Volume 101, Issue 12 p. 2079-2087
Ecology
Free Access

Effects of apical meristem mining on plant fitness, architecture, and flowering phenology in Cirsium altissimum (Asteraceae)

Subodh Adhikari

Subodh Adhikari

Department of Land Resources and Environmental Sciences, Montana State University, PO Box 173120, Bozeman, Montana, 59717-3120 USA

Search for more papers by this author
F. Leland Russell

Corresponding Author

F. Leland Russell

Department of Biological Sciences, Wichita State University, 1845 Fairmount, Wichita, Kansas, 67260-0026 USA

Author for correspondence (e-mail: [email protected])Search for more papers by this author
First published: 01 December 2014
Citations: 10

The authors thank J. Beck, J. Dunlap, J. Hadle, G. Houseman, M. Jameson, O. Keller, M. Martino, R. McMinn, W. Parcell, and D. Wickell for insightful comments on early drafts of the manuscript. The authors thank A. Adhikari and R. McMinn for help with field and laboratory work. This is contribution #23 of the WSU Biological Field Station.

Abstract

Premise of the study: Interactions that limit lifetime seed production have the potential to limit plant population sizes and drive adaptation through natural selection. Effects of insect herbivory to apical meristems (apical meristem mining) on lifetime seed production rarely have been quantified experimentally. We studied Cirsium altissimum (tall thistle), whose meristems are mined by Platyptilia carduidactyla (artichoke plume moth), to determine how apical damage affects plant maternal fitness and evaluate both direct and indirect mechanisms underlying these effects.

Methods: In restored prairie, apical mining was manipulated on tall thistles by applying insecticide, water, or no spray to apical meristems. We quantified effects on lifetime seed production, plant architecture, and flowering phenology. Seed germinability and seedling mass were evaluated in a greenhouse.

Key results: Apical meristem miners decreased lifetime seed production of C. altissimum, but not seed quality. Higher mortality rates of damaged plants contributed to reduced seed production. Apical damage reduced plant height and increased the proportion of blooming flower heads in axial positions on branches. Apical damage delayed flowering and shortened flowering duration.

Conclusions: Apical meristem mining reduced plant maternal fitness. The shift in the identity of blooming flower heads from terminal to axial positions contributed to this reduction because axial heads are less fecund. Shorter, meristem-mined plants may have been more susceptible to competition, and this susceptibility may explain their higher mortality rates. The kinds of changes in architecture and phenology that resulted from apical damage to C. altissimum have been shown to affect floral visitation in other plant species.

Understanding the magnitude of herbivores’ effects on plant lifetime seed production is central to the ecological and evolutionary study of plant–herbivore interactions. For example, evidence is increasing that the availability of seeds, as opposed to the availability of safe microsites for establishment, can limit plant population sizes (Turnbull et al., 2000; Ehrlen et al., 2006), especially for monocarpic plants (Louda and Potvin, 1995). Therefore, herbivores that limit seed production can restrict the abundance and influence the habitat distribution of plants they consume. Knowledge of which herbivores limit lifetime seed production is of applied significance for developing strategies to limit weed populations. Crawley (1989) hypothesized that insect herbivores generally have little effect on plant lifetime seed production because individual insects remove little tissue and plants often have strong ability to regrow lost tissue. However, examples from a diverse set of insect herbivore guilds, including folivores (Rausher and Feeny, 1980), root-feeders (Maron, 2001), and florivores (Louda and Potvin, 1995), demonstrate that insect herbivory can reduce lifetime seed production.

Effects of apical meristem herbivory on plant fitness and mechanisms by which plants compensate for apical damage have interested ecologists for a long time. Strauss and Agrawal (1999) defined a damaged plant's compensatory response as the degree to which the plant is able to match the fitness of an undamaged plant. Compensatory responses vary from lower fitness in damaged plants (undercompensation) to equal fitness (full compensation) to greater fitness (overcompensation) as compared with undamaged plants. For monocarpic plants, examples of apical damage resulting in undercompensation (Boalt and Lehtila, 2007; Brody and Irwin, 2012), full compensation (Hendrix, 1979; Naber and Aarssen, 1998; Rautio et al., 2005; Hamback et al., 2011), and overcompensation in lifetime seed production (Maschinski and Whitham, 1989; Huhta et al., 2000) have all been documented. Within a single plant species, the degree of reproductive compensation achieved by apically damaged plants is affected by resource availability (Maschinski and Whitham, 1989; Wise and Abrahamson, 2008), seasonal timing of herbivory (Maschinski and Whitham, 1989; Lennartsson et al., 1998), and amount of tissue removed with the apical meristem (Huhta et al., 2000). Recent research also suggests that indirect interactions involving pollinators or herbivores that visit plants after the apical meristem is damaged can affect the plant's compensatory response to apical damage (Hamback et al., 2011; Brody and Irwin, 2012). By contrast, differences in plant compensatory responses when apical damage is imposed by mammals vs. by insects are poorly understood because the effects of insect apical meristem herbivory on plant maternal fitness have been quantified for few plant species (e.g., Hendrix, 1979; Naber and Aarssen, 1998; Hamback et al., 2011). Nevertheless, apical meristem herbivory by insects can be common in natural plant populations. In the three studies that we found that quantified the frequency of insect damage to apical meristems, all reported >30% of ramets in populations were attacked (Hendrix, 1979; Anderson et al., 1989; Hamback et al., 2011).

