Biol Invasions https://doi.org/10.1007/s10530-019-02072-z (0123456789().,-volV) (0123456789().,-volV) ORIGINAL PAPER Habitat alteration by invasive European green crab (Carcinus maenas) causes eelgrass loss in British Columbia, Canada Brett R. Howard . Fiona T. Francis . Isabelle M. Côté . Thomas W. Therriault Received: 31 August 2018 / Accepted: 5 August 2019  Springer Nature Switzerland AG 2019 Abstract Dominant, habitat-forming plant species, such as seagrasses, are key components of coastal ecosystems worldwide. Multiple stressors, including invasive species that directly alter, remove, or replace the foundation plant species, threaten these ecosystems. On the Atlantic coast of North America, ecosystem engineering by invasive European green crab (Carcinus maenas) has been linked to the loss of some eelgrass (Zostera marina) beds. However, the interaction of the same co-occurring species on the Pacific coast has not been investigated. We conducted an enclosure experiment in Barkley Sound, British Columbia, to determine if the engineering impacts of green crabs on Pacific eelgrass ecosystems mirror those previously identified on the Atlantic coast. Eelgrass shoot density declined rapidly over 4 weeks, with a 73–81% greater loss in enclosures with high crab density compared to the low-density and control Electronic supplementary material The online version of this article (https://doi.org/10.1007/s10530-019-02072-z) contains supplementary material, which is available to authorized users. B. R. Howard (&)
F. T. Francis
I. M. Côté Earth to Ocean Research Group, Department of Biological Sciences, Simon Fraser University, Burnaby, BC V5A 1S6, Canada e-mail: brett.howard@sfu.ca T. W. Therriault Fisheries and Oceans Canada, Pacific Biological Station, Nanaimo, BC V9T 6N7, Canada treatments. The low ratio of eelgrass blades to rhizomes in the high-density treatment suggests that blade shredding, rather than bioturbation of whole plants, was the main mechanism of eelgrass loss. Eelgrass was detected in green crab stomach contents, consistent with observations from the Atlantic coast. Crab density did not have a detectable effect on the biomass or community composition of benthic fauna associated with eelgrass over the duration of the experiment. The eelgrass loss we observed was consistent with losses observed on the Atlantic coast, which raises management concerns on the Pacific coast, particularly in areas where green crabs co-occur with other coastal stressors and with ecologically and economically important species such as salmon. Keywords Habitat alteration
Aquatic conservation
Disturbance
Ecosystem engineering
Enclosure
Zostera marina Introduction Invasive species can negatively affect ecosystems by physically altering the habitats they invade. When these alterations impact the composition or availability of resources required by other species, the invasive species is described as an ecosystem engineer (Jones et al. 1994). Ecosystem engineering may be autogenic, 123 B. R. Howard et al. where the invasive species itself creates habitat, or allogenic, where the habitat is transformed by the invasive organism through its activities (Fei et al. 2014). For example, the invasive Pacific oyster (Crassostrea gigas), a reef-forming sessile invertebrate, is an autogenic ecosystem engineer of marine ecosystems throughout parts of its invasive range, including British Columbia (Canada), Washington (USA), and France (Cognie et al. 2006; Kelly et al. 2008; Padilla 2010; Wagner et al. 2012). In contrast, allogenic ecosystem engineers are often mobile invertebrates (Fei et al. 2014; Romero et al. 2015). Along the east coast of North America, grazing by the invasive herbivorous European periwinkle (Littorina littorea) can sufficiently disrupt sedimentation in saltmarshes to convert them into rocky intertidal habitat (Bertness 1984). Similarly, in California, bioturbation by the invasive isopod Sphaeroma quoianum causes bank collapse and erosion, leading to the loss of marshland (Talley et al. 2001). The impacts of an invasive ecosystem engineer depend on context-specific factors, including the temporal and spatial scales of the introduction, per capita impact, and population density (Molnar et al. 2008; de Moura Queirós et al. 2011; Guy-Haim et al. 2018). An invasive species considered to be a major ecosystem engineer in one location may have much less consequential impacts in a different setting. For example, although the Pacific oyster is an autogenic ecosystem engineer in some regions, in other areas they simply replace functionally similar native species or remain in low densities after establishment, resulting in minimal ecosystem impacts (Padilla 2010; Green and Crowe 2014; Zwerschke et al. 2018). Therefore, we cannot assume that the observed engineering impacts of an invasive species will be consistent from one region to another or over time. The European green crab (Carcinus maenas) has been introduced to coastal ecosystems around the globe, including on both the Atlantic and Pacific coasts of North America. To date, research on invasive green crabs in eelgrass habitats has been focused on the Atlantic coast, where green crabs are considered allogenic engineers (Klassen and Locke 2007; Matheson et al. 2016). Davis et al. (1998) first showed that green crabs in eelgrass mesocosms both shredded blades and dislodged whole plants through bioturbation while foraging for prey. The density of green crabs predicted eelgrass damage, with densities of 4 123 crabs per m2 or more having a significant impact after just 1 week (Davis et al. 1998). Shredding damage has also been attributed to direct consumption of eelgrass blades by juvenile green crabs in mesocosms (Malyshev and Quijón 2011). Regardless of the exact mechanism (i.e., shredding or bioturbation), subsequent experimental work supported a causal link between widespread eelgrass declines along the east coast of North America and high densities of green crabs (Malyshev and Quijón 2011; Garbary et al. 2014; Neckles 2015; Matheson et al. 2016). The consequence of ecosystem engineering on eelgrass communities may be extensive along the Atlantic coast of North America (Thompson 2007; Matheson et al. 2016). However, whether green crabs