Non-adaptive phenotypic plasticity: the effects of terrestrial and aquatic herbicides on larval salamander morphology and swim speed

Abstract

Phenotypic plasticity, although ubiquitous, may not always be advantageous. Non-adaptive plasticity is likely to occur in response to novel environmental stress. Anthropogenic contaminants, such as herbicides, are novel stressors that are not present in the evolutionary history of most species. We investigated the pattern and consequences of phenotypic plasticity induced by four glyphosate-based herbicides (two terrestrial and two aquatic) in larvae of the spotted salamander, Ambystoma maculatum, by determining (1) whether the herbicides induced different morphologies; (2) if different morphologies translated to differences in burst swim performance; and (3) how induced individuals performed relative to non-induced controls. Different herbicide formulations led to the production of significantly different head and tail morphologies, and tail morphology correlated with fastest escape speed. However, escape speed did not vary among treatments. In addition, three out of four herbicide treatments experienced accelerated growth rates, in terms of the lateral size of tails, although the tail shapes were either similar to preliminary controls or intermediate between preliminary and final controls. These observations suggest that herbicide-induced morphology is a case of non-adaptive phenotypic plasticity, and that there is potentially a trade-off between growth and development for larvae exposed to different formulations. Understanding the functional significance of induced phenotypes is important for determining their importance in shaping an organism's ecological interactions and evolutionary trajectories. Furthermore, under different conditions, the morphological changes that we observed in response to exposure to herbicides might affect salamander fitness and influence population dynamics.

Sexual dimorphism modifies habitat‐associated divergence: Evidence from beach and creek breeding sockeye salmon

“…Statistical testing was based on a nonparametric multivariate analysis of covariance (MANCOVA) procedure based on a randomized residual permutation procedure (RRPP) developed by Collyer et al. () for high‐dimensional morphometric data and extended by Levis, Schooler, Johnson, and Collyer () to allow the inclusion of random effects. To perform these analyses, we used the function procD.lm from the geomorph package (Adams & Otarola‐Castillo, ; Collyer et al., ).…”

“…Spawning habitat (beach or creek), lake system (Iliamna or Wood River) and their interaction were included as fixed factors. To determine whether population of origin influenced shape within a habitat type, we used log‐likelihood ratio tests to compare models with and without a random variable of sampling site (Levis et al., ). Following Levis et al.…”

“…Following Levis et al. (), when testing the importance of the random site effect, we included only the RWs that together accounted for roughly the first 95% of shape variation (RW1‐RW10), which avoided difficulties with model convergence. Models without the random site effect included all 112 RWs as response variables.…”

“…Given this result, and the statistical challenges associated with mixed MANCOVA (e.g. see supplemental information for Levis et al., ), we excluded the random effect of site. Thus, we analysed a model that included only fixed factors (habitat, lake system and their interaction) and the centroid size covariate.…”

Making Heads or Tails of Combined Landmark Configurations in Geometric Morphometric Data

“…(In this example, the separate sub-configurations are not fixed in terms of their spatial relationship to each other.) These data were originally summarized in Levis et al (2016), and are available in the R package, geomorph (Adams et al 2020). The data consist of landmarks and semilandmarks for both heads (26 points) and tails (64 points) of 114 specimens.…”