INTRODUCTION

Invasive nonnative species disrupt both fundamental ecosystem processes and ecosystem structure in their introduced ranges (Mooney and Hobbs 2000; Pejchar and Mooney 2009; Gutierrez et al. 2014). Invasive species typically lack the checks and balances, such as pathogens and predators, that constrain them in their native range (Hejda et al. 2009). They also frequently possess adaptations that permit rapid growth (Grotkopp and Rejmanek 2007; Van Kleunen et al. 2010; Graebner et al. 2012) and improve reproductive success (Engelhardt and Anderson 2011; Vervoort et al. 2011), and other competitive advantages, such as tolerance of a wide range of environmental conditions (Hejda et al. 2009; Higgins and Richardson 2014). The forests of the eastern United States have been negatively impacted by a number of invasive species in the modern era, from chestnut blight (Anagnostakis 2001) and Dutch elm disease in the early 1900s (Schlarbaum et al. 1998) to the much more recent emerald ash borer (Herms and McCullough 2014), as well as many problematic nonnative plant species (Devine and Fei 2011).

Invasive plant species often grow rapidly and aggressively, outcompeting native species (Vila and Weiner 2004) for resources such as light (Allred et al. 2010; McKinney and Goodell 2010), water (Cavaleri and Sack 2010; Nicotra and Davidson 2010), and nutrients (Mooney and Hobbs 2000; Matzek 2011; Gioria and Osborne 2014; Jo et al. 2015). As invasive plant species become dominant in an ecosystem, they can dramatically reduce the abundance and diversity of native species (Aguilera et al. 2010; Aronson and Handel 2011; Powell et al. 2013), disrupting ecosystem functions and structure (Pritekel et al. 2006; Hejda 2009). Many invasive plant species actively restrict the germination and growth of their native competitors by secreting allelopathic substances into the soil (Hierro and Callaway 2003; Pisula and Meiners 2010; Bauer et al. 2012; Cipollini et al. 2012; Del Fabbro and Prati 2015). Furthermore, the competitive adaptations of invasive species frequently render them difficult to remove from an ecosystem once they have become established (Simberloff 2003). Attempts at removal are costly and may not be effective at restoration of affected ecosystems (Zavaleta et al. 2001; Kettenring and Adams 2011; Frank et al. 2018).

Amur honeysuckle (Lonicera maackii Rupr.), native to east Asia, was introduced to North America in 1896 and was widely used for landscaping and wildlife habitat improvement (Luken and Thieret 1996). The fast-growing shrub can reach up to 6 m tall and is recognizable by its light-colored, attractive flowers and bright red berries (Luken and Thieret 1996), which are readily dispersed by wildlife (Castellano and Gorchov 2013; Guiden et al. 2015). It dominates the understory shrub layer of the affected forest (Medley 1997; Hartman and McCarthy 2008; Henkin et al. 2013) and, since its introduction, has spread throughout most of the eastern United States (Luken and Thieret 1996). This shrub is among the first plants to leaf out in the spring and one of the last species to lose its leaves in the fall (McEwan et al. 2009), which gives it a significant competitive advantage over native species for light (Luken et al. 1997; Chen and Matter 2017). This lengthy photosynthetic period also creates a longer shading period for the understory, altering the growing conditions of understory plants (Gorchov and Trisel 2003; Smith 2013; Chen and Matter 2017). Another competitive advantage of L. maackii is its allelopathic properties—allelopathic chemicals of L. maackii have been found to inhibit the germination of several native species (McEwan et al. 2010; Bauer et al. 2012; Cipollini et al. 2012). L. maackii is also able to grow across a wide range of environmental conditions (Lieurance and Landsbergen 2016), allowing it to successfully establish in ecosystems across the eastern United States (Luken et al. 1997; Parker-Gibson 2016). For a recent review of the ecology and impacts of L. maackii, see McNeish and McEwan (2016).

