**4. Gene flow from herbicide-resistant crops to wild or weedy relatives**

Hybridization and introgression are normal processes, which have continuously occurred between crops and wild or weedy relatives [49, 50], as well as between relative populations of weedy and/or wild species [51, 52]. Even though the hybridization of crops and weeds has an important role in the evolution of many weed species [53], it can also result in the extinction of certain species related to the crops or the rise of new weed forms, which are more aggressive and better adapted to artificial habitats [30]. There are three types of gene flow: *vertical* (between sexually compatible individuals), *horizontal* (between distant related species), and *diagonal* (between related but incompletely incompatible species) [54], but introgression of genes from cultivated to wild or weedy forms of the same species is possible through *vertical* and *diagonal* gene flow.

The ecological consequences of gene transfer from crops to their wild relatives are determined by the quantity of genes, which are being transferred into the populations of wild plants and weeds and the phenotypic characteristics controlled by these genes. Some of the characteristics are insignificant for the fitness of wild relatives, while others (herbicide resistance, disease resistance, and tolerance to the environmental stress factors) mostly improve it. For example, the first generation crop-wild hybrids produced through hybridization between cultivated and wild radish populations [53, 55, 56] was relatively fecund, produced large quantities of seeds and rapidly evolved increased pollen fertility. Contrary to this, if the introduced genes weaken the fitness of their wild relatives, their invisibility will also decrease. This process can be accelerated by introgression and the introduction of new genes from neighboring crops, which ultimately leads to the extinction of the initial populations of wild relatives [57]. Except ecological consequences, gene flow from crops to weedy relatives is associated with many problems in crop production. Namely, the development of HR crops has given rise to the situation where the hybridization is often seen as a problem, particularly when it relates to the hybridization between GM crops and related species. Also, it is important to bear in mind that in some countries coexist different cropping systems, which cultivate conventional, organic, and GM crops. In that situation, there is risk for gene flow between GM and non-GM cultivars through crossfertilization due to pollen flow between neighboring fields. Progeny of HR crops and weedy/ wild relatives or volunteers will be resistant weeds, in which control is difficult.

Genes responsible for crop's herbicide resistance can be spread in the environment as a result of three mechanisms, including gene transfer across a pollen (as a result of allogamy), seeds (as a result of their dispersal) and for perennial species by the vegetative propagules. Potential for pollen-mediated gene flow is higher for both wind and insect pollinated out-crossing crops than for self-pollinated crops [58]. Although gene flow across a pollen is more studied, gene flow by seeds during commerce may be very important for the long-distance dispersal of genes responsible for resistance to herbicides [59]. The both ways of gene flow from HR crops including both GM and conventionally bred HR crops have been confirmed in many cases [37–40, 60, 61].

The transfer of genes from HR crops to their relatives is dependent on multiple factors (**Figure 2**), such as the coexistence and proximity of the crop and its close relatives, their biology and phenology, type of vector, development of F1 generation, which is fertile and capable of survival, the production of fertile subsequent generations, the potential for gene transmission, chromosome recombination and movement of genes of one species into the genome of another, due to introgressive hybridization and gene persistence in volunteer crop populations [58, 62]. Also, in study about gene flow from glufosinate-resistant rice to improved rice cultivars and weedy rice in China, the conclusion was that gene flow depends on the height of pollen recipient plants [63]. They found that the gene flow was lesser if recipients were taller than in situation when they were shorter.

Cross-pollination between HR crops and sexually compatible wild or conventional cultivated crops of the same species is the major pathway for gene escape. Therefore, transfer of genes responsible for HR between sexually compatible individuals is most often done through pollen, whether within the same population or between different populations [38, 64]. This occurrence is dependent on different factors of which autoincompatibility that enhances allogamy in wild forms, environmental conditions (wind speed and direction, temperature, light intensity, and humidity) as well as the type (wind and/or insect) of pollination vector [37, 38, 65, 66]. In addition to this, the crucial role in gene transfer through pollen lies in the coincidence of the flowering period between the HR crop and its wild relatives. Although experimental data suggest

