The response of herbivorous insects to plant drought stress can range from positive to negative, and it has been challenging to understand the causes of this variation. We tested whether plant trait values associated with aridity gradients might underlie this variation and how effects vary among two insect-feeding guilds. Here we propose that plant trait values associated with adaptation to arid environments would result in positive effects of experimental drought on herbivores, with such plant species adaptively shifting resources away from resistance to maintain performance under stress. In contrast, plants with trait values associated with adaptation to mesic environments would result in negative effects of drought because such species lose vigor and thus decline in their host-quality. We tested these predictions using experimental manipulations in 13 milkweed species (genus Asclepias) adapted to a wide range of environmental conditions, and herbivore performance of herbivores. We exposed plants to species-specific watering regimes physiologically calibrated to maximize (100%) or reduce (50%) stomatal conductance and then monitored the performance of the herbivores. Here we present milkweed trait values associated with aridity gradient [specific leaf area (SLA), relative water content (RWC), water use efficiency (WUE), and maximum stomatal conductance (g_smax)] and changes of trait values associated with host-plant quality to herbivores [nutrients content (protein, nitrogen, and carbohydrates) and plant defensive traits(latex and cardenolides)]. 

Trait measurements were obtained from 3-5 newly expanded and undamaged leaves collected from each experimental plant at the conclusion of the greenhouse experiment. Excised leaves were transferred to a paper envelope and immediately frozen at -20 °C for later characterization of plant traits. For water-use-associated traits, we measured leaves from plants in the control treatment as representative of their constitutive trait values. To measure SLA, leaf area (cm2)/dry mass (mg), frozen leaves were scanned, weighed, dried at 60 °C for 2 days, and reweighed to the nearest mg. Leaf water content was calculated as a percentage estimated from the difference between leaf fresh weight and leaf dry weight. Dried leaves were pulverized using a Mixer Mill (Retsch MM 400) for stable isotope and macronutrient analyses. Foliar carbon isotopes ratio (13C and 12C) was estimated from pulverized foliar tissue as an indicator of WUE for a subset of samples (n=46) at the UC-Irvine Mass Spectrometry facility. Maximum stomatal conductance (gs max) was the raw stomatal conductance values achieved by each species when at its maximum level of stomatal conductance.

To quantify drought-induced changes in hostplant quality, we measured plant defenses and nutritional content from plants in the control and drought treatments. Latex exudation and cardenolides concentration were measured as the typical plant defensive traits in milkweeds. Latex exudation was estimated by excising 2-3 mm off the tip of a new, undamaged, and fully extended leaf. The latex produced within 30 seconds was collected on pre-weighed 1 cm2 filter paper discs which were then placed in pre-weighted 2 mL Eppendorf tubes. The tubes were immediately reweighed to estimate the mass of wet latex collected. Cardenolides concentrations were analyzed from a subsample of plants (n=47) from pulverized foliar tissue. We determined cardenolide concentration (mg/g dry tissue) from pulverized foliar tissue (n=47) by high-performance liquid chromatography following the methods of Züst et al. (2019) in the Ecology and Evolutionary Biology department at Cornell University.

Milkweeds nutritional profiles were characterized from pulverized leaf tissue by quantifying nitrogen, protein, and total non-structural carbohydrate concentrations for a subset of samples (n=46). Nitrogen content (mg/g dry tissue) was estimated during stable isotope analysis at the UC-Irvine Mass Spectrometry facility. Soluble carbohydrates and carbohydrates from starch were extracted separately following protocols described by Chown and Landhässeur (2004) and quantified colourimetrically using the phenol-sulfuric acid assay (DuBois et al., 1956) optimized for microplate reading (Masuko et al., 2005). Protein was extracted by sonication following protocols described in Lenhart et al. (2015) and quantified colourimetrically using the Bio-Rad Bradford microassay (Bradford, 1976) in a microplate reader. Total carbohydrates and protein content were also estimated on a per-mass basis (mg/g dry mass). 

Excel