*For offprint requests: fax: (614)292-6345, email: snow.1@osu.edu

SUMMARY

Crop-to-wild hybridization has the potential to introduce beneficial traits into wild populations, thereby causing weedy plants to become more vigorous and abundant. With the commercialization of genetically engineered crops, in particular, the transfer of single genes can introduce traits such as resistance to herbicides, insect herbivores, disease, and environmental stress. Cultivated sunflower (H. annuus) hybridizes spontaneously with wild/weedy populations (also H. annuus), but little is known about the relative fitness of F₁ hybrids. In order to examine the extent to which crop genes can persist in wild sunflower populations, we compared fitness-related characteristics of F₁ wild-crop progeny with those of purely wild genotypes. First, we conducted crosses involving two cultivated lines and wild plants from three different regions (Texas, Kansas, and North Dakota), yielding wild x wild versus wild x crop offspring from each region. Seed burial experiments in the region of origin showed that wild-crop seeds had somewhat higher germination rates than did wild seeds from Kansas and North Dakota, while no differences were seen in Texas seeds (all had >90% germination). Progeny from each type of cross were grown outdoors in Ohio and Kansas to quantify lifetime fecundity. In Ohio, wild-crop plants from North Dakota were resistant to a rust that infected 53% of the purely wild genotypes. At both sites, flowering periods of hybrid and wild progeny overlapped most in plants from North Dakota and Texas, indicating that these hybrids are very likely to backcross with wild plants. In general, hybrid plants had fewer branches and produced fewer flower heads than wild plants, but in two crosses the fecundity of hybrids was not significantly different from that of purely wild genotypes. In summary, these results suggest that F₁ wild-crop hybrids had lower fitness than wild progeny, especially when grown under favorable conditions, but the F₁ barrier to the introgression of crop genes into wild populations is quite "permeable". Therefore, episodes of crop-to-wild hybridization are likely to lead to the long-term persistence of nondeleterious crop genes in wild populations.

Key words: Gene flow, wild sunflower, hybridization, introgression, escaped transgenes

INTRODUCTION

Crop-to-wild hybridization has the potential to influence the evolutionary ecology of related wild/weedy taxa, but little is known about the persistence or ecological effects of crop genes that enter wild populations via pollen movement (Small, 1984; Snow and Moran-Palma, 1996). Weedy relatives are likely to acquire genes from commercial cultivars when they co-occur, have overlapping flowering periods, share pollen vectors, and do not have strong reproductive barriers that prevent hybridization and introgression. Examples of crops that hybridize spontaneously with wild/weedy populations include sunflower (Arias and Rieseberg, 1994), squash (Kirkpatrick and Wilson, 1988), radish (Klinger et al., 1992), foxtail millet (Till-Boutraud et al., 1992), sorghum (Arriola and Ellstrand, 1996), and canola (Crawley et al., 1993; Jeorgensen and Andersen, 1995). In sunflower, foraging bees carried crop-specific genetic markers as far as 1000 m from small experimental stands of cultivated sunflower (Arias and Rieseberg, 1994). In addition, a 6.4 km isolation zone is recommended to protect commercial sunflower seed nurseries from unwanted wild sunflower pollen (e.g., Smith, 1978). Thus, pollen from cultivated sunflower is certain to spread to adjacent wild populations due to the movements of foraging bees.

The long-term goals of our research on sunflower are to determine (1) the extent to which crop genes persist in wild/weedy populations and (2) whether beneficial crop traits could cause wild/weedy varieties to become more abundant and invasive. Here we report results from the first phase of our research on the fitness of F₁ hybrids between wild and cultivated sunflower, both Helianthus annuus. Wild H. annuus is a native, annual weed that is widespread throughout much of the USA (Heiser, 1954). Natural populations from different regions of the country exhibit a great deal of morphological variation, but distinct subspecies are difficult to categorize (Heiser, 1954; 1976). Recent isozyme and molecular surveys suggest that gene flow among populations has been extensive (Rieseberg and Seiler, 1990, Arias and Rieseberg, 1995), and populations near commercially grown sunflower harbor crop-specific DNA markers due to crop-wild hybridization (Whitton et al., in prep.; Rieseberg et al., in prep.). These findings are not surprising, given that this species is a human-dispersed weed that is fully compatible with cultivated varieties.