resistance, and tolerance to the environmental stress factors) mostly improve it. For example, the first generation crop-wild hybrids produced through hybridization between cultivated and wild radish populations [53, 55, 56] was relatively fecund, produced large quantities of seeds and rapidly evolved increased pollen fertility. Contrary to this, if the introduced genes weaken the fitness of their wild relatives, their invisibility will also decrease. This process can be accelerated by introgression and the introduction of new genes from neighboring crops, which ultimately leads to the extinction of the initial populations of wild relatives [57]. Except ecological consequences, gene flow from crops to weedy relatives is associated with many problems in crop production. Namely, the development of HR crops has given rise to the situation where the hybridization is often seen as a problem, particularly when it relates to the hybridization between GM crops and related species. Also, it is important to bear in mind that in some countries coexist different cropping systems, which cultivate conventional, organic, and GM crops. In that situation, there is risk for gene flow between GM and non-GM cultivars through crossfertilization due to pollen flow between neighboring fields. Progeny of HR crops and weedy/

wild relatives or volunteers will be resistant weeds, in which control is difficult.

than in situation when they were shorter.

44 Herbicide Resistance in Weeds and Crops

Genes responsible for crop's herbicide resistance can be spread in the environment as a result of three mechanisms, including gene transfer across a pollen (as a result of allogamy), seeds (as a result of their dispersal) and for perennial species by the vegetative propagules. Potential for pollen-mediated gene flow is higher for both wind and insect pollinated out-crossing crops than for self-pollinated crops [58]. Although gene flow across a pollen is more studied, gene flow by seeds during commerce may be very important for the long-distance dispersal of genes responsible for resistance to herbicides [59]. The both ways of gene flow from HR crops including both GM and conventionally bred HR crops have been confirmed in many cases [37–40, 60, 61].

The transfer of genes from HR crops to their relatives is dependent on multiple factors (**Figure 2**), such as the coexistence and proximity of the crop and its close relatives, their biology and phenology, type of vector, development of F1 generation, which is fertile and capable of survival, the production of fertile subsequent generations, the potential for gene transmission, chromosome recombination and movement of genes of one species into the genome of another, due to introgressive hybridization and gene persistence in volunteer crop populations [58, 62]. Also, in study about gene flow from glufosinate-resistant rice to improved rice cultivars and weedy rice in China, the conclusion was that gene flow depends on the height of pollen recipient plants [63]. They found that the gene flow was lesser if recipients were taller

Cross-pollination between HR crops and sexually compatible wild or conventional cultivated crops of the same species is the major pathway for gene escape. Therefore, transfer of genes responsible for HR between sexually compatible individuals is most often done through pollen, whether within the same population or between different populations [38, 64]. This occurrence is dependent on different factors of which autoincompatibility that enhances allogamy in wild forms, environmental conditions (wind speed and direction, temperature, light intensity, and humidity) as well as the type (wind and/or insect) of pollination vector [37, 38, 65, 66]. In addition to this, the crucial role in gene transfer through pollen lies in the coincidence of the flowering period between the HR crop and its wild relatives. Although experimental data suggest

**Figure 2.** Comparison of the requirements and factors affecting gene flow via pollen, seed, and vegetative propagules, modified figure from reference [58].

that the flowering period of wild populations is generally longer than the flowering period of crops, which makes the overlap highly likely [67], in some cases, gene flow between HR crops and relatives was disabled due to flowering period not overlapping or time of overlapping was short. For example, hybridization between imazamox-resistant and weedy sunflower was not confirmed in experiments in Serbia when period of flowering overlapping was short [42]. Also, it was confirmed that the gene transfer from the cultivated onto the wild sunflower in Argentina depended on the overlap between the flowering period and the presence of common pollinators [68, 69]. Pollen dispersal from HR crops onto their wild relatives is also dependent on their mutual distance, the size of populations from which the pollen originates and where it is delivered, plant density, number of flowers per plant, and the location of wild relatives in relation to the crop [70].