also act as allogenic ecosystem engineers in eelgrass beds on the Pacific coast has not been assessed previously. There are important differences between the two coasts that could make the impact of green crabs on eelgrass different. Although eelgrass beds on both coasts are dominated by Zostera marina (Short et al. 2007), this marine angiosperm occurs at shallower depths, experiences less frequent annual disturbance, and flowers earlier in the summer on the Pacific coast than it does at the same latitude on the Atlantic coast of North America (Phillips et al. 1983a, b; Robertson and Mann 1984; Moore and Short 2006). Zostera marina in Pacific Canada also frequently co-occurs with the invasive Z. japonica (dwarf eelgrass), which is absent on the Atlantic coast of North America (Shafer et al. 2014). While eelgrass beds on both coasts are important habitats for many species, including economically important and threatened fish and bird species (Gotceitas et al. 1997; Heck et al. 2003; Kennedy et al. 2018), the species assemblages and interactions are different. On the Pacific coast, eelgrass beds function as both spawning habitat and shelter for Pacific herring (Clupea pallasii) (Hosack et al. 2006; Pikitch et al. 2014; Shelton et al. 2014), and as foraging grounds for out-migrating juveniles of several species of Pacific salmon, including the critically important Chinook salmon (Oncorhynchus tshawytscha) (Moore et al. 2016; Kennedy et al. 2018). Finally, while the Pacific coast of North America is typically more species rich (Archambault et al. 2010; Costello et al. 2010), it is also more heavily invaded than the Atlantic coast (Choi et al. 2016). These differences could result in different impacts of green Habitat alteration by invasive European green crab (Carcinus maenas) causes eelgrass… crabs in eelgrass ecosystems on the Atlantic and Pacific coasts. To determine the potential impacts of allogenic ecosystem engineering by green crabs on Pacific coast eelgrass ecosystems, we conducted an enclosure experiment in an eelgrass bed in Barkley Sound, British Columbia, Canada. We expected eelgrass loss to increase over time and with increasing densities of green crabs (e.g., Davis et al. 1998; Garbary et al. 2014; Neckles 2015). We also examined the mechanism of habitat alteration by green crabs. If eelgrass is lost primarily due to bioturbation (Fei et al. 2014), we hypothesized that the ratio of eelgrass blades to rhizome biomass would not differ significantly across green crab density treatments, as both shoots and rhizomes would be lost at a similar rate. In contrast, if eelgrass is lost primarily due to above-ground shredding (Davis et al. 1998; Malyshev and Quijón 2011; Garbary et al. 2014), then the ratio of eelgrass blades to rhizomes should be smaller in treatments with green crabs present, as blades would be lost at a faster rate than rhizomes. Finally, any habitat modification by green crabs might impact native species diversity (e.g., Thompson 2007; Matheson et al. 2016). Thus, we examined the abundance and size of benthic fauna in the enclosures, to test if higher densities of green crabs resulted in decreasing biomass of eelgrass-associated benthic fauna and shifts in community composition. Fig. 1 a Map of Barkley Sound, British Columbia, Canada. The arrow indicates the southeast corner of Mayne Bay where this study was conducted. The white diamond indicates the town of Bamfield. White denotes sea, grey indicates land; b embayment where green crab enclosure experiment was established; c location of experimental blocks. Each block contained one plot of each of four treatments: no-crab enclosure (0 crabs m-2), low density (1.4 crabs m-2), high density (5.6 crabs m-2), and an enclosure-control plot. Continuous dotted line indicates mean low low water (MLW) Methods Field site Our field site was located at the head of an unnamed embayment in the southeast corner of Mayne Bay, Barkley Sound, British Columbia, Canada (latitude 48.974, longitude - 125.288) (Fig. 1a). The site was sheltered and there were no obvious anthropogenic or natural stressors (i.e., sedimentation, disease, aquaculture, etc.) in the immediate area. The embayment was approximately 1 km long with a continuous intertidal area of approximately 0.03 km2, one-third of which consisted of eelgrass (British Columbia Marine Conservation Analysis 2011). The eelgrass band extended from the high intertidal zone to 1.1 m below mean low water (MLW) at its deepest point. The slope of the eelgrass bioband ranged from 3.1 to 123 B. R. Howard et al. 8.0% along the steeper southern shore and flattened out at the head of the embayment to 0.9%. While the highest part of the bioband consisted of invasive Zostera japonica (see also Shafer et al. 2016), most of the bed, including the area used for this experiment, was predominantly or exclusively Z. marina. Green crabs were first reported in Barkley Sound in 1999 and specifically in Mayne Bay in 2006 (Gillespie et al. 2007). The catch-per-unit-effort over three consecutive summers (2013–2015) ranged from 0.18 to 0.56 crab/trap-day, i.e. 100 to 339 crabs caught per day. This density is moderate compared to catch rates at the most heavily invaded sites in Barkley Sound (i.e., as high as 2.5 crab/trap-day, * 1400 crabs per day; BRH, unpublished data). Enclosure design and sampling We installed experimental enclosures along the low intertidal zone of the eelgrass bed (* 1.0 m above MLW) (Fig. 1b, c). Enclosures measured 1.2 m 9 1.2 m 9 1 m (i.e., benthic area enclosed was 1.44 m2) and were constructed of rebar and plastic deer fencing with a mesh size of 2.54 cm, which prevented the movement of large, mobile epifauna including crabs [ 50 mm carapace width (CW). The tops of the enclosures were also covered with fencing material to prevent crabs from climbing out or being predated. The walls of the enclosures were buried approximately 15 cm into the sediment to prevent crabs from escaping by burrowing. There was sufficient vertical height in the enclosures that the eelgrass was not impeded at high tide when the enclosures were fully submerged. The top of each enclosure had a small mesh (2.0 cm) collar that functioned as an access point for data collection. The collar was held closed with cable ties between sampling events. We arranged enclosures in a randomized block design, with a total of six blocks along the south- and eastfacing shoreline. Each block included three enclosure plots, one per treatment (described below), and one enclosure-control plot marked only by cedar stakes (to estimate the enclosure effect), for a total of 24 plots. Plots in a block were spaced 1 m apart and blocks were spaced[ 5 m apart (Fig. 1c). No enclosures were lost during the experiment, which ran for 4 weeks (1–29 August 2015). 