Ecological Value of Native Plant Cultivars Versus Wild-Type Native Plants for Promoting Hemipteran Diversity in Suburban Areas

Ecological Value of Native Plant Cultivars Versus Wild-Type Native Plants for Promoting... Abstract Because of concerns over recent declines in overall biodiversity in suburban areas, homeowners are attempting to improve the ecological functioning of their landscapes by incorporating native plants. Native plants are important for supporting native herbivorous insects, but it is unknown whether the native plants that are commercially available, typically cultivated varieties (cultivars) of a single genotype, are equally effective as food sources as the local, wild-type plants. We compared the hemipteran communities feeding on cultivars and wild-propagated plants for four species of native perennials commonly used as ornamentals. Of 65 hemipteran species collected, 35 exhibited a preference for some plant species over others, indicating a high degree of host-plant specialization. Moreover, the insect community associated with cultivars was distinct from the insect community associated with wild-type plants for each plant species, with three to four insect species accounting for most of the observed difference. Total insect abundance and insect biomass differed between cultivars and wild-propagated plants, but the direction of the difference changed over time and was not consistent among plant species. Species richness and a diversity index (the Q statistic) did not differ between cultivars and wild-type plants. These data suggest that abundance and diversity of hemipteran insects does not depend on the source of the plant material per se, but rather on the particular characteristics of cultivars that distinguish them from the wild type. wildlife value, biodiversity, hemiptera, plant–insect interaction, leafhopper As the population of the United States expands, more land is converted for residential use. As of 2012, approximately 70 million acres were classified as urban use, and 106 million acres were classified as rural residential use in the United States (Bigelow and Borchers 2017). Landscaping in residential areas typically does not reflect the natural plant community that existed in that area before development; in fact, urban areas are associated with a replacement of native flora by exotic species (McKinney 2006), many of which also invade surrounding areas (Gavier-Pizarro et al. 2010). Most of the exotics that replace natives in landscapes are purchased through the plant trade industry for their aesthetic value (Mack and Lonsdale 2001). There has been some speculation that this prevalence of non-native vegetation in residential landscapes could contribute to a loss of biodiversity in suburban areas (Tallamy 2004). The purported mechanism by which landscaping with exotic plants would lead to a loss of biodiversity has roots in plant–insect interaction theory. Most insect herbivores specialize on a narrow group of plants, usually plants in three or fewer families (Bernays and Graham 1988). One theory predicts this specialization is a result of coevolution between plants and insects. Plants produce a diverse array of secondary metabolites that act as feeding deterrents, and herbivorous insects develop specialized physiological adaptations for detoxifying those metabolites (Fraenkel 1959, Ehrlich and Raven 1964). When large areas of native vegetation are replaced with exotic vegetation, as is often the case in suburban landscapes, the exotic vegetation may contain secondary metabolites that are not present in the native vegetation, and hence native insects are not equipped to deal with these novel feeding deterrents. Because arthropods often make up the largest proportion of animal biomass in a given ecosystem, removing one of their primary food sources—native plants—could catastrophically disrupt food webs, with effects cascading up to higher trophic levels (Wilson 1987). Recent research has provided strong empirical evidence that some exotic plant species support less herbivorous insect biomass and less diverse herbivorous insect communities than native plants (Zuefle et al. 2008, Burghardt et al. 2010, Tallamy et al. 2010, Burghardt and Tallamy 2013). There is also evidence that the diversity of organisms in higher trophic levels, especially birds, is highly correlated with the abundance and diversity of insects in suburban habitats (Kim et al. 2007, Burghardt et al. 2009, Narango et al. 2017). The concern that a decline in insect biomass could cause a concomitant decline in overall biodiversity in suburban areas, together with a desire to improve the ecological functioning of landscapes, has spurred an interest in ‘gardening for wildlife’ by replacing exotics with native ornamental plants in landscapes. But are the native plants available commercially, typically cultivated varieties (cultivars) of a single genotype, equally effective as the local, wild-type plants in providing food for native herbivorous insects? There are several lines of evidence that suggest cultivars and wild-type plants could differ in their abilities to support herbivorous insects. Cultivars are often asexually-propagated, and insect diversity is known to correlate with the genetic diversity of their host plants (Wimp et al. 2004, Johnson et al. 2006). Cultivars are also selected for some characteristic that distinguishes them from the wild form. Pest resistance, altered growth habitat, changes in flower or leaf color, and sterility are common goals of plant breeders, and any of these could influence an insect’s ability to feed on the plant (e.g., Tenczar and Krischik 2007, Mphosi and Foster 2010). While many of these traits imply negative consequences for insect communities, it is important to note that some characteristics may actually improve a cultivar’s ecological value relative to the wild form. In the simplest case, selecting for hybrid vigor would make more plant biomass available to support more insects. We investigated whether these theoretical consequences of cultivar selection actually affect herbivorous insects in a garden setting. We chose several native herbaceous perennials that occur locally in natural areas near the study site and have cultivars available commercially. We determined whether the cultivars differed from plants grown from wild-collected seed in their ability to serve as a food source for native hemipterans (Auchenorrhyncha and Heteroptera), a highly abundant and speciose group associated with grasslands and plants growing in open, disturbed areas (Wallner et al. 2013). We measured hemipteran biomass, abundance, diversity, and community composition over the course of one growing season. Materials and Methods Study Site and Plant Material Plots were established in an open field at the Mimsie Lanier Center for Native Plant Studies at the State Botanical Garden of Georgia in Athens, Clarke County, Georgia. The Botanical Garden was an ideal study site for this research because it reflects a typical residential setting. It is a mixture of woodlands, lawns, and gardens located just on the outskirts of a medium-sized city. The field in which plots were established is located within the floodplain of the Oconee River and is characterized by periodic flooding and loamy, alluvial soils. There are open and forested areas, both upland and bottomland, immediately surrounding the site. All wild-type plants were grown from seed locally-collected from populations occurring in natural areas. Here, we define natural areas to mean open sites comprised of mostly native, early-successional vegetation; this includes relatively undisturbed sites, such as granitic outcrops, and sites with man-made disturbance, such as utility rights-of-way maintained by mowing. Seeds of Amsonia tabernaemontana were collected from a population at Currahee Mountain in Stephens County, Georgia. Seeds of Monarda fistulosa were collected from a natural area within the State Botanical Garden. Seeds of Coreopsis grandiflora, Oenothera fruticosa, and Schizachyrium scoparium were collected from the Rock and Shoals Natural Area, a granitic outcrop at the end of Rock and Shoals Dr. in Clarke County, Georgia. For all plant species, a few seeds were collected from many individuals (usually >50) in order to capture as much of the genetic variation within the population as possible. The common names and plant families of the plants used in the study are provided in Table 1. Table 1. Descriptions of plant species and cultivars used in this experiment Plant species Common name Cultivar Family Cultivar origin Difference from wild-type Amsonia tabernaemontana (Walter, Gentianales: Apocynaceae) Eastern bluestar ‘Blue Ice’ Apocynaceae Interspecific hybrida Longer bloom, darker flowers, compact form Coreopsis grandiflora (Hogg ex Sweet, Asterales: Asteraceae) Large-flowered tickseed ‘Tequila Sunrise’ Asteraceae Interspecific hybrid Variegated leaves, compact form Monarda fistulosa (Linnaeus, Lamiales: Lamiaceae) Wild bergamot ‘Claire Grace’ Lamiaceae Straight species Powdery mildew resistant, darker flowers Oenothera fruticosa (Linnaeus, Myrtales: Onagraceae) Southern Sundrops ‘Cold Crick’ Onagraceae Interspecific hybrida Sterile, compact form Schizachyrium scoparium ((Michaux) Nash, Poales: Poaceae) Little bluestem ‘Prairie Blues’b Poaceae Straight species Blue-green foliage turning wine-red in fall Plant species Common name Cultivar Family Cultivar origin Difference from wild-type Amsonia tabernaemontana (Walter, Gentianales: Apocynaceae) Eastern bluestar ‘Blue Ice’ Apocynaceae Interspecific hybrida Longer bloom, darker flowers, compact form Coreopsis grandiflora (Hogg ex Sweet, Asterales: Asteraceae) Large-flowered tickseed ‘Tequila Sunrise’ Asteraceae Interspecific hybrid Variegated leaves, compact form Monarda fistulosa (Linnaeus, Lamiales: Lamiaceae) Wild bergamot ‘Claire Grace’ Lamiaceae Straight species Powdery mildew resistant, darker flowers Oenothera fruticosa (Linnaeus, Myrtales: Onagraceae) Southern Sundrops ‘Cold Crick’ Onagraceae Interspecific hybrida Sterile, compact form Schizachyrium scoparium ((Michaux) Nash, Poales: Poaceae) Little bluestem ‘Prairie Blues’b Poaceae Straight species Blue-green foliage turning wine-red in fall aPurported hybrid. bPropagated from seed. View Large Table 1. Descriptions of plant species and cultivars used in this experiment Plant species Common name Cultivar Family Cultivar origin Difference from wild-type Amsonia tabernaemontana (Walter, Gentianales: Apocynaceae) Eastern bluestar ‘Blue Ice’ Apocynaceae Interspecific hybrida Longer bloom, darker flowers, compact form Coreopsis grandiflora (Hogg ex Sweet, Asterales: Asteraceae) Large-flowered tickseed ‘Tequila Sunrise’ Asteraceae Interspecific hybrid Variegated leaves, compact form Monarda fistulosa (Linnaeus, Lamiales: Lamiaceae) Wild bergamot ‘Claire Grace’ Lamiaceae Straight species Powdery mildew resistant, darker flowers Oenothera fruticosa (Linnaeus, Myrtales: Onagraceae) Southern Sundrops ‘Cold Crick’ Onagraceae Interspecific hybrida Sterile, compact form Schizachyrium scoparium ((Michaux) Nash, Poales: Poaceae) Little bluestem ‘Prairie Blues’b Poaceae Straight species Blue-green foliage turning wine-red in fall Plant species Common name Cultivar Family Cultivar origin Difference from wild-type Amsonia tabernaemontana (Walter, Gentianales: Apocynaceae) Eastern bluestar ‘Blue Ice’ Apocynaceae Interspecific hybrida Longer bloom, darker flowers, compact form Coreopsis grandiflora (Hogg ex Sweet, Asterales: Asteraceae) Large-flowered tickseed ‘Tequila Sunrise’ Asteraceae Interspecific hybrid Variegated leaves, compact form Monarda fistulosa (Linnaeus, Lamiales: Lamiaceae) Wild bergamot ‘Claire Grace’ Lamiaceae Straight species Powdery mildew resistant, darker flowers Oenothera fruticosa (Linnaeus, Myrtales: Onagraceae) Southern Sundrops ‘Cold Crick’ Onagraceae Interspecific hybrida Sterile, compact form Schizachyrium scoparium ((Michaux) Nash, Poales: Poaceae) Little bluestem ‘Prairie Blues’b Poaceae Straight species Blue-green foliage turning wine-red in fall aPurported hybrid. bPropagated from seed. View Large Four of the cultivars were purchased as liners from North Creek Nurseries in Landenberg, Pennsylvania. These were Amsonia ‘Blue Ice’, Coreopsis ‘Tequila Sunrise’, Monarda fistulosa ‘Claire Grace’, and Oenothera ‘Cold Crick’. All four cultivars were propagated vegetatively; i.e., they were clones from a single source. The fifth cultivar, Schizachyrium scoparium ‘Prairie Blues’, was donated by Hoffman Nursery in Bahama, NC. This cultivar was propagated by seed. The cultivars were chosen to represent both variety in their genetic origins and their traits of interest. For example, the cultivars of Coreopsis, Oenothera, and Amsonia are all interspecific hybrids or likely hybrids, whereas the cultivars of Monarda and Schizachyrium are selections of the straight species. Likewise, Schizachyrium and Coreopsis were selected for traits that involve a change in leaf chemistry (viz., leaf color), while Oenothera, Amsonia, and Monarda were selected for traits such as form, sterility, and disease resistance. We provide a full summary of the purported differences between the wild-type plants and cultivars (Table 1). Both cultivars and wild-type plants were planted in the field in April 2013. At that point, both wild-type plants and cultivars had root balls that filled square nursery pots approximately 3.5 inches wide and 3 inches deep. By the late summer of 2013, many plants were fully-grown and flowering. By the summer of 2014, just before data were collected, all the plants had reached flowering size and most plots had achieved 100% cover. Experimental Design The experiment followed a fully-randomized two-way analysis of variance (ANOVA) design (schematic diagram in Supplementary Appendix S1). The first factor was plant species and included five levels: Amsonia, Coreopsis, Monarda, Oenothera, and Schizachyrium. The second factor was plant source and included two levels: cultivar and wild-type. There were five replicates for each treatment, giving a total of 50 experimental units. Each experimental unit was a 2 × 2 m plot containing 16 plants evenly spaced, and plots were placed 1.5 m apart. The planting density per plot was chosen to be 16 individuals so that percent cover would be close to 100% by the time insects were collected. Plots were kept weed-free throughout the growing season and wood mulch was used in-between plots. Data Collection We collected insects from all the plant species on three dates in 2014: 8 July, 15 August, and 20 September. The collection was broken up into two consecutive days for each sample period because all 50 plots could not be sampled in 1 d. We only sampled on days that were sunny with low wind speed. Insects were vacuumed for 1 min from plots with a modified leaf vacuum (Stihl BG 55) following the design in Wilson et al. (1993). The period of 1 min was based on previous trials and chosen to maximize the number of insects vacuumed while minimizing the number of vacuumed insects escaping. The order in which the plots were sampled was randomized to reduce any systematic bias caused by insects that escaped the vacuum and moved to other plots. Sampling began at 11:00 a.m. and ended by 2:00 p.m. to coincide with peak xylem flow. The insects were killed with ethyl acetate, sorted into hemipterans and non-hemipterans, and stored dry in boxes. Representative specimens of each species were pinned for subsequent identification. A combination of family-specific regional keys (e.g., DeLong 1948 for leafhoppers) and genera-specific keys were used to identify insects to species level. All identifications were subsequently confirmed by a taxonomist (Dr. Rick Hoebeke, Natural History Museum, University of Georgia). We also categorized insects into feeding guilds based on whether they primarily feed on xylem, phloem, mesophyll, or seeds/fruits. A full list of the taxonomic references and resources for information about insects’ life histories can be found in Supplementary Appendix S7. We measured two response variables for each experimental unit (plot) at each sampling date: total dry biomass of hemipterans and abundance of each adult hemipteran species. Total dry biomass included both adult hemipterans and nymphs and was measured to the nearest 0.0001 g. For the count data, nymphs were not included because most could not be identified to species. From the abundances of each species, we constructed several other response variables for our statistical analyses. These were total overall abundance of hemipterans, species richness, and the Q-statistic. The Q-statistic is a diversity index that measures the interquartile slope of the species accumulation curve, excluding both low- and high-abundance species from the measure of diversity (Kempton and Taylor 1978). We chose the Q-statistic over more common indices such as Simpson’s Index or Fisher’s Alpha because many of these indices are highly influenced either by low-abundance species or by dominant species. In the context of this experiment, infrequently-collected insects more likely represent ‘tourists’ (Gaston 1996) rather than rare species that are actually associated with a given plant, and hence should carry less weight instead of more. The Q-statistic provides a good compromise for measuring diversity when it is desirable to reduce the influence of both low- and high-abundance species. Standardization by Plant Biomass Because plant species, and