Although numerous studies have confirmed the transfer of genes relevant for HR to their wild relatives, hybridization level mainly was low. Some authors [60] studied the transfer of genes responsible for imazethapyr-resistance, from the rice cultivars to the weedy rice species in 22 field sites. They confirmed that even though gene transfer occurs, in the majority of sites (18) less than 1% of hybrid progeny was present, while in the remaining four sites that percentage was somewhat higher (up to 3%). Also, low levels of hybridization (1–2%) were confirmed between rice and its wild congener *Oryza rufipogon* [71]. Similarly, Ref. [72] confirmed a low level of hybridization between HR rapeseed and related weed species *Raphanus raphanistrum*. Their research has shown that the proportion of crop-weed hybrids in the F1 generation was at the level of 10−7 to 3 × 10−5, depending on the geographic position of the weed species in the experimental plot. Gene flow from glyphosate-resistant canola to *B. rapa* in commercial fields was confirmed, but the genes were apparently not fully introgressed [73]. Contrary to that, in study of transgene escaping from canola to *B. rapa*, the gene frequency in the first backcross generation was 50%. But, in the fourth backcross generation, it was 0.1% in conditions without herbicide application, while in conditions with glyphosate application, gene frequency was about 5.5% within six successive backcross generations [44].

Despite the fact that the gene transfer from crops to their wild relatives is widely studied, there are no detailed data available on what happens with these genes, which have been introduced into wild populations after a longer period of time. Namely, the majority of this research concludes with the first generation of hybrids. However, genes originating from the cultivated sunflower can persist in wild populations over the five-year period, following the hybridization [43]. Some authors [74] have also studied the effects of a 40-year long gene transfer from the cultivated to the wild sunflower populations.

Importance of crop-weed hybrids produced as result of gene flow from HR crops to wild or weedy relatives for future crop production can be different depending on traits introduced into progeny. Therefore, assessment of gene flow occurrence requires not only estimating the degree of gene flow, but also evaluating the relative fitness of hybrids. It long dominated the view that crop-wild hybrids have a lower fitness than their wild parent [75, 76]. But, many studies confirmed that some hybrids display increased [47], while the other display reduced [77] fitness in comparison with their parents. Displayed fitness depends not only on the crop traits introduced to wild relatives, but also on environmental conditions. Namely, fitness of hybrids between crop and wild sunflower increases in stressful conditions common to conventional agroecosystem like competition and herbicide application [77].

The role of seeds in the transfer of HR genes from crops to their wild relatives is evident in their spread into new areas where volunteer populations are formed. After that HR genes can be transferred from these volunteer populations to their wild relatives through the pollen. Also, hybrids resulting from spontaneous crosses of HR crops and their wild relatives through seeds can be carried into new areas, where they subsequently present a source of pollen, which carries the resistance genes. Unlike pollen, the seeds usually remain in the close proximity of the plants from which they have originated. But, as seeds are more persistent than pollen, movement of seeds is possible to further distances by human activities then pollen movement [59]. In general, seed dispersal of HR crops or progeny created through their spontaneous crossing with wild relatives, depends on the biological properties of the crop, the ecological conditions, crop production technology and the agrotechnical measures applied on these fields, following with harvest. Nevertheless, it is possible to monitor the dispersal of these seeds in space and time. Some authors [78] have confirmed the gene transfer of sugar beet to their wild relatives through the seeds whose dispersal resulted from soil transport. Namely, although spontaneous spatial dispersal is often considered as irrelevant since the seeds of a majority of crop cultures have lost this ability, seed dispersal is also possible as a result of spillage during the harvest and their transport and storage operations, which enables the spread to great distances. The dispersal of seeds containing the genes responsible for resistance over time depends on the dormancy characteristics and the seed's longevity in the soil, as well as the ecological requirements for its germination. Also, it should be considered that, in addition to pollen and seeds, soil seedbank has an important role in the plant dispersal [79]. Namely, when considering different life forms of sugar beet (cultivated, wild, and weedy), it is well known that they form long-term seedbanks [80], which, over a longer period of time, can provide the plants which are then a source of HR genes.

Argentina depended on the overlap between the flowering period and the presence of common pollinators [68, 69]. Pollen dispersal from HR crops onto their wild relatives is also dependent on their mutual distance, the size of populations from which the pollen originates and where it is delivered, plant density, number of flowers per plant, and the location of wild rela-