Predictions from studies of plant compensatory responses to mammalian herbivory are not readily transferred to the effects of apical damage by insects. General differences between mammalian and insect herbivory that can affect plant compensation include the specificity and duration of tissue removal (Kotanen and Rosenthal, 2000). These differences clearly apply to apical meristem damage. Whereas insect damage to the apical meristem may be confined to the stem apex, mammals often remove much of the stem, including lateral meristems that might have been released from apical dominance (Huhta et al., 2000). In Gentianella campestris, clipping little tissue, as might resemble insect herbivory, resulted in overcompensation by damaged plants whereas severe clipping, which removed 75% of the stem, resulted in merely full compensation (Huhta et al., 2000). Further, a mammal damages apical tissue over a short interval. One bite can remove the apical meristem. By contrast, insect herbivory removes tissue gradually, especially when larvae developing inside the apical meristem impose the damage (Fay et al., 1996). The rarity of studies that quantify fitness effects of insect apical meristem herbivory limits our understanding of the consequences of these differences for plant maternal fitness.

Host plant architectural (Lortie and Aarssen, 2000; Rautio et al., 2005) and phenological (Huhta et al., 2000; Freeman et al., 2003) changes are hypothesized to both directly and indirectly mediate the degree of reproductive compensation for apical meristem herbivory. Variation in the extent to which apical damage releases lateral meristems, thereby increasing the number of branches, flowers and seeds produced, is the most common direct mechanism cited for explaining variation in compensatory seed production (Naber and Aarssen, 1998; Rautio et al., 2005). Compensation through altered branching patterns, however, can reduce seed quality because seed produced by more basal or higher-order branches can be smaller and germinate at a lower rate (Benner, 1988; Hendrix and Trapp, 1989). Apical damage also may delay flowering phenology (Freeman et al., 2003; Huhta et al., 2009). For species that flower late in the growing season, delayed flowering could produce incomplete compensation by damaged plants when at the end of the growing season all flowering ceases (Huhta et al., 2000; Freeman et al., 2003).

Indirect effects of apical damage on plant fitness may arise through altered interactions with other antagonists and mutualists. Changes in plant height or the structure of the flowering display may affect plant apparency to floral visitors. For example, apical inflorescences produced by plants with an intact apical meristem may be more conspicuous than the more basal inflorescences produced by damaged plants (West, 2012). Phenological changes may mediate indirect effects on host plant fitness by affecting synchrony between peak activity of mutualist or antagonist floral visitors and the stages of inflorescence development that they use (Brody and Irwin, 2012). Further, changes in flowering duration affect the length of opportunity for pollen transfer. Therefore, fitness effects of apical damage may arise through multiple direct and indirect mechanisms.

Here we quantify the effects of apical meristem herbivory by larvae of the artichoke plume moth (Platyptilia carduidactyla) on maternal fitness in a monocarpic, perennial plant, tall thistle [Cirsium altissimum (L.) Spreng., Asteraceae]. We refer to this form of apical meristem herbivory, which involves developing larvae consuming tissue from within the apical meristem, as apical meristem mining. We also evaluate changes in tall thistle reproductive architecture and phenology to explore direct and indirect mechanisms that may cause these fitness effects. The C. altissimumP. carduidactyla interaction is of interest because apical damage by this herbivore affects a majority of C. altissimum plants in our region (F. L. Russell, personal observation). Further, populations of this monocarpic plant are often seed-limited (Russell et al., 2010), so reduced lifetime seed production could affect population size. Our research questions were: (1) Does apical meristem mining by insects reduce, increase, or leave unchanged lifetime seed production, seed quality, and seedling vigor of C. altissimum? (2) In what ways does apical meristem mining by insects alter the architecture and flowering phenology of C. altissimum?

MATERIALS AND METHODS

Study species

Cirsium altissimum (L.) Spreng. (tall thistle) is a short-lived, perennial, monocarpic plant that occurs in grasslands, pastures, and roadsides. Reproductive stalks of adults are 0.5–2.5 m tall and determinate flowering. Stems are usually branched and include both primary branches off the main stem and higher-order branches (secondary or tertiary) off the primary branches (Fig. 1). Spiny, purple flower heads can occur at the apex of the main stem (apical flower head), the apices of primary branches (terminal flower heads), and the apices of higher-order branches (axial flower heads) (Fig. 1). Cirsium altissimum is native throughout the eastern United States as far west as central Kansas. It is the most common thistle in tall grass prairies of the eastern Great Plains (Great Plains Flora Association, 1986). In southern Kansas, adult plants begin producing a reproductive stalk (“bolting”) in May, flower August–October, and disperse seeds September–November (S Adhikari, personal observation).