sometimes cultivars and wild-type plants, differed in size, we standardized total insect biomass and total abundance by the dry weight of the plants they were collected from. We could only measure plant dry weight in September, however, because it involved destructively sampling the plants. Within 2 wk following the September sampling, we harvested all of the above-ground plant material from each plot and stored it in large paper bags. We dried all of the plant material in a forced-air drying oven at 60°C for 7 d. After drying, we recorded the total dry weight of each plot to the nearest 0.1 kg. The dry weights were used to transform the September insect data as total abundance or total insect biomass per kilogram plant biomass. Statistical Analyses We used repeated measures two-way ANOVA to analyze total abundance and total dry biomass of insects. Several variants of repeated measures ANOVA exist. We used the variant that performs multivariate analysis of variance (MANOVA) on the differences between adjacent time periods (a.k.a. profile analysis) to test for the effects of plant species, plant source, time, and their interactions. We used the function manova() in R (R Core Team 2013). The standardized abundance and insect biomass data from September were analyzed with univariate two-way ANOVA using function aov(). Rather than analyze species richness and the Q-statistic by sampling date (i.e., with repeated measures ANOVA), we pooled data over all three sampling periods to compare diversity among treatments for the entire growing season. We analyzed richness and the Q-statistic with two-way ANOVA using function aov() in R (R Core Team 2013). The normality assumption of ANOVA was tested with function shapiro.test() and the homogeneity of variances assumption was tested with function bartlett.test(). To visually compare species richness among treatments, we constructed a sample-based rarefaction curve using function specaccum() in package ‘vegan’ in R (Oksanen et al. 2013). Although all the experimental units were the same size and sampling effort was equal among treatments, we rescaled the x-axis by number of individuals to account for differences in insect density among treatments (Gotelli and Colwell 2001). We also used the pooled count data to test whether the insect communities differed among treatments. Although distance-based analyses are often used for addressing community-level questions, they suffer from major problems (Anderson 2001, Warton et al. 2012). We instead used multivariate analyses based on generalized linear models (GLMs) to model the counts of each insect species directly. In particular, we fit the abundance data for each insect species to separate negative binomial models using function manyglm() in package ‘mvabund’ in R (Wang et al. 2014). We specified the identity matrix as the correlation structure for our multivariate tests and used the likelihood ratio test for Analysis of Deviance. The function provides a multivariate test for the entire community and univariate tests for each insect species, with P-values for the univariate tests corrected for multiple comparisons using a resampling-based procedure. We used 1,000 permutations to obtain P-values. Results Overall, we collected almost 12,000 adult hemipterans representing 130 different species. Exactly half of the species had abundances less than or equal to five in the entire dataset, so we excluded those species from all analyses. Those species were likely ‘transients’ or ‘tourists’, and not actually feeding on the plants they were collected from. Because arthropods are highly mobile, it is not surprising that such a large proportion of the species collected would be transients (Pimentel and Wheeler (1973) for a similar result). Although including these species in the analyses would make little difference for response variables such as total abundance, they would be highly influential for diversity measures such as species richness. Even the Q-statistic, which is meant to exclude both infrequent and highly abundant species, can be biased when the majority of the species are singletons (Magurran 2004). We provide a list of all the insect species excluded from the analyses and their abundances (Supplementary Appendix S2). Only one species of hemipteran was found feeding on the Amsonia wild-type plants and cultivars, and only nine individuals of this species were collected over the entire growing season, so the Amsonia wild-type and cultivar treatments were excluded from all analyses. The species found feeding on Amsonia was the broad-headed sharpshooter, Oncometopia orbona. Total Abundance and Insect Biomass Repeated measures ANOVA indicated a three-way interaction among plant species, plant source, and time for both total abundance of insects and total insect biomass. We broke up both analyses at each level of plant species and plotted these results as a function of time (Fig. 1a and b). For most plant species, there were significant interactions between plant source and time. Hence, we compared abundance or biomass of cultivars and wild-type plants for each plant species at each date. P-values were corrected for multiple comparisons using the Bonferroni method. The Bonferroni method is the most conservative multiple comparison adjustment so we displayed significant differences both at an experiment-wise α = 0.05 (P-value < 0.0042) and at α = 0.10 (P-value < 0.0083), denoted by three asterisks and one asterisk, respectively. Fig. 1. View largeDownload slide Plots of (a) total abundance of adult hemipterans and (b) total hemipteran biomass for cultivars and wild-type plants of each plant species at three sampling dates. Asterisks indicate significant differences in means between cultivars and wild-type plants for a given plant species on a given date. Three asterisks represent differences that were significant after a Bonferroni correction for multiple comparisons with an experiment-wise α = 0.05 (P-value < 0.0042). One asterisk represents differences that were significant after a Bonferroni correction with an experiment-wise α = 0.10 (P-value < 0.0083). Error bars are SD. Fig. 1. View largeDownload slide Plots of (a) total abundance of adult hemipterans and (b) total hemipteran biomass for cultivars and wild-type plants of each plant species at three sampling dates. Asterisks indicate significant differences in means between cultivars and wild-type plants for a given plant species on a given date. Three asterisks represent differences that were significant after a Bonferroni correction for multiple comparisons with an experiment-wise α = 0.05 (P-value < 0.0042). One asterisk represents differences that were significant after a Bonferroni correction with an experiment-wise α = 0.10 (P-value < 0.0083). Error bars are SD. Generally, there was close correspondence between insect abundance and insect biomass. The most notable exception was Monarda. The Monarda wild-type plants consistently had two to five times higher insect abundance than the cultivars, but insect biomass was not significantly different between cultivars and wild-type plants at any sampling date. Another notable pattern was that some treatments increased in abundance and biomass over the growing season, whereas others decreased (and some changed little). In particular, insect abundance and biomass increased on the Coreopsis and Schizachyrium cultivars over the growing season, but decreased on the Coreopsis and Oenothera wild-type plants. For the Oenothera cultivar and wild-type comparison, abundance and biomass were three times higher on wild-type plants early in the season, but there were no differences by the end of the season. For Schizachyrium, there were no differences in abundance or biomass early in the season, but insect abundance and biomass were approximately 2 and 3.5 times higher, respectively, on the cultivar by the last sampling date. The Coreopsis had perhaps the most striking pattern. Abundance and biomass were approximately 1.5 times higher on the wild-type plants on the first sampling date, but were three to four times higher on the cultivar by the end of the season. However, this difference in pattern is most likely due to the fact that many of the Coreopsis wild-type plants had died by the last sampling date. Species Diversity Two-way ANOVA of species richness indicated no significant interactions between plant species and plant source, but a significant overall effect of plant species (plant source: F1,32 = 2.01, P = 0.17; plant species: F3,32 = 7.26, P < 0.001; interaction: F3,32 = 2.41, P = 0.085). The results of two-way ANOVA of the Q-statistic were similar (plant source: F1,32 = 0.01, P = 0.92; plant species: F3,32 = 9.90, P < 0.001; interaction: F3,32 = 1.44, P = 0.25). That is, there was no indication of differences in richness or the Q-statistic between wild-type plants and cultivars for any plant species. Mean richness and mean Q-statistic were significantly higher for Oenothera than all other plant species after correcting for multiple comparisons with Tukey’s HSD test (Fig. 2a and b, respectively). Oenothera had on average five to six more insect species than the other plants. There were no differences in richness or the Q-statistic among the other plant species. Fig. 2. View largeDownload slide Plots of (a) species richness and (b) Q-statistic for cultivars and wild-type plants of each plant species pooled over three sampling dates. Species richness and the Q-statistic were used as measures of species diversity of adult hemipterans for each treatment. Different letters denote means of plants species that were significantly different at α = 0.05 after correcting for multiple comparisons with Tukey’s HSD test. Oenothera had higher average species richness and values of the Q-statistic than the other three plant species. There was no evidence for a difference in richness or the Q-statistic between wild-type plants and cultivars for any plant species. Error bars are SD. Fig. 2. View largeDownload slide Plots of (a) species richness and (b) Q-statistic for cultivars and wild-type plants of each plant species pooled over three sampling dates. Species richness and the Q-statistic were used as measures of species diversity of adult hemipterans for each treatment. Different letters denote means of plants species that were significantly different at α = 0.05 after correcting for multiple comparisons with Tukey’s HSD test. Oenothera had higher average species richness and values of the Q-statistic than the other three plant species. There was no evidence for a difference in richness or the Q-statistic between wild-type plants and cultivars for any plant species. Error bars are SD. The rarefaction curve of species richness generally agreed with the results of ANOVA with the exception of the Monarda cultivar and wild-type plants (Supplementary Appendix S3). For example, the curves for the Oenothera cultivar and wild-type are above all the other treatments, indicating that species richness is higher on Oenothera than the other plant species. Also, the wild-type and cultivar curves overlap for Coreopsis, Oenothera, and Schizachyrium, indicating that there was no effect of plant source. However, for Monarda, the curve for the cultivar lies above the curve for the wild-type. This discrepancy is due to differences in the density of insects on the Monarda wild-type plants versus the cultivars. The pooled abundance for the Monarda wild-type treatment was approximately four times higher than the cultivar treatment, despite that the area sampled for each treatment was the same. The Monarda wild-type treatment actually had higher absolute richness, and hence would explain why ANOVA indicated no difference in average richness between the cultivar and wild-type. Insect Community The multivariate GLM test for the entire insect community indicated a significant interaction between plant species and plant source in the full model (P ≈ 0.001). Although the interaction was significant for the multivariate test, this does not imply that the interaction would be significant for each univariate test. We also provide the results of univariate tests under the full model and plots of the abundances of each species (Supplementary Appendices S5 and S7, respectively). There was a significant effect of plant species (and no interaction) for over half of the insect species; i.e., most of the insect species prefer to feed on some plant species over others. These results fit well with plant–insect interaction theory, but are not the main interest of this experiment. Hence, we only tested the effect of plant source at each level of plant species in our follow-up analyses. There was a significant effect of plant source for every plant species (Coreopsis: P ≈ 0.003; Monarda: P ≈ 0.01; Oenothera: P ≈ 0.006; Schizachyrium: P ≈ 0.024). That is, there was strong evidence that the hemipteran community differed between cultivars and wild-type plants for every plant species. After correcting P values for multiple comparisons, univariate tests indicated that three to four insect species were responsible for the differences in insect communities between cultivars and wild-type plants (species with asterisks in Figs. 3–5). Fig. 3. View largeDownload slide Abundances of phloem-feeding insects that showed the strongest differences between cultivars and wild-type plants for Coreopsis, Monarda, Oenothera, and Schizachyrium. Asterisks indicate plant species where there was a significant difference between the cultivar and wild-type at α = 0.05 after correcting for multiple comparisons. Error bars are SD. Fig. 3. View largeDownload slide Abundances of phloem-feeding insects that showed the strongest differences between cultivars and wild-type plants for Coreopsis, Monarda, Oenothera, and Schizachyrium. Asterisks indicate plant species where there was a significant difference between the cultivar and wild-type at α = 0.05 after correcting for multiple comparisons. Error bars are SD. Fig. 4. View largeDownload slide Abundances of mesophyll-feeding insects that showed the strongest differences between cultivars and wild-type plants for Coreopsis, Monarda, Oenothera, and Schizachyrium. Asterisks indicate plant species where there was a significant difference between the cultivar and wild-type at α = 0.05 after correcting for multiple comparisons. Error bars are SD. Fig. 4. View largeDownload slide Abundances of mesophyll-feeding insects that showed the strongest differences between cultivars and wild-type plants for Coreopsis, Monarda, Oenothera, and Schizachyrium. Asterisks indicate plant species where there was a significant difference between the cultivar and wild-type at α = 0.05 after correcting for multiple comparisons. Error bars are SD. Fig. 5. View largeDownload slide Abundances of seed- and fruit-feeding insects that showed the strongest differences between cultivars and wild-type plants for Coreopsis, Monarda, Oenothera, and Schizachyrium. Note that Lygus feeds on both fruits and mesophyll. Asterisks indicate plant species where there was a significant difference between the cultivar and wild-type at α = 0.05 after correcting for multiple comparisons. Error bars are SD. Fig. 5. View largeDownload slide Abundances of seed- and fruit-feeding insects that showed the strongest differences between cultivars and wild-type plants for Coreopsis, Monarda, Oenothera, and Schizachyrium. Note that Lygus feeds on both fruits and mesophyll. Asterisks indicate plant species where there was a significant difference between the cultivar and wild-type at α = 0.05 after correcting for multiple comparisons. Error bars are SD. We assigned each insect we collected to a feeding guild based on information about its life history (Supplementary Appendix S7, which also contains food plant records for each insect). Insects were classified as feeding on xylem, phloem, leaf/stem mesophyll, or fruits/seeds, and an insect species could belong to multiple guilds. Of the many xylem-feeding insect species collected, few showed a preference between cultivars and wild-type plants, and there were none with significantly different abundances after adjusting P values. However, there were examples of insect species that preferred either the cultivar or the wild-type plant for each of the other three feeding guilds. For phloem-feeding and mesophyll-feeding insects, some insect species appeared to prefer the wild-type plants, while others appeared to prefer the cultivars (Figs. 3 and 4). In most cases, the differences in abundance were quite large. For example, the abundance of Empoasca sp. was over eight times higher on the Monarda wild-type than the cultivar, and Sixeonotus unicolor was almost six times higher on the Coreopsis cultivar than the wild-type. Differences for the other insect species ranged from approximately four-fold to almost eightfold. Although the abundances of many of these insects were relatively high, a few showed statistically significant differences between wild-types and cultivars despite having low abundances. For example, the average abundances for Blissus leucopterous on the Schizachyrium wild-type and Ceratocapsus punctulatus on the Oenothera wild-type were 6.2 and 3.6, respectively, yet both were significantly higher than their corresponding cultivars. For seed-feeding insects, there were five instances in which insect abundance differed significantly between cultivar and wild-type, and, in each case, it was the wild-type plants that were preferred (Fig. 5). The two species for which