Invasive grass (Microstegium vimineum) indirectly benefits spider community by subsidizing available prey

Abstract Invasive plant species cause a suite of direct, negative ecological impacts, but subsequent, indirect effects are more complex and difficult to detect. Where identified, indirect effects to other taxa can be wide‐ranging and include ecological benefits in certain habitats or locations. Here, we simultaneously examine the direct and indirect effects of a common, invasive grass species (Microstegium vimineum) on the invertebrate communities of understory deciduous forests in the eastern United States. To do this, we use two complementary analytic approaches to compare invaded and reference plots: (a) community composition analysis of understory arthropod taxa and (b) analysis of isotopic carbon and nitrogen ratios of a representative predatory spider species. Invaded plots contained a significantly greater abundance of nearly all taxa, including predators, herbivores, and detritivores. Spider communities contained over seven times more individuals and exhibited greater species diversity and richness in invaded plots. Surprisingly, however, the abundant invertebrate community is not nutritionally supported by the invasive plant, despite 100% ground cover of M. vimineum. Instead, spider isotopic carbon ratios showed that the invertebrate prey community found within invaded plots was deriving energy from the plant tissue of C3 plants and not the prevalent, aboveground M. vimineum. Synthesis and applications. We demonstrate that invasive M. vimineum can create non‐nutritional ecological benefits for some invertebrate taxa, with potential impacts to the nutritional dynamics of invertebrate–vertebrate food webs. These positive impacts, however, may be restricted to habitats that experience high levels of ungulate herbivory or reduced vegetative structural complexity. Our results highlight the importance of fully understanding taxon‐ and habitat‐specific effects of invading plant species when prioritizing invasive species removal or management efforts.

tified, indirect effects to other taxa can be wide-ranging and include ecological benefits in certain habitats or locations.
2. Here, we simultaneously examine the direct and indirect effects of a common, invasive grass species (Microstegium vimineum) on the invertebrate communities of understory deciduous forests in the eastern United States. To do this, we use two complementary analytic approaches to compare invaded and reference plots: (a) community composition analysis of understory arthropod taxa and (b) analysis of isotopic carbon and nitrogen ratios of a representative predatory spider species.
3. Invaded plots contained a significantly greater abundance of nearly all taxa, including predators, herbivores, and detritivores. Spider communities contained over seven times more individuals and exhibited greater species diversity and richness in invaded plots. 4. Surprisingly, however, the abundant invertebrate community is not nutritionally supported by the invasive plant, despite 100% ground cover of M. vimineum.
Instead, spider isotopic carbon ratios showed that the invertebrate prey community found within invaded plots was deriving energy from the plant tissue of C 3 plants and not the prevalent, aboveground M. vimineum.

Synthesis and applications.
We demonstrate that invasive M. vimineum can create non-nutritional ecological benefits for some invertebrate taxa, with potential impacts to the nutritional dynamics of invertebrate-vertebrate food webs. These positive impacts, however, may be restricted to habitats that experience high levels of ungulate herbivory or reduced vegetative structural complexity. Our results highlight the importance of fully understanding taxon-and habitat-specific effects of invading plant species when prioritizing invasive species removal or management efforts.

