Open Access
How to translate text using browser tools
20 February 2024 Assessment of leaf defense hypotheses at the intraspecific scale in three Helianthus (Asteraceae) species
Nora Mitchell, Madilyn N. Vetter, Michael Bylander, Thu A. Nguyen, Lydia McNabb, Aleks R. Leonardson, Chloe E. Meyer
Author Affiliations +

Interactions between plants and insects have shaped biodiversity at multiple ecological and evolutionary scales in both diverse taxonomic groups. Antagonistic interactions (such as herbivory) can impose strong selective pressures on plants, resulting in increased defense levels or diverse defense strategies. The nature and extent of plant defenses can vary both within and between species, potentially reflecting trade-offs between defense and growth strategy that are associated with environmental resources, known as the resource availability hypothesis. Global interspecific patterns may or may not reflect mechanisms acting within species at more restricted geographic scales. Here, we ask whether there is evidence for associations between growth strategies and defense levels, resource availability, and herbivore damage within three Helianthus species (H. giganteus, H. grosseserratus, and H. maximiliani) in a restricted region of the Upper Midwest in the USA. We measure growth traits, leaf defense traits, and leaf herbivore damage levels in wild sunflower populations to assess patterns across populations and in two common gardens to assess patterns within populations. We estimate associations between growth traits and defense traits, defense traits and environmental resources, and herbivore damage and environmental resources. Overall, we find that slower growth strategies are associated with increased levels of defenses, though these higher defense levels are not associated with lower-resource environments in wild populations, and some of these patterns are detected in the commons. We also find that herbivore damage levels are not associated with these resource levels. We conclude that defenses in these species are related to growth strategy even at this intraspecific scale and are largely in line with macroevolutionary patterns across the genus. Both defense levels and herbivore damage are not strongly related to resource availability, which may reflect limited amounts of resource variation in this restricted region and follow prediction within a general framework for intraspecific defense trait associations.

The interactions between insects and plants have helped to shape biodiversity both within and among species, potentially contributing to the extraordinarily high levels of species richness in both groups (Ehrlich and Raven 1964; Farrell et al. 1992; Janz 2011). Antagonistic plant–herbivore interactions have been a focus for the study of potential reciprocal adaptation and coevolution driving biodiversity (reviewed in Futuyma and Agrawal 2009). Plant defenses against herbivore pressures are numerous, including structural traits such as the immense diversity of trichomes (including glandular and nonglandular), which can serve as a major form of defense against herbivore attack (Esau 1953; Levin 1973; Wagner 1991; Werker 2000; Wagner et al. 2004), or general scleromorphy, which reduces palatability/digestibility (Hanley et al. 2007). Defense traits have been linked to either reduced herbivore damage or reduced herbivore presence, but associations are not always detectable and vary broadly (e.g., McKey et al. 1978; Agrawal 2005; Carmona et al. 2011). Studies spanning multiple taxonomic scales, functional and experimental assays, and community ecological work are necessary to more fully understand what drives the diversity and levels of plant defenses, including how environmental context affects plant strategies (Agrawal 2011).

Variation in levels and types of plant defenses can be attributed to multiple factors and may exist at different scales (Stamp 2003). The resource availability hypothesis is perhaps the most prominent theory explaining how abiotic environmental factors can affect plant defenses against herbivores. This hypothesis states that plants in resource-rich environments have faster growth (and regrowth) rates and lower investment in antiherbivore defenses since resources are plentiful to facilitate regrowth (Coley et al. 1985). In resource-poor environments, plants are expected to have slower growth rates and higher investment in antiherbivore defenses as the reinvestment of resources lost to herbivory in slow-growing plants is more costly, and any loss of biomass represents a greater proportion of investment than in faster-growing plants (Coley et al. 1985). Herbivore damage levels are therefore also higher in high-resource environments (due to the lack of defenses).

At the interspecific level, the resource availability hypothesis is supported by observational and experimental studies documenting the predicted patterns and trade-offs (e.g., Coley 1987; Endara and Coley 2011). Examining multiple species within lineages, predicted associations along environmental gradients are supported in a phylogenetic context across oak species on the global scale (Pearse and Hipp 2012) and congeneric salt marsh species pairs from high and low latitudes (Pennings et al. 2007). However, other meta-analyses do not find strong support for these ideas, including a lack of consistent latitudinal associations with defenses and herbivory (Moles et al. 2011).

Beyond the conflicting evidence at this scale, interspecific patterns are not always detectable within species, where patterns can be highly dependent on spatial and temporal variation (Bazzaz et al. 1987). There can be substantial intraspecific variation in plant traits, especially among populations found along environmental gradients (Albert et al. 2010). The extent and sources of intraspecific variation are also highly variable, ranging from phenotypic plasticity and physiology (detectable among individuals) to microevolutionary differences (detectable among populations) (Westerband et al. 2021). In terms of the resource availability hypothesis, different drivers may be affecting plant defenses at the inter-versus intraspecific scales (Hahn and Maron 2016). Across species, wide ranges in resources and the general use of species mean trait values can result in the predicted negative relationship between resource and defenses, whereas selective pressures within species (that represent more restricted portions of the resource continuum) may vary, with low-resource species exhibiting predicted patterns and high-resource species demonstrating opposite trends driven by biotic environment rather than abiotic (Hahn and Maron 2016). The scale of study therefore matters, where analyses across populations may indicate local adaptation and those among individuals within a population may reflect physiological trade-offs.

Evidence for intraspecific predictions associated with the resource availability hypothesis also varies. Multiple studies use latitudinal gradients as proxies for resource availability and examine associations across populations, with overall evidence for lower damage, increased palatability, and lower damage at higher latitudes (Pennings et al. 2001, 2007; Pennings and Silliman 2005; Woods et al. 2012; Nunes et al. 2016), whereas others test for clinal patterns with more specific environmental factors (such as temperature) (Abdala-Roberts et al. 2016; Hahn et al. 2019). These studies exhibit nuance beyond the straightforward expectations of the resource availability hypothesis, and results from reviews and meta-analyses at the intraspecific level also stress the lack of clear patterns and context-dependent nature of these patterns (Rasmann et al. 2014; Hahn and Maron 2016).