differences were observed were Coreopsis and Oenothera. Oenothera ‘Cold Crick’ is marketed as a sterile hybrid (Table 1), while Coreopsis ‘Tequila Sunrise’ is apparently sterile, as no seeds were observed during the study. Wild-type plants of both species were fertile. Like with the other feeding guilds, the observed differences in abundances of seed-feeding insects were very large. Neopamera bilobata was over 17 times more abundant on the Oenothera wild-type than on the cultivar. Lygus lineolaris was four times more abundant on the Coreopsis wild-type than the cultivar and over 10 times more abundant on the Oenothera wild-type than the cultivar. The abundances of the other two seed-feeding insects were relatively low. Xyonysius californicus and Homaemus proteus had average abundances of 3.0 and 1.8, respectively, on the Coreopsis wild-type, but no individuals of either species were found on the Coreopsis cultivar. Standardized Data We observed significant differences in total insect abundance and total insect biomass between cultivars and wild-type plants for several plant species in September. Because this was our last sampling date, we were able to harvest all of the plant material and standardize our insect counts by the amount of plant biomass in plots to determine whether the observed differences were due to an inherent effect of plant source or simply the amount of plant biomass available. Most of the Coreopsis individuals in the wild-type treatment were dead by this point, so we excluded both the Coreopsis wild-type and cultivar treatments from the analyses. The results of two-way ANOVAs of standardized abundance and standardized insect biomass revealed that standardization actually accentuated the observed differences between cultivars and wild-type plants (Supplementary Appendix S5). For example, for Monarda the raw abundance was approximately four times higher on wild-type plants than cultivars in September, but the standardized abundance was almost six times higher on wild-type plants than the cultivars. For Schizachyrium, the raw abundance was approximately two times higher on the cultivars than the wild-type plants in September, but the standardized abundance was almost four times higher on cultivars. The same general result was true for raw insect biomass versus standardized insect biomass. These results indicate plant source, and not differences in the amount of plant biomass available to insects, better explain the number of insects (and insect biomass) collected from plants. Discussion Different Hemipteran Species Associate With Different Plant Species Given that an estimated 90% or more of phytophagous insect species are host-plant specialists (Bernays and Graham 1988), we expected to find different insect communities associated with each plant species; here, we refer to host-plant specialists as insect species that feed on plants in three or fewer families, and generalists as species that feed on plants in more than three families. Our results agreed with this expectation, providing strong evidence that different hemipteran species are associated with different plant species. Although we suggest that the differences in the insect communities are a result of host-plant specialization among the insects we captured, note that we did not observe the feeding behavior of the insects directly. We measured the abundances of insects on different species of plants, which implies, but does not directly measure, which plants the insects were feeding on. Therefore, our data give an indication of host plant preferences for individual insect species, but these preferences do not preclude the possibility of the insects having much broader host ranges. In order to actually classify insects as specialists or generalists empirically from the data, direct measures of feeding behavior (and a larger suite of plants to test) would be necessary. Unfortunately, we were also unable to use host plants records from the literature to classify insects as specialists or generalists because records for most species were ambiguous, insufficient, or nonexistent (e.g., in Supplementary Appendix S7, the best available food plant records for several species was ‘herbaceous plants’). Although we were unable to classify insects as specialists and generalists, the patterns in their abundances clearly indicate that many insect species prefer certain plant species to others. The actual number of insect species observed to prefer some plants over others was 35 of 65 (Supplementary Appendix S4). Although this suggests that the majority of insects discriminate in their choice of food plants, this proportion is well below the expected estimate of 90% (if most insects are specialists). One possible reason for this discrepancy is that our sample size was small, so power may not have been sufficient to detect differences for some of the other insect species, especially those species with low overall abundances. Another possibility is that insect species were included in the analyses that were not actually feeding on the plants; these species would more likely show a random pattern of host-plant association rather than being found on one particular plant species. We attempted to reduce the number of these ‘tourist’ species by excluding all species with an overall abundance of five or less. We chose this conservative threshold to reduce the possibility of excluding from the analyses any rare species that were actually feeding on the plants. Another explanation is that the proportion of specialists among hemipterans is not as high as the proportion of specialists among herbivorous insects in general. Many of the hemipteran species we collected feed on xylem. There may be less host-plant specialization among those species because xylem generally contains fewer secondary metabolites that might act as feeding deterrents (Peck and Thompson 2008). In contrast, insect feeding guilds such as leafminers and gall-makers, and taxonomic groups such as lepidopterans, come into direct contact with cellular contents during the process of feeding, exposing them to secondary metabolites that potentially drive more specialized feeding behavior. Effect of Plant Source on Feeding Guilds Xylem-Feeding Insects Our main interest in this study was not whether different insect species are associated with different species of ornamental plants, but rather does plant source affect the insect community associated with a particular plant. For each of the plant species we tested, we found strong evidence that the hemipteran community differed between cultivars and wild-propagated plants. However, xylem-feeding insects contributed least to the observed differences in the insect communities, as none had large enough differences in abundance to meet the adjusted level of significance in the individual tests. Again, xylem-feeding insects likely experience fewer secondary metabolites, and this may explain why they were less discriminating between cultivars and wild-types. We also note that the total number of insect species that contributed most strongly to differences in the insect communities of wild-type plants and cultivars was relatively small (12 out of 65). Because hemipterans have haustellate mouthparts, they may be less exposed to plant secondary metabolites than are insects with mandibulate mouthparts, such as lepidopterans and orthopterans. Consequently, hemipterans may be less sensitive to changes in their food source, and the observed proportion of species driving community differences between cultivars and wild-type plants may be conservative relative to the pattern exhibited by other insect groups. Phloem-Feeding Insects Four phloem-feeding insect species contributed strongly to the observed differences in the insect community associated with cultivars versus wild-type plants. The species included two leafhoppers (Cicadellidae: Megophthalminae and Deltocephalinae), one fulgoroid (Derbidae), and one seed bug (Lygaeidae). The seed bug, Blissus leucopterous, feeds on the phloem of grasses rather than the seeds, as do all other members of the genus (Slater and Baranowski 1990). For three of the species, the wild-type plants were preferred. However, Balclutha neglecta exhibited a strong preference for the Schizachyrium cultivar. When attempting to explain why certain insect species preferred either the cultivar or wild-type, it is important to note that the individual tests for each species were performed on the pooled dataset, and therefore do not incorporate any information about the seasonal abundances of species. B. neglecta did not become abundant until the last sampling date. By this time, most of the wild-type Schizachyrium were starting to enter dormancy, whereas the cultivars had a slightly delayed phenology. This difference in phenology of the host-plants in combination with late-season peak abundance of B. neglecta could explain the observed preference for the cultivar. For the other three phloem-feeding species, it is not clear what differences between cultivars and wild types might be driving the observed preference for wild types. Phloem-feeding insects are known to respond strongly to the overall nitrogen content of the phloem as well as to particular amino acids that act as feeding stimulants (Cook and Denno 1994). However, we did not measure the chemical composition of the phloem (or any other parameters influencing host-plant quality), so any causal mechanisms are entirely speculative. Mesophyll-Feeding Insects Among the mesophyll-feeding insects exhibiting the largest differences in abundance between cultivars and wild-types, three were plant bugs (Miridae) and one was a leafhopper (Cicadellidae: Typhlocybinae). Both of these groups use the ‘lacerate and flush’ feeding strategy, in which they use their stylets to pierce the leaf tissue, lacerate the cells, and imbibe the cell contents (Wheeler 2001). Although mesophyll contains more nutrients than xylem or phloem, lacerating the cells also exposes the insect to whatever secondary metabolites that plant may produce as feeding deterrents, suggesting that insects with this feeding strategy would be particularly sensitive to differences in leaf chemistry. Because neither plant bugs nor typhlocybine leafhoppers were overly represented among the species showing preferences between cultivars and wild-type plants, this suggests either that the cultivars and wild-type plants have similar leaf chemistry or that other characteristics of the plants are playing an equally important role in influencing the host-plant preferences of insects. As with the phloem-feeding species, three species preferred the wild-type plants and only one preferred the cultivar (though there was weak evidence that Ceratocapsus punctulatus preferred the Coreopsis cultivar to the wild-type). The species preferring the cultivar, Sixeonotus unicolor, has been recorded to feed on ‘garden coreopsis’ (Wheeler 2001). It is not clear from the host plant record whether this refers to a cultivated variety of C. grandiflora or simply a Coreopsis species commonly used for ornamental purposes. Note also that none of the four mesophyll-feeding species were collected from Schizachyrium (except for two individuals of C. punctulatus, which were likely tourists). A probable explanation is that the lacerate and flush feeding strategy is not compatible with the particular defenses used by grasses. Grasses typically contain lignin and high levels of silica targeted at mechanically abrading the mouthparts of chewing insects (Cook and Denno 1994). Although hemipterans are not chewing insects, lignin and silica likely defend against the lacerate component of the feeding strategy with similar efficacy. Seed- and Fruit-Feeding Insects The seed- and fruit-feeding insects included two seed bugs (Lygaeidae and Rhyparochromidae), one shield-backed bug (Scutelleridae), and one plant bug (Miridae). The plant bug, Lygus lineolaris, can also feed on leaf and stem mesophyll, but appears to prefer the nectar, pollen, and immature fruits from flowers of herbaceous plants (Wheeler 2001). It is considered a serious pest for some agricultural crops and has an unusually broad host range, having been recorded from more than 300 plant species in many different families (Young 1986). In contrast to phloem- and mesophyll-feeders, the seed-feeding insects observed to have the largest differences in abundance between cultivars and wild-types consistently preferred the wild-type plants to the cultivars. Moreover, the only plant species in which differences were observed were species where the cultivar was sterile. That is, the seed-feeding insects we collected appeared to prefer the Coreopsis and Oenothera wild-types over the cultivars because the cultivars were sterile, whereas we observed no differences between Monarda and Schizachyrium cultivars and wild-types, presumably because the cultivars produced viable seeds. The seed bugs we collected were species that climb plants to feed on seeds, which is a less common strategy among seed bugs than feeding on seeds that fall to the ground (Slater and Baranowski 1990). Our sampling method would not have collected ground-dwelling insects very effectively, so we may be underestimating the actual number of species of seed-feeding insects that sterile cultivars fail to support. In addition to finding no difference in the numbers of Xyonysius californicus and L. lineolaris on Monarda cultivars and wild-types (Fig. 5), we collected several specialist seed-feeding species on both Monarda and Schizachyrium that did not discriminate between the cultivars and wild-types. For example, the scentless plant bug Arhyssus nigristernum (Rhopalidae), a specialist on Monarda and other mints (Schaefer and Chopra 1982), and the seed bug Paromius longulus (Rhyparochromidae), a specialist on grasses (Slater and Baranowski 1990), were found in similar numbers on cultivars and wild-types (Supplementary Appendix S6). It is clear why seed-feeding insects would be less abundant on sterile plants, but an additional point to note is that both of the sterile cultivars were hybrids, whereas both fertile cultivars were selections of the straight species. A common mechanism by which hybrid plants become sterile is crossing two species with an even, but different ploidy level to produce offspring with an odd ploidy level. Although we did not determine the chromosome number of our plants, this or another genetic mechanism is likely causing the sterility observed in the Coreopsis and Oenothera cultivars. We originally speculated that the genetic diversity of the host plants may influence the insects that feed on those plants, and while these results do not necessarily imply this is the case, the results do support the idea that the genetic origin of cultivars is important. Our conclusions regarding plant sterility and seed-feeding insects are analogous to the results obtained by others who compared cultivars and wild-type plants in terms of their abilities to support native insects, though their research focused exclusively on pollinators. Comba et al. (1999) and Corbet et al. (2001) observed that cultivars with altered flower morphology were visited by fewer pollinators or fewer species of pollinators than the wild-type plants. In some cases, the change in morphology reduced nectar secretion. For the cases in which nectar secretion was unaffected, the authors speculated that reduced accessibility caused the observed reduction in pollinator visits. In either case, fewer insects were supported by cultivars because the insect’s food resource was unavailable. Similarly, we found fewer seed-feeding insects on sterile cultivars because their food resource was absent. The codependency between insects and flowers for plant reproduction has long been recognized (Sprengel 1793). These mutualistic relationships are a result of a long history of coevolution. Our results provide additional evidence that selecting for traits that alter the reproductive biology of the plants, and hence perturb the result of tens-of-thousands or millions of years of increasing specialization, has a strong negative impact on the insects that utilize those plants. Insect Total Abundance, Biomass, and Diversity In addition to determining whether there were differences in the insect community, we wanted to know whether cultivars and wild-type plants differed for more general parameters, such as total insect abundance, biomass, and diversity. The most prominent trend apparent in the abundance and biomass data is that, for the most part, differences between cultivars and wild-types were not consistent across sampling date. Many of the patterns can be explained by qualitative observations regarding phenology and plant health combined with close examination of trends for individual insect species. For example, more insects and more insect biomass were collected from the Coreopsis wild-type on the first sampling date, but insect abundance and biomass were much higher on the cultivar by the last sampling date. This pattern can be explained by the fact that very few individuals of the Coreopsis wild-type were alive by the end of the growing season. Some individuals of the Coreopsis cultivar also died during the experiment, but more were alive on the last sampling date than had died. An obvious question is: did the higher feeding pressure on the Coreopsis wild-type plants early in the year cause their decline? While it is certainly possible, Coreopsis grandiflora is a short-lived perennial