| INTRODUC TI ON
Direct effects of one species on another can be relatively easy to quantify, particularly when considering simplified, two species interactions. However, such straightforward exchanges are rarely found in nature, with secondary and tertiary effects pervading into subsequent trophic levels and affecting multiple species (Preisser, Bolnick, & Benard, 2005;Werner & Peacor, 2003). Indirect interactions of one species on another through a third species can affect species' density and can impose behavioral changes to the indirectly affected species (Abrams, 1995;Werner & Peacor, 2003). Given the prevalence and strength of indirect species interactions, these relationships can often exert more influence over ecological communities that can direct interactions between a predator and herbivore (Seibold, Cadotte, MacIvor, Thorn, & Müller, 2018) and have been documented in various aquatic and terrestrial habitats (Fletcher et al., 2019;Vilá et al., 2011). Despite their ecological significance, indirect effects among species can be complex and difficult to detect, and as a result are often understudied and overlooked.
The ecological consequences of the invasion of certain non-native plant species are well-supported throughout the literature, with direct, negative effects documented on a wide range of native taxa.
Beyond directly impacting both the behavior and density of other taxa, invasive plant species can also cause negative indirect effects mediated through alteration of the behavior and density of other, often herbivorous, species (Vilá et al., 2011;White, Wilson, & Clarke, 2006). Nevertheless, there exists an expanding literature detailing both direct and indirect impacts considered positive or beneficial for a species or ecological community resulting from non-native plant invasion (McCary, Mores, Farfan, & Wise, 2016;Tymkiw, Bowman, & Shriver, 2013). Some invading plant species, often with large and showy inflorescences, can increase floral density and food availability for pollinators throughout the season and potentially at times of reduced floral availability (Davis, Kelly, Maggs, & Stout, 2018;Russo, Nichol, & Shea, 2016). The complex and novel architecture of many invasive plant species can also provide enhanced structure for primary consumers to effectively hide from predators (Dutra, Barnett, Reinhardt, Marquis, & Orrock, 2011;Malo et al., 2012). Contrastingly, changes in plant structure can also directly benefit actively hunting predators (Loomis, Cameron, & Uetz, 2014) or those that use passive hunting techniques (Dudek, Michlewicz, Dudek, & Tryjanowski, 2016;Pearson, 2009 Pearson, 2009Pearson, , 2010 or on specific predatory functional guilds (Smith-Ramesh, 2017) and not on predatory communities of the forest understory as a whole.
Microstegium vimineum is a grass species from eastern Asia that invades forest edges and disturbed habitats in the eastern half of the U.S. (Flory, Long, & Clay, 2011;Huebner, 2010). The species forms dense mats across the forest floor by spreading from stolons, ultimately reducing native tree seedling density, growth, and diversity (Oswalt, Oswalt, & Clatterbuck, 2007;Brewer, 2011), as well as overall native plant species cover (Adams & Engelhardt, 2009). The forest floor in invaded habitats in Maryland, often in areas with high densities of vertebrate herbivores, is generally simplified in structure, due in part to the growth habit of M. vimineum (Civitello, Flory, & Clay, 2008;Landsman pers. obs.). Given its growth habit and stand density, the presence of M. vimineum also alters abiotic characteristics of the near-ground forest environment. The forest floor experiences an increase in solar irradiation within M. vimineum stands, which decreases relative humidity and increases microhabitat temperatures (Civitello et al., 2008). Such climatic changes subsequently alter the ability of M. vimineum patches to host invertebrate populations: M. vimineum ground cover has been shown to reduce the overall diversity of soil microarthropods by greatly increasing the abundance of mites and subsequently reducing community evenness (McGrath & Binkley, 2009). The abundance of cicadellidae planthoppers, as well as acridid and gryllid Orthoptera, was also found to be higher in M. vimineum patches (Marshall & Buckley, 2009).
Conversely, M. vimineum can also reduce the abundance of Blattodea and chrysomelid beetles, as well as the abundance and survival of hard tick species in the Ixodidae (Civitello et al., 2008;Marshall & Buckley, 2009).
Microstegium vimineum directly affects forest floor invertebrates by physically altering their habitat; however, knowledge of subsequent indirect effects to the predators that utilize those affected invertebrates resulting from invasion-driven forest floor changes is lacking. Physiognomic changes in structural complexity resulting from invasive plant species with similar, mat-forming growth habits have been shown to alter community structure, composition, and species abundance of the spider community (Bultman &
Here, our objective was to better understand the direct and indirect impacts that the invasive M. vimineum has on the invertebrate prey and predator communities and to demonstrate whether such invasions have the potential to affect predator-prey interactions in the forest understory. As M. vimineum invades, it can alter near-ground vertical plant structure and the availability of palatable food resources, both of which may directly and indirectly alter the composition of the invertebrate community. We hypothesized that the abundance of herbivorous insects would decrease in M. vimineum patches due to suppression of native vegetation and food resources, and that, conversely, dipteran species in detrital food webs would prefer the sheltered microhabitat created by dense M. vimineum stands. We similarly predicted changes to the spider community in invaded habitats: While spider richness and diversity would not change within M. vimineum, we expected to see changes in community composition as alterations to near-ground plant structure negatively impact web-building taxa and benefit active hunting spiders. Finally, we hypothesized that web spider isotopic nitrogen signatures would reflect the greater proportion of available detritivorous insects in spider diet and that spider isotopic carbon would reflect the greater relative contribution of C 4 M. vimineum biomass in invaded areas. To study invertebrate community response to M. vimineum, we used a paired plot design across our study area. Within each of the three parks, we opportunistically located 16 patches of M. vimineum at least 10 m by 10 m in size, and at least 10 m away from the forest edge. Sites were selected that contained greater than or equal to 80% visual ground cover of M. vimineum. We established the center of a 6 m by 6 m square plot at the approximate center of the patch.

