Defense with benefits? Ducking plants outperformed erect plants in the goldenrod Solidago gigantea in the absence of herbivory

Study system

Along with several other goldenrods, Solidago gigantea Aiton (“giant goldenrod”) is an herbaceous perennial that often dominates old fields, roadsides, and other disturbed areas (Abrahamson et al., 2005). This species is native to eastern North America, but it has become an important invasive weed in much of Europe and eastern Asia (Weber, 2001; Schlaepfer et al., 2008; Hull-Sanders et al., 2009; Szymura and Szymura, 2016; Uesugi and Kessler, 2016). Solidago gigantea flowers in late summer-to-early fall, with branching inflorescences (panicles) bearing numerous flower heads (capitula) that contain female ray florets and perfect (cosexual) disk florets (Abrahamson and Weis, 1997). The florets are pollinated by a wide diversity of insects, and each floret produces a single-seeded fruit called an achene. In addition to prodigious sexual reproduction, S. gigantea spreads vegetatively through perennial rhizomes.

Source of plants

On 17–18 March of 2015, seeds were collected from 56 senesced ramets (stems) of Solidago gigantea that had flowered the previous summer in Green Hill Park (a 79.6-ha public park in Roanoke County, Virginia, USA; 37.27°N, 80.11°W). These ramets were from widely separated, discrete clumps of goldenrod stems, which ensured that each ramet was from a different genet (i.e., genetic individual). Achenes that were still attached to a ramet were collected into a zip-seal plastic bag. All of the seeds in one bag thus shared the same maternal parent, although they likely originated from numerous different sires. Thus, the seeds in each bag represented a family of half-sibs and full-sibs.

On 21 April 2015, seeds from the source plants were planted in 5-cm (2-in) square plastic pots in Miracle-Gro potting mix (The Scotts Company LLC, Marysville, Ohio, USA). The contents of the zip-seal bags were first shaken to dislodge achenes from their capitula. Then seeds from a single bag were sprinkled onto the surface of the potting mix in a single pot. The pots were placed in plastic trays on wooden pallets in an outdoor plot that was separated from natural goldenrod populations. The plants remained in this outdoor plot for the duration of the study.

When seedlings for a family reached the one-to-two true-leaf stage (beginning 13 May), they were transplanted singly into 5-cm square pots in Miracle-Gro potting mix. Between 12 and 21 (inclusive) seedlings were transplanted for each maternal family (mean of 20.5 seedlings per family). Then from 9–14 June, the seedlings from each pot were transplanted singly into 15-cm (6-in) round plastic pots in Miracle-Gro Moisture-Control potting mix. A total of 739 seedlings from 36 maternal families of S. gigantea were transplanted for this study.

Experimental design and data collection

The plants were assigned randomized positions in the outdoor plot, as diagrammed in Fig. 1. The plot consisted of 12 rows, each with 12 plastic trays (with drain holes), on wooden pallets. Six pots were positioned in each plastic tray. The row number, tray number, and pot-position number were used as covariates in the analyses. The rows represented a north-to-south gradient, the trays represented a west-to-east gradient, and the pot positions represented exterior-to-interior sites within pairs of rows (Fig. 1). For example, positions 1 and 4 were on the exterior, positions 3 and 6 were on the interior, and positions 2 and 5 were intermediate for each pair of rows. The 739 pots of S. gigantea seedlings completely filled the first 10 rows plus the first three trays and one position of the 11th row. The remaining positions in the 11th and 12th rows were occupied by pots of S. altissima seedlings, which were part of a separate study not addressed in this paper.

The plants were watered as needed (almost daily) throughout the growing season. Plants were surveyed roughly every 10 days from 22 June through 3 August to note which ramets were ducking and which were erect. It was generally obvious whether a ramet was ducking or not, but cloudy days tended to straighten stems slightly, and windy days tended to bend stems slightly. Thus, for a ramet to be considered ducking, the apical-leaf bud of the stem had to be pointed below the horizontal (i.e., bent >90°) on at least two of the surveys. In addition, plants were checked daily during the flowering period (starting 16 August) to note the first day of anthesis for each plant.

Despite the separation of the study site from natural goldenrod populations, one species of goldenrod herbivore did show up in appreciable numbers. Specifically, on the surveys of the plants in late July and early August, spittle masses produced by nymphs of the sunflower spittlebug (Clastoptera xanthocephala) were observed on some plants. The presence of these nymphs was recorded to determine whether ducking and erect morphs differed in susceptibility to this herbivore. These spittlebugs, as with other herbivores, were removed to minimize their damage to the plants.