Wild sunflower occurs in disturbed habitats such as roadsides and agricultural areas, often reaching heights of 2-3 m in optimal growing conditions. Sunflower was first domesticated by Native Americans (e.g., Smith, 1989) and modern cultivars typically have 6-fold larger seeds than wild plants and an oil content of about 45% as compared to 24% in wild seeds (Seiler, 1984). In contrast to wild populations, commercial varieties lack several fitness-related traits - branching, an extended flowering period, self-incompatibility, and seed dormancy - and often exhibit greater resistance to disease (Seiler, 1992 and references therein). In this study we focus on key characteristics of F₁ wild-crop hybrids that could influence the persistence of crop genes in wild populations: seed dormancy, flowering phenology, and lifetime fecundity of wild vs. wild-crop hybrids.

MATERIALS AND METHODS

Source of wild and wild-crop seeds. In order to study a variety of wild-crop sunflower hybrids, we made crosses using wild plants from three regions (Texas, Kansas, and North Dakota; Table 1) and two cultivated lines. This was desirable because wild sunflower is quite variable across its range, making it difficult to generalize based on data from any one locale. Also, we wanted to determine whether there are major differences between wild-crop hybrids from different commercial lines. The cultivated lines we chose were Triumph #565 (Triumph Seed Co., Ralls, TX) and the standard USDA #894 used in agronomic sunflower research.

Plants from these wild and cultivated seed sources were used in hand-pollinations that were carried out in an insect-free greenhouse during the summer of 1994. To imitate the flow of crop pollen into wild populations, wild plants were used as pollen recipients and wild or crop plants were the pollen donors. None of the wild plants set seed autonomously, being self-incompatible, and unintentional cross-pollination was not a problem as evidenced by the fact that only hand-crossed flower heads (= capitula) produced seeds. On each maternal plant, an equal number of capitula were pollinated with pollen from (1) other wild plants from the region, (2) Triumph #565, or (3) USDA #894. At least 15 plants from each cultivar were used as pollen donors for wild plants from each region, and 23 - 38 wild plants from each region served as donors and/or recipients (Table 1). This crossing program yielded a total of nine cross types (3 regions x 3 pollen sources; see Table 2).

Pollen was applied using a clean Q-tip at the stage when all of the stigmas on the capitulum were exerted and therefore receptive (florets are protandrous and mature centripetally). Several weeks later, mature seeds (= achenes) were collected and stored at room temperature. Seed set from all crosses appeared to be close to 100% and will not be considered further. Approximately equal numbers of seeds from maternal plants in each of the nine cross types were mixed for subsequent experiments, resulting in at least 2,000 seeds per cross type.

Seed germinability. Germinability was assayed by burying seeds in bridal-veil mesh bags at agricultural fields in their region of origin (Austin, Texas; Lawrence, Kansas; or Fargo, North Dakota) for about 2-4 months (exact dates given in Table 1). For each cross type, 25 or 30 bags containing 40 or 50 seeds each were buried at a depth of approximately 20 cm. Bags from the three cross types per region were randomly positioned at 0.5 m intervals along three transects of equal length. The bags were excavated in the spring of 1995 and shipped overnight to Ohio State University. Germinated seeds (i.e., those with hollow seed coats, open at one end) were counted and ungerminated seeds were placed in a trays of moist, sterilized sand at room temperature for approximately two weeks. The total number of seeds that had germinated by this time was used as an index of germinability. All ungerminated seeds were hard-coated appeared to be viable, as described in Teo-Sherrell (1996). Therefore, we assume that ungerminated seeds were probably viable but dormant.

Field experiments with F₁ progeny. Freshly germinated seeds from the burial experiment were used in two field experiments. In late May, we planted seedlings from all nine cross types in a recently disked field at the Kansas Ecological Reserves, University of Kansas. Seedlings from each region were planted together in a grid design, with 5 m between plants and an alternating arrangement of cross types. These plants were not fertilized or irrigated and were allowed to compete with local weeds (mainly ragweed). A similar experiment involving all nine cross types was carried out in Ohio, except that the plants were grown outdoors in 5-liter pots, with slow-release fertilizer (Osmocote) and frequent watering. Ample nutrients and water allowed these plants to become much larger than those in Kansas, as seen in Figure 3. Final sample sizes for the two experiments are shown in Table 2.

