- Top of page
- Materials and methods
- Supporting Information
The ability of an introduced species to become established in a community depends on an array of direct and indirect interactions between species and their environment (see Chesson 2000; Mack et al. 2000; Shea & Chesson 2002; Mitchell et al. 2006 for reviews). The strength and direction of these species interactions are context dependent, and their outcome depends on such factors as resource supply, physical stress, disturbance and life-history stage (e.g. see Chesson 2000). Ultimately, once a species is introduced into a new location via natural or anthropogenic means, its persistence and impact will depend on how it interacts with resident species as well as local environmental conditions (Chesson 2000; Seabloom et al. 2003). Thus, understanding the interplay between the environment and interactions between novel and resident species is important for predicting the success and long-term impact of an introduced species.
Dispersal limitation, environmental conditions and biotic interactions can be thought of as a set of successive filters that determine whether a species from the regional pool is able to persist at a given location. Natural dispersal patterns (and anthropogenic introductions) act as the first filter by allowing only a subset of species to colonize a local site. Once present in a new site, local environmental conditions act as a secondary filter by preventing the establishment of ill-adapted species. Biotic interactions mediated by resident species can act as a final filter by (i) preventing the establishment of introduced species that are well suited to cope with local environmental conditions via competitive exclusion or (ii) allowing the establishment of introduced species that are poorly adapted to local environmental conditions via facilitation. In reality, these filters do not always act sequentially and independently, but simultaneously and interactively because the strength and direction of biotic interactions are often mediated by local environmental conditions (Bertness & Hacker 1994; Mack et al. 2000; Callaway et al. 2002; Tylianakis et al. 2008). One way to evaluate the support for the combined effects of environmental filtering and species interactions is to measure how species interact across a relevant environmental gradient. If neither local environmental conditions nor species interactions are obstacles for the successful invasion of a species, then lack of introduction by natural or human-assisted means (i.e. dispersal limitation) is a likely candidate explanation.
Research on the mechanisms leading to native–non-native species coexistence has focused on competition avoidance mechanisms (Shea & Chesson 2002), apparent competition (Borer et al. 2007) and the role of spatial heterogeneity (Melbourne et al. 2007). However, facilitation can be as important as competition in structuring communities by promoting coexistence among (i) native species (Bertness & Callaway 1994; Hacker & Gaines 1997; Bruno, Stachowicz & Bertness 2003; Gouhier, Menge & Hacker 2011) and (ii) native and non-native species (Palmer & Maurer 1997; Richardson et al. 2000; MacDougall & Turkington 2005; Hacker & Dethier 2006; Mitchell et al. 2006; Rodriguez 2006; Wolkovich, Bolger & Cottingham 2009; Altieri et al. 2010). Indirect effects are also important factors mediating invader success or native–non-native coexistence, but these are just beginning to be recognized. Examples include indirect effects from multiple interactions between enemies, mutualists and competitors (Mitchell et al. 2006), refuge-mediated apparent competition (Orrock, Baskett & Holt 2010) and indirect effects that arise from pathogen interactions (e.g. a disease that promotes invader dominance: Tompkins, White & Boots (2003); and a disease-mediated reduction in native–non-native competition that results in coexistence: Borer et al. (2007)). In reality, the mechanisms driving native–non-native species coexistence are likely to be complex – including a combination of positive and negative direct and indirect interactions that are context dependent.
Here we use field data, manipulative experiments and mathematical models to (i) explore the context dependency of invasions by uncovering the direct and indirect effects of environmental conditions and biotic interactions on the establishment of a non-native species into a resident community and (ii) understand the potential long-term outcomes for community composition. Our study system is composed of beach grass communities found on dunes along the Pacific Northwest coast of the United States. These dunes are dominated by three beach grass species: the native grass, Elymus mollis (Trin.), and two non-native grasses, Ammophila arenaria (L.) Link and A. breviligulata Fernald (Fig. 1). Both non-native species were introduced for sand stabilization to the US Pacific Northwest beginning over a century ago (A. arenaria in 1868, A. breviligulata 1935) and have subsequently invaded dune-backed beaches, which comprise nearly 50% of the coastline (Seabloom & Wiedemann 1994; Wiedemann & Pickart 2004). Each invader has different effects on the shape of foredunes, which are linear dune ridges aligned parallel and adjacent to the shoreline (Hacker et al. 2012; Zarnetske et al. 2012). Foredunes dominated by A. breviligulata are lower and wider than those dominated by A. arenaria (Seabloom & Wiedemann 1994; Hacker et al. 2012). These foredune shapes are influenced by species-specific dune-building ability (where A. arenaria is superior to A. breviligulata; Zarnetske et al. 2012) and also provide different degrees of protection from wave overtopping and inundation (Seabloom et al. 2013). Ocean-derived sand supply also contributes to foredune shape and is an important environmental driver in this system (Cooper 1958; Ruggiero et al. 2005, 2011; Zarnetske 2011; Hacker et al. 2012).
