Interactions of larval dynamics and substrate preference have ecological significance for benthic biodiversity and Ostrea edulis Linnaeus, 1758 in the presence of Crepidula fornicata

Funding information Blue Marine Foundation Abstract 1. Populations of the European flat oyster Ostrea edulis have experienced catastrophic declines across Europe and subsequent spread of the non-native species Crepidula fornicata has led to its occurrence in exceptionally high densities in some areas previously dominated by O. edulis. 2. Spatial and temporal concurrence of C. fornicata larvae within the zooplankton community occurs throughout the O. edulis spawning season. A C. fornicata larval peak density of 374.7 ± 96.5 larvae/ml (mean ± SD) was observed in Langston Harbour sympatrically with O. edulis density of 45.7 ± 20.1 larvae/ml in early August. Overall oyster larvae contribution to the zooplankton community was higher in Portsmouth Harbour (12%) than C. fornicata contribution (9.6%), whilst the opposite occurred in Langstone (oysters, 11.7%; C. fornicata, 12%). 3. Larval abundance is not reflected in recruitment on the seabed, owing to the conspecific substrate preference of O. edulis. Settlement of O. edulis spat was significantly greater on settlement discs covered with recently deceased oyster shells; 6.7 ± 1.2 (mean ± SE) spat/disc vs old smooth oyster shells, 2.7 ± 1.3, C. fornicata shell 1.7 ± 0.3, cemented discs 2 ± 1 or the plastic control disc 0.7 ± 0.7. 4. Settlement substrate type matters in the presence of high benthic and larval densities of C. fornicata. The Solent has become a substrate-limited system for O. edulis; substrate management or reef deployment is required to restore a selfrecruiting population. 5. Finally, although C. fornicata may provide functional equivalence in terms of filtering services, it supports a significantly different and less biodiverse faunal community from that of O. edulis. Therefore C. fornicata does not provide equivalence as an ecosystem engineer and mechanisms of ecological phase shift are occurring within areas dominated by this invasive species. Received: 6 October 2019 Revised: 20 June 2020 Accepted: 12 July 2020 DOI: 10.1002/aqc.3446


| The loss of an ecosystem engineer
The term 'ecosystem engineer' is used to describe any organism that directly or indirectly modulates the availability of resources to other species, by causing physical state changes in biotic or abiotic materials. The European flat oyster Ostrea edulis (along with other oyster species), epitomizes the classification of an autogenic engineer, whereby the physical structure provided whilst alive and by the remaining shells when deceased change the environment (Jones, Lawton, & Shackak, 1994). Typically, O. edulis inhabits coastal and estuarine environments, which range from the intertidal down to 80 m depth, within a salinity range of 18-40‰ (Jackson, 2007).

