Impacts of Marine Invasive Species on Subtidal and Intertidal Food Web Dynamics

Invasive species can have a variety of impacts on food web structure and interspecific interactions. They can impact recruitment rates of associated marine species, influence behavior of potential prey items, and alter predator-prey relationships. This research is designed to assess 1) the relationship between the recruitment of Lacuna vincta and two species of algal epiphytes, the native Ceramium virgatum and the invasive Neosiphonia harveyi, in the shallow subtidal zone; 2) the spatial and temporal distribution of the invasive Hemigrapsus sanguineus in the intertidal zone of cobble beaches; and 3) the top-down effects and predator-prey interactions of H. sanguineus. Through manipulative field experiments, we found that the presence of algal epiphytes facilitated the recruitment rate of Lacuna vincta, regardless of the epiphyte species composition. We also found a positive relationship between the number of L. vincta present and epiphyte recruitment, which is disproportionately driven by higher recruitment of Neosiphonia harveyi than Ceramium virgatum. Long-term monitoring can be used to understand population trends of invasive species. Through monthly surveys in Narragansett Bay, Rhode Island, we determined that Hemigrapsus sanguineus densities are highest in the early summer and early fall months. Juvenile H. sanguineus are most abundant in June and July and gravid females are most prevalent in August and September. H. sanguineus exhibited a density gradient with the highest densities in the northern section of Narragansett Bay and declining toward the mouth at the southern end of Narragansett Bay. Invasive species can outcompete native and established species, thereby altering food web dynamics through changes in top-down and predator-prey interactions. Through mesocosm studies, we found that while Hemigrapsus sanguineus has an impact on Littorina littorea behavior, it does not alter the perwinkles’ grazing rates. By contrast, the combined presence of H. sanguineus and L. littorea results in a greater decrease in algal biomass than only L. littorea. In field tethering experiments, we observed that abiotic but not biotic factors were the dominant force in structuring the vertical distribution of H. sanguineus. Overall, we found that H. sanguineus does not occupy the same ecological niche as Carcinus maenas, the previously dominant crab in the intertidal zone of cobble beaches. This research provides insight into how invasive species shape the suband intertidal zones by influencing the recruitment rate and behavior of native and established species. Given that marine invasions are occurring at an increasing rate due to international marine transportation, human-mediated introductions, and global climate change, fully understanding the impacts of these invasive species is critical to mitigating and adapting to changes in species composition and abundance.

Through mesocosm studies, we found that while Hemigrapsus sanguineus has an impact on Littorina littorea behavior, it does not alter the perwinkles' grazing rates. By contrast, the combined presence of H. sanguineus and L. littorea results in a greater decrease in algal biomass than only L. littorea. In field tethering experiments, we observed that abiotic but not biotic factors were the dominant force in structuring the vertical distribution of H. sanguineus. Overall, we found that H. sanguineus does not occupy the same ecological niche as Carcinus maenas, the previously dominant crab in the intertidal zone of cobble beaches.
This research provides insight into how invasive species shape the sub-and intertidal zones by influencing the recruitment rate and behavior of native and established species. Given that marine invasions are occurring at an increasing rate due to international marine transportation, human-mediated introductions, and global climate change, fully understanding the impacts of these invasive species is critical to mitigating and adapting to changes in species composition and abundance.
iv ACKNOWLEDGMENTS First and foremost, I thank my major professor, Carol Thornber. Through her guidance, support, and training, I became a better researcher, vastly improved my writing skills, and discovered an enjoyment of teaching. Perhaps even more importantly than that, she allowed me the leeway to pursue a nontraditional graduate career that took me from Rhode Island to California and Washington DC. These meandering travels eventually led to my current career path. My committee members Evan Preisser, Graham Forrester, Kenny Raposa, and Scott McWilliams provided honest and insightful recommendations on manuscripts, grant proposals, and scientific presentations. Even though they often did not hear from me for long stretches of time as I went on interdisciplinary excursions, they always knew I would finish the graduate journey. A special thanks to Kenny for working side-by-side with me on Prudence Island each month as we braved all weather to collect, count, and measure tens of thousands of crabs.
There are no words to effectively convey my gratitude to IGERT, which was the single most transformative experience of my graduate career. It exposed me to alternatives to a life in academia and provided the skills I needed to enter into a career in environmental policy. Thank you to Pete August, Q Kellogg, and Judith Swift, who saw something in me during my IGERT interview that inspired them to accept me into the Fellowship program; it was at that point my two roads diverged, and I -I took the one less traveled by, and that has made all the difference. My fellow co-07ers, Carrie Byron, Kim Lellis-Dibble, and Nate Vinhaterio stand out among my support system. We haven't been CIIPers for three years, but I still seek out their advice over beers and gossip.
v I was extremely blessed to meet a slew of friends who helped keep me sane during the busy times, prevented me from taking graduate school too seriously, and were always easy to convince to grab a happy hour beer at The Mews -thank you. Josh Atwood, I do not know how we survived the summer of 2007, but I hope when we are old we sit together on a beach in Hawaii and reminisce about the good ole' days at URI. Laura Ingwell, Pam Reitsma, and I bonded over girl-time and wine on Thursday nights.
My research would not have been completed without the endless efforts of the Thornber Undergraduate Army, a never-ending supply of intelligent students who rarely complained about getting up at 4:00 AM to do manual labor or observe the behavior of snails. Thank you to Emily Jones and Heather Miceli, my first labmates in the dark, dank, and dingy office in BISC. Michele Guidone and Chris Newton were more than just my labmates, they are the friends who are a part of every fun experience and great memory I have from URI. We traveled together, mentored each other, and experienced the roller coaster of graduate school.