and most likely had reached the end of its life cycle. Another example was the sharp decline in insect abundance and biomass on the Oenothera wild-type plants over the growing season. Neopamera bilobata, a seed-feeding insect, was more abundant on the Oenothera wild-type than the cultivar and made up a large proportion of both the total abundance of insects and the total insect biomass collected from Oenothera wild-types. Oenothera blooms in early May and sets seed shortly after, with most seeds being dispersed by late summer. Although it is not apparent in the pooled data, N. bilobata was especially abundant on the first sampling date and declined substantially on each subsequent sampling, suggesting its population size was responding to the phenology of the Oenothera wild-types. There was no evidence for a difference in hemipteran diversity between cultivars and wild-types for any of the plant species we tested (Fig. 2). Given the other variables we measured, this is not surprising. The differences we observed in total abundance and total biomass were largely due to single, numerically dominant species. Because species richness does not measure the dominance or evenness of the community, single species that have a large effect on parameters such as total abundance and biomass would not have an undue influence on diversity. Also, when we investigated which insect species were contributing to the differences we observed in the overall community, we found that some species preferred cultivars and others preferred wild-type plants (Figs. 3–5). Together, the overall differences in the insect community and the lack of difference in diversity suggest that cultivars and wild-types support a similar number of insect species, but the particular species are different. Species richness is often criticized as a poor measure of diversity (Magurran 2004). To confirm that other methods would give similar results, we also used individual-based rarefaction (Hurlbert’s (1971) formulation) and a diversity index called the Q-statistic to assess diversity (Kempton et al. 1978). Like species richness, the Q-statistic is robust to the presence of over-dominant species because it excludes both the most and least abundant species from the measure. Unlike species richness, however, it incorporates information about the relative abundances of insects. Both the Q-statistic and rarefaction resulted in the same overall conclusion as species richness. Unanswered Questions and Future Directions Future research should not only address the differences in the insect communities associated with cultivars and wild-type plants, but also what characteristics of the plants are driving these differences. Chemical defenses, such as secondary metabolites, play a prominent role in deterring herbivory (Fraenkel 1959). Besides chemical defenses, plants also produce numerous physical defenses, including glandular and hooked trichomes. Rather than feeding deterrents, insects could have instead been responding to differences in feeding stimulants. Hemipterans are known to prefer host plants with higher concentrations of amino acids in the phloem, and specific amino acids sometimes act as feeding stimulants (Cook and Denno 1994). Because hemipterans in particular are known to respond strongly to nitrogen content, this aspect of host plant quality may be equally important as more frequently cited drivers of host-plant specialization, such as secondary metabolite chemistry, and should be investigated in any future research that aims to determine why herbivorous insects discriminate between cultivars and wild-type plants. All of the plants we used in this experiment were outbreeding species (Carman and Hatch 1982, Crawford and Smith 1984, Cruden et al. 1984, Godfrey and Johnson 2014), and so should contain much higher genetic diversity than the cultivars propagated asexually. Although this research did not investigate the question directly, we mentioned that the amount of genetic diversity contained in cultivars versus wild-type plants propagated from seed could be driving differences in the insect communities (see Johnson and Agrawal (2005), Johnson and Agrawal (2007), and Johnson (2008) for examples of positive correlation between host plant genetic diversity and variation in the associated arthropod community). However, our results did not support this hypothesis. We found different hemipteran communities associated with cultivars and wild-type plants, but the diversity of those communities did not differ. These results suggest that plants of the same species can support different herbivorous insects when their genetic origins differ, but that host plant genetic diversity has a much smaller relative effect (or no effect) on the diversity of the insect community. Although cultivars and wild-type plants of every plant species we tested differed in terms of their associated hemipteran communities, they did not differ in hemipteran species diversity. In addition, the differences in total hemipteran biomass and abundance we observed appeared to be related to single species of insect that preferred the wild-type over the cultivar (or vice versa), and these differences were not consistent over the growing season. These results suggest hemipteran abundance and diversity does not depend on the source of the plant material per se, but rather on the particular characteristics of the cultivar that distinguish it from the wild type. Certain characteristics are likely detrimental for wildlife, such as altering the reproductive biology of the plants by changing flower morphology or selecting for sterility. Other characteristics, perhaps nitrogen utilization or plant architecture, may actually benefit wildlife, and identifying these characteristics should be a priority for future research. This knowledge could enable plant breeders to select varieties that potentially outperform the wild-type plants in terms of their abilities to support wildlife. Possibly the best way to improve landscapes for wildlife is to include more plant species, regardless of whether they are cultivars or wild-type plants propagated from seed. We observed differences in abundance for more insect species when we compared different plant species than we did when we compared cultivars and wild-types within a plant species. These results fit both with the expectations of plant–insect interaction theory and with empirical evidence that plant species diversity drives arthropod diversity. Our results also suggest that planting unrelated species will do more to enhance insect diversity than will planting species within one family or within a few genera. Supplementary Data Supplementary material is available at Environmental Entomology online. Acknowledgments Funding for this research was provided by the Georgia Native Plant Society and the Garden Club of America. Cultivars of Schizachyrium were donated by Hoffman Nursery in Bahama, NC. Dr. Rick Hoebeke, curator of the Natural History Museum at the University of Georgia, confirmed the identifications of the insects. Thanks also to all the staff at the State Botanical Garden of Georgia, especially Heather Alley, whose advice and support made this research possible. References Cited Anderson , M. J . 2001 . A new method for non-parametric multivariate analysis of variance . Austral. Ecol . 26 : 32 – 46 . Bernays , E. , and M. Graham . 1988 . On the evolution of host specificity in phytophagous arthropods . Ecology . 69 : 886 – 892 . Google Scholar CrossRef Search ADS Bigelow , D. P. , and A. Borchers . 2017 . Major uses of land in the United States, 2012. EIB-179 . U.S. Department of Agriculture, Economic Research Service, Washington, DC . Burghardt , K. T. , and D. W. Tallamy . 2013 . Plant origin asymmetrically impacts feeding guilds and life stages driving community structure of herbivorous arthropods . Divers. Distrib . 19 : 1553 – 1565 . Google Scholar CrossRef Search ADS Burghardt , K. T. , D. W. Tallamy , and W. Gregory Shriver . 2009 . Impact of native plants on bird and butterfly biodiversity in suburban landscapes . Conserv. Biol . 23 : 219 – 224 . Google Scholar CrossRef Search ADS PubMed Burghardt , K. T. , D. W. Tallamy , C. Philips , and K. J. Shropshire . 2010 . Non-native plants reduce abundance, richness, and host specialization in lepidopteran communities . Ecosphere . 1 : 1 – 22 . Google Scholar CrossRef Search ADS Carman , J. G. , and S. L. Hatch . 1982 . Aposporous apomixis in schizachyrium (Poaceae, Andropogoneae) . Crop Sci . 22 : 1252 – 1255 . Google Scholar CrossRef Search ADS Comba , L. , S. A. Corbet , A. Barron , A. Bird , S. Collinge , N. Miyazaki , and M. Powell . 1999 . Garden flowers: insect visits and the floral reward of horticulturally-modified variants . Ann. Bot.-London . 83 : 73 – 86 . Google Scholar CrossRef Search ADS Cook , A. G. , and R. F. Denno . 1994 . Planthopper/plant interactions: feeding behavior, plant nutrition, plant defense, and host plant specialization , pp. 114 – 139 . In R. F. Denno and T. J. Perfect (eds.), Planthoppers: their ecology and management . Chapman and Hall , New York, NY . Corbet , S. A. , J. Bee , K. Dasmahapatra , S. Gale , E. Gorringe , B. La Ferla , T. Moorhouse , A. Trevail , Y. Van Bergen , and M. Vorontsova . 2001 . Native or exotic? Double or single? Evaluating plants for pollinator-friendly gardens . Ann. Bot.-London . 87 : 219 – 232 . Google Scholar CrossRef Search ADS Crawford , D. J. , and E. B. Smith . 1984 . Allozyme divergence and intraspecific variation in Coreopsis grandiflora (Compositae) . Syst. Botany . 9 : 219 – 225 . Google Scholar CrossRef Search ADS Cruden , R. W. , L. Hermanutz , and J. Shuttleworth . 1984 . The pollination biology and breeding system of Monarda fistulosa (Labiatae) . Oecologia . 64 : 104 – 110 . Google Scholar CrossRef Search ADS PubMed DeLong , D. M . 1948 . The leaf-hoppers, or Cicadellidae, of Illinois (Eurymelinae-Balcluthinae) . Ill. Nat. Hist. Surv. Bull . 24 : 97 – 376 . Ehrlich , P. R. , and P. H. Raven . 1964 . Butterflies and plants: a study in coevolution . Evolution . 18 : 586 – 608 . Google Scholar CrossRef Search ADS Fraenkel , G. S . 1959 . The raison d’ĕtre of secondary plant substances . Science . 129 : 1466 – 1470 . Google Scholar CrossRef Search ADS PubMed Gaston , K. J . 1996 . Species richness: measure and measurement , pp. 77 – 113 . In K. J. Gaston (ed.), Biodiversity: a biology of numbers and difference . Blackwell Science , Oxford . Gavier-Pizarro , G. I. , V. C. Radeloff , S. I. Stewart , C. D. Huebner , and N. S. Keuler . 2010 . Housing is positively associated with invasive exotic plant species richness in New England, USA . Ecol. Appl . 20 : 1913 – 1925 . Google Scholar CrossRef Search ADS PubMed Godfrey , R. M. , and M. T. Johnson . 2014 . Effects of functionally asexual reproduction on quantitative genetic variation in the evening primroses (Oenothera, Onagraceae) . Am. J. Bot . 101 : 1906 – 1914 . Google Scholar CrossRef Search ADS PubMed Gotelli , N. J. , and R. K. Colwell . 2001 . Quantifying biodiversity: procedures and pitfalls in the measurement and comparison of species richness . Ecol. Lett . 4 : 379 – 391 . Google Scholar CrossRef Search ADS Hurlbert , S. H . 1971 . The nonconcept of species diversity: a critique and alternative parameters . Ecology . 52 : 577 – 586 . Google Scholar CrossRef Search ADS PubMed Johnson , M. T. J . 2008 . Bottom-up effects of plant genotype on aphids, ants, and predators . Ecology . 89 : 145 – 154 . Google Scholar CrossRef Search ADS PubMed Johnson , M. T. J. , and A. A. Agrawal . 2005 . Plant genotype and environment interact to shape a diverse arthropod community on evening primrose (Oenothera biennis) . Ecology . 86 : 874 – 885 . Google Scholar CrossRef Search ADS Johnson , M. T. J. , and A. A. Agrawal . 2007 . Covariation and composition of arthropod species across plant genotypes of evening primrose (Oenothera biennis) . Oikos . 116 : 941 – 956 . Johnson , M. T. , M. J. Lajeunesse , and A. A. Agrawal . 2006 . Additive and interactive effects of plant genotypic diversity on arthropod communities and plant fitness . Ecol. Lett . 9 : 24 – 34 . Google Scholar PubMed Kempton , R. A. , and L. R. Taylor . 1978 . The Q statistic and the diversity of floras . Nature . 275 : 252 – 253 . Google Scholar CrossRef Search ADS Kim , J. , J. Chae , and T. H. Koo . 2007 . Variation in bird diversity in relation to habitat size in the urban landscape of Seoul, South Korea . Acta Ornithol . 42 : 39 – 44 . Google Scholar CrossRef Search ADS Mack , R. N. , and W. M. Lonsdale . 2001 . Humans as global plant dispersers: getting more than we bargained for . Bioscience . 51 : 95 – 102 . Google Scholar CrossRef Search ADS Magurran , A. E . 2004 . Measuring biological diversity . Blackwell Pub ., Maldan, MA . McKinney , M. L . 2006 . Urbanization as a major cause of biotic homogenization . Biol. Conserv . 127 : 247 – 260 . Google Scholar CrossRef Search ADS Mphosi , M. S. , and S. P. Foster . 2010 . Female preference and larval performance of sunflower moth, Homoeosoma electellum, on sunflower pre-breeding lines . Entomol. Exp. Appl . 134 : 182 – 190 . Google Scholar CrossRef Search ADS Narango , D. L. , D. W. Tallamy , and P. P. Marra . 2017 . Native plants improve breeding and foraging habitat for an insectivorous bird . Biol. Cons . 213 : 42 – 50 . Google Scholar CrossRef Search ADS Oksanen , J. , F. G. Blanchet , R. Kindt , P. Legendre , P. R. Minchin , R. B. O’Hara , G. L. Simpson , P. Solymos , M. H. H. Stevens , and H. Wagner . 2013 . vegan: Community Ecology Package. R package version 2.0–10 . http://CRAN.R-project.org/package=vegan Peck , D. C. , and V. Thompson . 2008 . Spittlebugs (Hemiptera:Cercopoidea) , pp. 3512 – 3516 . In J. L. Capinera (ed.), Encyclopedia of entomology , 2nd ed . Springer , Dordrecht, Netherlands . Pimentel , D. , and A. G. J. Wheeler . 1973 . Species and diversity of arthropods in the alfalfa community . Environ. Entomol . 2 : 659 – 668 . Google Scholar CrossRef Search ADS R Core Team . 2013 . R: a language and environment for statistical computing . R Foundation for Statistical Computing , Vienna, Austria . http://www.R-project.org/ Schaefer , C. W. , and N. P. Chopra . 1982 . Cladistic analysis of the Rhopalidae (Hemiptera, Coreoidea) with a list of food plants . Ann. Entomol. Soc. Am . 75 : 224 – 233 . Google Scholar CrossRef Search ADS Slater , J. A. , and R. M. Baranowski . 1990 . Lygaeidae of Florida (Hemiptera: Heteroptera) , vol. 14 . Arthropods of Florida and Neighboring Land Areas. Gainesville, FL . Spengel , C. K . 1793 . Das entdeckte Geheimnis der Natur im Bau und in der Befruchtung der Blumen . Friedrich Vieweg, Berlin, Germany . Tallamy , D. W . 2004 . Do alien plants reduce insect biomass ? Conserv. Biol . 18 : 1689 – 1692 . Google Scholar CrossRef Search ADS Tallamy , D. W. , M. Ballard , and V. D’Amico . 2010 . Can alien plants support generalist insect herbivores ? Biol. Invasions . 12 : 2285 – 2292 . Google Scholar CrossRef Search ADS Tenczar , E. G. , and V. A. Krischik . 2007 . Effects of new cultivars of ninebark on feeding and ovipositional behavior of the specialist ninebark beetle, Calligrapha spiraeae (Coleoptera: Chrysomelidae) . Hortscience . 42 : 1396 – 1399 . Wallner , A. M. , B. Molano-Flores , and C. H. Dietrich . 2013 . Using Auchenorrhyncha (Insecta: Hemiptera) to develop a new insect index in measuring North American tallgrass prairie quality . Ecol. Indic . 25 : 58 – 64 . Google Scholar CrossRef Search ADS Wang , Y. , U. Naumann , S. Wright , and D. Warton . 2014 . mvabund: statistical methods for analysing multivariate abundance data. R package version 3.9.3 . http://CRAN.R-project.org/package=mvabund Warton , D. I. , S. T. Wright , and Y. Wang . 2012 . Distance-based multivariate analyses confound location and dispersion effects . Methods Ecol. Evo. L . 3 : 89 – 101 . Google Scholar CrossRef Search ADS Wheeler , A. G . 2001 . Biology of the plant bugs (Hemiptera: Miridae): pests, predators, opportunists . Cornell University Press , Ithaca, NY . Wilson , E. O . 1987 . The little things that run the world (the importance and conservation of invertebrates) . Conserv. Biol . 1 : 344 – 346 . Google Scholar CrossRef Search ADS Wilson , S. W. , J. L. Smith , and A. H. Purcell . 1993 . An inexpensive vacuum collector for insect sampling . Entomol. News . 104 : 203 – 208 . Wimp , G. M. , W. P. Young , S. A. Woolbright , G. D. Martinsen , P. Keim , T. G. Whitham , and T. Meagher . 2004 . Conserving plant genetic diversity for dependent animal communities . Ecol. Lett . 7 : 776 – 780 . Google Scholar CrossRef Search ADS Young , O. P . 1986 . Host plants of the tarnished plant bug, Lygus lineolaris (Heteroptera, Miridae) . Ann. Entomol. Soc. Am . 79 : 747 – 762 . Google Scholar CrossRef Search ADS Zuefle , M. E. , W. P. Brown , and D. W. Tallamy . 2008 . Effects of non-native plants on the native insect community of Delaware . Biol. Invasions . 10 : 1159 – 1169 . Google Scholar CrossRef Search ADS © The Author(s) 2018. Published by Oxford University Press on behalf of Entomological Society of America. All rights reserved. For permissions, please e-mail: journals.permissions@oup.com. This article is published and distributed under the terms of the Oxford University Press, Standard Journals Publication Model (https://academic.oup.com/journals/pages/open_access/funder_policies/chorus/standard_publication_model) http://www.deepdyve.com/assets/images/DeepDyve-Logo-lg.png Environmental Entomology Oxford University Press