Figure 1. Distribution of (a) two non-native grasses, Ammophila arenaria and Ammophila breviligulata, and the native grass, Elymus mollis (as mean dry biomass kg m−2 from 84 transects along the front of the foredune in 2009) and (b) sand deposition (measured as dune vertical growth rate from 1997–2009 (m year−1)) along the Oregon and Washington coasts.
Download figure to PowerPoint
The three beach grass species co-occur in some regions of the coast, but not in others (Fig. 1a, Zarnetske 2011; Hacker et al. 2012). Prior to the introduction of A. breviligulata, A. arenaria was the dominant beach grass invader along the coast (Cooper 1958; Seabloom & Wiedemann 1994). However, the 1935 A. breviligulata introductions to coastline adjacent to the Columbia River led to subsequent invasions in the northern part of the study area that were previously dominated by A. arenaria (Seabloom & Wiedemann 1994; Hacker et al. 2012). Today, A. breviligulata occurs across a range of sand supply rates, but is restricted to northern foredunes with higher sand supply rates, A. arenaria occurs across a range of sand supply rates, but dominates more southern foredunes with lower sand supply rates, and E. mollis occurs across the range in low abundance (Fig. 1a). While it is believed that the geographical segregation of A. breviligulata is a consequence of dispersal limitation (see Hacker et al. 2012), it remains unclear whether A. breviligulata will continue to expand its range to the south and whether this invasion will result in the displacement of the resident beach grass species. These potential community outcomes are important given the species-specific influence on the structural and functional characteristics of foredunes and the associated implications for coastal protection ecosystem services (Zarnetske et al. 2012).
We parameterized a Lotka–Volterra model using experimental and observational data to determine to what extent the long-term distribution of beach grass communities and potential invasion of A. breviligulata is constrained by environmental filtering (difference in sand supply suitability), competitive exclusion (by the resident beach grass community, A. arenaria and E. mollis) and dispersal limitation. To answer this question, we used the model to determine whether native and non-native beach grass species can coexist, whether the potential for coexistence is mediated by environmental conditions and what types of species interactions are involved in driving these results.
- Top of page
- Materials and methods
- Supporting Information
Our models indicated that across all sand supply rates, coexistence among all three species was the most common outcome (26 of 30 cases, Fig. 2a–c). The extinction of the native, E. mollis, and coexistence of both Ammophila invaders was the only other outcome from the models (4 of 30 cases, Fig. 2). In the models, the time to reach long-term field abundances (position of t3) influenced whether a community was composed of 2 or 3 species, and altered the abundances of the three species (Fig. 2 and Appendix S3-E, F).
Figure 2. Species coexistence and relative abundance patterns (y-axis) by the fixed time to reach long-term abundances (x-axis), per sand supply rate. For each sand supply rate, the time to reach long-term abundance (t3) was allowed to vary over 10 linearly spaced time points (between t2 + 1 year and t4−1 year) to allow species to take different amounts of time to achieve long-term field abundances, before reaching the life span of a foredune (see Appendix S3). In this manner, we tracked the consistency of model outcomes, depending on how long it takes for species to achieve this long-term abundance. Panels (a–c) show species richness by sand supply rate. Panels (d–f) show the relative abundance of each species by sand supply rate. Two- and 3-species communities per sand supply rate were selected for further assessments – the 2-species communities are highlighted with light grey bars, and the 3-species communities are highlighted with dark grey bars.
Download figure to PowerPoint
The initial species interaction coefficients (αij) from the mesocosm experiment represent the transient dynamics in the system and showed that after a year of low sand deposition rate, A. breviligulata and A. arenaria competed strongly, but A. arenaria had a stronger effect on A. breviligulata (αAB = 0.6988, αBA = 0.8841, Appendix S1-F3). At mid- and high sand supply rates, A. breviligulata had slightly positive effects on A. arenaria (αAB mid = −0.2190, αAB high = −0.1446), while A. arenaria had relatively weak negative effects on A. breviligulata (αBA mid = 0.3500, αBA high = 0.0034) (Appendix S1-F3). Effects on E. mollis by Ammophila species were mostly negative across sand supply rates, and strongest from A. arenaria, though effects were positive at high sand supply from A. arenaria and at mid-sand supply from A. breviligulata (Appendix S1-F3). Elymus mollis had slightly negative effects on both Ammophila species at low sand supply, but positive effects on them at mid- and high sand supply rates (Appendix S1-F3). In general, the mesocosm experiment showed that there was a tendency towards weaker competitive and stronger facilitative interactions with increasing sand supply (Appendix S1-F3).