Statistical analyses. Because of differences in the times of hand-pollination, seed excavation, and subsequent planting and harvesting, data from each region were considered to be separate experiments and statistical interactions across regions will not be presented. The main results of interest are how the wild plants perform relative to the wild-crop hybrid from the same region. Unless noted otherwise, the data were analyzed using the GLM procedure in SAS followed by Tukey tests (SAS, 1994).

RESULTS

Genes inherited from cultivated sunflower inhibited a moderate amount of seed dormancy found in wild plants from North Dakota and Kansas. Crop-wild hybrids from all regions had germination rates of 90-95%, while wild seeds from Kansas and North Dakota reached germination rates of only 72% and 64%, respectively (Figure 1). In the Texas group, however, percent germination of wild seeds was very high and similar to those of the wild-crop hybrids.

Cross type did not affect plant survival (data not shown), which was lower in Kansas than in Ohio due to browsing by deer. In both field experiments, flowering times of wild and wild-crop hybrids overlapped extensively, but wild plants from Kansas flowered considerably later than the wild-crop hybrids from this region (Figure 2, Table 2). In contrast, there were fewer differences in the flowering times of wild and wild-crop hybrids from North Dakota, and minor differences between cross types from Texas.

In one case, disease resistance may have been inherited by wild-crop hybrids. By early August, 53% of the North Dakota plants grown in pots in Ohio showed symptoms of rust (brown spots and brittle leaves), while none of the wild-crop hybrids showed symptoms (N = 39-40). The frequency of rust symptoms seen in wild and hybrid plants from Kansas and Texas ranged from 0-3%, so presumably these plants were resistant to the rust. Among the wild plants from North Dakota, those that were free of rust symptoms produced 24% more capitula than infected plants (means were 47.6 vs. 37.9, P < 0.02, t-test, N=23, 17).

The relative fitness of wild vs. wild-crop hybrids was measured as the number of capitula per plant. Wild plants clearly produced more capitula than hybrids when the plants were grown in pots with slow-release fertilizer, but at the unmanaged field site in Kansas, these differences were much less pronounced (Figure 3). Plants grown amongst weeds were much smaller, and in two cases differences between wild and hybrid plants were not statistically significant. Capitula of wild plants were smaller than those of wild-crop hybrids, but these differences were not great enough to have a major impact on total seed production per plant (Snow et al., in prep.). In other words, the number of capitula per plant was strongly correlated with total seed production.

DISCUSSION

This study demonstrates regional differences in the relative performance of wild plants vs. wild-crop hybrids, and general consistency in comparisons between progeny from the two cultivated lines. Overall, we found that wild-crop hybrids are very likely to germinate and interbreed with wild plants. Possible disadvantages of the hybrids include a lack of seed dormancy, which was seen in 28% and 36% of the wild seeds from North Dakota and Kansas, respectively (seed dormancy was not detected in wild plants from Texas). This wild trait may promote the long-term persistence of populations in Kansas and North Dakota. Another disadvantage of hybrids was seen in the flowering times of hybrids from Kansas, which began flowering 4-8 weeks earlier than wild genotypes. This limits opportunities for interbreeding between F₁ hybrids and wild neighbors, although we expect less asynchrony to occur under natural conditions, where the timing of seedling establishment is highly variable.

The magnitude of fitness differences between wild vs. wild-crop hybrids was strongly affected by local growing conditions. Wild plants from all regions produced 2-3 times more seeds than hybrids when grown in the absence of competition (in pots in Ohio). In contrast, differences in seed production were smaller when plants were grown at a weedy field site in Kansas, especially for plants from North Dakota and Texas. Taken together, our results suggest that the F₁ generation may act as a weak and temporary barrier to the spread of crop genes after episodes of hybridization. During subsequent generations of backcrossing with wild plants, natural selection should gradually purge highly deleterious crop traits from wild populations, while selectively neutral alleles would persist and favorable alleles would spread.