123 Our experimental treatments included high-density (5.6 green crabs per m2), low-density (1.4 green crabs per m2), and no-crab enclosures (0 green crabs per m2). The first two densities approximated the high and average densities, respectively, of green crabs at sites in Barkley Sound (personal observation, BRH, TWT, and IMC). The green crabs used were caught on site using baited Fukui fish traps and were all males with intact claws, between 50 and 72 mm CW (notch-tonotch). We chose to use males to control for differences in behaviour and activity levels between sexes during the breeding season (Behrens Yamada 2001). To evenly distribute the range of crab sizes caught, the low-density treatment consisted of two average-sized crabs (approximately 50–65 mm CW), and the highdensity treatment contained eight crabs, two large ([ 65 mm CW) and six average-sized crabs. We surveyed eelgrass shoot density in all plots by snorkeling at high tide every 4–6 days over the course of the experiment, for a total of six surveys. We haphazardly threw a circular frame (201 cm2) into each plot and counted the shoots inside the frame. This subsampling was repeated three times per plot on our first site visit, which took place over 2 days (August 4 and 5), after which we increased sampling to five subsamples per plot for all following visits. After counting shoots, we counted the number of crabs in each enclosure. At the end of the experiment, we took five replicate sediment cores, 5.5 cm in diameter and 10 cm in depth (238 cm3), including the associated above-ground eelgrass blades, in each plot. The locations of the cores were predetermined using randomized x–y coordinates and were at least 10 cm from the perimeter of the plots to avoid edge effects. We also collected as many of the enclosed green crabs as we could find. We placed the green crabs and cores in seperate, sealed plastic bags, which were kept frozen until processing. Sample processing We defrosted and sieved the cores over a fine mesh screen and all eelgrass and benthic fauna visible to the naked eye were retained for further analysis. Fauna were stored in 95% ethanol until further processing. We also removed the remaining organic detritus in Habitat alteration by invasive European green crab (Carcinus maenas) causes eelgrass… each core (e.g., shells, terrestrial plant material, etc.). After cleaning the individual eelgrass plants of gravel and sand by rinsing them with fresh water we then cut the sheath to separate the blades from rhizomes. Both blades and rhizomes were dried to constant weight for a minimum of 48 h at 60 C and we took the total dry weight of blades and, separately, of rhizomes (to the nearest 0.0001 g) for each core. We weighed the benthic fauna individually (wet weight to the nearest 0.001 g) and identified each to the lowest taxonomic level possible based on macroscopic characteristics. Analysis All analyses were done using R (version 3.3.2) (R Development Core Team 2008). We generated generalized linear mixed-effects models (GLMMs) using the lme4 package for three response variables: average eelgrass shoot density, average blade to rhizome dry weight ratio, and average benthic fauna wet weight. In all cases, the within-plot variance of the samples was first analysed using Bartlett’s or Levene’s test, as appropriate, and determined to be non-significant. This allowed us to average samples within each plot without losing statistically important variance (Murtaugh 2007). We opted to analyse averages rather than raw data as a means to deal with the true-zero-inflated fauna and eelgrass weight data, as recommended by Bolker (2008). To determine if average eelgrass shoot density differed among treatments over time, we included treatment (categorical), time (days since onset of experiment; continuous), and the interaction between the two as fixed effects. We included a variance structure allowing for different variances by treatment to account for heteroscedasticity in the residuals (Zuur et al. 2009). Because we resampled shoot density in the enclosures over time, we tested for auto-correlation between sampling dates but found no detectable effect (Zuur et al. 2009). The average ratio of blades to rhizomes was calculated for each plot based on dry weights. We modelled this response using treatment as the only fixed effect. The model was fitted using logged ratios to account for heteroscedastic residuals, and differences among pairs of treatments were examined using HSD post-hoc tests. To determine if there was a difference in benthic fauna biomass among treatments, we calculated the average wet weight per plot by summing weights of individual taxa per core and then averaging across cores per plot. As above, we included treatment as the only fixed effect and logged the response variable to correct the heteroscedasticity. All models included block as a random effect. The final model for each response variable was tested against a null model using a likelihood ratio test to determine the significance of the fixed effects. We used non-metric multi-dimensional scaling (nMDS) to determine if community composition of benthic fauna varied with treatment, using the package vegan (Oksanen et al. 2012). Due to the low taxonomic resolution and low numbers of fauna collected (see ‘‘Results’’ section), we used functional groups and relative size, rather than taxonomic groupings, to create a Bray–Curtis dissimilarity matrix based on abundance. In doing so, we assumed that taxa in the same functional group might respond similarly to the risk of crab predation (Langerhans and DeWitt 2004; Sih et al. 2010). We conducted an analysis of similarity (ANOSIM) to determine if the dissimilarity in