Introduction
Recruitment is a key process in the population dynamics of many marine species (Bertness et al. 1992, Gaines & Bertness 1992, Miron et al. 1995, White 2007 Gribben et al. 2009b), and the marine Antarctic (Siegel & Loeb 1995). The presence or absence of other macrobenthic species on a suitable substratum is one biotic factor that may influence larval settlement dynamics (Rodriguez et al. 1993, Huggett et al. 2005. One group of macrobenthic species that can influence larval settlement is epiphytic macroalgae (Hall & Bell 1993, Swanson et al. 2006. Epiphytic macroalgae are small, often filamentous species that live attached to larger host macroalgal species. Algal epiphytes can increase the structural complexity of their host algal species, which may, in turn, increase the suitability of the host as habitat for small invertebrates, including herbivores (Martin-Smith 1993). Epiphytes can provide these herbivores protection from predators while also serving as a food source (Pavia et al. 1999). The epiphytes can also buffer the associated herbivores from abiotic stress such desiccation (Salemaa 1986, Bostrom & Mattila 1999. As a result, these herbivores may benefit the host macroalgae by preventing epibiont overgrowth (Stachowicz & Whitlatch 2005) and/or promoting algal growth via nitrogen excretion (Fong et al. 1997, Bracken et al. 2007). ( attached to the shallow subtidal with marine epoxy (A-788 Splash Zone Compound). We left the mimics in the field for one month to allow for epiphyte recruitment. One month later, when the first L. vincta appeared as recruits in the field, we removed the mimics from the intertidal and brought them to the lab in order to manipulate the epiphyte abundance and composition prior to L. vincta recruitment. We divided the 60 mimics evenly into six groups of ten each. Three groups were used for epiphyte abundance 6 experiments and three for epiphyte composition experiments. Mimics were removed from the field for a maximum of twelve hours and were stored in ambient temperature, flowthrough seawater systems at the URI Graduate School of Oceanography when not being processed.

Lacuna vincta
Within the epiphyte abundance experiments, ten mimics retained a high epiphyte density (75% -100% of mimic surface area covered with epiphytes), ten were pruned to have a low epiphyte density (25% -50% of mimic surface area covered with epiphytes), and ten were pruned to remove all visible epiphytes. Within the epiphyte composition experiments, we selectively pruned the mimics such that ten mimics contained only Ceramium virgatum epiphytes, ten had only Neosiphonia harveyi epiphytes, and ten had a mixture of half C. virgatum and half N. harveyi epiphytes.
We photographed each mimic and measured its wet mass after blotting off excess moisture. We then reattached the mimics in the shallow subtidal to allow for natural Lacuna vincta recruitment. After two weeks, we removed the mimics and brought them to the lab for analysis. We photographed each mimic, measured the total wet mass (after blotting the mimics), removed and identified all of the epiphytes, and measured the mass of each epiphyte species. We also removed and counted the L. vincta found on each of the mimics. Data were log transformed and analyzed for differences in L. vincta recruitment among epiphyte abundance and composition with one-way ANOVAs and post-hoc Tukey tests (JMP v. 7; www.sas.com).

Impacts of Lacuna vincta on epiphyte recruitment
During May of 2008, before yearly Lacuna vincta recruitment occurred, we placed one macroalgal mimic in each of 30 one-liter semi-transparent plastic containers with one-mm mesh sides and lids. These containers were then secured directly to the rocky subtidal, using the same method as described above. In the same manner, we also attached mimics in ten open containers, and ten mimics without containers, to additional PVC rings, as controls. After L. vincta were added, we left the containers and mimics in the field for two weeks (this duration was based on pilot data collected in the summer of 2007) to allow for epiphyte recruitment. After these two weeks, we removed the mimics and containers, placed them in individual plastic bags to retain all L. vincta, and brought them to the lab.
We photographed each mimic, measured its wet mass (mimics were first blotted to remove excess moisture), removed and indentified all of the epiphytes, and measured the mass of each epiphyte species. We also removed and counted the Lacuna vincta in the closed containers, on the mimics in the open containers, and on mimics with no 8 containers. Data were analyzed for a relationship between Lacuna density and epiphyte biomass using correlation techniques (JMP v. 7;www.sas.com).

