Ongoing ecological and evolutionary consequences by the presence of transgenes in a wild cotton population
In this study, we showed that the expression of cry and cp4-epsps genes in wild cotton altered the secretion of EFN, the associations with different ant species, and the levels of herbivore damage on target plants. Wcry constantly maintained a high production of EFN, regardless of the MeJA treatment, but nectar production was minimal in Wcp4-epsps. These changes in nectar inducibility seem to modify the composition of ant communities, foster the dominance of the generalist and defensive species C. planatus in Bt plants and the presence of ants without defensive role, M. ebeninum, in the herbicide tolerant genotype, while W plants had both defending species (C. planatus, C. rectangularis aulicus and P. gracilis) and invasive ant species (P. longicornis) in the same proportion. Furthermore, herbivore damage and its associated ant community were different according to the introgressed transgene.
Wild and introgressed cotton do not display phenotypic equivalence in natural conditions
In general, it has been assumed that introgressed and wild genotypes should display similar phenotypes in the absence of the selection agents targeted by transgenes. However, when we compared the control group and the three genotypes, we registered different nectar secretion patterns among them (Fig. 1). Similar results have been registered in populations of bt rice and glyphosate-tolerant sunflowers living in natural conditions where introgressed plants are different from their wild relatives5.
Transgene expression modified indirect induced defences in wild cotton
Most plants are able to induce responses after herbivore damage and/or phytohormone exogenous application (i.e. jasmonic acid, JA; methyl jasmonate, MeJA; and salicylic acid, SA)11,28,29. However, unlike wild plants without transgenes, individuals with transgenes were not sensitive to the induction treatment with MeJA for increasing their EFN production (Fig. 1). These results contrast with previous reports on cultivated varieties, such as Bt and glyphosate-resistant (cp4-epsps), in which direct defences such as gossypol terpenoids (160%), hemigossypolone (160%), helicoids 1|4 (213%) and indirect defenses, such as volatile compounds (VOCs) (171.2%) and extrafloral nectar (EFN) (133%), were reported to increase in plants sprinkled with JA and MeJA21,28,29,30.
The inability of plants with transgenes to have the production of extrafloral nectar induced in them was related to different processes dependent on the identity of the transgenes in question. Whereas Wcry control plants had a high EFN production equivalent to the induced state of W plants, EFN production in Wcp4-epsps plants was inhibited. Contrasting these findings with results obtained under controlled conditions (i.e. greenhouse and crop conditions)3,21, we suggest that EFN production is linked to genotypes with transgenes and abiotic stress in the coastal dunes, because transgenes are connected to main metabolic pathways that respond to stressful conditions21.
Wild cotton with cp4-epsps
In the absence of herbicides acting as a selection agent, wild plants with cp4-epsps exhibited large differences compared to wild plants without them. Their low nectar production ( > 8 µg/mL) (Fig. 1) could be linked to the crosstalk between the jasmonate and the salicylate (SA) pathways (Fig. 4, orange and purple section). In G. hirsutum and other species, SA signalling has been proven to negatively affect JA signalling (e.g. Zea mays, Solanum lycopersicum, Nicotiana tabacum and Arabidopsis thaliana)31,32,33: therefore, we suggest an interference between the SA and JA pathways given previous reports that an over-expression of the cp4-epsps gene modifies the second part of the shikimate pathway (post-chorismate), which leads to the synthesis of essential amino acids as phenylalanine, tryptophan, or tyrosine, the latter being a precursor of benzoic acid BE, and SA34,35 (Fig. 4, purple section). This evidence highlights that hidden crosstalk effects among different metabolic pathways can scale up and modify plant phenotypes (e.g. extrafloral nectar production).