A number of studies have found that L. maackii presence and/ or abundance was associated with lower understory species richness and/or diversity. In southwestern Ohio forests, Hutchinson and Vankat (1997) found that herb cover was inversely related to L. maackii density in stands with ≥20% coverage. Similarly, Hartman and McCarthy (2008) found that L. maackii invasion was associated with a 4% decrease in herb layer species richness and a 57% reduction in herbaceous cover. In addition, Collier et al. (2002) reported effects on tree seedlings, with a 41% reduction in species richness and 68% reduction in density of seedlings of species with canopy potential.

These observations led to a series of experimental studies testing the impacts of L. maackii on planted understory species or evaluating the effects of L. maackii removal on understory recovery. For example, L. maackii significantly reduced survival and fecundity of transplanted native annuals in Ohio (Gould and Gorchov 2000). Similarly, although effects on survival of perennial herbs did not persist over five seasons, effects on growth and fecundity were persistent, indicating long-term population-level impacts of L. maackii on native understory plants (Miller and Gorchov 2004). Furthermore, L. maackii presence (compared to sites from which L. maackii was removed and sites in which L. maackii was absent) was related to lower abundance (but not diversity and richness) of understory herbaceous species (Christopher et al. 2014), although Peebles-Spencer et al. (2017) reported significant effects of L. maackii on herbaceous species richness. Finally, in addition to experimental L. maackii impacts on herbaceous species, Loomis et al. (2015) reported that L. maackii presence was associated with significantly lower sugar maple (Acer saccharum) survival.

The current study was conducted to evaluate the association of L. maackii with flowering spring flora in Raven Run Nature Sanctuary, a 297 ha nature sanctuary located in the Inner Bluegrass region of Kentucky. Raven Run falls within the Kentucky River Palisades, an area considered to be of high conservation value (Campbell et al. 1995; Barnes 2002). Among other plants of conservation interest, the spring wildflowers at Raven Run have great ecological and cultural value and attract many visitors to the nature sanctuary (Campbell et al. 1995). Some studies suggest that L. maackii impacts spring flora more intensely than it affects summer plant species (Miller and Gorchov 2004; Christopher et al. 2014; Shields et al. 2015; Chen and Matter 2017; Peebles-Spencer et al. 2017), suggesting that the diverse spring wildflower community at Raven Run may be particularly vulnerable to L. maackii encroachment. This study investigates patterns of abundance, species richness, and species diversity of flowering spring wildflowers, as well as total herbaceous groundcover, across an L. maackii density gradient (high, low, and no L. maackii), from March to April 2019. These data contribute to understanding of L. maackii-invaded plant communities in the sensitive and ecologically valuable Kentucky River Palisades region.