community composition between treatments was significant. Finally, we dissected the stomachs of green crabs collected on the last day of the experiment to determine if crabs ingested eelgrass, and whether crab size affected eelgrass consumption. After inspecting the stomach contents visually, we homogenized each sample by agitation in 99% ethanol with a sterilized tungsten bead. After centrifuging and decanting the excess ethanol, the remaining tissue samples were sent to the Canadian Centre for DNA Barcoding (University of Guelph, Ontario) for metabarcoding (see Online Resource 1 for details). Samples were compared to a custom Barcode of Life Database (BOLD) reference database for rbcLa marker in plants and assigned an identity using the BLAST algorithm. This allowed us to potentially detect both Z. marina and Z. japonica. Taxonomic identifications for species detected in each sample were accepted as genuine if they were supported by at least 100 reads that matched a reference sequence with at least 98% identity across at least 100 base pairs. Results Over the course of the experiment we observed crabs actively digging in the sediment in the enclosures and 123 B. R. Howard et al. consuming eelgrass rhizomes, benthic fauna, and detritus. Observations of fresh remnants of bivalves and small green crabs (i.e., \ 50 mm CW, small enough to enter and exit enclosures) suggested that the enclosed green crabs were eating, although we cannot definitively attribute this physical evidence of predation exclusively to our experimental green crabs. We did not observe all of our enclosed crabs at every site visit; however, this was likely due to crabs burying into the sediment, not because of migration or mortality. During the last field survey, we observed all of the crabs in the low-density enclosures and 39 (of 48) of the crabs in the high-density enclosures. Ultimately, we recaptured 33 out of 60 crabs: 22 (of 48) from the high-density plots and 11 (of 12) from the low-density plots before the incoming tide restricted our ability to keep searching the enclosures. Eelgrass shoot density decreased over time in all treatments, from an initial estimated average density of 796 shoots m-2, based on counts from our undisturbed enclosure-control plots. Average eelgrass shoot density was better predicted by the model including treatment than by the null model (Likelihood ratio test: X2 = 110.93, df = 7, p \ 0.001). Eelgrass shoot density declined at an average rate of 4.1 to 7.5 shoots m-2 day-1 across the low-density, no-crab, and enclosure-control treatments, with extensive overlap among these treatments (Fig. 2). However, in the high-density treatment, eelgrass shoot density declined 2.4 to 4 times faster, by an average of 17.6 shoots m-2 day-1 (Fig. 2). The average ratio of blades to rhizome biomass was also better predicted by the treatment model than the null model (Likelihood ratio test: X2 = 15.57, df = 6, p = 0.001). The ratios across all treatments were below 1, meaning that above-ground (blade) biomass was consistently less than below-ground (rhizome) biomass. The average ratio of blades to rhizomes in the enclosure-control plots was the highest of the four treatments (Fig. 3a). The observed average ratio of blades to rhizomes in the high-density treatment represents a 77.2% relative decrease in eelgrass blade biomass compared to the enclosure-control treatment (Tukey’s HSD post-hoc test, p \ 0.001) and a 42.7% relative decrease compared to the low-density treatment (Tukey’s HSD post-hoc test, p = 0.002) (Fig. 3a). There was also a 58.9% relative decrease in blade biomass in the no-crab treatment compared to the enclosure-control treatment (Tukey’s HSD post- 123 Fig. 2 Change in average eelgrass shoot density (scaled to m-2) in Mayne Bay, British Columbia, over a 4-week period in plots containing variable numbers of invasive green crabs. Lines represent the predicted average derived from the generalized linear mixed-effects model. Shaded areas represent 95% confidence intervals. Raw data (average of all samples per plot) are indicated by open points. In addition to low and high green crab density treatments (1.4 and 5.6 green crabs m-2, respectively), there was a no-crab enclosure treatment (0 crabs m-2), and an enclosure-control treatment that experienced the ambient density of green crab present at the field site hoc test, p = 0.03) (Fig. 3a). There were no significant differences in the ratio of blades to rhizomes for any other HSD post-hoc contrasts. The model of average benthic fauna wet weight that included a treatment effect was not significantly different than the null model (Likelihood ratio test: X2 = 5.06, df = 6, p = 0.17). There was no significant difference in benthic fauna weight among treatments (Fig. 3b). Moreover, there was no detectable dissimilarity in community composition between treatments (ANOSIM, R = - 0.05, p = 0.80; Fig. 4). We visually identified eelgrass rhizomes in the stomachs of 12 crabs (of 33). This was consistent with observations of crabs injesting rhizomes during the experiment and with barcoding results. Of the 19 individual stomach samples successfully barcoded, 42% contained Z. marina and 26% contained Z. japonica, with one crab stomach containing both species. There was 75% agreement between our visual identification of eelgrass rhizomes and barcoding, and barcoding detected Zostera in an additional 10 stomachs where material was too digested for visual Habitat alteration by invasive European green crab (Carcinus maenas) causes eelgrass… Fig. 3 Regression coefficients of the a average ratio of dry weights of eelgrass blades to rhizomes, and b average benthic fauna wet weight, from plots containing variable densities of invasive green crabs at the end of a 4-week period. Points are the unlogged predicted coefficients from the generalized mixedeffects models and errors bars the 95% confidence interval (n = 6 in all cases). In addition to low and high green crab density treatments (1.4 and 5.6 green crabs m-2, respectively), there was a no-crab enclosure treatment (0 crabs m-2), and an enclosure-control treatment that experienced the ambient density of green crab present at the field site low-density enclosures (45%, n = 11 processed samples) (X2 = 9.68, p \ 0.01). There was no significant difference in the size of crabs (CW) with or without Zostera in their stomach contents (t17 = - 0.75, p = 0.46). Discussion Fig. 4 Non-metric multidimensional scaling (nMDS) plot of benthic fauna community composition in plots containing variable densities of invasive green crabs at the end of a 4-week period. Fauna were classified based on functional group and relative size class, rather than taxonomy (Online Resource 2). Each point represents a plot (n = 6 plots per treatment). In addition to low and high green crab density treatments (1.4 and 5.6 green crabs m-2, respectively), there was a no-crab enclosure treatment (0 crabs m-2), and an enclosure-control treatment that experienced the ambient density of green crab present at the field site. The stress value of 0.15 suggests an acceptable degree of distortion required to plot multidimensional dissimilarity rankings in two dimensions (Clarke and Ainsworth 1993) identification. Zostera was found in a significantly higher proportion of crabs from high-density enclosures (88%, n = 8 processed samples) compared to Anthropogenic stressors such as invasive species can cause drastic declines of habitat-forming species in coastal ecosystems, but the severity of these impacts may vary geographically (Molnar et al. 2008; Padilla 2010; Guy-Haim et al. 2018). Despite the presence of European green crabs and eelgrass on both the Atlantic and Pacific coasts of North America, the contextdependent differences between these two regions led us to ask whether the impacts of green crabs on eelgrass ecosystems on the Pacific coast were different than the impacts previously observed on the Atlantic coast. Broadly speaking, we found that the impacts of invasive green crabs on the Pacific coast were consistent with those seen on the Atlantic coast, where green crabs are capable of significantly altering 123 B. R. Howard et al. eelgrass habitats, particularly at high green crab densities (Davis et al. 1998). Eelgrass shoot loss in our high-density treatment was three times greater than natural, seasonal senescence. We also observed that the removal of above-ground eelgrass biomass, through shredding by green crabs, was more common than the removal of whole plants (i.e., bioturbation). Both mechanisms have previously been observed (Davis et al. 1998; Malyshev and Quijón 2011; Garbary et al. 2014); however, our additional observations of direct consumption of below-ground biomass (rhizomes) by large adult male crabs are novel. Contrary to Thompson (2007) who found that enclosed green crabs had significant direct and indirect effects on invertebrate fauna in Atlantic eelgrass beds (with no change in eelgrass biomass) over a 4-week period, we did not detect changes in benthic fauna biomass or community composition at any density of green crab. The rapid loss of eelgrass at high densities of green crabs suggests that these invaders can be ecosystem engineers on the Pacific coast and could have a similar, or potentially greater, long-term negative impact on Pacific than on Atlantic eelgrass ecosystems. In many cases, the density of an invasive species is a well-established determinant of impact severity (Parker et al. 1999; Thomsen et al. 2011). While green crabs had no discernable effect on shoot density in low-density enclosures or enclosure-control plots exposed to ambient densities of green crabs, the rate of eelgrass loss in our high-density treatment was rapid. After accounting for natural, site-wide declines in eelgrass due to seasonal and/or cage effects (i.e., an average decline of 4.1 shoots m-2 day-1 in our nocrab enclosures), we estimated that green crabs in high-density plots removed an average of 13.5 eelgrass shoots m-2 day-1. This is a much faster rate of eelgrass loss than previously detected on the Atlantic coast, where similar densities of enclosed green crabs (4.4 crabs m-2) reduced shoot density by only 4.1 shoots m-2 day-1 on average over 7 days (Garbary et al. 2014). The difference in eelgrass loss between the two studies may be partially due to study duration as our study ran longer than that of Garbary et al. (2014), potentially magnifying the localized impact of crabs on both prey and habitat (Steele 1996; Fernandes et al. 1999). Another major difference was that the Atlantic bed Garbary et al. (2014) studied was only 20% as dense as the bed in our study (Garbary 123 et al. 2014: 172 shoots m-2; this study: 796 shoots m-2). When described in terms of percent bed loss after 1 week, our results (14% loss) are comparable to those of Garbary et al. (2014) (15% loss). At high green crab densities, eelgrass on the Pacific coast may therefore be lost at a much faster rate than natural senescence, which suggests a risk of severe eelgrass ecosystem degradation and loss at several invaded sites in the region. The mechanism of green crab disturbance in our study was likely blade shredding, not bioturbation. This conclusion is supported by the reduced ratio of above- to below-ground eelgrass biomass in enclosures with high green crab densities, which points mainly to the removal of above-ground material. A consistent ratio of blade to rhizome weight would have indicated the removal of entire plants and suggested bioturbation. While crab density had an effect on the ratio of above- to below-ground eelgrass biomass, some of the reduction in above-ground eelgrass biomass in enclosures may have been due to shading or other enclosure artifacts as reflected by differences between our no-crab and enclosure-control plots. Blade shredding might initially be less detrimental to eelgrass bed persistence than bioturbation, because the latter results in the uprooting of whole plants, including the rhizomes (Harrison 1979). However, the loss of rhizomes may be particularly important only in stable environments, when Z. marina is a perennial plant that maintains beds predominantly through clonal growth (Boese et al. 2009). In environments with greater natural disturbance (i.e., temperature or salinity fluctuations, ice scouring, grazing), Z. marina tends to be an annual plant, flowering and releasing seeds which regrow the bed each year (Ruesink et al. 2010). In such cases, blade shredding could compromise