Results and Discussion
Across all of our treatments, we found a significant positive correlation between the proportion of Neosiphonia harveyi (0.02 + 0.006 grams per cm 2 of macroalgal mimic) and the total epiphyte mass (r = 0.92, p > 0.0001). Ceramium virgatum densities were never greater than 0.005 grams per cm 2 of macroalgal mimic regardless of total epiphyte mass, indicating that recruitment densities of C. virgatum were low, or that the C.
virgatum that did recruit to the mimics was rapidly consumed by L. vincta.
Macroalgal mimics containing high or low epiphyte densities had significantly more L. vincta (4.10 + 1.16 per cm 2 macroalgal mimic and 3.86 + 0.88 per cm 2 macroalgal mimic respectively) than mimics with no epiphytes present (0.67 + 0.30 L. vincta per cm 2 macroalgal mimic; post hoc Tukey Kramer test), and there was no significant difference between the high and low epiphyte densities. Data are means + 1 SE. vincta larvae were able to pass through the mesh covering of the containers: recruitment by these 'accidental' individuals produced a range of L. vincta densities in our closed buckets that reached 5x's the level of our stocking density. Due to this, we examined the relationship between the number of L. vincta and epiphyte density in all of the treatments.
When N. harveyi and Ceramium virgatum were analyzed individually, we found a weak but significant, positive relationship between the number of L. vincta and N. harveyi recruitment ( Figure 2B, r = 0.46, p = 0.0026) but no significant relationship between L.
Our results support the hypothesis that Lacuna vincta recruitment is influenced by the presence of epiphytes, and lend support to the growing body of evidence on the importance of biotic interactions in recruitment dynamics. This relationship may be explained, in part, by protection from predators provided by the epiphytes' threedimensional structure (Williams et al. 2002, Henninger et al. 2009). There may be additional interactions taking place on a shorter time frame than our two-week experiments, but we are primarily concerned with longer-scale, community-wide impacts that persist and ultimately shape the intertidal algal community.
Alternatively, epiphytes may provide a source of food for L. vincta. Prior feeding studies in our study location have shown that L. vincta preferentially consume native show correlative support that there is no reciprocal effect of L. vincta on epiphyte recruitment, but further research is needed in this area.  intertidal crabs is more accurate and efficient than using semi-permanent trays buried in the substrate. We found peak densities of H. sanguineus in the early summer and early fall months, with the summer peak driven by small, juvenile crabs. Densities also tended to be highest in the northern section of Narragansett Bay and declined toward the mouth of this estuary. These surveys provide a solid foundation for long-term monitoring of this invasive species and can provide valuable context for future scientific experiments.

Introduction
Marine invasive species commonly occur in coastal waters, where they can have serious ecological and economic impacts (Carlton 1992. Invasive species may experience lower rates of predation than native species, be superior competitors, and may cause declines in native biodiversity (Mack et al. 2000, Keane & Crawley 2002, Callaway & Ridenour 2004, Vermeij et al. 2009). Invasive species can also facilitate the recruitment and survival of native species, but the long-term benefits of these facilitative interactions are unclear (Gribben & Wright 2006, Rodriguez 2006 Invasive species surveillance typically occurs in areas where species are most likely to become established (Buchan & Padilla 2000), but this method is only effective when there is sufficient knowledge of environmental factors that can impact invasive species and may overlook additional locations that introduced species can inhabit. Longterm environmental monitoring can provide information on the biotic and abiotic characteristics of potential invasion sites. Researchers consistently conducting surveys are more likely to notice changes to the biotic community and identify nonnative species before they become established, making control and/or eradication viable options (Mehta et al. 2007). However, the impacts of eradication should be carefully assessed in order to avoid unintended consequences (Bergstrom et al. 2009). Even if eradication does not occur, continuation of long-term monitoring can provide critical information on how the invasive species interacts with and impacts the native community, and changes can be 25 better anticipated and/or mitigated (Lovett et al. 2007, Henry et al. 2008, Lindenmayer & Likens 2010).
Long-term environmental monitoring can provide extremely useful information but is often quite costly (Caughlan & Oakley 2001) and difficult to fund (Lodge et al. 2006, Lovett et al. 2007 (Menge 1976, Trussell et al. 2003, Trussell et al. 2006.
Hemigrapsus sanguineus frequently outcompetes Carcinus maenas for both available habitat and preferred prey (Jensen et al. 2002, Griffen 2007, 2011, and it may have a lower mortality rate due to predators a lack of predators and/or the substrate complexity of cobble beaches, which provides refuge from predators (Kim & O'Connor 2007, Heinonen & Auster 2012 Detailed demographic profiles such as these can also provide context for manipulative experiments by scientific researchers, and inform decisions by local resource managers who strive to protect native species while containing invasives.