The genus Helianthus L. (Asteraceae; sunflowers) is a model system in plant ecology and evolution because of taxonomic richness (some 50 species), morphologic diversity (stark differences in growth form, life history, secondary metabolites, reproductive morphology, etc.), agricultural importance, sequenced genome, well-studied examples of hybridization, and interactions with insects via both pollination and herbivory (Seiler and Gulya 2004). Helianthus has substantial diversity in secondary metabolites, which have previously been used in systematic studies, as well as multiple types of trichomes used as both physical and chemical defenses (Seaman 1982; Spring et al. 1987; Spring 1991; Aschenbrenner et al. 2013; Brentan Silva et al. 2017). At the interspecific scale, Mason et al. (2016) used a phylogenetic comparative approach to assess herbivore defenses across 28 species of diploid Helianthus. They found that degree of defense was related to leaf economic strategy, with higher defenses in resource-conservative species, but that this was not necessarily tied to environmental resources. At the intraspecific scale, a study on the common sunflower H. annuus using seeds collected from a latitudinal gradient grown in a greenhouse setting, populations from lower latitudes had elevated resistance to herbivory and slower growth strategies (Beaton 2020). Whether or not these patterns hold true at a more restricted geographic scale associated with specific environmental factors and in perennial species, to our knowledge, remains untested.

Here, we ask whether there is evidence for the resource availability hypothesis within species in three closely related sunflower species using observational evidence from wild populations at the regional scale and evidence from two common gardens established using commercially available wild-collected seeds. The use of multiple species, wild populations, and common gardens are well suited to test intraspecific predictions related to the resource availability hypothesis (Hahn and Maron 2016). Specifically, we hypothesize that (H1) high levels of leaf defenses are associated with slower growth strategy both across populations and within common gardens; (H2) high levels of leaf defenses are associated with resource-poor environments; and (H3) levels of leaf herbivore damage are lower in resource-poor environments (Fig. 1).

Fig. 1.

Conceptual diagram of hypothesized predictions relating defense levels and growth rates, defense levels and resources, and damage levels and resources.


Materials and Methods.

Study Species.

We investigated three closely related species of herbaceous perennial Helianthus found in Wisconsin and into Minnesota: the giant sunflower H. giganteus L., the sawtooth sunflower H. grosseserratus M. Martens, and Maximilian's sunflower H. maximiliani Schrad. All three Helianthus species have broad distributions across North America, with H. giganteus largely found in the Midwest and Northeast of the USA; H. grossesseratus in the Midwest, Great Plains, and farther south; and H. maximiliani largely overlapping the range of the other two. Additionally, there have been documented cases of hybridization between these species either under natural conditions or via experimental crosses, largely made in the Midwest (Long 1955, 1959, 1960, 1961).

Wild Study Sites and Common Gardens.

We identified wild populations of all three species across Wisconsin and into nearby northeastern Minnesota using locations found in the Consortium of Wisconsin Herbaria and the community science database iNaturalist. We sampled a total of 20 populations (between 5 and 7 populations per species) between July and September 2020 (Fig. 2, Appendix 1). Populations included roadside stands, preserved prairies, and restored prairies.

Fig. 2.

Map of sampling locations. Pink = Helianthus giganteus, green = H. grosseserratus, light blue = H. maximiliani, orange = common gardens (overlapping).


We established two common gardens in Eau Claire, WI using commercially available seeds: Jeffers (44.8565°N, 91.5235°W) and Cornell (44.8399°N, 91.4925°W) (Fig. 2, overlapping orange points). Although located a short distance apart (approximately 3 km), the two sites differ in soil nutrients (see environmental data methods and results below). We obtained wild-collected seeds from Prairie Moon Nursery. Helianthus giganteus seeds were sourced from Wisconsin, H. grosseserratus from Illinois, and H. maximiliani from North Dakota. We exposed seeds to a 30-day wet/cold treatment before germination. In late April 2020, we germinated seeds in petri dishes, transferred seedlings to peat pellets (Jiffy), and established them in the University of Wisconsin–Eau Claire greenhouse. We watered seedlings, rotated trays daily, and used a standing rotating fan to simulate wind conditions until transplantation into common gardens.

We tractor-tilled the sites used for common gardens and planted seedlings during the first week of June. We randomly assigned locations to 75 seedlings of each species in each of the two common gardens. We planted seedlings 90 cm apart in rows 1 m apart. Because of unusually hot and dry conditions, we hand-watered the common garden plants for approximately 3 wk after transplantation. We performed routine maintenance on the common gardens, including hand-removal of weeds and installation of deer fencing around the gardens to deter large vertebrate predation.

Leaf Measurements.

We measured leaf traits, leaf damage, and trichome densities on plants from the common gardens and wild populations (Table 1). The leaf traits assessed are related to either growth strategy or putatively related to resistance or palatability. Although we are unaware of bioassays directly relating these traits to defense in sunflowers, there is evidence from other Helianthus species that these traits are under selection (Whitney et al. 2006) and they are also commonly used measures of leaf palatability (Hanley et al. 2007). We assessed all living plants from the common gardens in late July and haphazardly selected 10 plants (located at least 1 m apart) from each of the natural populations (see Appendix 1 for sampling dates). We harvested two recently fully expanded leaves per plant—we used the first to measure leaf investment traits and the second to estimate trichome densities and C:N ratio. For leaf traits, we measured the area of the leaf using a CI-202 portable laser leaf area meter (CID Bio-Science, Camas, WA) and the fresh mass of the leaf (without the petiole) using a microbalance. We estimated the leaf thickness (mm) using a digital micrometer (Mitutoyo, Kawasaki, Japan). We placed leaves in a drying oven for at least 48 hr before measuring leaf dry mass. We calculated specific leaf area (SLA, cm2 g–1) as leaf area/leaf dry mass and leaf dry matter content (LDMC, leaf dry mass/leaf fresh mass).

Table 1.

Descriptions of variables and abbreviations used in this study for plant traits, herbivore damage, and environmental variables.


For the second leaf sample, we used cork borers to extract leaf disks, which were then dried for at least 48 hr. We estimated the densities of glandular (GlandDens) and nonglandular (HairDens) trichomes on the abaxial side of a single dried leaf disk. We counted trichomes using a Leica s9i stereomicroscope (Leica, Wetzlar, Germany) under 5.5 × magnification with a 1 cm × 1 cm reticle and converted these raw values to densities (measuring a 0.18 cm × 0.18 cm area, 0.0325 cm2). To estimate leaf C:N ratio and leaf N content, we weighed 3–4 mg of material from another leaf disk for processing at the University of New Mexico Center for Stable Isotopes.

To assess herbivore damage, for each plant, we scored the three oldest nonsenesced leaves for insect damage as percentage of leaf area caused by different types of insect herbivores and calculated as a damage index D (see Whitney et al. 2006 for details). We constructed composite damage indices for leaf chewers (ChewDam: Orthoptera, Lepidoptera, Diptera) and for leaf vascular tissue feeders (SuckDam: Hemiptera, Homoptera) by summing D scores for each of the component taxa.

Environmental Data.