Although numerous studies have confirmed the transfer of genes relevant for HR to their wild relatives, hybridization level mainly was low. Some authors [60] studied the transfer of genes responsible for imazethapyr-resistance, from the rice cultivars to the weedy rice species in 22 field sites. They confirmed that even though gene transfer occurs, in the majority of sites (18) less than 1% of hybrid progeny was present, while in the remaining four sites that percentage was somewhat higher (up to 3%). Also, low levels of hybridization (1–2%) were confirmed between rice and its wild congener *Oryza rufipogon* [71]. Similarly, Ref. [72] confirmed a low level of hybridization between HR rapeseed and related weed species *Raphanus raphanistrum*. Their research has shown that the proportion of crop-weed hybrids in the F1 generation was at the level of 10−7 to 3 × 10−5, depending on the geographic position of the weed species in the experimental plot. Gene flow from glyphosate-resistant canola to *B. rapa* in commercial fields was confirmed, but the genes were apparently not fully introgressed [73]. Contrary to that, in study of transgene escaping from canola to *B. rapa*, the gene frequency in the first backcross generation was 50%. But, in the fourth backcross generation, it was 0.1% in conditions without herbicide application, while in conditions with glyphosate application, gene frequency was

Despite the fact that the gene transfer from crops to their wild relatives is widely studied, there are no detailed data available on what happens with these genes, which have been introduced into wild populations after a longer period of time. Namely, the majority of this research concludes with the first generation of hybrids. However, genes originating from the cultivated sunflower can persist in wild populations over the five-year period, following the hybridization [43]. Some authors [74] have also studied the effects of a 40-year long gene

Importance of crop-weed hybrids produced as result of gene flow from HR crops to wild or weedy relatives for future crop production can be different depending on traits introduced into progeny. Therefore, assessment of gene flow occurrence requires not only estimating the degree of gene flow, but also evaluating the relative fitness of hybrids. It long dominated the view that crop-wild hybrids have a lower fitness than their wild parent [75, 76]. But, many studies confirmed that some hybrids display increased [47], while the other display reduced [77] fitness in comparison with their parents. Displayed fitness depends not only on the crop traits introduced to wild relatives, but also on environmental conditions. Namely, fitness of hybrids between crop and wild sunflower increases in stressful conditions common to con-

The role of seeds in the transfer of HR genes from crops to their wild relatives is evident in their spread into new areas where volunteer populations are formed. After that HR genes can be transferred from these volunteer populations to their wild relatives through the pollen. Also, hybrids resulting from spontaneous crosses of HR crops and their wild relatives through seeds can be carried into new areas, where they subsequently present a source of pollen, which carries

tives in relation to the crop [70].

46 Herbicide Resistance in Weeds and Crops

about 5.5% within six successive backcross generations [44].

transfer from the cultivated to the wild sunflower populations.

ventional agroecosystem like competition and herbicide application [77].

Gene flow by the vegetative propagules (stolons, rhizomes, roots, crowns, and bulbs) is possible on short distance via natural means or on equipment moved between fields, while longdistance movement could be possible only with human activities or through the waterways [58]. As HR crops are mainly annual species, gene flow via vegetative propagules can be interesting only for perennials like glyphosate-resistant alfalfa (commercially available) and creeping bentgrass (*Agrostis stolonifera*, under consideration) [58].

*Gene flow risk assessment* is a procedure, which helps determine whether the transfer will occur, and if it will, in which degree, with a goal to reduce such a risk to the minimal possible level. Furthermore, such estimates are also significant due to the possibility that the transfer of genes responsible for HR will lead to an increase in the survival and adaptability of the introduced weed species. Also, it is considered that certain plants can attain the traits of invasive species as a result of introduced genes, making the assessments of long-term consequences of gene transfer from crops to their wild relatives a necessity. There is no same potential for gene flow for all HR crops. For example gene flow from maize is theoretically possible to teosinte, but these species only exist in Mexico and Central America and not yet been reported for contamination with transgenes [81]. On the other hand, there is good potential for introgression from sunflower [74], sugarbeet [82] and rice [83] to wild relatives. Namely, the dangers of the transfer of genes responsible for HR and the ecological consequences of this must be evaluated individually for each specific case (herbicide, plant, wild relatives, etc.), regardless of the fact whether the crop resistance has been achieved through conventional breeding methods or genetic engineering. Crucial steps in the rational assessment of ecological consequences of this phenomenon include the understanding of the following: (1) gene transfer from crops to wild relatives; (2) gene expression and inheritance in hybrids, which have resulted from the gene transfer from crops to wild relatives; (3) changes in fitness in wild relatives caused by the introduction of genes; and (4) the dynamics of the gene transfer from crops to wild populations. The invasibility of crop-weed hybrids, which have originated as a result of gene transfer from HR crops to wild relatives, is dependent on all of these aspects.