General survey methods
We conducted all surveys on intertidal cobble and small boulder beaches (diameter 20-60 cm; hereafter referred to as 'cobble') in Narragansett Bay, RI (see subsequent sections for particular survey descriptions). All sites were readily accessible from the landward side and were open to the general public. Surveys were conducted monthly during spring low tides + two hours along a transect parallel to, and just above, the low water line, and sample areas (see subsequent sections) were randomly selected along a 30m transect using a random number generator. We collected all crabs > 4mm carapace width from each sample area. We identified each crab to the species level, measured its carapace width (mm), recorded its sex, and, if the crab was female, recorded whether or not it was gravid. For Hemigrapsus sanguineus, all crabs 4mm to 9mm carapace width were recorded as juveniles (for this size range, sex was not able to be determined visually), while crabs > 10 mm carapace width were recorded as adult and either male or female based on sexual dimorphism in the abdomen. Over 99% of the crabs collected from each site were H. sanguineus (N. Rohr, unpubl. data); therefore, we only analyzed data on this species.

Survey methods comparison
In conjunction with NBNERR's ongoing long-term biological monitoring We conducted a concurrent survey using quadrat sampling in June and July 2007 at the same four sites. This was to determine any potential sampling bias due to monthly cobble removal and re-addition. Each month, we sampled three 1m 2 quadrats per site randomly placed along a 30m transect parallel to the water line during spring low tides.
We removed all of the crabs from the sampling area by removing all cobble from the quadrat until the underlying coarse sand substrate was completely revealed. After we removed the cobble and captured and measured the crabs (see General Survey Methods), we replaced the cobbles and released the crabs in the same location.
The number of Hemigrapsus sanguineus collected from each tray was standardized to crabs m -2 . H. sanguineus density data was square root transformed. Data were analyzed for differences in density and mean carapace width between method types (tray and quadrat), among sites, and among months using three-way fixed factor ANOVAs (JMP v.9, www.sas.com). Significant differences were analyzed using a Student's t or Tukey post-hoc analysis, as appropriate. Williams University, Sabin Point, and Save The Bay (Figure 1). These sites were randomly selected from known, publically accessible, cobble intertidal sites. We conducted monthly quadrat surveys (described above) at each site during spring low tides, through October 2008. Data were analyzed for differences in density and mean carapace width among months and sites using two-way fixed factor ANOVAs.

Figure 1 Mean
Significant differences were analyzed using Tukey post-hoc analyses.

Population demographics
We analyzed the data from our ''Narragansett Bay Surveys' for differences between numbers of males and females, adults and juveniles, and gravid and not gravid females to better understand intrapopulation patterns. Data were analyzed using contingency analyses among months and sites with correspondence analyses to evaluate groupings. Results are expressed as percentages to more clearly show comparisons.

Survey methods comparison
The density of H. sanguineus was significantly higher when sampled with the tray method than the quadrat method (52 + 7 vs. 23 + 3 crabs m -2 ; Figure 2, Table 1). We found higher densities of H. sanguineus in June 2007 than July 2007, which is consistent with temporal patterns seen across additional sites and years (described below). When combined across survey methods, we also significantly fewer H. sanguineus at T-Wharf (16 + 4 crabs m -2 ) than at Nag Creek or Stone Wharf (41 + 6 crabs m -2 and 55 + 10 crabs m -2 , respectively).

). When Sabin Point is excluded, which is located in the
Providence River at the northern end of the Bay (Figure 1), H. sanguineus densities generally decreased from north to south (R 2 = 0.5603, p = 0.0125; Figure 5).

Figure 5
A latitudinal gradient of Hemigrapsus sanguineus mean density in Narragansett Bay, averaged from June through October 2008.

Population demographics
We found more male than female Hemigrapsus sanguineus during all months across all sites. Males comprised no less than 56% of the population and peaked at 77% Frequency (# H. sanguineus) ( Figure 7A). We found significant differences in the ratio of males to females among sites (χ 2 = 48.21, p < 0.0001) and months (χ 2 = 22.97, p = 0.0001). We found a higher ratio of males to females in June, July, and September 2008 than in August and October.
Freebody Street had an average of 77% adult male H. sanguineus over all months, which was 10% higher than any other site.