In the model, the long-term (asymptotic) outcomes in the 2- or 3-species communities were maintained by a mixture of competitive and facilitative interactions (Fig. 3). Although both Ammophila species coexisted in the 2- and 3-species communities, they differed slightly in their interactions across community type. In both communities, A. breviligulata directly facilitated A. arenaria in low sand supply, but competed against A. arenaria at high sand supply, while A. arenaria increasingly facilitated A. breviligulata with increasing sand supply in the 3-species community, but only strongly facilitated A. breviligulata at the mid-sand supply in the 2-species community (Fig. 3). Although the sand supply gradient did not affect the final outcome of coexistence, it did mediate the strength of the underlying α species interactions (Fig. 3) and magnitudes of parameters r and K (Appendix S3-F). Specifically, we found that greater sand supply rate reduced the intrinsic rate of growth (r) for all species, reduced carrying capacity (K) for both Ammophila invaders in the 3-species community and increased the strength of species interactions across both communities (Fig. 3, Appendix S3-F). The stable 2- and 3-species communities at equilibrium show consistent dominance by A. breviligulata across sand supply rates (Fig. 2). E. mollis was consistently excluded in the 2-species community, probably because at low and mid-sand supplies it received stronger competition from A. arenaria (which was promoted through facilitation by A. breviligulata) than direct facilitation from A. breviligulata, and at high sand supply, experienced direct and indirect competition from A. breviligulata (Figs 2, 3 and Appendix S3). E. mollis was the lowest abundance species in the 3-species community, and there were no stable solutions that led to E. mollis dominance (Figs 2, 3 and Appendix S3).
Figure 3. Path diagrams for 2- and 3-species communities per sand supply rate, showing the strength (arrow thickness) and direction (facilitation, competition) of all α values (both inter- and intraspecific interactions where αAA, αBB, αMM all equal 1).
Download figure to PowerPoint
Sensitivity analyses showed that sand supply mediated the relative influence of inter- and intraspecific interactions on community composition. The abundance of A. arenaria in the 2-species community was most sensitive to interspecific interactions at low sand supply (facilitation from A. breviligulata), intraspecific interactions at mid-sand supply (i.e. its own carrying capacity) and both inter- and intraspecific interactions at high sand supply (i.e. competition from A. breviligulata, and its own carrying capacity) (Fig. 3, Appendix S4). Conversely, in the same 2-species community, the abundance of A. breviligulata was more sensitive to its own carrying capacity at low and high sand supply and interspecific facilitation from A. arenaria at mid-sand supply (Fig. 3 and Appendix S4). The abundance of each species in the 3-species community was most sensitive to its own carrying capacity at low sand supply, but became increasingly sensitive to interspecific interactions in mid-sand supply (facilitative) and high sand supply (both direct and indirect, positive and negative) (Fig. 3 and Appendix S4).
The relative abundance of each species varied by sand supply rate (Fig. 2), owing to the sand supply–mediated interactions (Fig. 3). In both communities, A. breviligulata achieved its highest long-term abundance at the mid-sand supply rate, while A. arenaria achieved its highest long-term abundance at low and mid-sand supply in the 2-species community, and at mid-sand supply in the 3-species community (Fig. 2). The direct and indirect interactions enabled species to exceed their carrying capacities at long-term equilibrium (Figs 2, 3 and Table 1). In the 2-species community at low and mid-sand supply rates, direct and indirect positive interactions enabled A. breviligulata and A. arenaria to exceed their carrying capacities, while direct and indirect competition between Ammophila species at the high sand supply rate reduced the abundance of A. arenaria to levels far below its carrying capacity and maintained A. breviligulata near its carrying capacity (Figs 2, 3 and Table 1). In the 3-species community, direct and indirect facilitation enabled A. breviligulata and E. mollis to exceed their carrying capacities across sand supply rates (Figs 2, 3 and Table 1). In contrast, direct and indirect competition restricted the long-term abundance of A. arenaria to near its carrying capacity in low and mid-sand supply rates, while direct and indirect facilitation enabled it to exceed its carrying capacity at high sand supply over the long term (Fig. 3 and Table 1).
