Journal of Experimental Marine Biology and Ecology 448 (2013) 289–297

Contents lists available at ScienceDirect

Journal of Experimental Marine Biology and Ecology

j ourna l homepage: www.e lsev ie r .com/ locate / jembe

Larval microhabitat associations of the non-native gastropod Crepidulafornicata and effects on recruitment success in the intertidal zone

Katrin Bohn ⁎, Christopher A. Richardson, Stuart R. JenkinsSchool of Ocean Sciences, Bangor University, LL59 5AB Anglesey, Wales, United Kingdom

⁎ Corresponding author at: Ocean and Earth Science,Southampton, Waterfront Campus, University of SouthamUnited Kingdom. Tel.: +44 (0) 2380 596599.

E-mail addresses: Katrin.Bohn@soton.ac.uk (K. Bohn),(C.A. Richardson), S.Jenkins@bangor.ac.uk (S.R. Jenkins).

0022-0981/$ – see front matter © 2013 Elsevier B.V. All rhttp://dx.doi.org/10.1016/j.jembe.2013.07.020

a b s t r a c t

a r t i c l e i n f o

Article history:Received 25 February 2013Received in revised form 29 July 2013Accepted 31 July 2013Available online 24 August 2013

Keywords:Crepidula fornicataHabitat associationLarval settlementMicrohabitatPost-settlement mortalityRecruitment

Habitat-specific distributions of marine benthic invertebrates can be caused by several processes acting prior to,during or after settlement, including differential settlement and varying levels ofmortality betweenhabitat typesfollowing adaptation of the benthic mode. The non-native gastropod Crepidula fornicata is known for its gre-garious settlement patterns, yet associations with other shellfish species are also common. In the presentstudy, a series of no-choice and choice laboratory assays were undertaken in which larvae were offered dif-ferent settlement substrata, separately and simultaneously, to investigate whether differential settlementof C. fornicata larvae occurs in favour of specific microhabitat types. A field experiment was also conductedto test if recruitment success in the intertidal differed between microhabitat types, by comparing densitiesof young (b2 weeks) and older (b8 weeks) settlers. The laboratory studies indicated that settlement occursin larger numbers in association with certain habitats. However, settlement in association with specific micro-habitat types was not observed in the intertidal. Instead, the distribution of C. fornicata recruits is establishedafter settlement, as the distribution of older recruits, but not younger ones, differed between microhabitattypes. Our findings show that the availability of certain complex structures in the intertidal zone is highly impor-tant in determining survival success of C. fornicata, due to varying levels of post-settlement mortality.

© 2013 Elsevier B.V. All rights reserved.

1. Introduction

The associations of species with specific habitat types can be theconsequence of a variety of settlement and post-settlement processes.For example, active selective settlement behaviour of larvae or passiveprocesses such as varying levels of ‘accessibility’ of the settlement sur-face may occur in favour of certain habitat types (Hale et al., 2008;Olabarria et al., 2002; Tait and Hovel, 2012; Webley et al., 2009).Gregarious settlement in close proximity to conspecifics or associativesettlement with other species, including microbial biofilms, may resultin non-random distributions (Jenkins, 2005; Kent et al., 2003; McGeeand Targett, 1989; Pawlik, 1992). An extreme example of this is the al-most total absence of barnacle settlement in some sheltered rocky shoresites, where despite equivalent levels of larval supply, barnacle settle-ment can be negligible where cues from adults are lacking (Jenkinsand Hawkins, 2003; Jenkins et al., 1999). In contrast, settlement mayalso occur randomly across various habitat types, and in this case,habitat-specific distributions may be established at a later stage due todifferences in the habitats' suitability to support later growth and sur-vival of the settled organism. For example, differences in foraging

National Oceanography Centrepton, Southampton SO14 3ZH,

C.A.Richardson@bangor.ac.uk

ights reserved.

opportunities, food availability or protection from biological and physi-cal stressors may cause highly variable levels of early post-settlementmortality (EPSM) (Gosselin and Chia, 1995a, 1995b; Loher andArmstrong, 2000; Moksnes et al., 1998), which, especially under inter-tidal conditions tends to be high due to the exposure to extreme envi-ronmental conditions (Delany et al., 2003; Gosselin and Qian, 1996,1997; Hunt and Scheibling, 1997; Power et al., 2006). Associative settle-ment occurring with certain microhabitats that may lower levels ofEPSM through for example the provision of refuges or protection fromthese extreme environmental conditions can thus drastically improvethe survival of an individual. The effects of settlement and post-settlement processes on species distributions are therefore unlikely act-ing in isolation, making it highly challenging to distinguish betweentheir relative importance when predicting recruitment dynamics of aspecies in certain habitat types.

The American slipper limpet Crepidula fornicata, a sessile gastropodnative to the North-West Atlantic coast, has now established non-native populations in most European waters (Blanchard, 1997). It ishighly gregarious and juvenile abundance has been found to be posi-tively correlated with adult abundance (Walne, 1956; Hoagland, 1978;McGee and Targett, 1989; but see Bohn et al., 2012, 2013). Settlementof planktonic larvae can be followed by relocation of crawling juvenilesand males up to a few weeks old (Werner, 1955). Whether establishedby the larval or the juvenile stage, permanent settlement often occurson top of conspecifics, resulting in the formation of stacks that are a nec-essary prerequisite for internal fertilisation to occur. Stacks with N10

290 K. Bohn et al. / Journal of Experimental Marine Biology and Ecology 448 (2013) 289–297

individuals are common (pers. obs.). However, C. fornicata is oftenfound attached to other hard substrata (Mineur et al., 2012), includingseveral commercially important shellfish species such as the culturedblue mussel Mytilus edulis (Korringa, 1942; McMillan, 1938; Thieltges,2005). This has greatly aided its first introductions and subsequent sec-ondary spread in its non-native range. The first documented specimensin Europeanwaters were found among the American oyster Crassostreavirginica (Crouch, 1893; McMillan, 1938), and the further transporta-tion ofM. edulis, Crassostrea gigas and C. virginicawith attendant epibontC. fornicata through aquaculture practices has facilitated its spreadwithin its non-native range (Blanchard, 1997; Korringa, 1942; Mineuret al., 2012). Potential resulting direct and indirect negative effects onits basibionts are well documented (Blanchard et al., 2008;Decottignies et al., 2007a, 2007b; Riera et al., 2002; Thieltges, 2005;Thouzeau et al., 2000).