Historically, O. edulis populated extensive areas of seabed in
European waters, equating to over 25,000 km 2 (Olsen, 1883)  Large-scale cultivation and management of the species extend back to the Roman Empire (Günther, 1897) and the continued largescale extraction throughout the industrial revolution is highlighted by the 120,000-strong fleet of oyster dredgers that, in 1864, supplied 700 million oysters to London alone (Philpots, 1890). The 80 million oysters harvested annually in the Bay of Biscay, prior to 1859, were valued at £10,000 (Sullivan, 1870), equivalent to £1.2 million today.
The long-standing impression that the ocean provided an inexhaustible source of fish and shellfish can be seen elsewhere, including the historical shell piles that are estimated to contain 5 × 10 12 shells in France (Gruet & Prigent, 1986 as cited by Goulletquer & Heral, 1997).
The unsustainable extraction resulted in catastrophic declines across all of Europe; the situation is arguably most severe in Germany where O. edulis is now classified as extinct and requires a reintroduction (Pogoda, 2019).
Until recently, the Solent contained one of the largest remaining self-sustaining O. edulis fisheries in Europe, with populations forming dense aggregations, predominantly occurring in the areas around Stanswood and Calshot (Key & Davidson, 1981;Palmer & Firmin, 2011). Between 1979 and1980, 15 million oysters (650-850 tonnes) were landed by 450 vessels and recorded seabed densities were as high as 32/m 2 (Key & Davidson, 1981). This extraction was not sustainable and resulted in the collapse of the fishery as the biological limit of the species was exceeded, in part owing to the removal of the reproductively viable population but also the settlement substratum for their larvae, provided by those mature oysters.
The availability of suitable substrata is key for the completion of the O. edulis life cycle. The veliger larvae display gregarious behaviour, preferentially settling and metamorphosing on conspecifics and other hard, clean substrata that has a high surface heterogeneity (Bayne, 1969;Cole & Knight-Jones, 1939;Cole & Knight-Jones, 1949;Walne, 1964;Walne, 1974). The nature of the settlement surface, biofilm formation and other cues may influence settlement behaviour (Walne, 1974). However, Smyth, Mahon, Roberts, and Kregting (2018) reported the availability of hard substrata rather than its type determined the settlement by O. edulis in Strangford Lough. Other research suggests that cultch (disarticulated shell) is an outcome of a selfrecruiting oyster reef and the presence of live or box shell (dead but not disarticulated cultch) is key to recruitment for some species of oyster (Powell, Hofmann, & Klinck, 2018). The invasive American slipper limpet Crepidula fornicata is also suggested as a suitable substrate for O. edulis within fisheries management (T. Cameron, pers. comm., 2019). The large-scale extraction of O. edulis habitat and associated substrate remains a serious concern for the recruitment and survival of this species.

| Ecosystem functions and services of native oyster reefs
Ostrea edulis provides benefits to commercial fisheries, and provides an important ecological role in providing habitat for other organisms (Korringa, 1946;Mistakidis, 1951). Facilitation of increased species diversity and abundance is one of the major and most relevant functions native oysters provide. Korringa (1946) and Mistakidis (1951) conducted studies to detail the associated epibiota. They found numerous species regularly inhabiting shells of O. edulis, considered as characteristic epifauna of the native oyster. The three-dimensional structures created by years of successive settlement of oyster larvae on adult shells provide structural complexity in systems dominated by soft, flatbottom habitats (Bartol, Mann, & Luckenbach, 1999;Micheli & Peterson, 1999). Mobile fish and decapod crustacean species utilize oyster reefs for numerous reasons, consuming the oysters or their associated epibiont community, using oyster shells as surfaces for spawning and finding refuge from predation within the oyster reef (Tolley & Volety, 2005), whereas sessile species use the reefs for settlement and attachment (Boudreaux, Stiner, & Walters, 2006). Fish produced on oyster reefs have significant economic value to coastal communities (Grabowski & Peterson, 2007). The lost habitats caused by decline in oyster reefs have a negative economic impact as they are linked to decreases in overall coastal and shelf sea biodiversity (Airoldi, Balata, & Beck, 2008;Lotze et al., 2006). Although there is an increasing acknowledgement that oyster reefs provide multiple ecosystem services, management objectives beyond harvest are not yet widespread (Beck et al., 2011). Many European oyster restoration projects go beyond biodiversity conservation as their focus; the Native Oyster Network, UK, and Ireland (2020) and European Native Oyster Restoration Alliance (2020) are jointly creating monitoring guidelines that include metrics that quantify ecosystem functions and services.
Oyster reef habitat provides a wide range of ecosystem services including water filtration, food, shoreline stabilization, coastal defence and fisheries (Grabowski & Peterson, 2007;Newell, Fisher, Holyoke, & Cornwell, 2005;NRC, 2010). As filter-feeders, particulate matter resuspended by tidal currents and storms is an important food source to O. edulis (Grant, Enright, & Griswold, 1990). By removing suspended solids from the water, the oysters increase water clarity. Although difficult to quantify in large bodies of water, localized effects of filtration, such as reduced turbidity, have been observed (Coen et al., 2007;Grabowski & Peterson, 2007). Indeed, oysters are able to reduce the volume of suspended solids and phytoplankton (Grizzle, Greene, Luckenbach, & Coen, 2006;Nelson, Leonard, Posey, Alphin, & Mallin, 2004). Healthy oyster reefs can therefore reduce the likelihood of harmful algal blooms occurring and prevent the negative economic and ecological impacts associated with harmful algal blooms, especially at the local scale (Cerrato, Caron, Lonsdale, Rose, & Schaffner, 2004;Newell & Koch, 2004). The improvement to water quality can increase recreational activities such as sport fisheries and tourism to the area (Lipton, 2004). Shellfish are also associated with nutrient remediation in coastal bays via denitrification in surrounding sediments (Newell et al., 2005). The nutrient remediation potential of oysters could translate into a high economic value (Watson, Preston, Beaumont, & Watson, 2020) since nutrient removal and achieving nitrate neutrality is a high priority for coastal stakeholders, including public bodies, housing developers and policy makers (Natural England, 2020).
Oyster reefs serve as natural coastal defences absorbing wave energy thus reducing erosion caused by boat waves, sea-level rise and storms (Meyer, Townsend, & Thayer, 1997;Piazza, Banks, & La Peyre, 2005). Currently ecosystem services provided by O. edulis are yet to be quantified. The potential services of a healthy oyster reef are widely understood from the quantification of ecosystem services of other oyster species. Quantifying services and functions of O. edulis reefs will be a key step in shifting the focus of management objectives.

