Do settlement dynamics influence competitive interactions between an alien tunicate and its native congener?

Abstract Variation in density of early stages, that is, larvae and juveniles, is a major determinant of the distribution and abundance of the adult population of most marine invertebrates. These early stages thus play a key role in competitive interactions, and, more specifically, in invasion dynamics when biologically similar native and non‐native species (NNS) come into contact in the same habitat. We examined the settlement dynamics and settlement rate of two important members of the fouling community that are common on human‐made infrastructures around the world: Ciona robusta (formerly known as Ciona intestinalis type A) and C. intestinalis (formerly known as C. intestinalis type B). In the western English Channel, the two species live in close syntopy following the recent introduction of C. robusta in the native European range of C. intestinalis. Using settlement panels replaced monthly over 2 years in four marinas (including one studied over 4 years) and species‐diagnostic molecular markers to distinguish between juveniles of both species (N = 1,650), we documented similar settlement dynamics of both species, with two settlement periods within a calendar year. With one exception, settlement times were highly similar in the congeners. Although the NNS showed lower settlement density than that of the native congener, its juvenile recruitment was high during the second settlement period that occurs after the warm season, a pattern also observed in adult populations. Altogether, our results suggest that species’ settlement dynamics do not lead to the dominance of one species over the other through space monopolization. In addition, we showed that changes over time are more pronounced in the NNS than in the native species. This is possibly due to a higher sensitivity of the NNS to changes of environmental factors such as temperature and salinity. Environmental changes may thus eventually modify the strength of competitive interactions between the two species as well as species dominance.

of settlement may also facilitate a shift in dominance, favoring NNS at the community scale (Stachowicz, Terwin, Whitlatch, & Osman, 2002). The presence of juveniles can in addition prevent settlement or increase postsettlement mortality in closely related species (e.g., between native and non-native tunicates; Rius, Turon, & Marshall, 2009).
Variation in settlement rate can thus be critical in determining the type and intensity of the interactions between biologically similar congeners living in the same locations and habitats (i.e., syntopy).
We here compared the settlement dynamics of two emblematic tunicate species, for which the taxonomic status has been recently re-evaluated (Brunetti et al., 2015): Ciona robusta and Ciona intestinalis (formerly named C. intestinalis type A and C. intestinalis type B, respectively). The two species are important members of fouling communities and well established in artificial habitats including marinas and harbors worldwide (Aldred & Clare, 2014;Bouchemousse, Lévêque, Dubois, & Viard, 2016;Lambert & Lambert, 1998;Ramsay, Davidson, Landry, & Arsenault, 2008;Tracy & Reyns, 2014). They live in syntopy in several marinas in the western English Channel and southern Brittany in the Northeast Atlantic (Bouchemousse, Lévêque, et al., 2016;Nydam & Harrison, 2011). This is, so far, the only confirmed region where both species coexist (Bouchemousse, Bishop, & Viard, 2016;Caputi et al., 2007;Zhan, Macisaac, & Cristescu, 2010). Their coexistence in this region is due to the recent introduction of C. robusta in the native European range of C. intestinalis (ca. 15-20 years ago; Bishop, Wood, Yunnie, & Griffiths, 2015;Nydam & Harrison, 2011). In this region, C. robusta has never been reported outside marinas (L.L., personal observation). In the English Channel, C. intestinalis is occasionally found in natural habitats (L.L. and F.V., personal observation) although its occurrence and abundance in natural habitats is very low as compared with marinas where flourishing populations develop-and often become dominant in the fouling communities.
Interestingly, the two species are interfertile in laboratory conditions (Bouchemousse, Lévêque, et al., 2016;Suzuki, Nishikawa, & Bird, 2005), but only a few F1 hybrids have been observed in the wild and there is so far no convincing evidence of contemporary introgressions between the two species (Bouchemousse, Haag-Liautard, Bierne, & Viard, 2016;Bouchemousse, Lévêque, et al., 2016;Nydam & Harrison, 2011). These results suggest the existence of efficient reproductive isolation mechanisms in the wild. Competitive interactions may thus be more important than evolutionary interactions (e.g., adaptive introgression) in determining the fate of these two congeneric species in their sympatric range. However, due to their recent taxonomic revision (Brunetti et al., 2015), their specific life cycle, environmental preferences, and population dynamics have yet to be examined in detail.
In a previous study (Bouchemousse, Lévêque, et al., 2016), we documented substantial seasonal variation in the relative abundance of adult populations of the NNS C. robusta compared with its congener, in the English Channel and Atlantic coasts of Brittany. In particular, we showed a very low relative abundance of C. robusta in spring compared with autumn. This variation can be attributed to differential settlement success and/or postsettlement competition according to temperature conditions, favoring C. robusta during the warmer season and the native species C. intestinalis during the colder season.
However, data are scarce on the life cycle and settlement dynamics of the two Ciona species in Brittany. Ciona species usually display a short life cycle, as documented for the two targeted species in other regions (e.g., one or two generations per year for C. intestinalis in Sweden (Dybern, 1965), and at least two generations per year for C. robusta in the Mediterranean Sea; Caputi, Crocetta, Toscano, Sordino, & Cirino, 2015). Based on the literature, we hypothesized that one to two settlement events per year occur in Brittany. Thus, considering putative environmental preferences and the observed variation in adult abundances of the two species (Bouchemousse, Lévêque, et al., 2016), the timing and intensity of settlement may differ between the two species during a year. The existence of a phenological shift may ultimately lead to an increase in competition between the two species, giving the advantage to the species settling first.
To test for differential postsettlement patterns, we gathered data monthly on the settlement dynamics of the two Ciona species in four marinas over 2 years (over 4 years in one of them) on settlement panels. Species-diagnostic molecular markers were used to reliably identify species at the juvenile stage.

