Introduced seaweeds are responsible for severe worldwide biological invasions, with important effects on native macroalgal and animal assemblages (Piazzi et al. 2001, Buschbaum et al. 2006, Schaffelke and Hewitt 2007, McKinnon et al. 2009, Byers et al. 2010, Pacciardi et al. 2011). Effects of invasion may be particularly serious when habitat-forming species are involved, as each change in population of these organisms may have severe effects on associated assemblages (Gribben et al. 2009). Macroalgae are important habitat-forming organisms in temperate coastal systems, providing a suitable habitat for many epiphytes and mobile invertebrates (Edgar and Moore 1986, Taylor and Cole 1994, Cacabelos et al. 2010) and influencing the structure and the biodiversity of coastal systems (Tanaka and Leite 2003, Bates and Dewreede 2007, Wikström and Kautsky 2007).

In the Mediterranean Sea, species of genus Cystoseira are the most important habitat-forming species in shallow rocky bottoms (Ballesteros 1990a, b), where they play a key role in determining patterns of diversity (Sales and Ballesteros 2009). The erect structure of Cystoseira thalli, like those of other canopy species, can limit the spread of most invasive seaweeds, constituting a mechanical barrier against the invasion (Bulleri et al. 2010). However, invaders such as the Rhodophyta Lophocladia lallemandii (Montagne) F. Schmitz (Bedini et al. 2011) seem to be facilitated by the presence of Cystoseira beds. This species is widespread in tropical and subtropical waters and was probably introduced into the Mediterranean Sea through the Suez Canal (Boudouresque and Verlaque 2002). Until now, in the Mediterranean Sea, invasive events by L. lallemandii have only been described in the Balearic Islands (Patzner 1998, Cebrian and Ballesteros 2010, Marbà et al. 2014) and in the Tuscan Archipelago (Bedini et al. 2011). In both areas, the alga is able to reach high values of percentage cover and biomass (Bedini et al. 2011) on rocky bottoms and in seagrass meadows (Ballesteros et al. 2007, Sureda et al. 2008, Marbà et al. 2014). Cystoseira beds are particularly subjected to invasion (Cebrian and Ballesteros 2007, Bedini et al. 2011), as thalli of these algae may offer a valuable substrate for L. lallemandii anchoring (Bedini et al. 2011). Negative effects of L. lallemandii invasion have been described for sessile invertebrates in meadows of the seagrass Posidonia oceanica (L.) Delile (Cabanellas-Reboredo et al. 2010, Deudero et al. 2010), while no information is available about effects of invasion on mobile macro-invertebrates.

The present study aimed to compare the structure of mobile epifaunal assemblages associated with Cystoseira beds between areas invaded and not invaded by Lophocladia lallemandii. The following hypotheses were tested: i) epifaunal assemblages associated with Cystoseira beds invaded by L. lallemandii differ in species composition and abundance from those colonizing non-invaded beds, ii) temporal patterns of assemblages vary between conditions, iii) differences between conditions are greater during the period of maximum vegetative growth of L. lallemandii.

The study was carried out around the Island of Pianosa, in the Tuscan Archipelago National Park (northwestern Mediterranean Sea) at 5 m depth (Fig. 1). Lophocladia lallemandii started to spread around the island in 2008, and in 2010 it colonized with variable coverage a stretch of about 10 km between 2 and 10 m depth (Bedini et al. 2011). The alga begins to grow in July, reaches its maximum abundance in November and completely disappears between January and June (Bedini et al. 2011). All around the island, the rocky bottom between 4 m and 8 m of depth is colonized by Cystoseira spp. assemblages (mostly C. crinita Duby and C. brachycarpa var. balearica (Savageau) Giaccone). In invaded C. crinita beds, the biomass of L. lallemandii in November was about 0.2 kg dw m^–2 (Bedini et al. 2011).