Details are in the caption following the image

Schematic representation of a tall thistle plant (Cirsium altissimum) with an undamaged apical meristem. Terminology for flower head positions and branch orders is illustrated. In the text, secondary and tertiary branches are collectively referred to as higher-order branches.

Many insects attack C. altissimum. Folivores include the native weevil Baris subsimilis, the exotic weevil Trichosirocalus horridus (Takahashi et al., 2009), larvae of the painted lady butterfly Vanessa cardui, a flea beetle Systena hudsonias (Russell and Spencer, 2010), grasshoppers, and several microlepidopterans (Guretzky and Louda, 1997). Insect herbivores on reproductive tissues include the meristem-mining moth Platyptilia carduidactyla and two species whose larvae attack developing flower heads: the tephritid fly Paracantha culta and the pyralid moth Homoeosoma eremophasma (Takahashi et al., 2009). As the only apical meristem miner of C. altissimum at our study site, P. carduidactyla is the focal herbivore of our study.

Study site

Study sites for this research were located at Wichita State University's Ninnescah Reserve, near Viola, Kansas USA. The Ninnescah Reserve is restored tall grass prairie that was converted from agriculture 35 yr before this study. This restored prairie is dominated by warm season grasses, including Andropogon gerardii (big bluestem), Schizachyrium scoparium (little bluestem), and Panicum virgatum (switch grass), and forbs, including Helianthus maximiliani (maximilian sunflower) and C. altissimum (tall thistle). The prairie is moderately invaded by shrubs, especially Symphoricarpos orbiculatus (coral berry) (S. Adhikari, personal observation). The Ninnescah Reserve has not been grazed for >35 yr.

Two sites within the Ninnescah Reserve (hereafter, sites A and B) were chosen because they had a sufficient number of tall thistles. Site A (37°32′00″N, 97°40′40″W) had not been mowed for several years before our study and had extensive shrub cover. Site B (37°32′19″N, 97°40′49″W) had been mowed regularly and had little woody vegetation. Soils at both sites were dry sandy loam with site A richer in humus than B (S. Adhikari, personal observation). Both sites are within the 100 yr floodplain of the Ninnescah River.

Experimental design

To choose naturally occurring adult tall thistles for our experiment, we established transects in both sites (112.5 m long in site A and 107.5 m in site B) along the longest axis through the tall thistle population in April 2012. At 2.5-m intervals along each transect, the two nearest adult tall thistles with undamaged apical meristems were chosen with the constraint that experimental plants had to be >2 m apart.

We randomly assigned levels of the apical meristem herbivory treatment to plants by tossing a coin. Once half of the plants at a site were assigned to either the “insecticide” or “control” treatment level, the remainder was assigned to the other treatment level. Insecticide plants were protected from apical damage and control plants were exposed to apical meristem mining by P. carduidactyla. Control plants were then divided into water controls and unmanipulated controls. This procedure resulted in 46 insecticide, 23 water control, and 23 unmanipulated control plants in site A and 44 insecticide, 22 water, and 22 unmanipulated control plants in site B.

Bifen I/T (Control Solutions Pasadena, Texas, USA), a nonsystemic synthetic pyrethroid insecticide, was mixed in a 1:15 ratio with water and applied only on the apical meristems of the 90 plants assigned to the insecticide treatment level using a hand-held sprayer. We sprayed insecticide every 2 wk between the last week of April and the first week of July when the apical flower head was visible. For the 45 water control plants, a different sprayer of the same type was used to apply water to apical meristems between late April and early July. We included this water control treatment level to evaluate potential confounding effects of the water used to dilute the insecticide. The remaining 45 plants were not sprayed. In 19 control plants (∼21%), there was no apical meristem mining by the time an apical flower head became apparent. The apical flower heads of these 19 plants were clipped to simulate meristem mining. Four insecticide plants (4.5%) suffered apical damage despite insecticide application. These four plants were excluded from analyses that addressed effects of apical meristem mining on plant performance, architecture, and phenology. Initial sizes of plants were measured when treatment levels were assigned. Root crown diameter was measured just below the shoot–root junction. Height was measured from the soil surface to the apical meristem.

Herbivore damage

Whether plants suffered apical meristem mining or not was determined from visual evidence of the damage, specifically dead leaves and frass around the apex. Because our study addressed effects of apical meristem damage, it was important that our insecticide application not influence insect herbivore guilds other than apical meristem miners. Number of leaves damaged on each experimental plant and percentage of area damaged on each damaged leaf were quantified on 15 June 2012 to evaluate effects of insecticide application on folivores. To evaluate effects of insecticide application on flower head-feeding insects, we recorded the proportion of flower heads damaged per plant during flower head dissections.

