INTRODUCTION
The preservation of species and the restoration of natural habitats are urgently needed measures to fight against species and habitat degradation caused by anthropogenic impacts. Invasive species have played a major role in global change and biodiversity loss in the 20th Century and early 21st Century (Soulé, Reference Soulé1990; Gray, Reference Gray1997; Vitousek et al., Reference Vitousek, D'Antonio, Loope, Rejmánek and Westbrooks1997; Sala et al., Reference Sala, Chapin, Armesto, Berlow, Bloomfield, Dirzo, Huber-Sanwald, Huenneke, Jackson, Kinzig, Leemans, Lodge, Mooney, Oesterheld, Poff, Sykes, Walker, Walker and Wall2000). Invasive species are capable of changing species composition, abundance and diversity, thereby profoundly altering host ecosystems. At a more general level, invasive species are capable of changing important ecosystem functions such as nutrient cycling and primary productivity, hydrology, disturbance regimes and trophic structures. In all cases, the phenomenon of species introductions represents an irreversible ecosystem change (Vitousek et al., Reference Vitousek, D'Antonio, Loope, Rejmánek and Westbrooks1997). Invasive species are characterized by a high capacity for spread and a negative ecological and/or economic impact (Alpert et al., Reference Alpert, Bone and Holzapfel2000; Boudouresque & Verlaque, Reference Boudouresque and Verlaque2002).
Apart from the abundant literature on descriptive studies of invasion impacts, there is a considerable lack of manipulative experimental evidence (Levine et al., Reference Levine, Vilà, D'Antonio, Dukes, Grigulis and Lavorel2003; Hulme & Bremner, Reference Hulme and Bremner2006). In order to assess the impact of an invasive species, invaded assemblages are commonly compared to non-invaded ones (Levine et al., Reference Levine, Vilà, D'Antonio, Dukes, Grigulis and Lavorel2003). However, due to the characteristics of invasive species that tend to monopolize large extents of substrate (Alpert et al., Reference Alpert, Bone and Holzapfel2000; Richardson et al., Reference Richardson, Pyšek, Rejmánek, Barbour, Panetta and West2000; Boudouresque & Verlaque, Reference Boudouresque and Verlaque2002), it is difficult to find non-invaded and invaded assemblages in the same study site and often the two assemblages are situated in different sites. Consequently, invasive species removal or addition may be an interesting trail to follow (Zavaleta et al., Reference Zavaleta, Hobbs and Mooney2001).
Invasive species removal is also performed in the frame of eradication and control campaigns in order to restore the native flora and fauna (Myers et al., Reference Myers, Simberloff, Kuris and Carey2000; Zavaleta et al., Reference Zavaleta, Hobbs and Mooney2001; Simberloff et al., Reference Simberloff, Parker and Windle2005). After removal of the invasive species the system is assumed to return to its state before invasion. In the case of a plant assemblage, the recovery is influenced by propagule pressure which itself is dependent on: (1) presence of a seed bank; (2) distance to propagule sources; and (3) propagule availability (period of the year and seasonal cycles of species). Thus the time for an assemblage to recover to its non-invaded state may greatly vary.
