If humans select crops that grow densely in monospecific stands, are those plants only a few allele changes away from becoming invasives or weeds ? If such is the case, careful and thorough evaluation should accompany the development of new crops designed to answer pressing societal needs, including new bio-energy feedstocks created with the goal of providing a stable domestic energy supply .Escape and establishment of crop species outside of agricultural systems, known as crop ferality, can be a concern if the escaped crop contains novel traits and establishes self-perpetuating populations within agricultural landscapes. A feral population can be found by the dispersal of seeds from the agricultural fields to adjacent habitats such as the roadsides. Human mediated dispersal, such as seed spill from farm machinery and seed transport trucks or hitchhiking on a vehicle, is the main vector for crop seed dispersal out of agricultural fields. Animals can also disperse seeds, but to a lesser extent than humans both in terms of the distance and number of propagule that they can disperse. The persistence of feral populations, similar to any plant population, can be influenced by the spatial and temporal heterogeneity of the habitat. In a heterogeneous landscape, certain sites could be unfavorable for the establishment of feral populations and at these sites the population may only sustain through continuous immigration of propagule from the source sites. Arrival of large number of propagules would increase the survival chance of the population.
While propagule dispersal is a critical first step for the establishment of feral crop populations in natural areas, cannabis dryer certain habitat characteristics such as vegetation density and drainage potential can also influence the establishment of feral populations in these areas. Additionally, seed dormancy and the ability to establish a seedbank would allow the feral populations to persist in natural habitats and enable recovery from stochastic fluctuations of the environment. Sorghum is a major crop in terms of production, ranking fifth worldwide and in the US. As a drought tolerant crop, sorghum can be grown in areas where the extremes of high temperatures and low soil moistures are unsuitable for the production of other row crops such as corn. Similar to many other feral crops, sorghum may have the ability to establish self sustaining populations outside of cultivated fields. The occurrence of feral crops along roadsides might be attributed to seed dispersal and the peculiarity of this habitat such as increased water runoff and low plant community richness; these characteristics may favor the initial establishment of feral sorghum in roadside habitats. The ferality potential in sorghum will be of concern because of sympatric presence of weedy relatives that can outcross with cultivated sorghum. In Southern US, the weedy relatives of S. bicolor include shattercane and johnsongrass. Of these, johnsongrass is known to be the most widely distributed and frequently found relative of sorghum in South Texas . Johnsongrass is one of the most troublesome weeds in the world, capable of spreading by both underground rhizomes and seeds.
Johnsongrass can cause severe yield losses in sorghum and many other crops. Both sorghum and johnsongrass are interfertile and can be hybridized under controlled conditions. Gene flow from sorghum to johnsongrass has also been observed in natural conditions. For sorghum cultivars bred or engineered with adaptive or herbicide resistance traits, ferality can be a concern as it can facilitate the establishment of these traits in the broader environment, causing ecological and/or agronomic issues. The co-occurrence of feral sorghum and johnsongrass in the proximity would increase the chances of cross-pollination between the two species. Pollen-mediated gene flow from johnsongrass to sorghum may enhance ferality in sorghum through de-domestication and the provision of adaptive alleles. The majority of commercial sorghum cultivars grown in the US are hybrids, and cytoplasmic male sterility is used to produce hybrid seeds. Male sterility is a recessive trait and male fertile F1 hybrids are actually in a heterozygous condition for alleles conditioning fertility restoration. Consequently, segregation for male sterility would be expected in progeny from the hybrid. In fact, approximately 25% of the feral sorghum plants established through seed dispersal will be male sterile. In these circumstances, the co-occurrence of feral sorghum and johnsongrass presents an increased chance for outcrossing. Texas is the second largest sorghum producer in the US, closely following Kansas. In South Texas, the majority of sorghum production is concentrated in the Rio Grande Valley, Coastal Bend and Upper Gulf Coast regions. Sorghum grown in this region is frequently transported to Mexico along highways and railroads through the Rio Grande Valley.