Mortality rates and seed production

To quantify the effect of apical meristem mining on seed production by C. altissimum, we randomly chose 120 (of 180) experimental plants. The other 60 plants were used to describe the community of floral visiting insects on tall thistle, and these results are reported elsewhere (Adhikari, 2013). Organza fabric bags were placed over postanthesis flower heads to catch seeds. Bagged heads were collected after seeds dispersed into the bag. Through this procedure, we collected 96.1% of flower heads that bloomed on experimental plants. We dissected the collected flower heads to determine the number and mass of viable seeds produced by each flower head. A seed was considered viable if its sides were convex and it was undamaged. Our criteria for visually assessing seed viability appear generally to have been accurate. Of 3810 randomly chosen seeds that were sown in the greenhouse (see next section), 86% germinated within 5 wk. For a monocarpic plant, total viable seed produced during its one reproductive episode represents an estimate of maternal fitness.

Greenhouse experiment for germination rate and seedling biomass

A germination experiment was run in the greenhouse to compare germination rate and seedling vigor among plants in different levels of the apical herbivory treatment. Only seeds from flower heads that produced ≥30 viable seeds were chosen for the greenhouse study. The apical flower head (Fig. 1) and the terminal flower head on the most basal primary branch were chosen from each plant. If there was no apical flower head, then the terminal flower head on the most apical primary branch was used instead. Flower heads were collected between 15 September and 28 October 2012. Of 321 flower heads (from 86 plants) dissected, 127 flower heads (94 from insecticide plants, 13 from water control plants, and 20 from unmanipulated control plants) fit the criteria for sowing seeds to assess seed and seedling quality. These flower heads were from 51 plants (36 insecticide, 6 water, and 9 unmanipulated controls).

Soil was collected from Wichita State University's Gerber Reserve (37°40′48″N, 97°57′0″W). We avoided collecting soil in areas where tall thistles had been observed in previous years to prevent including seeds from thistles that had not received the experimental treatments. Soils from different collection locations and different depths were mixed together to make a homogenized soil in all pots. Labeled, square, 1.78 L pots were filled with soil. Thirty seeds from a single flower head were sown in each pot on 25 December 2012. Because >30 seedlings never germinated in a pot, the field-collected soil apparently did not contain tall thistle seeds. Pots were watered regularly as needed, and ambient light was augmented for 12 h a day with a 1000 W halogen bulb. To avoid confounding effects of environmental variation in the greenhouse, we rotated all pots clockwise weekly for 6 wk.

Proportion of seeds germinating was quantified for each pot (flower head). Seedlings were recorded as germinated once cotyledons were visible. The first four seedlings to germinate in each pot, which we labeled with colored pins, were kept to quantify seedling biomass, and all other seedlings were pulled. Germination rate was calculated as the proportion of seeds that germinated among the 30 seeds sown. Five weeks after sowing or 4 wk after the mean germination date for the first four seedlings, all four seedlings in a pot were harvested to determine the total seedling biomass for each pot and so for each flower head. All seedlings in a pot were harvested simultaneously to allow growth for the same time period (average 4 wk per pot) from their germination date. Because of the different timing of germination in different pots, it took 11 d (1–10 February 2013) to harvest seedlings of all pots. Plants were dried at 60°C for 3 d and weighed to quantify the total dry biomass.

Plant architecture and flowering phenology

Architectural and phenological measurements were made on all 180 experimental plants. Measurements to quantify the effect of apical meristem mining on plant architecture were made when the most apical flower head on the plant reached full bloom. Architecture variables included final height, number of primary branches, number of flower heads initiated, and number of flower heads blooming. Final height for each plant was measured as distance from the ground to the top of the highest flower head on the plant. For each plant, we counted number of primary branches, number of flower heads initiated, and number of flower heads blooming.

To quantify effects of apical meristem mining on flowering phenology, we censused flower heads on all plants each week from early July through the end of flowering in fall 2012. At each census, we recorded the number of flower heads present on each plant and each flower head's developmental stage. Flower head developmental stages were: 1 = small bud, 2 = large bud, 3 = early flowering (flowers open but stigmas not completely exserted), 4 = late flowering (flowers open and stigmas completely exserted), 5 = mature (flower wilting), and 6 = dispersing (Adhikari and Adhikari, 2010). Diameter of each flower head was recorded when it was in full bloom (i.e., stages 3 or 4).