To characterize the climate of each wild population site, we downloaded historical climate data from WorldClim at 30-sec resolution (Fick and Hijmans 2017). We chose to analyze mean annual temperature and mean annual precipitation, as these two variables are commonly analyzed in the literature and broadly capture the variation across the region. We used the CGIAR Consortium for Spatial Information database to download aridity index (AI) values (at 30 arc seconds) for each of our populations (Zomer et al. 2022) ( Here, higher values correspond to higher humidity and lower values correspond to higher aridity. To obtain information on soil nutrient content, we revisited sites in August–September 2021 and collected a surface soil sample from each wild population and three samples from across each common garden. Soil samples were processed at Midwest Laboratories (Omaha, NE) for soil texture, the “S1AN” soil nutrient package. We chose to analyze the effects of mean annual temperature (MAT), mean annual precipitation (MAP), soil P content, soil N content, and soil % organic matter on the plant defense traits (Table 1). In this study area, we would generally characterize “resource-rich” environments as those with higher levels of precipitation (MAP and AI), warmer temperatures, and higher levels of soil N, P, and organic matter (Fig. 1).

Statistical Analyses.

We conducted all analyses in R v4.0.2 (R Core Development Team 2016). To assess overall relationships between leaf defenses and growth strategy across populations, we standardized major axis regressions (SMAR) across populations separately for each species with the population mean values for each defense trait as the response variables and the population mean values for each growth trait as predictors using the sma() function in the R package smatr (Warton et al. 2018). We chose to use SMAR because these analyses allow for uncertainty in the estimation of both x and y variables, and we chose to use population means to examine patterns across populations (rather than within). For the common gardens, we ran similar SMARs but did not take common garden averages; instead, both response and predictor variables were individual plant measurements, to assess patterns across individuals. We also included common garden (Jeffers versus Cornell) as “groups” and used a likelihood ratio test to determine whether the SMA lines shared a common slope between gardens and to estimate the confidence intervals for the slope (where the null hypothesis is that the slopes are the same between common gardens). We examined multivariate trait space by performing principal components analysis (PCA) on standardized traits separately for each species in both the wild populations and common gardens using the princomp() function.

To test for relationships between both leaf defense traits and herbivore damage with environmental variables, we ran univariate linear mixed effects models separately for each species with each trait as a response variable and each environmental variable as a predictor with site as a random effect using the lme() function in the R package nlme (Pinheiro et al. 2015). We used site as a random effect to account for the fact that multiple individuals were measured per population. We also assessed differences between the common gardens in resource traits, defense traits, and herbivore damage levels using Welch two-sample t tests using the t.test() function, separately for each species.


Variation in Traits And Environments.

We found substantial variation both within and among the three species in terms of growth traits, defense traits, and amount of leaf damage experienced both in the wild populations and in the common gardens (Appendix 2; Figs. S1, S2). For one example characterizing trait variation, in the wild populations, values for SLA (cm2 g–1) are within the global range and close to the median value for herbaceous plants (Poorter et al. 2009), with H. giganteus having higher values (mean = 185.9, SD = 59.1) and H. grosseserratus and H. maximiliani having similar values (mean = 109.8, SD = 17.2; mean = 107.0, SD = 21.5, respectively). Common garden mean SLA in H. giganteus was similar to that in wild populations (mean = 175.7, SD = 40.2) but was higher in H. grosseserratus and H. maximiliani (mean = 163.5, SD = 31.3; mean = 157.5, SD = 49.8, respectively). See Appendix 2 for trait data summaries for all traits and Figs. S1, S2 for visualization of trait data.

Wild sites and common gardens generally had environmental characteristics associated with high resources. Mean annual temperature ranged from 3.8°C to 8.0°C, with H. grosseserratus tending to occupy warmer areas (Fig. S3). Mean annual precipitation ranged from 739 mm to 846 mm and AI ranged from 0.720 to 0.868. Soil P ranged from 2.0 parts per million (ppm) to 83.0 ppm (with one high value for H. maximiliani), soil N ranged from 1.0 ppm to 57.0 ppm (with one high value for H. giganteus), and soil organic matter content ranged from 0.60% to 19.1%. Climatic values for the two common gardens were essentially the same, but Jeffers had higher levels of soil P than Cornell (51.3 ppm versus 16.3 ppm, averaged across three samples per garden). See Appendix 2 for full environmental data summaries and Fig. S3 for full visualization of environmental data.

Leaf Defenses and Growth Strategies.

We tested for a total of 30 possible associations between defense traits and growth traits (10 for each species) in each setting (wild populations and common gardens). Higher values for all five defense traits correspond to increased defenses, whereas higher values for the two growth traits correspond to faster growth strategies, so we expected negative relationships (fast growth traits with low defenses and vice versa). We found evidence for six significant negative associations in the wild populations, including negative associations between SLA and leaf thickness, LDMC, and glandular trichome densities in H. grosseserratus and negative associations between leaf N and leaf C:N ratio in all three species (Fig. 3, Appendix 3). We found two significant positive associations: higher glandular trichome density with faster growth strategy in H. giganteus (SLA) and H. maximiliani (leaf N) (Fig. 3, Appendix 3). See Appendix 3 for full statistical results.

Fig. 3.

Associations between defense traits and growth traits (A) specific leaf area (SLA, green) and (B) leaf N (pink) for each species. Length of bar corresponds to the marginal R2 value; bars to the left of the dashed zero line indicate a negative association and bars to the right of the dashed zero line indicate a positive association. Associations significant at P < 0.05 are deeply shaded.


We found six significant negative associations between growth and defense traits at the Cornell common garden, where SLA and LDMC were negatively associated across all three species, SLA was negatively associated with leaf thickness in H. grosseserratus, and leaf N and C:N ratio were negatively associated in H. giganteus and H. maximiliani (Table 2). In contrast, there was only one negative association at the Jeffers common garden, where leaf N and C:N ratio were negatively related in H. grosseserratus (Table 2). Furthermore, 19 of the 30 associations had significantly different slopes between the common gardens, and most of the time the relationship was stronger at Cornell than at Jeffers. Thirteen of these differences were cases where the direction of the relationship was the same (11 both negative, 2 both positive), but the slope differed, and in 12 of these 13 cases, the slope was steeper at Cornell (Appendix 4). In the remaining six significant slope differences between common gardens, the signs of the slopes differed. See Appendix 4 for full statistical results.

Table 2.

Common garden standardized major axis regressions results. Includes slope estimate, R2 values, and associated P-values for all combinations of growth and defense traits across all three species of Helianthus and two common gardens. Relationships significant at P < 0.05 are in bold. See Appendix 4 for full statistical results; see Table 1 for abbreviations.