Ecological Value of Native Plant Cultivars Versus Wild-Type Native Plants for Promoting Hemipteran Diversity in Suburban Areas

Loading next page...
 
/lp/ou_press/ecological-value-of-native-plant-cultivars-versus-wild-type-native-hvIjEpzBrK
Publisher
Entomological Society of America
Copyright
© The Author(s) 2018. Published by Oxford University Press on behalf of Entomological Society of America. All rights reserved. For permissions, please e-mail: journals.permissions@oup.com.
ISSN
0046-225X
eISSN
1938-2936
D.O.I.
10.1093/ee/nvy057
Publisher site
See Article on Publisher Site

Abstract

Abstract Because of concerns over recent declines in overall biodiversity in suburban areas, homeowners are attempting to improve the ecological functioning of their landscapes by incorporating native plants. Native plants are important for supporting native herbivorous insects, but it is unknown whether the native plants that are commercially available, typically cultivated varieties (cultivars) of a single genotype, are equally effective as food sources as the local, wild-type plants. We compared the hemipteran communities feeding on cultivars and wild-propagated plants for four species of native perennials commonly used as ornamentals. Of 65 hemipteran species collected, 35 exhibited a preference for some plant species over others, indicating a high degree of host-plant specialization. Moreover, the insect community associated with cultivars was distinct from the insect community associated with wild-type plants for each plant species, with three to four insect species accounting for most of the observed difference. Total insect abundance and insect biomass differed between cultivars and wild-propagated plants, but the direction of the difference changed over time and was not consistent among plant species. Species richness and a diversity index (the Q statistic) did not differ between cultivars and wild-type plants. These data suggest that abundance and diversity of hemipteran insects does not depend on the source of the plant material per se, but rather on the particular characteristics of cultivars that distinguish them from the wild type. wildlife value, biodiversity, hemiptera, plant–insect interaction, leafhopper As the population of the United States expands, more land is converted for residential use. As of 2012, approximately 70 million acres were classified as urban use, and 106 million acres were classified as rural residential use in the United States (Bigelow and Borchers 2017). Landscaping in residential areas typically does not reflect the natural plant community that existed in that area before development; in fact, urban areas are associated with a replacement of native flora by exotic species (McKinney 2006), many of which also invade surrounding areas (Gavier-Pizarro et al. 2010). Most of the exotics that replace natives in landscapes are purchased through the plant trade industry for their aesthetic value (Mack and Lonsdale 2001). There has been some speculation that this prevalence of non-native vegetation in residential landscapes could contribute to a loss of biodiversity in suburban areas (Tallamy 2004). The purported mechanism by which landscaping with exotic plants would lead to a loss of biodiversity has roots in plant–insect interaction theory. Most insect herbivores specialize on a narrow group of plants, usually plants in three or fewer families (Bernays and Graham 1988). One theory predicts this specialization is a result of coevolution between plants and insects. Plants produce a diverse array of secondary metabolites that act as feeding deterrents, and herbivorous insects develop specialized physiological adaptations for detoxifying those metabolites (Fraenkel 1959, Ehrlich and Raven 1964). When large areas of native vegetation are replaced with exotic vegetation, as is often the case in suburban landscapes, the exotic vegetation may contain secondary metabolites that are not present in the native vegetation, and hence native insects are not equipped to deal with these novel feeding deterrents. Because arthropods often make up the largest proportion of animal biomass in a given ecosystem, removing one of their primary food sources—native plants—could catastrophically disrupt food webs, with effects cascading up to higher trophic levels (Wilson 1987). Recent research has provided strong empirical evidence that some exotic plant species support less herbivorous insect biomass and less diverse herbivorous insect communities than native plants (Zuefle et al. 2008, Burghardt et al. 2010, Tallamy et al. 2010, Burghardt and Tallamy 2013). There is also evidence that the diversity of organisms in higher trophic levels, especially birds, is highly correlated with the abundance and diversity of insects in suburban habitats (Kim et al. 2007, Burghardt et al. 2009, Narango et al. 2017). The concern that a decline in insect biomass could cause a concomitant decline in overall biodiversity in suburban areas, together with a desire to improve the ecological functioning of landscapes, has spurred an interest in ‘gardening for wildlife’ by replacing exotics with native ornamental plants in landscapes. But are the native plants available commercially, typically cultivated varieties (cultivars) of a single genotype, equally effective as the local, wild-type plants in providing food for native herbivorous insects? There are several lines of evidence that suggest cultivars and wild-type plants could differ in their abilities to support herbivorous insects. Cultivars are often asexually-propagated, and insect diversity is known to correlate with the genetic diversity of their host plants (Wimp et al. 2004, Johnson et al. 2006). Cultivars are also selected for some characteristic that distinguishes them from the wild form. Pest resistance, altered growth habitat, changes in flower or leaf color, and sterility are common goals of plant breeders, and any of these could influence an insect’s ability to feed on the plant (e.g., Tenczar and Krischik 2007, Mphosi and Foster 2010). While many of these traits imply negative consequences for insect communities, it is important to note that some characteristics may actually improve a cultivar’s ecological value relative to the wild form. In the simplest case, selecting for hybrid vigor would make more plant biomass available to support more insects. We investigated whether these theoretical consequences of cultivar selection actually affect herbivorous insects in a garden setting. We chose several native herbaceous perennials that occur locally in natural areas near the study site and have cultivars available commercially. We determined whether the cultivars differed from plants grown from wild-collected seed in their ability to serve as a food source for native hemipterans (Auchenorrhyncha and Heteroptera), a highly abundant and speciose group associated with grasslands and plants growing in open, disturbed areas (Wallner et al. 2013). We measured hemipteran biomass, abundance, diversity, and community composition over the course of one growing season. Materials and Methods Study Site and Plant Material Plots were established in an open field at the Mimsie Lanier Center for Native Plant Studies at the State Botanical Garden of Georgia in Athens, Clarke County, Georgia. The Botanical Garden was an ideal study site for this research because it reflects a typical residential setting. It is a mixture of woodlands, lawns, and gardens located just on the outskirts of a medium-sized city. The field in which plots were established is located within the floodplain of the Oconee River and is characterized by periodic flooding and loamy, alluvial soils. There are open and forested areas, both upland and bottomland, immediately surrounding the site. All wild-type plants were grown from seed locally-collected from populations occurring in natural areas. Here, we define natural areas to mean open sites comprised of mostly native, early-successional vegetation; this includes relatively undisturbed sites, such as granitic outcrops, and sites with man-made disturbance, such as utility rights-of-way maintained by mowing. Seeds of Amsonia tabernaemontana were collected from a population at Currahee Mountain in Stephens County, Georgia. Seeds of Monarda fistulosa were collected from a natural area within the State Botanical Garden. Seeds of Coreopsis grandiflora, Oenothera fruticosa, and Schizachyrium scoparium were collected from the Rock and Shoals Natural Area, a granitic outcrop at the end of Rock and Shoals Dr. in Clarke County, Georgia. For all plant species, a few seeds were collected from many individuals (usually >50) in order to capture as much of the genetic variation within the population as possible. The common names and plant families of the plants used in the study are provided in Table 1. Table 1. Descriptions of plant species and cultivars used in this experiment Plant species Common name Cultivar Family Cultivar origin Difference from wild-type Amsonia tabernaemontana (Walter, Gentianales: Apocynaceae) Eastern bluestar ‘Blue Ice’ Apocynaceae Interspecific hybrida Longer bloom, darker flowers, compact form Coreopsis grandiflora (Hogg ex Sweet, Asterales: Asteraceae) Large-flowered tickseed ‘Tequila Sunrise’ Asteraceae Interspecific hybrid Variegated leaves, compact form Monarda fistulosa (Linnaeus, Lamiales: Lamiaceae) Wild bergamot ‘Claire Grace’ Lamiaceae Straight species Powdery mildew resistant, darker flowers Oenothera fruticosa (Linnaeus, Myrtales: Onagraceae) Southern Sundrops ‘Cold Crick’ Onagraceae Interspecific hybrida Sterile, compact form Schizachyrium scoparium ((Michaux) Nash, Poales: Poaceae) Little bluestem ‘Prairie Blues’b Poaceae Straight species Blue-green foliage turning wine-red in fall Plant species Common name Cultivar Family Cultivar origin Difference from wild-type Amsonia tabernaemontana (Walter, Gentianales: Apocynaceae) Eastern bluestar ‘Blue Ice’ Apocynaceae Interspecific hybrida Longer bloom, darker flowers, compact form Coreopsis grandiflora (Hogg ex Sweet, Asterales: Asteraceae) Large-flowered tickseed ‘Tequila Sunrise’ Asteraceae Interspecific hybrid Variegated leaves, compact form Monarda fistulosa (Linnaeus, Lamiales: Lamiaceae) Wild bergamot ‘Claire Grace’ Lamiaceae Straight species Powdery mildew resistant, darker flowers Oenothera fruticosa (Linnaeus, Myrtales: Onagraceae) Southern Sundrops ‘Cold Crick’ Onagraceae Interspecific hybrida Sterile, compact form Schizachyrium scoparium ((Michaux) Nash, Poales: Poaceae) Little bluestem ‘Prairie Blues’b Poaceae Straight species Blue-green foliage turning wine-red in fall aPurported hybrid. bPropagated from seed. View Large Table 1. Descriptions of plant species and cultivars used in this experiment Plant species Common name Cultivar Family Cultivar origin Difference from wild-type Amsonia tabernaemontana (Walter, Gentianales: Apocynaceae) Eastern bluestar ‘Blue Ice’ Apocynaceae Interspecific hybrida Longer bloom, darker flowers, compact form Coreopsis grandiflora (Hogg ex Sweet, Asterales: Asteraceae) Large-flowered tickseed ‘Tequila Sunrise’ Asteraceae Interspecific hybrid Variegated leaves, compact form Monarda fistulosa (Linnaeus, Lamiales: Lamiaceae) Wild bergamot ‘Claire Grace’ Lamiaceae Straight species Powdery mildew resistant, darker flowers Oenothera fruticosa (Linnaeus, Myrtales: Onagraceae) Southern Sundrops ‘Cold Crick’ Onagraceae Interspecific hybrida Sterile, compact form Schizachyrium scoparium ((Michaux) Nash, Poales: Poaceae) Little bluestem ‘Prairie Blues’b Poaceae Straight species Blue-green foliage turning wine-red in fall Plant species Common name Cultivar Family Cultivar origin Difference from wild-type Amsonia tabernaemontana (Walter, Gentianales: Apocynaceae) Eastern bluestar ‘Blue Ice’ Apocynaceae Interspecific hybrida Longer bloom, darker flowers, compact form Coreopsis grandiflora (Hogg ex Sweet, Asterales: Asteraceae) Large-flowered tickseed ‘Tequila Sunrise’ Asteraceae Interspecific hybrid Variegated leaves, compact form Monarda fistulosa (Linnaeus, Lamiales: Lamiaceae) Wild bergamot ‘Claire Grace’ Lamiaceae Straight species Powdery mildew resistant, darker flowers Oenothera fruticosa (Linnaeus, Myrtales: Onagraceae) Southern Sundrops ‘Cold Crick’ Onagraceae Interspecific hybrida Sterile, compact form Schizachyrium scoparium ((Michaux) Nash, Poales: Poaceae) Little bluestem ‘Prairie Blues’b Poaceae Straight species Blue-green foliage turning wine-red in fall aPurported hybrid. bPropagated from seed. View Large Four of the cultivars were purchased as liners from North Creek Nurseries in Landenberg, Pennsylvania. These were Amsonia ‘Blue Ice’, Coreopsis ‘Tequila Sunrise’, Monarda fistulosa ‘Claire Grace’, and Oenothera ‘Cold Crick’. All four cultivars were propagated vegetatively; i.e., they were clones from a single source. The fifth cultivar, Schizachyrium scoparium ‘Prairie Blues’, was donated by Hoffman Nursery in Bahama, NC. This cultivar was propagated by seed. The cultivars were chosen to represent both variety in their genetic origins and their traits of interest. For example, the cultivars of Coreopsis, Oenothera, and Amsonia are all interspecific hybrids or likely hybrids, whereas the cultivars of Monarda and Schizachyrium are selections of the straight species. Likewise, Schizachyrium and Coreopsis were selected for traits that involve a change in leaf chemistry (viz., leaf color), while Oenothera, Amsonia, and Monarda were selected for traits such as form, sterility, and disease resistance. We provide a full summary of the purported differences between the wild-type plants and cultivars (Table 1). Both cultivars and wild-type plants were planted in the field in April 2013. At that point, both wild-type plants and cultivars had root balls that filled square nursery pots approximately 3.5 inches wide and 3 inches deep. By the late summer of 2013, many plants were fully-grown and flowering. By the summer of 2014, just before data were collected, all the plants had reached flowering size and most plots had achieved 100% cover. Experimental Design The experiment followed a fully-randomized two-way analysis of variance (ANOVA) design (schematic diagram in Supplementary Appendix S1). The first factor was plant species and included five levels: Amsonia, Coreopsis, Monarda, Oenothera, and Schizachyrium. The second factor was plant source and included two levels: cultivar and wild-type. There were five replicates for each treatment, giving a total of 50 experimental units. Each experimental unit was a 2 × 2 m plot containing 16 plants evenly spaced, and plots were placed 1.5 m apart. The planting density per plot was chosen to be 16 individuals so that percent cover would be close to 100% by the time insects were collected. Plots were kept weed-free throughout the growing season and wood mulch was used in-between plots. Data Collection We collected insects from all the plant species on three dates in 2014: 8 July, 15 August, and 20 September. The collection was broken up into two consecutive days for each sample period because all 50 plots could not be sampled in 1 d. We only sampled on days that were sunny with low wind speed. Insects were vacuumed for 1 min from plots with a modified leaf vacuum (Stihl BG 55) following the design in Wilson et al. (1993). The period of 1 min was based on previous trials and chosen to maximize the number of insects vacuumed while minimizing the number of vacuumed insects escaping. The order in which the plots were sampled was randomized to reduce any systematic bias caused by insects that escaped the vacuum and moved to other plots. Sampling began at 11:00 a.m. and ended by 2:00 p.m. to coincide with peak xylem flow. The insects were killed with ethyl acetate, sorted into hemipterans and non-hemipterans, and stored dry in boxes. Representative specimens of each species were pinned for subsequent identification. A combination of family-specific regional keys (e.g., DeLong 1948 for leafhoppers) and genera-specific keys were used to identify insects to species level. All identifications were subsequently confirmed by a taxonomist (Dr. Rick Hoebeke, Natural History Museum, University of Georgia). We also categorized insects into feeding guilds based on whether they primarily feed on xylem, phloem, mesophyll, or seeds/fruits. A full list of the taxonomic references and resources for information about insects’ life histories can be found in Supplementary Appendix S7. We measured two response variables for each experimental unit (plot) at each sampling date: total dry biomass of hemipterans and abundance of each adult hemipteran species. Total dry biomass included both adult hemipterans and nymphs and was measured to the nearest 0.0001 g. For the count data, nymphs were not included because most could not be identified to species. From the abundances of each species, we constructed several other response variables for our statistical analyses. These were total overall abundance of hemipterans, species richness, and the Q-statistic. The Q-statistic is a diversity index that measures the interquartile slope of the species accumulation curve, excluding both low- and high-abundance species from the measure of diversity (Kempton and Taylor 1978). We chose the Q-statistic over more common indices such as Simpson’s Index or Fisher’s Alpha because many of these indices are highly influenced either by low-abundance species or by dominant species. In the context of this experiment, infrequently-collected insects more likely represent ‘tourists’ (Gaston 1996) rather than rare species that are actually associated with a given plant, and hence should carry less weight instead of more. The Q-statistic provides a good compromise for measuring diversity when it is desirable to reduce the influence of both low- and high-abundance species. Standardization by Plant Biomass Because plant species, and sometimes cultivars and wild-type