Statistical analyses

We first attempted parametric analyses, but nonparametric tests are reported when the ANCOVA assumptions of normality of residuals and homoscedasticity were not met by either untransformed or transformed data. In each analysis, water and unmanipulated controls were compared. If they did not differ significantly, then they were pooled to compare the controls with insecticide plants. If they did differ, then we performed pairwise comparisons among the levels of the herbivory treatment with Bonferroni adjustments to significance thresholds. In ANCOVAs, site was a random effect. For nonparametric analyses, we tested for an effect of site, and if there was no significant difference between sites then observations from the two sites were pooled. Otherwise, sites were analyzed separately with a Bonferroni adjustment to the significance threshold. Statistical analyses were conducted using SAS version 9.1 (SAS Institute, Cary, North Carolina, USA).

Herbivore damage

To assess whether insecticide prevented apical meristem mining, the frequency of apical meristem damage in experimental plants was analyzed with a χ2 test of a 2 × 2 contingency table. There were two damage categories: apical meristem damaged vs. undamaged. To determine whether the insecticide application affected damage by flower-head-feeding insects, the proportion of flower heads damaged per plant was analyzed with a χ2 analysis of a 2 × 3 contingency table. The three categories for flower head damage were: plants with no damaged flower heads, plants with <50% of flower heads damaged, and plants with ≥50% of flower heads damaged. To evaluate effects of insecticide application on leaf-feeding insects, the proportion of leaves damaged per plant and proportion of leaf area damaged per damaged leaf were analyzed with ANCOVAs. Total leaves per plant was the covariate. Proportion of leaves damaged per plant was arcsine square-root transformed and proportion of leaf area damaged per damaged leaf was natural log-transformed to meet ANCOVA assumptions.

Mortality rates and seed production

Treatment effects on mortality rates of experimental plants were analyzed using a χ2 analysis of a 2 × 2 contingency table. Treatment effects on total number of viable seeds produced per plant and total mass of seeds produced per plant were analyzed with Kruskal–Wallis tests. These analyses were performed for the data set that included all experimental plants, including those that died before flowering and for a smaller data set that included only plants that survived to flower. Treatment effects on number of viable seeds per flower head that bloomed were analyzed with a Kruskal–Wallis test.

We used ANCOVA, with initial root crown diameter as the covariate, to compare mean individual seed mass among levels of the herbivory treatment. Estimates of individual seed mass were based upon all viable seeds produced by an experimental plant.

Greenhouse experiment for seed viability and seedling biomass

We evaluated treatment effects on germination rate of viable seeds and seedling mass using Kruskal–Wallis tests. “Plant” was the unit of observation for analyses of germination rates and seedling masses because we analyzed the mean of the two heads from each plant whose seeds were sown in the greenhouse.

Effects of flower head position within a plant

We examined effects of axial vs. terminal flower head position within a plant on flower head diameter, number of viable seeds per flower head, and seed mass. We identified a subset of plants that had at least one branch where both the terminal flower head and one or more axial flower head bloomed. If a branch had more than one axial flower head that bloomed, then we analyzed the mean of flower head diameter or number of viable seeds or seed mass across all axial flower heads on that branch. If a plant contained more than one branch that had blooming terminal and axial flower heads, then we analyzed the mean of flower head diameter or number of viable seeds or seed mass across all axial heads on those branches and the mean across all terminal heads on those branches. We used paired t tests or nonparametric sign tests to compare flower head diameter, viable seed production, and individual seed mass between axial and terminal flower heads within a plant. For effects of flower head position, we pooled across levels of the herbivory treatment and sites because plants that had branches with blooming terminal and axial flower heads were rare in some treatment levels and sites.

Architecture and flowering phenology

Plant architecture variables were analyzed with Kruskal–Wallis tests. These variables included initial root crown diameter, initial and final plant heights, number of primary branches, number of total flower heads, number of flower heads that bloomed, and proportion of blooming flower heads that were axial vs. terminal.

Flowering phenology variables were analyzed with χ2 analyses of three-way contingency tables. These variables were treated as categorical because with weekly censuses of flowering condition and a flowering period for the entire experimental population of 10 wk, they had a narrow range of values. Phenology variables included initial date of flowering (date on which a plant first had a blooming flower head), latest date of flowering (date on which a plant last had a flower head with purple florets), peak flowering date (date on which a plant had the largest number of flower heads blooming), flowering duration, and size of maximum floral display (number of flower heads blooming simultaneously). For peak flowering date, if the plant had the same number of flower heads blooming on several census dates, then the earliest date with the peak number of blooming heads was used.

RESULTS

Initial root crown diameter (χ2 = 35.6, df = 1, p < 0.0001) and height (χ2 = 22.83, df = 1, p < 0.0001) were both greater for plants at site B than site A. However, there were no differences between plants that were assigned to insecticide application and plants that were assigned to be controls in initial root crown diameter at either site (site A: χ2 = 1.503, df = 1, p = 0.22; site B: χ2 = 0.035, df = 1, p = 0.85) or in height at site A (χ2 = 0.54, df = 1, p = 0.46). At site B, control plants were taller than plants that were assigned to receive insecticide at the beginning of the experiment (χ2 = 6.459, df = 1, p = 0.011).

