Interspecific variation in resistance and tolerance to herbicide drift reveals potential consequences for plant community co-flowering interactions and structure at the agro-eco interface

Abstract

Background and aims: When plant communities are exposed to herbicide 'drift', wherein particles containing the active ingredient travel off-target, interspecific variation in resistance or tolerance may scale up to affect community dynamics. In turn, these alterations could threaten the diversity and stability of agro-ecosystems. We investigated the effects of herbicide drift on the growth and reproduction of 25 wild plant species to make predictions about the consequences of drift exposure on plant-plant interactions and the broader ecological community.

Methods: We exposed potted plants from species that commonly occur in agricultural areas to a drift-level dose of the widely used herbicide dicamba or a control solution in the glasshouse. We evaluated species-level variation in resistance and tolerance for vegetative and floral traits. We assessed community-level impacts of drift by comparing the species evenness and flowering networks of glasshouse synthetic communities comprised of drift-exposed and control plants.

Key results: Species varied significantly in resistance and tolerance to dicamba drift: some were negatively impacted while others showed overcompensatory responses. Species also differed in the way they deployed flowers over time following drift exposure. While drift had negligible effects on community evenness based on vegetative biomass, it caused salient differences in the structure of co-flowering networks within communities. Drift reduced the degree and intensity of flowering overlap among species, altered the composition of groups of species that were more likely to co-flower with each other than with others and shifted species roles (e.g. from dominant to inferior floral producers, and vice versa).

Conclusions: These results demonstrate that even low levels of herbicide exposure can significantly alter plant growth and reproduction, particularly flowering phenology. If field-grown plants respond similarly, then these changes would probably impact plant-plant competitive dynamics and potentially plant-pollinator interactions occurring within plant communities at the agro-ecological interface.

Keywords: Co-flowering; agro-eco interface; anthropogenic stress; community; dicamba; dicamba drift; drift; flowering time; herbicide; herbicide drift; interspecific variation; network; pesticide; phenology; resistance; tolerance; weeds; wild flowers.

Fig. 1.

Conceptual framework for assessing the impact of herbicide drift on co-flowering interaction networks. (A) Hypothetical flowering phenologies for four plant species (different coloured lines) in a herbicide drift-unexposed (left) and exposed (right) community over a growing season. (B) Corresponding co-flowering interaction networks for the four hypothetical plant species (different coloured flower icons) based on flowering deployment shown in (A). Links between species represent co-flowering interactions (flowering overlap between species). The thickness of the lines reflects the strength of interactions (duration and intensity of flowering overlap). Different coloured filled circles represent different modules (groups of species that interact more strongly, i.e. are more likely to co-flower, with each other than with other species); different coloured lines indicate when species within modules (green or pink) or from different modules (grey) are interacting. The size of the circles reflects species betweenness centrality (the average percentage of shortest paths in the co-flowering network that must go through a species, i.e. the relative importance of a species to network stability).

Fig. 2.

Plant species vary in resistance and tolerance to dicamba drift and in how drift affects floral traits. (A) Estimated marginal means ±95 % confidence intervals show the proportion of undamaged leaves 48 h after dicamba drift treatment by species, i.e. resistance scores. The vertical dashed line at 1 is a reference for no damage. (B–F): Contrast estimates ±95 % confidence intervals show the difference between dicamba drift-treated plants and control plants, in short-term tolerance (i.e. plant size at 21 d post-treatment; B), long-term tolerance (i.e. final biomass at 145 d post-treatment; C), day of first flower (D), flowering duration (E) and biomass per flower (F). Red denotes species that (A–C) were significantly negatively impacted by dicamba drift, (D) dicamba drift delayed the day of first flower, (E) shortened flowering duration or (F) decreased biomass per flower. Light blue shows significant effects in the opposite direction and black indicates no significant change. See Supplementary data Tables S2 and S5–S9 for results of tests of significance. Species are designated by four-letter codes as in Table 2. Values plotted are back-transformed (see Supplementary data Fig. S4 for transformed data used in statistical models).

Fig. 3.

Co-flowering networks of control and dicamba drift-exposed synthetic glasshouse communities. (A and B) Full networks when all flowering species (n = 22) are represented in the control (A) and drift (B) synthetic plant community. Each plant species is represented as a circle, and links between them represent co-flowering interactions. The thickness of the lines reflects the strength of co-flowering overlap (duration and intensity), and circle size reflects species betweenness centrality (the relative importance of species for network stability). (C and D) Betweenness centrality for each species according to the full networks in rank order for the control (C) and drift (D) community. High values reflect higher relative importance in the network. (A–D) Different colours represent different modules (groups of species that co-flower more strongly with each other than with other species). See Table 2 for species codes noted in circles (A and B) and on y-axes (C and D), and Supplementary data Fig. S11 for results of sub-set networks that only show species that flowered in both communities.

Fig. 4.

Species-level tolerance is correlated with a change in co-flowering interactions between dicamba drift-exposed and control synthetic communities. Species (blue points labelled with four-letter codes; Table 2) and long-term tolerance scores (Table 1; Fig. 2C) correlated with the change in (control subtracted from drift) weighted degree (A; Table 3) and log-transformed betweenness centrality (B) between the dicamba drift and control glasshouse communities.