Discussion
Quadrat sampling is widely accepted among scientists for conducting long-term surveys of mobile organisms such as crabs that live on the benthos, (for review, see McIntyre & Eleftherou (eds) 2005). Previous sampling of these species using the quadrat method resulted in a 90-100% sampling efficiency (Lohrer and Whitlatch 2002). While the tray sampling method described here has been used for sampling crabs in other habitat types (Riggs 2003), we do not recommend it for sampling Hemigrapsus sanguineus on cobble intertidal beaches. The tray sampling method inflated crab densities over the quadrat sampling method, and was biased toward larger H. sanguineus, possibly by providing a degree of habitat complexity that protects larger crabs from predators (Lohrer et al. 2000a, Hovel & Lipcius 2001, Lohrer & Whitlatch 2002, Ochwada et al. 2009). Inclement weather also occasionally dislodged trays, and there was evidence of human removal as well. Since uneven sample sizes can inhibit data analysis (Underwood, 2004), it is preferable to utilize a reliable sampling method, such as quadrats, for longterm monitoring. However, if managers are monitoring for H. sanguineus in a previously non-invaded area, a tray may be a better option to attract crabs from a broader area and thereby more likely to detect crabs at a low density.
Along New England shorelines, Hemigrapsus sanguineus have been shown to outcompete the European green crab, Carcinus maenas, which previously was the primary competitor for food resources and space in the rocky intertidal zone (Lohrer et al. 2000b, Jensen et al. 2002, Lohrer & Whitlatch 2002, Griffen 2007. While smaller than C. maenas, H. sanguineus can be found at densities up to 30 times higher than C. maenas ) and thus could have a greater impact on other species. We found that densities of H. sanguineus peaked due to an influx of small, juvenile crabs in the early summer. This influx also resulted in a smaller mean carapace width than in later months. In the Middle Atlantic Bight, newly metamorphosed juveniles have a growth rate of approximately 0.06 mm day -1 (Epifanio et al. 1998). This growth rate could explain the peak in 4-9mm carapace width juveniles, followed approximately 100 days later by a peak in 10-14 mm carapace width crabs.
Gravid females were found in Narragansett Bay, R.I. from June through September, which is consistent with Hemigrapsus sanguineus populations found further south (Epifanio et al. 1998, McDermott 1998. We found ovigerous females as small as 10mm carapace width, but it was not possible to determine if these eggs were viable or successfully metamorphosed. Even though females were gravid in the summer months, we did not see the recruitment of juveniles until the following June. This is due to the cessation of growth during the coldest winter months, and sexual maturity is expected one year after metamorphosis (Epifanio et al. 1998). We found the most gravid females at two of our northern-most sites: Colt State Park and Save the Bay.
Within Narragansett Bay, we found a general increase in Hemigrapsus sanguineus densities as we moved north. However, this did not hold true at one location, Sabin Point, which is located in the Providence River. This could be due to a number of environmental factors including salinity, cobble size, and wave exposure. The Providence River is more heavily influenced by freshwater input from rivers and precipitation runoff from the surrounding urban areas during storm events; salinities as low as 13psu have to determine if H. sanguineus will eventually colonize cobble beaches that currently have low densities of crabs as more adult crabs move onto the beach. If not, then these sites may have restricted recruitment due to ecological and/or physical factors.
Our results provide a solid overview of the temporal and spatial variability in Hemigrapsus sanguineus population density and demographics. Because invasive species are contributing to the rapid alteration of coastal ecosystems, expanded monitoring efforts should be implemented in thoughtful and efficient manners in order to capture this change. In addition, targeted experimental research spurred by the survey patterns would also help to determine the finer intricacies of reproduction and recruitment. If the data 42 collected are managed, analyzed, and reported in a consistent manner, then the return on the effort and investment can be extremely high (Lindenmayer & Likens 2010).

