Structure of intertidal sessile communities before and after the invasion of Isognomon bicolor ( C . B . Adams , 1845 ) ( Bivalvia , Isognomonidae ) in southeastern Brazil

Scientists recognize the importance of ecological data prior to invasion by non-native species in order to evaluate changes in the recipient community. Here we assess the potential impact of the invasion of the bivalve Isognomon bicolor (C.B. Adams, 1845) on Brazilian rocky shores through the use of surveys both before and after the arrival of this non–native species. The invader was mostly distributed across the mid and low shore levels of the intertidal zone with relative abundance ranging from 9.0 to 36.7 percent cover. The mid shore, previously dominated by the native barnacle Tetraclita stalactifera (Lamarck, 1818), was co-dominated by this barnacle species and I. bicolor after invasion. The relative abundance of these species, and presumably the interaction strength between them, differed between sites. At the site where I. bicolor reached the highest abundance (around 30% on average), the abundance of T. stalactifera decreased on average 70% compared to baseline values obtained before the I. bicolor invasion. Finally, conspicuous and extensive I. bicolor beds such as those observed in this study have not been reported in its original distribution range. Beds of I. bicolor may create a much more intricate biogenic matrix than the extents of bare rock and barnacle clumps it replaced. This bivalve may act as an ecosystem engineer and, thus a functionally different component of the intertidal community in its invaded range compared to its native distribution.


Introduction
Invasion of marine and freshwater systems by non-native species can cause significant change in the abundance and distribution of species in recipient communities (see review in Carlton 1996;Ricciardi and MacIsaac 2011).The effect of invasive species on native species is now the second most addressed topic in studies of biological invasion (Lowry et al. 2013).In marine habitats, most of these impacts have been investigated once the exotic species have already become invasive either by comparisons of sites with different levels of invasion (e.g Grosholz and Ruiz 1995;Crooks and Khim 1999;Rilov et al. 2004;Bownes and McQuaid 2009) or by field experiments in which the presence or density of invasive species were manipulated (e.g Ross et al. 2003;Byers 2005;Eastwood et al. 2007;Salvaterra et al. 2013).The analysis of pre-and post-invasion datasets is central to unequivocally identify changes in the recipient community associated with the establishment of the invader (Grosholz et al. 2000;Grosholz 2005;Sánchez et al. 2005;Cuevas et al. 2006;Robinson et al. 2007;Hanekom 2008).Unfortunately, pre-invasion baseline data are not available in most cases (Forrest and Taylor 2002;Junqueira 2013).In this work, we present data from the southeast coast of Brazil on the structure of rocky intertidal sessile assemblages both before and after the shores were invaded by the bivalve Isognomon bicolor (C. B. Adams, 1845).
The bicolor purse oyster Isognomon bicolor, originally distributed in the Gulf of Mexico, south of Florida and the Caribbean region, has been recorded as an exotic species at several sites along 4,956 km of the Brazilian coast, from Piauí to Santa Catarina (Domaneschi and Martins 2002;Loebmann et al. 2010;Dias et al. 2013).Isognomon bicolor was first reported for the Brazilian coast in 1994 on a rocky shore at Barequeçaba (23º49′ S; 45º26′ W) beach in the state of São Paulo where this exotic species was found in dense aggregations at shore levels previously occupied by oysters and the bivalve Brachidontes sp.(Domaneschi and Martins 2002).At Arraial do Cabo, our study area, this species was not detected in surveys carried out before 1991 (Lavrado 1992) but it was recorded on two rocky shores, Ponta da Fortaleza and Prainha, during a survey performed between 1994 and 1999 along the southeast coast of Brazil (Domaneschi and Martins 2002). Later, in 2002-2003, I. bicolor densities of more than 1 × 10 5 individuals m -2 were reported for shores of Rio de Janeiro state, including Ponta da Fortaleza at Arraial do Cabo (Breves-Ramos et al. 2010).
The published information on the I. bicolor invasion is focused on the population ecology of this species, including its distribution, abundance, and size structure (Fernandes et al. 2004;Oliveira and Creed 2008;Breves-Ramos et al. 2010;Ignacio et al. 2010;Zamprogno et al. 2010;Martinez 2012;Dias et al. 2013).Nevertheless, recent experimental studies have reported biological interactions of I. bicolor with some native species such as the positive association with the macroalgae Sargassum sp.(López and Coutinho 2010) and its consumption by two guilds of predators (López et al. 2010).Despite great concern on the possible negative effects of I. bicolor, there is no information about broader consequences of this invasion at the community level.Since substrate is often a limiting resource, and competition among sessile organisms is one of the main ecological processes structuring communities on rocky shores (Connell 1983), changes in the structure of intertidal sessile communities can be expected as a consequence of the I. bicolor invasion.In this study, our hypotheses were: (i) modifications in the structure of sessile assemblages are related to changes in the abundance of those resident species that overlap most of their vertical distribution with that of the invasive bivalve and (ii) I. bicolor and bare substrate cover percent will be negatively correlated.We compared the abundance and distribution of the dominant resident sessile invertebrates before (1990)(1991) and after (2002)(2003) the invasion by I. bicolor at two rocky-shore sites.Throughout this study, we refer to existing communities as resident or recipient, avoiding the use of the terms 'native' and 'indigenous' since there are other species in the intertidal assemblages on the sampled rocky shores which may represent older, naturalized, introductions such as the brown mussel Perna perna (Linnaeus, 1758) and the titan acorn barnacle Megabalanus coccopoma (Darwin, 1854) (Junqueira et al. 2009).