Table 1. Model equilibrium abundance per species and community type (2- or 3-species), as it relates to carrying capacity (K) of each species. The equilibrium abundances and carrying capacities are from the 2- and 3-species communities that were selected for analysis (see Fig. 2: 2-species communities in light grey bars; 3-species communities in dark grey bars)
|Species||Low sand supply||Mid-sand supply||High sand supply|
|2 species||3 species||2 species||3 species||2 species||3 species|
| A. arenaria ||>>KA||≈ KA||>KA||≈ KA||<<KA|| >K A |
| A. breviligulata ||>KB||>KB||>>KB||>>KB||≈ KB||>>KB|
| E. mollis ||–||>KM||–||>KM||–|| >K M |
- Top of page
- Materials and methods
- Supporting Information
Strong correlations between the geographical distribution of multiple species and local environmental conditions in the short term often suggest direct relationships. However, while such short-term correlations provide important insights about community composition, they may not hold over longer periods of time because they may reflect transitionally co-occurring species. Understanding the long-term distribution of species and community structure thus requires a holistic approach that integrates the effects of both local environmental conditions and biotic interactions over the long term (Seabloom et al. 2003; Siepielski & McPeek 2010; Gravel, Guichard & Hochberg 2011). Here, we used empirical and theoretical approaches to assess whether long-term community composition and associated spatial distributions were especially sensitive to a combination of environmental filtering and biotic interactions. Our results showed that environmental filtering does not play a direct role in determining either the long-term community composition or the spatial distribution of species because of overriding effects of positive and indirect biotic interactions. However, environmental conditions mediated the strength and sign of these biotic interactions and thus had an indirect influence on the community over the long term.
The strong patterns of covariation between species abundance and coastal environmental conditions in the US Pacific Northwest suggest that dispersal limitation and the environmental context of sand supply limits the distribution and spread of the invasive beach grass species, A. breviligulata (Fig. 1, Hacker et al. 2012). However, these patterns provide little insight into the role of species interactions and their potential effects on (i) the long-term distribution of species and (ii) the potential spread and impact of A. breviligulata. To address this unknown, we parameterized a Lotka–Volterra model with experimental and field data and evaluated the model at equilibrium to generate predictions about the effects of environmental conditions and species interactions on the long-term community structure. Given the parameterization and model assumptions, we predict that co-occurring beach grasses are more likely to result in native–non-native coexistence than exclusion. Although we can only speculate about the nature of the processes underlying these outcomes, we can use the model to (i) understand how these processes manifest themselves through changes in species interactions in response to different levels of sand supply, (ii) identify which community outcomes are more likely and (iii) predict whether A. breviligulata, once introduced, may be able to invade a resident beach grass community in new areas along the coast.
The model results suggest that once A. breviligulata is introduced to new coastal areas where the other two grasses currently exist, it could persist and dominate the community across the range of sand supply rates investigated here. However, the potential for coexistence and the long-term community composition was not directly mediated by sand supply. Instead, we found that sand supply can play an important but subtle role through its influence on (i) intra- vs. interspecific community regulating processes, and (ii) the strength of positive and indirect effects among beach grasses (Fig. 3), and thus ultimately shape the long-term abundances of the beach grass species (Fig. 2 and Table 1).
Although our model suggests that sand supply does not have a strong direct effect on long-term beach grass community composition or the success of a non-native invasion (i.e. both non-natives or all three grass species persist across sand supply regimes; Fig. 2), sand supply appears to have a strong impact on the types of interactions that drive community structure (Fig. 3). The intraspecific parameters (r, K) and interspecific parameters (α) covaried with sand supply (Appendix S3-D, F). Thus, sand supply plays an important intermediary role in community composition for both 2- and 3-species communities because at lower sand supply, species abundance is most sensitive to intraspecific processes such as self-regulating carrying capacities, whereas interspecific community-level processes (especially direct and indirect facilitation) have a stronger influence on species abundance at higher sand supply conditions (Appendix S4). If our model assumptions are correct, these findings, together with the decrease in the intrinsic rate of growth r with increasing sand supply, indicate that intraspecific competition regulates these populations in benign environments characterized by low sand deposition, while interspecific interactions are more influential in adverse environments characterized by high sand deposition.
