Low toxin doses change plant size distribution in dense populations - Glyphosate exposed Hordeum vulgare as a greenhouse case study.

Numerous intentionally released toxins persist in agricultural or natural environments at low concentrations. Such low toxin doses are regularly associated with hormesis, i.e., growth stimulation, and they are suspected to affect mortality and within-population plant size distribution in dense plant stands. However, it is not known whether all these low-dose effects exist when plants grow in soil. We exposed barley to a range of low glyphosate doses and let the plants grow in dense stands for several weeks in soil. Six experiments were done that contained altogether 10,260 seedlings in 572 pots. We evaluated if the changes in average biomass and shoot length occur at the same concentrations as do the effects on slow- and fast-growing individuals, if seed size or early vigor explains variation in the response to glyphosate, and if low toxin doses change within-population mortality. Plant biomass, length and survival of subpopulations changed at doses that did not affect mean biomass. Effects of early vigor faded early, but differences in seed size and particularly vegetative growth had impacts: fast-growing plants hardly showed hormesis, whereas hormesis was particularly strong among slow-growing individuals. Compared to the population mean, glyphosate effects started at lower doses among slow-growing individuals and at higher doses among fast-growing individuals. Several times higher doses were needed before the fast-growing individuals showed the same toxicity as most of the population. Low toxin doses regularly enhanced the growth of the smallest individuals, which reduced size variation within populations and was associated with a higher number of surviving plants. Indeed, in one experiment self-thinning was not observed at low doses that stimulated the growth of slow-growing plants. As glyphosate levels in this study match those observed in agricultural fields and natural environments, we conclude that even low-levels of agro-environmental contamination are likely to shape phenotypic response, which might lead to adaptation and cascading ecological impacts.