Due to its geographical position at the intersection between the Atlantic and Indo-Pacific Oceans, the opening of the Suez Canal, intensive aquaculture and maritime traffic, the Mediterranean Sea is a worldwide hotspot of marine macrophyte introductions (Ribera Siguan, Reference Ribera Siguan, Ruiz and Carlton2003; Boudouresque & Verlaque, Reference Boudouresque and Verlaque2005). Nearly 600 plant and animal species, representing 5% of the known flora and fauna, are estimated to have been introduced into the Mediterranean Sea (Boudouresque & Verlaque, Reference Boudouresque and Verlaque2005; Boudouresque et al., Reference Boudouresque, Ruitton, Verlaque, Velikova and Chipev2005; Zenetos et al., Reference Zenetos, Çinar, Pancucci-Papadopoulou, Harmelin, Furnari, Andaloro, Bellou, Streftaris and Zibrowius2005). Since 1990, the introduced Caulerpa racemosa var. cylindracea (Sonder) Verlaque, Huisman & Boudouresque (hereafter Caulerpa racemosa) has invaded the coasts of 12 Mediterranean countries at an almost unprecedented rate of spread (Verlaque et al., Reference Verlaque, Boudouresque, Meinesz and Gravez2000, Reference Verlaque, Afonso-Carillo, Gil-Rodríguez, Durand, Boudouresque and Le Parco2004; Piazzi et al., Reference Piazzi, Meinesz, Verlaque, Akçali, Antolić, Argyrou, Balata, Ballesteros, Calvo, Cinelli, Cirik, Cossu, D'Archino, Djellouli, Javel, Lanfranco, Mifsud, Pala, Panayotidis, Peirano, Pergent, Petrocelli, Ruitton, Žuljević and Ceccherelli2005; Klein & Verlaque, Reference Klein and Verlaque2008). With its high rate of spread—colonization of nearly the entire Mediterranean Sea within 17 years—and its capability to alter benthic assemblages, Caulerpa racemosa may be considered as one of the most problematic invasive species in the Mediterranean Sea (Klein & Verlaque, Reference Klein and Verlaque2008).
Different approaches have been explored to investigate the invasion impact of this Australian species in the Mediterranean Sea. Several studies have attempted to evaluate the impact by comparing the flora and fauna associated with Caulerpa racemosa meadows to other Mediterranean rhizophytic macrophytes, such as the native Caulerpa prolifera (Forsskål) J.V. Lamouroux and seagrasses (Di Martino & Giaccone, Reference Di Martino and Giaccone1996; Buia et al., Reference Buia, Gambi, Terlizzi, Mazzella, Gravez, Ruitton, Boudouresque, Le Direac'h, Meinesz, Scabbia and Verlaque2001). Yet, the largest part of C. racemosa impact studies has dealt with comparisons between C. racemosa meadows and non-invaded Posidonia oceanica (Linnaeus) Delile dead ‘matte’ or hard substrata (rock and coralligenous assemblages) (Piazzi et al., Reference Piazzi, Ceccherelli and Cinelli2001, Reference Piazzi, Balata, Cecchi and Cinelli2003, Reference Piazzi, Meinesz, Verlaque, Akçali, Antolić, Argyrou, Balata, Ballesteros, Calvo, Cinelli, Cirik, Cossu, D'Archino, Djellouli, Javel, Lanfranco, Mifsud, Pala, Panayotidis, Peirano, Pergent, Petrocelli, Ruitton, Žuljević and Ceccherelli2005, Reference Piazzi, Balata and Cinelli2007; Carriglio et al., Reference Carriglio, Sandulli, Deastis, Gallo d'Addabbo and Grimaldi de Zio2003; Balata et al., Reference Balata, Piazzi and Cinelli2004; Casu et al., Reference Casu, Ceccherelli, Palomba, Curini-Gelletti and Castelli2005; Klein & Verlaque, Reference Klein and Verlaque2009a, Reference Klein and Verlaqueb).
The aim of the present study was to experimentally test the effect of C. racemosa on Mediterranean macrophyte assemblages. Therefore, a manual removal experiment has been carried out over a period of 18 months. A comparison was made between three macrophyte assemblages on dead Posidonia oceanica ‘matte’: (1) an assemblage invaded by C. racemosa (Invaded); (2) an assemblage where C. racemosa has been removed and excluded for 18 months (Removed); and (3) a non-invaded assemblage (Non-invaded).
MATERIALS AND METHODS
Area description
The present study was carried out in the bay of Marseilles (France, north-west Mediterranean Sea) (43°13′44″N 5°20′ 41″E) where Caulerpa racemosa has been present since 1997 (Jousson et al., Reference Jousson, Pawlowski, Zaninetti, Meinesz and Boudouresque1998). Caulerpa racemosa forms meadows exhibiting a marked seasonal development pattern with a maximum in late summer with 100% cover and highest biomass (September 2002: 82 ± 3 g DW m−2) and a minimum in winter (Ruitton et al., Reference Ruitton, Verlaque and Boudouresque2005).