The relative influence of interspecific (i.e. species interactions) and intraspecific (i.e. growth rate and carrying capacity) processes on patterns of abundance across different levels of sand supply predicted by the model are consistent with those observed in the field. In beach grass communities found along the Pacific Northwest, each species' abundance tends to be higher on foredunes with low and mid-sand supply rates than on foredunes with higher sand supply rates (Appendix S1-F). This reflects the higher tiller density growth at lower sand supply rates and lower tiller density growth at higher sand supply rates (Zarnetske et al. 2012). Lower density growth under high sand supply likely reflects horizontal spreading (Zarnetske et al. 2012). In turn, this spreading could lead to more encounters with other species. If this is the case, interspecific effects could increase with sand supply rate. Our model results show that environmental conditions can have subtle yet critical impacts on communities by modulating the relative influence of intra- and interspecific processes that govern patterns of species abundance and richness without altering the patterns themselves.
Here we investigated the spatial variation in sand supply rate rather than its temporal variation. Recent studies from other systems show that temporal variation in disturbance can have strong effects on community composition (García Molinos & Donohue 2011; Miller, Roxburgh & Shea 2011), and this could be the case in this system as well. Along the US Pacific Northwest coast, sand supply rates vary through space and time, but the variation in space has been relatively consistent since the 1950s (Ruggiero et al. 2005, 2011), leading to a consistent spatial environmental stress gradient for the plant community. Although this spatial variation in sand supply rate has persisted for over 50 years, temporal variability could affect species interactions in complex ways and potentially result in different outcomes for beach grass communities. Our study thus represents an important first step that should be extended by future field or mesocosm experiments that investigate both spatial and temporal variations in sand supply rate.
Facilitation and indirect interspecific effects increased in importance in the more stressful high sand deposition environments (Fig. 3) as has been shown in coastal dunes in other regions of the world (Franks & Peterson 2003; Martinez & Garcia-Franco 2008; Forey, Lortie & Michalet 2009). This increased facilitation could be a reflection of several mechanisms including increased structural above- and below-ground support, shelter from sand scour, reduced sand compaction and increased aeration (Franks & Peterson 2003), shared benefits from mycorrhizal fungi (Kowalchuk, De Souza & Van Veen 2002) and escape from harmful nematodes and fungi (de Rooij-Van der Goes et al. 1997). For example, given that each beach grass species has different modes of above- and below-ground growth in response to sand supply rate (Hacker et al. 2012; Zarnetske et al. 2012), it is possible that they could assist one another in structural support and shelter from sand scour, both of which would increase with increasing sand supply rates.
Facilitation, whether direct, or indirect, is predicted to be the predominant factor enabling species to coexist in this system; however, its importance to each species depends not only on sand deposition, but also on the resident-community context. For example, the model shows that facilitation allowed A. arenaria to far exceed its carrying capacity at low sand supply in a 2-species community (Fig. 3a and Appendix S3) and to coexist at high sand deposition in a 3-species community despite competition from A. breviligulata (Figs 2d and 3f). In contrast, model results suggest that facilitation allowed A. breviligulata to far exceed its carrying capacity at high sand deposition in a 3-species community (Fig. 3f and Table 1). We found that the native grass, E. mollis, is potentially a key player in facilitating the non-native invaders and may have farther-reaching effects on the community. The model shows that the native species, despite its low abundance, can play an important role in maintaining coexistence and, together with A. arenaria, could potentially facilitate further invasion of A. breviligulata.
Our study and other recent research show instances of native and invasive species facilitating one another – these findings could help explain why we rarely see invaders completely excluding native species (Palmer & Maurer 1997; MacDougall & Turkington 2005; Hacker & Dethier 2006; Rodriguez 2006; Wolkovich, Bolger & Cottingham 2009; Altieri et al. 2010). For example, Davis et al. (2011) have emphasized that native–non-native interactions rarely end with the extinction of native species. There is also evidence that species diversity may increase after an invasion (Sax, Gaines & Brown 2002). We have anecdotal evidence that Ammophila invasions have increased species richness in the back dune, where sand stabilization has promoted the creation of wetlands and forests (Seabloom & Wiedemann 1994; Wiedemann & Pickart 2004; Hacker et al. 2012), although on the foredune proper, richness and cover of native endemic dune species have declined (Zarnetske, Seabloom & Hacker 2010; Hacker et al. 2012).
The species interactions within the model provide possible explanations for historical trends in beach grass distributions. For example, the small increase in E. mollis abundance within the last two decades is concurrent with the A. breviligulata invasion (Hacker et al. 2012) and may be explained by the model-predicted direct facilitation from A. breviligulata (Fig. 3d). In addition, A. arenaria once dominated throughout much of the US Pacific Northwest, but is now only found sparsely in the northern regions of its distribution (Fig. 1). Our model results suggest that competitive displacement by A. breviligulata may be responsible for the observed contraction in the northern range of A. arenaria.