Methods
This study was carried out on intertidal rocky shores at the Cabo Frio upwelling region in southeastern Brazil (Figure 1).The tidal regime is semi-diurnal with a maximum range of 1.3 m during spring tides.We surveyed two sites, 650 m away from each other: Forno (FN, 22º58′05″S; 42º00′20″W) and Ponta da Fortaleza (PF,22º58′12″S;42º00′43″W).An assessment of the structure of intertidal communities at both sites was conducted in spring 1990 and summer 1991, before the invasive bivalve was detected (Lavrado 1992).We used these datasets as baseline, pre-invasion information, on the vertical distribution and abundance of sessile organisms.We re-surveyed these sites in spring 2002 and summer 2003, after the I. bicolor invasion.
The same sampling protocol was used during both sampling periods.Three transects were randomly positioned on the shore perpendicular to the water line.Along each transect, 0.2 × 0.2 m quadrats were consecutively placed from the low to the high shore level encompassing the intertidal zone.Each quadrat contained a grid of one-hundred equally spaced intersections.The organisms or bare substrate below each intersection was recorded to estimate the percent cover.The width of the intertidal zone varied from 1.6 to 2.0 m and from 3.0 to 4.6 m at FN and PF, respectively.This resulted in a surveyed area of approximately 20 m 2 and 58 m 2 at FN and PF, respectively.

Data analyses
The vertical distribution of sessile invertebrates was qualitatively analyzed by graphically representing the percent cover of each species from the low to the high shore level.Changes in invertebrate assemblages between seasons (spring vs. summer) and alterations associated with the establishment of I. bicolor (before vs. after comparisons) were evaluated for each shore separately where the exotic bivalve had been reported at the study region before this study was carried out (Domaneschi and Martins 2002).using multivariate analysis.The relative abundance was calculated as the sum of the intersection points where each species was recorded divided by the sum of all the intersection points where sessile invertebrate species occurred in the whole transect.After the square root transformation of the relative abundances, a resemblance matrix based on the Bray-Curtis similarity index was used to perform the non-metric multidimensional scaling (nMDS) ordination and the permutational analysis of variance (PERMANOVA) (Clark and Warwick 2001; Anderson et al. 2008).The data transformation was applied to down-weight the contribution of dominant species to the between-sample similarities (Clarke and Warwick 2001).When a significant effect was detected using PERM-ANOVA, the permutational analysis of homogeneity of multivariate dispersions (PERMDISP) (both based on 999 permutations) was performed to investigate the nature of the multivariate effects (i.e.location and/or dispersion effects).Isognomon bicolor was not included in the analyses to avoid magnification of differences between assemblages due to the absence of the invader in 1990-1991.The contribution of each species to dissimi-larities among groups was assessed using the Similarity Percentage Procedure (SIMPER).
Additionally, we analyzed the correlation between I. bicolor and those resident species that overlapped their vertical distribution with the invasive bivalve (i) Tetraclita stalactifera, (ii) Perna perna, (iii) Brachidontes solisianus as well as bare substrate.Each correlation was performed using the percent cover (arcsin p transformation) along the shore level where each resident species (or bare substrate) overlapped with the invader after invasion.We included the three transects from the two sites and the two seasons that were sampled only after the invasion by I. bicolor (n=12).We pooled data from sites and seasons to increase the number of observations in each correlation analysis.Further-more, the scatterplots for each season displayed similar patterns of relationship, even at PF where the invertebrate assemblages differed between seasons (see results below).Pearson product-moment correlation was used when data met the assumption of normality; otherwise the non-parametric Spearman's rank correlation was performed.Statistical analyses were done with STATISTICA 1 0.0 (for univariate analysis) and PRIMER V6 and PERMANOVA+ for PRIMER software packages (for multivariate analysis) (Clarke and Gorley 2006;Anderson et al. 2008).