plants, differed in size, we standardized total insect biomass and total abundance by the dry weight of the plants they were collected from. We could only measure plant dry weight in September, however, because it involved destructively sampling the plants. Within 2 wk following the September sampling, we harvested all of the above-ground plant material from each plot and stored it in large paper bags. We dried all of the plant material in a forced-air drying oven at 60°C for 7 d. After drying, we recorded the total dry weight of each plot to the nearest 0.1 kg. The dry weights were used to transform the September insect data as total abundance or total insect biomass per kilogram plant biomass. Statistical Analyses We used repeated measures two-way ANOVA to analyze total abundance and total dry biomass of insects. Several variants of repeated measures ANOVA exist. We used the variant that performs multivariate analysis of variance (MANOVA) on the differences between adjacent time periods (a.k.a. profile analysis) to test for the effects of plant species, plant source, time, and their interactions. We used the function manova() in R (R Core Team 2013). The standardized abundance and insect biomass data from September were analyzed with univariate two-way ANOVA using function aov(). Rather than analyze species richness and the Q-statistic by sampling date (i.e., with repeated measures ANOVA), we pooled data over all three sampling periods to compare diversity among treatments for the entire growing season. We analyzed richness and the Q-statistic with two-way ANOVA using function aov() in R (R Core Team 2013). The normality assumption of ANOVA was tested with function shapiro.test() and the homogeneity of variances assumption was tested with function bartlett.test(). To visually compare species richness among treatments, we constructed a sample-based rarefaction curve using function specaccum() in package ‘vegan’ in R (Oksanen et al. 2013). Although all the experimental units were the same size and sampling effort was equal among treatments, we rescaled the x-axis by number of individuals to account for differences in insect density among treatments (Gotelli and Colwell 2001). We also used the pooled count data to test whether the insect communities differed among treatments. Although distance-based analyses are often used for addressing community-level questions, they suffer from major problems (Anderson 2001, Warton et al. 2012). We instead used multivariate analyses based on generalized linear models (GLMs) to model the counts of each insect species directly. In particular, we fit the abundance data for each insect species to separate negative binomial models using function manyglm() in package ‘mvabund’ in R (Wang et al. 2014). We specified the identity matrix as the correlation structure for our multivariate tests and used the likelihood ratio test for Analysis of Deviance. The function provides a multivariate test for the entire community and univariate tests for each insect species, with P-values for the univariate tests corrected for multiple comparisons using a resampling-based procedure. We used 1,000 permutations to obtain P-values. Results Overall, we collected almost 12,000 adult hemipterans representing 130 different species. Exactly half of the species had abundances less than or equal to five in the entire dataset, so we excluded those species from all analyses. Those species were likely ‘transients’ or ‘tourists’, and not actually feeding on the plants they were collected from. Because arthropods are highly mobile, it is not surprising that such a large proportion of the species collected would be transients (Pimentel and Wheeler (1973) for a similar result). Although including these species in the analyses would make little difference for response variables such as total abundance, they would be highly influential for diversity measures such as species richness. Even the Q-statistic, which is meant to exclude both infrequent and highly abundant species, can be biased when the majority of the species are singletons (Magurran 2004). We provide a list of all the insect species excluded from the analyses and their abundances (Supplementary Appendix S2). Only one species of hemipteran was found feeding on the Amsonia wild-type plants and cultivars, and only nine individuals of this species were collected over the entire growing season, so the Amsonia wild-type and cultivar treatments were excluded from all analyses. The species found feeding on Amsonia was the broad-headed sharpshooter, Oncometopia orbona. Total Abundance and Insect Biomass Repeated measures ANOVA indicated a three-way interaction among plant species, plant source, and time for both total abundance of insects and total insect biomass. We broke up both analyses at each level of plant species and plotted these results as a function of time (Fig. 1a and b). For most plant species, there were significant interactions between plant source and time. Hence, we compared abundance or biomass of cultivars and wild-type plants for each plant species at each date. P-values were corrected for multiple comparisons using the Bonferroni method. The Bonferroni method is the most conservative multiple comparison adjustment so we displayed significant differences both at an experiment-wise α = 0.05 (P-value < 0.0042) and at α = 0.10 (P-value < 0.0083), denoted by three asterisks and one asterisk, respectively. Fig. 1. View largeDownload slide Plots of (a) total abundance of adult hemipterans and (b) total hemipteran biomass for cultivars and wild-type plants of each plant species at three sampling dates. Asterisks indicate significant differences in means between cultivars and wild-type plants for a given plant species on a given date. Three asterisks represent differences that were significant after a Bonferroni correction for multiple comparisons with an experiment-wise α = 0.05 (P-value < 0.0042). One asterisk represents differences that were significant after a Bonferroni correction with an experiment-wise α = 0.10 (P-value < 0.0083). Error bars are SD. Fig. 1. View largeDownload slide Plots of (a) total abundance of adult hemipterans and (b) total hemipteran biomass for cultivars and wild-type plants of each plant species at three sampling dates. Asterisks indicate significant differences in means between cultivars and wild-type plants for a given plant species on a given date. Three asterisks represent differences that were significant after a Bonferroni correction for multiple comparisons with an experiment-wise α = 0.05 (P-value < 0.0042). One asterisk represents differences that were significant after a Bonferroni correction with an experiment-wise α = 0.10 (P-value < 0.0083). Error bars are SD. Generally, there was close correspondence between insect abundance and insect biomass. The most notable exception was Monarda. The Monarda wild-type plants consistently had two to five times higher insect abundance than the cultivars, but insect biomass was not significantly different between cultivars and wild-type plants at any sampling date. Another notable pattern was that some treatments increased in abundance and biomass over the growing season, whereas others decreased (and some changed little). In particular, insect abundance and biomass increased on the Coreopsis and Schizachyrium cultivars over the growing season, but decreased on the Coreopsis and Oenothera wild-type plants. For the Oenothera cultivar and wild-type comparison, abundance and biomass were three times higher on wild-type plants early in the season, but there were no differences by the end of the season. For Schizachyrium, there were no differences in abundance or biomass early in the season, but insect abundance and biomass were approximately 2 and 3.5 times higher, respectively, on the cultivar by the last sampling date. The Coreopsis had perhaps the most striking pattern. Abundance and biomass were approximately 1.5 times higher on the wild-type plants on the first sampling date, but were three to four times higher on the cultivar by the end of the season. However, this difference in pattern is most likely due to the fact that many of the Coreopsis wild-type plants had died by the last sampling date. Species Diversity Two-way ANOVA of species richness indicated no significant interactions between plant species and plant source, but a significant overall effect of plant species (plant source: F1,32 = 2.01, P = 0.17; plant species: F3,32 = 7.26, P < 0.001; interaction: F3,32 = 2.41, P = 0.085). The results of two-way ANOVA of the Q-statistic were similar (plant source: F1,32 = 0.01, P = 0.92; plant species: F3,32 = 9.90, P < 0.001; interaction: F3,32 = 1.44, P = 0.25). That is, there was no indication of differences in richness or the Q-statistic between wild-type plants and cultivars for any plant species. Mean richness and mean Q-statistic were significantly higher for Oenothera than all other plant species after correcting for multiple comparisons with Tukey’s HSD test (Fig. 2a and b, respectively). Oenothera had on average five to six more insect species than the other plants. There were no differences in richness or the Q-statistic among the other plant species. Fig. 2. View largeDownload slide Plots of (a) species richness and (b) Q-statistic for cultivars and wild-type plants of each plant species pooled over three sampling dates. Species richness and the Q-statistic were used as measures of species diversity of adult hemipterans for each treatment. Different letters denote means of plants species that were significantly different at α = 0.05 after correcting for multiple comparisons with Tukey’s HSD test. Oenothera had higher average species richness and values of the Q-statistic than the other three plant species. There was no evidence for a difference in richness or the Q-statistic between wild-type plants and cultivars for any plant species. Error bars are SD. Fig. 2. View largeDownload slide Plots of (a) species richness and (b) Q-statistic for cultivars and wild-type plants of each plant species pooled over three sampling dates. Species richness and the Q-statistic were used as measures of species diversity of adult hemipterans for each treatment. Different letters denote means of plants species that were significantly different at α = 0.05 after correcting for multiple comparisons with Tukey’s HSD test. Oenothera had higher average species richness and values of the Q-statistic than the other three plant species. There was no evidence for a difference in richness or the Q-statistic between wild-type plants and cultivars for any plant species. Error bars are SD. The rarefaction curve of species richness generally agreed with the results of ANOVA with the exception of the Monarda cultivar and wild-type plants (Supplementary Appendix S3). For example, the curves for the Oenothera cultivar and wild-type are above all the other treatments, indicating that species richness is higher on Oenothera than the other plant species. Also, the wild-type and cultivar curves overlap for Coreopsis, Oenothera, and Schizachyrium, indicating that there was no effect of plant source. However, for Monarda, the curve for the cultivar lies above the curve for the wild-type. This discrepancy is due to differences in the density of insects on the Monarda wild-type plants versus the cultivars. The pooled abundance for the Monarda wild-type treatment was approximately four times higher than the cultivar treatment, despite that the area sampled for each treatment was the same. The Monarda wild-type treatment actually had higher absolute richness, and hence would explain why ANOVA indicated no difference in average richness between the cultivar and wild-type. Insect Community The multivariate GLM test for the entire insect community indicated a significant interaction between plant species and plant source in the full model (P ≈ 0.001). Although the interaction was significant for the multivariate test, this does not imply that the interaction would be significant for each univariate test. We also provide the results of univariate tests under the full model and plots of the abundances of each species (Supplementary Appendices S5 and S7, respectively). There was a significant effect of plant species (and no interaction) for over half of the insect species; i.e., most of the insect species prefer to feed on some plant species over others. These results fit well with plant–insect interaction theory, but are not the main interest of this experiment. Hence, we only tested the effect of plant source at each level of plant species in our follow-up analyses. There was a significant effect of plant source for every plant species (Coreopsis: P ≈ 0.003; Monarda: P ≈ 0.01; Oenothera: P ≈ 0.006; Schizachyrium: P ≈ 0.024). That is, there was strong evidence that the hemipteran community differed between cultivars and wild-type plants for every plant species. After correcting P values for multiple comparisons, univariate tests indicated that three to four insect species were responsible for the differences in insect communities between cultivars and wild-type plants (species with asterisks in Figs. 3–5). Fig. 3. View largeDownload slide Abundances of phloem-feeding insects that showed the strongest differences between cultivars and wild-type plants for Coreopsis, Monarda, Oenothera, and Schizachyrium. Asterisks indicate plant species where there was a significant difference between the cultivar and wild-type at α = 0.05 after correcting for multiple comparisons. Error bars are SD. Fig. 3. View largeDownload slide Abundances of phloem-feeding insects that showed the strongest differences between cultivars and wild-type plants for Coreopsis, Monarda, Oenothera, and Schizachyrium. Asterisks indicate plant species where there was a significant difference between the cultivar and wild-type at α = 0.05 after correcting for multiple comparisons. Error bars are SD. Fig. 4. View largeDownload slide Abundances of mesophyll-feeding insects that showed the strongest differences between cultivars and wild-type plants for Coreopsis, Monarda, Oenothera, and Schizachyrium. Asterisks indicate plant species where there was a significant difference between the cultivar and wild-type at α = 0.05 after correcting for multiple comparisons. Error bars are SD. Fig. 4. View largeDownload slide Abundances of mesophyll-feeding insects that showed the strongest differences between cultivars and wild-type plants for Coreopsis, Monarda, Oenothera, and Schizachyrium. Asterisks indicate plant species where there was a significant difference between the cultivar and wild-type at α = 0.05 after correcting for multiple comparisons. Error bars are SD. Fig. 5. View largeDownload slide Abundances of seed- and fruit-feeding insects that showed the strongest differences between cultivars and wild-type plants for Coreopsis, Monarda, Oenothera, and Schizachyrium. Note that Lygus feeds on both fruits and mesophyll. Asterisks indicate plant species where there was a significant difference between the cultivar and wild-type at α = 0.05 after correcting for multiple comparisons. Error bars are SD. Fig. 5. View largeDownload slide Abundances of seed- and fruit-feeding insects that showed the strongest differences between cultivars and wild-type plants for Coreopsis, Monarda, Oenothera, and Schizachyrium. Note that Lygus feeds on both fruits and mesophyll. Asterisks indicate plant species where there was a significant difference between the cultivar and wild-type at α = 0.05 after correcting for multiple comparisons. Error bars are SD. We assigned each insect we collected to a feeding guild based on information about its life history (Supplementary Appendix S7, which also contains food plant records for each insect). Insects were classified as feeding on xylem, phloem, leaf/stem mesophyll, or fruits/seeds, and an insect species could belong to multiple guilds. Of the many xylem-feeding insect species collected, few showed a preference between cultivars and wild-type plants, and there were none with significantly different abundances after adjusting P values. However, there were examples of insect species that preferred either the cultivar or the wild-type plant for each of the other three feeding guilds. For phloem-feeding and mesophyll-feeding insects, some insect species appeared to prefer the wild-type plants, while others appeared to prefer the cultivars (Figs. 3 and 4). In most cases, the differences in abundance were quite large. For example, the abundance of Empoasca sp. was over eight times higher on the Monarda wild-type than the cultivar, and Sixeonotus unicolor was almost six times higher on the Coreopsis cultivar than the wild-type. Differences for the other insect species ranged from approximately four-fold to almost eightfold. Although the abundances of many of these insects were relatively high, a few showed statistically significant differences between wild-types and cultivars despite having low abundances. For example, the average abundances for Blissus leucopterous on the Schizachyrium wild-type and Ceratocapsus punctulatus on the Oenothera wild-type were 6.2 and 3.6, respectively, yet both were significantly higher than their corresponding cultivars. For seed-feeding insects, there were five instances in which insect abundance differed significantly between cultivar and wild-type, and, in each case, it was the wild-type plants that were preferred (Fig. 5). The two species for which differences were observed were Coreopsis and Oenothera. Oenothera ‘Cold Crick’ is marketed as a sterile hybrid (Table 1), while Coreopsis ‘Tequila Sunrise’ is apparently sterile, as no seeds were observed during the study. Wild-type plants of both species were fertile. Like with the other feeding guilds, the observed differences in abundances of seed-feeding insects were very large. Neopamera bilobata was over 17 times more abundant on the Oenothera