Herbivore damage

Applying insecticide to the apical meristem reduced the frequency of damage to apical meristems and did not affect damage by other insect herbivore guilds. The apical meristem on 4.5% of insecticide plants was damaged before producing an apical flower head, whereas the apical meristem on 78.7% of control plants was damaged (χ2 = 55.84, df =1, p < 0.0001). Throughout the Results section, “control plants” refers to the pooled water control and unmanipulated control plants when these two treatment levels did not differ. Proportion of flower heads damaged per plant was not significantly different between insecticide and control plants (χ2 = 3.6, df =2, p = 0.15). Neither measure of folivore damage differed among treatment levels (proportion of leaves damaged per plant: F1,1 = 1.66, P = 0.42; proportion of area damaged on leaves that suffered damage: F1,1 = 5.05, P = 0.27).

Mortality rates and seed production

The mortality rate of insecticide plants was significantly lower (Fig. 2; χ2 = 15.68, df = 1, p < 0.0001) than for control plants. No insecticide plants died at site B, and this empty cell in the contingency table forced us to pool across sites to analyze mortality rate. Including experimental plants that died before flowering, number of viable seeds produced per plant by insecticide plants were significantly higher than for control plants in both sites (Fig. 3; site A: χ2 = 12.45, df = 1, p = 0.0004; site B: χ2 = 17.15, df = 1, p < 0.0001). Similarly, when plants that died before flowering were included in the analysis, total mass of viable seeds produced per plant was greater for insecticide plants at both sites (Fig. 3; site A: χ2 = 12.54, df = 1, p = 0.0004; site B: χ2 = 17.62, df = 1, p < 0.0001). Among the subset of plants that survived to flower, insecticide plants produced significantly more viable seeds per plant than control plants in site B (χ2 = 5.54, df = 1, p = 0.019). At site A, the difference in viable seed production was marginally significant after a Bonferroni adjustment to the significance threshold for testing the sites separately (χ2 = 3.86, df = 1, p = 0.049). Total mass of viable seeds produced per plant was greater for insecticide plants than for control plants at site A when only plants that survived to bloom were analyzed (χ2 = 5.86, df = 1, p = 0.015). The difference was marginally significant, after Bonferroni adjustment, at site B (χ2 = 4.2, df = 1, p = 0.04). Number of viable seeds per flower head that bloomed was greater for insecticide plants than for control plants (χ2 = 10.07, df = 1, p = 0.0015). Protection of the apical meristem did not affect the mean mass of individual viable seeds (F1,1 = 0.91, P = 0.52).

Details are in the caption following the image

Number of plants of Cirsium altissimum that lived to disperse seeds vs. number of plants that died before dispersing seeds for insecticide and control plants. Plants were pooled across sites. Water control and unmanipulated control plants were pooled.

Details are in the caption following the image

Mean (±SE) number and total mass of undamaged, viable seeds produced by insecticide and control plants of Cirsium altissimum. (A) Mean number of viable seeds of all flower heads of each plant. (B) Mean total mass of viable seeds of all flower heads of each plant. Means include plants that died before reproducing. Plants were pooled across water and unmanipulated controls.

Greenhouse experiment for seed viability and seedling biomass

There was no difference between insecticide and control plants in germination rate of viable seeds (χ2 = 0.0031, df = 1, p = 0.96). Biomass of seedlings produced by insecticide plants was marginally significantly less than biomass of seedlings produced by control plants (χ2 = 2.979, df = 1, p = 0.084).

Effects of flower head position within a plant

Terminal flower heads (Fig. 1) had significantly greater diameters (sign test: M = −7.5, P < 0.0001) and produced 78.7% more viable seeds per flower head (t = 3.19, df = 14, P = 0.007) than axial flower heads on the same branch. Individual seed mass did not differ between terminal vs. axial flower heads (t = −1.62, df = 11, P = 0.13).

Architecture and flowering phenology

For final height, plants from site B remained significantly taller than plants from site A (χ2 = 20.14, df = 1, p < 0.0001). Control plants were shorter (site A: χ2 = 14.08, df = 1, p = 0.0002; site B: χ2 = 8.05, df = 1, p = 0.0046) than insecticide plants at both sites.

The herbivory treatment did not significantly affect number of primary branches per plant, number of flower heads per plant or number of flower heads bloomed per plant at either site, after Bonferonni corrections for separate comparisons at the two sites. Plants at site A produced more branches (χ2 = 17.87, df = 1, p < 0.001), flower heads (χ2 = 23.75, df = 1, p < 0.001), and flower heads that bloomed (χ2 = 14.1, df = 1, p = 0.002) than did the plants at site B. Proportion of blooming flower heads per plant that were axial, rather than terminal, was greater in control plants than in insecticide plants (Fig. 4; χ2 = 7.17, df = 1, p = 0.0074).