Between 24 September and 13 October, samples of five to seven flowering capitula per ramet were collected from a subset of plants and taken into the lab for dissection to count the number of ray (female) and disk (perfect) florets. In total, floral-sex-ratio data were recorded for 116 ducking and 29 erect ramets across 33 maternal families.

The proportion of ducking ramets differed greatly among maternal families, with four families producing exclusively ducking and one family producing exclusively erect stems. In total, 72% of the progeny had ducking stems, and 28% had erect stems. In mid-October, 18 maternal families that produced at least four ramets of each stem morph were chosen for further data collection on growth and seed production. In total, these data were taken on 99 ducking ramets and 97 erect-stemmed ramets.

Once flowering was finished on a ramet, and before seeds could be dispersed (19 October through 19 November), all of the capitula on these 196 ramets were counted. A sample of seven intact capitula was taken from across the inflorescence of each ramet. The seven capitula were individually dissected to count the number of seeds per capitulum (i.e., “capitulum size”) and obtain a mean value for capitulum size for each ramet. Total seed production per ramet was then estimated by multiplying the number of capitula a ramet produced by the mean number of seeds its capitula contained. In addition, the final height of each ramet was measured from the base of the stem to the top of the stem apex, and the diameter of each ramet's stem at 10 cm above the soil surface was measured using dial calipers.

Following shoot senescence, the rhizomes for these 196 ramets were removed from their pots, cleaned with water, dried in an oven at 60°C for at least four days (at which point they had attained a constant mass), and weighed to the nearest 0.1 g.

Statistical analysis

All statistical analyses in this paper were performed using JMP-IN 4.0.4 (SAS Institute, Cary, North Carolina, USA).

Stem and rhizome growth

There was no evidence that ducking ramets paid any cost in terms of reduced growth. In fact, at maturity, the stems of ramets that had ducked were 4% taller than erect morphs, and 5% thicker (Fig. 2A, B; Table 1). Rhizome growth (as measured by dry mass) did not differ between ducking and erect-stemmed morphs (Fig. 2C).

Sexual reproduction

Among ramets, dates of first flowering ranged from 16 August through 7 November. Despite this 83-day among-ramet range, the plant-family means in day of first flowering ranged across only 23 days (31 August to 23 September). Still, plant family was the only significant predictor of first-flowering date, accounting for 38% of the variation (Table 2). In total, the 426 ducking ramets took an average of ~1.6 days longer to begin flowering than the 153 erect ramets, but this difference was not statistically significant (Table 2).

The sex ratio of florets within capitula did not differ among plant families or between stem morphs. Specifically, the capitula of ducking ramets contained an average of 56% ray florets, while the capitula of erect ramets contained an average of 55% ray florets (Table 2).

Ducking ramets produced 20% more seeds on average than erect ramets (Table 2; Fig. 2F). This difference was entirely due to the fact that ducking ramets produced 20% more capitula (Fig. 2D), because the number of seeds per capitulum did not differ between stem morphs (17.5 for ducking, and 17.8 for erect ramets; Table 2, Fig. 2E).

Susceptibility to herbivores

Because the experimental plot was isolated from any natural goldenrod populations, herbivore damage was very minor. Nevertheless, during the checks of the plants, I found a few red aphids, a few adult Trirhabda beetles, and occasional grasshoppers and lepidopteran caterpillars. These herbivores were removed immediately and caused negligible damage. A white-tailed deer visited the experiment one evening and damaged seven ramets severely enough that they had to be removed from the experiment.

The only herbivore that was found with regularity was the sunflower spittlebug (Clastoptera xanthocephala). Specifically, nymphs were found in spittle masses on 32 ramets. Erect ramets were significantly more susceptible to colonization by these spittlebugs (10.1%) than were ducking ramets (3.4%) (likelihood-ratio χ² = 11.609, P = 0.0007).

Costs and benefits of resistance-by-ducking

Previous studies have demonstrated that resistance-by-ducking in the goldenrod Solidago altissima is an effective defense mechanism against apex-galling insects—reducing the likelihood of galling by about half (Wise and Abrahamson, 2008; Wise, 2009; Wise et al., 2010a). Nevertheless, nonducking (erect) morphs persist in natural populations—a pattern suggesting that ducking must entail some sort of cost that counteracts its benefit. Despite this expectation of a cost to ducking, neither previous studies (Wise, 2009; Wise et al., 2009, 2010b) nor the current study was able to find compelling evidence of costs to ducking. In fact, the current study on S. gigantea found that ducking seemed to have only positive side effects.