wild-type than on the cultivar. Lygus lineolaris was four times more abundant on the Coreopsis wild-type than the cultivar and over 10 times more abundant on the Oenothera wild-type than the cultivar. The abundances of the other two seed-feeding insects were relatively low. Xyonysius californicus and Homaemus proteus had average abundances of 3.0 and 1.8, respectively, on the Coreopsis wild-type, but no individuals of either species were found on the Coreopsis cultivar. Standardized Data We observed significant differences in total insect abundance and total insect biomass between cultivars and wild-type plants for several plant species in September. Because this was our last sampling date, we were able to harvest all of the plant material and standardize our insect counts by the amount of plant biomass in plots to determine whether the observed differences were due to an inherent effect of plant source or simply the amount of plant biomass available. Most of the Coreopsis individuals in the wild-type treatment were dead by this point, so we excluded both the Coreopsis wild-type and cultivar treatments from the analyses. The results of two-way ANOVAs of standardized abundance and standardized insect biomass revealed that standardization actually accentuated the observed differences between cultivars and wild-type plants (Supplementary Appendix S5). For example, for Monarda the raw abundance was approximately four times higher on wild-type plants than cultivars in September, but the standardized abundance was almost six times higher on wild-type plants than the cultivars. For Schizachyrium, the raw abundance was approximately two times higher on the cultivars than the wild-type plants in September, but the standardized abundance was almost four times higher on cultivars. The same general result was true for raw insect biomass versus standardized insect biomass. These results indicate plant source, and not differences in the amount of plant biomass available to insects, better explain the number of insects (and insect biomass) collected from plants. Discussion Different Hemipteran Species Associate With Different Plant Species Given that an estimated 90% or more of phytophagous insect species are host-plant specialists (Bernays and Graham 1988), we expected to find different insect communities associated with each plant species; here, we refer to host-plant specialists as insect species that feed on plants in three or fewer families, and generalists as species that feed on plants in more than three families. Our results agreed with this expectation, providing strong evidence that different hemipteran species are associated with different plant species. Although we suggest that the differences in the insect communities are a result of host-plant specialization among the insects we captured, note that we did not observe the feeding behavior of the insects directly. We measured the abundances of insects on different species of plants, which implies, but does not directly measure, which plants the insects were feeding on. Therefore, our data give an indication of host plant preferences for individual insect species, but these preferences do not preclude the possibility of the insects having much broader host ranges. In order to actually classify insects as specialists or generalists empirically from the data, direct measures of feeding behavior (and a larger suite of plants to test) would be necessary. Unfortunately, we were also unable to use host plants records from the literature to classify insects as specialists or generalists because records for most species were ambiguous, insufficient, or nonexistent (e.g., in Supplementary Appendix S7, the best available food plant records for several species was ‘herbaceous plants’). Although we were unable to classify insects as specialists and generalists, the patterns in their abundances clearly indicate that many insect species prefer certain plant species to others. The actual number of insect species observed to prefer some plants over others was 35 of 65 (Supplementary Appendix S4). Although this suggests that the majority of insects discriminate in their choice of food plants, this proportion is well below the expected estimate of 90% (if most insects are specialists). One possible reason for this discrepancy is that our sample size was small, so power may not have been sufficient to detect differences for some of the other insect species, especially those species with low overall abundances. Another possibility is that insect species were included in the analyses that were not actually feeding on the plants; these species would more likely show a random pattern of host-plant association rather than being found on one particular plant species. We attempted to reduce the number of these ‘tourist’ species by excluding all species with an overall abundance of five or less. We chose this conservative threshold to reduce the possibility of excluding from the analyses any rare species that were actually feeding on the plants. Another explanation is that the proportion of specialists among hemipterans is not as high as the proportion of specialists among herbivorous insects in general. Many of the hemipteran species we collected feed on xylem. There may be less host-plant specialization among those species because xylem generally contains fewer secondary metabolites that might act as feeding deterrents (Peck and Thompson 2008). In contrast, insect feeding guilds such as leafminers and gall-makers, and taxonomic groups such as lepidopterans, come into direct contact with cellular contents during the process of feeding, exposing them to secondary metabolites that potentially drive more specialized feeding behavior. Effect of Plant Source on Feeding Guilds Xylem-Feeding Insects Our main interest in this study was not whether different insect species are associated with different species of ornamental plants, but rather does plant source affect the insect community associated with a particular plant. For each of the plant species we tested, we found strong evidence that the hemipteran community differed between cultivars and wild-propagated plants. However, xylem-feeding insects contributed least to the observed differences in the insect communities, as none had large enough differences in abundance to meet the adjusted level of significance in the individual tests. Again, xylem-feeding insects likely experience fewer secondary metabolites, and this may explain why they were less discriminating between cultivars and wild-types. We also note that the total number of insect species that contributed most strongly to differences in the insect communities of wild-type plants and cultivars was relatively small (12 out of 65). Because hemipterans have haustellate mouthparts, they may be less exposed to plant secondary metabolites than are insects with mandibulate mouthparts, such as lepidopterans and orthopterans. Consequently, hemipterans may be less sensitive to changes in their food source, and the observed proportion of species driving community differences between cultivars and wild-type plants may be conservative relative to the pattern exhibited by other insect groups. Phloem-Feeding Insects Four phloem-feeding insect species contributed strongly to the observed differences in the insect community associated with cultivars versus wild-type plants. The species included two leafhoppers (Cicadellidae: Megophthalminae and Deltocephalinae), one fulgoroid (Derbidae), and one seed bug (Lygaeidae). The seed bug, Blissus leucopterous, feeds on the phloem of grasses rather than the seeds, as do all other members of the genus (Slater and Baranowski 1990). For three of the species, the wild-type plants were preferred. However, Balclutha neglecta exhibited a strong preference for the Schizachyrium cultivar. When attempting to explain why certain insect species preferred either the cultivar or wild-type, it is important to note that the individual tests for each species were performed on the pooled dataset, and therefore do not incorporate any information about the seasonal abundances of species. B. neglecta did not become abundant until the last sampling date. By this time, most of the wild-type Schizachyrium were starting to enter dormancy, whereas the cultivars had a slightly delayed phenology. This difference in phenology of the host-plants in combination with late-season peak abundance of B. neglecta could explain the observed preference for the cultivar. For the other three phloem-feeding species, it is not clear what differences between cultivars and wild types might be driving the observed preference for wild types. Phloem-feeding insects are known to respond strongly to the overall nitrogen content of the phloem as well as to particular amino acids that act as feeding stimulants (Cook and Denno 1994). However, we did not measure the chemical composition of the phloem (or any other parameters influencing host-plant quality), so any causal mechanisms are entirely speculative. Mesophyll-Feeding Insects Among the mesophyll-feeding insects exhibiting the largest differences in abundance between cultivars and wild-types, three were plant bugs (Miridae) and one was a leafhopper (Cicadellidae: Typhlocybinae). Both of these groups use the ‘lacerate and flush’ feeding strategy, in which they use their stylets to pierce the leaf tissue, lacerate the cells, and imbibe the cell contents (Wheeler 2001). Although mesophyll contains more nutrients than xylem or phloem, lacerating the cells also exposes the insect to whatever secondary metabolites that plant may produce as feeding deterrents, suggesting that insects with this feeding strategy would be particularly sensitive to differences in leaf chemistry. Because neither plant bugs nor typhlocybine leafhoppers were overly represented among the species showing preferences between cultivars and wild-type plants, this suggests either that the cultivars and wild-type plants have similar leaf chemistry or that other characteristics of the plants are playing an equally important role in influencing the host-plant preferences of insects. As with the phloem-feeding species, three species preferred the wild-type plants and only one preferred the cultivar (though there was weak evidence that Ceratocapsus punctulatus preferred the Coreopsis cultivar to the wild-type). The species preferring the cultivar, Sixeonotus unicolor, has been recorded to feed on ‘garden coreopsis’ (Wheeler 2001). It is not clear from the host plant record whether this refers to a cultivated variety of C. grandiflora or simply a Coreopsis species commonly used for ornamental purposes. Note also that none of the four mesophyll-feeding species were collected from Schizachyrium (except for two individuals of C. punctulatus, which were likely tourists). A probable explanation is that the lacerate and flush feeding strategy is not compatible with the particular defenses used by grasses. Grasses typically contain lignin and high levels of silica targeted at mechanically abrading the mouthparts of chewing insects (Cook and Denno 1994). Although hemipterans are not chewing insects, lignin and silica likely defend against the lacerate component of the feeding strategy with similar efficacy. Seed- and Fruit-Feeding Insects The seed- and fruit-feeding insects included two seed bugs (Lygaeidae and Rhyparochromidae), one shield-backed bug (Scutelleridae), and one plant bug (Miridae). The plant bug, Lygus lineolaris, can also feed on leaf and stem mesophyll, but appears to prefer the nectar, pollen, and immature fruits from flowers of herbaceous plants (Wheeler 2001). It is considered a serious pest for some agricultural crops and has an unusually broad host range, having been recorded from more than 300 plant species in many different families (Young 1986). In contrast to phloem- and mesophyll-feeders, the seed-feeding insects observed to have the largest differences in abundance between cultivars and wild-types consistently preferred the wild-type plants to the cultivars. Moreover, the only plant species in which differences were observed were species where the cultivar was sterile. That is, the seed-feeding insects we collected appeared to prefer the Coreopsis and Oenothera wild-types over the cultivars because the cultivars were sterile, whereas we observed no differences between Monarda and Schizachyrium cultivars and wild-types, presumably because the cultivars produced viable seeds. The seed bugs we collected were species that climb plants to feed on seeds, which is a less common strategy among seed bugs than feeding on seeds that fall to the ground (Slater and Baranowski 1990). Our sampling method would not have collected ground-dwelling insects very effectively, so we may be underestimating the actual number of species of seed-feeding insects that sterile cultivars fail to support. In addition to finding no difference in the numbers of Xyonysius californicus and L. lineolaris on Monarda cultivars and wild-types (Fig. 5), we collected several specialist seed-feeding species on both Monarda and Schizachyrium that did not discriminate between the cultivars and wild-types. For example, the scentless plant bug Arhyssus nigristernum (Rhopalidae), a specialist on Monarda and other mints (Schaefer and Chopra 1982), and the seed bug Paromius longulus (Rhyparochromidae), a specialist on grasses (Slater and Baranowski 1990), were found in similar numbers on cultivars and wild-types (Supplementary Appendix S6). It is clear why seed-feeding insects would be less abundant on sterile plants, but an additional point to note is that both of the sterile cultivars were hybrids, whereas both fertile cultivars were selections of the straight species. A common mechanism by which hybrid plants become sterile is crossing two species with an even, but different ploidy level to produce offspring with an odd ploidy level. Although we did not determine the chromosome number of our plants, this or another genetic mechanism is likely causing the sterility observed in the Coreopsis and Oenothera cultivars. We originally speculated that the genetic diversity of the host plants may influence the insects that feed on those plants, and while these results do not necessarily imply this is the case, the results do support the idea that the genetic origin of cultivars is important. Our conclusions regarding plant sterility and seed-feeding insects are analogous to the results obtained by others who compared cultivars and wild-type plants in terms of their abilities to support native insects, though their research focused exclusively on pollinators. Comba et al. (1999) and Corbet et al. (2001) observed that cultivars with altered flower morphology were visited by fewer pollinators or fewer species of pollinators than the wild-type plants. In some cases, the change in morphology reduced nectar secretion. For the cases in which nectar secretion was unaffected, the authors speculated that reduced accessibility caused the observed reduction in pollinator visits. In either case, fewer insects were supported by cultivars because the insect’s food resource was unavailable. Similarly, we found fewer seed-feeding insects on sterile cultivars because their food resource was absent. The codependency between insects and flowers for plant reproduction has long been recognized (Sprengel 1793). These mutualistic relationships are a result of a long history of coevolution. Our results provide additional evidence that selecting for traits that alter the reproductive biology of the plants, and hence perturb the result of tens-of-thousands or millions of years of increasing specialization, has a strong negative impact on the insects that utilize those plants. Insect Total Abundance, Biomass, and Diversity In addition to determining whether there were differences in the insect community, we wanted to know whether cultivars and wild-type plants differed for more general parameters, such as total insect abundance, biomass, and diversity. The most prominent trend apparent in the abundance and biomass data is that, for the most part, differences between cultivars and wild-types were not consistent across sampling date. Many of the patterns can be explained by qualitative observations regarding phenology and plant health combined with close examination of trends for individual insect species. For example, more insects and more insect biomass were collected from the Coreopsis wild-type on the first sampling date, but insect abundance and biomass were much higher on the cultivar by the last sampling date. This pattern can be explained by the fact that very few individuals of the Coreopsis wild-type were alive by the end of the growing season. Some individuals of the Coreopsis cultivar also died during the experiment, but more were alive on the last sampling date than had died. An obvious question is: did the higher feeding pressure on the Coreopsis wild-type plants early in the year cause their decline? While it is certainly possible, Coreopsis grandiflora is a short-lived perennial and most likely had reached the end of its life cycle. Another example was the sharp decline in insect abundance and biomass on the Oenothera wild-type plants over the growing season. Neopamera bilobata, a seed-feeding insect, was more abundant on the Oenothera wild-type than the cultivar and made up a large proportion of both the total abundance of insects and the total insect biomass collected from Oenothera wild-types. Oenothera blooms in early May and sets seed shortly after, with most seeds being dispersed by late summer. Although it is not apparent in the pooled data, N. bilobata was especially abundant on the first sampling date and declined substantially on each subsequent sampling, suggesting its population size was responding to the phenology of the Oenothera wild-types. There was no evidence for a difference in hemipteran diversity between