| Ecological invasion by the American slipper limpet
Non-native marine species are of special concern when they become invasive and displace native species. Negative impacts include biotic homogenization, modification of habitats and alteration of community structures and ecosystem functions (Bax, Williamson, Aguero, Gonzalez, & Geeves, 2003;Katsanevakis et al., 2014;Viard, David, & Darling, 2016). When these impacts impede the provision of ecosystem services it can detrimentally affect human health and cause substantial economic losses (Grosholz, 2002;Perrings, 2002;Wallentinus & Nyberg, 2007).

and the Pacific oyster
Crassostrea gigas (Blanchard, 1997). First appearing in Liverpool during the 1880s (Moore, 1880in McMillan, 1938 and the east coast and Thames estuary in the 1890s (Cole, 1915;Crouch, 1893), C. fornicata is now a well-established invasive non-native species. The loss of oyster habitat has further exacerbated the spread and the abundance of C. fornicata and is a major concern across Europe ( Blanchard, 1997;Boyle, 1981), particularly in the Solent . In rare instances C. fornicata 'stimulates zoobenthic community diversity and abundance' in muddy sediments (de Montaudouin & Sauriau, 1999).
In contrast to the wide range of ecological benefits provided by O. edulis, C. fornicata has been shown to be detrimental to habitat suitability for juvenile fish (Le Pape, Guérault, & Désaunay, 2004;Le Pape et al., 2007) and suprabenthic biodiversity (Vallet, Dauvin, Hamon, & Dupuy, 2001). The shell growth and survival of other commercially important bivalves, such as Mytilus edulis (Thieltges, 2005), are also impacted. Habitat modification in the presence of C. fornicata is also an issue in many areas. This occurs through the production of mucoidal pseudofaeces, which converts predominantly sandy substrata into mud-dominated substrata with a high organic content that rapidly becomes anoxic and unsuitable for other species (Streftaris & Zenetos, 2006). This includes oysters that prefer less silty and muddy waters (Barnes et al., 1973;Bromley, McGonigle, Ashton, & Roberts, 2016;Fulford, Breitburg, & Luckenbach, 2011;Walne, 1979).
Ostrea edulis populations are also negatively impacted through a reduction in suitable substrata available for larval settlement (Blanchard, 1997), hindering recruitment and potentially oyster restoration efforts on the seabed.

