The odds ratio analysis showed that the likelihood of occurrence of feral sorghum along a gravel or dirt road is 1.5 or 2.4 times greater than that of a paved road, respectively . Dirt and gravel roads are prevalent in rural areas surrounding farmlands and high likelihood for the presence of feral sorghum populations along these road types suggests that movement of farm equipment and production activities greatly contribute to sorghum seed dispersal into roadside habitats. Further, wash boarding, corrugation and any potholes on the surface of unpaved roads can increase vehicle bouncing and thus increase the chances of seed spill from seed transport trucks and farm equipment. Contrary to our findings, the higher frequency of feral oilseed rape along the paved compared to the gravel and dirt roads in France was attributed to the higher traffic intensity with commodity transport on paved roads. Although the Chi-square test suggested significant differences between the road types , the odds ratio analysis failed to detect such differences . This may suggest that all road types are equally likely to accommodate feral sorghum, following a seed immigration event. Although previous studies have found a strong relationship between the presence of feral crops and road type, this was not the case for feral sorghum in South Texas. One notable exception in the present survey was that feral sorghum populations were common along U.S. Highway 77 between Kingsville and Raymondville, TX where sorghum and crop production in generalis very sparse. The high intensity of grain transport via truck movement is likely contributing to sorghum seed dispersal along this highway. Feral sorghum was more likely to be present at the road shoulders as shown by the odds ratio values , but they were also present in the field shoulders and at field edges . However, cannabis drying system as the road and field shoulders are only separated by few meters in many cases, feral sorghum found along the road verges could also be sourced by sorghum production activities in the adjacent fields.

This postulation is supported by a lack of significant difference between the odds of feral sorghum presence at field shoulders relative to that of the road shoulders, as shown by the odds ratio estimates . However, a logical question could be why feral sorghum is less abundant at field shoulders or field edges if the adjacent fields could contribute to propagule immigration. Field edges are typically disturbed , disrupting the establishment and persistence of feral sorghum in these sites. Nevertheless, the microhabitats at the roadside could provide more moisture for the establishment of plants compared to field edges. The nearby land use had large effects on the presence of feral sorghum . Results showed that the odds for the occurrence of feral sorghum in sites adjacent to sorghum cultivation was larger than that of all other land uses; the likelihood of finding feral sorghum in a location closer to a sorghum field was almost twice as high as a location contiguous to corn, hay, pasture, shrubland, urban or fallow lands . Sorghum is one of the major crops grown in South Texas and results suggest that sorghum cultivation and seed transport activities in the region contribute to seed immigration and establishment of feral sorghum on roadside habitats. Further, the sorghum seed dispersed following harvest might germinate instantly due to the lack of seed dormancy and the warm environmental conditions in South Texas may allow feral sorghum to produce viable seed prior to killing frost and establish self-perpetuating populations. There is also a possibility for spring establishment of feral sorghum from the seeds entered into the soil post-harvest should they be able to survive during the fall and winter. Data on seed survival rate of sorghum coupled with early spring monitoring are needed to address this question.The population size of the feral sorghum at each site was scored based on visual estimations.

The results from the analysis of variance showed no significant relationship between all the measured factors and the population size of the feral sorghum at sampled sites expect for the vegetation cover . The largest feral sorghum population sizes were associated with the highest vegetation cover , a finding that is unexpected given that vegetation with higher canopy cover should be more resistant to invasion than those with low canopy cover. One possible explanation is that the roadsides are regularly treated with herbicides by the Department of Transportation and in some cases by county weed control specialists for controlling tall vegetation such as johnsongrass. It is likely that herbicides might have been recently applied at sites with low vegetation cover, thus reducing the chance of observing feral sorghum individuals.The co-occurrence of feral sorghum and johnsongrass was rare and both species were found together only in 48 of the 2,077 survey sites visited . Fig 4 shows co-occurrence of feral sorghum and johnsongrass in a roadside site near Corpus Christi, TX. Results from the logistic model showed a negative relationship between the occurrence of johnsongrass and feral sorghum . The likelihood of detecting feral sorghum at locations without johnsongrass was 4.3 times greater than the locations where johnsongrass was present. Three possible scenarios might explain this finding: the presence of johnsongrass in the site may have a negative influence on the germination and establishment of feral sorghum ; however, the data collected in this study was not sufficient to establish any causal relationship or there is no anecdotal evidence to support such a scenario in production fields, the dispersal of sorghum seed on roadsides as a function of intensive sorghum cultivation and seed transport occurs primarily in the much Southern parts of Texas from Victoria towards Brownsville, an environmental gradient increasingly less suited for johnsongrass as well as in the habitat suitability map for johnsongrass; thus, the cooccurrence of both species was perhaps naturally limited, and/or roadside herbicide applications that target johnsongrass may eliminate any feral sorghum plants present within these sites, while johnsongrass could regrow from rhizomes. It is very likely that the second and third scenarios have substantial influence on the co-occurrence of these two species. A followup observation conducted in summer 2017 has revealed supporting evidence for the third scenario in that several of the feral sorghum-johnsongrass complex sites we identified in the 2014 survey were severely impacted by roadside herbicide applications that typically target johnsongrass. In these sites, several johnsongrass plants survived, but almost all feral sorghum plants were eliminated. Given that feral sorghum and johnsongrass can hybridize, the co-occurrence of these two species may facilitate the persistence of feral sorghum through gene flow and introgression of adaptive traits from johnsongrass. In addition, cultivation of diverse sorghum lines, including sudangrass and sorghum-sudangrass hybrids, in the vicinity may enrich the diversity within the feral sorghum populations and thereby increase the adaptive ability of feral sorghum. Such an outcome has been reported for feral populations of oilseed rape and alfalfa. Since this is the first record of the presence of feral sorghum innature, no information is available on the diversity, population genetic structure and longterm persistence of feral sorghum populations.Although most of the feral sorghum populations were observed along the roadsides, they may have the potential for spread to their contagious natural and unmanaged areas. To investigate the potential for broader distribution of feral sorghum in South Texas, we calibrated a model using nearby land use type and regional habitat suitability for johnsongrass as reliable predictors.