We used PCAs to examine multivariate trait space in defense and growth traits in both the wild populations and common gardens. In the wild populations, growth traits (SLA and leaf N) were sometimes orthogonal, loading onto principal component (PC)1 versus PC2 (Fig. 4A). SLA generally traded off with defense traits like leaf thickness, LDMC, and hair density along PC1, whereas leaf N generally traded off with C:N ratio along PC2 (Fig. 4A). Glandular trichome densities did not consistently fit into these patterns in H. giganteus or H. maximiliani, whereas patterns in the latter species were less clear overall (Fig. 4A). In the common gardens, growth traits and defense traits largely traded off along PC1 in all three species, whereas trichomes loaded onto PC2 in different ways (Fig. 4B). For instance, in H. giganteus, nonglandular trichome density and leaf thickness loaded heavily onto PC2, whereas glandular trichome density loaded onto PC2 in H. grosseserratus, and both trichome types loaded onto PC2 in H. maximiliani (Fig. 4B). See Appendix 5 for details of the PCAs, including loadings and percentage of variation explained by each axis.

Fig. 4.

Principal components analysis results for defense and growth traits, analyzed separately for each species in both the (A) wild populations and (B) common gardens. Ellipses are 95% confidence intervals. Arrows point to direction of increasing values for the labeled trait variable. Pink = Helianthus giganteus, green = H. grosseserratus, light blue = H. maximiliani.


Leaf Defenses and Environmental Variables.

In line with the resource availability hypothesis, we expected negative relationships between defense traits and environmental variables; however, there were only a few detectable relationships between defenses and environment in these plant populations. Of the 90 associations tested (three species × five traits × six environmental variables), only six were significant (five positive and one negative) (Table 3, Appendix 6). Mean annual temperature was positively associated with nonglandular trichome density in H. giganteus (b = 719.056, t = 3.075, d.f. = 6, P = 0.022, marginal R2 = 0.414). Mean annual precipitation was positively associated with both leaf thickness (b = 0.001, t = 3.303, d.f. = 5, P = 0.021, marginal R2 = 0.233) and leaf dry matter content in H. grosseserratus (b < 0.001, t = 3.409, d.f. = 5, P = 0.019, marginal R2 = 0.255). Aridity index was positively related to nonglandular trichome density in H. grosseserratus (b = 105798, t = 3.248, d.f. = 5, P = 0.023, marginal R2 = 0.313). In terms of soil variables, in H. grosseserratus, soil N was negatively related to leaf C:N ratio (b = –0.199, t = –4.201, d.f. = 5, P = 0.008, marginal R2 = 0.457), whereas soil organic matter was positively related to glandular trichome density (b = 68.757, P = 0.024, t = 3.191, d.f. = 5, marginal R2 = 0.306). See Appendix 6 for full statistical results.

Table 3.

Defense trait–environment relationship linear mixed effect modeling results for three Helianthus species. Numbers are marginal R2 values from the model; + or – in parentheses refer to positive or negative relationship. Relationships significant at P < 0.05 are in bold. See Table 1 for description of abbreviations and Appendix 6 for full statistical results.


Herbivore Damage and Environmental Variables.

We expected higher levels of herbivore damage in higher-resource environments (positive relationships)—of 36 potential associations, we found 8 significant ones. We detected positive associations between leaf chewing damage and MAT, MAP, AI, and soil organic matter, all in H. grosseserratus (Fig. 5, Appendix 7). With sucking damage, we found positive associations with soil P in H. maximiliani, positive associations with soil N in H. giganteus and H. maximiliani, and with soil organic matter in H. giganteus. See Appendix 7 for full statistical results.

Fig. 5.

Associations between environmental variables and damage traits (A) ChewDam (blue) and (B) SuckDam (yellow) for each species. Length of bar corresponds to the marginal R2 value; bars to the left of the zero line indicate a negative association and bars to the right of the zero dashed line indicate a positive association. Associations significant at P < 0.05 are deeply shaded.


Trait Comparisons between Common Gardens.

We compared trait values between the two common gardens as a preliminary means of examining trait differences in low- versus high-resource environments, where Cornell has naturally lower soil P levels and is thus considered a lower-resouce site, and Jeffers has naturally higher soil P levels and is therefore considered a higher-resource site (Fig. S3; Appendix 2). In line with the resouce availability hypothesis, we expected higher values for resource traits, lower values for defense traits, and higher levels of herbivore damage at Jeffers compared with Cornell. In all three species, both resource-related traits (SLA and leaf N) were higher at Jeffers than at Cornell (Fig. 6). Of the 15 potential differences in defense traits, 10 were significant, and in all 10 the defense trait values were lower at Jeffers than at Cornell (Fig. 6). Only nonglandular trichome density was not significantly different between gardens in any of the species. Finally, of the six potential differences in herbivore damage levels (ChewDam and SuckDam), only one was significant, where chewing damage was higher at Jeffers than at Cornell (Fig. 6). See Appendix 8 for full statistical results.

Fig. 6.

Comparisons between plant traits at common gardens. Box plots represent the mean, quarterlies, and outliers with jittered data points in gray. Trait values are on the y-axis, with the three different species (GIG = Helianthus giganteus, GRO = H. grosseserratus, MAX = H. maximiliani) on the x-axis, and colors indicate the common garden (light orange = Cornell, left; dark orange = Jeffers, right). Asterisks next to the species abbreviation denote significant differences between gardens (P < 0.05).



We asked whether interspecific patterns between leaf defenses and growth strategy, leaf defenses and environmental resources, and herbivore damage and environmental resources were also detectable at the intraspecific scale in three species of Helianthus within a single geographic region. We found evidence for associations between lower defense levels and faster growth strategies (negative relationships) in support of the first prediction of the resource availability hypothesis, but these were largely unrelated to environmental variables.

High Levels of Defense Are Generally Associated with Slower Growth Strategy Among Populations.

Overall, we found associations between slower growth strategies and increased levels of defenses, in line with expectations of the resource defense hypothesis (Coley et al. 1985). These associations were detectable across wild populations (within species, where of the eight detectable associations between leaf defense traits and growth traits, six were in the expected direction and two were opposite to that [both associated with glandular trichome density]) (Fig. 3, Appendix 3). Principal components analyses in the wild populations also reflected trade-offs between SLA and defense traits along one axis and leaf N and leaf C:N ratio along a second axis (Fig. 4, Appendix 5). These trade-offs across populations are potentially indicative of local adaptation, though population-level differences may also be driven by phenotypic plasticity in the field associated with environmental differences, herbivore pressure, or other biotic interactions. Herbivore-induced plasticity in floral traits, for example, is highly species specific and reflects the interactions among all interacting members of the floral community (Rusman et al. 2019).

Although not explicitly defined by the resource availability hypothesis, we also tested for associations between defenses and growth across individuals in each of our common gardens to ask whether intraspecific variation within a population follows the same patterns as among populations. In the common gardens, we found few significant associations, but all of these were in the expected directions (Table 2, Appendix 4). At this scale, growth–defense patterns are more likely the result of physiological trade-offs rather than adaptive variation. Similar trade-offs between growth and defense were found across individuals within populations of Pinus pinaster grown in common garden settings, though the strength of these relationships varied among populations (indicative of genetic differences) (Vázquez-González et al. 2020).