Settlement of C. fornicata occurs as a direct consequence of meta-morphosis following the loss of the swimming organ, the velum(Werner, 1955). Several chemical, physical and/or biological cuesmay induce metamorphosis in well developed, late-stage competentC. fornicata larvae, including the presence of juvenile C. fornicata andfragments of Mercenaria mercenaria shells (Pechenik, 1980), adult-conditioned seawater (McGee and Targett, 1989; Pechenik andHeyman, 1987), the presence of the species Crepidula plana and thehermit crab Pagurus pollicaris occupying shells of Busycon carica(McGee and Targett, 1989), short exposure to heat shocks (Gaudetteet al., 2001), a naturally-produced halogenated compound (Taris et al.,2010) and elevated KCl concentrations (Pechenik and Gee, 1993;Pechenik and Heyman, 1987). Biofilmed surfaces were found to suc-cessfully induce metamorphoses in other species of the same genus(Chiu et al., 2007; McGee and Targett, 1989; Zhao and Qian, 2002)and are most likely also an inducer for C. fornicata larvae (Eyster andPechenik, 1988; Pechenik and Gee, 1993). However, it remains unclearif the abovementioned aggregative distribution of juveniles and adultsamong conspecifics and other basibionts is a result of differential pat-terns established directly at settlement of the larvae due to behaviouralresponses to any one of these cues. Alternatively, the documentedclumped distribution of juveniles may be a result of passive depositionof the larvae following metamorphosis, triggered after contact with asettlement inducing cue. It is also possible that recruitment differs be-tween certain microhabitat types due to post-metamorphic movementor mortality (Crowe and Underwood, 1998). Previous studies showedthat the distribution of C. fornicata is modified some time after settle-ment due to such post-colonisation processes (McGee and Targett,1989; Shenk and Karlson, 1986). Particularly in the intertidal, recruit-ment of C. fornicata is highly variable, possibly due to the exposure ofthe newly settled individuals to sub-optimal conditions causing highlevels of post-settlement mortality (Bohn et al., 2012, 2013). Previouswork has also shown that this might be attenuated whenmicrohabitatsprovide protection from such physical stressors (Bohn et al., 2013).

In the present study, we conducted a series of laboratory experi-ments to investigate if settlement of the larvae of C. fornicata differsbetween two microhabitats, i.e. shells of conspecific C. fornicata andshells of the mollusc species M. edulis that is also often used for at-tachment by C. fornicata. We also determined whether the early dis-tribution of juvenile C. fornicata changes due to post-metamorphicmovement. Lastly, a field experiment was designed to monitor set-tlement rates in association with certain microhabitat types, and ifrecruitment success differs between these.

2. Methods

2.1. Study site

Field experiments and collection of adult C. fornicata broodstock forlarval rearing were undertaken in the low intertidal zone (~1.0–1.3 mabove Chart Datum) at Pennar in the Milford Haven Waterway

(MHW) in SouthWestWales, UK (Fig. 1). The low intertidal of this nat-ural ria is characterised by muddy–gravel banks that are interspersedwith sandy beaches and rocky outcrops (Nelson-Smith, 1965). Marineconditions prevail in the ria due to the low input of fresh water,resulting in the occurrence of a diverse marine flora and fauna(Nelson-Smith, 1967). Tides are semi-diurnal and low water duringspring tides occurs around mid-day and mid-night, causing the regularexposure of the lower shore biota to extreme environmental conditionsduring spring tide emersion. Further details on the physical and biolog-ical environment can be found elsewhere (Nelson-Smith, 1965, 1967).

C. fornicata was first observed in the MHW in the early 1950s (Coleand Baird, 1953), has spread rapidly across most of the ria by the1960s (Crothers, 1966) and now reaches localised densities ofN1000 ind m−2 in the low intertidal and subtidal zone of the middlereaches of the ria (Bohn, 2012). Densities vary greatly at a scale of me-tres and between habitat types. At Pennar, a shore typical for the MHWdue to the composition of the substrata of mud and gravel, a verticalgradient in adult densities of 1031 ± 943, 343 ± 360 and 76 ±125 ind m−2 (mean ± SD) has been observed at tidal heights of ~0.6,1.2 and 1.8 m above C.D., respectively. The reproductive season ofC. fornicata in the MHW extends from early spring to late autumnwith the main period of settlement occurring between the end ofJune/July and September (Bohn et al., 2012).

2.2. Laboratory settlement assays

A series of no-choice and choice assays were undertaken in whichlarvae of C. fornicatawere offered two substratum types (shells of themussel M. edulis, shells of C. fornicata) separately or simultaneouslyto test whether larval settlement occurs in greater numbers on cer-tain substratum types. No-choice assays were done to test the effec-tiveness of experimental treatments in inducing metamorphosis(and thus settlement) and to test for potential artefact effects of exper-imental conditions. For these, larvaewere offered only one of each shelltype (i.e.M. edulis or C. fornicata shell). Choice experimentswere under-taken by exposing the larvae to both shell types simultaneously to as-sess whether differential settlement between various shell types mayoccur.

Olabarria et al. (2002) suggest that equally designed choice and no-choice assays should be undertaken simultaneously to unambiguouslydistinguish behavioural processes (i.e. larval preference of a specific set-tlement site) from passive processes due to differences in accessibility(e.g. the simplicity with which attachment occurs over a given time).If following this design, active selective settlement behaviour of the lar-vae can be established when the proportion of larvae observed on eachsubstratum type differs when larvae are offered a choice compared towhen larvae are offered no choice (Olabarria et al., 2002). This wasnot possible within the scope of this study due to difficulties in rearingsufficient amounts of larvae for the simultaneous undertaking of choiceand no-choice assays. The experiments presented here are thusdesigned to identify patterns in settlement (and the potential forthese early processes in determining the observed juvenile and adultdistribution in the field), but not to identify causative mechanisms forthe observed patterns (i.e. if due to passive or active larval settlement).

2.2.1. Larval rearingLarvae for all experiments were reared in the laboratory. An adult

C. fornicata broodstock was collected at Pennar. Stacks (N6 adultseach) were individually placed in cylindrical tubes (diameter~12 cm) with 250 μm-mesh attached to the bottom to ensure the re-tention of the released larvae and enable the identification of the pa-rental stack of each larval brood. Each adult stack was supplied withconstantly flowing, aerated and 120 μm-filtered seawater at ~15–16 °C and fed a diet of Isochrysis galbana (clone T-ISO). Larvae werecollected within 12 h following release and immediately transferredto 1 μm- and UV-filtered seawater at ~19–21 °C. Larvae were reared

Fig. 1.Map of theMilford HavenWaterway (MHW) in SouthWestWales, UK (top left inset). Field experiments and collection of adult broodstock were undertaken at ~1.0–1.3 m aboveC.D. at Pennar.

291K. Bohn et al. / Journal of Experimental Marine Biology and Ecology 448 (2013) 289–297

in batch cultures at densities of ~0.25–0.5 larvae mL−1, and fed adiet of I. galbana (clone T-ISO, cell density ~1.8 ∗ 105 cells mL−1).Seawater and food were changed every other day. Larvae from a sin-gle batch were released on the same day and were from the samestack, but not necessarily from the same female.

To estimate the development of each batch of larvae, growth ratesand morphological development of 15 randomly chosen larvae weremonitored every other day (data not shown). Once a batch reached anaverage size of N650 μmand showed first signs of morphological differ-entiation to the competent stage (flattened shell geometry and pres-ence of a “brimmed shell”, see Pechenik, 1980; Pechenik and Lima,1984), subsamples of 40 larvae, divided into 4 replicates of 10 larvae,were frequently exposed to an artificial inducer (20 mM K+ enrichedfiltered sea water, see Pechenik and Heyman, 1987) for 4–6 h to estab-lish the percentage of metamorphically competent larvae in the batch.When metamorphic competency of a batch was nearing 50%, a no-choice or choice settlement experiment was started within the next48 h. All larvae were between 14 and 19 days old at the beginning ofthe experiments. Experiments were run under the same conditions asthe larval rearing, but in the absence of food to avoid the presence of apotentially confounding cue. Each larva was only used once.

