Skip to main content
Dryad logo

Climatic history, constraints, and the plasticity of phytochemical traits under water stress


Diethelm, Aramee et al. (2022), Climatic history, constraints, and the plasticity of phytochemical traits under water stress, Dryad, Dataset,


Environmental stress can induce changes in organismal traits and in resulting intraspecific variation. The nature of such effects will depend on the plasticity of trait expression and on any ecological constraints to such expression. Plants can mitigate abiotic stress, like drought, by changing their chemistry, but the ability to induce costly metabolites may be under strong local selection and ecologically constrained. Here we asked whether climate at the seed source predicts plant chemical plasticity in response to water stress and what the consequences are for intraspecific variation in phytochemical traits. To this end, we used common gardens of two widespread species of western milkweed (Asclepias fascicularis and Asclepias speciosa) that had been collected from sites across an aridity gradient. Both species produce high concentrations of leaf flavonols, which are hypothesized to mitigate water stress by functioning as antioxidants. These compounds were found in higher constitutive concentrations in plants sourced from drier sites, and both species responded to water stress in the common garden by increasing leaf flavonol concentrations. Interestingly, flavonol plasticity was higher in plants sourced from wetter sites in A. fascicularis, with similar, but weaker, patterns in A. speciosa. These opposing patterns in constitutive and induced flavonol expression reduced the variation between populations in leaf flavonol concentrations under water stress. These results suggest that local adaptation in plants can shape phytochemical strategies for water limitation but that the cost of metabolite production may ultimately limit the range of phytochemical variation.


To seed the experiments, we collected seeds from six sites spanning 385 km of the Great Basin Desert, USA, in fall 2016. From highest to lowest climatic water deficit (mm), our seed-source sites were: Fallon, NV (FN; 988.98 mm), Auburn, CA (CA; 985.68 mm), Pyramid Lake, NV (PL; 919.62 mm), Battle Mountain, NV (BM; 847.74 mm), Reno, NV (RN; 587.19 mm),  and Verdi, NV (VE; 460.3 mm).

To determine whether intraspecific trait variation could be predicted by seed-source CWD and how such variation is affected by water limitation, we conducted a drought experiment with A. fascicularis and A. speciosa in a glasshouse. We germinated 12 A. fascicularis seeds from each of three maternal families from all six sites (N = 216), and 12 A. speciosa seeds from each of three maternal families from four of the sites (FN, PL, BM, and RN; N = 144). We randomly assigned six plants from each maternal family to the control (well watered) treatment and the other six plants to the dry treatment (n = 36 per seed-source site). Prior to beginning the dry treatments, mortality was higher for A. speciosa than for A. fascicularis, such that the final average n was ~33 for A. fascicularis but ~17 for A. speciosa. We used a gravimetric dry-down treatment to expose plants to drought stress for four weeks in March–April 2017. Seeds were germinated in Petri dishes under lights (L18:D6) at 25° C in November 2016. Plants were grown in 4x9.5-in treepots containing 1500 g of a mixture of 2:1:1 parts sand:peat moss:composted bark. Pots were completely randomized on tables in the glasshouse and fertilized weekly with 24:8:16 N:P:K fertilizer. We calculated gravimetric soil water content using 13 treepots with 1500 g of the same soil. Saturated mass was measured 2 h after fully saturating the pots; dry mass was measured after oven drying for 48 h at 90° C. 100% soil saturation was estimated as: saturated mass dry mass. We allowed control plants to dry to 70% soil saturation and dry plants to 10% soil saturation.

To verify the efficacy of our drought treatment and to explore differences in physiological responses to drought, which may mediate the plant’s metabolic allocation, we measured the following plant traits: change in plant height; whole-plant dry biomass of roots and shoots; root:shoot ratio; leaf mass per area (LMA); and stomatal conductance. Plant height was recorded prior to beginning the dry treatment and again prior to harvesting the plants. Roots, stems, and leaves were harvested separately, washed, and dried at 60° C for 72 h before weighing. Tissues were weighed in microcentrifuge tubes, and the weight of the tube was subtracted. Prior to weighing, 41 microcentrifuge tubes that had contained stem tissue were accidentally discarded; we thus present results on both root:shoot ratios and root:leaf ratios. A leaf in position three of the phylotaxis was collected for LMA, which was estimated by dividing the dry weight of the leaf in mg by the estimated leaf area in mm2 (length x width). Stomatal conductance was measured between 1200–1400 h using an SC1 Porometer (Decagon).

To determine how water stress affected phytochemical trait expression, we measured plant UV-absorbent secondary chemistry. Prior to harvest, leaves and fine roots were collected and stored at -80° C. These tissues were later freeze-dried, ground, and extracted in 100% methanol with a cardenolide internal standard (digitoxin). UV-absorbent peaks were measured on a high-performance liquid chromatography (HPLC) system with a diode array detector recording peaks that absorbed between 200–330 nm. We retained peaks for our analysis that could be consistently identified from mass fragments using low-resolution mass spectrometry and/or that were present in a majority of the samples of a given species (A. fascicularis or A. speciosa) and respective tissue (leaf or root).

All analyses were conducted in R version 4.0.5 (R Core Team 2021). To determine whether trait plasticity under acute water stress depended on climatic history at the seed source, we used generalized linear mixed models (GLMMs) from the glmmTMB package. Preliminary analyses showed strong trait differences and a lack of correspondence in phytochemical compounds between species, such that we ran separate models for each species. Each saturated model started with the fixed effects of water treatment, seed-source CWD, the water x CWD interaction, and the random intercept effects of plant maternal family nested within the seed-source site. To compare effect sizes among response variables, we normalized responses and CWD values using the BBmisc package, and we report beta coefficients (β) with standard errors. We assessed the residuals of each fitted model, and we square-root or log-transformed response variables when these transformations provided a better fit to the Gaussian distribution.

Usage Notes

A ReadMe file will be attached with the data.


University of Nevada, Reno

Alexander von Humboldt-Stiftung