Spatial overlapping between Isognomon bicolor and resident species
The intertidal zone was wider at PF than at FN.To illustrate the vertical profile of the sessile invertebrates, only one of the sampled transects in each shore, season and period was graphically represented because the three sampled transects varied little in terms of species distribution (Figure 2).In the pre-invasion period, there were three zones dominated by: (i) Chthamalus bisinuatus and Brachidontes solisianus in the high shore; (ii) T. stalactifera in the mid shore and (iii) Megabalanus spp.and P. perna in the low shore (Figure 2).In 2002-2003 (after invasion), Isognomon bicolor was distributed from the mid to the low intertidal at PF and from the high to the low intertidal at FN.At PF, the mid-intertidal zone dominated by T. stalactifera before invasion was co-dominated by this barnacle and I. bicolor after invasion.At FN, I. bicolor was conspicuous (20 to 40 percent cover) along the same shore level as T. stalactifera but also overlapped with the lower distribution ends of B. solisianus and Chthamalus bisinuatus (Pilsbry, 1916).During summer, I. bicolor extended its vertical distribution downward on the shore and increased the spatial overlap with the mussel Perna perna and barnacles of the genus Megabalanus on both shores.

Temporal variation in sessile invertebrate assemblages
In all surveys, the barnacle Chthamalus bisinuatus was the most abundant sessile species.The relative abundance of the other sessile species on each shore changed between periods and seasons (Table 1).Before the invasion, in 1990-1991, the barnacle Tetraclita stalactifera was the second most abundant species except during summer at PF when the mussel Perna perna ranked second.
In general, P. perna abundance was low (less than 5%, on average) during this period, but a 6fold increase of its cover was detected at PF from spring to summer.In 2002-2003, the invasive bivalve Isognomon bicolor ranked second and third at PF and FN, respectively.The relative abundances of this species ranged from 9.1 to 28.5% at FN and from 19.1 to 36.7% at PF.After the invasion T. stalactifera was still the second most abundant species at FN, but at PF it was one of the least abundant species.In contrast, Brachidontes solisianus was one of the most abundant species before invasion but one of the least abundant species after invasion at FN.At PF, however, it ranked similarly between periods.Multivariate analysis of sessile invertebrate assemblages (omitting the invasive species Isognomon bicolor) did not show interaction between the factors 'period' and 'season' (Table 2) and differences between periods, before (1990)(1991) and after (2002)(2003) the invasion, were clear and significant on both shores (Figure 3; Table 2).Regardless of whether the sampling was conducted before or after the invasion, seasonal differences were detected only at PF (Figure 3; Table 2. Summary of PERMANOVA results testing for differences in the intertidal sessile assemblages on the two surveyed shores according to the factors period (P; two levels: before and after the I. bicolor invasion) and season (Se; two levels: spring and summer).The contribution of I. bicolor was excluded from the analysis to avoid over magnification of the differences.Analyses were based on Bray-Curtis dissimilarities, and each test was performed using 999 permutations.Significant p-values are given in bold.Table 2).In addition, differences in the multivariate dispersion (PERMDISP analysis) of invertebrate assemblages were detected between periods (F = 6.93, p =0.043) and season (F = 5.34, p = 0.023) at PF.Then, at this site, groups differed not only in their location but also in their dispersion.The variability of the assemblages at PF (mean of deviations from the centroid ± SE) was higher before than after the invasion (before: 10.24 ± 1.21; after: 6.58 ± 0.69), and also larger in spring compared to summer (spring: 9.92 ± 0.86; summer: 7.24 ± 0.78) (Figure 3).Since the taxonomic composition stayed the same, the differences detected among sessile assemblages over the surveys were a consequence of changes in the abundance of species.At PF, the SIMPER analysis revealed an average dissimilarity of 16.7% between the before and after invasion periods.The species T. stalactifera, P. perna, B. solisianus and C. bisinuatus contributed 35.9, 25.4, 13.8 and 13.7% respectively to this dissimilarity.Tetraclita stalactifera, B. solisianus and C. bisinuatus were less abundant while P. perna was more abundant after invasion than before invasion.At FN, the average dissimilarity between periods was 17.6% and the species P. perna, B. solisianus, C. bisinuatus and Megabalanus spp were the main contributors to this dissimilarity with 29.4, 23.3, 20.4 and 18.0%, respectively.Here, the abundances of B. solisianus and C. bisinuatus were lower while the abundance of P. perna and Megabalanus spp.were higher after than before invasion.The average dissimilarity between spring and summer at PF was 16.5%.Perna perna, T. stalactifera, Megabalanus spp and B. solisianus contributed with 32.1, 27.9, 13.6 and 13.5% to this distinction.Perna perna increased its abundance in summer but T. stalactifera, Megabalanus spp.and B. solisianus were more abundant in spring.