Conclusions
Long-term environmental monitoring serves a critical role in understanding ecosystem trends that occur on yearly or decadal timescales. Our data quantify the temporal and spatial variability, distribution, and life-cycle characteristics of an invasive crab species in a highly impacted estuary. Ongoing monitoring of these sites by NBNERR will increase the knowledge base of this crab's population dynamics, community impacts, and potential for continued spread. There are many cost-effective options to conducting ongoing monitoring studies (Hauser & McCarthy 2009), including recent utilization of citizen science (Delaney et al. 2008, Conrad & Hilchey 2011 and partnering with regional researchers, local agencies, or nongovernmental organizations.
This study utilized academic researchers and a local government agency to maximize time, funds, and effort.
The utility of having long-term, robust data sets has been illustrated repeatedly in ecosystems impacted by natural or anthropogenic disasters, such as the Exxon Valdez spill in 1989 or the BP Deepwater Horizon spill in 2010. A major spill occurred at the entrance of Narragansett Bay in 1989, releasing almost 300,000 gallons of home heating oil and costing $567,000 in damage to the natural environment (NOAA 2009). In these instances, long-term data served as a baseline from which to assess the damages in natural resources. In addition, long-term data sets can influence global policy, such as the now-famous 'Keeling Curve' that illustrates increasing carbon dioxide concentrations in 43 the atmosphere and its connection to global climate change (Sundquist & Keeling 2009).
While no one could have predicted the significance of -or need for -these data, the commitment of researchers to long-term monitoring proved invaluable.
Invasive species have the ability to reshape intertidal marine food webs . Prevention of the introduction of novel species is the first line of defense for protecting our coastal and estuarine ecosystems (Simberloff et al. 2005). However, once an introduction occurs, it becomes important to identify new species while there is still the opportunity to prevent their spread, understand the long-term effects of specific invasive species, and prepare for potential impacts .  (Paine 1966, Carpenter et al. 1987, Kimbro et al. 2009, Newcombe & Taylor 2010, Sieben et al. 2011. Predators can impact community structure either via direct predation, chemical cues that influence the behavior of cooccurring species, and/or the alteration of the physical environment (Paine 1966, Estes et al. 1998, Menge 2000, Robinson et al. 2011. The disruption of this dynamic through species extirpations or introductions can result in severe consequences that have repercussions for the abiotic (i.e. bioturbidation) and biotic (i.e. loss of predators) environments, as well as the coastal communities that depend on them (Levin et al. 2002, Williams & Grosholz 2008, McGeoch et al. 2010).
The susceptibility of trophic cascades to disruption is at least partially influenced by biodiversity (Finke & Denno 2004, Altieri et al. 2010. A high level of biodiversity increases ecosystem resilience by creating redundant functional roles that promote community renewal following a disturbance (Peterson et al. 1998. For this reason, biodiversity is an important indicator of healthy and resilient marine ecosystems, and the maintenance of this ecological health indicator has been highlighted as a management goal for ecosystems around the world (Crowder & Norse 2008, Williams & Grosholz 2008, Palumbi et al. 2009). Biodiversity is naturally dynamic but can have enhanced fluctuations due to human-caused extinctions and species introductions ).
Introductions of exotic species throughout the world's coastal and marine ecosystems are widely recognized as serious biological threats with major ecological and anthropological consequences . For example, San Francisco Bay, CA, is widely recognized as the most invaded estuarine habitat in the world . Particularly aggressive species such as the Asian clam, Potamocorbula amurensis, and shipworms, Teredo navalis, cause hundreds of millions of dollars worth of damage a year (Pimentel et al. 2005). In contrast, estuaries on the northeast coast of the United States do not yet have as high a richness of invasive species, but they are still significantly impacted by nonnative species introductions (see review by Lockwood et al. 2007).
Carcinus maenas (Linnaeus), the European green crab, was introduced to the Atlantic shoreline of the United States approximately 200 years ago ) and became an important predator along shorelines in New England, USA. C.
maenas is a key species in structuring intertidal communities of New England through consumptive and non-consumptive impacts on several species, including Littorina littorea (Linnaeus), the marsh periwinkle , Trussell et al. 2003, Trussell et al. 2006. L. littorea is one of the dominant intertidal invertebrates on northwest Atlantic shorelines (Lubchenco 1978). It preferentially consumes ephemeral green algae such as Ulva (Lubchenco 1978); when the green algae are removed, the algal community shifts towards perennial species such as Chondrus crispus. Since its introduction, C. maenas has contributed to the alteration of intertidal algal abundance and composition by changing L. littorea densities and feeding behaviors via both lethal (direct predation) and non-lethal (release of chemical cues) effects (Trussell 1996, Trussell et al. 2002, Trussell et al. 2006).
More than 150 years after the invasion of Carcinus maenas, another introduced crab has become established in this same region. Hemigrapsus sanguineus (De Haan), the Asian shore crab, was first observed on the coast of New Jersey in 1988  and has since expanded its range from South Carolina through the coast of Maine.
H. sanguineus frequently outcompetes C. maenas for both habitat and preferred prey (Jensen et al. 2002), leading to the displacement of C. maenas from the intertidal zone of cobble beaches (N. Rohr pers. obs., Lohrer & Whitlatch 2002a). The replacement of C. maenas by H. sanguineus could have significant impacts on the already altered native marine flora and fauna in the coastal northwestern Atlantic (e.g. , Freeman & Byers 2006, Griffen & Byers 2009).
There is evidence in laboratory settings that H. sanguineus consume L. littorea up to 13 mm in length (Gerard et al. 1999), but the predation pressure of H. sanguineus in the field is unknown. Chemical cues from H. sanguineus elicit a shell-thickening response in Mytilus edulis in areas of New England where they co-occur, but M. edulis outside of the invaded range of H. sanguineus fail to express this induced trait (Freeman & Byers 2006). By contrast, the non-lethal effects of H. sanguineus on L. littorea are unknown. H. sanguineus does not consume prey as efficiently as C. maenas, but they occur at much higher densities (Lohrer & Whitlatch 2002b. If H. sanguineus impacts L. littorea populations differently than C. maenas, this could shift intertidal algal abundance and species composition.
Invasive species can also alter predator-prey interactions. Crabs and other crustaceans are consumed by predators such as fish (Clark et al. 2006, Kim & O'Connor 2007 and seabirds (Ellis et al. 2005). Juvenile H. sanguineus are consumed by Fundulus heteroclitus and Fundulus majalis (Kim & O'Connor 2007. By contrast, field manipulations of Cancer spp. and C. maenas in New England indicate that gull predation on H. sanguineus is unlikely given their small size (J. Ellis pers. comm., Ellis et al. 2007), but they may be potential prey for crows, shorebirds, and/or terrestrial mammals (Carlton & Hodder 2003, Placyk Jr. & Harrington 2004. Currently, there is little known about the predatory pressures on H. sanguineus on cobble beaches.