METHODS AND MATERIALS

Raven Run is a nature sanctuary located in the Inner Bluegrass region, just outside of the city of Lexington, Kentucky, along the Kentucky River Palisades (Campbell et al. 1995; Raven Run Nature Sanctuary 2020). The lithology of the study area is Ordovician limestone interbedded with shale and sandstone, which gives rise to highly fertile soils primarily of the McAfee and Fairmount series (Campbell et al. 1995). Steeper slopes at Raven Run are dominated by a mature forest (Campbell et al. 1995), but L. maackii is abundant, especially near the forest edges and in old-field sites undergoing succession (Wilson et al. 2013). Researchers worked with Raven Run personnel to select three sites, each of which exhibited a clear L. maackii density gradient ranging from high density to low density to no L. maackii (Figure 1). Sites were selected prior to L. maackii greenup, in February 2019. While high-density L. maackii plots were consistently closest to, and no-L. maackii plots consistently farthest away from, the forest edge, sites were selected based on the abundance of mature trees throughout the site (indicative of similar landuse history), and plots were established a minimum of 50 m from the forest edge to minimize edge effects. At each site, two 5 × 2 m plots were established at each of the three L. maackii densities (high, low, or no L. maackii) for a total of 18 plots (2 plots × 3 densities × 3 sites). Plots were established a minimum of 10 m apart. One corner of each plot was marked using a stamped aluminum tree tag anchored to a 30 cm piece of rebar to enable resampling in the future. This paired-plot experimental design was chosen to support L. maackii removal in the future (one of each plot pair would be cleared of L. maackii, the other left alone as a control). To evaluate selection of “high” and “low” density plots, the canopy area of all L. maackii individuals within each plot was estimated in Spring 2019. The crown diameter of every L. maackii individual rooted within a given plot was measured twice (two diameter measurements approximately perpendicular to each other). An estimated radius was calculated from the average diameter and used to estimate the canopy area for the individual. Total L. maackii canopy area estimates were calculated by adding the individual area estimates for all L. maackii individuals measured in the given plot. (Given that all L. maackii individuals rooted in the plot were measured, estimated total canopy area included canopy extending beyond the plot boundaries and thus exceeded plot area in some cases.) Plots designated as low density had fewer L. maackii individuals (mean = 5.6, SE = 1.3) and lower total L. maackii canopy area (mean = 3.98 m², SE = 0.71) than plots designated as high density (mean abundance = 9.3, SE = 0.60; mean total canopy area = 17.7 m², SE = 0.43), supporting this categorization.

Spring flora were surveyed by undergraduate students enrolled in HON 152 (Restoration Ecology in the Commonwealth) as part of a class research project. Surveys were conducted four times from March to April of 2019, using six quadrats (0.61 m × 0.61 m) randomly placed in each plot without overlapping (6 quadrats × 18 plots = 108 quadrats). Quadrats were relocated at each sample period. In each quadrat, percent groundcover by herbaceous vegetation (excluding woody plants, but including native and nonnative, flowering and nonflowering herbaceous plants) was visually estimated in multiples of 5%. All flowering plants within each quadrat were counted and identified according to Barnes and Francis (2004), with naming authorities added and species names updated according to the USDA Plants Database (USDA NRCS 2021). Nonflowering individuals were excluded from the count to ensure accurate identification. Tallying individual plants was straightforward for some species, but note that the spreading habit of some species, such as Claytonia virginica, rendered identifying independent individuals difficult, and is likely to have biased abundance and diversity estimates. Abundance of flowering individuals was calculated by summing the total number of flowering individuals observed. Similarly, flowering species richness was calculated by summing the total number of species of flowering individuals observed. Finally, flowering species diversity was calculated as Shannon's Diversity Index (H′).

Figure 1.

Satellite view of Raven Run Nature Sanctuary, located along the Kentucky River Palisades, within the Inner Bluegrass region, central Kentucky. Location of surveyed sites within Raven Run marked with pins.

RESULTS

Over the course of the study, a total of 18 species of flowering spring flora were observed, with Cardamine douglassii Britton, Collinsia verna Nutt., Claytonia virginica L., Delphinium tricorne Michx., and Trillium sessile L. the most abundant—individuals from the remaining 13 species comprised approximately 15% of the total flowering individuals recorded (Table 1). Percent groundcover ranged from 4% to 86% (mean: 23 ± 2 SE). There were 0–84 individual flowering plants per plot (21.7 ± 2.3 SE) from 0–7 species (2.8 ± 0.2 SE) with a diversity of 0–1.6 (0.6 ± 0.1 SE). Residuals were patterned for the richness analyses, which was fixed by using a Poisson distribution instead of a normal distribution. The richness model also had trouble converging, so we fit the model using all available optimizers (Bates et al. 2014). These optimizers all produced the same results, indicating a reliable model. All other models converged and had homoscedastic residuals. Groundcover, abundance, richness, and diversity all tended to increase over the course of the study.

Table 1.

Mean abundance of flowering spring wildflowers at Raven Run Nature Sanctuary, under varying densities of Lonicera maackii (high, low, and no L. maackii), surveyed over four weeks.