Figure 4
A diagram illustrating how the expression of cry (A) and cp4-epsps (B) in absence of their selection agent (pests and glyphosate) can affect the extrafloral nectar production. The extrafloral nectar (EFN) production is an induced defence that can be triggered by foliar herbivory, mechanical damage, and exogenous application of phytohormones (i.e. jasmonic acid, methyl jasmonate, and salicylic acid). These factors activate the octadecanoid pathway, and therefore, the production of extrafloral nectar, (A) aqua rectangle. The (C) section is an example of this reaction in a wild cotton plant (without transgenes). After damage, the key genes (yellow mesh) of the octadecanoid pathway are activated and produce extrafloral nectar. Another scenario is when the wild cotton expresses cry genes (A section), in this case, the key genes of the octadecanoid pathway interact synergistically with the cry transgene (green mesh). This triggers an over-expression of the production of EFN (aqua thick arrow), switching from inducible to constitutive responses. When the plants express cp4-epsps (B section), the production of extrafloral nectar is reduced or inhibited. A possible answer is an over-expression of the epsps gene (gold curve arrow), that increased production of salicylic acid which creates a crosstalk between shikimate and octadecanoid pathways (black cross-talk arrow). When the shikimate pathway is activated, the principal inducible defence is the production of volatile organic compounds (VOCs) (pink rectangle).
Full size image
Wild cotton with cry
Wild cotton plants with cry genes continuously produced EFN as a constitutive defence (Fig. 1), in equivalent quantities as the induced state of W plants. EFN production is regulated by the octadecanoid signalling pathway, which can be activated by herbivore damage, mechanical damage, and phytohormones, such as JA and MeJA21,28 (Fig. 4, green section). However, for cotton, a specific elicitor is not necessary36. Four key genes for the synthesis of JA and MeJA have been described: AOS, AOC, HPL, and COI137. In Bt maize, studies comparing GM corn and its isogenic lines report an increase of 24% in phenols and 63% of DIMBOA (2,4-dihidroxi-7-metoxi-1,4-benzoxazin-3-ona; natural defences against lepidopteran herbivores)11. This is consistent with observations of a synergy between maize direct defences and Bt genes, after exogenous applications of JA (Fig. 4, orange section). Considering the latter, we suggest that Wcry cotton may present a similar response, as the genes activating the JA pathway are GhAOS and GhCOI1 (homologs to maize JA biosynthesis genes: ZmAOS and ZmCOI1), in addition to Ghppo1, which confers natural resistance to lepidopteran pest, such as H. armigera38. The interaction of cry with other genes could modify the production of EFN in Wcry plants.
Effect of the transgenes’ expression on ants associated to wild cotton
We identified eight species of ants harvesting EFN (Table 2), but with distinctive communities as a function of the plant genotype. This result suggests that the change in quantity, and possibly the composition and quality of EFN, can influence the ant community associated with G. hirsutum39,40,41.
Changes in plant reward production could potentially compromise the attraction of natural enemies of herbivores42. In our study, the availability of EFN was modified. Although species richness was the same as in W plants (Table 2), the most abundant ant species associated with Wcp4-epsps plants, M. ebeninum, is considered a generalist species. Moreover, due to the lack of aggressive behaviour, this species does not represent an effective biotic defence43. The high abundance of this non-defensive species could be associated with the greater herbivore damage observed in Wcp4-epsps plants (Fig. 2). In contrast, W or Wcry plants showed a greater abundance of more aggressive ant species such as C. planatus, C. rectangulatus, and P. brunneus and significantly less herbivore damage.
In Wcry cotton, the community of patrolling ants was mainly dominated by C. planatus, in both treatments (control and induction). Interestingly, although the amount of nectar did not vary between treatments, the abundance of ants was significantly different. The dominance of a single ant species could have benefited the plants with increased indirect defence, reducing herbivore damage and promoting a greater seed production per plant, as described in Turnera ulmifolia44, Schomburgkia tibicinis45, and Opuntia stricta42. However, considering the aggressive and dominant behaviour of C. planatus, there may be ecological costs through antagonistic relationships with pollinators. Ants can interrupt pollination and affect plant fitness25,46,47. The outcome of these mutualistic and antagonistic interactions requires further study.
Effects of transgenes on herbivore damage
Considering that the type of mutualism that cotton sustains with ants is defensive, we suggest that the change we observed in the composition of ants is likely to have influenced herbivore damage in the different genotypes, which in turn has the potential to reduce fitness as shown by other studies of cotton48,49,50. However, a study carried out on wild upland cotton reported that plants tolerate intermediate levels of leaf damage inflicted by leaf-chewing insects ( More