2.2.2. No-choice assaysMetamorphically competent larvae were exposed either to an ex-

perimental treatment (i.e. various shell types) or a control treatment(i.e. no shell present) for 6 h to compare larval settlement behaviourin the absence of choice. We aimed at standardising and maximisingthe total number of larvae metamorphosing in the experimental treat-ments to improve comparability between groups and allow assessmentof active larval choice. Experimental treatments were therefore run inseawater conditioned by adult C. fornicata stacks (adult conditionedseawater, hereafter ACSW) which is known to be effective in inducingmetamorphosis (McGee and Targett, 1989; Pechenik and Heyman,1987). Seawater was conditioned by putting three stacks (5–7 adultseach) in 2 L of 1 μm- and UV-filtered seawater for 2.5-3 h. ACSW waspassed through a 100 μm filter to remove any debris and faeces. Exper-imental treatments were undertaken in glass dishes (diameter ~7 cm,height ~3.5 cm) filled with 50–70 mL ACSW. Pilot trials showed thatC. fornicata larvae may settle on the glass dish instead of the substrata.To encourage settlement of the larvae on top of the substrata, insteadof the surrounding glass, the bottom of the dish was covered to adepth of ~1 cm with sand, which we considered an unsuitable settle-ment substratum for C. fornicata. The sand was collected from the lowintertidal in the Menai Strait, North Wales, washed and sieved to125–500 μm.

Empty C. fornicata shells (treatment I— Crepidula) or emptyM. edulisshells (treatment II—Mussel) were used as substratum types in the ex-perimental treatments, to offer settlement surfaces identical to thosetypically encountered by the larvae in the field (i.e. treatment I — con-specifics during stack formation, treatment II — a shellfish species thatis abundant in the same habitat and is often colonised). An individualshell was placed in the middle of each dish and gently pushed into thesand until the shell margins were covered. Live C. fornicata or M. eduliswere not used because suspension feeding adults are known to ingestC. fornicata larvae (Pechenik et al., 2004). Only “fresh” and clean shellsof each species were used after removing the animal and scrubbingoff all other epibiota (including potential biofilms) from the shellthe day before the experiments started. Shells were then left in fil-tered seawater over night, in separate containers for each species,to standardise conditions the shells were exposed to prior to the begin-ning of the experiment. All shells were of similar size (~4.5–5.5 cmmaximum length).

Several control treatments were run in parallel to establishmetamorphic competency of the larvae at the time the experimentwas undertaken and to control for potential effects of experimentalconditions on metamorphosis and settlement behaviour of the larvae.Using the same glass dishes, the efficacy of ACSW in providing the cuethat induces metamorphosis was investigated by exposing larvae toACSW alone, not using substrata or sand (treatment III — ACSW). Also,a positive control and a negative control were established by exposinglarvae to 20 mM excess KCl in the absence of any other physical or wa-terborne cues (Pechenik and Gee, 1993; Pechenik and Heyman, 1987)to establish percentage competency of the batch (treatment IV — K+),or filtered seawater only (treatment V — FSW), respectively. To ascer-tain that the sand did not contain cues that might induce metamorpho-sis and attract settlement of the larvae, we established anothertreatment in which the bottom of the glass dish was covered withsand in the same way as in the experimental treatments, but usingFSW and offering no substrata (treatment VI — Sand).

Four replicateswere used for each experimental or control treatment.Fifteen larvae were introduced to each replicate at the beginning of theexperiment by pipetting them evenly across the water surface. Thedishes were left undisturbed for 6 h to allow sufficient time for meta-morphosis (and hence settlement) to occur. After 6 h, the glass disheswere examined under a dissectingmicroscope (×40 or 63). The numberof settled, metamorphosed individuals on each surface type (shell, glassor sand) and the number of free-swimming larvae were noted. Larvaeare known to crawl for long periods of time prior to metamorphosing(Werner, 1955). To ensure that larvae had successfully metamorphosed,we therefore checked for the absence of the swimming organ and the

292 K. Bohn et al. / Journal of Experimental Marine Biology and Ecology 448 (2013) 289–297

presence of gills in addition to lack of swimming behaviour. This exper-iment was run four times.

Data were analysed using one-way ANOVA to test for differencesamong experimental and control treatments following Levene's test toestablish homogeneity of variances. The proportions of larvae settledon top of the shells (instead of the glass dishes and the sand, hereafterreferred to as ‘other substrata’) between treatments I and II were com-pared each time the experiment was run. One-way ANOVA was alsoused on the proportions of the total number of settled larvae, regardlessof whether settlement had occurred on top of the shells or the ‘othersubstrata’. This aimed at testing if cues in the experimental treatments(ACSW, shell types offered) were efficient in inducing metamorphosisand adoption of a benthic mode in competent larvae by comparingproportions with the overall level of metamorphic competency,which was estimated by the percentage of larvae metamorphosingin response to the K+-treatment. Also, this tested for potential arte-fact effects of experimental conditions (sand, glass dishes) on meta-morphosis and settlement behaviour. Data transformations neverachieved homogeneity of variances for the first and the second runof the experiment due to the presence of only zero values in certaintreatments (treatment V — FSW in run 1, treatment III — ACSW inrun 2), so these treatments were removed for statistical analyses.Data from the third and the fourth run were arcsine-square roottransformed to achieve homogeneous variances. Tukey's post-hoctest established which treatments differed.

Table 1No-choice assays. Results of one-way ANOVA to test for the effects of different treatments(Crepidula shell, mussel shell, ACSW, K+, FSW, sand) on the percentage of Crepidulafornicata larvae metamorphosing within 6 h. Treatments with no metamorphosed larvaewere removed for statistical analyses as there were no variances to calculate (FSW inrun 1 and ACSW in run 2). Data from run 3 and run 4 were arcsine-square roottransformed to achieve homogeneous variances. Significant differences are shown in bold.

df MS F p

Run 1 4 0.216 27.514 b0.001Residual 15 0.008Run 2 4 0.293 72.439 b0.001Residual 15 0.004Run 3 5 2400.381 14.513 b0.001Residual 18 165.401Run 4 5 4263.235 29.651 b0.001Residual 18 143.780

2.2.3. Choice assaysShells of C. fornicata andM. edulis, prepared in the sameway as for

the no-choice assays, were simultaneously offered to C. fornicata lar-vae in 8 experimental mesocosms (12.5 cm × 12.5 cm × 6 cm,length × width × height) that were filled with sand (depth ~1 cm)and ~350 mL ACSW. Sand and ASCW were prepared as describedfor the no-choice assays. Two shells from each species were addedto each mesocosm and arranged in a chequered mosaic patternwith gaps N1 cm between shell margins. At the beginning of the ex-periment, 50 larvae were introduced by evenly distributing themacross the whole surface of the mesocosm. Four mesocosms weresampled after 6 h and the remaining four mesocosms after 24 h inorder to determine whether the distribution of early settlers differedto that of older juveniles, as a result of post-metamorphicmovement.As in the no-choice assays, the number of individuals that had settledon each surface type (C. fornicata shells, M. edulis shells, themesocosm wall, sand) was noted as well as the number of free-swimming larvae. Metamorphic competencies of the larval batchesused in these experiments were monitored in parallel by exposing15 larvae to excess KCl for 6 h or 24 h, proceeding in the same wayas in the no-choice assays. This experiment was run twice.