cultivars and wild-types for any of the plant species we tested (Fig. 2). Given the other variables we measured, this is not surprising. The differences we observed in total abundance and total biomass were largely due to single, numerically dominant species. Because species richness does not measure the dominance or evenness of the community, single species that have a large effect on parameters such as total abundance and biomass would not have an undue influence on diversity. Also, when we investigated which insect species were contributing to the differences we observed in the overall community, we found that some species preferred cultivars and others preferred wild-type plants (Figs. 3–5). Together, the overall differences in the insect community and the lack of difference in diversity suggest that cultivars and wild-types support a similar number of insect species, but the particular species are different. Species richness is often criticized as a poor measure of diversity (Magurran 2004). To confirm that other methods would give similar results, we also used individual-based rarefaction (Hurlbert’s (1971) formulation) and a diversity index called the Q-statistic to assess diversity (Kempton et al. 1978). Like species richness, the Q-statistic is robust to the presence of over-dominant species because it excludes both the most and least abundant species from the measure. Unlike species richness, however, it incorporates information about the relative abundances of insects. Both the Q-statistic and rarefaction resulted in the same overall conclusion as species richness. Unanswered Questions and Future Directions Future research should not only address the differences in the insect communities associated with cultivars and wild-type plants, but also what characteristics of the plants are driving these differences. Chemical defenses, such as secondary metabolites, play a prominent role in deterring herbivory (Fraenkel 1959). Besides chemical defenses, plants also produce numerous physical defenses, including glandular and hooked trichomes. Rather than feeding deterrents, insects could have instead been responding to differences in feeding stimulants. Hemipterans are known to prefer host plants with higher concentrations of amino acids in the phloem, and specific amino acids sometimes act as feeding stimulants (Cook and Denno 1994). Because hemipterans in particular are known to respond strongly to nitrogen content, this aspect of host plant quality may be equally important as more frequently cited drivers of host-plant specialization, such as secondary metabolite chemistry, and should be investigated in any future research that aims to determine why herbivorous insects discriminate between cultivars and wild-type plants. All of the plants we used in this experiment were outbreeding species (Carman and Hatch 1982, Crawford and Smith 1984, Cruden et al. 1984, Godfrey and Johnson 2014), and so should contain much higher genetic diversity than the cultivars propagated asexually. Although this research did not investigate the question directly, we mentioned that the amount of genetic diversity contained in cultivars versus wild-type plants propagated from seed could be driving differences in the insect communities (see Johnson and Agrawal (2005), Johnson and Agrawal (2007), and Johnson (2008) for examples of positive correlation between host plant genetic diversity and variation in the associated arthropod community). However, our results did not support this hypothesis. We found different hemipteran communities associated with cultivars and wild-type plants, but the diversity of those communities did not differ. These results suggest that plants of the same species can support different herbivorous insects when their genetic origins differ, but that host plant genetic diversity has a much smaller relative effect (or no effect) on the diversity of the insect community. Although cultivars and wild-type plants of every plant species we tested differed in terms of their associated hemipteran communities, they did not differ in hemipteran species diversity. In addition, the differences in total hemipteran biomass and abundance we observed appeared to be related to single species of insect that preferred the wild-type over the cultivar (or vice versa), and these differences were not consistent over the growing season. These results suggest hemipteran abundance and diversity does not depend on the source of the plant material per se, but rather on the particular characteristics of the cultivar that distinguish it from the wild type. Certain characteristics are likely detrimental for wildlife, such as altering the reproductive biology of the plants by changing flower morphology or selecting for sterility. Other characteristics, perhaps nitrogen utilization or plant architecture, may actually benefit wildlife, and identifying these characteristics should be a priority for future research. This knowledge could enable plant breeders to select varieties that potentially outperform the wild-type plants in terms of their abilities to support wildlife. Possibly the best way to improve landscapes for wildlife is to include more plant species, regardless of whether they are cultivars or wild-type plants propagated from seed. We observed differences in abundance for more insect species when we compared different plant species than we did when we compared cultivars and wild-types within a plant species. These results fit both with the expectations of plant–insect interaction theory and with empirical evidence that plant species diversity drives arthropod diversity. Our results also suggest that planting unrelated species will do more to enhance insect diversity than will planting species within one family or within a few genera. Supplementary Data Supplementary material is available at Environmental Entomology online. Acknowledgments Funding for this research was provided by the Georgia Native Plant Society and the Garden Club of America. Cultivars of Schizachyrium were donated by Hoffman Nursery in Bahama, NC. Dr. Rick Hoebeke, curator of the Natural History Museum at the University of Georgia, confirmed the identifications of the insects. Thanks also to all the staff at the State Botanical Garden of Georgia, especially Heather Alley, whose advice and support made this research possible. References Cited Anderson , M. J . 2001 . A new method for non-parametric multivariate analysis of variance . Austral. Ecol . 26 : 32 – 46 . Bernays , E. , and M. Graham . 1988 . On the evolution of host specificity in phytophagous arthropods . Ecology . 69 : 886 – 892 . Google Scholar CrossRef Search ADS Bigelow , D. P. , and A. Borchers . 2017 . Major uses of land in the United States, 2012. EIB-179 . U.S. Department of Agriculture, Economic Research Service, Washington, DC . Burghardt , K. T. , and D. W. Tallamy . 2013 . Plant origin asymmetrically impacts feeding guilds and life stages driving community structure of herbivorous arthropods . Divers. Distrib . 19 : 1553 – 1565 . Google Scholar CrossRef Search ADS Burghardt , K. T. , D. W. Tallamy , and W. Gregory Shriver . 2009 . Impact of native plants on bird and butterfly biodiversity in suburban landscapes . Conserv. Biol . 23 : 219 – 224 . Google Scholar CrossRef Search ADS PubMed Burghardt , K. T. , D. W. Tallamy , C. Philips , and K. J. Shropshire . 2010 . Non-native plants reduce abundance, richness, and host specialization in lepidopteran communities . Ecosphere . 1 : 1 – 22 . Google Scholar CrossRef Search ADS Carman , J. G. , and S. L. Hatch . 1982 . Aposporous apomixis in schizachyrium (Poaceae, Andropogoneae) . Crop Sci . 22 : 1252 – 1255 . Google Scholar CrossRef Search ADS Comba , L. , S. A. Corbet , A. Barron , A. Bird , S. Collinge , N. Miyazaki , and M. Powell . 1999 . Garden flowers: insect visits and the floral reward of horticulturally-modified variants . Ann. Bot.-London . 83 : 73 – 86 . Google Scholar CrossRef Search ADS Cook , A. G. , and R. F. Denno . 1994 . Planthopper/plant interactions: feeding behavior, plant nutrition, plant defense, and host plant specialization , pp. 114 – 139 . In R. F. Denno and T. J. Perfect (eds.), Planthoppers: their ecology and management . Chapman and Hall , New York, NY . Corbet , S. A. , J. Bee , K. Dasmahapatra , S. Gale , E. Gorringe , B. La Ferla , T. Moorhouse , A. Trevail , Y. Van Bergen , and M. Vorontsova . 2001 . Native or exotic? Double or single? Evaluating plants for pollinator-friendly gardens . Ann. Bot.-London . 87 : 219 – 232 . Google Scholar CrossRef Search ADS Crawford , D. J. , and E. B. Smith . 1984 . Allozyme divergence and intraspecific variation in Coreopsis grandiflora (Compositae) . Syst. Botany . 9 : 219 – 225 . Google Scholar CrossRef Search ADS Cruden , R. W. , L. Hermanutz , and J. Shuttleworth . 1984 . The pollination biology and breeding system of Monarda fistulosa (Labiatae) . Oecologia . 64 : 104 – 110 . Google Scholar CrossRef Search ADS PubMed DeLong , D. M . 1948 . The leaf-hoppers, or Cicadellidae, of Illinois (Eurymelinae-Balcluthinae) . Ill. Nat. Hist. Surv. Bull . 24 : 97 – 376 . Ehrlich , P. R. , and P. H. Raven . 1964 . Butterflies and plants: a study in coevolution . Evolution . 18 : 586 – 608 . Google Scholar CrossRef Search ADS Fraenkel , G. S . 1959 . The raison d’ĕtre of secondary plant substances . Science . 129 : 1466 – 1470 . Google Scholar CrossRef Search ADS PubMed Gaston , K. J . 1996 . Species richness: measure and measurement , pp. 77 – 113 . In K. J. Gaston (ed.), Biodiversity: a biology of numbers and difference . Blackwell Science , Oxford . Gavier-Pizarro , G. I. , V. C. Radeloff , S. I. Stewart , C. D. Huebner , and N. S. Keuler . 2010 . Housing is positively associated with invasive exotic plant species richness in New England, USA . Ecol. Appl . 20 : 1913 – 1925 . Google Scholar CrossRef Search ADS PubMed Godfrey , R. M. , and M. T. Johnson . 2014 . Effects of functionally asexual reproduction on quantitative genetic variation in the evening primroses (Oenothera, Onagraceae) . Am. J. Bot . 101 : 1906 – 1914 . Google Scholar CrossRef Search ADS PubMed Gotelli , N. J. , and R. K. Colwell . 2001 . Quantifying biodiversity: procedures and pitfalls in the measurement and comparison of species richness . Ecol. Lett . 4 : 379 – 391 . Google Scholar CrossRef Search ADS Hurlbert , S. H . 1971 . The nonconcept of species diversity: a critique and alternative parameters . Ecology . 52 : 577 – 586 . Google Scholar CrossRef Search ADS PubMed Johnson , M. T. J . 2008 . Bottom-up effects of plant genotype on aphids, ants, and predators . Ecology . 89 : 145 – 154 . Google Scholar CrossRef Search ADS PubMed Johnson , M. T. J. , and A. A. Agrawal . 2005 . Plant genotype and environment interact to shape a diverse arthropod community on evening primrose (Oenothera biennis) . Ecology . 86 : 874 – 885 . Google Scholar CrossRef Search ADS Johnson , M. T. J. , and A. A. Agrawal . 2007 . Covariation and composition of arthropod species across plant genotypes of evening primrose (Oenothera biennis) . Oikos . 116 : 941 – 956 . Johnson , M. T. , M. J. Lajeunesse , and A. A. Agrawal . 2006 . Additive and interactive effects of plant genotypic diversity on arthropod communities and plant fitness . Ecol. Lett . 9 : 24 – 34 . Google Scholar PubMed Kempton , R. A. , and L. R. Taylor . 1978 . The Q statistic and the diversity of floras . Nature . 275 : 252 – 253 . Google Scholar CrossRef Search ADS Kim , J. , J. Chae , and T. H. Koo . 2007 . Variation in bird diversity in relation to habitat size in the urban landscape of Seoul, South Korea . Acta Ornithol . 42 : 39 – 44 . Google Scholar CrossRef Search ADS Mack , R. N. , and W. M. Lonsdale . 2001 . Humans as global plant dispersers: getting more than we bargained for . Bioscience . 51 : 95 – 102 . Google Scholar CrossRef Search ADS Magurran , A. E . 2004 . Measuring biological diversity . Blackwell Pub ., Maldan, MA . McKinney , M. L . 2006 . Urbanization as a major cause of biotic homogenization . Biol. Conserv . 127 : 247 – 260 . Google Scholar CrossRef Search ADS Mphosi , M. S. , and S. P. Foster . 2010 . Female preference and larval performance of sunflower moth, Homoeosoma electellum, on sunflower pre-breeding lines . Entomol. Exp. Appl . 134 : 182 – 190 . Google Scholar CrossRef Search ADS Narango , D. L. , D. W. Tallamy , and P. P. Marra . 2017 . Native plants improve breeding and foraging habitat for an insectivorous bird . Biol. Cons . 213 : 42 – 50 . Google Scholar CrossRef Search ADS Oksanen , J. , F. G. Blanchet , R. Kindt , P. Legendre , P. R. Minchin , R. B. O’Hara , G. L. Simpson , P. Solymos , M. H. H. Stevens , and H. Wagner . 2013 . vegan: Community Ecology Package. R package version 2.0–10 . http://CRAN.R-project.org/package=vegan Peck , D. C. , and V. Thompson . 2008 . Spittlebugs (Hemiptera:Cercopoidea) , pp. 3512 – 3516 . In J. L. Capinera (ed.), Encyclopedia of entomology , 2nd ed . Springer , Dordrecht, Netherlands . Pimentel , D. , and A. G. J. Wheeler . 1973 . Species and diversity of arthropods in the alfalfa community . Environ. Entomol . 2 : 659 – 668 . Google Scholar CrossRef Search ADS R Core Team . 2013 . R: a language and environment for statistical computing . R Foundation for Statistical Computing , Vienna, Austria . http://www.R-project.org/ Schaefer , C. W. , and N. P. Chopra . 1982 . Cladistic analysis of the Rhopalidae (Hemiptera, Coreoidea) with a list of food plants . Ann. Entomol. Soc. Am . 75 : 224 – 233 . Google Scholar CrossRef Search ADS Slater , J. A. , and R. M. Baranowski . 1990 . Lygaeidae of Florida (Hemiptera: Heteroptera) , vol. 14 . Arthropods of Florida and Neighboring Land Areas. Gainesville, FL . Spengel , C. K . 1793 . Das entdeckte Geheimnis der Natur im Bau und in der Befruchtung der Blumen . Friedrich Vieweg, Berlin, Germany . Tallamy , D. W . 2004 . Do alien plants reduce insect biomass ? Conserv. Biol . 18 : 1689 – 1692 . Google Scholar CrossRef Search ADS Tallamy , D. W. , M. Ballard , and V. D’Amico . 2010 . Can alien plants support generalist insect herbivores ? Biol. Invasions . 12 : 2285 – 2292 . Google Scholar CrossRef Search ADS Tenczar , E. G. , and V. A. Krischik . 2007 . Effects of new cultivars of ninebark on feeding and ovipositional behavior of the specialist ninebark beetle, Calligrapha spiraeae (Coleoptera: Chrysomelidae) . Hortscience . 42 : 1396 – 1399 . Wallner , A. M. , B. Molano-Flores , and C. H. Dietrich . 2013 . Using Auchenorrhyncha (Insecta: Hemiptera) to develop a new insect index in measuring North American tallgrass prairie quality . Ecol. Indic . 25 : 58 – 64 . Google Scholar CrossRef Search ADS Wang , Y. , U. Naumann , S. Wright , and D. Warton . 2014 . mvabund: statistical methods for analysing multivariate abundance data. R package version 3.9.3 . http://CRAN.R-project.org/package=mvabund Warton , D. I. , S. T. Wright , and Y. Wang . 2012 . Distance-based multivariate analyses confound location and dispersion effects . Methods Ecol. Evo. L . 3 : 89 – 101 . Google Scholar CrossRef Search ADS Wheeler , A. G . 2001 . Biology of the plant bugs (Hemiptera: Miridae): pests, predators, opportunists . Cornell University Press , Ithaca, NY . Wilson , E. O . 1987 . The little things that run the world (the importance and conservation of invertebrates) . Conserv. Biol . 1 : 344 – 346 . Google Scholar CrossRef Search ADS Wilson , S. W. , J. L. Smith , and A. H. Purcell . 1993 . An inexpensive vacuum collector for insect sampling . Entomol. News . 104 : 203 – 208 . Wimp , G. M. , W. P. Young , S. A. Woolbright , G. D. Martinsen , P. Keim , T. G. Whitham , and T. Meagher . 2004 . Conserving plant genetic diversity for dependent animal communities . Ecol. Lett . 7 : 776 – 780 . Google Scholar CrossRef Search ADS Young , O. P . 1986 . Host plants of the tarnished plant bug, Lygus lineolaris (Heteroptera, Miridae) . Ann. Entomol. Soc. Am . 79 : 747 – 762 . Google Scholar CrossRef Search ADS Zuefle , M. E. , W. P. Brown , and D. W. Tallamy . 2008 . Effects of non-native plants on the native insect community of Delaware . Biol. Invasions . 10 : 1159 – 1169 . Google Scholar CrossRef Search ADS © The Author(s) 2018. Published by Oxford University Press on behalf of Entomological Society of America. All rights reserved. For permissions, please e-mail: journals.permissions@oup.com. This article is published and distributed under the terms of the Oxford University Press, Standard Journals Publication Model (https://academic.oup.com/journals/pages/open_access/funder_policies/chorus/standard_publication_model)

Journal

Environmental EntomologyOxford University Press

Published: Aug 1, 2018

There are no references for this article.

You’re reading a free preview. Subscribe to read the entire article.


DeepDyve is your
personal research library

It’s your single place to instantly
discover and read the research
that matters to you.

Enjoy affordable access to
over 18 million articles from more than
15,000 peer-reviewed journals.

All for just $49/month

Explore the DeepDyve Library

Search

Query the DeepDyve database, plus search all of PubMed and Google Scholar seamlessly

Organize

Save any article or search result from DeepDyve, PubMed, and Google Scholar... all in one place.

Access

Get unlimited, online access to over 18 million full-text articles from more than 15,000 scientific journals.

Your journals are on DeepDyve

Read from thousands of the leading scholarly journals from SpringerNature, Elsevier, Wiley-Blackwell, Oxford University Press and more.

All the latest content is available, no embargo periods.

See the journals in your area

DeepDyve

Freelancer

DeepDyve

Pro

Price

FREE

$49/month
$360/year

Save searches from
Google Scholar,
PubMed

Create lists to
organize your research

Export lists, citations

Read DeepDyve articles

Abstract access only

Unlimited access to over
18 million full-text articles

Print

20 pages / month

PDF Discount

20% off