Details are in the caption following the image

Mean (±SE) proportion of axial heads among the total heads bloomed in insecticide and control plants of Cirsium altissimum. Axial flower heads are heads that are not at the apex of the main stem or a lateral branch. Plants were pooled across sites. Water control and unmanipulated control plants were pooled.

Experimental tall thistles flowered between 13 August and 21 October 2012. Apical meristem mining delayed initial date of flowering (Fig. 5; χ2 = 27.7, df = 2, p < 0.0001). Sixty-two percent of insecticide plants began flowering by 19 August, whereas only 27% of control plants began flowering by 19 August. Latest date of flowering was unaffected by apical meristem mining or site (apical meristem mining: χ2 = 3.29, df = 2, p = 0.19; site: χ2 = 0.11, df = 1, p = 0.74).

Details are in the caption following the image

Flowering phenology of insecticide and control plants of Cirsium altissiumum. (A) Initial dates of flowering. (B) Peak flowering date. (C) Duration of flowering. (D) Maximum floral display size. Plants were pooled across water and unmanipulated controls for all four panels. Plants were pooled across sites for initial dates of flowering and peak flowering date.

Peak flowering date was significantly earlier in insecticide plants than in control plants (Fig. 5; χ2 = 4.62, df = 1, p = 0.032). Twenty-eight percent of insecticide plants reached peak flowering by 19 August, whereas only 10% of control plants reached peak flowering by that date. Insecticide plants had smaller maximum floral display sizes than control plants (χ2 = 6.24, df = 2, p = 0.044). This pattern was consistent across sites. Plants at site B had smaller maximum floral displays than plants at site A (χ2 = 10.69, df = 2, p = 0.0048).

Flowering duration was significantly longer for insecticide plants than control plants (Fig. 5; χ2 = 13.96, df = 2, p = 0.0009). This pattern was consistent across sites. For insecticide plants, the modal flowering duration was 2 wk with a maximum of 10 wk. For control plants, the modal flowering duration was 1 wk with a maximum of 5 wk. Flowering duration was longer at site A than at site B (χ2 = 11.12, df = 2, p = 0.0038).

DISCUSSION

While apical meristem herbivory has received considerable attention in research that explores plant compensatory mechanisms, there are few studies quantifying maternal fitness effects of insect damage to apical meristems (Hendrix, 1979; Hendrix and Trapp, 1989, Naber and Aarssen, 1998; Lortie and Aarssen, 2000; Hamback et al., 2011). In C. altissimum, apical meristem mining (consumption of meristem tissue by larvae developing inside the apical meristem) reduced lifetime maternal fitness whether it is quantified as seed production or mass of seeds produced. These reductions arose through higher mortality rates of plants with damaged apical meristems and through undercompensation in reproduction by damaged plants that survived. Undercompensation by surviving plants results, at least in part, from a shift in the position of flower heads that successfully produced seed within the plant. Further, changes in plant height and flowering phenology provide mechanisms by which indirect effects of apical meristem mining on lifetime seed production could arise through changes in pollinator visitation. Indirect effects of apical meristem mining on damage by insects that feed inside developing flower heads seem less likely to influence the compensatory response by tall thistle because we found no difference in damage to flower heads between insecticide and control treatment levels.

The proportion of adult tall thistles that died before producing seed was greater for plants that suffered apical meristem mining. This effect on mortality rate was critical to reducing mean maternal fitness. Nevertheless, few studies have reported effects of apical damage by either vertebrate or invertebrate herbivores on plant mortality rates. This lack may reflect generally small effects of apical damage on plant survival (Lennartsson et al., 1998). Alternatively, it may reflect a research focus on plant compensatory mechanisms, which requires that plants live to reproduce, rather than on quantifying the total effect of apical damage on maternal fitness. Ehrlen (1995) found that slug herbivory of Lathyrus vernus meristems reduced plant survival rate. Delayed growth by slug-damaged L. vernus may have prevented these plants from taking advantage of high light availability before leaf expansion by the forest canopy (Ehrlen, 1995). Reduced access to light could be a common mechanism by which apical damage affects plant survival. Light competition likely has been an important selective force in the evolution of apical dominance (Aarssen and Irwin, 1991), and so interactions that impede the production of a tall primary stem might be expected to reduce fitness through an interaction with competition. Shorter height of tall thistles with mined apical meristems may have increased the competitive effect of neighboring grasses. Previously, light competition from grasses at the Ninnescah Reserve has been shown to significantly reduce tall thistle rosette survival (Russell and Spencer, 2010).

