The nature of FLCP/domesticate hybrids and their progeny

 Initial connections between FLCP and domesticated C. pepo were established when it became evident the F1 hybrids between the two types were readily produced and fully fertile (Whitaker and Bemis, 1964). Eventual recognition that FLCP and domesticated forms constitute a conspecific unit resulted from demonstration of normal mendelian genetic behavior of the F1 hybrid (Kirkpartick et al., 1985) and experimental proof that crop/weed hybridization can be expected to result if crop and weed populations are growing in proximity (a radius of at least 1300 meters) and within the distributional range of specialized 'squash' bees (Kirkpartick and Wilson, 1988), which occur throughout the FLCP range of distribution. This perspective has been supported by the discovery of first-generation gene flow from cultivars to FLCP, as indicated by concordant heterozygosity at multiple isozyme loci among progeny taken from FLCP populations (Decker and Wilson, 1987; Decker-Walters et al., 1993). Allozyme frequency distributions and distinctive patterns of variation in fruit structure, color, and bitterness within populations of FLCP clearly indicate that past hybridization events have resulted in permanent gene transfer, or introgression, between domesticates and FLCP in the eastern U.S. (Decker and Wilson, 1987; Decker-Walters et al., 1993;Smith, 1992;Wilson, 1989;1990).

 Given current theory regarding genetic relationships among crop/weed systems as they relate to crop plant evolution (Harlan, 1992), this pattern of sporadic crop/weed genetic interaction is to be expected. It provides a unique foundation for "mixing the gene pool" (Harlan, 1965;1969) and injecting high levels of genetic variation that are needed to produce the types of extreme structural adaptations that characterize domesticated plants. However, as indicated by "wild carrots" (Queen Ann's Lace-Daucus carota) in the northeastern United States, "wild cucumbers" (Bur-Gherkin, Dudaim Melon-Cucumis spp.) in the southeast, and "wild oats" (Avena fatua) in the plains and West, this pattern of gene exchange can function to increase adaptive fitness of the non-domesticated participant.

 Relative fitness of C. pepo crop/weed hybrids and F2 progeny has not been examined. However, Kirkpatrick (1983), in an effort to define general linkage relationships between structural and molecular characters, conducted a detailed quantitative study of crop/weed F1 hybrids, parental types, and F2 progeny generated by self-pollinating the F1. The hybrids were developed using ssp. ovifera var. texana from Fayette County, Texas as the staminate parent and a zucchini cultivar (ssp. pepo) as the pistillate parent. The following discussion is based on data extracted from that study.

The F1 hybrids generated from this study were clearly heterotic, i.e., more robust and vigorous than either parent. This is reflected by data depicted in Fig. 7. The texana parent was a typical vining plant with relatively small fruit and leaves, whereas the zucchinis were 'bush' types with suppressed lateral shoots, larger fruits and larger leaves. The F1 plants combined features of both parents, retaining the longer, more numerous lateral shoots of the texana parent and the larger fruits and leaves of the zucchini parent. Expression of the 'wild type' features was weighted, mainly because these are generally genetic dominants in the Mendelian sense (Robinson et al. 1976). Thus, a combination of heterosis resulting from a relatively 'wide' intraspecific cross and simple genetic dominance produces a crop/weed F1 that is essentially greater than the sum of its parts. One would assume that F1 plants would, at the very least, compete well with typical FLCP in a 'weedy' agricultural habitat and, more likely, show greater fitness in terms of progeny production. This, however, has not been tested. One would also expect a reduced heterotic response from a crop/weed F1 plant if the crop parent was more closely related to FLCP, i.e., an element of ssp. ovifera var. ovifera such as crookneck or acorn squash. This has not been tested.