| Experimental design and Ciona spp. counts
The settlement dynamics of Ciona spp. were investigated in four marinas located along the northern coasts of Brittany and the Bay of Brest (Figure 1a). Previous observations in 2012 (see Table S1 for details) showed that the four marinas harbor various proportions of adult-stage C. robusta relative to C. intestinalis: The Roscoff and Château marinas are characterized by low to moderate relative proportions of C. robusta, and Trébeurden and Moulin Blanc marinas are characterized by moderate to high relative proportions of C. robusta (Bouchemousse, Lévêque, et al., 2016; Table S1 for details).
In each marina, three panels (Figure 1b), located at 2-m intervals from each other, were placed at 1.5 m depth under a pontoon, close to the local Ciona spp. adult population. Each panel had a horizontal surface (15 × 15 cm, Figure 1b) to provide a substrate orientated in the same direction as the pontoons. They were deployed without preconditioning, that is, not soaked in sea water, to allow natural biofilm to develop. They were replaced every month; therefore, settled juveniles were at most 4 weeks old. Note that with such monthly replacement, we did not examine properly the very early postsettlement stages (e.g., a few days old) during which several processes can occur (e.g., predation, competition). The panels were secured and transported back to the laboratory in a cooler filled with seawater taken from the marina.
In all marinas, panels were retrieved and brought back to the labora- March 2015. In this marina, the settlement dynamics were thus investigated over 52 months labeled from 0 (December 2010) to 51 (March 2015). In the laboratory, for each panel, the number of juveniles was recorded under a stereomicroscope using a grid composed of 11 × 11 cells of 1.21 cm 2 each (1.1 × 1.1 cm, Figure 1c). To avoid edge effects, the first row of cells along the edges was excluded; thus, only 81 cells were used for juvenile counts (Figure 1c). Panels were counted on the day or the day after their collection in the field.
To compare the density of the two species between juvenile and adult stages, samples of adult Ciona spp. populations were carried out in all four marinas in spring and autumn 2014 (i.e., May and October). These adult populations likely correspond to two different generations and to parents of the juveniles that settled during the year. Sampling was carried out by scuba-diving under two adjacent pontoons, close to the settlement panels using 10 quadrats of 0.1 m 2 (40 × 25 cm) per pontoon along a 50-m-long transect with one quadrat every 5 m. Individuals sampled were identified at the species level using the morphological criteria described in Sato, Satoh, and Bishop (2012).