Replicated G-tests of goodness-of-fit were undertaken to comparethe magnitudes of settlement on both substratum types (Lee et al.,2004; Sokal and Rohlf, 1995). We tested the hypothesis that the distri-bution of settlers between Crepidula shells andM. edulis shells is uneven(i.e. deviates from an expected ratio of 1:1). For this, analyses were runseparately for each trial (i.e. for 6 h and 24 h exposure for each of thetwo runs of the experiment). A second hypothesis was tested whichpredicts that the distribution of settlers may change with length of ex-posure to experimental conditions, due to post-metamorphic move-ment. Here, for each run separately, we tested the distribution ofsettlers after 24 h against the distribution after 6 h, instead of using anexpected ratio of 1:1 (even distribution). All replicated G-tests ofgoodness-of-fit were run by undertaking G-tests on the replicates sepa-rately (replicates 1–4), followed by a G-test on all the data from eachreplicated experiment together to test whether the expected ratio isfit by all data (Gtotal), on the pooled data set of each replicated experi-ment (Gpooled) and a heterogeneity test (Ghetero) (Lee et al., 2004;Sokal and Rohlf, 1995).

2.3. Field settlement experiment

A field experiment was undertaken at Pennar (~1.0–1.3 m aboveC.D.) to investigate whether rates of settlement and recruitment variedbetween different microhabitat types under intertidal conditions. Forthis, 2 sets of 16 slate plates (11 cm × 11 cm) were deployed in midJuly 2011 during spring low tides. The surfaces of the plates were ma-nipulated to simulate various naturally occurring microhabitat types.Whilst 2 × 4 plateswere left bare (referred to as treatment ‘Panel’ here-after), the same amount of plates were covered in flat stones (‘Stone’),empty C. fornicata shells (‘Crepidula’) or empty M. edulis shells (‘Mus-sel’). Stones and shells (unbroken and not eroded) used for the experi-ments were collected at the same site and cleaned thoroughly of allepibionts the day before deployment. Wired mesh (mesh size2 cm × 2 cm) was wrapped around the plates with the stones or shellsaligned on top to keep the substrata in position. Plateswere deployed onmetal frames identical to those used in previous studies at the same site(Bohn et al., 2012, 2013), with one panel of each microhabitat type at-tached to each frame. One set of the plates was retrieved and replacedwith new plates every two weeks to measure the densities of young(b2 weeks) C. fornicata settlers on the variousmicrohabitat types (here-after referred to as ‘biweekly’). The other set was sampled only once atthe end of the experiment inmid September to estimate the densities ofolder C. fornicata recruits (b8 weeks, ‘two-monthly’) (n = 4 per sub-strata type and panel deployment duration). At the end of deploymentin the field, plates were collected and transported to the laboratory bysuspending them on horizontal metal stakes that rested in coolingboxes. All C. fornicata present on the panels as well as on the substratawere counted and measured under a dissectingmicroscope with an oc-ular micrometre (×63).

One-way ANOVA was undertaken on the calculated densities foreach sampling event and panel deployment duration separately to testfor potential differences in settlement rates (biweekly plates) and re-cruitment success (seasonal plates) between the four different micro-habitat types. Levene's test was used to check for homogeneity invariances and transformed where appropriate. Tukey post-hoc testingwas performed where significant differences were detected.

3. Results

3.1. Laboratory settlement assays

3.1.1. No-choice assaysComparisons among treatments and controls for the general settle-

ment response (i.e. the total number of larvaemetamorphosing and set-tling, irrespective of settlement site) showed variable patterns over thefour different runs of the no-choice experiment (Table 1, Fig. 2A). Thisoccurred even though metamorphic competency of the larval batcheswas generally high (percentage metamorphosis K+-treatment N60% inall four runs, Fig. 2A), demonstrating that all larval batches used during

0

20

40

60

80

100

120

0

20

40

60

80

100

120

0

20

40

60

80

100

120

0

20

40

60

80

100

120

140

0

20

40

60

80

100

120

140

0

20

40

60

80

100

120

140

0

20

40

60

80

100

120

140

0I) Crepidula shell

I) Crepidula shell

II) Mussel shell

II) Mussel shell

III) ACSW

IV) K+V) FSW

VI) Sand

20

40

60

80

100

120

aa a a

a

aaa

a

a

b

b

acdcd d

ac

b

bc

b

c

bb

p = 0.496

p = 0.018

p = 0.002

p = 0.314

Run 1

Run 2

Run 3

Run 4

Met

amo

rph

ose

d la

rvae

(%

)

Lar

vae

sett

led

on

su

bst

ratu

m (

%)

A B

Fig. 2. Results from no-choice settlement assays where Crepidula fornicata larvae were exposed for 6 h to one of two experimental treatments: I) Crepidula shell, II) mussel shell, or one offour control treatments: III) adult conditioned seawater (ACSW), IV) 20 mM K+ enriched filtered sea water (K+), V) filtered seawater (FSW) or VI) sand. A) Percentage of all larvae thathad metamorphosed, including those that had settled on the sand or glass dishes. Different letters above bars indicate significant differences between treatments. B) Percentage of newlymetamorphosed juveniles that had settled on top of the empty Crepidula shell (treatment I) or the empty mussel shell (treatment II), instead of the sand or glass, of all metamorphosedlarvae. Proportions were significantly different in run 2 and run 3. Mean ± SD. n = 4, with 15 larvae introduced to each mesocosm.

293K. Bohn et al. / Journal of Experimental Marine Biology and Ecology 448 (2013) 289–297

the experiments were in a developmental stage suitable to observesettlement if metamorphosis-inducing cues were present. TreatmentsI and II (Crepidula shell andmussel shell) were generally effective in in-ducing metamorphosis and settlement, although the percentage meta-morphosis in these treatments was relatively low in run 1 and run 2.ACSWwas only effective in inducing metamorphosis in competent lar-vae in large numbers in run 3 when percentage metamorphosis wassimilar to the K+-treatment (Tukey p N 0.05, Fig. 2A). Although thelow response to ACSW indicated that experimental conditions werenot always successful in maximising settlement of competent larvae,the number of larvae metamorphosing in response to treatments Iand II did not differ (Tukey p N 0.05 in all four runs, Fig. 2A). Thisshowed that the cues present in both treatments were similarly effec-tive in inducing metamorphosis, even if this was not due to the provi-sion of ACSW. Only few larvae responded when exposed to FSW(treatment V) and sand (treatment VI), ensuring that the mesocosmconditions (glass jar, sand) did not induce larval metamorphosis andsettlement.

Although similar numbers of larvaemetamorphosedwhen presentedwith either shell type, there was a trend that larger proportions of larvaein the Crepidula treatment settled on top of the shell instead of on thesurrounding substrata compared to in the mussel shell treatment. Thiswas significant in runs 2 and 3 (Table 2, Fig. 2B). In total, 76 larvae settledon Crepidula shells and 37 on mussel shells.

3.1.2. Choice settlement assaysReplicates 1–4 from each trial always fitted the same ratios, resulting

in non-significant heterogeneity tests (Ghetero, Table 3) and thusallowing the pooling of all the data of each experiment (Sokal andRohlf, 1995). There was a general trend that more larvae had settledon the Crepidula shells,with approximately twice asmany settled larvaeobserved on top of the Crepidula shells than on top of the mussel shells(Fig. 3A and B). G-tests on the pooled data (Gpooled) from each trial re-vealed that deviations from an expected even distribution were signifi-cant twice: in run 1 after 24 h and in run 2 after 6 h (Fig. 3A and B,Table 3).