| Interspecific competition between O. edulis and C. fornicata
An association of species characterizes benthic fauna in the Solent, with C. fornicata dominating the benthic community in many locations throughout the area, regardless of depth and substratum (Barnes et al., 1973). It is well known that invasive species have detrimental effects on the growth and survival of native species (Thieltges, Strasser, & Reise, 2006), especially if they occupy the same niche.
Owing to C. fornicata's suspension feeding regime and preference for similar habitats to O. edulis, this invasive species can quickly exert a detrimental effect on oyster populations and habitat (de Montaudouin, Audemard, & Labourg, 1999): 'they have a detrimental effect upon oyster culture' (Chipperfield, 1951); 'Crepidula is an oyster-pest' (Korringa, 1951;Walne, 1956). It is essential to understand the ecological interactions between the two species to recognize the negative effects caused by the presence of C. fornicata. This will help restoration efforts, by enabling adaptive management strategies in locations where C. fornicata are present and informing site selection criteria for restoration projects.
Current research investigating competition between C. fornicata and O. edulis is limited, especially at the planktonic larval stage, but the topic receives increasing attention for its ecological consequences. Blanchard, Pechenik, Giudicelli, Connan, and Robert (2008) found that C. fornicata larvae ingested phytoplankton over a larger range of cell sizes and at increased rates compared with C. gigas. This laboratory study was a com-

| Recruitment substrate characterization
Settlement substrata availability for, and preference of, O. edulis and C. fornicata in the eastern Solent harbours (Portsmouth, Langstone and Chichester, Figure 1) were assessed. Settlement substratum was recorded for each individual O. edulis from the three harbours, all of which were purchased from the commissioned fisheries. Settlement substratum of C. fornicata chains was recorded for individuals collected during surveys of the three harbours (see Helmer et al., 2019 for collection methods and locations). The settlement substratum was determined for O. edulis as the organism or material the oyster was attached to near the hinge/umbo. When a clear scar was present but no material remained, it was recorded as 'absent'. The attachment substratum for each individual chain of C. fornicata was recorded as the substratum that the last living individual at the base of the chain was settled upon. Chains were considered separate when the substratum had multiple chains attached to it and live individuals did not interconnect these chains.

| Sample collection and preservation
Seawater samples were collected using a plankton net (300 mm diameter, 64 μm mesh, NHBS). Surface tows were conducted at high tide ±1 h at a speed of 1.5 kn for 1 min, with three replicate samples collected at each location. Using this method, a volume of 3.27 m 3 of seawater was filtered by the plankton net during each tow. Plankton sampling was carried out at two locations (Langstone and Portsmouth) at approximately weekly intervals throughout the spawning season (May to August 2016; Table 1). Immediately after collection samples were filtered across a 64 μm sieve and fixed in 4% formalin in seawater (borax-buffered 5 g/L), stained with Rose Bengal (0.05 g/L), then preserved in 70% ethanol after 1 week (Goswami, 2004). Once in ethanol, samples were split into two sub-samples to be used for larval quantification and scanning electron microscopy (SEM).
This procedure was replicated in triplicate for each sampling F I G U R E 2 Settlement plates deployed in 2016 comprising (a) blank plastic discs, (b) plastic discs dipped in cement, (c) plastic discs covered in old, smooth Ostrea edulis valves, (d) plastic discs covered in recently deceased O. edulis valves and (e) plastic discs covered in Crepidula fornicata shells. (f) Each structure contained three replicates of each substratum placed in random order. Photos: Luke Helmer. Schematic of disc deployment provided on the right T A B L E 1 Sample collection dates and labels for both Portsmouth and Langstone harbours period/location. The values of larval abundance were then averaged and used to calculate the larval density (larvae/ml).

