Present address: School of Biological Sciences, University of Sydney, Sydney, New South Wales 2006 Australia. E-mail: email@example.com
Corresponding Editor: G. A. Fox.
Plants serve as both basal resources and ecosystem engineers, so plant invasion may exert trophic influences on consumers both via bottom-up processes and by altering the environmental context in which trophic interactions occur. To determine how these mechanisms affect a native predator we used a mark–recapture study in eight pairs of 58-m2 field enclosures to measure the influence of Japanese stilt grass invasion on 3200 recently metamorphosed American toads. Toad survivorship was lower in invaded habitats despite abiotic effects that favor amphibians. Prey densities were also lower in invaded habitats, but growth was unaffected. Frequent spider predation events in invaded habitats led us to use factorial field cage manipulations of stilt grass and lycosid spiders to determine if invasion increases predation rates. Spiders persisted at higher densities in the presence of stilt grass, and toad survival was lowest in cages with both grass and spiders. Invasion alone did not significantly reduce toad survival. Our results demonstrate that despite prey reductions and abiotic effects, it is increased spider persistence that reduces toad survival in invaded habitats. Invasion therefore affects resident forest floor consumers by modifying trophic interactions between native species, causing structurally mediated reductions in intraguild predation rates among spiders, with cascading implications for toad survival.
Indirect effects are increasingly recognized as integral to the functioning of ecosystems, contributing to the structuring of trophic relationships and influencing basic services (Schmitz 2010). However, as these influences can be complex, involving interactions between trophic dynamics, habitat structure, and behavior, understanding the mechanisms behind them can be challenging (Schmitz 2010). Plant invasion is a pervasive process that has the potential to drive indirect effects on consumers living in invaded habitats. As some exotic plants are relatively herbivore-free in their introduced range (Keane and Crawley 2002), these species have limited potential to drive direct, positive, bottom-up effects via herbivory. However, invaders may indirectly influence resident species through other mechanisms. For example, arthropod communities may respond to losses in native plant diversity following invasion (Carvalheiro et al. 2010, Simao et al. 2010), invasion-driven alterations in nutrient inputs can alter the structure and function of belowground communities (Kourtev et al. 2002), and invaders can affect abiotic properties in ways that impact consumer performance (Byers et al. 2010).
Structural alterations associated with invasion may also drive indirect effects on resident consumers. Invaders often differ structurally from native plant communities. As invasive plants grow, the spatial architecture of invaded ecosystems is frequently altered (Crooks 2002), with potential implications for the trophic connections within these landscapes (Jones et al. 1997). For example, indirect effects on arthropod prey have been documented where invasion increases attachment points for web-building spiders (Miyashita and Takada 2007, Pearson 2010), but structural effects may also influence other consumers. An important trophic linkage that may be affected by this process is the intraguild or cannibalistic linkage between species or individuals filling similar trophic positions. The strength of this linkage, in which predators prey upon each other as well as a shared prey base, is often especially responsive to contextual changes (reviewed by Janssen et al. 2007). When structural complexity increases, positive effects are expected for the persistence of intraguild prey (Langellotto and Denno 2004, Janssen et al. 2007), potentially increasing pressure on shared prey (Finke and Denno 2002). Therefore, by modifying ecosystem structure, invading plants have the potential to promote one group via positive, structurally mediated effects on intraguild prey through refuge provisioning, to the detriment of another, via negative, density mediated effects on their shared prey.
We investigated the detritally mediated bottom-up, structurally mediated top-down, and abiotic effects of exotic grass invasion on the performance of recently metamorphosed American toads (Anaxyrus [Bufo] americanus). We first measured toad survival and growth in field enclosures spanning natural invasion fronts, and related observed patterns to abiotic conditions and prey availability. Upon observing frequent spider predation events in invaded habitats we used factorial manipulations of invasive plant and lycosid spider presence in field cages to measure the effect of plant invasion on spider persistence and resulting spider effects on toad survival. We hypothesized that toad growth and survival in field enclosures would be lower in invaded habitats and associated with reduced prey and increased predator abundance. We also hypothesized that the presence of the invasive grass in experimental cages would increase the persistence of spiders via structurally mediated decreases in intraguild predation rates, resulting in the lowest toad survival in cages with both stilt grass and spiders present.