In both settings, aimed at detecting associations across or within populations, we found similar patterns between growth and defense. Hahn et al. (2021) compared defense–growth correlations in populations of Monarda fistulosa at multiple scales, including between high- and low-resource environments (Wisconsin and North Dakota) and among populations in Wisconsin. They found that scale mattered where they generally detected negative associations between regions but positive associations among populations found in Wisconsin (the high-resource region) where our study was also conducted. We found generally negative associations or failed to detect associations among populations. Hahn et al. (2021) also found neutral correlations within populations in common garden settings, in line with no significant associations in our gardens.

In multivariate trait space, our wild populations and common gardens differed in terms of relationships between the resource-associated traits. We found that SLA and leaf N load more or less congruently along the same PC axis in the common gardens, as expected from the leaf economics spectrum (Wright et al. 2004), but largely orthogonally in the wild populations (Fig. 4). This goes against the ideas of Zhou et al. (2022), who defined the SLA–leaf N relationships as coincident coordination, with the expectation of no relationship within species. These differences may reflect physiological trade-offs in the common gardens and potentially different adaptive drivers in the wild populations, and may also be driving differences in other relationships between traits.

These intraspecific patterns may also vary from species to species or across time (Bazzaz et al. 1987). Among our species, some patterns differed. For instance, across the wild populations glandular trichome density had relationships with SLA in opposite directions in H. giganteus and H. grosseserratus and no detectable relationship in H. maximiliani (Fig. 3). Likewise, although major PC axes were similar across species, individual traits are loaded onto the PC axes in different ways in our three species (Fig. 3, Appendix 5). These differences may reflect slightly different strategies among species in terms of defenses, in line with interspecific predictions and findings (Mason et al. 2016). We note that the three Helianthus studied here are perennials, where temporal variation across years could also affect ontogenetic allocation of resources (Boege and Marquis 2005; Mason and Donovan 2015), and we do not know the age of the plants measured.

We examined largely mechanical defenses or trichomes that contain secondary metabolites, rather than analyzing the specific composition of these metabolites, which are common and diverse across Helianthus. Additional associations between growth strategies and chemical defenses may differ from associations with the traits analyzed here (Aschenbrenner et al. 2013; Mason et al. 2016; Brentan Silva et al. 2017). However, we expect findings similar to those of Mason et al. (2016), where there were strong relationships between leaf economics spectrum traits and secondary metabolites across Helianthus.

Increased Leaf Defenses Are Not Associated with Environmental Resources.

In line with the resource availability hypothesis, we expected lower levels of leaf defenses in resource-rich environments (Coley 1987; Endara and Coley 2011). In our study area, we would generally characterize resource-rich environments as those with higher levels of precipitation (MAP and AI), warmer temperatures, and higher levels of soil N, P, and organic matter. We found only a handful of detectable relationships between leaf defenses and environmental variables in the wild populations, and these were generally in the direction opposite to the resource availability hypothesis predictions; we found higher defenses in resource-rich areas (Fig. 5, Appendix 6). These patterns were fairly consistent across the three focal species, though glandular trichome density in H. giganteus was positively associated with SLA, whereas this trait was not associated with any growth traits in the other species (Fig. 2). These findings opposite to the expectations of the interspecific resource availability hypothesis are, however, in line with the intraspecific framework of Hahn and Maron (2016), which predicts that positive relationships between resource and defenses in high-resource environments are driven by herbivore pressure rather than by physiological stresses. Mason et al. (2016) also did not find support for this prediction at the macroevolutionary scale across the genus. We did find that the common garden with the higher resources (in terms of soil P only) had higher levels of defense traits in most cases (Fig. 6, Appendix 8).

Across a latitudinal gradient, H. annuus populations from lower latitudes (higher resources) had increased herbivore resistance and slower growth, in line with predictions (Beaton 2020). It may be that these intraspecific patterns are only detectable across a larger breadth of climatic conditions, and that the location of our study, largely in Wisconsin and into Minnesota, may result in both reduced and different gradients of stressors in plants. For instance, MAT ranged from 6.9°C to 24.4°C across the latitudinal gradient of Beaton (2020), and only 3.8°C to 8.0°C at our sites, whereas MAP ranged from 73.7 mm to 1443.0 mm in the H. annuus study and only 739 mm to 846 mm in our study (Fig. S3; Appendix 2). We also did not find strong associations between soil nutrients and defense traits in the wild populations, which have been detected at (relatively) finer spatial scales within species (Sanczuk et al. 2021), despite the variation in soils across our sites (Fig. S3). Our common garden sites, however, had some extreme levels of soil P compared with the wild populations, with the Jeffers site having nearly four times higher P levels than Cornell and the wild populations. These high levels of P may remove any constraint imposed by nutrients in the framework of Hahn and Maron (2016).

Herbivore Damage Levels Are Sometimes Associated with Higher Resource Levels.

We expected higher levels of herbivore damage in locations with more plentiful resources. Both leaf chewing damage (inflicted by Orthoptera, Lepidoptera, and Diptera) in H. grosseserratus and sucking damage (inflicted by Hemiptera and Homoptera) in H. giganteus and H. maximiliani had some positive associations with aspects of temperature, moisture, and soil nutrients (Fig. 5). In the common gardens, only chewing damage in H. grossesseratus differed between the two locations, with higher damage levels at the higher-resource site (Jeffers), in line with expectations for both the interspecific and intraspecific resource availability hypotheses (Fig. 6, Appendix 8) (Coley 1985; Hahn and Maron 2016). The overall lack of strong patterns here could be linked to the lack of patterns between defense traits and resources across these environmental gradients. Both hypothesized relationships hinge on the idea that it is less costly for plants with access to plentiful resources to replace damaged tissue, so it is not as essential to invest in protection (Coley et al. 1985; Endara and Coley 2011). This reduced investment should reduce damage, so the lack of patterns in defense traits would logically lead to few patterns with damage traits.

Evidence for higher levels of herbivore damage associated with higher resources is inconsistent. For instance, studies report higher damage levels in Sal-via nubicola found at lower versus higher altitudes (but with different overall strategies in different populations) (Dostálek et al. 2016), a lack of differences in damage along a latitudinal gradient within Acacia falcata (Andrew and Hughes 2005), and increased damage at drier sites (opposite to the expectation) in 13 tropical tree species (Muehleisen et al. 2020). Our results are consistent with the lack of strong detectable relationships between herbivore damage levels and environmental resources.

Differences between Common Gardens Stress THE Importance of Ecological Scale.