| MATERIAL S AND ME THODS
Paired with each M. vimineum plot, we also established a reference plot of the same size, without M. vimineum. We used random integers to select an azimuth from which to establish reference plots, 20 m from the edge of the M. vimineum patch, at least 10 m from the forest edge, and within the same vegetation community type.
Within individual square plots for both invaded and reference habitats, we estimated ground cover of M. vimineum in four 1 m 2 square subplots, 1 m away from the plot center in the four cardinal directions, and used the mean ground cover estimate for each plot. We also measured understory vegetative structure using a 2.0 m tall profile board placed in the center of the plot. We estimated the percentage of the board that was obscured by vegetation between 0.5 and 2.0 m in height and used the mean of the four values for each plot in analyses.
We conducted vacuum sampling within each individual square plot in mid-July 2017. We vacuumed insects and arachnids between 0.5 and 2.0 m above the ground surface in order to avoid forest floor and fossorial taxa. We used a commercially available leaf blower and vacuum (Black+Decker LSWV36) with a 2-gallon paint strainer bag affixed to the intake tube to vacuum vegetative surfaces, spider webs, and other spaces between vegetation in each plot. We vacuumed throughout the entirety of each 36 m 2 plot for a standardized 7 min.
After sampling, we euthanized collected arthropods using ethyl acetate, removed vegetative debris, and placed invertebrates in 70% ethyl alcohol. We identified insects to order using Triplehorn and Johnson (2005). However, certain orders were further subdivided if palatability to forest spiders greatly differed within groups: Weevils (Curculionidae) were classified separately from other Coleoptera, predatory assassin bugs (Reduviidae) and damsel bugs (Nabidae) were separated from herbivorous Hemiptera, ants (Formicidae) were considered distinct from the other Hymenoptera, and Lepidoptera were subdivided into caterpillars and adults. Spiders were identified to genus or species when possible using Ubick, Paquin, Cushing, and Roth (2017). Any specimens not identified to this taxonomic resolution, including those too damaged for identification or recently hatched individuals, were excluded from community analysis and analyses of diversity and richness.
For a closer examination of changes in nutritional dynamics, we also specifically sampled a representative orb-weaving spider common to eastern deciduous forests (Tetragnathidae: Leucauge venusta Walckenaer 1841) prior to vacuum sampling for the remaining invertebrate community. We sampled L. venusta as this species was commonly found throughout the study area and has a wide geographic distribution across much of eastern North America. This species spins a relatively horizontal orb web with attachment points in low-growing vegetation in wooded areas. L. venusta and other relatively small orb-weaving spiders prey mostly on flies (Diptera), leafhoppers (Hemiptera: Cicadellidae), and other small, alate true bugs, and beetles (Coleoptera). We collected mature female spiders by hand from plots in early July 2017. Samples were immediately frozen and individually dried for 24 hr at 60°C. We then weighed the dry spider samples to obtain body mass. Spiders were then individually ground, homogenized, and encapsulated. We analyzed individual spiders for δ 15 N and δ 13 C using a continuous flow isotope ratio mass spectrometer (DELTA V Plus; Thermo Fisher Scientific) and elemental analyzer (NC 2,500, Carlo Erba; 95% CI ±0.5‰).
Isotopic nitrogen ratios can provide information on a predator's diet, while isotopic carbon can reveal photosynthetic pathways in sampled organisms: M. vimineum, a C 4 species, maintains δ 13 C levels between −13‰ and −15‰, while C 3 plants generally have δ 13 C values near −27‰ (Bradford et al., 2010;Hyodo, 2015 Thus, we excluded this metric from all analyses. Similarly, we also regressed the Shannon-Weiner diversity, abundance, and richness of spider and insect communities in models with the same independent variables. For those linear models and variables that exhibited non-normality, including models with spider body mass and isotopic carbon and nitrogen, we employed a square root transformation of the dependent variable. For count data, including invertebrate abundance and richness, we used generalized linear models with a negative binomial probability distribution using glm.nb in the MASS package (Venables & Ripley, 2002). We obtained p-values from the likelihood ratio test statistic using anova.glm. For spider and insect community data, we calculated the Euclidean distance between taxa within community matrices after Hellinger transformation (Borcard, Gillet, & Legendre, 2011;Legendre & Legendre, 2012;Rao, 1995).