Microstegium vimineum is an invasive annual grass that is widespread in moist, shaded, forest-floor habitats throughout the eastern United States. This plant can influence belowground carbon dynamics (Strickland et al. 2011) and alter arthropod diversity and abundance (Simao et al. 2010), giving it the potential to drive bottom-up effects. Herbivory on this plant is limited (but see Bradford et al. 2010) and unlikely to contribute to trophic resources utilized by the predators in this research (Flowers and Graves 1995, Lensing and Wise 2006); therefore, any bottom-up effects are likely to manifest through influences on detrital food webs. Invasion also increases forest floor plant cover and complexity. This effect is pronounced in the deciduous forests of the southeastern United States, where the plant communities of the habitats utilized by this invader are relatively depauperate from excessive deer browse and historic agriculture (see Appendix A).
Response species: Anaxyrus (Bufo) americanus (American toad)
The American toad, A. americanus, is a common forest floor predator throughout the eastern United States. Compared to other anurans, toads undergo metamorphosis at a small size and prioritize terrestrial growth (Werner 1986). Newly metamorphosed toads actively feed on small litter invertebrates such as mites and ants (Flowers and Graves 1995). This life history strategy renders them vulnerable to predation during this period of rapid growth.
Intermediate species: Toad prey and lycosid spiders
Metamorphic toads feed principally on litter invertebrates produced through detrital food webs (Flowers and Graves 1995). As these food webs are responsive to changes in nutrient inputs, cascading bottom-up effects on toad prey are possible, and, in fact, reduced arthropod production has been experimentally demonstrated following M. vimineum invasion (Simao et al. 2010).
As small, active forest floor residents, metamorphic toads are also susceptible to spider predation. Lycosid spiders are abundant predators that persist at higher densities in structurally complex habitats (Denno et al. 2002). Greater habitat complexity and resulting increases in spider densities can result in suppression of the shared prey of this group, with complex environments associated with increased prey susceptibility to predation (Finke and Denno 2002).
Examining invasion effects on the field survival and growth of toads
To determine the effects of M. vimineum on toad performance, abiotic soil properties, and prey availability we identified eight sites in the Georgia piedmont with independent, active, invasion fronts: four in Whitehall Forest, Athens, Georgia; two in Hard Labor Creek State Park, Rutledge, Georgia; and two in the Oconee National Forest, Eatonton, Georgia. We then constructed open-topped, 7.6 × 7.6 m pens on either side of each front using 1 m high silt fencing (see Appendix A for site photos). Fencing was buried to a depth of 15 cm, preventing toad escape, but crawling invertebrates could disperse over the woven material. From May to June, we released toads that we had raised to metamorphosis into these pens in two cohorts of 50 (2007) or four cohorts of 25 each (2008), for a total of 100 toads·pen−1·yr−1 (see Appendix B for detail). Wet masses of each cohort were recorded prior to release, and all toads were given a single toe clip to indicate cohort affiliation. At 45–55 days after the final release, we used three- to four-day closed capture periods to estimate survival and growth of toads, during which we searched the enclosures daily, measured each toad, identified them to cohort, and (at first capture) added a unique set of clips (≤2) to the original mark to allow for subsequent identification of individuals and construction of individualized capture strings. Processing was completed in the field and toads were immediately rereleased. Capture probabilities were high (0.89 d−1; 95% CI, 0.79–0.96) and consistent between treatments (see Appendix B), so the number of individuals known alive per cohort was taken as the number of survivors (uncaptured individuals were presumed dead). Survival was analyzed as a binomial response via logistic regression (1, alive; 0, dead; glmer) in R (R Development Core Team 2013) to test the fixed effects of M. vimineum, year, and a covariate of days lapsed between release and recapture (mean 56 days; range 45–69). Site was included as a random effect within which year, invasion status, and cohort were nested. One Oconee site was excluded due to complete mortality.
As the initial sizes of the randomly assigned cohorts did not differ by treatment (∼69.5 mg, F1,53 = 0.064, P = 0.94; see Appendix B) and mass gain greatly exceeded initial mass (increasing by 1420% ± 240% during the study period; mean ± 95% CI, n = 14 field enclosures) we used the average mass at recapture of the surviving individuals within each cohort as a proxy for growth. Analyses were conducted via a linear mixed model in R (lme) with year and M. vimineum presence as fixed effects within the random effect of site; covariates accounted for variation between the cohorts in growth period (days between release and recapture) and initial size (average mass at release).
