Common ragweed (A. artemisiifolia
) is a C3
plant (a plant that uses a 3 carbon compound for CO2
fixation during photosynthesis which should thrive in enriched CO2
atmospheres) common to roadsides and disturbed habitats throughout most of the United States and Canada (Bassett and Crompton 1975
). It is monoecious, with separate male and female flowers borne on the same plant on distinct axillary branches, allowing for independent control of allocation to sexes (Payne 1963
Seeds of A. artemisiifolia collected from wild populations in Woodstock, Illinois, were vernalized by sowing seeds in six growth containers containing compost (Pro-Mix, Red Hill, PA) and storing in a refrigerator at 4°C until ready for germination. Two trays at a time were transferred from cold conditions to the glasshouses at three 15-day intervals, creating three temporal cohorts that would simulate variability in the onset of the growing season and would include anticipated advances of spring several decades into the future. One tray from each cohort was placed in 380 ppm (ambient) and the other at 700 ppm (elevated) CO2 concentration. From each pair of trays, seedlings were chosen that all germinated on the same day; the germination dates (23 May 2002, 7 June 2002, and 22 June 2002) were also at 15-day intervals. The middle cohort approximates the germination date of plants in the Boston area (Rogers C, personal observation).
Approximately 15 days after their germination, we transplanted 24 seedlings from each tray into 6-dry-quart–capacity growth containers (22.23 cm diameter × 21.59 cm deep). Soil in each container was composed of a 4:1 mix of Pro-Mix compost and washed sand (Quickrete Co., Atlanta, GA). The soil mixture was amended with slow-release 14:14:14 nitrogen:phosphorous:potassium fertilizer (Osmocote; Scott’s, Marysville, OH), and plants were watered daily.
The glasshouses consist of six modules structured as three blocks, each block having two modules of differing CO2 concentrations (380 and 700 ppm). Containers were arranged in the modules according to their CO2 and temporal cohort (i.e., eight plants in each of three temporal cohorts, in each of three glasshouse modules, at both low and high CO2, for a total of 144 plants). Day/night temperatures were maintained at 26/21°C. Ambient glasshouse light levels were approximately 70% of full sun, supplemented with 6 hr of light daily (1000 to 1600 hr) from overhead metal halide lamps, thus allowing plants to experience natural variation in day length. Temperature, CO2, and light were computer-controlled for all modules, and we used corn plants (Zea mays) to help maintain a constant CO2 concentration in the low-CO2 modules. In each module, temporal cohorts of ragweed plants were separated, and the positions of the containers within each treatment were randomized at intervals to minimize edge effects. Cohorts were grown at a foliar density of approximately nine plants per square meter. We recorded measurements of flower phenology and date of first pollen release for each ragweed plant throughout the experiment.
We chose five male floral spikes at random from each plant in the first two cohorts and three from each plant in the third cohort at each CO2
level and placed a 5 cm × 25 cm polyethylene bag over each selected spike, similar to the procedure described in Ziska and Caulfield (2000)
. On one side of the bag near the bottom, we cut a small slit and placed the spike inside. The slit was then taped shut and the bag left to collect pollen shed by the spike, with the tops of the bags left open for ventilation. After pollen production had stopped, we measured the length of the bagged flower spikes, cut each at the base, and stored the spike in the collection bag at −20° C until ready for evaluation. Bags in which water accumulated due to watering or heavy condensation were discarded, leaving 477 individual inflorescences.
After senescence, we harvested plants over 3 days from 16 through 18 September. Plant height and number of inflorescences were recorded, the plants were cut at the base, and all flower spikes were removed and placed in bags separate from the vegetative material. We measured the length of each floral spike on each plant. Roots were washed clean of dirt and also placed in separate bags. All plant material was dried at 70°C for 48 hr, and we recorded separate dry weight measurements for all roots, flowers, and vegetative material.
For each bagged flower spike, pollen was recovered by twice repeated 30-sec vortexing in a wash solution (distilled water with 0.02% Tween 20) in 15 mL Falcon tubes, followed by 5-min centrifugation (2,500 rpm; relative centrifugal force = 600). Pollen recoveries from the spike and pollen rinsed from the polyethylene bag were combined in a total volume of 2.0 mL wash solution. We determined the number of pollen grains per spike by calculating the pollen concentration in the wash suspension from microscopic counts using a glass hemacytometer (Hausser Scientific, Horsham, PA).
For each inflorescence, we estimated pollen production pij from an allometric model based on log inflorescence length, time of dormancy release, CO2 concentration, total number of inflorescences, total weight of inflorescences, and days to anthesis:
where μ is a constant and j indexes each inflorescence of log length lj on plant i with number of inflorescences ni, total inflorescence weight wi, and days to anthesis ai, dormancy release at time tr, and grown under CO2 concentration cq. Additional interaction terms did not improve model prediction. We estimated whole-plant pollen production, pi, as the sum of pollen production over all inflorescences on each plant.
We used a two-way factorial design with time of dormancy release crossed with CO2 treatment and CO2 nested within glasshouse wing to assess the responses to the timing of dormancy release and CO2, and we modeled estimated pollen count, inflorescence number, inflorescence weight, aboveground biomass, plant height, days to anthesis, and date of anthesis. Time was included as a fixed term. Glasshouse wing and CO2 within wing were included as random terms to permit broad inference. We included the time × CO2 interaction as a fixed term because plants were individually randomized.