58
In order to better understand how Hemigrapsus sanguineus impacts intertidal and subtidal ecosystems we used a combination of laboratory mesocosm and field experiments to assess: 1) H. sanguineus impacts on Littorina littorea (via lethal or nonlethal effects) and macroalgae, and 2) predation pressure on H. sanguineus from intertidal predators. Our findings yield insight into how invasive species may alter trophic cascades, and how the replacement of one invader by another may further change these dynamics.

L. littorea algal consumption rates
To investigate the lethal and nonlethal effects of Hemigrapsus sanguineus on the algal consumption rate of Littorina littorea, we conducted outdoor mesocosm experiments from June through October 2010 at the University of Rhode Island's Narragansett Bay Campus in Narragansett, Rhode Island. Mesocosm tanks were supplied with ambient temperature, free flowing filtered seawater from Narragansett Bay. We collected L. littorea ( > 4mm shell height), H. sanguineus ( > 10mm carapace width), and Ulva rigida (C. Agardh) in Narragansett Bay; the identity of U. rigida was confirmed via microscopic analysis in the laboratory with molecular voucher specimens (Hofmann et al. 2010, M. Guidone et al. unpubl. data). Ulva lactuca is the algal species typically included in the New England food web based on morphological characteristics (Lubchenco 1978), but recent molecular assessments have determined that U. rigida is the most common Ulva species from our field sites (Hofmann et al. 2010, M. Guidone et al. unpubl. data).
We used mesh cages (30cm x 30cm x 28cm, mesh size 4mm) placed singly inside mesocosm tanks (60cm in diameter). We installed a 10cm tall (2cm diameter) standpipe in the center of each tank and covered all standpipes with 1mm window screening to prevent the escape of mobile organisms. We placed eight textured PVC tiles (10cm x 10cm) at the bottom of each cage and randomly designated four to have one 2.0-3.0g piece of Ulva rigida attached to its center, to simulate algae attached to a hard substrate in the intertidal zone. The remaining four tiles had no algae attached. We then randomly assigned each cage (tank) to one of five treatments: 1) No invertebrates, as a control, 2) For each tank, we recorded the initial wet mass (g) of each Ulva rigida piece. All algae were spun 20 times in a salad spinner to remove excess water prior to weighing.
Each trial ran for four days. Because we only had 10 tanks, we ran six trials during the summer, with two tanks per treatment per trial, using new organisms each time. Trials were ran in sets of two and, after confirming there was no difference between sets, were combined into one treatment with n = 4. The treatments were randomized among tanks over time, and the tanks were scrubbed and allowed to air dry for at least 48 hours between trials. At the end of each trial, we measured the final mass (g) of each algal piece; all algae were again spun prior to weighing. We analyzed the change in mass of U. rigida with a two-way ANOVA and a post-hoc Tukey analysis (JMP v7.0,www.sas.com) to assess differences in change of algal mass among treatments and time. Data met assumptions for normality and homogeneity of variances.

L. littorea feeding behavior
In the same cages as above and concurrent with the previous study, we monitored the behavior of Littorina littorea every six hours, beginning at 12:00 AM, for the first 24 hours of each trial. We recorded the number of L. littorea that were: 1) on the Ulva rigida, consuming it; 2) on the bottom of the tank, neither feeding nor fleeing; and 3) on the sides of the mesh cage, exhibiting a fleeing response by vertically moving out of the reach of the Hemigrapsus sanguineus. We analyzed our results using contingency analyses (JMP v7.0, www.sas.com) to assess differences in snail behavior among treatments and across time of day. Results are expressed as percentages to more clearly show comparisons.