eelgrass bed persistence, and indeed, the loss of above-ground material has led to bed collapse (Holdredge et al. 2009; Nowicki et al. 2018). The impact of blade shredding by green crabs on eelgrass bed persistence may therefore be predicated on the dominant reproductive strategy of a bed. Regardless of the mechanism of removal (i.e., shredding or bioturbation), the loss of eelgrass due to green crab on the Pacific coast is not just an indirect consequence of behaviours such as sheltering or foraging, but also due to direct consumption. We detected Zostera both visually and genetically in crab stomachs, especially in the high-density treatment, Habitat alteration by invasive European green crab (Carcinus maenas) causes eelgrass… and we were able to confirm for the first time that eelgrass ingestion is not limited to juvenile green crabs. Moreover, we were able to visually identify rhizomes in a third of crab stomachs and incidentally observed rhizome consumption during our field surveys. Juvenile green crabs on the Atlantic coast are only known to consume eelgrass blades and meristems (Malyshev and Quijón 2011). A rigorous test of green crab consumption preferences (blades vs. rhizomes) and rates may be an important avenue of future research, especially if green crab have a preference for (or against) invasive Z. japonica, as has been shown for ducks and geese (Baldwin and Lovvorn 1994). Despite evidence of active predation by green crabs on benthic fauna in our enclosures and drastic losses in eelgrass cover in some enclosures, we did not detect differences in the biomass or composition of benthic fauna among treatments. This is contrary to similar enclosure experiments, both inside and outside of eelgrass beds on both coasts, which found significant effects of green crabs on a wide range of invertebrate species (Thompson 2007; Whitlow 2010; Estelle and Grosholz 2012). Our inability to detect a community shift in benthic fauna as a result of green crab activity might have been caused by the low faunal density at our site (Online Resource 2), the low taxonomic resolution of the data, or the duration of the experiment, which may have been long enough to allow for recolonization of fauna (Fernandes et al. 1999). In addition, there may have also been changes to epifauna on the eelgrass itself and nektonic fauna, including fish, that were not tested here. On the Pacific coast, critically important fish species, including Pacific salmon species (e.g., chinook) and Pacific herring that rely on eelgrass ecosystems extensively for survival and reproduction (Hosack et al. 2006; Kennedy et al. 2018). The loss of eelgrass habitat may contribute to declines in salmon abundance, as the growth of out-migrating juveniles in these estuarine ecosystems predicts future adult abundance (Tomaro et al. 2012). When juvenile salmon are forced to forage in less productive, more risky ecosystems, their growth rates are slower, potentially resulting in worse survival outcomes (David et al. 2016; Kennedy et al. 2018). The impacts of an invasive ecosystem engineer will depend on factors specific to both the invasion, such as the density of the invader, and of the habitat itself, such as the reproductive strategy of a specific eelgrass bed. Despite this context-dependency, our study indicates that on the Pacific coast, green crab impacts on eelgrass should be similar, if not greater, to those observed on the Atlantic coast, especially at sites with high densities of green crabs. On the Atlantic coast, large declines in eelgrass and subsequent changes to local biodiversity indicate that European green crabs are ecosystem engineers capable of ecosystem-wide impacts (Klassen and Locke 2007; Matheson et al. 2016). Although green crabs have been established at high densities in areas on the Pacific coast for more than two decades, declines in eelgrass beds coincident with high green crab densities have not been reported to date. However, this may represent a failure of detection rather than a true difference in Atlantic and Pacific green crab impacts; the densest populations of green crabs on the Pacific coast are found in remote areas and therefore less closely observed than populations on the Atlantic coast. A case in point is eelgrass loss in Benoit Cove, Nova Scotia, which was noted by local residents prior to the start of formal research in the area (Garbary et al. 2014). It is possible that high population densities of green crabs in British Columbia may have already had a significant impact on eelgrass habitat, but that a habitat shift has gone unobserved (or at least unreported). On the other hand, isolation from human populations may mean that invaded Pacific eelgrass beds are under less anthropogenic stress from other sources (e.g., anchors, pollution, shading) than their Atlantic counterparts. Fewer overall stressors could increase the resilience of Pacific eelgrass beds to green crab disturbance. As green crabs on the Pacific coast of North America spread to more populated areas, it is possible that the combined effect of high green crab densities with other coastal stressors will result in observations of ecosystem-wide impacts of green crabs in eelgrass habitats, as has been observed on the Atlantic coast. Acknowledgements We thank Dickson Wong, Sarah Calbick, Kyla Jeffrey, Emma Atkinson, Andrew Bateman, Helen Yan, and Elizabeth Oishi for their help, and the Bamfield Marine Sciences Centre for logistical support. Funding The study was funded by the Second Canadian Aquatic Invasive Species Network (CAISN II) (BRH, IMC, TWT), a Natural Sciences and Engineering Research Council of Canada (NSERC) Discovery Grant (IMC), an NSERC - Canada Graduate Scholarships-Doctoral award (FTF), and the Fisheries and Oceans Canada’s Aquatic Invasive Species program (TWT). 