| Sampling and molecular species identification of the juveniles
Due to their small size and large numbers, the juveniles of C. robusta and C. intestinalis cannot be reliably discriminated morphologically. For juveniles sampled in 2013, we used the degenerate primers developed by Nydam and Harrison (2007) that have been shown to be reliable for distinguishing the two species in the study populations (Bouchemousse, Lévêque, et al., 2016). Species identification is based on a PCR-RFLP protocol (i.e., enzymatic digestion of the PCR product) detailed in Nydam and Harrison (2011). Because amplification success was low for juveniles sampled in 2014, we developed species-specific primers that increased the barcoding success significantly. These species-specific primer pairs each amplify mtDNA of only one of the two species. The PCR fragment includes the region with the restriction site targeted with the degenerate nonspecific primers previously used. Thus, both PCR-RFLP and direct PCR essays can be carried out with these primers. The forward and reverse primer sequences for amplifying C. robusta are, respectively, CiA-mtCOIshort-F: 5′-ACAGTTTATCCTCCTTTATCTGCA-3′ and CiA-mtCOIshort-R: 5′-TGGATCTCTTCTCCCATTCGG-3′. The forward and reverse primer sequences specific to C. intestinalis are, respectively, CiB-mtCOIshort-F: 5′-CTTTGCATTTAGCTGGGGTTTC-3′, CiB-mtCOIshort-R: 5′-AGGATCCCTTCTTCTGTTAGGA-3′. The amplifications with these primers were carried out in a total reaction volume of 20 μl, with 5 μl of template DNA diluted to 1:10, 1× buffer (Thermoprime, ABGene ® ), 0.2 mmol/L of each dNTP, 1.5 mmol/L of MgCl 2 , 0.05 μg/μl of bovine serum albumin, 0.05 μmol/L of each primer, and 0.2 U of Taq Polymerase (Thermoprime, ABGene ® ).
A touchdown PCR program was used, starting with 5 min at 95°C followed by 5 cycles with 30 s at 95°C, 30 s at an initial temperature of 56°C (then decreasing by 1°C per cycle), 30 s at 72°C and then 30 cycles at 95°C for 30 s, 52°C for 30 s, 72°C for 30 s, and a final elongation step at 72°C for 10 min. Species determination was easily carried out on an agarose gel (2%) because CiA-mtCOIshort primers amplify a fragment of 306 base pairs (bp) in C. robusta and CiB-mtCOIshort primers amplify a fragment of 239 bp in C. intestinalis. Both nonamplification and the PCR product size can be used to determine the species. Note that every individual was amplified using the two PCR primer pairs to check for consistency of the results.

| Statistical analyses
We analyzed the number of juveniles settled per dm 2 (i.e., juvenile density). We first investigated the annual settlement dynamics of Ciona spp. in each marina (i.e., from March of year y to March of year y + 1). Assuming that each major settlement event follows a Gaussian distribution, the number and characteristics (e.g., time of settlement) of discrete settlement events of juveniles were estimated using a modal decomposition analysis carried out with the R package MIXDIST (Green & Macdonald, 1985;R Development Core Team, 2014). This package contains maximum-likelihood-based methods to find the best fit between the observed data and a mixture of Gaussian distributions (Green & Macdonald, 1985). The density data were smoothed using a weighted moving average at the third order to rule out irregularities (Frontier & Pichod-Viale, 1991). Prior to the modal decomposition analysis, the normality of each dataset modified by the weighted moving average was tested using a Shapiro test in R (the shapiro.test function). A significant deviation from normality is a prerequisite before attempting to find the best mixture of Gaussian curves explaining the data.
To compare the settlement rate among marinas and settlement periods, we used a two-way analysis of variance (ANOVA) on the average density value computed across the month showing the highest density of juveniles (m) and the preceding and following months (m − 1 and m + 1). For months characterized by an overlap between two settlements events within a calendar year, density values were divided by two. "Marinas" (n = 4) and "settlement periods" (n = 4) were considered as random factors. Prior to the ANOVA, the values of density were transformed using the square root function to meet the normality and homoskedasticity conditions (i.e., using the Shapiro and Levene tests, respectively). For factors showing a significant effect, pairwise comparisons were carried out with Student-Newman-Keuls (SNK) post hoc tests.
Following the DNA-based species identification, except in Roscoff, where only C. intestinalis was found during the 2013-2014 survey, the same analyses were carried out for each species separately (i.e., modal decomposition, comparison of the intensity of settlement among sites, and settlement periods). The density of each of study species was computed for each panel and each month by multiplying the density value of Ciona spp. by the relative proportion of C. robusta (or C. intestinalis) obtained with the DNA-based species identification.

