Greenhouse experiment
We performed a common garden experiment in a greenhouse with a subset of 15 populations (Fig. 2, Table S1). The experiment spanned almost the entire geographical and environmental native range, and included three non-native populations to increase the breadth of source environmental conditions (Appendix S2). We pooled all the collected seeds at the population level. We sowed 2,728 seeds (180-200 per population) and obtained 1485 seedlings in individual pots after 25 days. Seedlings were then exposed to treatments with two levels of water supply crossed with three levels of light availability (one block with six treatment combinations). We used 18 seedlings per treatment combination for each population (except for BG, RO and TW populations with, respectively, 14, 10 and 8 seedlings per treatment combination; Table S1). The treatments were chosen to compare their effects with those of two source environmental drivers: Aridity (related to water availability) and Vegetation cover (related to light availability). These treatments also represent parameters likely affected by climate and land-use change. For the water treatments, half of the plants were watered every three days (“wet” treatment), and the other half every nine days (“dry” treatment), by flooding the supporting trays until soil was soaked with water. Each water treatment level was divided into three light levels: 1) 100% light, 2) 64% light and 3) 33% light (Appendix S2). Watering and light levels were designed to span a wide environmental range, characteristic of cosmopolitan plants.
To collect trait data in the greenhouse, we measured plant leaves, flowering status and inflorescences 2.5 months after the onset of treatments in the same way as in field populations. To account for possible maternal effects, usually more manifest in early life stages (Roach & Wulff 1987), control leaf measurements were also taken 1 month after the onset of treatments. At the end of the experiment, the longest healthy leaf was collected from each of 10 individuals per population and treatment combination. Leaves were scanned to estimate leaf area, oven-dried (60ºC), and weighed to calculate SLA. Root:shoot ratio (RSR) was also calculated in the individuals used for SLA measurements, but only for eight populations (Table S1) and excluding the intermediate light treatment due to logistical constraints. To measure RSR, the remaining leaves and the roots were collected, roots were washed, and both leaves and roots were oven-dried.