There was stronger evidence for growth–defense trade-offs in the low-resource common garden (Cornell) than in the high-resource garden (Jeffers) (Table 2, Appendix 4), where nutrient stress may have resulted in this pattern because growth is constrained by resources at Cornell, whereas those at Jeffers were not subject to this constraint. This is in line with both predictions and observations made in the intraspecific defense framework of Hahn and Maron (2016). Jeffers plants had more resource-acquisitive trait values, lower defenses, and (in only one case) more damage than Cornell plants, consistent with predictions related to resource availability. Of interest, the only marked difference between these sites was soil P levels, though nutrients alone can markedly alter defense strategies. Across genotypes of the tall goldenrod Solidago altissima, those with artificially higher nutrient levels had lower constitutive plant defenses, but increased induced resistance after experimental exposure to herbivory (Burghardt 2016). Differences between constitutive and induced defenses could also be important, as they may represent different responses to herbivore pressures (energetic tradeoffs versus plastic responses) (Sun et al. 2023). Although we did not measure induced defenses in this study, experimental treatments could be applied to observe if this pattern can be replicated in sunflowers.


Overall, we found that faster growth strategies were associated with lower defense levels against herbivores within three Helianthus species, but that defense traits were not linked to the availability of resources in their growth environments. Evidence for trade-offs between defense and growth across populations may indicate local adaptation, whereas those in the common gardens within populations point to physiological constraints, which were stronger in the lower-resource garden. Thus, the main predictions of the resource availability hypothesis were not detectable at this regional scale but are in line with the intraspecific framework of Hahn and Maron (2016). Our findings within closely related species across Wisconsin were largely consistent with patterns found across Helianthus species at the macroevolutionary scale (Mason et al. 2016). These consistencies suggest that similar evolutionary or ecological processes may be acting in similar ways across geographic and phylogenetic scales. Results here provide a solid basis for understanding the complexities in this system and future work related to specific secondary metabolites, the ontogeny of defenses, and insights into intraspecific variation at both broad and restricted spatial scales.

©Copyright 2024 by The Torrey Botanical Society

Literature Cited


Abdala-Roberts, L., S. Rasmann, J. C. Berny-Mier Y Terán, F. Covelo, G. Glauser, and X. Moreira. 2016. Biotic and abiotic factors associated with altitudinal variation in plant traits and herbivory in a dominant oak species. American Journal of Botany 103: 2070–2078. Google Scholar


Agrawal, A. A. 2005. Natural selection on common milkweed (Asclepias syriaca) by a community of specialized insect herbivores. Evolutionary Ecology Research 7: 651–667. Google Scholar


Agrawal, A. A. 2011. Current trends in the evolutionary ecology of plant defence. Functional Ecology 25: 420–432. Google Scholar


Albert, C. H., W. Thuiller, N. G. Yoccoz, A. Soudant, F. Boucher, P. Saccone, and S. Lavorel. 2010. Intraspecific functional variability: Extent, structure and sources of variation. Journal of Ecology 98: 604–613. Google Scholar


andrew, N. R. and L. Hughes. 2005. Herbivore damage along a latitudinal gradient: Relative impacts of different feeding guilds. Oikos 108: 176–182. Google Scholar


Aschenbrenner, A.-K., S. Horakh, and O. Spring. 2013. Linear glandular trichomes of Helianthus (Asteraceae): Morphology, localization, metabolite activity and occurrence. AoB Plants 5: plt028–plt028. Google Scholar


Bazzaz, F. A., N. R. Chiariello, P. D. Coley, and L. F. Pitelka. 1987. Allocating resources to reproduction and defense. BioScience 37: 58–67. Google Scholar


Beaton, L. L. 2020. A latitudinal gradient in herbivore resistance in common sunflower, Helianthus annuus (Asteraceae). Plant Ecology and Evolution 153: 199–207. Google Scholar


Boege, K. and R. J. Marquis. 2005. Facing herbivory as you grow up: The ontogeny of resistance in plants. Trends in Ecology & Evolution 20: 441–448. Google Scholar


Brentan Silva, D., A.-K. Aschenbrenner, N. P. Lopes, and O. Spring. 2017. Direct analyses of secondary metabolites by mass spectrometry imaging (MSI) from sunflower (Helianthus annuus L.) trichomes. Molecules 22: 774. Google Scholar


Burghardt, K. T. 2016. Nutrient supply alters goldenrod's induced response to herbivory. Functional Ecology 30: 1769–1778. Google Scholar


Carmona, D., M. J. Lajeunesse, and M. T. J. Johnson. 2011. Plant traits that predict resistance to herbivores. Functional Ecology 25: 358–367. Google Scholar


Coley, P. D. 1987. Interspecific variation in plant anti-herbivore properties: The role of habitat quality and rate of disturbance. New Phytologist 106: 251–263. Google Scholar


Coley, P. D., J. P. Bryant, and F. S. Chapin. 1985. Resource availability and plant antiherbivore defense. Science 230: 895–899. Google Scholar


Dostálek, T., M. B. Rokaya, P. Maršík, J. Rezek, J. Skuhrovec, R. Pavela, and Z. Münzbergová. 2016. Trade-off among different anti-herbivore defence strategies along an altitudinal gradient. AoB PLANTS 8: plw026. Google Scholar


Ehrlich, P. R. and P. H. Raven. 1964. Butterflies and plants: A study in coevolution. Evolution 18: 586–608. Google Scholar


Endara, M.-J. and P. D. Coley. 2011. The resource availability hypothesis revisited: A meta-analysis. Functional Ecology 25: 389–398. Google Scholar


Esau, K. 1953. Plant Anatomy. Wiley, New York, NY. 735 pp. Google Scholar


Farrell, B. D., C. Mitter, and D. J. Futuyma. 1992. Diversification at the insect–plant Interface. BioScience 42: 34–42. Google Scholar


Fick, S. E. and R. J. Hijmans. 2017. WorldClim 2: New 1-km spatial resolution climate surfaces for global land areas. International Journal of Climatology 37: 4302–4315. Google Scholar


Futuyma, D. J. and A. A. Agrawal. 2009. Macroevolution and the biological diversity of plants and herbivores. Proceedings of the National Academy of Sciences of the United States of America 106: 18054–18061. Google Scholar


Hahn, P., A. Agrawal, K. Sussman, and J. Maron. 2019. Population variation, environmental gradients, and the evolutionary ecology of plant defense against herbivory. American Naturalist 193: 20–34. Google Scholar


Hahn, P. G., K. Keefover-Ring, L. M. N. Nguyen and J. L. Maron. 2021. Intraspecific correlations between growth and defence vary with resource availability and differ within and among populations. Functional Ecology 35: 2387–2396. Google Scholar


Hahn, P. G. and J. L. Maron. 2016. A framework for predicting intraspecific variation in plant defense. Trends in Ecology & Evolution 31: 646–656. Google Scholar