| Settlement analyses of Ciona spp. reveal two settlement periods within a year
The settlement surveys carried out over 52 months in the Moulin Blanc marina, and over 25 months in the three other marinas, showed important changes in the number of juveniles observed monthly over the study period ( Figure 2; Tables S2 and S3 for details). In all marinas, juveniles were generally observed from May to June until November-December. Practically, no juveniles were observed from January to March. In addition, regardless of the year or the marina, the distribution of the density of Ciona spp. juveniles differed significantly from a normal distribution (Fig. S1). For each study year, modal decomposition analyses clearly identified two Gaussian curves corresponding to two major periods of settlement with a short overlap. The average time of each settlement period (i.e., determined by the mode of the Gaussian curve) is provided in Table 1. Observed distribution and Gaussian curves for Ciona spp.
are shown in Fig. S1, with parameters characterizing the Gaussian curves detailed in Table S4. Over a given study year, the first settlement period covers juveniles settled between April and August (spring-summer) and the second settlement period corresponds to juveniles settled between August and December (late summer-autumn; Figure 2, Table 1).

| Variations in Ciona spp. settlement patterns among sites
Density data in Figure (Table S2). In comparison, in Roscoff, the density values were always lower, with a maximum mean density value of 54.4 (±20.1 SD) juv/dm 2 for June 2013.  (Table 2a) with the exception of the Moulin Blanc marina. There was however a highly significant difference between sites (p < .001; Table 2a).
Pairwise comparisons (i.e., SNK post hoc tests) indicated that most of After the low densities observed in spring 2011 and 2012 (Table   S2), the intensity of spring settlement steadily increased in 2013 and 2014.  Table S3. Throughout the survey, juveniles of C. robusta were recorded in three of the four marinas surveyed, namely Trébeurden, Château, and Moulin Blanc, but surprisingly absent in Roscoff (over 272 juveniles barcoded). Adults of C. robusta were reported in 2012 in the Roscoff marina before starting the settlement panel experiment (Bouchemousse, Lévêque, et al., 2016).  The unit for the settlement period is the month, with values ranging thus from 0 (December 2010) to 51 (March 2015). Settlement periods were identified using modal decomposition analyses carried out separately for each year on the monthly distribution of the mean juvenile density on panels (n = 3). The settlement period indicated is the modal value of each identified Gaussian curve. Details of the results of the modal decomposition analyses are given in Figure S1 and Table S4 for Ciona spp. and in Figure  S2 and Table S5 for the two species separately. Note that no Ciona robusta juveniles were identified in the Roscoff marina.

| Comparison of settlement patterns between C. robusta and C. intestinalis
T A B L E 1 Mean settlement period for each settlement event identified over the course of the study (i.e., 52 months) for Ciona spp. and for Ciona robusta and Ciona intestinalis separately during spring and early summer for the first settlement period and during end of summer and autumn for the second one ( Figure 4). The average settlement time per settlement period and marina is provided in Table 1 for C. intestinalis and C. robusta. Details on the parameters of the modal decomposition outcomes are given in Table S5. Interestingly, the settlement timing of the two species was very similar regardless of the settlement period (Table 1, Figure 4). There was however one ex- T A B L E 2 Result of the two-way ANOVA testing the effect of the settlement period (n = 4, 1st-13: first settlement period in 2013, etc., see

| Comparison of density patterns at adult stages between C. robusta and C. intestinalis
Throughout the study, the same trend observed for juveniles was observed between adults of the two species sampled in 2014 in the three marinas where both species were present ( Figure 5), that is, (1) a higher density for both species in autumn compared to spring generation, (2) a higher density for C. intestinalis compared to C. robusta whatever the season, and (3)