Experimental procedures
Three different assemblages were considered: Invaded, Removed and Control. All assemblages were situated at 17 m depth on dead Posidonia oceanica ‘matte’Footnote 1.
For the Removed assemblage, Caulerpa racemosa was removed in August 2003 from five randomly positioned 20 cm × 20 cm plots. This manual removal was carried out very delicately, by pulling out all small fragments visible to the naked eye within the plot and in a margin around the plots without removing the other macrophytes. Then C. racemosa was excluded for 18 months until March 2005. In March 2005, the five Removed plots and five randomly positioned 20 cm × 20 cm plots were sampled both in the Invaded and the Control assemblages. Sampling was carried out by SCUBA diving. Samples were preserved in 4% formaldehyde–seawater until further treatment.
As a consequence of the miniaturization of the Mediterranean flora, the qualitative minimum sampling area is very small compared to other regions of the world (Coppejans, Reference Coppejans, Price, Irvine and Farnham1980). This area was determined for various subtidal and intertidal Mediterranean macrophyte assemblages and was always found to be comprised between 64 cm2 and 200 cm2 (Boudouresque, Reference Boudouresque1974; Cinelli et al., Reference Cinelli, Boudouresque, Fresi, Marcot and Mazzella1977a, Reference Cinelli, Fresi, Idato and Mazzella1977b; Boudouresque & Belsher, Reference Boudouresque and Belsher1979a, Reference Boudouresque and Belsher1979b; Coppejans, Reference Coppejans, Price, Irvine and Farnham1980). The size of the sampling plots used in the present study was fixed at 400 cm2 (20 cm × 20 cm) which is largely representative.
Eighteen months of removal/exclusion have been chosen, because a preliminary experimental study on the macrophyte assemblages on dead Posidonia oceanica ‘matte’ of the Port-Cros National Park showed recovery after complete destruction within 12 months (P. Helman, personal communication).
Sample analysis
An exhaustive macrophyte species list (Chlorophyta, Ochrophyta (Heterokontophyta), Rhodophyta) was drawn up for each plot, by analysing the entire sample under a binocular microscope and light microscope. For all species the taxonomic nomenclature of the database AlgaeBase (Guiry & Guiry, Reference Guiry and Guiry2009) was used. Introduced species were identified using the checklists of Boudouresque & Verlaque (Reference Boudouresque and Verlaque2002) and Wallentinus (Reference Wallentinus, Leppäkoski, Gollasch and Olenin2002). Cover was measured for each species, by spreading out the thalli and measuring the surface covered (Boudouresque, Reference Boudouresque1971; Ballesteros, Reference Ballesteros1984). The dry weight of Caulerpa racemosa and other macrophytes was quantified for each plot by drying the macrophytes in an oven at 60°C until constant weight. Each macrophyte species was classified in one of the following functional form groups (Littler & Littler, Reference Littler and Littler1980; Steneck & Dethier, Reference Steneck and Dethier1994): articulated calcareous (= jointed calcareous), encrusting, epiphytes (very small microscopic epiphytes < 1 mm high), filamentous (filamentous and thin sheets/foliose) and erect (more complex structure or bigger than filamentous).
Data analysis
Normality of data was evaluated using the Kolmogorov–Smirnov test. If data were normally distributed, a test of homogeneity of variances (Bartlett's test) and a 1-way ANOVA were carried out. Significant differences between assemblages were detected by the Student–Newman–Keuls (SNK) test. An analysis of similarities (ANOSIM) was carried out on the similarity matrix (Bray–Curtis similarity) calculated from macrophyte species cover for each sample. A non-metric multidimensional scaling (nMDS) plot was constructed using as input data macrophyte species cover (including Caulerpa racemosa) for each sample.
For the ANOSIM and the nMDS plot the program Primer 6 has been used and the program Statistica 6.1 StatSoft® was used for all other statistical analyses.