Hanley, M. E., B. B. Lamont, M. M. Fairbanks, and C. M. Rafferty. 2007. Plant structural traits and their role in anti-herbivore defence. Perspectives in Plant Ecology, Evolution and Systematics 8: 157–178. Google Scholar


Janz, N. 2011. Ehrlich and Raven revisited: Mechanisms underlying codiversification of plants and enemies. Annual Review of Ecology, Evolution, and Systematics 42: 71–89. Google Scholar


Levin, D. A. 1973. The role of trichomes in plant defense. Quarterly Review of Biology 48: 3–15. Google Scholar


Long, R. W. 1955. Hybridization in perennial sunflowers. American Journal of Botany 42: 769–777. Google Scholar


Long, R. W. 1959. Natural and artificial hybrids of Helianthus maximiliani × H. grosseserratus. American Journal of Botany 46: 687–692. Google Scholar


Long, R. W. 1960. Biosystematics of two perennial species of Helianthus (Compositae). I. Crossing relationships and transplant studies. American Journal of Botany 47: 729–735. Google Scholar


Long, R. W. 1961. Biosystematics of two perennial species of Helianthus (Compositae), II. Natural populations and taxonomy. Brittonia 13: 129–141. Google Scholar


Mason, C. M. and L. A. Donovan. 2015. Does investment in leaf defenses drive changes in leaf economic strategy? A focus on whole-plant ontogeny. Oecologia 177: 1053–1066. Google Scholar


Mason, C. M., A. W. Bowsher, B. L. Crowell, R. M. Celoy, C.-J. Tsai, and L. A. Donovan. 2016. Macroevolution of leaf defenses and secondary metabolites across the genus Helianthus. New Phytologist 209: 1720–1733. Google Scholar


Mckey, D., P. G. Waterman, C. N. Mbi, J. S. Gartlan, and T. T. Struhsaker. 1978. Phenolic content of vegetation in two African rain forests: Ecological implications. Science 202: 61–64. Google Scholar


Moles, A. T., S. P. Bonser, A. G. B. Poore, I. R. Wallis, and W. J. Foley. 2011. Assessing the evidence for latitudinal gradients in plant defence and herbivory. Functional Ecology 25: 380–388. Google Scholar


Muehleisen, A. J., B. M. J. Engelbrecht, F. A. Jones, E. Manzané-Pinzón, and L. S. Comita. 2020. Local adaptation to herbivory within tropical tree species along a rainfall gradient. Ecology 101: e03151. Google Scholar


Nunes, K. A., C. M. Cassin, and P. M. Kotanen. 2016. Variation in herbivory along a latitudinal gradient for native and exotic Asteraceae. Plant Ecology 217: 481–493. Google Scholar


Pearse, I. S. and A. L. Hipp. 2012. Global patterns of leaf defenses in oak species. Evolution 66: 2272–2286. Google Scholar


Pennings, S. C. and B. R. Silliman. 2005. Linking biogeography and community ecology: Latitudinal variation in plant–herbivore interaction strength. Ecology 86: 2310–2319. Google Scholar


Pennings, S. C., E. L. Siska, and M. D. Bertness. 2001. Latitudinal differences in plant palatability in Atlantic Coast salt marshes. Ecology 82: 1344–1359. Google Scholar


Pennings, S. C., M. Zimmer, N. Dias, M. Sprung, N. Davé, C.-K. Ho, A. Kunza, C. Mcfarlin, M. Mews, A. Pfauder, and C. Salgado. 2007. Latitudinal variation in plant–herbivore interactions in European salt marshes. Oikos 116: 543–549. Google Scholar


Pinheiro, J., D. Bates, S. Debroy, D. Sarkar, and R Core Develpment Team. 2015. nlme: Linear and nonlinear mixed effects models. R Package, Version 3. Google Scholar


Poorter, H., Ü. Niinemets, L. Poorter, I. J. Wright, and R. Villar. 2009. Causes and consequences of variation in leaf mass per area (LMA): A meta-analysis. New Phytologist 182: 565–588. Google Scholar


R Core Development Team. 2016. R: A language and environment for statistical computing. R Foundation for Statistical Computing, Vienna, Austria. Google Scholar


Rasmann, S., L. Pellissier, E. Defossez, H. Jactel, and G. Kunstler. 2014. Climate-driven change in plant–insect interactions along elevation gradients. Functional Ecology 28: 46–54. Google Scholar


Rusman, Q., D. Lucas-Barbosa, E. H. Poelman, and M. Dicke. 2019. Ecology of plastic flowers. Trends in Plant Science 24: 725–740. Google Scholar


Sanczuk, P., S. Govaert, C. Meeussen, K. De Pauw, T. Vanneste, L. Depauw, X. Moreira, J. Schoelynck, M. De Boevre, S. De Saeger, K. Bollmann, J. Brunet, S. A. O. Cousins, J. Plue, M. Diekmann, B. J. Graae, P.-O. Hedwall, G. Iacopetti, J. Lenoir, A. Orczewska, Q. Ponette, F. Selvi, F. Spicher, P. Vermeir, K. Calders, H. Verbeeck, K. Verheyen, P. Vangansbeke, and P. De Frenne. 2021. Small scale environmental variation modulates plant defence syndromes of understorey plants in deciduous forests of Europe. Global Ecology and Biogeography 30: 205–219. Google Scholar


Seaman, F. C. 1982. Sesquiterpene lactones as taxonomic characters in the Asteraceae. Botanical Review 48: 121–594. Google Scholar


Seiler, G. J. and T. J. Gulya. 2004. Exploration for wild Helianthus species in North America: Challenges and opportunities in the search for global treasures, pp. 43–68. International Sunflower Conference Proceedings. Google Scholar


Spring, O. 1991. Trichome microsampling of sesquiterpene lactones for the use of systematic studies, pp. 319–345. In N. H. Fischer, M. B. Isman, and H. A. Stafford, eds. Modern Phytochemical Methods. Recent Advances in Phytochemistry, Vol. 25. Springer, Boston, MA. Google Scholar


Spring, O., U. Bienert, and V. Klemt. 1987. Sesquiterpene lactones in glandular trichomes of sunflower leaves. Journal of Plant Physiology 130: 433–439. Google Scholar


Stamp, N. 2003. Out of the quagmire of plant defense hypotheses. Quarterly Review of Biology 78: 23–55. Google Scholar


Sun, X., Y. Sun, X. Cao, X. Zhai, R. M. Callaway, J. Wan, S. L. Flory, W. Huang, and J. Ding. 2023. Trade-offs in non-native plant herbivore defences enhance performance. Ecology Letters 00: 1–13. Scholar


Vázquez-González, C., L. Sampedro, V. Rozas, and R. Zas. 2020. Climate drives intraspecific differentiation in the expression of growth-defence trade-offs in a long-lived pine species. Scientific Reports 10: 10584. Google Scholar


