The Invasive Seaweed Asparagopsis Taxiformis Erodes the Primary Productivity and Biodiversity of Native Algal Forests in the Mediterranean Sea

F. Paolo Mancusoa (  francesco.mancuso@unipa.it ) University of Palermo https://orcid.org/0000-0001-7217-5131 Riccardo D’Agostaro University of Palermo Marco Milazzo University of Palermo Fabio Badalamenti Institute of Anthropic Impacts and Sustainability in Marine Environment (CNR-IAS) Luigi Musco University of Salento Barbara Mikac University of Bologna Sabrina Lo Brutto University of Palermo Renato Chemello University of Palermo

e.g. soft bottoms) the introduction of non-native seaweeds enhances structural complexity that may favor the increase of biodiversity and food web length (Dijkstra et al., 2017). Conversely, the introduction in well-structured habitats (e.g. seagrass meadows, algal canopies) may alter the diversity and functioning, depending on the structural features of the recipient habitat (Engelen et al., 2013; Veiga et al., 2014Veiga et al., , 2018. It is interesting to observe that the same invasive species can determine opposite effects. For example, Veiga et al. (2018) found that the invasive Sargassum muticum (Yendo) Fensholt hosted a low diverse epifaunal assemblage compared to the native Sargassum avifolium Kützing. These results were in contrast with previous studies which suggested only weak to no impact of S. muticum on native faunal diversity (Wernberg et    In this study, we evaluated the consequences of the habitat shift from the native Ericaria brachycarpa (J.Agardth) Orellana & Sansón to the invasive A. taxiformis analyzing the epifaunal community associated with three plausible alternative states of the transition between native to invasive seaweeds habitats. In particular, we characterized and compared the biomass and the diversity (richness, evenness, structure and composition) of the epifauna associated with the fronds of homogenous and mixed stands of E. brachycarpa and A. taxiformis. Moreover, we explored the variation of the epifaunal diversity in relation to the structural features of the two algae (dry weight, thallus volume, canopy volume and interstitial volume). In this study, we evaluated the consequences of the habitat shift from the native Ericaria brachycarpa (J.Agardth) Orellana & Sansón to the invasive A. taxiformis analyzing the epifaunal community associated with three plausible alternative states of the transition between native to invasive seaweeds habitats. In particular, we characterized and compared the biomass and the diversity (richness, evenness, structure and composition) of the epifauna associated with the fronds of homogenous and mixed stands of E. brachycarpa and A. taxiformis. Moreover, we explored the variation of the epifaunal diversity in relation to the structural features of the two algae (dry weight, thallus volume, canopy volume and interstitial volume).

Materials And Methods
Loading [MathJax]/jax/output/CommonHTML/fonts/TeX/fontdata.js Study area and algal species characteristics The research was performed on the southwestern, shallow rocky shore of the Favignana Island (Sicily, Italy), within the Egadi Islands Marine Protected Area (MPA) in June 2011 (Fig. 1). The region consists of gently sloping (5°-10°) carbonate rocky platforms and scattered boulders (Pepe et al., 2018) that provide substrates for well-developed macroalgal vegetation.
In this area A. taxiformis was rst recorded in 2000 ( Barone et al., 2003). Since then, no studies have been explored the temporal effects of this invasive species on native habitats. Although today A. taxiformis is well established in the area, previous surveys allowed the identi cation of three sites with distinctive habitats corresponding to three possible alternative states of the transition from native to invasive seaweed habitats: "Scoglio Corrente" (37° 55' 2.0778" N, 12° 17' 6.0432" E) characterized by stands of E. brachycarpa (100% coverage); "Scoglio Palumbo" (37° 55' 10.4226" N, 12° 18' 41.097" E) hosting stands of A. taxiformis (100% coverage), and "Cala Grande" (37° 55' 35.385'' N, 12° 16' 39.514'' E) with mixed stands of E. brachycarpa (~ 50% coverage) and A. taxiformis (~ 50% coverage) (Fig. 1). In this study, we decided to use these three sites to highlight changes in the epifaunal communities associated to seaweeds caused by a shift from native E. brachycarpa to the invasive A. taxiformis.