Total herbaceous groundcover increased over the season and, at the end of the season, varied by L. maackii density. There was 1.8–2.6 times more groundcover during the fourth week than the first week in all three densities, with some differences among intermediate weeks in the low- and no-density plots (Figure 2). In week four, no-density plots had 1.7 times more groundcover than high-density plots (Table 2, Figure 2).

Abundance of flowering plants increased over time and varied among densities. For high-density plots, the only difference was that abundances in week 4 were 1.6 times higher than in week 1 (Figure 2). For low- and no-density plots, abundances tripled between weeks 1 and 4, with differences among all weeks except weeks 2 and 3 (Figure 2). The no-density plots had 2.8–3.8 times higher abundance than the high-density plots during all weeks, and the low-density plots had 2.0–3.1 times greater abundance than the low-density plots during all but the first week (Table 2, Figure 2).

Species richness of flowering plants increased over time and was highest in the no-density plots. Richness was 4.5 times higher in week 4 than in week 1 in the high-density plots (Figure 2). In the low-density plots, week 4 had 4.2 times higher richness than week 1 and had 2.1 times higher richness than week 2. Week 3 had 3.3 times higher richness than week 1 (Figure 2). Richness did not vary between weeks for the no-density plots (Figure 2). No-density plots had 2.5–4.0 times greater richness than the high-density plots during weeks 1–3, while no-density plots had 2.3–2.7 times greater richness than the low-density plots during weeks 1 and 2 (Table 2, Figure 2).

Diversity of flowering plants was 9.9 times higher in week 4 than week 1 and 2.6 times higher in week 4 than week 2 in the high-density plots (Figure 2). In the low-density plots, weeks 3 and 4 had 11.1–12.2 times higher diversity than week 1, and 2.5–2.7 times greater diversity than week 2 (Figure 2). There were no differences among weeks in the no-density plots (Figure 2). No-density plots had 2.1–9.5 times higher diversity than high-density (weeks 1, 2, and 3) and 2.8–9.3 times higher diversity than low-density plots (weeks 1 and 2; Table 2, Figure 2).

DISCUSSION

This study found that flowering spring flora differed across an L. maackii density gradient. Flowering individuals of Viola blanda, Delphinium tricorne, Trillium sessile, and Cardamine douglassii exhibited similar abundances across the gradient of L. maackii density. These observations suggest that the flowering of particular spring wildflower species may be relatively unaffected by L. maackii presence at this site. In contrast, average species richness of flowering wildflower species was higher in no-L. maackii plots than in both low- and high-density plots, but similar between low- and high-density plots. In our study, flowering individuals of several species were found only in no-L. maackii plots: Viola pubescens, Dicentra cucullaria, Jeffersonia diphylla, and Stellaria pubera. Miller and Gorchov (2004) reported that V. pubescens populations transplanted under L. maackii exhibited suppressed growth and reduced flowering. The species in this study for which flowering individuals were found only in plots without L. maackii may be similarly suppressed; however, experimental studies will be necessary to elucidate whether L. maackii presence or some uncontrolled variable is causing this association at this site.

Figure 2.

Total herbaceous ground cover, and abundance, species richness, and diversity (Shannon's H′) of flowering spring flora, surveyed over four weeks and over a Lonicera maackii density gradient (high, low, and no L. maackii). Error bars represent ± 1 SE; means with different lowercase letters are statistically different (P < 0.05) across L. maackii density categories. If no differences were found, letters are not included.

