Consequently, crop-specific genes and associated traits are easily purged from the wild background genome, as shown by Shim & Jørgensen (2000) and Papa & Gepts (2003) in Daucus and Phaseolus, respectively.

Alternatively, crop genes and associated traits can behave recessively in hybrids between closely related or conspecific species (Bussell et al. 2002). Consequently, hybrid lines may phenotypically be more similar to the wild parental species than expected. The difference with linkage disequilibrium is that crop-specific genes might still be retained for future segregation.

Potential consequences of gene flow between crops and wild relatives have mostly been studied on predominantly out-crossing species. Surprisingly few studies have been conducted on predominantly autogamous species with inherently low out-crossing rates (Papa & Gepts 2003). More recently, after transgenic rice has become available, annual autogamous species have received more attention (Messeguer et al. 2001; Madsen, Valverde & Jensen 2002; Song et al. 2004).

Most autogamous species exhibit some degree of out-crossing (Song et al. 2003). Consequently, even after a rare out-crossing event, one F1 individual produces numerous selfed hybrid progeny (Song et al. 2003; Chen et al. 2004). Assuming that new (trans) genes exhibit neutral or simple dominance relationships and are therefore subjected to selection (Klinger & Ellstrand 1994), autogamous species should exhibit similar patterns as predominant out-crossers. This was recently shown by Papa & Gepts (2003), who demonstrated a hybrid zone of intermediate genotypes in Phaseolus, and by Song et al. (2004), detecting high performance of F1 hybrids in Oryza. Preliminary stochastic modelling of our demographic data showed that low out-crossing rates create a time lag only, but do not predominantly alter the long-term consequences (D. A. P. Hooftman et al., unpublished data).

Measuring demographic vital rates is of prime importance for studying ecological consequences of crop–wild hybridization (Bergelson 1994b; Hails 2000; Gray 2004). Our data clearly indicate that the performance advantage of hybrids would have been missed if vital rates had not been measured. Vital rates encompass a range of single components, including the net effect of interactions between them, and seem therefore a more reliable estimate of overall performance (Conner, Glare & Nap 2003). To date, only a few examples of similar approaches exist, such as Hauser, Jørgensen & Østergard (1998), who analysed a wide array of single fitness components. Several measures of fecundity-related vital rates, such as seed and seedling performance, proved to be useful as well (Linder & Schmitt 1995; Alexander et al. 2001).

We would like to stress the strong need for a more general approach, preferably based on combinations of demographic vital rates and fecundity (Bergelson 1994b; van Tienderen 2004): the measure used in this study, i.e. the number of seeds returned seed sown−1, is a primitive example of such approach. Predictive tools based on population models, measuring the risk of plant invasion by its finite rate of population increase (λ), seem highly promising (Gray 2004; van Tienderen 2004). Such models will provide useful insights in the sensitivity of the complete life cycle to novel traits (Bullock 1999). Only then will we be able to better evaluate and assess the future ecological consequences of (GM) crop to wild hybridization and, importantly, compare different species, life histories and added traits. Furthermore, including potentially different intraspecific density-dependence relations of hybrids compared with parental species (sensuHauser, Damgaard & Jørgensen 2003) would provide an opportunity to distinguish between performance during initial colonization of L. serriola bare habitat, as presented here, and long-term dynamics including both hybrids and L. serriola at one site. In the latter situation, the large amount of seeds returned per seed sown (Bullock 1999), representing a founder event, would probably decrease to more equilibrium levels.

Crop to wild gene flow seems likely for a wide array of species; good monitoring programmes, techniques and (computer-) applications to predict its consequences, expected occurrences and distribution of hybrids, are therefore important (Ellstrand 2003; den Nijs, Bartsch & Sweet 2004). For Lactuca, we have shown that the presence of indistinguishable but potentially invasive hybrids is possible, and should be anticipated when new (transgenic) traits would convey higher performance. This is the more relevant when crops and wild relatives are both diploid and genotypically similar (van Tienderen 2004).