 The F1 hybrid is an important, but ephemeral, element of the crop/weed introgressive process. More important is the nature of the F2 generation. This would derive from either self pollination of the F1 or back crossing from parental types. Given the breeding system of FLCP and a probable 'home garden' pollen source, the generation immediately following a crop/weed hybridization event would most likely result from self-pollination of the F1. As indicated by the 'F2' panel in Fig. 7 self pollination of the texana/zucchini F1 produces an F2 progeny that shows a broad pattern of variation that reflects the results of "mixing the gene pool."

A more detailed perspective on the nature of variation among F2 plants is presented in Fig. 8. Analysis of this F2 family was conducted to test the notion that differentiation between crop and weed forms of C. pepo is based on 'supergenes' or tightly linked gene clusters. This involved measurement of 14 characters, representing both vegetative and floral morphology, from 272 field-grown plants. The pattern of variation among measured plants was examined by principal component analysis, a procedure that reduces a complex pattern of variation to 'factors' which allow visualization via two-dimensional plots such as Fig. 8. The diffuse pattern of data points present in Fig. 8, each representing a single plant, indicates that genetic segregation is mostly independent, i.e., loci that carry differentiated crop/weed alleles are probably dispersedthroughout the genome and not localized in linkage groups or centered on specific chromosomes.

 The data presented in Fig. 8 roughly define that nature of progeny that might be produced by crop/weed hybridsgenerated under 'natural' field conditions. It is clear that F1 hybrids (Fig. 8 - 'F') and their progeny (Fig. 8. -'circle') tend to express the genetic constitution of the FLCP parent (Fig. 8 - 'T'). Adaptations characteristic ofthe domesticate are typically the product of homozygous, recessive genetic expression (Robinson et al., 1976). Thus, F1 hybrids ('F') cluster with the FLCP parents ('T') in the ordination (Factor 1) provided by the suite of variables that separate domesticate ('P') from weed ('T'). Variables influencing Factor 1 are mostly vegetative (shoot and leaf characters). Ordination of samples along Factor 2, which is weighted by floral characters, places the F1 plants ('F') in a unique position, i.e., the variables to not separate parental crop ('P') and weed ('T') plants along the axis of Factor 2. This unique phenetic placement of the F1 plants is concordant with data presented in Fig. 7 as an indication that first generation hybrids constitute, in terms of phenotypic expression, a unique entity that transcends either parent. Weighted variables for Factor 2 include measurements of reproductive structures. Higher values along the axis of Factor 2 signify larger staminal columns, both anther and filament. Thus, in terms of general reproductive potential, F1 plants and a subset of their progeny can be expected to offer more pollen to pollination vectors than either parent. Consequently, over the long term history of 'hybrid swarm' populations in nature, plants expressing a genetic 'mix' of crop and weed genomes will probably show a slight reproductive edge which would function maximize introgressive gene flow and recombination.

 The pattern of variation among F2 plants depicted in Fig. 8 does not reflect hybrid intermediacy. Structural variation ordinated by both Factors is clearly skewed toward the FLCP condition in both the F1 plants ('F') and their progeny ('circle'). This pattern, probably also a function of 'wild type' genetic dominance, demonstrates the tendency of both first generation and segregating progeny to move toward the FLCP phenotype. This suggests that the frequency of plants that are ill-adapted for existence in the wild, i.e., those that resemble the domesticated parent, will be minimal in a hybrid swarm population. Conversely, the percentage of plants in a given hybrid swarm that either resemble the FLCP parent, or carry unique features associated with the FLCP condition that might enhance fitness in the wild, is maximized. The pattern of variation among both F1 and F2 plants in Fig. 8 therefore suggests that the products of 'natural' hybridization events between domesticated and free-living C. pepo are genetically 'pre-adapted' for life in the wild or weedy habitat.

 It should be noted that the data generated by Kirkpatrick (1983) were not produced to approach the problem of hybrid fitness, and the analyses were not conducted with this aspect of the crop/weed interface in mind. However, the data do reflect phenotypic variation as indicated by measurements of both vegetative and floral characters. If plant size is an element of fitness, then the patterns of variation depicted in Figs. 7 and 8 are relevant to the question of hybrid fitness.