Table 2No-choice assays. Results of one-way ANOVA to test for the attractiveness of Crepidulashells (treatment I) or mussel shells (treatment II) as settlement substrata for Crepidulafornicata larvae. Tests were done on the proportions of the newly metamorphosedjuveniles that were observed on the shells instead of the surrounding substrata (glass orsand) after 6 h. Significant differences are shown in bold.

df MS F p

Run 1 1 0.146 0.538 0.496Residual 5 0.271Run 2 1 0.516 39.450 0.002Residual 5 0.013Run 3 1 0.424 10.415 0.018Residual 6 0.041Run 4 1 0.114 1.210 0.314Residual 6 0.094

0

10

20

30

40

50Crepidula shell

Mussel shell

0

10

20

30

40

50

6h 24h

Pro

po

rtio

n o

f la

rvae

set

tled

on

su

bst

rata

(%

)

A

B

Fig. 3. Proportions of newly metamorphosed Crepidula fornicata juveniles settled on theempty Crepidula shells and the empty mussel shells of the total number of metamor-phosed juveniles in the first run (A) and the second run (B) of the choice settlement as-says. Larvae were simultaneously offered the choice between both substratum types for6 h or 24 h. Metamorphic competency of the larval batches was always N90%. Mean ±SD. n = 4, with 50 larvae introduced to each mesocosm.

294 K. Bohn et al. / Journal of Experimental Marine Biology and Ecology 448 (2013) 289–297

In run 2 after 24 h, similar numbers had settled on themussel shells(36 ± 14% of all settled larvae, mean ± SD) and the Crepidula shells(42 ± 5%), but this had not been the case after 6 h when proportionsdiffered between the substratum types (Fig. 3, Table 3). This indicatesthat here, the newly metamorphosed juveniles made an active choicetowards the mussel shells between 6 h and 24 h. These temporalchanges were marginally significant at p = 0.05 (Table 4). No changesover time were observed during run 1. Total levels of metamorphoses(i.e. including those larvae that were observed on the jar or sand) onlyincreased slightly between hours 6 and 24 in run 1 (mean ± SD after6 h: 59 ± 7%, after 24 h: 68 ± 11%). In run 2, however, only 36 ± 8%of the larvae had successfully metamorphosed after 6 h, but 62 ± 10%had metamorphosed in the containers that were sampled after 24 h.

3.2. Field experiments

Biweekly settlement densities showed no consistent pattern amongthe four substratum types over the four sampling dates and at no pointwas any significant effect of substratum detected (Table 5, Fig. 4). At theend of the experiment, however, densities of juvenile slipper limpetsdiffered between substratum types on the two-monthly panels(Table 5, Fig. 4). Post-hoc testing revealed that densities on the two-monthly bare slate panels were significantly lower compared to thoseon the two-monthly mussel shell panels (Tukey p = 0.05).

4. Discussion

The combined laboratory and field studies of the present studyshowed that although settlement of the American slipper limpetC. fornicata occurs in greater numbers in association with certainhabitat types under laboratory conditions, the intertidal distribution

Table 3Choice assays. ReplicatedG-test of goodness-of-fit to test the hypothesis of differential settlemeseparate runs of the experiment. Comparisons were done against the expected ratio of an evedetails. Significant differences are shown in bold.

Experiment Replicates

1 2

Run 1 — 6 h vs even distribution d.f. 1 1G 0.287 2.039p 0.592 0.153

Run 1 — 24 h vs even distribution d.f. 1 1G 0.699 1.828p 0.403 0.176

Run 2 — 6 h vs even distribution d.f. 1 1G 5.062 2.039p 0.024 0.153

Run 2 — 24 h vs even distribution d.f. 1 1G 0.476 1.007p 0.490 0.316

of juvenile settlers is determined during processes acting after set-tlement. We found that settlement of C. fornicata larvae tended tobe higher on Crepidula shells than on mussel shells in laboratory as-says. Interestingly, however, such associations with specific micro-habitat types are not obvious soon (b2 weeks) after settlementunder intertidal conditions. Post-colonisation processes were foundto re-structure juvenile distribution, as differences in juvenile densi-ties betweenmicrohabitat types were observed for older (b8 weeks)but not younger recruits (b2 weeks). Post-settlement mortality istherefore highly important in shaping the intertidal distribution ofthe species. Our results thus indicate that although associative settle-ment occurs, adult distributional patterns in the intertidal zone maybe established post-settlement.

The results from our laboratory study suggest that processesother than passive fall-out of the larvae determine the distribution

nt of Crepidula fornicata larvae on Crepidula shells andmussel shells after 6 h or 24 h in twon distribution of settlers between substratum types (1:1). See Sokal and Rohlf (1995) for

Gtotal Gpooled Ghetero

3 4

1 1 4 1 30.335 0.077 2.737 2.136 0.6010.563 0.781 0.603 0.144 0.8961 1 4 1 34.717 5.017 12.260 11.181 1.0790.030 0.025 0.016 b0.001 0.7821 1 4 1 30.335 0.077 7.513 4.478 3.0350.563 0.781 0.111 0.034 0.3861 1 4 1 30.430 2.165 4.077 0.728 3.3490.512 0.141 0.396 0.393 0.341

0.0

0.5

1.0

1.5

2.0

2.5

3.0

30-Jul

0.0

0.5

1.0

1.5

2.0

2.5

3.0

13-Aug

(103

juve

nile

s m

-2)

Table 4Choice assays. ReplicatedG-test of goodness-of-fit to test the hypothesis of temporal changes in the distribution of newlymetamorphosed Crepidula fornicata larvae between 6 h and 24 hof exposure to Crepidula shells andmussel shells in two separate runs of the experiment. The observed distribution of settlers between both substrata types after 24 hwas tested against anexpected ratio that was derived from the observed distribution after 6 h (Crepidula:mussel = 34:23 in run 1, 33:18 in run 2). See Sokal and Rohlf (1995) for details. Significant differencesare shown in bold.

Experiment Replicates Gtotal Gpooled Ghetero

1 2 3 4

Run 1 — 6 h vs. 24 h d.f. 1 1 1 1 4 1 3G 0.019 0.242 1.643 2.144 4.048 2.970 1.078p 0.889 0.623 0.200 0.143 0.400 0.085 0.782

Run 2 — 6 h vs. 24 h d.f. 1 1 1 1 4 1 3G 0.376 6.299 0.512 0.003 7.190 3.841 3.349p 0.540 0.012 0.474 0.959 0.126 0.050 0.341

295K. Bohn et al. / Journal of Experimental Marine Biology and Ecology 448 (2013) 289–297

of newly-metamorphosed C. fornicata juveniles. Passive fall-outshould have resulted in a fully random distribution of the newly set-tled juveniles across all surface types (i.e. the shell, the sand and thejar) in both the Crepidula shell and the mussel shell treatments of theno-choice assays. However, only b3.5% of the settled individualswere ever observed on the sand in the experimental treatments ineach run of the no-choice assays, suggesting that settlement didnot occur randomly. The observed differential spatial settlement pat-terns could occur as a result of varying concentrations or differentlevels of effectiveness of waterborne cues released by conspecificsor other species. Biofilms associated with shells may attract settle-ment and stimulate metamorphosis and this has been reported forseveral species of the genus Crepidula (Chiu et al., 2007; McGee andTargett, 1989; Zhao and Qian, 2002). We cleaned all shells thorough-ly to reduce any such potentially confounding effects of differentbiofilms associated with both shell types. Experiments were carriedout in the absence of adults due to the problem of ingestion of larvaeby adults (Pechenik et al., 2004). ACSW was added to experimentaltreatments instead, to increase and standardise levels of metamor-phoses. The response of the larvae to ACSWwas mostly low, possiblyas the duration for conditioning the seawater was relatively short.Nevertheless, overall levels of metamorphoses were equal betweenCrepidula and mussel shells, but proportions of larvae settling onthe actual shell differed. For these reasons, it is unlikely that the un-derlying process of metamorphosis and the availability or efficacy ofthe inducing cue determined the distribution patterns of settlersamong these substrata types.