| Similar timing of settlement for the non-native and native congeners in syntopic localities
In all marinas, except Roscoff, both C. robusta and C. intestinalis recruited on the settlement panels that we set out. When both species were present in a given marina and month, the juveniles of both species were found on each panel without any evidence of spatial clustering (i.e., relative proportions of the two species were similar on all three panels). That the two species are well mixed on experimental substrates reflects observations of adults growing under pontoons (see figure 1 in Bouchemousse, Lévêque, et al., 2016) (Bouchemousse, Lévêque, et al., 2016).
Previous studies of these two Ciona species have reported juvenile settlement dynamics for one or the other of the two species independently. Although these studies use the names "Ciona intestinalis" or "C. intestinalis type B" and "C. intestinalis type A" (Nydam & Harrison, 2007), we were able to determine the identity of the spe-  (Carver, Chisholm, & Mallet, 2003;Howes, Herbinger, Darnell, & Vercaemer, 2007;Vercaemer, Sephton, Nicolas, Howes, & Keays, 2011), in Sweden (Dybern, 1965), and in a Danish fjord (Petersen & Svane, 1995), where two settlement periods were identified and are similar to those found in our study. In the English Channel, information is available on adults only (i.e., in Plymouth, Orton, 1914Orton, , 1920, but also fit with the assumption of two generations, and thus two settlement periods, per year. For C. robusta, our study is the first to document settlement patterns in the NE Atlantic. There are however data available for San Francisco Bay, another region where the species has been presumably introduced: Two settlement periods are also observed, one occurring in August and one in October-November (Blum et al., 2007). Other Consequently, competitive interactions for space may occur during juvenile growth in localities where the recruitment is intense for both species, such as the Château marina ( Figure 3). Such early competition may influence adult abundance because recruitment success is a key feature in space monopolization for sessile organisms (e.g., between ascidian species; Rius, Potter, Aguirre, & Stachowicz, 2014). There is however one exception to the general pattern observed, in Trébeurden in 2014 (Table 1, Figure 3), where recruitment of C. robusta juveniles was delayed by at least 4 months compared with their native congeners. We have no clear explanation for this singularity, especially considering that we did not observe any obvious differences in gamete production in either species in this marina (Bouchemousse, Lévêque, et al., 2016). However, this finding may indicate the onset of a phenological shift between the two species in this marina, which is also one of the few marinas where the NNS C. robusta can have higher relative abundance than C. intestinalis at the adult stage (Bouchemousse, Lévêque, et al., 2016). In any case, we conclude that phenological shifts are not driving any potential competitive exclusion between the NNS C. robusta and its native congener in situations of syntopy.

| Can larval supply or environmental conditions explain variation in settlement success between marinas?
The overall settlement timing was similar for the two species, but we observed important differences in settlement rate of juveniles at dif- Higher densities for C. intestinalis were observed in the two marinas in the Bay of Brest, and for C. robusta in Château.
Local hydrodynamic features can have a strong impact on the density and the aggregation of Ciona spp. juveniles (Havenhand & Svane, 1991), and more generally on marine invertebrates with a benthopelagic life cycle (Hunt & Scheibling, 1997). Both the Château and Moulin Blanc marinas are located in the Bay of Brest, a semiclosed system, which may contribute to retaining larvae close to the sites where they have been released and thus enhance larval supply near resident adult populations (Gaines & Bertness, 1992). Moreover, close to estuaries of important rivers (i.e., Aulne and Elorn, Figure  to be developed to examine in further detail the role of larval supply/ larval export to/from urban marine environments. In contrast to C. intestinalis, we found a significant settlement period effect on the density of juveniles for C. robusta (Table 2b). As pictured in Figure 3, the abundance of settled C. robusta juveniles increased over time within a year in two of the three study marinas where this species was present (i.e., Trébeurden and Château). This increase in C. robusta abundance within a year was also documented for adults ( Figure 5) with up to a 10-fold increase in the density of C. robusta in autumn 2014 compared with spring 2014. This increase is most likely due to warmer conditions from July to September (Figure 2), that are more favorable to C. robusta recruitment, considered to be a warmtemperate species compared with the cold-temperate species C. intestinalis (Bouchemousse, Lévêque, et al., 2016;Procaccini, Affinito, Toscano, & Sordino, 2011). This presumed difference in temperature preferences may also explain the disappearance of C. robusta in the Roscoff marina after 2012 (i.e., 1 year after its building). Adult individuals of C. robusta have indeed been observed in 2012 in this marina (Table S1) but have completely disappeared in 2013 and 2014 (Table S1, Figure 5). Note that more recent surveys (July 2016) confirmed the absence of C. robusta in this locality (L. Leveque & F. Viard, unpublished data). This marina is located in the coldest part of the Brittany coastline (Gallon et al., 2014), and its maximal temperatures are below 17°C ( Figure 2). These temperature conditions may favor reproduction and sustainable population establishment of C. intestinalis at the expense of C. robusta (Bouchemousse, Lévêque, et al., 2016).
Other abiotic factors such as large changes in surface salinity due to rainfall and river inputs are known to influence the dynamics of ascidian communities established under pontoons and floating docks (Lambert & Lambert, 1998. In our study, we observed dramatic  (Bouchemousse, Lévêque, et al., 2016). Similar population collapses notably due to rainfall events have been recently documented in another introduced ascidian, S. plicata (Pineda, Turon, Perez-Portela, & Lopez-Legentil, 2016 and references therein). Although located close by (5 km), the Château marina seems less exposed to such events, perhaps because it is further away from the estuaries of two important rivers (i.e., Aulne and Elorn, Figure 1a) than Moulin Blanc.
Similarly, Trébeurden and Roscoff being located in fully marine parts of the Brittany coastline are even less affected by rainfall. Apart from stochastic events, whose effects may have an impact over several years,