Ericaria brachycarpa is a brown seaweed (Fucales) characterized by caespitosus thalli up to 20-25 cm in height with several perennial axes, up to 2-6 cm in height, connected to the substratum by a more or less compact discoid base formed by haptera (Molinari Nova and Gury, 2020 Samples were collected by scuba diving at a depth of 5-7 m. For each site (hereafter referred as habitat), two areas (5 x 5 m) were haphazardly selected. For each area, 10 thalli of E. brachycarpa from homogenous stands (100% algal coverage), 10 thalli of E. brachycarpa form mixed stands and 10 gametophytes of A. taxiformis from homogenous stands (100% algal coverage) were collected (n = 20 per habitat). Underwater, each thallus and the associated epifauna were enveloped with a plastic bag, then the alga was detached from the substrate and the plastic bag was immediately closed to prevent the escape of vagile fauna. After collection, each sample was carefully drained from seawater in order to prevent escape of small epifauna and stored at -20°C until laboratory analysis. In the laboratory, each thallus of E. brachycarpa and gametophytes of A. taxiformis were transferred into buckets lled with tap water and shaken vigorously, allowing the associated fauna to detach from the algae. Then, the water was sieved through a 1 mm mesh. After sorting, molluscs, amphipods and annelids were stored in 70% seawater ethanol solution and subsequently counted and identi ed to species, or nearest possible taxonomic level. Taxonomy and nomenclature were updated according to the World Register of Marine Species database (T Horton et al., 2021).

Seaweeds structural attributes
For each thallus of E. brachycarpa and gametophyte of A. taxiformis collected, we measured 4 structural features (thallus volume, canopy volume, interstitial volume, and biomass), to explore their relationships with the diversity indices calculated for the epifaunal assemblages. Thallus volume (measured as the variation of volume, in ml, after the immersion of a thallus into a graduate cylinder lled with seawater); canopy volume (the volume, in ml, created by the overall dimension of a thallus submerged in seawater) and interstitial volume (the volume, in ml, of water among the fronds of the alga) were estimated according to Hacker and Steneck (1990). The canopy volume was de ned as the volume of a theoretical cylinder (CV = π × r 2 × h), where π = 3.14, h is the length of thallus from the base to the apical portion of the frond, including epiphytes, and r is the radius calculated as averaged measure of the radius of the thallus measured with a ruler (+/-1 mm) at the apical, median and basal parts. The interstitial volume (IV) was obtained by subtracting the thallus volume (TV), and the axis volume (caV, estimate as the volume of cylinder obtained measuring the height and the radius of the perennial axis) from the canopy volume CV(IV = (CV − TV) − caV).
Finally, the biomass of the macroalgae was calculated as dry weight (DW, gr) after drying in stove at 60°C for 48 h (Stein-Taylor et al., 1985). Biomass was used as a proxy of primary production of each habitat.

Data analysis
For each epifaunal species, we calculated total abundance (N), Frequency (F%; the percentage of samples in which a particular species is present) and Dominance index (D%; the percentage of the rate between the percentage of individuals of a particular species and the total number of individuals within the sample) (Magurran, 1988). The epifaunal assemblages of each habitat were characterized according to total abundance of individuals (N), total number of species (S), Shannon-Wiener diversity index (H') and Loading [MathJax]/jax/output/CommonHTML/fonts/TeX/fontdata.js Pielou's Evenness index (J). A two-way analysis of variance (ANOVA) was used to test differences in the epifaunal indices (N, S, H', J) between habitats ( xed and orthogonal with 3 levels: E. brachycarpa, E. brachycarpa in mixed stands and A. taxiformis) and area (random and nested within habitat with 2 levels: area 1 and area 2). Cochran's test was used to check for the homogeneity of variances (Underwood 1997). Tukey's HSD procedure was used to separate means (at α = 0.05) following signi cant effects in the ANOVAs (Underwood, 1996). The hierarchical structure of the taxonomic classi cations of the epifaunal assemblages of E. brachycarpa, E. brachycarpa in mixed stands and A. taxiformis was visualized using the "heat_tree" function in the "Metacoder" R-package (Foster et al., 2017).
SIMPER analysis (Clarke, 1993) was performed to identify those taxa that contributed to the dissimilarity of the epifaunal assemblages between habitats (δi%). The ratio δi/SD (δi) was used to measure the consistency of the contribution of a particular taxon to the average dissimilarity in the comparison between habitats. A cut-off value of 70% was used to exclude low contributions.