Four areas of about 100 m² were selected in C. crinita beds along the southern coast of the island, two of them invaded by L. lallemandii and two non-invaded; areas were randomly chosen among those available for each condition (Fig. 1). On four dates during a one-year period (May 2010, August 2010, November 2010, May 2011), three samples were collected in each area. Samples were constituted by all organisms present within an area of 400 cm². All mobile macro-invertebrates present in each sample were identified and the abundance of each species was expressed as number of individuals m^–2. Epifaunal assemblages were analyzed by Permutational Analysis of Variances (PERMANOVA, Anderson 2001). A three-way model was used with Condition (Invaded vs. Non-invaded) as a fixed factor, Date (4 levels) as a random factor crossed with Condition and Area (2 levels) as a random factor nested in Condition. Data were log(x+1) transformed before calculation of the Bray-Curtis index of dissimilarity. The Monte-Carlo procedure was used when the number of permutations was low. A two-dimensional multidimensional scaling (n-MDS) was used for a graphical representation of results. The SIMPER routine was performed to establish which taxa most contributed to the dissimilarity between conditions and dates.

The number of taxa per sample and the abundance of organisms were detected by analyses of variance (ANOVA), with the same factors and levels used for multivariate analyses; Cochran’s C-test was utilised before each analysis to check for homogeneity of variance and data were transformed when necessary (Underwood 1997).

A total of 150 taxa were identified: 42 Polychaeta, 78 Arthropoda, 26 Mollusca and 4 Echinodermata (Table 1).

ANOVA analyses detected a significant difference among dates for the abundance of organisms (F=80.7, p=0.003) and the mean number of taxa per sample (F=20.6, p=0.001), while differences between conditions were not significant (F=2.6, p=0.120 and F=44.6, p=0.071 respectively). Both variables were higher on spring dates than on the others (Fig. 2).

PERMANOVA detected a significant interaction between Date and Condition (Table 2). The pairwise test showed that differences between conditions were significant in November 2010 but not on the other sampling dates (Table 2). In invaded condition, May 2010 and May 2011 differed from August and November 2010; in non-invaded condition, November 2010 differed from the other dates. MDS showed a greater disjunction between invaded and non-invaded assemblages in November 2010 than in the other sampling dates (Fig. 3).

The SIMPER test showed that the main differences between assemblages in November 2010 were the greater abundance of amphipods (Caprella acanthifera, Ampithoe ramondi, Dexamine spiniventris), isopods (Idotea granulosa) and polychaetes (Platynereis dumerilii) in invaded areas and the greater abundance of molluscs (Rissoa variabilis, Barleeia unifasciata) and decapods (Calcinus tubularis, Hippolyte longirostris, Cestopagurus timidus) in non-invaded areas (Table 3).

The main differences between spring sampling dates (May 2010 and May 2011) and autumn ones (November 2010) were a higher abundance of organisms in spring, especially the molluscs Barleeia unifasciata and Bittium latreillii and the amphipods Hyale schmidti, Ampithoe ramondi, Phtisica marina and Caprella spp.; only a few taxa were more abundant in autumn, including the decapods Cestopagurus timidus and Calcinus tubularis (Table 3).

Results of the study described the structure of epifaunal assemblages associated with Cystoseira crinita beds and highlighted differences between areas invaded by Lophocladia lalemandii and non-invaded areas related to the vegetative cycle of Rhodophyta.

Epifaunal assemblages associated with C. crinita were characterized by high abundance and diversity, compared with those described for other seaweed habitats (Gestoso et al. 2012, Janiak et al. 2012, Engelen et al. 2013). Macroalgal assemblages associated with Mediterranean Cystoseira beds are well known (Boudouresque 1972, Sales and Ballesteros 2010), while epifaunal assemblages have been less investigated and knowledge is limited to particular taxa (Arrontes and Anadon 1990, Chemello and Milazzo 2002, Fraschetti et al. 2002). The present study, analysing the whole epifaunal assemblages, confirms the important ecological role of Cystoseira beds in Mediterranean coastal systems. Cystoseira thalli, like those of other structurally complex algae (Tanaka and Leite 2003, Wikström and Kautsky 2007), may create a high number of microhabitats, hosting organisms with different requirements (Russo 1990, Gee and Warwick 1994, Taylor 1998). Moreover, Cystoseira beds may offer an effective refuge from predators and abundant food availability (Buschmann 1990, Holmlund et al. 1990, Martin-Smith 1993).

