Chemical analysis

Samples were placed on ice and transported to −80 °C storage the same day as harvest. They were lyophilized for 48 h, and ground using 2 mm beads in a Qiagen TissueLyser II (Qiagen, Hilden, Germany) with bead mill at 30 rpm. The HCl-butanol method (Hagerman and Butler 1980; Porter et al. 1985) was used to analyse both leaf and root samples for condensed tannins. Ground tissue was weighed to within 5 % of 30 mg and the exact weight was recorded. This ground tissue was then extracted twice using 70 % acetone with ascorbic acid. For aspen leaf tissue, initial tests showed that the concentration of condensed tannins in the supernatant was above the detection limit of the spectrophotometer, so the leaf tissue supernatant was diluted to 20 %. Sample extracts were then reacted with the butanol-HCl reagent (5 % 12 M HCl in 95 % butanol) as well as a solution of 2 g ferric ammonium sulfate dissolved in 100 mL of 2 M HCl as recommended by Porter et al. (1985). The ratio of reagents was 5:30:1 (sample supernatant:butanol-HCl:ferric solution). This sample was left to react for 50 min at 95 °C, then absorbance at 550 nm was measured using a BioTek PowerWave XS Spectrophotometer (BioTek Instruments, Inc., Winooski, VT, USA). Condensed tannins were purified from leaves of natural trembling aspen collected on the University of Alberta campus (~150 km from the study site) following the method of Hagerman and Butler (1980) and used as standards. The spectrophotometer readings were converted into milligrams of tannins per gram of tissue using a standard curve, with each sample value adjusted for initial sample weight and any dilution.