| SEM of oyster larvae
Larval analysis by SEM was used to confirm the light microscope identification of oyster larvae and monitor O. edulis larval survival and growth in the column water. Larval measurements were used to calculate the percentage frequency of each shell size class in both locations. The two species, O. edulis and C. gigas, were distinguished using morphological features clearly visible from the micrographs.
Five to 10 oyster larvae were selected from each sample and placed in sodium hypochlorite (5%) for 48 h to disarticulate the two valves of each individual (Rees, 1950). Samples were then dehydrated through a series of increasing ethanol concentrations (50, 60, 70, 80, 90 and 100%), followed by submersion in hexamethyldisilazane (100% HMDS solution).
Samples were mounted on 12 mm SEM stubs, which were then coated in gold/palladium (Leica EM ACE600; Turner & Boyle, 1975).
Electron micrographs of the larval shells were obtained using a Zeiss Evo MA10 SEM. Identification was confirmed morphologically using the features of the D veliger, umbo and left valve hinge, according to Hu et al. (1993) and Waller (1981). The maximum shell length and height (μm) of each larva were calculated using ImageJ software ( Figure 4). The percentage frequency of each shell size was also calculated for each location and each time point.

| Associated epibiont diversity and abundance
The epibiont succession and community assemblage supported by

| Associated epibiont diversity and abundance
The univariate and non-parametric multivariate techniques using ordination from PCO with S17 Bray Curtis similarity matrices contained in   Figure 5b).
F I G U R E 5 (a) Proportion of O. edulis (n/harbour = 700) retrieved from the fisheries within Portsmouth, Langstone and Chichester harbours, as well as the total (n = 2,100), settled to C. fornicata shell, oyster shell or with no obvious attachment point observed.
(b) Proportion of attachment substrata for live C. fornicata chains observed within Portsmouth (n = 221), Langstone (n = 127) and Chichester (n = 584) harbours, as well as the total for all locations (n = 932). Each chain of C. fornicata was defined as the individuals attached to one another in a single mass, irrespective of the direction of attachment and excluding any deceased shells used as attachment substratum. Chains were considered separate when the substratum had multiple chains attached to it and these chains were not interconnected by live individuals

| Planktonic larval densities
Since the percentage contribution of C. gigas to the total abundance of oyster larvae was very low in both sites (<5.5%), the overall oyster densities are referred to as O. edulis larval densities.
In Langstone Harbour C. fornicata larvae dominated, contributing the highest density at 374.7 ± 96.5 (mean ±SD) larvae/ml, whilst O. edulis was lowest at 1 ± 0 larvae/ml (Figure 6a). In Portsmouth Harbour, O. edulis occurred at the highest density at 67.7 ± 29.3 larvae/ml and C. fornicata was lowest at 6 ± 4.6 larvae/ml (Figure 6b). In Langstone, during the entire spawning season (June to August 2016), O. edulis larval density varied between 8 ± 1.7 and 92.3 ± 12.9 larvae/ml. Two possible spawning events are suggested by peaks in O. edulis, on 28 July (92.3 ± 12.9 larvae/ml) and 12 August (45.7 ± 20.1 larvae/ml). Crepidula fornicata larval density ranged between 6 ± 3 and 374.7 ± 96.5 larvae/ml. This last value indicates that a massive spawning event took place around 12 August. The 65.3 ± 14.2 (mean ±SD) larvae/ml observed on 19 August could be a second event, or more likely larvae still present in the water column from a previous spawning event.
In Portsmouth, between the end of June and the end of August, O. edulis larval density ranged between 6.7 ± 3.5 and 67.7 ± 29.3 (mean ±SD) larvae/ml. A first spawning event on 30 June and a possible second one on 12 August corresponded to peaks of 67.7 ± 29.3 and 38.7 ± 26.8 larvae/ml respectively. Crepidula fornicata larval density was lower in Portsmouth than in Langstone, ranging between 6 ± 4.6 and 44 ± 18.1 larvae/ml, respectively. The only peak in larval density corresponding to a probable spawning event was found in Portsmouth on 12 August, with 44 ± 18.1 larvae/ml.
From the two-way ANOVA performed on O. edulis larval density (variable V 1 ), no significant differences were found between the two sites, except on the 30 June and 28 July, with significantly higher densities in Portsmouth and Langstone, respectively (Figure 7a). The difference of larval density was significant between dates (F 5, 35 = 13.6, P ≤ 0.001) and for the combination of factors (site × date) (F 5,35 = 7.8, P ≤ 0.001), with levels 6 and 7 of the factor 'Date' (28 July and 4 August) mostly responsible for this significant difference (post-hoc Tukey's pairwise test, P ≤ 0.05). Larval density of C. fornicata (variable V 2 ) did not vary significantly between sites and dates, except on 12 August (Figure 7), when a massive spawning event occurred, par-  Differences between data labels indicate significant differences (two-way ANOVA, P < 0.05) F 5,24 = 6.9, P ≤ 0.001) and for the combination of factors were found (pseudo-F 5,24 = 2.55, P ≤ 0.05). No significant differences were found either between sites within each level of factor 'date' and between dates within both levels of factor 'site' (post-hoc pair-wise test).
The average similarity of community composition was >80% within each level of both factors (SIMPER analysis). This is mainly due to the presence of Copepoda (larvae and adults), since their contribution to the similarity of each level ranged between 40 and 50%. The dissimilarity between either dates and sites was no higher than 27%.
The PCO analysis, explaining 72.2% of the total variation between dates in Langstone Harbour, showed a separation of the planktonic communities sampled on 31 May and 12 August from the rest of the samples (Figure 8b solid circles). Significant differences in community composition were found between dates (F 9,20 = 11.08, P ≤ 0.001) in Langstone (PERMANOVA main test performed with one factor). Nonetheless the post-hoc pair-wise test did not produce any significant difference between each level of factor 'date'.
The average dissimilarity was <25% between most of the dates. It The cages were grouped together to increase the replicates ( Table 2).
Neither of the sites yielded an even population of rank abundance.
Both sites were dominated by a few species.
Species diversity was significantly different between location T A B L E 2 Mean number of species and total abundance found associated with Ostrea edulis and Crepidula fornicata, showing the results of the high-and low-density conditions and then an average as there was found to be no significant difference between the two densities