In addition to significant associations of L. maackii density with richness and abundance of flowering plants, our analysis found significant effects of time on flowering spring wildflower abundance, species richness, and diversity. The significant species diversity effect was particularly interesting, indicating that the relationship between L. maackii density and flowering spring wildflower diversity was most pronounced in the early spring, becoming negligible over time. Because diversity is directly related to both species evenness and richness, low diversity can be driven by low richness or by community dominance by only a few species. In our study, diversity of flowering plants was relatively high in no-L. maackii plots in the first two weeks, but was similar to plots with L. maackii by weeks 3 and 4. These patterns in diversity were related to (1) increasing richness over time, driving increasing diversity in L. maackii plots, and (2) dominance by only a few species in no-L. maackii plots in later weeks (e.g., Claytonia virginica in week 3 and Collinsia verna in week 4), driving lower diversity. While diversity data alone would suggest that further research should target L. maackii-associated flowering plants in the early season, species with flowering individuals found only in no-L. maackii plots were recorded predominantly during later surveys. Thus, while further research is necessary to characterize the temporal patterns of the association of L. maackii with the apparent suppression of flowering in particular spring flora, such research should seek to characterize the entire period of the spring wildflower emergence, rather than focusing on the early season to the exclusion of later time periods.

Overall, this study reports the association of L. maackii abundance with lower richness and abundance, as well as lower early-season species diversity, of flowering spring wildflowers. This relationship is likely tied to the early leaf-out of L. maackii, which has been found to reduce understory light availability and significantly reduce soil temperatures in the early season (Chen and Matter 2017). Gorchov and Trisel (2003) found that aboveground competition was a significant contributing factor to the negative effects of L. maackii on native tree seedlings. Furthermore, Miller and Gorchov (2004) found that L. maackii reduced growth and fecundity of three native understory herbs, an effect that they attributed to light competition. Similarly, McKinney and Goodell (2010) reported that shading by L. maackii significantly reduced reproduction in a native understory plant. While this mechanism appears fairly consistent in the literature, further research is necessary to elucidate the particular mechanism of the observed associations between L. maackii and flowering plants in the present study.

Table 2.

P-values comparing differences among Lonicera maackii densities during a study of ground cover, abundance, species richness, and diversity (Shannon's H′) for flowering spring wildflowers. Plots were surveyed over four weeks and over a Lonicera maackii density gradient (high, low, and no L. maackii). Comparisons are considered significant if P ≤ 0.05.

In some cases, removing L. maackii has led to increased species richness and abundance in understory herbaceous plants. For example, Thompson and Poindexter (2011) reported a threefold increase in species richness and significant native species recovery after removal of L. maackii from a central Kentucky site. Similarly, Conover and Sisson (2016) reported significant recovery of native understory plants after removal of L. maackii, and Hartman and McCarthy (2004) found that L. maackii removal enhanced survival of planted native tree seedlings. While these studies support L. maackii removal as an approach to ecological restoration of impacted sites, they also caution that one-time removal may not be sufficient (Conover and Sisson 2016). Because of significant L. maackii presence in the seedbank (Thompson and Poindexter 2011) and persistence of L. maackii and other invasive species (e.g., Alliaria petiolata and Euonymus fortunei) in surrounding areas (Conover and Sisson 2016), continuous vigilance over time is likely necessary to prevent recurring invasions of protected sites. Further research will be necessary to understand how management activities such as L. maackii removal will influence recovery of native species, especially those not found in L. maackii plots in this study, on the sites surveyed in the present study (especially given the prevalence of L. maackii at Raven Run and our anecdotal observations that other invasive species, such as A. petiolata and E. fortunei, are also present at some of these sites).

Overall, this study is consistent with both the observational and experimental literature associating L. maackii presence with negative impacts on native understory plants. Our study specifically characterizes associations of L. maackii with flowering understory plants in an area of conservation concern in central Kentucky—the Kentucky River Palisades. However, because of the nature of our study, causation cannot be inferred, and potential contributions of uncontrolled variables must be considered. For example, the observed effects could be related to underlying environmental structures favoring invasion by L. maackii, such as distance to forest edge and soil conditions. Further research will be necessary to elucidate the mechanisms of the observed associations in these sites.