Pollen flow from crop to the relative seems as relatively simply process, but gene introgression is complex, occurring in several steps which mean several hybrid generations, which can exchange genes among themselves and coexist many years simultaneously (**Figure 3**). The likelihood of gene transfer from crops to their wild relatives depends on the genetic characteristics of crops and their wild relatives, as well as the homology of their genomes [62]. In the cases where the degree of the homology between the crops and their wild relatives is higher, as in the case of *Beta vulgaris* × *B. maritima* or *Raphanus sativus* × *R. raphanistrum*, the likelihood that the introduction of transferred genes will occur is higher. Additionally, the introduction of genes is dependent on the part of the genome, in which the gene is positioned. Some authors [30] found a possibility of gene introduction from 13 most important crops into wild relatives and determined that 12 of the studied crops can hybridize with their wild relatives. Of the 12 listed crops, cases of introduction have been confirmed for 7, while in the remaining five there is a possibility that the introduction will occur. Also, based on the potential danger of transgenic introgression into their wild relatives, some authors [76] have grouped GM crops based on their risk levels into four categories: high, middle, low, and very low (**Table 3**). A similar categorization pertaining to the risk assessment was also applied by other authors [84, 85].

In order to *prevent or reduce* the unwanted transfer of pollen from HR crops onto their relatives, different barriers can be used, although there is no absolute guarantee that the gene transfer can be prevented in this manner. The most often used barriers are isolation in space or time, protective vegetation barriers made up of one or more different species, male sterility as a genetic mechanism for the prevention of gene transfer, etc.

*Spatial* (*distance*) *isolation* means increasing the distance between fields sown by HR crops and populations of its relatives. Also, spatial isolation is applied as preventive measure in production of GM and non-GM crops in coexistence with the aim to avoid contamination products of non-GM crops. It has been known that by increasing the distance between crops

**Figure 3.** Gene flow and its potential resources, modified from reference [76].


populations. The invasibility of crop-weed hybrids, which have originated as a result of gene

Pollen flow from crop to the relative seems as relatively simply process, but gene introgression is complex, occurring in several steps which mean several hybrid generations, which can exchange genes among themselves and coexist many years simultaneously (**Figure 3**). The likelihood of gene transfer from crops to their wild relatives depends on the genetic characteristics of crops and their wild relatives, as well as the homology of their genomes [62]. In the cases where the degree of the homology between the crops and their wild relatives is higher, as in the case of *Beta vulgaris* × *B. maritima* or *Raphanus sativus* × *R. raphanistrum*, the likelihood that the introduction of transferred genes will occur is higher. Additionally, the introduction of genes is dependent on the part of the genome, in which the gene is positioned. Some authors [30] found a possibility of gene introduction from 13 most important crops into wild relatives and determined that 12 of the studied crops can hybridize with their wild relatives. Of the 12 listed crops, cases of introduction have been confirmed for 7, while in the remaining five there is a possibility that the introduction will occur. Also, based on the potential danger of transgenic introgression into their wild relatives, some authors [76] have grouped GM crops based on their risk levels into four categories: high, middle, low, and very low (**Table 3**). A similar categorization pertaining to the risk assessment was also applied by other authors [84, 85].

In order to *prevent or reduce* the unwanted transfer of pollen from HR crops onto their relatives, different barriers can be used, although there is no absolute guarantee that the gene transfer can be prevented in this manner. The most often used barriers are isolation in space or time, protective vegetation barriers made up of one or more different species, male sterility

*Spatial* (*distance*) *isolation* means increasing the distance between fields sown by HR crops and populations of its relatives. Also, spatial isolation is applied as preventive measure in production of GM and non-GM crops in coexistence with the aim to avoid contamination products of non-GM crops. It has been known that by increasing the distance between crops

as a genetic mechanism for the prevention of gene transfer, etc.

**Figure 3.** Gene flow and its potential resources, modified from reference [76].

transfer from HR crops to wild relatives, is dependent on all of these aspects.