123 B. R. Howard et al. Compliance with ethical standards Conflict of interest The authors declare that they have no conflict of interest. References Archambault P, Snelgrove PVR, Fisher JAD, Gagnon JM, Garbary DJ, Harvey M, Kenchington EL, Lesage V, Levesque M, Lovejoy C, Mackas DL, McKindsey CW, Nelson JR, Pepin P, Piché L, Poulin M (2010) From sea to sea: Canada’s three oceans of biodiversity. PLoS ONE. https:// doi.org/10.1371/journal.pone.0012182 Baldwin JR, Lovvorn JR (1994) Expansion of seagrass habitat by the exotic Zostera japonica, and its use by dabbling ducks and brant in Boundary Bay, British Columbia. Mar Ecol Prog Ser 103:119–127. https://doi.org/10.3354/ meps103119 Behrens Yamada S (2001) Global invader: the European green crab. Oregon State University, Corvallis Bertness MD (1984) Habitat and community modification by an introduced herbivorous snail. Ecology 65:370–381. https:// doi.org/10.2307/1941400 Boese BL, Kaldy JE, Clinton PJ, Eldridge PM, Folger CL (2009) Recolonization of intertidal Zostera marina L. (eelgrass) following experimental shoot removal. J Exp Mar Biol Ecol 374:69–77. https://doi.org/10.1016/j.jembe.2009.04. 011 Bolker BM (2008) Modeling variance. In: Ecological models and data in R. Princeton University Press, Princeton, pp 316–336 British Columbia Marine Conservation Analysis (2011) Marine atlas of Pacific Canada. British Columbia marine conservation analysis, Vancouver, BC Choi FMP, Murray CC, Therriault TW, Pakhomov EA (2016) Intertidal invasion patterns in Canadian ports. Mar Biol 163:1–12. https://doi.org/10.1007/s00227-016-2957-0 Clarke KR, Ainsworth M (1993) A method of linking multivariate community structure to environmental variables. Mar Ecol Prog Ser 92:205–219. https://doi.org/10.3354/ meps092205 Cognie B, Haure J, Barillé L (2006) Spatial distribution in a temperate coastal ecosystem of the wild stock of the farmed oyster Crassostrea gigas (Thunberg). Aquaculture 259:249–259. https://doi.org/10.1016/j.aquaculture.2006. 05.037 Costello MJ, Coll M, Danovaro R, Halpin P, Ojaveer H, Miloslavich P (2010) A census of marine biodiversity knowledge, resources, and future challenges. PLoS ONE. https:// doi.org/10.1371/journal.pone.0012110 David AT, Simenstad CA, Cordell JR, Toft JD, Ellings CS, Gray A, Berge HB (2016) Wetland loss, juvenile salmon foraging performance, and density dependence in Pacific Northwest estuaries. Estuaries Coasts 39:767–780. https:// doi.org/10.1007/s12237-015-0041-5 Davis R, Short F, Burdick D (1998) Quantifying the effects of green crab damage to eelgrass transplants. Restor Ecol 6:297–302 123 de Moura Queirós A, Hiddink JG, Johnson G, Cabral HN, Kaiser MJ (2011) Context dependence of marine ecosystem engineer invasion impacts on benthic ecosystem functioning. Biol Invasions 13:1059–1075. https://doi.org/10. 1007/s10530-011-9948-3 Estelle V, Grosholz ED (2012) Experimental test of the effects of a non-native invasive species on a wintering shorebird. Conserv Biol 26:472–481. https://doi.org/10.1111/j.15231739.2011.01820.x Fei S, Phillips J, Shouse M (2014) Biogeomorphic impacts of invasive species. Annu Rev Ecol Evol Syst 45:69–87. https://doi.org/10.1146/annurev-ecolsys-120213-091928 Fernandes TF, Huxham M, Piper SR (1999) Predator caging experiments: a test of the importance of scale. J Exp Mar Biol Ecol 241:137–154. https://doi.org/10.1016/S00220981(99)00076-3 Garbary DJ, Miller AG, Williams J, Seymour NR (2014) Drastic decline of an extensive eelgrass bed in Nova Scotia due to the activity of the invasive green crab (Carcinus maenas). Mar Biol 161:3–15. https://doi.org/10.1007/s00227-0132323-4 Gillespie GE, Phillips AC, Paltzat DL, Therriault TW (2007) Status of the European green crab, Carcinus maenas, in British Columbia - 2006. Can Tech Rep Fish Aquat Sci 2700:vii-39 Gotceitas V, Fraser S, Brown JA (1997) Use of eelgrass beds (Zostera marina) by juvenile Atlantic cod (Gadus morhua). Can J Fish Aquat Sci 54:1306–1319. https://doi.org/10. 1139/f97-033 Green DS, Crowe TP (2014) Context- and density-dependent effects of introduced oysters on biodiversity. Biol Invasions 16:1145–1163. https://doi.org/10.1007/s10530-0130569-x Guy-Haim T, Lyons DA, Kotta J, Ojaveer H, Queirós AM, Chatzinikolaou E, Arvanitidis C, Como S, Magni P, Blight AJ, Orav-Kotta H, Somerfield PJ, Crowe TP, Rilov G (2018) Diverse effects of invasive ecosystem engineers on marine biodiversity and ecosystem functions: a global review and meta-analysis. Glob Change Biol. https://doi. org/10.1111/gcb.14007 Harrison PG (1979) Reproductive strategies in intertidal populations of two co-occurring seagrasses (Zostera spp.). Can J Bot 2:2635–2638 Heck KLJ, Hays G, Orth RJ (2003) Critical evaluation of nursery hypothesis for seagrasses. Mar Ecol Prog Ser 253:123–136. https://doi.org/10.3354/meps253123 Holdredge C, Bertness MD, Altieri AH (2009) Role of crab herbivory in die-off of New England salt marshes. Conserv Biol 23:672–679. https://doi.org/10.1111/j.1523-1739. 2008.01137.x Hosack GR, Dumbauld BR, Ruesink JL, Armstrong DA (2006) Habitat associations of estuarine species: Comparisons of intertidal mudflat, seagrass (Zostera marina), and oyster (Crassostrea gigas) habitats. Estuaries Coasts 29:1150–1160. https://doi.org/10.1007/BF02781816 Jones CG, Lawton JH, Shachak M (1994) Organisms as ecosystem engineers. Oikos 69:373–386 Kelly JR, Proctor H, Volpe JP (2008) Intertidal community structure differs significantly between substrates dominated by native eelgrass (Zostera marina L.) and adjacent to the introduced oyster Crassostrea gigas (Thunberg) in Habitat alteration by invasive European green crab (Carcinus maenas) causes eelgrass… British Columbia, Canada. Hydrobiologia 596:57–66. https://doi.org/10.1007/s10750-007-9057-6 Kennedy LA, Juanes F, El-Sabaawi R (2018) Eelgrass as valuable nearshore foraging habitat for juvenile Pacific salmon in the early marine period. Mar Coast Fish 10:190–203. https://doi.org/10.1002/mcf2.10018 Klassen G, Locke A (2007) A biological synopsis of the European green crab, Carcinus maenas. Can Manuscr Rep Fish https://doi.org/10.1007/ Aquat Sci 2818:vii-75. BF00348935 Langerhans RB, DeWitt TJ (2004) Shared and unique features of evolutionary diversification. Am Nat 164:335–349. https://doi.org/10.1086/422857 Malyshev A, Quijón PA (2011) Disruption of essential habitat by a coastal invader: new evidence of the effects of green crabs on eelgrass beds. ICES J Mar Sci 68:1852–1856. https://doi.org/10.1093/icesjms/fsr126 Matheson K, McKenzie CH, Gregory RS, Robichaud DA, Bradbury IR, Snelgrove PVR, Rose GA (2016) Linking eelgrass decline and impacts on associated fish communities to European green crab Carcinus maenas invasion. Mar Ecol Prog