Although fitness-related traits are expected to be under strong selection, traits related to reproduction are often quite variable within plant populations. We used data from two large greenhouse experiments to quantify phenotypic, genetic, and environmental variation, as well as genetic tradeoffs that might help explain the maintenance of within-population variation in four traits related to sexual or vegetative reproduction in tall goldenrod (Solidago altissima). The goldenrod population exhibited high levels of both phenotypic and genetic variation for capitulum (flower head) number and size, seed production, and rhizome growth. Significant negative genetic correlations were present between the number of capitula and size of capitula—but only at high-nutrient levels—and between seed production and rhizome growth when nutrients were more limiting. In total, negative genetic correlations may act to maintain variation in fitness-related traits in goldenrod populations—a phenomenon we suspect may be shared by other herbaceous plant species as their populations experience variation in environmental factors, such as nutrient levels, among sites or over the course of ecological succession within a site.

Plant source and propagation methods

In the early spring of 2003, I excavated rhizome samples from 26 widely separated Solidago altissima (“tall goldenrod”) genets from a 3-ha old-field population in Union County, Pennsylvania, USA (40°57.9´ N, 76°57.3´ W). The rhizomes were located by the presence of clumps of senesced goldenrod stems (ramets) from the previous year. Different genets were inferred by large gaps between the clumps of stems. I transplanted rhizomes into 27-cm diameter plastic pots in commercial growing medium (ProMix BX^TM; Premier Horticulture Ltd, Dorval, QC, Canada) in a greenhouse at Bucknell University in Lewisburg, Pennsylvania. New sections of rhizomes from these pot-grown genets were stored in refrigeration over winter and repotted in the same fashion each year through 2009. Vegetative progeny (from new rhizome growth) from these genets were used in experiments at Bucknell from 2003-2009. These procedures enabled my colleagues and me to use multiple ramets from each of multiple genets to estimate broad-sense heritabilities in a number of different traits.

Experiment 1 (2005):

In late April of 2005, rhizomes of 26 goldenrod genets were removed from cold storage to begin plant propagation. For each genet, healthy rhizomes were cut into at least 20 equal-sized (2 cm³) segments, as measured by water displacement of 2 mL in a 100 mL-graduated cylinder containing 98 mL of water. Rhizome segments were planted into plastic flats containing ProMix BXTM, and shoots (ramets) began emerging within about one week. In late May, 18 ramets from each genet were transplanted individually from the flats into 16.5-cm diameter plastic azalea pots containing ProMix BXTM. These 468 pots were placed in two randomized blocks on greenhouse benches at Bucknell University (representing east and west sides of the greenhouse), with one ramet per genet per each of nine spittlebug treatments per block. The spittlebug treatments consisted of placing 0, 1, 2, 3, 4, 5, 6, 7, or 8 nymphs on a ramet. Early-instar spittlebug nymphs were transferred to ramets on 3 June and removed upon eclosion to adults, which concluded on 29 June.

Plants were fertilized monthly using Peters Professional 15-16-17 NPK water-soluble fertilizer (J.R. Peters, Allentown, PA). This fertilizer was mixed to a concentration of 3.9 mL per liter of water (i.e., 1 tablespoon per gallon), and 59 mL (1/4 cup) of the mixture was applied to each pot. In elemental terms, this rate works out 34.5 mg of N, 16.0 mg of P, and 32.5 mg of K per application.

Once a ramet finished flowering and setting seeds, its capitula were counted by hand, one inflorescence branch at a time (from 12 October to 14 November). In addition, ten capitula for each ramet were collected across different locations in the inflorescence. These capitula were dissected to count achenes and obtain a mean number of seeds per capitulum for each ramet. This mean was multiplied by the number of capitula to estimate the total number of seeds produced by each ramet. In late winter, each pot was emptied and the underground plant material was cleaned of growing medium. Roots were removed from the rhizomes, and the new rhizomes (initiated and grown in 2005) were collected into paper bags, dried to a constant mass in a drying oven at 60 °C, and weighed to the nearest mg.

Experiment 2 (2006):

In April of 2006, a random subset of 15 of the 26 goldenrod genets from the 2005 experiment were randomly selected for a follow-up study to investigate the effects of nutrient levels on the expression of phenotypic and genetic variation and potential correlations among the same four fitness-related traits as were examined in the previous experiment (viz., number of capitula, seeds per capitula, seed production per ramet, an rhizome mass). Goldenrod ramets were propagated from cold-stored rhizomes using the same procedures as detailed for Experiment 1. For the current study, twelve ramets per genet were grown in 16.5 cm plastic azalea pots in ProMix BXTM. These plants were assigned randomly to a two-way factorial design with nutrient level (n = 2) crossed by spittlebug treatment (n = 2). Each genet had three ramets per treatment combination, for a total of 180 ramets in the experiment.

The high-nutrient treatment received weekly fertilization at the dosage described above for the plants in Experiment 1. The low-nutrient ramets received no fertilizer for the duration of the experiment. On 2 June, 10 field-collected, early-instar spittlebug nymphs were placed on half of the ramets. Each spittlebug was removed upon adult eclosion, which took as short as one week to as long as nearly three weeks of feeding. Measurements of capitula number, capitulum size, seed number, and rhizome mass were taken with the same procedures as described above for Experiment 1.