Estimating impacts of invasion on toad prey availability and abiotic conditions
We measured litter invertebrate abundance to account for differences in prey availability between invaded and uninvaded habitats in May 2008, immediately prior to toad release, by extracting three 0.25-m2 quadrats of litter from randomly assigned positions in each pen on Tullgren funnels. Based on observations and a report of metamorphic toad diets (Flowers and Graves 1995) we confined our prey analysis to the following orders: Acari, Araneae, Coleoptera, Collembola, Diptera, Hemiptera, and Hymenoptera. Macro- and mesoarthropods (>2 mm length) were excluded, as we did not observe prey that large in metamorphic toads (J. L. DeVore, unpublished data). Due to sample corruption we were only able to utilize data from six of the eight sites. Prey counts were analyzed in Statistica 10 (StatSoft, Tulsa, Oklahoma, USA) via quasi-Poisson generalized linear model.
To account for abiotic changes with potential to affect toad performance, we measured soil moisture (volumetric water content), temperature, and pH monthly from May to August of 2008. Soil moisture and temperature were taken in triplicate from three haphazard positions within each pen, and pH was measured from 8 cm diameter, 0–10 cm depth soil cores taken from each of these positions. Replicate cores were sieved together (4 mm), transported on ice, and stored fresh at 5°C until analysis. The pH was measured in two analytical replicates using a 1:1 soil : H2O volumetric ratio. Data were analyzed via repeated measures linear models in Statistica.
Factorial spider × Microstegium vimineum field cage experiment
We used a 2 × 2 factorial field cage experiment to detect possible isolated and combined effects of M. vimineum and lycosid spider presence on toad survival. We installed five blocks of four cages each in invaded deciduous forest in Whitehall using 175 gallon reptariums (122 × 74 × 74 cm). Cages were buried to 10 cm in November 2008. We froze all soil and leaf litter in a freezer for >2 weeks to kill predatory invertebrates prior to placement in cages and added equal litter quantities to each cage. We assigned two cages in each block to M. vimineum present treatments and added 50 seed heads collected from adjacent plants. We then left the cages to mature until the following spring, and used weeding and supplemental plantings as appropriate to maintain treatments.
In mid May and early June of 2009, we stocked cages that we randomly assigned to spider treatments with wild caught cursorial spiders. We added spiders from several size classes to each cage: 10 at <0.01 g, 10 at 0.01–0.05 g, five at 0.1–0.2 g, and five at 0.3–0.8 g in May, followed by an additional two at 0.1–0.2 g and three at 0.3–0.8 g three weeks later. Spiders ≥0.10 g were also paired by size and species within a block (Hogna carolinensis and Hogna helluo were used). Spider abundance was then allowed to self-regulate. We considered spiders ≥0.10 g to be potential predators and subsequently only monitored their abundance. Two weeks after the second pulse of introductions, each of the 20 cages was stocked with 20 individually marked toads. Five days after release, toads were recaptured, identified, and rereleased daily over a three-day period to determine abundances while accounting for capture probabilities. Each cage was also searched for spiders after sunset for 10 nights, and spiders ≥0.10 g were individually marked using nontoxic neon paint writers (Elmer's Brand, Columbus, Ohio, USA). No new captures were made after the fourth night. Toad capture probabilities were also high (>95% per search) so all analyses were conducted on the proportion known alive of the initial 20. Spider counts were initially modeled using a Poisson distribution, but were underdispersed (Pearson χ2/df = 0.033), so data were square-root-transformed and the influences of treatment and block analyzed in Statistica via a linear mixed model. The effect of M. vimineum presence on toad survival was analyzed in R in both the presence and absence of spiders via logistic regression with a random blocking factor.
Pen survival and growth
The presence of M. vimineum significantly reduced the odds of survival, such that toads were 1.75 times as likely to survive where M. vimineum was absent (95% CI = 1.33–2.30; z = 3.978, P < 0.0001; see Fig. 1). There was also a significant effect of year (the odds of surviving were lower in 2008; odds ratio [OR] = 0.60, 95% CI = 0.36–0.98, z = −2.029, P = 0.0424) and a negative association between survival and the number of days elapsed between release and census (OR = 0.94, 95% CI = 0.93–0.96, z = −6.761, P < 0.0001; see Appendix C: Table C1).