We can also use the model to understand potential future changes in invaded foredune grass communities, given current sand supply conditions. At this time, A. breviligulata is restricted to the northern coast (Fig. 1), likely due to dispersal limitation (Hacker et al. 2012). However, if the dispersal barrier is lifted via planting or successful propagule settlement from wind or ocean currents (Maun 1984; Baye 1990; Konlechner & Hilton 2009), our model suggests that A. breviligulata could spread further south into regions where it is presently absent. Certainly, other factors not investigated here would affect a potential further invasion (e.g. more or less favourable conditions at new sites owing to improved nutrients or novel enemies). However, both our parameterized model and our mesocosm experiments suggest that the low sand supply rates commonly observed in southern regions of the Pacific Northwest are unlikely to prevent the establishment of A. breviligulata.
If A. breviligulata continued to spread, our model suggests that over the long term, A. arenaria and A. breviligulata would co-dominate under the lower sand deposition conditions of the southern coast and may occasionally lead to the exclusion of E. mollis. The potential invasion of A. breviligulata into new regions will undoubtedly have strong ecological implications for biodiversity by altering both species richness and abundance in beach grass (A. arenaria and E. mollis) and native plant species (Hacker et al. 2012). For instance, a recent study in coastal dunes of the Great Lakes suggests that species and genetic diversity of the dominant A. breviligulata have strong, community-wide effects on abundance and ecosystem functioning (Crawford & Rudgers 2012). In the Pacific Northwest, foredunes dominated by A. breviligulata have lower species richness than those dominated by A. arenaria (Seabloom & Wiedemann 1994; Hacker et al. 2012), and thus, there is anecdotal evidence that further invasions may negatively affect dune biodiversity. Future work is needed to investigate the combined effects of species diversity and Ammophila genetic diversity on the composition of coastal dune communities in the Pacific Northwest.
Potential invasions of A. breviligulata into southern regions of the coast could affect dune geomorphology, particularly dune height. In companion studies, we have shown that through its lateral spreading growth habit, A. breviligulata generates lower, wider dunes (Seabloom and Wiedemann 1994; Hacker et al. 2012; Zarnetske et al. 2012). Therefore, if A. breviligulata invades taller foredunes in low sand supply regions of the coast where A. arenaria dominates, dune height may decline over time, compromising coastal protection from the overtopping of large waves generated by storms and tsunamis (Sallenger 2000; Liu et al. 2005; Mascarenhas & Jayakumar 2008; Zarnetske et al. 2012; Seabloom et al. 2013). However, our model predicts that A. arenaria will co-dominate with A. breviligulata under lower sand supply conditions and thereby potentially mitigate large changes in dune shape.
This field- and experiment-parameterized model framework provided a means to assess the spread and impact of an introduced species in resident beach grass communities along an environmental gradient of sand supply. In this case, beach grass interactions most often resulted in long-term coexistence across the sand supply gradient. This outcome would have been difficult to predict without this approach. Indeed, it is impractical to introduce an invasive species within a field-based manipulation and to wait years for coexistence or exclusion outcomes, especially because rapid management recommendations are typically needed in order to reduce the costs of controlling the spread of invasive species. In this case, a strictly field-based approach would have resulted in even longer delays before management recommendations could be made because of the prevalence of indirect effects, which typically play out over longer periods of time than direct effects (Menge 1995). Additionally, the model framework demonstrated that environmental conditions and species interactions jointly influence community composition. This interactive effect of abiotic and biotic factors on community composition could not have been gleaned from patterns of covariation between beach grass species abundances and sand supply rates along the coast.
Overall, we found that environmentally mediated facilitation and indirect effects can be important factors that promote invasion success as well as long-term native–non-native coexistence. Facilitation and indirect effects are potentially more widespread than currently documented within invaded systems and could be key factors contributing to the observed lack of invader-induced extinction of native species. In some cases, they may promote coexistence by alleviating stressful environmental conditions that would otherwise lead to extinction. In turn, these interactions can have far-reaching consequences on ecosystem function and services through their influence on community composition. Ultimately, by uncovering these important factors that regulate the beach grass communities, we can generate more robust predictions about the biological and physical consequences of current community composition patterns as well as potential future invasions across valuable but fragile coastal ecosystems.