Differences in the epifaunal community structure (which takes into account species identity and relative abundance) and composition (presence/absence, which only takes into account species identity) among habitats and areas were assessed by Permutational Multivariate Analysis of Variance (PERMANOVA).
The analyses were based on a Bray-Curtis distance matrix of square-root transformed epifaunal abundances (structure) and on Jaccard distances matrix of presence/absence data (composition) using 9999 permutations. A principal coordinate analysis (PCoA) plot was generated to visualize the variation of the epifaunal community structure (based on a Bray-Curtis distance matrix) and composition (based on Jaccard distance matrix).
Differences in each of the structural attributes (CV, IV, TV, DW) among habitats and areas were analyzed by two-way ANOVAs according to the above mentioned design. Cochran's test was used to check for the homogeneity of variances (Underwood, 1996).
Linear regression (LM) analysis was used to test which algal structural attributes explained better the variation of total abundance (N), species richness (S), Shannon-Wiener diversity (H') and Pielou's Evenness (J) of the epifaunal assemblages. In addition, a distance-based redundancy analysis (dbRDA, Legendre and Anderson, 1999) was used to investigate the relationship between structural attributes and the epifaunal multivariate structure. Since dbRDA is susceptible to multicollinearity (i.e. high correlation between environmental variables), draftsman plots were used to verify skewness or identify clear correlations between structural attributes. A log(x + 1) transformation was used to correct right-skewness of thallus volume (TV) and biomass (DW). Moreover, due to the high correlation between canopy volume (CV) and interstitial volume (IV) we removed CV from the subsequent analyses. Then, the structural attributes were normalised using a z-score transformation due to their varying measurement scales. Finally, forward selection was used to identify the structural properties that mostly contributed to the heterogeneity in the multivariate structure of the epifaunal assemblages.
Total abundance (N) and species richness (S), differed signi cantly among habitats with values that were higher in E. brachycarpa compared to E. brachycarpa in mixed stands and A. taxiformis (Fig. 3, Table S3). Shannon-Wiener diversity (H') was similar between E. brachycarpa and E. brachycarpa in mixed stands, and signi cantly higher compared to A. taxiformis. Conversely, Pielou's evenness (J) was higher in A. taxiformis compared to the other two habitats, which showed comparable values (Fig. 3, Table S3).
PERMANOVA showed that the structure and composition of the epifaunal assemblages differed signi cantly among habitats (Table S4). PERMDISP analysis revealed a high dispersion of samples within habitats, especially for E. brachycarpa in mixed stands and A. taxiformis (Fig. 4). Notwithstanding this high dispersion, the epifaunal assemblages of the three habitats were clearly separated as showed by the PCoA ordination plot (Fig. 4). The proportion of variance explained by the rst two axes was 62.8% for structure and 45.6% for composition. The rst axis accounted for the larger part of the variance (structure = 49.5% and composition = 36.1%) and highlighted a shift, in both structure and composition, from E. brachycarpa to A. taxiformis, with E. brachycarpa in mixed stands placed between the two homogeneous stands of native and invasive seaweeds (Fig. 4). The second axis explained lower variation (structure = 13.3% and composition = 9.5%) and separated E. brachycarpa and A. taxiformis from E. brachycarpa in mixed stands (Fig. 4).
SIMPER analysis revealed that 28 taxa contributed 70% to the dissimilarity between E. brachycarpa and A. taxiformis; 37 taxa contributed 70% to the dissimilarity between E. brachycarpa and E. brachycarpa in mixed stands; and 30 taxa contributed 70%to the of dissimilarity between A. taxiformis and E.
brachycarpa (Table S5). Most of the species contributing to the dissimilarities belonged to amphipods. The polychaete Amphiglena mediterranea (Leydig, 1851) was the species mostly contributing to the differences observed between both E. brachycarpa and A. taxiformis and between E. brachycarpa and E. brachycarpa in mixed stands contributing respectively to 8% and 6% of the observed differences. The amphipod Apherusa alacris Krapp-Schickel, 1969 was the species mostly contributing to the differences (7%) between E. brachycarpa in mixed stands and A. taxiformis. In addition, gastropod Obtusella macilenta (Monterosato, 1880) was the species that contributed consistently (higher δi/SD(δi) values) to the difference between E. brachycarpa and A. taxiformis (Table S5), while the amphipod Stenothoe monoculoides (Montagu, 1813) and the gastropod Eatonina cossurae (Calcara, 1841) were the species that contributed consistently to the differences between E. brachycarpa in mixed stands and E. brachycarpa, and between E. brachycarpa in mixed stands and A. taxiformis (Table S5). The polychaete S. prolifera was among the rst 5 species contributing to the differences between each couple of habitats (Table S5).