The sampling design of the study was not suitable for assessing the temporal dynamics of the assemblages. However, in non-invaded areas, epifaunal assemblages associated with C. crinita showed great differences between sampling dates, suggesting the occurrence of seasonal patterns which should be investigated through further studies. Seasonal variations in epifaunal assemblages associated with Cystoseira spp. as a consequence of taxa life cycles and modifications in seaweed structure have already been described (Fraschetti et al. 2002, Gozler et al. 2010). In fact, Cystoseira are perennial species with seasonal cycles of vegetative growth: they reach their maximum size in spring-summer period, while in autumn they lose secondary branches, changing their habitus (Sales and Ballesteros 2012). Temporal changes of epifaunal associated with Cystoseira spp. can also be caused by changes of macroalgal epiphyte assemblages. In fact, Cystoseira species host an abundant assemblage of macroalgae, mostly constituted by seasonal filamentous species (Ballesteros et al. 2009, Sales and Ballesteros 2010), which may change greatly throughout the year following the growth cycles of the main taxa.

The seasonal development of L. lallemandii overlaps this scenario. In fact, the study results showed that epifaunal assemblages associated with Cystoseira crinita beds differed between areas invaded and not invaded by Lophocladia lallemandii in November, when the invasive species reached maximum values of cover and biomass (Bedini et al. 2011), while assemblages showed no differences between conditions in other periods of the year.

The main effects of the presence of L. lallemandii were an increase in amphipods and polychaetes and a decrease in decapods and molluscs. Species more linked to the architecture of Cystoseira thalli may be damaged by substrate modification; in fact, many epifaunal organisms are related to particular macroalgae and may be strongly influenced by the presence of invasive species (Viejo 1999, Gestoso et al. 2010). On the other hand, polychaetes are not specifically facilitated by the morphology of canopy seaweeds, and food preference and/or different substrate requirements may well cause their increase in invaded areas; in fact, several polychaete species may be facilitated by turfs created by L. lallemandii, where sediment and organic matter could be trapped. Caprellid amphipods need cylindrical substrates with a small diameter to be encircled by pereopods in order to avoid being dislodged by water movements (Aoki and Kikuchi 1990), and the presence of L. lallemandii can increase the substrate available for anchoring. Moreover, herbivore amphipods, ampithoids in particular, may also be influenced by the increase in food availability in invaded areas (Duffy 1990, Duffy and Hay 2000, Poore et al. 2008).

The results show that the effects of L. lallemandii colonization on mobile organisms are related more to changes in species composition than to changes in patterns of diversity. This finding agrees with those described for other introduced seaweeds, suggesting that, while macroalgal invasions strongly affect diversity of sessile assemblages (Ribera and Boudouresque 1995, Piazzi et al. 2001, Schaffelke and Hewitt 2007, Baldacconi and Corriero 2009, Zuljevic and Nikolic 2008), the effects of invasions on mobile organisms are more related to changes in the structure of assemblages (Vázquez-Luis et al. 2009, Gestoso et al. 2012, Janiak et al. 2012, Pacciardi et al. 2011, Engelen et al. 2013).

Differences between invaded and non-invaded beds were not evident five months after the disappearance of L. lallemandii. The effects of invasion on Cystoseira-associated assemblages seem to be limited to the period of vegetative growth of the alga and reversible during the period of its vegetative rest. Recovery of assemblages could be related both to the intrinsic characteristics of organisms and to the lack of damage to C. crinita thalli. Macro-invertebrate assemblages are able to respond rapidly to various kinds of impacts (Teixeira et al. 2009), but they are also able to recover their original structure quickly after disturbance because of the short life cycles of the organisms (Lu and Shio-Sun Wu 2007, Pacciardi et al. 2011). Moreover, recovery followed the return of the habitat to its original structure. In fact, until now, no evidence of Cystoseira regression has been observed in invaded areas of Pianosa Island (Bedini et al. 2011). Although L. lallemandii completely cover Cystoseira thalli during the period of spread, several months without the invasive alga seem to be enough to avoid severe damage to Cystoseira beds.

The effects of L. lallemandii invasion at Pianosa Island seem to be less severe than those described in the Balearic Islands. However, the colonization of L. lallemandii in the Tuscan Archipelago has recently started and more severe effects could be hypothesized after longer periods of colonization on both Cystoseira beds and its associated assemblages, indicating the importance of monitoring the spread of the invasive alga.