Predation rates on H. sanguineus
To determine the predation pressure on Hemigrapsus sanguineus, we conducted a randomized tethering experiment at Bear Point, Prudence Island, in the Narragansett Bay National Estuarine Research Reserve (NBNERR; 41° 39.631ʹ′ N 71° 20.527ʹ′ W). Bear Point has an intertidal zone with a tidal amplitude of approximately 1.2 m, has low wave disturbance, and the substrate is dominated by cobbles less than 50cm in diameter.
We constructed predator exclusion cages (25cm x 25cm x 18cm) from 1.3 cm PVC pipe covered with 4mm mesh. Full exclusion cages were covered with mesh on all sides except the bottom; benthic predator exclusion cages were open on the bottom and top, to allow access by pelagic predators; control cages had no mesh covering over the PVC frame to allow access by both benthic and pelagic predators. We conducted experimental trials at three tidal heights: high intertidal, low intertidal, and shallow subtidal, with approximately five meters between tidal heights. However, due to unavoidable complications in the shallow subtidal (i.e. sharp rocks and barnacles that compromised crab tethers), data from this tidal height were excluded from our analyses.
At each tidal height, we secured nine tethered Hemigrapsus sanguineus to the cobble substrate. We randomly assigned each crab to one of three treatments: full exclusion cages, benthic predator exclusion cages, and control 'cages' (one crab per cage). Each cage was approximately 1m from the next. This was repeated six times during the summer of 2010.
Tethers consisted of a 6lb monofilament harness tied around each Hemigrapsus sanguineus between the claws and first walking appendages and secured to the carapace with marine epoxy (Eclectic Products, Marine Goop Adhesive). Each tether was 25cm in length and attached to one side of a 4lb stainless steel double swivel; the other side of the swivel was placed over an 18cm metal stake buried a minimum of 10cm in the substrate.
H. sanguineus were able to move freely throughout the radius of their tether and were able to conceal themselves beneath cobble. Once crabs were tethered, cages were placed on top of them; the bottom edges of all exclusion cages were then buried in the cobble to prevent organisms from burrowing underneath the mesh and/or frame. After three days, we removed the cages and recorded all crabs as present, absent, or desiccated (deceased).
Results were analyzed contingency analyses (JMP v7.0, www.sas.com) to assess differences among cage types, tidal heights, and time. Results are expressed as percentages to more clearly show comparisons.

L. littorea algal feeding rates
Littorina littorea grazing rates were not reduced by the presence of, or chemical cues from, Hemigrapsus sanguineus. Ulva rigida in the presence of only L. littorea, or with both L. littorea and H. sanguineus' chemical cue, exhibited a four-fold decrease in algal growth compared to the control (F 4,90 = 40.41, p < 0.0001; Figure 1). The greatest decrease in algal mass was observed when both consumers were present, with over a sixfold decrease in algal mass versus the control. However, H. sanguineus alone also had a negative impact on U. rigida biomass (Figure 1). There was also a significant difference in the change in algal mass among trials (F 7,90 = 19.63, p < 0.0001) with a significant interaction term (F 28,90 = 2.02, p = 0.0069). During the entire experiment, less than 2% of L. littorea were unaccounted, with no evidence of snail consumption by a predator (i.e. broken shells).

L. littorea feeding behavior
The behavior of Littorina littorea in the mesocosm tanks varied significantly among Hemigrapsus sanguineus treatments (presence, absence, or chemical cue; χ 2 4 = 13.51, p = 0.0090) and across daily cycles (χ 2 6 = 60.87, p < 0.0001; Figure 2 littorea behaviors at 6:00 A.M. and 6:00 P.M. were very similar to each other. The number of L. littorea observed at 6:00 PM was lower than at the other times of day due to environmental conditions that prevented us from sampling.

Predation rates on H. sanguineus
There was a significant difference in the distribution of alive, removed, and desiccated Hemigrapsus sanguineus between tidal heights (χ 2 2 = 14.47, p = 0.0007) but not among cage types ((χ 2 4 = 8.75, p = 0.0676), with no interaction effect (χ 2 4 = 0.39, p = 0.9836; Figure 3). In the field, the vertical distribution of Hemigrapsus sanguineus was influenced by physical factors in the upper-range and biological factors in its lower range, which is the traditional model of species' distributional limits in intertidal habitats (Connell 1961 Replacements of one invasive species by another have the ability to reshape intertidal marine food webs , Griffen & Byers 2009), which can have biological implications through changes to predator-prey interactions, behavior modification of conspecifics, and temporary release from consumption by novel predators , Steinberg & Epifaunio 2011. As globalization increases and maritime transportation becomes more efficient, species are being introduced to new environments at an increasing rate (Ruiz et al. 1997) and, once they are introduced, their establishment and spread is facilitated by the ever-increasing threat of global climate change (Harley et al. 2006, Hellmann et al. 2008, Walther et al. 2009). While prevention of the introduction of novel species is important to protecting the ecologic and economic viability of our coastal and estuarine ecosystems, it is also important to identify the introduction of new species while there is still the opportunity to prevent their spread, understand the long-term effects of specific invasive species, and prepare for the potential economic impact ).