In future risk assessment of crops with new traits, transgenic or not, every scenario should be tested case-by-case, depending on the nature of the added trait (Beringer 2000), the type of species (Stewart, Halfhill & Warwick 2003), and the presence of any sympatric wild relative (Ellstrand 2003; den Nijs, Bartsch & Sweet 2004). Rather than being based on single fitness components alone, a better indication of future potential invasiveness can be obtained by incorporating into stage-dependent models both empirical and estimated changes in vital rates that might result from new traits. At the same time this would make extensive multi-year field surveys less necessary.

Acknowledgements

  1. Top of page
  2. Summary
  3. Introduction
  4. Methods
  5. Results
  6. Discussion
  7. Acknowledgements
  8. References

We like to thank Peter van Tienderen, Thure Hauser, François Felber and all ANGEL members for fruitful discussions. Mark van Kleunen, Johannes Kollmann, Phil Hulme and three anonymous referees are acknowledged for their constructive comments on early versions. We thank all (field) assistants and technical staff for their hard work. We gratefully acknowledge Achim Gathmann and Ingolf Schuphan for providing the plots in Aachen; Justus Houthuesen established and maintained the plots in Sijbekarspel and Amsterdam. This study is part of the EC-funded project ‘ANGEL’ (EU-QLK3-2001-01657 to H. C. M. den Nijs).