Multivariate analyses conducted separately for the three dominant epifaunal groups (molluscs, annelids and amphipods) revealed patterns of variation comparable to that of the whole epifaunal assemblage (Table S6). Only, amphipods showed less variability among habitats (Table S6).
Seaweeds structural attributes and relationships with the epifaunal assemblages.
Canopy volume (CV) and interstitial volume (IV) differed signi cantly among habitats with higher values in A. taxiformis compared to E. brachycarpa in mixed stands and E. brachycarpa ( Fig. 5a-b, Table S7). Biomass (DW) and thallus volume (TV) showed similar values between E. brachycarpa and E. brachycarpa in mixed stands and were signi cantly higher compared to that of A. taxiformis (Fig. 5c-d, Table S7).
Linear regression analysis revealed that biomass (DW) was the attribute that explained better the variation of abundance (R 2 N = 0.51), species richness (R 2 S = 0.61) and Shannon-Wiener diversity (R 2 H' = 0.54) of the epifaunal assemblages (Table S8). The variance explained by algal biomass increased if we considered a quadratic relationship between those variables (Fig. 6). Otherwise, canopy volume (CV) interstitial volume (IV) and thallus volume (TV) explained less variation (although highly signi cant p < 0.001) of the epifaunal attributes (R-squared < 0.5, Table S8). The analysis conducted separately on the three dominant epifaunal groups (molluscs, annelids and amphipods) revealed similar results, albeit with some differences. In fact, the variation of abundance, species richness and diversity explained by algal biomass decreased in both annelids (R 2 N = 0.35, R 2 S = 0.51, R 2 H' = 0.49) and amphipods (R 2 N = 0.23, R 2 S = 0.43, R 2 H' = 0.31) although resulting the most related explanatory variable for both groups, while molluscs revealed patterns of variation similar to the whole assemblage (R 2 N = 0.5, R 2 S = 0.53, R 2 H' = 0.48) (Table S8). Moreover, annelids showed a weaker and not signi cant relationship with the canopy and interstitial volumes in respect to amphipods and molluscs (Table S8). Loading [MathJax]/jax/output/CommonHTML/fonts/TeX/fontdata.js Biomass (DW) was also the structural attribute selected for constrained db-RDA, explaining 24.7% of the variation of the structure of the epifaunal assemblages (Table S9). The rst two axes of the dbRDA plot explained the 15.6% of the total variance of the multivariate structure of the epifaunal assemblages, with 12.4% for axis 1 and 3.2% for axis 2 (Fig. 7).

Discussion
The biodiversity and the socio-economic value of marine ecosystems are threatened by biological invasions around the world (Bax et al., 2003;Molnar et al., 2008). Understanding how invasive seaweeds modify the functioning of recipient ecosystems may allow to better understand large scale effects on native rocky shore habitats. Here we investigated the effects of the invasive A. taxiformis on the native E. brachycarpa by comparing the epifaunal assemblage associated with to three alternative states of the transition between native and invasive seaweeds, homogenous and mixed stands of the two seaweeds.