Possible impacts from crop-to-weed gene flow

 Given the demonstrated potential for crop/weed hybridization within the FLCP range of distribution, as well as the documentation of both first generation and introgressant plants in extant FLCP populations within this range, it is reasonable to assume that genetic contact between domesticated and free-living C. pepo has occurred within the FLCP range, and that this has resulted in the production of self-perpetuating, introgressed population systems. Given the agricultural history of eastern North America (Harlan, 1992;MacNeish, 1992;Watson, 1993), and the possible role of C. pepo (Smith, 1992;Cowan and Smith, 1993), it is quite possible that some of these populations are extant manifestations of ancient crop/weed interactions in the area (Wilson, 1990).

This set of circumstances suggests that the introduction of new C. pepo strains into the FLCP range of distribution (Fig. 5) will genetically impact FLCP populations in that those genetic features that mark the strain as 'new' will be 'captured' by FLCP populations via crop/weed hybridization events. All available evidence indicates that this has happened in the past, it will happen during the growing season of 1993, and there is no reason to believe that it will not happen in the future.

If crop/weed genetic interaction has occurred within the C. pepo complex of domesticated and free-living forms throughout the 3,000 year history of human agricultural activity in the eastern U.S., then why be concerned about the possible involvement of transgenic strains? All available evidence, both archaeological and botanical, indicates that new, domesticated elements of the C. pepo complex have been sequentially introduced into the agricultural system of eastern North America over the past 3,000 to 7,000 years (Smith, 1992). However, these introductions have not carried genetic material that is that has been obtained from phyletic lineages that are not part of the C. pepo complex. Thus, the source of transgenes, and unknown interactions between these unique genetic elements and the C. pepo genome, represent, within the biological and historical context of C. pepo, an unknown and untested factor. The process of injecting a foreign genetic element, a functional gene that has no precedent within the phylogenetic history of a complex crop/weed system such as C. pepo, constitutes a biological risk. The dynamics of this risk, in terms of level and nature of impacts, are difficult - if not impossible - to predict. Specific negative impacts, if any, cannot be determined with any accuracy. However, the historical record of both intentional and accidental human genetic manipulations and subsequent impacts on natural populations (most recently the 'Africanized' honey bee) suggests caution.

A cautious approach is reinforced in this instance by the action of the unique genes carried by the a transgenic C. pepo strain. Calgene's flavr-savr tomato, for instance, poses minimal risk in that action of the gene has no obvious selective relevance to free-living elements of the tomato primary gene pool of Central and South America. However, any genetically transmissible trait that provides enhanced fitness in the wild is cause for concern. This is based on the simple fact that any selective advantage that might be 'captured' from a transgenic domesticated line by FLCP could alter existing ecological and genetic balances in such a way that individual recipient plants, and FLCP progeny that carry the advantage, would expand their numbers and their range of distribution. A transgene-mediated range extension would have negative impacts in two critical areas of the human agricultural enterprise:

• crop losses due to FLCP infestation could increase
• potentially valuable genetic diversity associated with an economically important crop plant could be diminished or extirpated through displacement of extant FLCP populations by plants representing the relatively narrow genetic base of the transgene-equipped, 'expanding' lineage


 Engineered resistance to either predators (Bt-toxin) or disease (multi-viral resistance) that impact both domesticated and free-living population systems represent the type of genetic trait that could produce these changes in population structure.

 It is important to keep in mind, for the purposes of this discussion, that interactions between plant viruses and host plant taxa are long-term, phylogenetic phenomena. This represents a basic host/parasite, co-evolving relationship that existed in the native flora long before the origin of domesticated plant species. It is therefore reasonable to assume that impacts on domesticated populations that are brought about by viral parasitism are fully comparable to similar impacts on populations of free-living relatives of the crop that exist in the native flora. This fundamental reality provides a rational justification for efforts, which are often successful, to find virus resistance among wild relatives of a given crop.