Sorghum seed spill along the transportation routes could lead to the establishment of feral sorghum populations in the byways along these routes and in fact, sorghum is seen commonly along the major highways in South Texas. However, no systematic survey has been conducted in the region to document the occurrence of feral sorghum along roadsides and the extent of sympatry with johnsongrass. The objective of this study was to document the prevalence of feral sorghum and johnsongrass along roadside habitats in South Texas and underpin, using GIS and logistic regression models, its association with several anthropogenic and environmental factors.A field survey was conducted along the roadsides of South Texas, from the Rio Grande Valley to the Upper Gulf Coast where grain sorghum cultivation is prevalent. The climate of the survey region is humid subtropical with mild winters and warm summers. Sorghum is planted from mid-February in the Rio Grande Valley to early April in the Upper Gulf Coast regions, with harvest occurring in about four months after planting. Due to the long growing seasons in South Texas, sorghum seed that disperses after harvest will have a chance to germinate and attain reproductive maturity prior to a killing frost during late fall season. In the Rio Grande Valley, frost occurs very rarely and a second sorghum crop can be planted in early August. The survey was conducted during late October- early November 2014 to allow the feral sorghum along the roadsides to establish and mature. However, the survey also included feral sorghum plants that recruited during spring. The survey area was divided into three regions based on distinct environmental conditions: 1) Upper Gulf Coast, from west of Houston, TX to Victoria, TX , 2) Coastal Bend, from Victoria, TX to Kingsville, TX , and 3) Rio Grande Valley, from Kingsville, TX to Brownsville, TX . Survey sites within each region were chosen using a semi-stratified survey methodology, as described by Bagavathiannan and Norsworthy. One hundred survey sites were selected at random within each region and survey routes were optimized using the ITN Converter software on a Google1 map layer. The ITN files were loaded to a GPS device to facilitate navigation to the pre-determined survey sites. In each site, the presence/absence of feral sorghum and johnsongrass was recorded. If feral sorghum was present, observations were carried out on the feral population size and site characteristics within a 25 m strip along the roadside site . If feral sorghum was absent in a pre-determined survey site, the first population found along the route to the next pre-determined site was used for characterization. No specific permissions were required for the activities carried out in this project. Moreover, cannabis growing systems the authors confirm that the field studies did not involve endangered or protected species.At each site where feral sorghum was present, observations were also carried out on the road body-type , micro-topography of the site , vegetation cover of the habitat , and nearby land use type . The micro-topography category ‘road shoulder’ represents the area immediately adjacent to the road margin towards the deepest point of the ditch, ‘field shoulder’ represents the area from the deepest point of the ditch to the field edge. The ‘field edge’ represents the edge of the cultivated field.
The road and field shoulders typically have high vegetation cover and minimally disturbed, whereas the field edges are often tilled and have relatively less vegetation. The co-occurrence of feral sorghum and johnsongrass was defined when both species were present within 50 meters of each other. To understand whether there is a relationship between the presence of feral sorghum and distance to grain sorting facilities, locations of such facilities were recorded during the survey. Data pertaining to the nearby land use type and the presence/absence of feral sorghum and johnsongrass were used for developing a projection model for species distribution, as described below.For each sampled site, the nearest road type was identified using Texas road maps obtained from Texas Natural Resources Information System online database . The road type classifications were county roads, highways, local streets, federal roads, functional classification streets and third-party toll roads. However, the recorded feral sorghum populations were predominantly found in county roads, highways and FC streets and not the other road type categories. As shown in Fig 1, only a sub-region of South Texas, representing the latitudinal and longitudinal limits of anticipated feral sorghum distribution, was included in the analysis. To determine the nearest road type to a given sample site, we first converted the road vector lines to a raster format using the Qgis software . Prior to rasterization, the character strings were converted to numerical values; for example, a highway road type was assigned a numeric value of 1while a county road was given a value of 2, and so forth. For each individual road type, a proximity map was produced which gives the distance of each grid cell on the map from the given road type . The georeferenced sampled sites were then overlaid on these road proximity maps and the raster value of each individual proximity map was appended to these points sequentially. The nearest road type to a given sample site was determined as the road with the smallest proximity value. To determine the nearby land use for the sampled points, we used the very high resolution CropScape database. We drew a buffering circle with a radius of 90 m around each sampling point and then converted the resultant circle polygons to a raster format using the values of CropScape raster layers as the required data field for the conversion. Similar to road type data, nearby land use data are presented using numerical values; for example, a sorghum crop is identified by the value ‘4’ . Using the Zonal Statistical feature of Qgis, the mode was calculated for each circle: mode represents the land cover with the highest frequency for the circle and thus dominant crop adjacent to the sampled sites. The distance of the sampling site to the nearest grain handling facility was calculated using the Matrix Distance feature of Qgis.Using a binary logistic model, we modelled the probability of the presence of feral sorghum as a function of the road type , road body-type , micro-topography of the surveyed site , the nearby land use , presence/ absence of johnsongrass and distance to grain sorting facilities. The above model was fitted using the PROC LOGISTIC procedure of SAS . The ‘odds ratio’ feature of PROC LOGISTIC was used to calculate the odds ratio and test for significance of the differences between the levels of predictors. Prior to analysis, all data were reordered to reduce the number of categories and avoid the effect of ‘quasi-complete separation of data points’. For example, 40 categories were initially recorded for nearby land use but were then reclassified to 16 groups. At 360 sites , the density of the feral sorghum population was recorded on scale of 1 to 5 . The effects of the road type, road body-type, micro-topography of the site, the nearby land use, the vegetation cover and presence/absence of johnsongrass on the scores of feral population density was investigated using PROC GLM of SAS and means were separated using the least significant differences test.A total of 2,077 sites were visited for the presence of feral sorghum and johnsongrass in our survey. Feral sorghum was found in 17% of the sites, whereas johnsongrass was more abundant and found in 45% of the sites visited. To our knowledge, this is the first documented account of the occurrence of sorghum as feral populations outside of cultivated fields. Results from the logistic model showed a significant association between the presence of feral sorghum with the road type and its body-type, micro-topography of the site, nearby land use and the presence/absence of johnsongrass, but showed no relationship with distance to the nearest grain sorting facility .