We used the Euclidean distance of Hellinger-transformed data as these data are metric and considered robust in ordination analyses (Legendre & Gallagher, 2001). We then conducted permutational multivariate analysis of variance using adonis2 in the vegan package to understand how communities differ in patches of M. vimineum after testing for multivariate homogeneity of variance using betadisper and anova.betadisper (Anderson, Ellingsen, & McArdle, 2006;McArdle & Anderson, 2001;Oksanen et al., 2017). To test for significance, we examined the marginal effects of variables after 20,000 permutations.
The abundance of prey palatable to forest spiders, including beetles, TA B L E 1 Mean abundance of invertebrate groups within invaded and reference plots throughout study area more prey (G 2 = 9.881; p = .0017). M. vimineum patches contained more orb web-building (G 2 = 11.013; p = .0009), space web-building (G 2 = 18.847; p < .0001), and hunting (G 2 = 9.765; p = .0018) spiders,   We had also hypothesized that, due to incongruent evolutionary history, native herbivore abundance would be depressed within M. vimineum carbon ratios are generally between −13 and −15‰ (Bradford et al., 2010). Considering a potential stepwise increase of 0.5‰ per trophic level, it is unlikely that potential prey in this invaded habitat, including detritus-based Diptera and phytophagous Hemiptera, are utilizing the invasive species. If so, spider δ 13 C values should be much closer to that of C 4 plants. Given the relative abundance of flies deriving from detrital food webs in invaded plots, it is likely that the spider community in our study area is largely supported by the detritivores (Hyodo et al., 2010;McNabb, Halaj, & Wise, 2001;Miyashita, Takada, & Shimazaki, 2003). These Our findings may be restricted to those areas with high ungulate densities and resulting depauperate understory vegetation: Deer densities in our sampling area were between 46 and 66 deer/km 2 while densities at much lower levels can cause significant impacts to understory forest vegetation and structure (Horsley, Stout, & de Calesta, 2003;Tilghman, 1989 Vegetation structure is a critical factor affecting the ability of a habitat to support understory spiders in forests with extensive ungulate herbivory (Landsman & Bowman, 2017;Miyashita et al., 2004;Takada et al., 2008). In habitats where additional native or non-native structure exists, the relationship between M. vimineum and the invertebrate community could reverse, as was noticeable in the Antietam plot pair and evident in the significance of the interaction term between Park and plant invasion on many of the response variables. This relationship has also been noted with forest birds in similar habitats (Tymkiw et al., 2013 which also exhibit enriched isotopic nitrogen, constitute an increased proportion of spider diet in invaded habitats (Hyodo, 2015). As guanotelic organisms such as spiders use nitrogen efficiently, diet shifts that include a greater relative percentage of nitrogen-poor prey may affect individual body condition and fecundity (Toft & Wise, 1999) (Hodkinson, Coulson, Harrison, & Webb, 2001;Kitchell et al., 1979;Schmitz, Hawlena, & Trussell, 2010). Changes to the arthropod food web, brought about through replacement of native plants by invasive species, are largely undocumented and potentially pervasive across forest habitats with introduced and invasive plants.

F I G U R E 1 Boxplots showing effects of
The results of this study clearly indicate the extent to which invading plant species can cause both direct and indirect effects on multiple taxa as well as different trophic and functional groups. In this particular example, the invading species benefited invertebrate communities, albeit with as yet unknown subsequent impacts on higher trophic levels. Such positive effects are certainly species-specific and not congruent with all species invasions nor are effects consistent across affected taxa (Fletcher et al., 2019).  (Mack et al., 2000;Qian & Ricklefs, 2006), land managers and conservation biologists must consider any subsequent, cascading impacts in order to prioritize invasive plant management efforts.

ACK N OWLED G M ENTS
We would like to thank D. Nelson and R. Paulman with the Central Appalachians Stable Isotope Facility at UMCES for stable isotope analysis and M. Hall, K. Haines, and B. Sammarco for assistance in data collection and insect identification. We would also like to thank the Maryland Native Plant Society for partial financial assistance and two anonymous reviewers for their extremely helpful and constructive review of this manuscript. The views and opinions represented herein are to be considered a work of the authors and not that of the U.S. Department of the Interior. Any use of trade, firm, or product names is for descriptive purposes only and does not imply endorsement by the U.S. Government.