An interesting advantage of hybrid plants was seen in the apparent inheritance of disease resistance that occurred in plants from North Dakota. Wild populations of H. annuus are known to vary in their resistance to strains of the rust Puccinia helianthi Schw. (Seiler, 1992), and resistance genes from wild plants have been bred into cultivated sunflower (e.g., Quresh et al., 1990). In our experiment, it appears that resistance genes may have been lacking in about half of the wild plants from North Dakota, although further studies are needed to confirm this. If this was the case, hybridization with the crop led to protection from disease. Wild plants that were free of rust symptoms produced 24% more capitula than infected plants, indicating that the rust had a significant effect on lifetime fecundity. This is just one example of the types of advantageous traits that will be transferred from transgenic plants to wild relatives, thereby increasing the seed production of the wild plants. If wild plants inherit several such fitness-related genes from cultivated relatives, whether transgenic or not, it is possible that these beneficial traits could contribute to weediness of the wild genotypes. Indeed, some of the striking regional variation in the size and fecundity of wild sunflowers may be due to varying degrees of wild-crop hybridization that has occurred over the past few decades.

In conclusion, the present study and our research involving genetic markers (Whitton et al., in prep.; Rieseberg et al, in prep.) provide strong evidence that crop genes continually move into wild populations and persist in these populations for many years. The next question of interest, then, is how do transgenic traits such as strong insect resistance affect seed production and population growth rates under natural conditions? We plan to investigate the ecological effects of beneficial transgenes in the next phase of our research.

ACKNOWLEDGEMENTS

We thank the Kansas Ecological Reserves for field assistance and the use of the Nelson Experimental Area. Helen Alexander, Norma Fowler, Annette Wszelaki, Sara Taliaferro, Laura McDonald, and Lawrence Spencer provided valuable assistance during various stages of the research. This research was funded by USDA-BRARG grant 92373007590 to LHR and AAS.

REFERENCES

Arias, D. M., and L. H. Rieseberg. 1994. Gene flow between cultivated and wild sunflower. Theor. Appl. Gen. 89:655-660.

Arias, D. M., and L. H. Rieseberg. 1995. Genetic relationships among domesticated and wild sunflowers (Helianthus annuus, Asteraceae). Econ. Bot. 49:239-248.

Arriola, P. E., and N. C. Ellstrand. 1996. Crop-to-weed gene flow in the genus Sorghum (Poaceae): spontaneous interspecific hybridization between johnsongrass, Sorghum halapense, and crop sorghum, S. bicolor. Am. J. Bot. 83:1153-1160.

Crawley, M. J., R. S. Hails, M. Rees, D. Kohn, and J. Buxton. 1993. Ecology of transgenic oilseed rape in natural habitats. Nature 363:620-623.

Heiser, C. B. 1954. Variation and subspeciation in the common sunflower, Helianthus annuus. Am. Midl. Nat. 51:287-305.

Heiser, C. B. 1976. Sunflowers. pp. 36-38 In: N. W. Simmonds (Ed.), Evolution of crop plants. Longman, London.

Jeorgensen, R., and B. Andersen. 1995. Spontaneous hybridization between oilseed rape (Brassica napus) and weedy Brassica campestris: a risk of growing genetically modified oilseed rape. Am. J. Bot. 81:1169-1175.

Kirkpatrick, K. J., and H. D. Wilson. 1988. Interspecific gene flow in Cucurbita: C. texana vs. C. pepo. Am. J. Bot. 75:519-527.

Klinger, T., and N. C. Ellstrand. 1994. Engineered genes in wild populations: fitness of weed-crop hybrids of Raphanus sativus. Ecol. Appl. 4:117-120.

Quresh, Z., C. C. Jan, and T. G. Gulya. 1990. Transferring rust resistance genes from wild Helianthus spp. into cultivated lines. In: Proc. Sunflower Research Workshop, National Sunflower Assoc., Bismark, ND, pp. 90-93.

Rieseberg, L. H., and G. Seiler. 1990. Molecular evidence and the origin and development of the domesticated sunflower (H. annuus L.). Econ. Bot. 44:79-91.

SAS, 1994. Statistical Analysis Systems, Cary, NC.

Seiler, G. J. 1984. Variation in agronomic and morphological characteristics of several populations of wild annual sunflower (Helianthus annuus L.). Helia 7:29-33.

Seiler, G. J. 1992. Utilization of wild sunflower species for the improvement of cultivated sunflower. Field Crops Res. 30:195-230.