Mean total abundance
High-density oysters 10.1 42.9 Low-density oysters 8.7 30.9 High-density limpets 5.4 17.3 Low-density limpets 4.9 15.4 All oysters 9.4 37.4 All limpets 5.2 16.5 F I G U R E 9 (a) Species diversity and abundance associated with O. edulis and C. fornicata. Data labels indicate significant differences between diversity and abundance associated with the two species (P < 0.05). (b) Principal component analysis illustrating the distribution of species abundance associated with O. edulis and C. fornicata populations in Langstone Harbour, and the faunal species that best characterize the respective communities species contributed to 70% of the faunal community on O. edulis, with P. serratus, P. triqueter, A. scabra and T. indivisa contributing 31.2, 16.4, 13.7 and 9.4%, respectively (SIMPER). Two species contributed to 70% of the faunal community associated with C. fornicata, with P. serratus and Spirorbis spirorbis (contributing 53.3 and 15.6%, respectively).
A PCO explaining 46.1% of the variation in Langstone Harbour revealed a significant difference between the faunal community associated with O. edulis and C. fornicata (PERMANOVA main test F 1, 28 = 5, P ≤ 0.001), also corroborated by an ANOSIM test (R = 0.47, P ≤ 0.001; Figure 9b). Twenty-five species characterized the faunal community associated with O. edulis, and eight species characterized the faunal communities associated with C. fornicata (DIVERSE Test). Settlement of O. edulis spat was significantly greater on settlement discs covered with recently deceased oyster shells, with 6.7 ± 1.2 (mean ± SE) spat/disc, more than double the number of spat associated with old smooth oyster shells, 2.7 ± 1.3. No significant difference in the number of settled spat was found between old smooth oyster shells, C. fornicata shell (1.7 ± 0.3), cemented discs (2 ± 1) or the plastic control disc (0.7 ± 0.7; Figure 10; one-way ANOVA, F 4, 10 = 5.6, P ≤ 0.05). which led to concern about mechanisms of competitive ecological exclusion of O. edulis by the invasive C. fornicata.