Ser 548:31–45. https://doi.org/10.3354/ meps11674 Molnar JL, Gamboa RL, Revenga C, Spalding MD (2008) Assessing the global threat of invasive species to marine biodiversity. Front Ecol Environ 6:485–492. https://doi. org/10.1890/070064 Moore KA, Short FT (2006) Zostera: biology, ecology, and management. In: Larkum AWD, Orth RJ, Duarte C (eds) Seagrasses: biology, ecology and conservation. Springer, Dordrecht, pp 361–386 Moore JW, Gordon J, Carr-Harris C, Gottesfeld AS, Wilson SM, Russell JH (2016) Assessing estuaries as stopover habitats for juvenile Pacific salmon. Mar Ecol Prog Ser 559:201–215. https://doi.org/10.3354/meps11933 Murtaugh PA (2007) Simplicity and complexity in ecological data analysis. Ecology 88:56–62 Neckles HA (2015) Loss of eelgrass in Casco Bay, Maine, linked to green crab disturbance. Northeast Nat 22:478–500. https://doi.org/10.1656/045.022.0305 Nowicki R, Fourqurean J, Heithaus M (2018) The role of consumers in structuring seagrass communities: direct and indirect mechanisms. In: Larkum A, Kendrick G, Ralph P (eds) Seagrasses of Australia. Springer, Berlin, pp 491–540 Oksanen et al (2012) Vegan: community ecology package. R package version 2.5–3. http://CRAN.R-project.org/package= vegan/ Padilla DK (2010) Context-dependent impacts of a non-native ecosystem engineer, the Pacific oyster Crassostrea gigas. Integr Comp Biol 50:213–225. https://doi.org/10.1093/icb/ icq080 Parker IM, Simberloff D, Lonsdale WM, Goodell K, Wonham M, Kareiva PM, Williamson MH, Von Holle B, Moyle PB, Byers JE, Goldwasser L (1999) Impact: toward a framework for understanding the ecological effects of invaders. Biol Invasions 1:3–19. https://doi.org/10.1023/A: 1010034312781 Phillips RC, McMillan C, Bridges KW (1983a) Phenology of eelgrass, Zostera marina L., along latitudinal gradients in North America. Aquat Bot 15:145–156. https://doi.org/10. 1016/0304-3770(83)90025-6 Phillips RCR, Grant WS, McRoy CP (1983b) Reproductive strategies of eelgrass (Zostera marina L.). Aquat Bot 16:1–20. https://doi.org/10.1016/0304-3770(83)90047-5 Pikitch EK, Rountos KJ, Essington TE, Santora C, Pauly D, Watson R, Sumaila UR, Boersma PD, Boyd IL, Conover DO, Cury P, Heppell SS, Houde ED, Mangel M, Plagányi É, Sainsbury K, Steneck RS, Geers TM, Gownaris N, Munch SB (2014) The global contribution of forage fish to marine fisheries and ecosystems. Fish Fish 15:43–64. https://doi.org/10.1111/faf.12004 R Development Core Team (2008) R: a language and environment for statistical computing. R Foundation for Statistical Computing, Vienna Robertson AI, Mann KH (1984) Disturbance by ice and lifehistory adaptations of the seagrass Zostera marina. Mar Biol 80:131–141. https://doi.org/10.1007/BF02180180 Romero GQ, Gonçalves-Souza T, Vieira C, Koricheva J (2015) Ecosystem engineering effects on species diversity across ecosystems: a meta-analysis. Biol Rev 90:877–890. https:// doi.org/10.1111/brv.12138 Ruesink JL, Hong JS, Wisehart L, Hacker SD, Dumbauld BR, Hessing-Lewis M, Trimble AC (2010) Congener comparison of native (Zostera marina) and introduced (Z. japonica) eelgrass at multiple scales within a Pacific Northwest estuary. Biol Invasions 12:1773–1789. https://doi.org/10. 1007/s10530-009-9588-z Shafer DJ, Kaldy JE, Gaeckle JL (2014) Science and management of the introduced seagrass Zostera japonica in North America. Environ Manag 53:147–162. https://doi.org/10. 1007/s00267-013-0172-z Shafer DJ, Swannack TM, Saltus C et al (2016) Development and validation of a habitat suitability model for the nonindigenous seagrass Zostera japonica in North America. Manag Biol Invasions 7:141–155. https://doi.org/10.3391/ mbi.2016.7.2.02 Shelton AO, Francis TB, Williams GD, Feist B, Stick K, Levin PS (2014) Habitat limitation and spatial variation in Pacific herring egg survival. Mar Ecol Prog Ser 514:231–245. https://doi.org/10.3354/meps10941 Short F, Carruthers T, Dennison W, Waycott M (2007) Global seagrass distribution and diversity: a bioregional model. J Exp Mar Biol Ecol 350:3–20. https://doi.org/10.1016/j. jembe.2007.06.012 Sih A, Bolnick DI, Luttbeg B, Orrock JL, Peacor SD, Pintor LM, Preisser E, Rehage JS, Vonesh JR (2010) Predator-prey naı̈veté, antipredator behavior, and the ecology of predator invasions. Oikos 119:610–621. https://doi.org/10.1111/j. 1600-0706.2009.18039.x Steele MA (1996) Effects of predators on reef fishes: separating cage artifacts from effects of predation. J Exp Mar Biol Ecol 198:249–267 Talley TS, Crooks JA, Levin LA (2001) Habitat utilization and alteration by the invasive burrowing isopod, Sphaeroma quoyanum, in California salt marshes. Mar Biol 138:561–573. https://doi.org/10.1007/s002270000472 Thompson WJ (2007) Population-level effects of the European green crab (Carcinus maenas, L.) in an eelgrass community of the southern Gulf of St. Lawrence. Dissertation. University of New Brunswick Thomsen MS, Wernberg T, Olden JD, Griffin JN, Silliman BR (2011) A framework to study the context-dependent 123 B. R. Howard et al. impacts of marine invasions. J Exp Mar Biol Ecol 400:322–327. https://doi.org/10.1016/j.jembe.2011.02.033 Tomaro LM, Teel DJ, Peterson WT, Miller JA (2012) When is bigger better? Early marine residence of middle and upper Columbia River spring Chinook salmon. Mar Ecol Prog Ser 452:237–252. https://doi.org/10.3354/meps09620 Wagner E, Dumbauld BR, Hacker SD, Trimble AC, Wisehart LM, Ruesink JL (2012) Density-dependent effects of an introduced oyster, Crassostrea gigas, on a native intertidal seagrass, Zostera marina. Mar Ecol Prog Ser 468:149–160. https://doi.org/10.3354/meps09952 Whitlow WL (2010) Changes in survivorship, behavior, and morphology in native soft-shell clams induced by invasive green crab predators. Mar Ecol 31:418–430. https://doi. org/10.1111/j.1439-0485.2009.00350.x 123 Zuur AF, Ieno EN, Walker NJ, Saveliev AA, Smith GM (2009) Mixed effects models and extensions in ecology with R. Springer, New York Zwerschke N, Hollyman PR, Wild R et al (2018) Limited impact of an invasive oyster on intertidal assemblage structure and biodiversity: the importance of environmental context and functional equivalency with native species. Mar Biol 165:1–13. https://doi.org/10.1007/s00227-018-3338-7 Publisher’s Note Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.