Non-random distributions such as the one we observed in the no-choice and choice assays can also be due to active choice (or, of course,rejection) of certain settlement substrata by the larvae, i.e. the larvae ex-press a preference. Our study was not designed to investigate whetheractive larval choice resulted in the aggregated distribution of juveniles,using the experimental approach suggested by Olabarria et al. (2002).

Table 5Field experiment. One-way ANOVA to test for differences in juvenile Crepidula fornicatadensities between microhabitat types after two weeks (biweekly) and eight weeks(seasonal). Slate plates were manipulated to simulate different microhabitats byattaching stones, Crepidula shells or mussel shells, or were left bare. Data from biweeklypanels on 13th September were arcsine-square root transformed to achievehomogeneous variances. Significant differences are shown in bold. Tukey post-hoctesting was performed where significant differences were detected (*).

df MS F p

Biweekly 1(30th July) 3 31872.6 0.091 0.963Residual 12 349474.9Biweekly 2 (13th August) 3 155954.076 1.094 0.389Residual 12 142578.551Biweekly 3 (28th August) 3 10672.267 0.470 0.986Residual 12 224967.464Biweekly 4 (13th September) 3 26.089 0.622 0.614Residual 12 41.926Seasonal (13th September) 3 510557.910 3.809 0.040⁎

Residual 12 134042.149⁎Tukey Panel b mussel

However, one can use the laboratory results to speculate on mecha-nisms causing the observed patterns. A change in the ratio of settlerabundance between two (or more) different substrata when offered achoice compared to when no choice is available is argued to demon-strate active preference sensu Olabarria et al. (2002). In our study theratios of settler abundance on mussel vs Crepidula shells between twono-choice assays and in a single choice assay were similar, suggestingthat although therewas greater settlement on Crepidula shells, no activepreference was taking place. The observed patterns of juvenile distribu-tion are thus likely due to other processes such as differences in ‘acces-sibility’ of the shells. Olabarria et al. (2002) refer to this as the ‘ease withwhich amicrohabitat can be found or be occupied’. In our study, unsuit-able surface complexity, for example, may have reduced the numbers oflarvae that could successfully attach to the mussel shell during settle-ment, even if equally ‘liked’ compared to other potential surfaces andpresenting similar metamorphosis-stimulating cues. Similarly, severalstudies have reported differences in the levels of settlement on certainsurface structures over others for barnacles (Berntsson et al., 2000;Hills and Thomason, 1998; Raimondi, 1988), even if the process (activebehavioural vs. passive processes such as accessibility or differential

0.0

0.5

1.0

1.5

2.0

2.5

3.0

28-Aug

0.0

0.5

1.0

1.5

2.0

2.5

3.0

13-Sep

biweekly

seasonal

Den

sity

Fig. 4. Juvenile Crepidula fornicata densities on the different substrata types that weredeployed for two weeks (biweekly, recovered 30th July, 13th August, 28th August and13th September 2011) or eight weeks (seasonal, recovered only on 13th September2011) during the settlement season of 2011 at Pennar in the Milford Haven Waterway(MHW). Ordinary roofing slates were used as a base and either left bare (‘Panel’), orwere covered in flat stones (‘Stone’), empty Crepidula fornicata shells (‘Crepidula’) orempty Mytilus edulis shells (‘Mussel’). Mean ± SD. n = 4.

296 K. Bohn et al. / Journal of Experimental Marine Biology and Ecology 448 (2013) 289–297

availability of inducing cues) has not always been explicitly tested. Suchstudies into which life cycle stages determine the distribution of a spe-cies are however crucial when aiming at predicting the dispersal poten-tial of non-native species and how larval supply, habitat availability andthe presence of already established adult populations may affect thespecies' potential distribution, even if underlying mechanisms remainunknown.

Besides those abovementioned findings of differential settlement infavour of Crepidula shells, we have found some evidence that this distri-bution that was established at settlement may change during the firstday following settlement. In run 2 of the choice experiment, changesin settlement patterns between 6 h and 24 h were apparent, indicatingthat post-metamorphicmovement alters the dispersal of settlerswithin24 h. Surprisingly, the change here occurred in favour towards a ‘moreeven’ distribution: after 24 h, settlers were observed in equal propor-tions between both shell types (indicated by non-significant deviationsfrom an expected even (1:1) distribution). This 24 h-distribution wasfound to deviate from the distribution determined after 6 h, indicatingthat some time after metamorphosis a choice is made by the juvenilesin favour of the mussel shells, levelling out any differences establishedduring settlement due to better accessibility of the Crepidula shells. Itshould be noted, however, that total levels of metamorphosis almostdoubled between 6 h and 24 h in run 2. Late metamorphoses of somelarvae instead of only post-metamorphic movement could account forsome of the observed changes in the distribution over time. Differentialmortality can be excluded as the causal post-colonisation process underthe experimental conditions of our laboratory studies. Mortality wasoverall low and 93% of all introduced individuals were recovered aliveat the end of this trial, either as settled (between 36 and 68% in the var-ious runs) or as free-swimming larvae (between 32 and 64%).

Interestingly, any evidence for aggregative settlement from labora-tory experiments was not reflected in our observations from the field,where biweekly settlement rates never differed between microhabitattypes. A lack of discrimination of microhabitat in field trials in contrastto what is predicted from laboratory experiments has also been ob-served in barnacles (Thompson et al., 1998). It is likely that local hydro-dynamics inhibit settlement in association with certain microhabitattypes under field conditions, even on the most accessible surfaces(i.e. the Crepidula shells, as laboratory assays suggest), resulting ina random deposition of the larvae. Juvenile movement between hab-itat types, although a potentially important process as revealed byour laboratory studies, is unlikely the main reason for this observeddistribution, as the setup used in our field experiment would haverequired the juveniles to move from the panels onto the surroundingsediments, providing a large obstacle for juvenile movement. EPSM,although previously shown to be high at the same study site (Bohnet al., 2013), is also unlikely to have caused the distribution amongmicrohabitat types on the biweekly panels. Whilst submersed, EPSMis low (Bohn et al., 2013), and sampling of the manipulated panels inthe present study had always been undertaken on the first day of springtides when the panels were accessible again, decreasing the potentialloss of juveniles due to mortality in this treatment.