| DISCUSSION
In the absence of plentiful live oyster substrate (1-8% were found associated with conspecific shells), a relatively low percentage  shells are rough and scaly in appearance (Perry & Jackson, 2017), whereas C. fornicata shells are much smoother (Rayment, 2008). Differences in CaCO 3 mineral composition may also explain this settlement preference by O. edulis larvae; C. fornicata shells are predominantly aragonitic (Pilkey & Goodell, 1964 (Walne, 1974). This study disputes the finding that settlement is determined by the availability of hard substrata alone (Smyth et al., 2018) resulted in delayed metamorphosis, during which the velum degenerated and the foot grew larger, there was also a decline in feeding rate and eventually the larvae were no longer able to feed (Bayne, 1965). In the polychaete Hydroides elegans, metamorphosis cannot be delayed without measurable negative effects on juvenile survival and growth (Qian & Pechenik, 1998). Echinometra larvae that experienced a prolonged delay in metamorphosis also had a reduced chance of survival, metamorphosis success and survival to juvenile stage (Rahman, Boon, Muntohar, Tanim, & Pakrashi, 2014). There is currently no evidence that delayed metamorphosis in O. edulis has these adverse effects; however, it is likely that there will be negative effects as observed in other species.
Both O. edulis and C. fornicata are filter feeding molluscs that potentially offer functional equivalence in their nutrient assimilation or water filtration services. They do not, however, provide ecological equivalence in terms of the ecological niche and suprabenthic communities they support. De  found that the presence of C. fornicata had no effect on the benthic community; however, this study demonstrates that the presence of C. fornicata has a significant negative effect on the epibiont biodiversity. Specifically, the biodiversity decreased in the presence of C. fornicata. As well as supporting a lower total abundance of species, in relation to O. edulis, C. fornicata also supported a significantly different community. It is now widely accepted that oyster shells show higher diversity than non-living hard substrata, and as oysters grow older and therefore larger, epibiotic diversity will increase (Smyth & Roberts, 2010). However, this study is one of the first (at least in recent years) to show that O. edulis substrate supports higher levels of biodiversity than C. fornicata.
As well as an increase in biodiversity, O. edulis also provides an increase in overall biomass, which in turn improves the health and quality of an ecosystem. Although increases in biomass and biodiversity themselves do not necessarily make an ecosystem more resilient to change, they are driving factors. The three main factors required to facilitate ecosystem resilience are diversity, connectivity within the ecosystem and adaptive capacity (Bernhardt & Leslie, 2013). Therefore, an increase in trophic complexity associated with O. edulis, compared with C. fornicata, will also increase the resilience and health of an ecosystem.
Non-native invasive species are a threat to the conservation of biodiversity and can negatively impact ecosystem services, with both ecological and economic impacts (Katsanevakis et al., 2014). Phaseshifts caused by the introduction of invasive species are becoming increasingly common, for example the introduction of Arcuatula senhousia (Asian date mussel) to San Diego, USA, changed the entire community composition (Grosholz, 2002;Lambert, Levin, & Berman, 1992). The mats of byssal threads produced by the mussel created a unique habitat that was not present in the otherwise largely unstructured mudflats, which as a result encouraged the development of a new community assemblage (Crooks, 1998;Crooks & Khim, 1999). Crepidula fornicata is a threat to native habitats and species; as a habitat engineer it has been reported to cause substantial large-scale changes in the recipient ecosystems, which could lead to phase shifts. These include modification of the trophic structure, changes in phytoplankton composition, enhanced siltation owing to accumulation of faeces and pseudofaeces, and changes in benthic sediments and near-bottom currents (Thieltges et al., 2006). This study demonstrates that the species assemblage of the community associated with C. fornicata was significantly different from the community associated with the native keystone species O. edulis, causing a shift in the coastal benthic biodiversity and ecosystem structure.