By the end of the growing season, ducking ramets (i.e., stems that had nodded then straightened out again prior to flowering) were larger and had greater reproductive output than erect ramets. Specifically, stems of ducking ramets ended up being 4–5% taller and thicker on average than stems of erect ramets. Ducking ramets also produced more capitula, resulting in 20% greater seed production than erect ramets. Ducking and erect ramets had the same proportion of disk (cosexual) and ray (female) florets; therefore, there was no evidence of a cost to ducking through loss of fitness through potential pollen production. Finally, biomass allocation to rhizomes did not differ between ducking and erect ramets. One might expect that rhizomes, as underground stems, would show the same size differences as did the aboveground stems. However, because each plant tended to initiate numerous rhizomes, each of which had multiple bends and branches, it was not possible to obtain a precise measurement of total length of rhizomes. Nevertheless, the fact that there was no difference in rhizome biomass suggests that ducking ramets did not pay a cost in terms of the potential for future growth and reproduction.

Although the current study was not designed to look at resistance benefits of ducking in S. gigantea, it is likely that ducking confers the same benefits as it does in S. altissima. Both species are attacked by apex-galling dipterans during the time when ducking stems are nodding (Dorchin et al., 2009; Dorchin et al., 2015). Although the fitness effects of galling has not been studied in S. gigantea, the same types of galls have consistently been found to reduce growth and reproduction in S. altissima (Hartnett and Abrahamson, 1979; Abrahamson and Weis, 1997; Wise et al., 2006).

In addition, erect ramets of S. gigantea in the current study were nearly three times as likely as ducking ramets to suffer attack from nymphs of the sunflower spittlebug (C. xanthocephala). In a previous study, attack by nymphs of the meadow spittlebug (Philaenus spumarius) on S. altissima was unrelated to stem morph (Wise, 2009). However, meadow spittlebugs are univoltine, and their nymphs hatch from overwintering eggs in the soil or leaf litter during spring and must crawl on the ground to find host plants (Wise et al., 2008). Thus, meadow spittlebugs must choose their host plants without sensing whether stems were ducking. In contrast, females of the sunflower spittlebug (at least of the summer generation) are able to oviposit directly onto herbaceous host plants of the nymphs (Hamilton, 1982). Oviposition in this experiment occurred during the time period when many stems were ducking. The fact that ducking stems were significantly less likely to harbor spittlebug nymphs strongly suggests that adult females preferentially oviposited on erect stems, and thus that ducking conferred a degree of resistance against sunflower spittlebugs. Notably, this is the first documentation of ducking conferring resistance against a nongalling insect.

Because the plants in this experiment were relatively isolated from natural sources of spittlebugs, the overall attack rate was light. Only 10% of the erect ramets were attacked by spittlebug nymphs, and there was no instance of more than two nymphs on a single ramet. Furthermore, although oviposition occurred in midsummer, feeding by the spittlebug nymphs occurred in late summer and autumn, after the plants were nearly fully grown. Finally, I removed the nymphs as soon as I encountered them on the plants. Therefore, it is highly unlikely that these spittlebugs (or any insect herbivores) had any measurable effect on growth or reproduction of plants in this experiment.

Frequency of stem morphs

Across the 36 maternal families included in this study, ducking stems were nearly three times more common than erect stems (72% vs. 28%, respectively). Because these families were sampled throughout a large (~80 ha) population, this ratio of nearly 3:1 likely reflects the ratio of stem morphs within the source population. In addition, the high proportion of ducking ramets is consistent with numerous personal observations of natural S. gigantea populations. That there was a preponderance of ducking morphs may be surprising because it is the reverse of what has consistently been reported for S. altissima (Wise and Abrahamson, 2008; Wise, 2009; Wise et al., 2010b). What factors might be responsible for this difference in stem-morph frequency between the two similar plant species is not obvious and is worthy of future investigation.

Both of these goldenrod species are attacked by apex-galling insects. However, it is not known whether the intensity of attack differs between the plant species. Thus, it is difficult to speculate whether selection for ducking as a resistance trait differs between the species—or whether it differed in the evolutionary history of the species. Comparative studies of the herbivore-community composition of the two goldenrod species across a wide geographic range, combined with data on the frequency of ducking stems, would be valuable in elucidating ecological factors that influence the evolutionary dynamics of ducking as a resistance trait. It is possible that there is a cost to ducking that is expressed in S. altissima, but is not expressed in S. gigantea. Therefore, it will be worthwhile to perform a study similar to the current one to examine internal costs of ducking in S. altissima.