RESULTS
Caulerpa racemosa cover and biomass
In the Non-invaded assemblage, Caulerpa racemosa was found with insignificant cover (0.88% of the cover of the plots) and biomass (Figure 1). In the Removed assemblage, during the 18 months of the experiment re-growth by creeping axes originating from the adjacent meadows was occasionally observed only in the margins around the plots. Moreover, few tiny fragments were occasionally found inside the plots. They may originate from microscopic C. racemosa filaments remaining in the ‘matte’ after removal, propagules or sexual reproduction from the surrounding meadows. Finally, in the Invaded assemblage, biomass of C. racemosa was 1.27 ± 0.22 g DW plot−1 (Figure 1) corresponding to a mean cover of 133.20 ± 25.89 cm2 plot−1 (mean ± SE), which represented up to 90% of the total macrophyte cover.
Fig. 1. Caulerpa racemosa dry weight (g DW plot−1) (mean ± SE) in March 2005. NI, Non-invaded; R, Removed; I, Invaded. Values not different at α = 5% are denoted with the same letter (Student–Newman–Keuls test).
Species richness, cover and biomass
Overall 124 taxa were identified, among them 85 Rhodophyta, 24 Chlorophyta and 15 Ochrophyta (Appendix). For the Non-invaded assemblage, significantly highest number of species (mean ± SE: 72 ± 2 species per plot; P < 0.001) and highest macrophyte biomass (1.42 ± 0.02 g DW plot−1; P < 0.001) (Figure 2A & B) were recorded. The highest value for the Shannon diversity index (H′ = 4.89) and high Pielou's evenness (mean = 0.62 ± 0.07) were found in samples of the Non-invaded assemblage (Table 1). Chlorophyta and Rhodophyta dominated (Table 2) and the erect perennial species Flabellia petiolata (Turra) Nizamuddin represented the major component.
Fig. 2. (A) Macrophyte species number and (B) dry weight (g DW plot−1) excluding Caulerpa racemosa (mean ± SE). NI, Non-invaded; R, Removed; I, Invaded. Values not different at α = 5% are denoted with the same letter (Student–Newman–Keuls test).
Table 1. Shannon diversity and evenness indices of the Non-invaded (NI), Removed (R) and Invaded (I) assemblage.
Table 2. Cover (cm2 plot−1) of the different macrophyte phyla and functional form groups of the Non-invaded (NI), Removed (R) and Invaded (I) assemblage (mean ± SE).
Of all assemblages, the lowest number of species (53 ± 3 species per plot), significantly lowest macrophyte biomass (0.20 ± 0.07 g DW plot−1; P < 0.001) (Figure 2A & B), the lowest values for H′ (maximum 1.85) and significantly lowest evenness (0.21 ± 0.03; P < 0.001) (Table 1) were found in the Invaded assemblage. No dominant algal phylum could be discerned. Compared to the Non-invaded assemblage, the abundance of erect and filamentous species was strongly reduced (Table 2) and a number of species were completely excluded (Appendix) in the Invaded assemblage. The other introduced species significantly decreased in cover (P = 0.01) (particularly ‘Falkenbergia rufolanosa (Harvey) F. Schmitz’ phase and Womersleyella setacea (Hollenberg) R.E. Norris) and several species were completely lacking (Acrothamnion preissii (Sonder) E.M. Wollaston, Antithamnion amphigeneum A.J.K. Millar and Asparagopsis armata Harvey) (Figure 3).
Fig. 3. Introduced species cover (cm2 plot−1) excluding Caulerpa racemosa (mean ± SE). NI, Non-invaded; R, Removed; I, Invaded. Values not different at α = 5% are denoted with the same letter (Student–Newman–Keuls test).
In the Removed assemblage, species richness remained significantly lower than in the Non-invaded assemblage (SNK test: P < 0.01) without a significant increase compared to the Invaded assemblage (60 ± 2 species per plot) (Figure 2A). On the other hand, macrophyte biomass significantly increased while still remaining lower than in the Non-invaded assemblage (0.53 ± 0.09 g DW plot−1; SNK test R-I: P < 0.05; R-NI: P < 0.01) (Figure 2B). The filamentous group clearly dominated whereas perennial erect species had not yet recovered (Table 2). No recruitment of the perennial erect species Flabellia petiolata occurred, a species which dominated in the Non-invaded assemblage. Rhodophyta dominated in terms of number of species and cover (Table 2). In contrast to native macrophytes, the total cover of other introduced species reached a level comparable to the Non-invaded assemblage (Figure 3).