| Does competition with its native congener limit the invasiveness of C. robusta?
Despite its advantage during the warmer season, we documented a consistently low recruitment success in C. robusta than in C. intestinalis. For example, the maximum value observed over the whole dataset was in Château in October 2014, with a density value of 361.6 juv/dm 2 , a value more than two times lower than for C. intestinalis (898.7 juv/dm 2 ) at the same place and date. The same result was observed for the adult stage in all sampling periods ( Figure 5).
Inhibition of settlement and subsequent growth by a superior competitor can strongly influence the abundance of species, as shown in numerous cases for marine invertebrates (Davis, Butler, & Vanaltena, 1991;Grosberg, 1981), including in ascidians (Rius et al., 2009;Svane & Young, 1989). For example, using ex situ experiments, Rius et al. (2009) showed that the presence of S. plicata juveniles, a non-native ascidian in Australia, increased juvenile mortality and reduced the growth of the native ascidian Microcomus squamiger, probably via competition for food. The opposite situation, that is, negative influence of native species on the settlement of NNS, has also been reported. For example, in Spain (Mediterranean Sea), predation of larvae by juveniles of the native mussel M. galloprovincialis was shown to stimulate settlement avoidance close to mussels of two non-native ascidians S. plicata and Microcomus squamiger (Ordonez et al., 2013 Considering that competition between the two species at the settlement stage does not currently appear to be a significant driving factor of population abundance, and because C. robusta settlement and adult density both increase in autumn, the abundance of this species is expected to increase from year to year. However, the exact opposite trend is observed, with a particularly low adult density and juvenile settlement rate of this species in spring. For instance, in Trébeurden, the highest density value for C. robusta during the second settlement event of 2013 was twice as low as that of C. intestinalis, but the adult density value was 27 times lower for C. robusta compared with C. intestinalis in spring 2014. We suggest that the autumn settlement of C. robusta is successful, but subsequent growth and/or survival is limited. The NNS seems to suffer from unfavorable environmental conditions during the winter season (i.e., coldest temperatures and strong variation of salinity, as explained above for Ciona spp.), which may either reduce juvenile growth and/or cause adult mortality before the onset of their reproductive period in spring. This finding is consistent with known differences in environmental preferences between the two species (i.e., C. robusta is considered to be a warmtemperate species and C. intestinalis is a more cold-temperate species, as mentioned before). Thus, the winter season seems to have a lesser impact on the mortality of C. intestinalis populations than on those of C. robusta and may therefore explain the much lower adult density and thus settlement density of C. robusta than C. intestinalis in the spring.
Altogether, our results suggest that competitive exclusion at early stages between the NNS C. robusta and its native congener in situation of syntopy is unlikely to play a major role in determining their patterns of coexistence in the wild. Regarding the density of adults