Our results showed that A. taxiformis can determine drastic shift in the epifaunal assemblages associated with the native E. brachycarpa, leading to the reduction of abundance, number of species and diversity. In particular, A. taxiformis hosted almost 6 times less epifaunal individuals compared to E. brachycarpa in mixed stands, and 10 folds less individuals compared to homogenous stands of E. brachycarpa. Also, the number of epifaunal species was more than 4 folds lower in the invasive compared to the native habitat, while diversity (Shannon-Wiener) reduced by a half. These results con rm the negative role played by invasive seaweeds in invaded habitats (Maggi et  We found that variation in diversity and multivariate structure of the epifaunal assemblages was related to changes in algal structural features. In particular, biomass was the variable better explaining the variation of abundance, number of species and the multivariate structure of the epifaunal assemblages. The role of the macroalgal complexity in shaping the associated biota has been highlighted in several studies, with complex algae hosting larger abundance and diversity of epifauna than simpler ones  Bedini et al. (2014) found that the invasive Lophocladia lallemandii (Montagne) F. Schmitz hosted a higher abundance of amphipods, isopods and polychaetes, while native habitats harbored a greater abundance of molluscs and decapods. Bivalves associated to the invasive S. muticum were more abundant compared to native seaweeds, which in contrast hosted more gastropods , and Gestoso et al. (2010) found that isopods and amphipods were more abundant in S. muticum than in native seaweeds. Moreover, the invasive Codium fragile subsp. fragile (Suringar) Hariot supported higher densities of nematodes, bivalves and specialist herbivores compared to fronds of the native kelp, that in contrast supported greater densities of gastropods and asteroids (Schmidt and Scheibling, 2006). Other authors revealed that differences among invasive and native seaweeds in single components of epifaunal assemblages changed depending on the site and the identity of the algal species (Navarro-Barranco et al., 2019). The fact that in our study, the A. taxiformis habitat showed lower abundance, species richness and diversity values for all the epifaunal organisms, regardless of the groups investigated, led us to hypothesize that a potential shift from the native (i.e. E. brachycarpa) to the invasive (i.e. A. taxiformis) habitat could cause large negative cascade effects within the benthic ecosystem.
Although differences in the epifaunal assemblages among native and invasive seaweeds have been already largely explored, our results also suggest that the presence of A. taxiformis affects the epifaunal assemblages associated to E. brachycarpa in mixed stands. This result could be explained by other attributes that differed between native and invasive seaweeds, such as the amount of epiphytes and/or the presence of chemical defenses, that have been related to the ability of seaweeds to shape their associated fauna (Hay et al., 1987;Viejo, 1999 . Other studies suggest that invasive seaweeds can alter the trophic web by changing the composition of epiphytes which reduces suitable habitat for many epifaunal species (Viejo, 1999;Wikström and Kautsky, 2004). Several authors suggested that the amount of epiphytes could explain the higher species richness found in the invasive S. muticum compared to native seaweeds (Viejo, 1999;Cacabelos et al., 2010). In our study, A. taxiformis had no or fewer epiphytes compared to E. brachycarpa (R.C. personal observation). As epifauna is mostly Loading [MathJax]/jax/output/CommonHTML/fonts/TeX/fontdata.js represented by microalgae grazers, we can hypothesize that differences in the abundance of epiphytes between A. taxiformis and E. brachycarpa could contribute to the variation in epifaunal assemblages observed in this study. It is therefore arguable that further studies analyzing the direct and indirect role of epiphyte abundance and secondary metabolites released by A. taxiformis in structuring its associated epifauna would allow to better clarify the effects of this seaweed on the recipient habitats.
Moreover, as suggested by other authors (Navarro-Barranco et al., 2019), landscape features could be another key aspect explaining the effect of A. taxiformis on E. brachycarpa associated assemblages in mixed stands. In fact, the presence of invasive seaweeds may contribute to the fragmentation of native habitats, reducing the patch size of native seaweeds, and at the same time increasing their isolation (Roberts and Poore, 2006;Lanham et al., 2015). It has been observed that the reduction of patch size of Cystoseira sensu lato habitats reduces the diversity of associated faunal assemblages (Mancuso et al., 2021). Thus, we can hypothesize that the presence of A. taxiformis in mixed stands can act as a physical barrier for the dispersal of vagile fauna, reducing connectivity at small scale and ultimately eroding the diversity of native habitats (Lanham et al., 2015).
In summary, our study suggests that shifting from native to invasive habitats may pose serious threat to biodiversity in coastal areas (Martin et al., 1992;Heck et al., 2003), potentially leading to bottom-up effects in rocky shore ecosystems. In addition, the low biomass supplied by the herein studied invasive species suggests that the shift from native canopy-forming algae to the invasive A. taxiformis habitat would also drastically reduce the biomass of primary producers of affected coastal areas. Predicting the ecological effects of invasive seaweeds is one of the main goals in the study of biological invasions.       Relationship between structural attributes and the multivariate structure of the epifaunal assemblages associated to the three habitats. Distance-based redundancy (dbRDA) plot illustrating the structural attribute better explaining the multivariate structure of the three habitats. DW.log = seaweeds biomass (log + 1).

Supplementary Files
This is a list of supplementary les associated with this preprint. Click to download.