The ANOSIM carried out pairwise on the three assemblages, with species cover (including C. racemosa) as input data, showed high R values for the Non-invaded and Invaded (0.992), Removed and Invaded assemblages (0.996) and lower values for the Non-invaded and Removed assemblages (0.352). These results suggest that in the Invaded assemblage the number of species and their cover were different from the other two. The Non-invaded and the Removed assemblages showed more similarity.
The nMDS plot showed a clear opposition of the Invaded assemblage dominated by C. racemosa at one end and Removed and Non-invaded assemblages at the other end (Figure 4). Inside the cluster of Non-invaded and Removed assemblages a change can be seen from a Non-invaded assemblage associated with the perennial species Flabellia petiolata, Codium bursa (Olivi) C. Agardh and Phyllophora sicula (Kützing) Guiry & L.M. Irvine to the Removed assemblage associated with the filamentous species Womersleyella setacea and Cladophora laetevirens (Dillwyn) Kützing.
Fig. 4. Non-metric multidimensional scaling plot of the macrophyte assemblage in the Non-invaded (•), Removed (■) and Invaded assemblage (▴).
DISCUSSION
In the Mediterranean Sea the invasive Caulerpa racemosa mainly colonizes dead Posidonia oceanica ‘matte’ and deep detritic assemblages (Klein & Verlaque, Reference Klein and Verlaque2008, Reference Klein and Verlaque2009a). Furthermore, C. racemosa is commonly found in urbanized areas near large cities and ports (Athens, Genoa, Livorno, Taranto, Toulon, etc) (Klein & Verlaque, Reference Klein and Verlaque2008). Therefore, the bay of Marseilles can be considered as representative of many other invaded areas in the Mediterranean Sea.
Non-invaded dead ‘matte’ assemblage
At Marseilles, in the Non-invaded assemblage, the macrophyte assemblage on dead Posidonia oceanica ‘matte’ were similar to the macrophyte assemblages studied in Italy and south-eastern France (Augier & Boudouresque, Reference Augier and Boudouresque1967; Cinelli & Salghetti-Drioli, Reference Cinelli and Salghetti-Drioli1983–1984; Salghetti-Drioli et al., Reference Salghetti-Drioli, Pardi and Cinelli1985, Reference Salghetti-Drioli, Pardi, Cinelli, Della Pieta, Boudouresque, Meinesz, Fresi and Gravez1989; Piazzi et al., Reference Piazzi, Ceccherelli and Cinelli2001). The samples were typical of the cold season in the north-western Mediterranean Sea with the dominance of Rhodophyta and Chlorophyta. In contrast with previous studies, species richness was high, and species composition differed with regard to the dominant species, probably due to greater depth and colder annual temperatures. The nearly absence of articulated calcareous and encrusting species was a general feature common to all the samples.
Caulerpa racemosa invasion impact
In contrast with previous years, when Caulerpa racemosa showed a seasonal development with a temporal complete withdrawal in winter (Ruitton et al., Reference Ruitton, Verlaque and Boudouresque2005) an important C. racemosa cover was present in March 2005. This tendency of C. racemosa to becoming a permanent part of the macrophyte assemblage in the bay of Marseilles has been confirmed in subsequent years (2006 and 2007; personal observation), indicating a possible acclimation to local conditions. The C. racemosa biomass was comparable to the biomass found during the same season in the warmer region of the French Riviera (Villefranche-sur-mer) (Capiomont et al., Reference Capiomont, Breugnot, Den Haan and Meinesz2005).