Size data was collected for all surviving toads, which represented 75 of the 96 cohorts of 25 (2008) or 50 (2007) toads that were initially released into the 16 field enclosures. Analysis of mean cohort mass at recapture (g) demonstrated that mass did not vary as a function of invasion status (0.024 ± 0.172 [mean ± 95%CI] for invaded; t = 0.373, P = 0.788), year (t = −0.721, P = 0.498), or mean mass at release (t = 1.444, P = 0.156), although the number of days that passed between release and recapture had a significant, positive effect (0.025 ± 0.011; t = 0.445, P = 0.0001; see Table C2). The mean growth rates of all surviving individuals within the enclosures ([final mass − mean mass at metamorphosis]/time) averaged 16.53 ± 2.84 mg/d (n = 14 field enclosures).
Prey availability and abiotic conditions
Prey densities in invaded habitats were 30% lower than in uninvaded habitats (reduced from 1379 ± 670 to 964 ± 444 prey/m2; χ2(1) = 4.69, P = 0.030; n = 6 habitats). Site also influenced prey densities (χ2(5) = 20.10, P = 0.001). Soil moisture was significantly higher in invaded plots during the study period (F1,7 = 9.14, P = 0.019), but temperature and pH did not differ significantly between treatments (F1,7 = 5.49, P = 0.052 and F1,7 = 3.40, P = 0.107, respectively), though there was a tendency toward higher temperatures and decreased soil acidity in invaded plots. Effects of invasion on temperature interacted significantly with month (F3,21 = 3.89, P = 0.023) such that the magnitude of this warming tendency varied through time (see Appendix C: Table C3).
The presence of M. vimineum significantly increased lycosid spider persistence in experimental cages (Fig. 2; β = 1.2 ± 0.4 [mean ± 95% CI], F1,4 = 32.13, P = 0.0048). Toad survival in the absence of spiders was high (0.96 ± 0.04; n = 10 cages) and not affected by M. vimineum (z = 1.116, P = 0.264, n = 5 cages; Table C1). In the presence of spiders, however, there was a significant effect of M. vimineum, such that the odds of surviving in an uninvaded cage were 3.52 (95% CI = 1.90–6.51) times those in an invaded cage (see Fig. 2 for means and Appendix C: Fig. C1 for a scatterplot; z = 4.007, P < 0.0001, n = 5 cages).
Although plant invasion can alter habitat structure, basal resource availability, and abiotic conditions, few studies have simultaneously explored how these mechanisms will affect native consumers. While changes in basal resources or abiotic conditions have potential to influence consumers, here we found little evidence for these effects. Although, through prior research, we found that M. vimineum reduced belowground carbon pools at our sites (Strickland et al. 2011), and here we show that invasion reduced toad prey, these resource reductions did not lead to reductions in growth. Instead, we found that dominant impact of this plant invader on metamorphic toads occurs via associated changes in environmental context, which modify the strength of a preexisting interaction between this species and an abundant predator. This modification occurs not through refuge provisioning for the prey species, but more indirectly, through the provisioning of refuge from intraguild predation by the cannibalistic predator. This structurally mediated decrease in intraguild predation by lycosid spiders and resulting increases in predation pressure on young toads lowers toad survival in invaded habitats and exemplifies the complex mechanisms through which plant invasion can modify species interactions (Fig. 3).
The potential of spiders to suppress prey populations has been well documented (Finke and Denno 2002); spiders are a highly abundant predatory force on the forest floor (Moulder and Reichle 1972). Population dynamics in lycosid spiders are responsive to habitat structure; their densities have been found to increase following Vinca minor invasion (Bultman and DeWitt 2008), thatch addition (Schmidt and Rypstra 2010), and provisioning of faux leaves (Bultman and Uetz 1984) or plastic plants (Schmidt and Rypstra 2010). Variation in densities is thought to be driven principally by the availability of refuge from intraguild predation and cannibalism, which is a major mortality factor in this group (Janssen et al. 2007). Here we found through pulsed stocking of lycosid spiders that, in closed populations, the effects of cannibalism can create these patterns in habitats where structural complexity increases following plant invasion.