Work by Shenk and Karlson (1986) and McGee and Targett (1989)showed that the distribution of C. fornicata established during settle-ment is altered by post-settlement processes, resulting in a differentialspread of early settlers to that of older juveniles and adults. Findingsfrom the present study confirm this. At the end of the field experiment,more larvae were observed on the mussel microhabitat panels com-pared to the bare panels, suggesting that the mussel shells offered amore suitable habitat for survival during the first days of intertidal ex-posure. Our combined results from laboratory and field studies thusshow that although settlement occurs in association with specific habi-tat types (shells of conspecifics) under laboratory conditions, its fine-scale distribution in the intertidal is not determined by such processes,but by variable levels of post-settlement mortality between microhabi-tat types following an even dispersion of settling larvae across habitat

types. Microhabitat choice by the early life cycle stages of a speciesmay be essential in determining its recruitment success, especially ifthe early stages are highly susceptible to mortality (Gosselin and Chia,1995b). If larvae lack the ability to distinguish between microhabitatsencountered during the process of metamorphosis, this may havedetrimental effects on later survival and population growth. Inter-estingly, C. fornicata attains highest densities in the subtidal andlow intertidal (Blanchard, 1997; Bohn, 2012). It is likely that the spe-cies is not well adapted to intertidal conditions, and also has notevolved the behavioural response necessary to counteract such pas-sive processes through controlled settlement in association with mi-crohabitats that would support successful recruitment of the speciesintertidally. If this is the case, reproductionwould also be impaired inthe intertidal, due to difficulties of larvae or juveniles to locate pre-existing stacks or EPSM overruling any potential associations in fa-vour of settlement on conspecifics. It is highly likely that processesthat determine the fine-scale distribution of C. fornicata undersubtidal conditions, where the species is exposed to less physicalstressors, differ from those acting under intertidal conditions. Fur-ther work is needed to investigate whether the different processesacting under intertidal and subtidal conditions may also cause varia-tions in C. fornicata's fine-scale distribution.

Acknowledgments

This study was funded by the Countryside Council for Wales (CCW,now part of Natural Resources Wales - NRW) and the Bangor MusselProducers Ltd via a PhD studentship to KB.We are grateful for the assis-tance of Ben Harvey and Jorge Dominguez in the field. [SS]

References

Berntsson, K.M., Jonsson, P.R., Lejhall, M., Gatenholm, P., 2000. Analysis of behavioural re-jection of micro-textured surfaces and implications for recruitment by the barnacleBalanus improvisus. J. Exp. Mar. Biol. Ecol. 251, 59–83.

Blanchard, M., 1997. Spread of the slipper limpet Crepidula fornicata (L. 1758) in Europe.Current state and consequences. Sci. Mar. 61 (Suppl. 2), 109–118.

Blanchard, M., Pechenik, J.A., Giudicelli, E., Connan, J.P., Robert, R., 2008. Competition forfood in the larvae of two marine molluscs, Crepidula fornicata and Crassostrea gigas.Aquat. Living Resour. 21, 197–205.

Bohn, K., 2012. The Distribution and Potential Northwards Spread of the Non-native Gas-tropod Crepidula fornicata in Welsh Coastal Waters, School of Ocean Sciences. BangorUniversity 154.

Bohn, K., Richardson, C.A., Jenkins, S.R., 2012. The invasive gastropod Crepidula fornicata:reproduction and recruitment in the intertidal at its northernmost range in Wales,UK, and implications for its secondary spread. Mar. Biol. 159, 2091–2103.

Bohn, K., Richardson, C.A., Jenkins, S.R., 2013. The importance of larval supply, larval habitatselection and post-settlement mortality in determining intertidal adult abundance ofthe invasive gastropod Crepidula fornicata. J. Exp. Mar. Biol. Ecol. 440, 132–140.

Chiu, J.M.Y., Thiyagarajan, V., Pechenik, J.A., Hung, O.S., Qian, P.Y., 2007. Influence of bacteriaand diatoms in biofilms on metamorphosis of the marine slipper limpet Crepidulaonyx. Mar. Biol. 151, 1417–1431.

Cole, H.A., Baird, R.H., 1953. The American slipper limpet (Crepidula fornicata) in MilfordHaven. Nature 172, 687.

Crothers, J.H., 1966. Dale Fort Marine Fauna. Field Studies Council.Crouch, W., 1893. On the occurrence of Crepidula fornicata in Essex. Proc. Malacol.

Soc. Lond. 1, 19.Crowe, T.P., Underwood, A.J., 1998. Testing behavioural “preference” for suitable microhab-

itat. J. Exp. Mar. Biol. Ecol. 225, 1–11.Decottignies, P., Beninger, P.G., Rince, Y., Riera, P., 2007a. Trophic interactions between

two introduced suspension-feeders, Crepidula fornicata and Crassostrea gigas, areinfluenced by seasonal effects and qualitative selection capacity. J. Exp. Mar. Biol.Ecol. 342, 231–241.

Decottignies, P., Beninger, P.G., Rince, Y., Robins, R.J., Riera, P., 2007b. Exploitation of nat-ural food sources by two sympatric, invasive suspension-feeders: Crassostrea gigasand Crepidula fornicata. Mar. Ecol. Prog. Ser. 334, 179–192.

Delany, J., Myers, A.A., McGrath, D., O'Riordan, R.M., Power, A.M., 2003. Role of post-settlement mortality and ‘supply-side’ ecology in setting patterns of intertidal distri-bution in the chthamalid barnacles Chthamalus montagui and C. stellatus. Mar. Ecol.Prog. Ser. 249, 207–214.

Eyster, L.S., Pechenik, J.A., 1988. Comparison of growth, respiration, and feeding of juve-nile Crepidula fornicata (L) following natural or KCl-triggered metamorphosis.J. Exp. Mar. Biol. Ecol. 118, 269–279.

Gaudette, M.F., Lowther, J.L., Pechenik, J.A., 2001. Heat shock induces metamorphosis inthe larvae of the prosobranch gastropod Crepidula fornicata. J. Exp. Mar. Biol. Ecol.266, 151–164.

297K. Bohn et al. / Journal of Experimental Marine Biology and Ecology 448 (2013) 289–297

Gosselin, L.A., Chia, F.S., 1995a. Characterizing temperate rocky shores from the perspec-tive of an early juvenile snail— themain threats to survival of newly-hatched Nucellaemarginata. Mar. Biol. 122, 625–635.

Gosselin, L.A., Chia, F.S., 1995b. Distribution and dispersal of early juvenile snails: effec-tiveness of intertidal microhabitats as refuges and food sources. Mar. Ecol. Prog. Ser.128, 213–223.

Gosselin, L.A., Qian, P.Y., 1996. Early post-settlement mortality of an intertidal barnacle: acritical period for survival. Mar. Ecol. Prog. Ser. 135, 69–75.

Gosselin, L.A., Qian, P.Y., 1997. Juvenile mortality in benthic marine invertebrates. Mar.Ecol. Prog. Ser. 146, 265–282.

Hale, R., Downes, B.J., Swearer, S.E., 2008. Habitat selection as a source of inter-specific dif-ferences in recruitment of two diadromous fish species. Freshw. Biol. 53, 2145–2157.

Hills, J.M., Thomason, J.C., 1998. The effect of scales of surface roughness on the settlementof barnacle (Semibalanus balanoides) cyprids. Biofouling 12, 57–69.