The negative effects caused by the invasion by C. racemosa observed in the present study were consistent with previous descriptive studies conducted in Italy and the bay of Marseilles (France). Macrophyte assemblages of dead Posidonia oceanica ‘matte’ showed reduced species number and abundance in the presence of C. racemosa (Piazzi et al., Reference Piazzi, Ceccherelli and Cinelli2001; Piazzi & Cinelli, Reference Piazzi and Cinelli2003; Balata et al., Reference Balata, Piazzi and Cinelli2004; Klein & Verlaque, Reference Klein and Verlaque2009b). Although C. racemosa biomass was more than five times lower at Marseilles than in Italy (Piazzi et al., Reference Piazzi, Ceccherelli and Cinelli2001), it is important to notice that the invasion effect was equally drastic.
Decreased species diversity through competitive exclusion is a widespread effect of invasive species (Morrison, Reference Morrison2000; Brown et al., Reference Brown, Mitchell and Graham2002; Branch & Steffani, Reference Branch and Steffani2004; Bando, Reference Bando2006). Thus the decrease in macrophyte species numbers and abundance in the assemblage invaded by C. racemosa may be explained by C. racemosa's competitive superiority. Caulerpa racemosa is also known to contain various secondary metabolites that are supposed to play a role in herbivore deterrence and may be in competition (Klein & Verlaque, Reference Klein and Verlaque2008).
In the Invaded assemblage, the reduction in biomass (‘Falkenbergia rufolanosa’ phase, Womersleyella setacea) or absence (Acrothamnion preissii, Antithamnion amphigeneum and Asparagopsis armata) of other introduced species found in the Non-invaded assemblage suggests that C. racemosa was competitively superior not only to the native species but also to the other introduced macrophytes present in the bay of Marseilles. Similar results have been found in Italy, where the abundance of the invasive species Womersleyella setacea was reduced in the presence of C. racemosa (Piazzi & Cinelli, Reference Piazzi and Cinelli2003).
Effect of Caulerpa racemosa removal
After 18 months of manually removing and excluding Caulerpa racemosa, the macrophyte assemblage of dead P. oceanica ‘matte’ was still significantly different from the Non-invaded assemblage, indicating that this time period was not long enough.
A similar study of removal was carried out by Piazzi & Ceccherelli (Reference Piazzi and Ceccherelli2006) over one year on rocky substrate. The results showed that in the Invaded assemblage, the erect and filamentous group was little affected by C. racemosa in contrast to the encrusting functional form group that was eliminated. After one year of removal, the erect and filamentous group recovered completely, whereas no sign of recovery was detected for the encrusting group and the levels of the Non-invaded assemblage had not been reached in terms of percentage cover and species number.
In the Removed plots of the present study, the development of other invasive species was more facilitated by the removal of C. racemosa than Mediterranean species. This may have important consequences for the subsequent development of the assemblage. Similar results have been found in some cases in terrestrial and fresh water environments where increased abundance of other introduced species was observed after removal or control of certain invasive species, a phenomenon called the Sisyphus-effect (Mack & Lonsdale, Reference Mack, Lonsdale, Veitch and Clout2002).
On the one hand, the relatively high species richness and partial recovery of the assemblage observed in the Removed plots showed the development potential of the macrophyte assemblage after removal of C. racemosa. On the other hand, the removal favoured first the development of other invasive species, in particular the filamentous Rhodophyta Womersleyella setacea, which recovered faster than the native species and become dominant in the assemblage.
ACKNOWLEDGEMENTS
The authors would like to acknowledge the field assistance of Sandrine Ruitton and thank Charles-François Boudouresque who provided helpful comments on an earlier draft. We also thank Michael Paul for reviewing the English text and Michèle Boudouresque for assistance with the literature. Funding was provided by grants from the Agence de l'Eau Rhône-Méditerranée-Corse, the Conseil Général 13 and the Ville de Marseille to J.C.K. We thank two anonymous referees for helpful comments on the manuscript.
Appendix. Presence (+) and absence (–) of each macrophyte species in the Non-invaded, Removed and Invaded assemblage. Phyla: C, Chlorophyta; O, Ochrophyta; R, Rhodophyta.