 If extant FLCP populations carried resistance to important cucurbit viruses, such as WMV-2 and ZYMV, then the presence of engineered resistance in transgenic domesticated lines would pose no obvious problem in that increased fitness would not result from crop/weed introgression. This, however, is not the case. Genetic resistance to ZYMV and WMV-2 is not present in the C. pepo primary gene pool. This is clearly demonstrated by the fact that breeders have not been able to produce resistant C. pepo cultivars using traditional methods.

FLCP, a sample of var. texana from DeWitt County, Texas, was included in a suite of 20 free-living and domesticated taxa that were screened for viral resistance by Provvidenti et al. (1978). Plants were subjected to virus attack by inoculation of six virus strains in the greenhouse, and exposure to native viruses in the field. The FLCP sample showed resistance to one, Tobacco Ringspot Virus, of the six viral strains, as did most Cucurbita taxa tested. While resistance to several viral strains was present in several wild Cucurbita species tested, FLCP showed a pattern similar to that of the domesticated C. pepo used in the screen; full and often extreme susceptibility to viral attack.

If multi-viral resistance entered FLCP populations by hybridization/introgression events involving a transgenic line, it is reasonable to assume that this would provide a selective advantage in that extant negative selective pressures produced by viral parasitism would be released. Fitness of those FLCP plants carrying transgenic resistance would increase and, if viral disease constituted a significant negative selective factor in the natural habitat, this could lead to increases in population size and range of distribution for resistant FLCP lineages. Clearly, FLCP populations have demonstrated an ability to expand and become weed problems within the agroecosystem. A relatively small shift in the selective forces that act on these populations could amplify this tendency.

 While data presented here point toward the clear presence of risk in two areas - increased weedyness and loss of crop plant biodiversity - the nature of risk, in terms of magnitude and dimensions, is difficult to access. Efforts to simplify the problem by experimental determination of levels of 'fitness' or 'selective pressure' and units of selection (gene vs. genome) are complicated by the multifaceted nature of the problem. The potential participants, FLCP and C. pepo domesticates, are not uniform entities in terms of genetic structure. Different results might obtain from different points of genetic contact (ssp. pepo vs. ssp. ovifera, var. texana vs. var. ozarkana), either natural or experimental. Possible selective arenas include both 'natural' (undisturbed) and human (cultivated fields), as well as third 'weedy' realm that extends, more or less as a continuum of human vs. non-human selective pressures, between these extremes. While Kirkpatrick's study (1983) provides a rough picture of what might be expected to result from the combination of two differentiated C. pepo genomes (free-living vs. domesticated), the selective dynamics of heterosis and hybrid recombination are unknown. When considered against this diverse comparative background, assessment of 'fitness' or selective advantage that might be associated with a given transgene would be extremely difficult, even if the foreign genetic material was expressed cleanly (no pleiotrophy or epistasis) and without any genetic or recombinational 'load' (non-coding elements).

 Given this set of circumstances, and that fact that these circumstances will exist when transgenic cultivars of Helianthus annuus are released into the U.S. and transgenic Zea mays is released into Mexico, one must consider the impact of this transgenic squash as a cultural precedent. If future agricultural development is to be insured, those responsible for environmental protection as it relates the development of biotechnology should - at the very least - move to insure that extant germplasm resources representing crop plant biodiversity in the United State (at a minimum) are located and salvaged as viable accessions as soon as possible. This would serve as a hedge against possible 'worse case' genetic interactions and also provide a foundation for eventual assessment of real risk over the long term.

How would you design an experiment if you wanted to determine whether virus resistant FLCP-domesticate hybrids were more 'weedy' than the parental species? Would gene copy number, homozygosity, or heterozygosity of the virus resistance gene influence the design or interpretation of results?