Hoagland, K.E., 1978. Protandry and evolution of environmentally-mediated sex change—

study of mollusca. Malacologia 17, 365–391.Hunt, H.L., Scheibling, R.E., 1997. Role of early post-settlementmortality in recruitment of

benthic marine invertebrates. Mar. Ecol. Prog. Ser. 155, 269–301.Jenkins, S.R., 2005. Larval habitat selection, not larval supply, determines settlement pat-

terns and adult distribution in two chthamalid barnacles. J. Anim. Ecol. 74, 893–904.Jenkins, S.R., Hawkins, S.J., 2003. Barnacle larval supply to sheltered rocky shores: a limit-

ing factor? Hydrobiologia 503, 143–151.Jenkins, S.R., Norton, T.A., Hawkins, S.J., 1999. Settlement and post-settlement interactions

between Semibalanus balanoides (L.) (Crustacea: Cirripedia) and three species of fu-coid canopy algae. J. Exp. Mar. Biol. Ecol. 236, 49–67.

Kent, A., Hawkins, S.J., Doncaster, C.P., 2003. Population consequences of mutual attrac-tion between settling and adult barnacles. J. Anim. Ecol. 72, 941–952.

Korringa, P., 1942. Crepidula fornicata's invasion in Europe. Basteria 7, 12–23.Lee, J.T., Widdows, J., Jones, M.B., Coleman, R.A., 2004. Settlement of megalopae and early

juveniles of the velvet swimming crab Necora puber (Decapoda: Portunidae) in flowconditions. Mar. Ecol. Prog. Ser. 272, 191–202.

Loher, T., Armstrong, D.A., 2000. Effects of habitat complexity and relative larval supply onthe establishment of early benthic phase red king crab (Paralithodes camtschaticusTilesius, 1815) populations in Auke Bay, Alaska. J. Exp. Mar. Biol. Ecol. 245, 83–109.

McGee, B.L., Targett, N.M., 1989. Larval habitat selection in Crepidula (L) and its effect onadult distribution patterns. J. Exp. Mar. Biol. Ecol. 131, 195–214.

McMillan, N.F., 1938. Early records of Crepidula in English waters. Proc. Malacol. Soc. 23, 236.Mineur, F., Cook, E.J., Minchin, D., Bohn, K., Macleod, A., Maggs, C.A., 2012. Changing

coasts: marine aliens and artificial structures. Oceanogr. Mar. Biol. Annu. Rev. 50,189–234.

Moksnes, P.O., Pihl, L., van Montfrans, J., 1998. Predation on postlarvae and juveniles ofthe shore crab Carcinus maenas: importance of shelter, size and cannibalism. Mar.Ecol. Prog. Ser. 166, 211–225.

Nelson-Smith, A., 1965. The marine biology of Milford Haven: the physical environment.Field Stud. 2, 155–188.

Nelson-Smith, A., 1967. Marine biology ofMilford Haven: the distribution of littoral plantsand animals. Field Stud. 2, 435–477.

Olabarria, C., Underwood, A.J., Chapman, M.G., 2002. Appropriate experimental design toevaluate preferences for microhabitat: an example of preferences by species ofmicrogastropods. Oecologia 132, 159–166.

Pawlik, J.R., 1992. Chemical ecology of the settlement of benthic marine invertebrates.Oceanogr. Mar. Biol. 30, 273–335.

Pechenik, J.A., 1980. Growth and energy-balance during the larval lives of 3 prosobranchgastropods. J. Exp. Mar. Biol. Ecol. 44, 1–28.

Pechenik, J.A., Gee, C.C., 1993. Onset of metamorphic competence in larvae of the gastro-pod Crepidula fornicata (L), judged by a natural and an artificial cue. J. Exp. Mar. Biol.Ecol. 167, 59–72.

Pechenik, J.A., Heyman, W.D., 1987. Using KCl to determine size at competence for larvaeof the marine gastropod Crepidula fornicata (L). J. Exp. Mar. Biol. Ecol. 112, 27–38.

Pechenik, J.A., Lima, G.M., 1984. Relationship between growth, differentiation, and lengthof larval life for individually reared larvae of the marine gastropod, Crepidula fornicata.Biol. Bull.-Us 166, 537–549.

Pechenik, J.A., Blanchard, M., Rotjan, R., 2004. Susceptibility of larval Crepidula fornicata topredation by suspension-feeding adults. J. Exp. Mar. Biol. Ecol. 306, 75–94.

Power, A.M., Delany, J., McGrath, D., Myers, A.A., O'Riordan, R.M., 2006. Patterns of adultabundance in Chthamalus stellatus (Poli) and C. montagui Southward (Crustacea:Cirripedia) emerge during late recruitment. J. Exp. Mar. Biol. Ecol. 332, 151–165.

Raimondi, P.T., 1988. Rock type affects settlement, recruitment, and zonation of the bar-nacle Chthamalus anisopoma Pilsbury. J. Exp. Mar. Biol. Ecol. 123, 253–267.

Riera, P., Stal, L.J., Nieuwenhuize, J., 2002. Delta C-13 versus delta N-15 of co-occurringmolluscs within a community dominated by Crassostrea gigas and Crepidula fornicata(Oosterschelde, The Netherlands). Mar. Ecol. Prog. Ser. 240, 291–295.

Shenk, M.A., Karlson, R.H., 1986. Colonization of a shell resource by calyptraeid gastropods:tests of habitat selection and preemption models. J. Exp. Mar. Biol. Ecol. 99, 79–89.

Sokal, R.R., Rohlf, F.J., 1995. Biometry: The Principles and Practice of Statistics in BiologicalResearch. WH Freeman and Co.

Tait, K.J., Hovel, K.A., 2012. Do predation risk and food availability modify prey andmesopredator microhabitat selection in eelgrass (Zostera marina) habitat? J. Exp.Mar. Biol. Ecol. 426, 60–67.

Taris, N., Comtet, T., Stolba, R., Lasbleiz, R., Pechenik, J.A., Viard, F., 2010. Experimental in-duction of larval metamorphosis by a naturally-produced halogenated compound(dibromomethane) in the invasive mollusc Crepidula fornicata (L.). J. Exp. Mar. Biol.Ecol. 393, 71–77.

Thieltges, D.W., 2005. Impact of an invader: epizootic American slipper limpet Crepidulafornicata reduces survival and growth in European mussels. Mar. Ecol. Prog. Ser.286, 13–19.

Thompson, R.C., Norton, T.A., Hawkins, S.J., 1998. The influence of epilithic microbial filmson the settlement of Semibalanus balanoides cyprids— a comparison between labora-tory and field experiments. Hydrobiologia 375–76, 203–216.

Thouzeau, G., Chauvaud, L., Grall, L., Guerin, L., 2000. Do biotic interactions control pre-recruitment and growth of Pecten maximus (L.) in the Bay of Brest? C. R. Acad. Sci.III 323, 815–825.

Walne, P.R., 1956. The Biology and Distribution of the Slipper Limpet (Crepidula fornicata)in Essex Rivers: With Notes on the Distribution of the Larger Epi-benthic Inverte-brates. H.M.S.O.

Webley, J.A.C., Connolly, R.M., Young, R.A., 2009. Habitat selectivity of megalopae and ju-venile mud crabs (Scylla serrata): implications for recruitment mechanism. Mar. Biol.156, 891–899.

Werner, B., 1955. Über die Anatomie, die Entwicklung und Biologie des Veligers und derVeliconcha von Crepidula fornicata L. (Gastropoda Prosobranchia). Helgol. Mar. Res. 5,169–217.

Zhao, B., Qian, P.Y., 2002. Larval settlement and metamorphosis in the slipper limpetCrepidula onyx (Sowerby) in response to conspecific cues and the cues from biofilm.J. Exp. Mar. Biol. Ecol. 269, 39–51.