Wagner, G. J. 1991. Secreting glandular trichomes: More than just hairs. Plant Physiology 96: 675–679. Google Scholar


Wagner, G. J., E. Wang, and R. W. Shepherd. 2004. New approaches for studying and exploiting an old protuberance, the plant trichome. Annals of Botany 93: 3–11. Google Scholar


Warton, D., R. Duursma, D. Falster, S. Taskinen, and M. R. Duursma. 2018. Package ‘smatr.’ CRAN-Software R (CRAN, 2015). Google Scholar


Werker, E. 2000. Trichome diversity and development, Advances in Botanical Research 31: 1–35. Google Scholar


Westerband, A. C., J. L. Funk, and K. E. Barton. 2021. Intraspecific trait variation in plants: A renewed focus on its role in ecological processes. Annals of Botany 127: 397–410. Google Scholar


Whitney, K. D., R. A. Randell, and L. H. Rieseberg. 2006. Adaptive introgression of herbivore resistance traits in the weedy sunflower Helianthus annuus. American Naturalist 167: 794–807. Google Scholar


Woods, E. C., A. P. Hastings, N. E. Turley, S. B. Heard, and A. A. Agrawal. 2012. Adaptive geographical clines in the growth and defense of a native plant. Ecological Monographs 82: 149–168. Google Scholar


Wright, I. J., P. B. Reich, M. Westoby, D. D. Ackerly, Z. Baruch, F. Bongers, J. Cavender-Bares, T. Chapin, J. H. C. Cornelissen, M. Diemer, J. Flexas, E. Garnier, P. K. Groom, J. Gulias, K. Hikosaka, B. B. Lamont, T. Lee, W. Lee, C. Lusk, J. J. Midgley, M.-L. Navas, U. Niinemets, J. Oleksyn, N. Osada, H. Poorter, P. Poot, L. Prior, V. I. Pyankov, C. Roumet, S. C. Thomas, M. G. Tjoelker, E. J. Veneklaas, and R. Villar. 2004. The worldwide leaf economics spectrum. Nature 428: 821–827. Google Scholar


Zhou, J., E. Cieraad, and P. M. Van Bodegom. 2022. Global analysis of trait–trait relationships within and between species. New Phytologist 233: 1643–1656. Google Scholar


Zomer, R. J., J. Xu, and A. Trabucco. 2022. Version 3 of the Global aridity index and potential evapotranspiration database. Scientific Data 9: 409. Google Scholar


Appendix 1

Sampling sites for wild populations of Helianthus and two common garden sites. All voucher specimens are deposited at the University of Wisconsin–Eau Claire herbarium (UWEC).


Appendix 2

Trait and environmental variable summaries. Means, standard deviations, and minimum and maximum values for each of the trait and environmental variables for wild populations and common gardens for each species of Helianthus. See Table 1 for description of abbreviations and Figs. S1, S2 for visualizations.


Appendix 2



Appendix 3

Defense–growth standardized major axis regression modeling results for wild populations of Helianthus; models run separately for each species on population means, with growth traits SLA and % N as response variables and defense traits leaf thickness, LDMC, HairDens, GlandDens, and CNratio as predictors (see Table 1 for description of abbreviations). Values indicate the elevation (intercept) estimates and 95% confidence intervals (CIs), slope estimates and 95% confidence intervals, and R 2 and P values.


Appendix 4

Defense–growth standardized major axis regression modeling results for common gardens; models run separately for each species of Helianthus on individual trait values, with growth traits SLA and % N as response variables and defense traits leaf thickness, LDMC, HairDens, GlandDens, and CNratio as predictors (see Table 1 for description of abbreviations). Values for each garden (Cornell and Jeffers) indicate the elevation (intercept) estimates and 95% confidence intervals (CIs), slope estimates and 95% CIs, and R 2 and P values. Comparisons indicate results from the likelihood ratio (LR) statistic, associated P value


Appendix 4



Appendix 4



Appendix 5

Growth and defense trait principal components analysis results. Analyses performed separately for each species of Helianthus for the wild populations and common gardens. Values are loadings for each principal component and proportion of variance (Prop of var) explained by each principal component. See Table 1 for description of abbreviations.


Appendix 5



Appendix 6

Defense–environment linear mixed effect modeling results. Models run separately for each species of Helianthus, with defense traits leaf thickness, LDMC, HairDens, GlandDens, and CNratio as response variables and environmental variables MAT, MAP, AI, Soil_P, Soil_N, and Soil_org as predictors (see Table 1 for description of abbreviations). Values indicate the b slopes, SE is the standard error of the value estimate, d.f. = degrees of freedom, t value = t statistic, associated P value and marginal R 2 values from ANOVA testing.


Appendix 6



Appendix 6



Appendix 7

Damage–environment linear mixed effect modeling results. Models run separately for each species of Helianthus, with damage traits ChewDam and SuckDam as response variables and environmental variables MAT, MAP, AI, Soil_P, Soil_N, and Soil_org as predictors (see Table 1 for description of abbreviations). Values indicate the b slopes; SE is the standard error of the value estimate, d.f. = degrees of freedom, t = t statistic, associated P and marginal R 2 values from ANOVA testing.


Appendix 8

Comparison of traits of Helianthus species between common gardens. t = t statistic, where positive values indicate higher trait values at Cornell and negative values indicate higher trait values at Jeffers; d.f. = degrees of freedom; PCI_low and CI_high are the confidence intervals for the mean under the alternative hypothesis, and Cornell and Jeffers means are the mean trait values at the respective common garden. Abbreviations and units follow Table 1.


Fig. S1.

Summaries for trait values associated with growth, defense, and herbivore damage in the wild populations. Boxplots display the medians, quartiles, upper and lower maxima, and outliers. Pink = H. giganteus, green = H. grosseserratus, blue = H. maximiliani.


Fig. S2.

Summaries for trait values associated with growth, defense, and herbivore damage in the common gardens. Boxplots display the medians, quartiles, upper and lower maxima, and outliers. Pink = H. giganteus, green = H. grosseserratus, blue = H. maximiliani.


Fig. S3.

Summaries for environmental variables at each site. Boxplots display the medians, quartiles, upper and lower maxima, and outliers. Pink = H. giganteus, green = H. grosseserratus, blue = H. maximiliani, orange = common garden (Jeffers and Cornell).

Nora Mitchell, Madilyn N. Vetter, Michael Bylander, Thu A. Nguyen, Lydia McNabb, Aleks R. Leonardson, and Chloe E. Meyer "Assessment of leaf defense hypotheses at the intraspecific scale in three Helianthus (Asteraceae) species," The Journal of the Torrey Botanical Society 151(1), 1-31, (20 February 2024).
Received: 17 April 2023; Published: 20 February 2024
intraspecific variation
Resource Availability Hypothesis
Back to Top