Source of variation
The percent cover of I. bicolor was negatively correlated with that of T. stalactifera (r = -0.79,p = 0.002) and bare substrate cover (r = -0.84,p = 0.001) while I. bicolor percent cover did not show any significant correlation with that of Perna perna (r = -0.27;p = 0.40) and B. solisianus (r s = 0.38; p = 0.22).

Discussion
In this study we report changes in the intertidal sessile community of rocky shores at two sites in southeastern Brazil over a period of 11 years.The most obvious difference was the inclusion of the bivalve Isognomon bicolor as a dominant component of the sessile assemblages.This bivalve has been classified as an invasive species on the Brazilian coast with the potential for a high impact on resident species (Junqueira et al. 2009).Local extinction of the most common sessile invertebrates was not observed (as yet) following the invasion, but our results suggest that I. bicolor may strongly interact with other major components by overlapping its vertical range on the shore.
Isognomon bicolor and the barnacle T. stalactifera co-dominated the mid-intertidal zone but before the invasion this barnacle was the main sessile species occupying the mid shore.Nearly matching distributions of the invasive bivalve and T. stalactifera, as well as the negative correlation between the percent cover of these two species, suggest that T. stalactifera was the species most affected by the invasion of I. bicolor.
Although the shore levels occupied by I. bicolor were similar at the two sites, there were differences in the magnitude of this invasion and on the changes of the resident sessile assemblage.On the shore where I. bicolor reached its maximum abundance, the barnacle T. stalactifera was around 70% less abundant after than before the invasion.In contrast, no changes in the abundance of T. stalactifera were evident at FN, where the invader was less abundant.At present, we cannot elucidate the mechanisms underlying the apparent negative effect on T. stalactifera populations.Recruits of this barnacle species are often observed on bare substrate (Skinner and Coutinho 2005), so one possibility is that the extensive beds formed by I. bicolor interfere with T. stalactifera larval settlement by reducing availability of bare substrate.In addition, bivalves are known to include cyprid larvae in their diet (Davenport et al. 2000); thus, I. bicolor may decrease the supply of potential settlers to the intertidal zone.Future research is needed to test whether either or both of these interactions is occurring.
Previous studies have suggested negative effects of I. bicolor on P. perna populations (Domaneschi and Martins 2002;Fernandez et al. 2004;Casarini and Henriques 2011).Isognomon bicolor has been observed to colonize the byssus threads of P. perna in high densities (Casarini and Heriques 2011), which could cause mussel detachment by strong wave action.In general, our study detected higher percent cover of Perna perna after invasion.At PF we observed a high recruitment of P. perna and the formation of an extensive mussel bed in the low intertidal zone.Increases in Perna perna cover are unlikely to be a direct effect of the I. bicolor invasion, since their abundances did not show any correlation and the invasive species was not a major component at the low shore level at the time of the survey.
The differences in I. bicolor abundances between the two sampled sites may be the consequence of differences in the stage of the invasion.Thus far, detailed information about the initial stages and the spreading direction of I. bicolor invasion is not known for our study region; therefore, it is difficult to interpret our results in this context.Moreover, site-specific factors can regulate the established populations of this invasive bivalve.For example, I. bicolor abundance and wave action were negatively associated in some invaded (Domaneschi and Martins 2002;Casarini and Henriques 2011) and native areas (Whorff et al. 1995;Malavé and Prieto 2007).However, in our study we observed higher abundances at PF, which is a moderately exposed shore.Similar to our results, Martinez (2012) found higher I. bicolor abundance on wave exposed shores in Santa Catarina, southern Brazil.Spatial variation in predation pressure on I. bicolor was observed in southeastern Brazil (López et al. 2010).Thus, the controlling mechanisms that explain the differences we observed between sites deserve further investigation with larger spatial replication.
Besides the negative effects of invasive bivalves on resident communities due to primary substrate monopolization (Branch and Steffani 2004;Rilov et al. 2004), species that form structurally complex beds can also host a diverse and abundant associated fauna (reviewed Sousa et al. 2009).For example, the diversity and density of intertidal invertebrates increased after the Mytilus galloprovincialis invasion in South Africa, because the matrix created by the mussel replaced bare rock or less complex biogenic substrates (Robinson et al. 2007).In our study, the mid intertidal shifted from a heterogeneous mosaic of barnacles and bare substrate to extensive beds of I. bicolor, with some individuals of the native barnacle T. stalactifera remaining in between.A diverse group of organisms have been found to inhabit I. bicolor beds (Fernandez et al. 2004;Ferreira and Rosso 2009), and some of these species, including the bivalve itself, recruit on this matrix (Lopez, MS pers.obs.).However, the impact of I. bicolor has not been fully investigated on the shores considered in the study.In addition, the importance of T. stalactifera as provider of habitat to other organisms has not been investigated in the study area and cannot be ruled out, since an exclusive fauna was found to be associated with both live and dead tests of Tetraclita stalactifera panamensis in the midlittoral zone of Panama (Reimer 1976).
Interestingly, the bed-forming colonization pattern of I. bicolor observed in this work, and in other studies carried out along the Brazilian coast (Breves-Ramos et al. 2010;Martinez 2012), has not been reported in its native range.Actually, ecological information of I. bicolor in the Caribbean region is scarce, and most of the published papers only report its occurrence.When referred to as an abundant component of intertidal communities, I. bicolor occupied mainly the mid shore (Almanza et al. 2004;Malavé and Prieto 2007), associated with heterogeneous substrates such as crevices (Almanza et al. 2004) and algal turfs (Whorff et al. 1995).Thus I. bicolor may act as a functionally different member of the intertidal community in its invaded range in Brazil compared to its native range.
In this study, as in others where pre-and postinvasion comparisons were attempted, we have had to tailor the post-invasion sampling design to match previous survey protocols.But even using low spatial replication, our data suggest important spatial variability in the establishment of I. bicolor beds between sites and shore heights.The conspicuous colonization of the mid shore by I. bicolor as well as the spatial overlapping and negative correlation between T. stalactifera and I. bicolor abundances strongly suggested this native barnacle as the species most affected by the invasion.However, since no continuous monitoring has taken place between the pre and post invasion survey and the lack of control sites (non-invaded areas), the extent to which the decrease in Tetraclita stalactifera abundance was merely a consequence of the I. bicolor invasion stands as an interesting subject for future studies.In this way, being aware of the limitations already mentioned, our study provided ecological information to support further long-term monitoring and experimental work to test processes associated with this recent invasion, and mechanisms that may scale up to a major potential threat to intertidal communities along the Brazilian coast.