 NOTE: The following section, pointing out Asgrow field tests of the transgenic strain (1994) at the site that had been used to demonstrate long distance crop/weed gene flow in 1983, was deleted from the report submitted to USDA-APHIS - at their request following USDA review of the first draft - and the discussion (continued below) was revised to accomodate this deletion:

 There are two ways to approach assessment of weedyness among FLCP plants that 'capture' viral resistance from a transgenic squash, irrational and rational. An irrational test of the phenomenon would be to simply plant several rows of transgenic squash in an area where crop/weed hybridization might be expected to occur. If local FLCP become viral resistant and weedy, then a problem exists. This approach is irrational in that it mediates uncontrolled genetic release of the trait under test and thereby essentially negates the need to conduct a test. Such an experiment is, however, underway in the Brazos bottomlands of central Texas. The Upjohn Company, with USDA approval, is conducting yield tests involving "about 150" transgenic plants, buffered from other Cucurbita by 500 meters (H. Quemada, pers. comm.) in fields that carried experimental populations that were used by Kirkpatrick and Wilson (1988) to document gene flow between free-living and domesticated populations of C. pepo. Gene flow, mediated by native squash bees, occured between FLCP and domesticates that were separated by 1300 meters. All elements necessary for the required hybridization event, including species of both "squash bee" genera and the 'Texas Gourd", are known to be present in the area. There are better places to conduct such a yield test. However, if the transgenic squash were isolated from other squash by a distance greater that ca. 1600 meters and the 1600 meter radius was surveyed to insure that native FLCP was not present, then experimental FLCP plants could be placed at various distances from the yield test plot of the transgenic crookneck cultivar and its parent type. This would allow a second experiment, i.e., a test for transgenic/FLCP gene flow under 'natural' conditions, and, if bee-mediated gene flow did occur, weed/crop hybrids for the experiment described below.

 Rational experimental assessment of the possibility for enhanced 'weedyness' in virus-resistant FLCP should be complicated by two factors: .......

 USDA-censored document continues:
 
 

Experimental assessment of the possibility for enhanced 'weedyness' in virus-resistant FLCP is complicated by two factors: 1) containment, and 2) the large, unwieldy nature of FLCP and FLCP/domesticate hybrids. These factors work to negate the use of field studies based on open pollinated plants that are grown to reproductive maturity. Both problems would be eliminated by an experimental design that is based on greenhouse-grown plants and an assumption that variation in vegetative growth, or production of biomass, roughly corresponds to 'fitness' in the wild. This assumption is probably valid in that:

• All available data indicate no loss of fertility or seed-set in FLCP/domesticate hybrids and progeny
• Solitary flowers are produced at nodes. As the number of nodes increases with overall plant size, the number of flowers will increase
• The number of potential fruit increases with increases in the number of flowers.


 Thus, an assumption of a direct relationship between seed production (reproductive fitness) and plant size (vegetative fitness) is reasonable.

• Domesticate A (ssp. pepo)
• Domesticate B (ssp. ovifera-non-transgenic cultivar - 'Domesticate C' without resistance)
• Domesticate C (transgenic, viral resistant)
•  FLCP A (texana type)
•  FLCP B (ozarkana type)
•  First Generation Hybrids(10) involving crossing between Domesticates A-C and FLCP A and B as the pistillate parent (the most likely direction of gene flow under field conditions).
•  F2 progenies resulting from self-pollination of the 10 F1s

• 1. control
• 2. virus inoculation (first priority should be the most virulent virus, the zucchini yellow mosaic virus [ZYMV])
• 3. competition with soybean

The experiment would require 500 10" pots. It would be replicated at least 4 times during the growing season. Consequently, to minimize the effect of varying photoperiod, light intensity, etc. encountered by different replicate plantings, data would be expressed as percentage biomass produced relative to the control. The results would provide a reasonable picture of relative vigor in vegetative growth under the varying experimental conditions. This, as indicated above, should directly correspond to relative reproductive potential among the different test groups.

prior section	Table of Contents		Literature Cited