References

  1. Top of page
  2. Summary
  3. Introduction
  4. Methods
  5. Results
  6. Discussion
  7. Acknowledgements
  8. References
  • Alexander, H.M., Cummings, C.L., Kahn, L. & Snow, A.A. (2001) Seed size variation and predation of seeds produced by wild and crop–wild sunflowers. American Journal of Botany, 88, 623627.
  • Arnold, M.L. & Hodges, S.A. (1995) Are natural hybrids fit or unfit relative to their parents? Trends in Ecology and Evolution, 10, 6771.
  • Bergelson, J. (1994a) The effects of genotype and the environment on costs of resistance in lettuce. American Naturalist, 143, 349359.
  • Bergelson, J. (1994b) Changes in fecundity do not predict invasiveness: a model study of transgenic plants. Ecology, 75, 249252.
  • Beringer, J.E. (2000) Releasing genetically modified organisms: will any harm outweigh any advantage. Journal of Applied Ecology, 37, 207214.
  • Bullock, J.M. (1999) Using population matrix models to target GMO risk assessment. Aspects of Applied Biology, 53, 205212.
  • Bussell, J.D., Waycott, M., Chappill, J.A. & James, S.H. (2002) Molecular phylogenetic analysis of the evolution of complex hybridity in Isotoma petraea. Evolution, 56, 12961302.
  • Candy, S.G. (1986) Fitting a parametric log–linear hazard function to grouped survival data. GLIM Newsletter, 13, 2831.
  • Chen, L.J., Lee, D.S., Song, Z.P., Suh, H.S. & Lu, B.R. (2004) Gene flow from cultivated rice (Oryza sativa) to its weedy and wild relatives. Annals of Botany, 93, 6773.
  • Con ner, A.J., Glare, T.R. & Nap, J.P. (2003) The release of genetically modified crops into the environment. Plant Journal, 33, 1946.
  • Ellstrand, N.C. (2003) Dangerous Liaisons? When Cultivated Plants Mate with Their Wild Relatives. John Hopkins University press, Baltimore, MD.
  • Ellstrand, N.C. & Schierenbeck, K.A. (2000) Hybridization as a stimulus for the evolution of invasiveness in plants? Proceedings of the National Academy of Sciences of the USA, 97, 70437050.
  • Fauré, N., Serieys, H. & Bervillé, A. (2002) Potential gene flow from cultivated sunflower to volunteer, wild Helianthus species in Europe. Agriculture Ecosystems and Environment, 89, 183190.
  • Frietema de Vries, F.T. (1996) Cultivated plants and the wild flora: effects analysis by dispersal codes . PhD Thesis. University of Leiden, Leiden, the Netherlands.
  • Fuchs, M., Chirco, E.M., McFerson, J.R. & Gonsalves, D. (2004) Comparative fitness of a wild squash species and three generations of hybrids between wild × virus-resistant transgenic squash. Environmental Biosafety Research, 3, 1728.
  • Gray, A.J. (2004) Ecology and government politics: the GM crop debate. Journal of Applied Ecology, 41, 110.
  • Hails, R.S. (2000) Genetically modified plants: the debate continues. Trends in Ecology and Evolution, 15, 1418.
  • Hauser, T.P., Damgaard, C. & Jørgensen, R.B. (2003) Frequency-dependent fitness of hybrids between oilseed rape (Brassica napus) and weedy B. rapa (Brassicaceae). American Journal of Botany, 90, 571578.
  • Hauser, T.P., Jørgensen, R.B. & Østergard, H. (1998) Fitness and backcross of F2 hybrids between weedy Brassica rapa and oilseed rape (B. napus). Heredity, 81, 436443.
  • Jongman, R.H., ter Braak, C.J.F. & van Tongeren, O.F.R. (1987) Data Analysis in Community and Landscape Ecology. Pudoc, Wageningen, the Netherlands.
  • Keller, M., Kollmann, J. & Edwards, P.J. (2001) Genetic introgression from distant provenances reduces fitness in local weed populations. Journal of Applied Ecology, 37, 647659.
  • Klinger, T. & Ellstrand, N.C. (1994) Engineered genes in wild populations: fitness of weed–crop hybrids of Raphanus sativus. Ecological Applications, 4, 117120.
  • Koopman, W.J.M., Zevenbergen, M.J. & van der Berg, R.G. (2001) Species relationships in Lactuca s.1. (Lactuceae, Asteraceae) inferred from AFLP fingerprints. American Journal of Botany, 88, 18811887.
  • Langton, F.A., Smith, J.W.M. & Edmondson, R.N. (1990) Heterosis in crisphead lettuce (Lactuca sativa L.) hybrids. Euphytica, 49, 1523.
  • Linder, C.R. & Schmitt, J. (1995) Potential persistence of escaped transgenes: performance of transgenic oil-modified Brassica seeds and seedlings. Ecological Applications, 5, 10561068.
  • Lu, C.M., Kato, M. & Kakihara, F. (2002) Destiny of a transgene escape from Brassica napus into Brassica rapa. Theoretical and Applied Genetics, 105, 7884.
  • Lynch, M. & Walsh, B. (1997) Genetics and Analysis of Quantitative Traits. Sinauer Associates, Sunderland, MA.