Figure 1 .
Figure 1.The study area, situated on the coast of Rio de Janeiro (RJ) state, is represented in the larger map on the right side.Ponta da Fortaleza (PF) and Forno (FN) are the two study sites indicated with black spots.The arrow points out the site where I. bicolor was first recorded on the southeast coast of Brazil (Barequeçaba -Bb, São Paulo state).Prainha (Pr) and Ponta da Fortaleza (PF) are the two shores where the exotic bivalve had been reported at the study region before this study was carried out (Domaneschi and Martins 2002).

Figure 3 .
Figure 3. Two dimensional nMDS ordination (on square root transformed relative abundances) of the sessile invertebrate assemblages on two rocky intertidal shores (PF: Ponta da Fortaleza and FN: Forno) in southeastern Brazil.Surveys were performed before (1990-91, open round symbol) and after (2002-03, filled triangle symbol) the Isognomon bicolor invasion.Sg: Spring.Sm: Summer.The contribution of I. bicolor was excluded from the analysis.

Table 1 .
Mean (SD; n=3)relative abundance of sessile invertebrates on two rocky intertidal shores in southeastern Brazil in spring and summer of1990 -1991  and 2002 -2003, before and after Isognomon bicolor invasion, respectively., before and after Isognomon bicolor invasion, respectively.