These two variables were chosen because they were statistically significant and the georeferenced data for the entire region was available. The combination of these two variables effectively predicted the distribution of feral sorghum, with an increasing trend in abundance from the Upper Gulf Coast towards the Rio Grande Valley, which corresponded to an increasing intensity of sorghum cultivation and seed transport activities in the landscape. Further, in the more Southern areas of Texas, sorghum seeds germinating after the harvest season will have a high chance to produce mature seed prior to killing frost, if any. The projected map for feral sorghum distribution is shown in Fig 5, which corroborates with the overall trend observed in the survey. Conversely, growing tray the distribution of johnsongrass showed an opposite trend, with more abundance in the Upper Gulf Coast region than in the Rio Grande Valley, attributable to its habitat suitability as evident in the habitat suitability map .The current survey showed that roadsides and field margins are the initial niches for feral sorghum to establish outside of cultivated fields. We found that the occurrence of feral sorghum in South Texas is highly associated with sorghum cultivation in the nearby area, providing propagules for the establishment of feral populations in field edges and roadsides during planting and grain transport operations. We did not find any relationship between the frequency of feral sorghum and road characteristics . Although johnsongrass can be found commonly along the roadsides in South Texas, the co-occurrence of feral sorghum and johnsongrass was infrequent. Yet, there are significant opportunities for outcrossing to occur between the two species outside of cultivated fields. More research is necessary to understand the frequency of outcrossing between the two species and fitness of the progenies. Experiments are on the way to characterize, using phenotypic and molecular markers, the progeny of seed harvested from feral sorghum plants during this survey in sites where both species co-existed. Further, field surveys and monitoring are being carried out to confirm and characterize potential hybrid progenies in nature in these feral sorghum-johnsongrass complex sites.When transgenic plants were initially developed, most plant evolutionary biologists and geneticists considered spontaneous hybridization between species to be rare and of little importance in terms of evolution. This view extended to both crops and their wild or weedy relatives, but has now radically changed. More than twenty years of gene-flow research has shown that interspecific hybridization is very common in some groups of vascular plants and may be of considerable evolutionary significance. Hybridization may occasionally result in the extinction of a population, may trigger the evolution of plant invasiveness, or initiate speciation. A substantial body of evidence has now accumulated, demonstrating the high potential for interspecific hybridization between agricultural crops and their wild or weedy relatives. Transgenic crops are no exception, and empirical studies have provided evidence of transgene dispersal from GM crops to their weedy relatives. Many factors have been shown to influence the rate of hybrid formation between crops and their wild or weedy relatives. Population effects such as the local densities of the parental types and their relative frequencies, have been demonstrated in several cases. Mating system differences at the individual level due to, for example, selfing rates and apomixis, have also been found to affect hybridization rates. Moreover, several studies have shown that overlap in the flowering periods of crop and weed plants affect opportunities for hybridization. The aim of this study is to gain insight into the impact of hybridization with transgenic crops on the evolution of the weedy relatives by verifying that hybridization opportunities for weedy plants depend on their phenotypic traits , measuring the relative fitness of hybridizing weeds, and searching for associations between the transgenic trait and the phenotypic traits increasing hybridization opportunities in the offspring of weedy plants. We studied hybridization opportunities, phenotypic traits and offspring phenotype of weedy individuals in experimental plant populations cultivated under glasshouse conditions. Experimental populations were composed of weeds and transgenic plants in a 1:1 ratio. Transgenic plants were crop plants of the Brassica genus , F1 hybrids between B. rapa and B. napus, or first-generation backcrosses. Crop plants were all homozygous for the Btcry1Ac transgene from Bacillus thuringiensis, F1 hybrids were all hemizygous and first-generation backcrosses and consisted of an equal mixture of hemizygotes and null homozygotes. Hybridization opportunities for each weedy individual was calculated as the expected proportion of pollen received from transgenic plants based on the observed flowering schedules. This experimental system was ideal for addressing the question of interest in this study, for three reasons. First, despite barriers to interspecific mating such as apomixis or preferential exclusion of hybrid zygotes, numerous studies have shown that B. napus and B. rapa readily hybridize under controlled conditions, but also in the field.