Among tall thistles that survived apical meristem mining, we found undercompensation in maternal fitness at one site and a trend in that direction at the second site. At both sites, lifetime seed production by surviving plants was reduced by >30%. Undercompensation for apical meristem damage in C. altissimum may reflect this species’ determinate flowering. As compared with species with indeterminate flowering, weaker apical dominance in determinate flowering species may lead to less difference in branching between damaged and undamaged plants and, hence, less compensatory seed production (Lortie and Aarssen, 2000). While apical meristem mining in tall thistle did not result in more primary branches per plant, mined plants may have had more higher-order branches. We did not quantify secondary and tertiary branching, but damaged plants did have more axial flower heads that bloomed. Benner (1988) found that clipping the main stem of Thlaspi arvense did not increase the number of primary branches per plant, but did increase secondary and tertiary branches. Alternately, tall thistle appears to produce “excess” flower heads that never mature, even when these flower heads are undamaged (F. L. Russell, personal observations). Rather than producing more higher-order branches, tall thistle's compensatory response may involve redirecting resources to develop axial flower heads that otherwise would not bloom.

Less seed production by surviving plants with apical meristem mining results from fewer seeds produced per flower head that bloomed. This reduction in viable seed production per flower head may reflect a shift in the identity of flower heads that bloomed toward flower heads in positions on the plant with lower seed production capacity. Axial flower heads constituted a larger proportion of the blooming flower heads on plants with apical damage than on plants with intact apical meristems. Further, axial flower heads are smaller and produce fewer viable seeds than terminal flower heads on the same branch. This shift in the position of flower heads that were primarily responsible for seed production, however, did not affect seed quality. This finding contrasts with responses in several other monocarpic species that have been used to study apical damage. In Pastinaca sativa (Hendrix, 1979) and Erysimum strictum (Rautio et al., 2005), a shift in seed production toward axillary or more basal branches was associated with lighter seeds.

Delayed flowering as a result of apical meristem herbivory has the potential to directly limit seed production by shortening the time available for replacement inflorescences to mature seed (Huhta et al., 2000; Freeman et al., 2003). Although initial flowering and peak flowering of tall thistle plants were delayed by apical meristem mining, it seems unlikely that delayed flowering directly limits development of replacement flower heads and seed production. Damaged plants finished blooming weeks before freezing temperatures. No damaged plants were blooming after 30 September and the earliest date of a killing freeze ever recorded in Wichita KS is 8 October (NOAA, 2014). Further, because several tall thistles with protected apical meristems flowered for weeks after damaged plants ceased flowering, development of additional flower heads is not constrained by seasonal cues that might be adaptive to ensure time for seed development. In fact, while apical damage affected initial and peak flowering, it had no effect on the latest date of flowering. Tall thistle's inability to extend flower head development, even when climate permits, likely reflects physiological constraints, perhaps related to the number of flower head primordia available to develop in response to damage or resource constraints.

The types of architectural and phenological changes that occurred in tall thistles with mined apical meristems have been shown to affect the magnitude of subsequent floral visitation in other plant species (Juenger et al., 2005; Brody and Irwin, 2012). Although architectural changes in damaged tall thistles generally were less than we expected, apically damaged plants were shorter than undamaged plants, and in some species, shorter inflorescences receive less floral visitation (Carromero and Hamrick, 2005; Gómez, 2008). Shorter flowering duration may limit the extent of pollen transfer. Self-incompatibility in tall thistle (Adhikari, 2013) further supports the possibility that less-effective pollinator service could contribute to reduced viable seed production per flower head. However, with its wide, horizontal flower heads, tall thistle is visited by a diverse set of floral insects. Eleven bee species were collected from flower heads of tall thistles outside of our experiment at the Ninnescah Reserve in 2012 (Adhikari, 2013), and lepidopterans and beetles also visited flower heads. This diversity of floral visitors may lessen the potential for indirect effects of apical meristem mining through reduced number of pollinator visits. Their efficacy in transferring pollen, however, is unknown, and changes in species composition of floral visitors across the growing season could interact with altered flowering phenology to influence the success of pollen transfer in damaged plants. Definitive conclusions concerning the contributions of modified interactions with pollinators to the degree of reproductive compensation that C. altissimum achieves for apical meristem mining require extensive observations of pollinator visitation and patterns of pollen transfer.

Our results suggest several productive research directions. First, we hypothesize that interaction effects between apical meristem mining and light competition reduced tall thistle survival. Greater understanding of the circumstances under which native insect herbivory and plant competition interact to reduce plant lifetime seed production is important for designing environmentally benign strategies for limiting populations of weedy species, like thistles (Suwa and Louda, 2012). Second, compensatory mechanisms for apical damage in tall thistle appear weak in comparison with compensatory responses in other monocarpic species. Weak compensatory mechanisms in tall thistle, which has determinate flowering, may be consistent with the hypothesis that indeterminate flowering species will show stronger compensatory responses than determinate flowering species because of stronger apical dominance in the undamaged condition. Alternately, the small amount of tissue consumed during apical meristem mining may have been insufficient to generate strong compensatory responses, but still may have reduced maternal fitness by eliminating the apical flower head, which is the most productive flower head on the plant. Disentangling the influences of plant development vs. the nature of meristem damage would improve predictions for the outcome of apical meristem mining for plant maternal fitness.