Small, E. 1984. Hybridization in the domesticated-weed-wild complex. Pages 195-210 in W. F. Grant, editor. Plant biosystematics. Academic Press, Toronto, Ontario, Canada.

Smith, B. D. 1989. Origins of agriculture in eastern North America. Science 246:1566-1571.

Smith, D. L. 1978. Planting seed production. pp. 371-372 in: Sunflower science and technology. Agronomy 19, American Society of Agronomy, Crop Science Society of America, and Soil Science Society of America. Madison, WI.

Snow, A. A., and P. Moran-Palma. 1996. Commercial cultivation of transgenic crops: potential ecological risks. BioScience, in press.

Teo-Sherrell, C. P. A. 1996. The fates of weed seeds. Ph. D. Dissertation, Department of Agronomy, University of Nebraska, Lincoln, Nebraska.

Till-Bouttraud, I., X. Reboud, P. Brabant, M. Lefranc, B. Rherissi, F. Vedel, and H. Darmency. 1992. Outcrossing and hybridization in wild and cultivated foxtail millets: consequences for the release of transgenic crops. Theoretical and Applied Genetics 83:940-946.

Table 1. Sources of wild seeds used in hand-pollinations to produce wild-crop hybrids. Seeds were collected from 25-30 widely spaced, open-pollinated plants in each population. Seed burial and excavation dates for germinability experiment are also shown.

REGION	COLLECTION SITE	NO. OF PLANTS USED IN CROSSES
Texas	L. Rieseberg, Collection #1022 (Karnes County), seeds buried 2/11/95-4/11/95	23
Kansas	L. Rieseberg, Collection #K-1 (Near Lawrence-west of Renner Rd on I-435), seeds buried 12/20/94-3/28/95	36
North Dakota	USDA Germplasm Collection #2102 (Wells County), seeds buried 12/1/94-4/21/95	38

Table 2. Characteristics of wild vs. wild-crop hybrids grown at two sites (field site in Kansas or outdoor pots in Ohio). Within regions, means followed by different superscripts are significantly different at P<0.05 (Tukey tests).

A. DAYS TO FIRST FLOWER

FIELD SITE

OUTDOOR POTS

Region	Cross type	Mean	SE, N	Mean	SE, N
Texas	x Wild x Triumph #565 x USDA #894	85 a 64 b 55 c	3, 36 2, 38 1, 33	69 a 70 a 63 b	1, 38 1, 37 1, 38
Kansas	x Wild x Triumph #565 x USDA #894	112 a 86 b 77 c	2, 34 4, 23 3, 31	135 a 93 b 79 c	2, 39 3, 40 2, 36
N. Dakota	x Wild x Triumph #565 x USDA #894	49 a 56 a 48 a	1, 42 2, 34 2, 38	64 b 70 a 63 b	1, 39 1, 40 1, 40

B. HEIGHT (in cm)

FIELD SITE

OUTDOOR POTS

Region	Cross type	Mean	SE, N	Mean	SE, N
Texas	x Wild x Triumph #565 x USDA #894	86 a 56 b 52 b	3, 36 3, 38 3, 33	145 a 129 b 110 c	3, 38 4, 37 3, 38
Kansas	x Wild x Triumph #565 x USDA #894	122 a 74 b 77 b	4, 34 2, 23 4, 31	272 a 174 b 142 c	6, 39 5, 40 6, 36
N. Dakota	x Wild x Triumph #565 x USDA #894	92 a 103 a 101 a	5, 42 4, 34 3, 38	131 b 147 a 123 b	3, 40 4, 40 3, 40

Figure 1. Percent of buried seeds that germinated the following spring. Cross types were wild x wild (W), wild x Triumph #565 (T), and USDA #894 (U). Within each region (Texas, Kansas, or North Dakota), means with different superscripts are significantly different at P < 0.05 (Tukey tests; percents were arcsin-transformed). N = 25-30; error bars represent 1 SE using nontransformed data.

Figure 2. Flowering phenology of each cross-type when grown outdoors in pots in Ohio. The total number of plants in each group was 37-40, as indicated in Table 2.

Figure 3. Mean number of flower heads per plant at A) the field site (Kansas) and B) in outdoor pots (Ohio). Labels and statistical tests as in Figure 1; see Table 2 for sample sizes.