  • Madsen, K.H., Valverde, B.E. & Jensen, J.E. (2002) Risk assessment of herbicide resistant crops: a Latin American perspective using Oryza sativa as a model. Weed Technology, 16, 215223.
  • Massinga, R.A., Al-Khatib, K., St Armand, P. & Miller, J.F. (2003) Gene flow from imidazolinone-resistant domesticated sunflower to wild relatives. Weed Science, 51, 854862.
  • van der Meijden, R. (1996) Heukels Flora van Nederland. Wolters-Noordhoff, Groningen, the Netherlands.
  • Messeguer, J., Fogher, C., Guiderdoni, E., Marfa, V., Catala, M.M., Baldi, G. & Mele, E. (2001) Field assessments of gene flow from transgenic to cultivated rice (Oryza sativa L.) using a herbicide resistance gene as tracer marker. Theoretical and Applied Genetics, 103, 11511159.
  • Nagata, R.T. (1992) Clip-and-wash method of emasculation for lettuce. Hortiscience, 27, 907908.
  • den Nijs, J.C.M., Bartsch, D. & Sweet, J. (2004) Introgression from Genetically Modified Plants Into Wild Relatives. CAB International, Wallingford, UK.
  • Papa, R. & Gepts, P. (2003) Asymmetry of gene flow and differential geographical structure of molecular diversity in wild and domesticated common bean (Phaseolus vulgaris L.) from Mesoamerica. Theoretical and Applied Genetics, 106, 239250.
  • Payne, R.W., Lane, P.W., Digby, P.G.N., Harding, S.A., Leech, P.K., Morgan, G.W., Todd, A.D., Thompson, G., Tunnicliffe, W.G., Welham, S.J. & White, P.R. (1995) GENSTAT 5. Lawes Agricultural Trust, Harpenden, UK.
  • Rieseberg, L.H., Archer, M.A. & Wayne, R.Y. (1999) Transgressive segregation, adaptation and speciation. Heredity, 83, 363372.
  • Ryder, E.J. (1999) Lettuce, Endive and Chicory. CAB International, Wallingford, UK.
  • Shim, S.I. & Jørgensen, R.B. (2000) Genetic structure in cultivated and wild carrots (Daucus carota L.) revealed by AFLP analysis. Theoretical and Applied Genetics, 101, 227233.
  • Snow, A.A. (2002) Transgenic crops: why gene flow matters. Nature Biotechnology, 20, 542.
  • Snow, A.A. & Palma, P.M. (1997) Commercialization of transgenic plants: potential ecological risks. Bioscience, 47, 8696.
  • Snow, A.A., Palma, P.M., Rieseberg, L.H., Wszelaki, A. & Seiler, G.J. (1998) Fecundity, phenology, and seed dormancy of F1 wild–crop hybrids in sunflower (Helianthus annuus, Asteraceae). American Journal of Botany, 85, 794801.
  • Snow, A.A., Pilson, D., Rieseberg, L.H., Paulsen, M.J., Pleskac, N., Reagon, M.R., Wolf, D.E. & Selbo, S.M. (2003) A Bt transgene reduces herbivory and enhances fecundity in wild sunflowers. Ecological Applications, 13, 279286.
  • Song, Z.P., Lu, B.R., Wang, B. & Chen, J.K. (2004) Fitness estimation through performance comparison of F1 hybrids with their parental species Oryza rufipogon and O. sativa. Annals of Botany, 93, 311316.
  • Song, Z.P., Lu, B.R., Zhu, Y.G. & Chen, J.K. (2003) Gene flow from cultivated rice to the wild species Oryza rufipogon under experimental field conditions. New Phytologist, 157, 657665.
  • Stewart, C.N., Halfhill, M.D. & Warwick, S.I. (2003) Transgene introgression from genetically modified crops to their wild relatives. Nature Reviews Genetics, 4, 806817.
  • Thompson, R.C., Whitaker, T.W., Bohn, G.W. & van Horn, C.W. (1958) Natural cross-pollination in lettuce. American Society for Horticultural Science Proceedings, 72, 403409.
  • van Tienderen, P.H. (2004) Hybridization in nature: lessons for the introgression of transgenes into wild relatives. Introgression from Genetically Modified Plants into Wild Relatives (eds J. C. M. den Nijs, , J. C. M., Bartsch, D. & J. Sweet), pp. 726. CAB International, Wallingford, UK.
  • Vos, P., Hogers, R., Bleeker, M., Reijans, M., van de Lee, T., Hornes, M., Frijters, A., Pot, J., Peleman, J., Kuiper, M. & Zabeau, M. (1995) AFLP: a new technique for DNA fingerprinting. Nucleic Acid Research, 23, 44074414.
  • de Vries, I.M. (1990) Crossing experiments of lettuce cultivars and species (Lactuca sect. Lactuca, Compositae). Plant Systematics and Evolution, 171, 233248.
  • de Vries, F.T., van der Meijden, R. & Brandenburg, W.A. (1992) Botanical files: a study of real chances for spontaneous gene flow from cultivated plants to the wild flora of the Netherlands. Gorteria, 1 (Supplement), 1100.
  • Watrud, L.S., Lee, E.H., Fairbrother, A., Burdick, C., Reichman, J.R., Bollman, M., Storm, M., King, G. & van de Water, P.K. (2004) Evidence for landscape-level, pollen-mediated gene flow from genetically modified creeping bentgrass with CP4 EPSPS as a marker. Proceedings of the National Academy of Sciences of the USA, 101, 1453314538.
  • Zohary, D. (1991) The wild genetic resources of cultivated lettuce (Lactuca sativa L.). Euphytica, 53, 3135.