48 Herbicide Resistance in Weeds and Crops

**Table 3.** The risk level of the introgression of genes from crops to their wild relatives ([86] made based on data reviewed by [76]).

and its relatives, the dispersal of pollen is reduced, i.e., the level of hybridization is reduced. Thus, the frequency of pollen originating from a transgene oilseed rape decrease from 1.5 to 0.00033%, as distance increase from 1 to 47 m [87]. Also, frequency of crop-wild relative hybrid decreased from 0.156 to 0.0038% with increase in distance from pollen source between 200 and 400 m [88]. Distances between pollen source and gene occurrence can be very valuable in the planning of spatial isolation of HR varieties, in order to prevent the gene flow to their relatives. This distance depends on many factors such as the presence of local barriers, the local climate, and the topography of the area. In the case of sunflower, the isolation distance should be greater than 1000 m [89]. Also, maize pollen can be detected at distances greater than 800 m from the pollen source [90]. But, pollen of maize has short flight range [91], after which it settles to the ground rapidly [92] due to relatively heavy and large grains. Due to that cross-fertilization mainly occurs within 50 m of the pollen source [93]. Therefore, measure for keeping seed purity of non-GM maize, which coexists with GM maize, suggests isolation distance between 10 and 50 m to achieve EU admissible threshold of 0.9% in the harvest [37, 93, 94].

*Temporal isolation* is a measure, which should prevent overlapping flowering times of crop and wild relatives with the aim to avoid gene flow. About 5 days lag in flowering of imazamoxresistant in comparison with tribenuron-methyl resistant sunflower resulted in lack of gene flow to weedy sunflower probably due to the short period of overlapping flowering time between the resistant hybrid and the weedy sunflower [42]. Temporal isolation is very suitable to prevent non-GM crop contamination with GM when grow in coexistence. Study of maize pollen mediated gene flow in Italy and showed that if time of flowering differs from 4 to 5 days the cross-pollination is reduced by 25%, while difference of 6 days provides 50% reduction [95]. Also, temporal separation in sowing days improves the coexistence of maize [96]. Contrary to that, temporal isolation based on selection of hybrid varieties in which flowering noncoincide achieves the same results although sowing date was the same [97]. Temporal separation and isolation distance together can be a good solution to minimize unwanted gene flow.

*Protective vegetation barriers*, made up of one or more different species, can prevent the gene flow by physically stopping pollen in the case of wind pollination. The sowing of conventional crops of the same species, also known as pollen traps, in the vicinity of HR hybrids is an efficient measure, as their role is to attract pollinating insects in order to leave pollen on these pollen traps. This type of barrier can be much more effective than isolation distance. Namely, the sowing of pollen traps between GM and conventional crops is the most efficient measure for the prevention of gene flow [98]. Also, gene flow through pollen from the HR oilseed rape decreases rapidly with the increase of pollen source distance, with the added necessity of a protective vegetation barrier or pollen traps [99]. Some authors [95] studied pollen-mediated gene flow between GM and non-GM maize and concluded that effect of two maize rows surrounding the recipient field in reduction of cross-fertilization is the same like effect of 12 maize rows surrounding the pollen donor.

The use of *biological barriers* achieves the best results in the prevention of gene flow, and so far the barriers based on cytoplasmatic male sterility, maternal inheritance, and seed sterility have mostly been used. Cytoplasmic male sterility is based on the inability of plants to produce viable pollen. This type of barrier is suitable option to reduce gene flow in sunflower and maize [43, 66, 100]. Maternal inheritance is successfully used in the prevention of gene flow across the pollen, in the case of several species, including tobacco and tomato [101, 102]. The control of embryo and seed fertility is known as GURT (Gene Use Restriction Technology), i.e., terminator technology, which is considered to be a better control measure, in comparison with sterile pollen production. However, this strategy is seen as the most controversial control measure for limiting genes flow. Additionally, strategies, which include apomixis (vegetative reproduction and asexual seed formation), cleistogamy (self-fertilization without the opening of flowers), genome incompatibility, chemical induction/deletion, etc., are also used in limiting the gene flow [103]. None of these strategies can be applied in all crops, therefore using combinations of different approaches for the prevention of unwanted gene flow is recommended.

All mentioned measures for prevention and reduction of gene flow are important separately, but their integration and combination with stewardship production system could be the best solution.