Although spiders benefit from refuge from intraguild predation within complex habitats, attack success on certain prey has been found to be independent of plant complexity in this group (Schmidt and Rypstra 2010), creating the potential for cascading effects, such as those seen here. Since these predators utilize a sit-and-pursue strategy (Schmitz and Suttle 2001) and prefer mobile prey (Moulder and Reichle 1972), species that utilize active foraging techniques (such as metamorphic toads) are especially vulnerable. Though we could not distinguish the direct, predatory influence of spiders from potential indirect effects, we did find that a 33% increase in their densities led to a 65% decrease in toad survival. While cages may have intensified predatory effects, the strength of this response could also be enhanced by increased structural hindrances to escape or stress-based responses to high predator densities (Preisser et al. 2007).
Increased top-down pressure can be ameliorated by bottom-up processes (Denno et al. 2002), but this did not occur in this system, as invasion also decreased prey availability. Despite prey reductions, toad growth rates were similar between treatments. The active foraging strategy utilized by juvenile toads is likely to minimize the potential for bottom-up effects by allowing toads to compensate for prey reductions through increased activity (Werner and Anholt 1993). However, this has the potential to further increase exposure to predation through trait-mediated indirect effects, in which augmented activity in these habitats increases susceptibility to sit-and-pursue predators; this has been demonstrated in hungry toads, leading to increased snake predation (Heinen 1994), and could have contributed to observed effects of invasion on survival.
In certain cases, structurally mediated modifications in trophic interactions can be secondary to effects driven by abiotic changes resulting from autotrophic invasion (e.g., Byers et al. 2010). As amphibians are ectotherms with permeable skin, and water uptake in this group can be further affected by pH, changes in soil parameters such as moisture, temperature, or soil acidity could drive a response in our model species (Feder and Burggren 1992). However, in the absence of spiders, we observed no effect of M. vimineum on toad survival. Additionally, observed abiotic influences of this grass included increased soil moisture and a tendency toward decreased soil acidity, both of which are expected to positively influence amphibian performance (Feder and Burggren 1992), making them unlikely to have contributed to survival reductions.
Although variation in survival during the metamorphic stage is thought to be a relatively important determinant of toad population growth (Biek et al. 2002), population level effects of M. vimineum invasion are yet unknown, as increasing body sizes and a transition to a less active foraging strategy reduce the risk of spider predation later in life. Any effects of invasion on later life stages are therefore likely to manifest through different mechanisms. However, increased spider densities are unlikely to only influence toads. Lycosid spiders prey heavily on detritivore communities and changes in their densities can affect ecosystem processes that may already be altered by invasion, such as decomposition (Lensing and Wise 2006). Therefore, although this study provides a unique perspective of the multiple mechanisms through which invasion affects a focal species, the cascading effects of this invader are likely to exceed those demonstrated here.
Invasive species are frequently implicated as a leading cause of species endangerment, but in many cases it is unlikely that they are directly driving these losses. Rather, indirect effects caused by their presence may render the habitat less suitable for natives. Accordingly, studies of the impacts of plant invasions have frequently noted differences in community composition across invasion boundaries, but the mechanisms driving these effects are rarely explored. This is especially true when indirect interactions are driving these changes (White et al. 2006). Although the impacts of plant invasion on trophic interactions have traditionally been considered to be relatively minor compared to those driven by animal invasion, plants can alter a number of habitat characteristics, and influences on native consumers should be expected. While a priori expectations may be that the influence of plant invasion will be manifested principally through bottom-up effects, in this case we found that a structurally mediated dampening of intraguild predation among wolf spiders made increased top-down pressure the dominant consequence of invasion for toad performance in this system, though whether these structural effects serve to offset (or overcompensate for) anthropogenic influences that previously depopulated this understory plant community remains unknown. Such engineering may frequently play a role in mediating trophic interactions between consumer species, but plant modifications of predator–prey dynamics are seldom considered in the context of plant community change (White et al. 2006). While direct impacts of invasion on prey, hosts, and competitors are demonstrably catastrophic in many systems, we argue that indirect influences such as those demonstrated here are likely to be widespread, with the potential to operate with similar strength as direct effects.
This study was conducted under animal use protocol A2007-10145. We thank the Oconee National Forest and Hard Labor Creek State Park for site access and the Maerz lab (especially V. Terrell) for their support. J. L. DeVore was funded by a Warnell assistantship during this research.