The island of São Miguel ( Figure 1 ), with approximately 750 km², is the largest of the archipelago of the Azores. The coastline is about 155 km in length and seashore is generally steeply sloping. The wave action is known to be generally stronger on north coast and responsible for its higher erosion rate (Borges, 2003). Before being invaded by the Rugulopteryx okamurae (E.Y.Dawson) I.K.Hwang, W.J.Lee & H.S.Kim 2009 (Dictyotales, Heterokontophyta) alga (Faria et al., 2022a, b), the seascape in São Miguel was characterized by rocky substrata mainly covered with other species of brown algae (see Neto, 2001, Martins et al., 2013; Sangil et al., 2018 for further details on the former habitat).

Sampling sites were chosen around the island of São Miguel, with different environmental conditions: geographical location (north/south/east/west sites), degree of disturbance (non-polluted sites/disturbed/polluted sites) and degree of exposure to the wave action (Wallenstein, 2002) (H-high exposure/M-medium exposure/S-sheltered sites). A total of 11 sites (26 replicates) were sampled during Autumn, and 20 sites (79 replicates) were sampled during two consecutive years, 1996 and 1997, at similar depths, ranging from 5.6 to 15.0 m ( Table 1 ). Ribeira Quente (RQT), Ladeira da Velha (LDV) and Ponta da Ferraria (FER) were considered as naturally disturbed sites, because of shallow-water subtidal vents nearby (see Ávila et al., 2007; Couto et al., 2015; Martins et al., 2021). Ribeira da Praia (RPR) was also considered as a naturally disturbed site because of the freshwater influence from stream mouth. Polluted sites were those located near factory outlets (e.g.: Atalhada (ATA), Cofaco (COF), Santa Clara (SCL), Lactoaçoreana (LAC) and Sinaga (SIN)), those located inside harbors (Porto de Vila Franca do Campo (PVF) and Pesqueiro (PES)) and the submarine outlet of the Ponta Delgada water treatment station (EMI). The remaining sites were classified as non-polluted: São Vicente Ferreira (SVC), Ribeirinha (RBA), Caloura (CAL), Ilhéu de Vila Franca do Campo (ILH), Faial da Terra (FAT), Mosteiros (MOS), Porto Formoso (PFR) and Nordeste (NOR).

Samples were collected by SCUBA divers. At each site, three replicates (in the Autumn campaign) and 5 replicates (in the Spring campaign) were taken. Sea water temperature was recorded during each sampling. The sampling area was located using a two-digit random method adapted from Fenwick (1984) (see also Costa and Ávila, 2001). At each replicate, ten plants were gently pulled off of the rock by hand and put in a 0.45 mm mesh labelled cotton drawstring bag. In the laboratory, the samples were washed several times and the animals were removed by pouring the washing water through a 0.50 mm mesh sieve. All samples were labelled and preserved in 70% ethanol. After draining for about 30 minutes, the wet weight of the algae (AWW) was determined ( ± 0.01 g). The algae were then dried for 48 hours at 60°C and re-weighted ( ± 0.01 g) (algal dry weight, ADW). The samples were sorted under a binocular dissecting microscope and separated into major invertebrate groups. The live-collected molluscs were then sorted at the species level, identified and counted. Species authorities and synonymy of mollusc species follow the WoRMS database (WoRMS Editorial Board, 2023). All samples received a code and were deposited in the DBUA (Department of Biology of the University of the Azores) marine molluscs reference collection (access numbers DBUA 896-1001).

Species diversity was calculated for each replicate using species richness (S), total number of molluscs per sample (N), species diversity indices of Margalef (d) (Schoch and Dethier, 1996), Pielou (J’) (Warwick and Clarke, 1993) and Shannon-Wiener (H’) (Pearson and Rosenberg, 1978), and Simpson’s dominance index (1-λ’=1-Σ(N_i.(N_i-1)/(N.(N-1))) (Carr, 1996). Mollusc density was also calculated for both algal wet weight (total number of specimens of a species (n_i) per 100 g of algal wet weight; n_i/100 g AWW) and algal dry weight (n_i/100 g ADW) and compared with previous studies.

The relationship between the epifaunal assemblages and a set of environmental factors was investigated using distance-based redundancy analysis (dbRDA, Legendre and Anderson, 1999). Multivariate multiple regression, using the DISTLM routine, tested the significance of these relationships by fitting a linear model based on Bray-Curtis dissimilarities on the square-root-transformed data. The number of species for each samples was calculated and similarly analysed (multiple regression), but based on Euclidean distances of untransformed data. To retain predictor variables with good explanatory power, both the AIC (An Information Criterion; Akaike, 1973) and BIC (Bayesian Information Criterion; Schwarz, 1978) routines were used and compared as selection criteria, the rationale being that AIC tends to be rather generous criterion, whilst BIC tends to be rather severe. All analyses were based on a “forward” selection procedure.

The discrepancy in sampling efforts between seasons (3 vs. 5 replicates), does not bias the comparisons for the following reasons: 1) like PERMANOVA, dbRDA can be used to analyse any balanced or unbalanced design, either with or without covariables, for fixed, random or mixed models, either with or without hierarchical nesting. The procedures (permutation test) implemented in these tests were made so that they cater for unbalanced designs. In these cases, the differences in weights (replicates) between groups are balanced by adjusting the corresponding probabilities; and 2) prior to analysis, predictor variables were examined for skewness and multi-collinearity. No transformations were deemed as necessary, but the variable season was removed from the model as it was highly correlated with seawater temperature. Seawater temperature was retained in the analyses which, does not suffer for such notable differences in replicates.

All analyses were run on the statistical package PRIMER 6 + PERMANOVA add-on (Plymouth Routines in Multivariate Ecological Research – Plymouth Marine Laboratory) (Anderson et al., 2008).

At the time when this study was conducted (1996-1997), the invertebrates inhabiting Halopteris scoparia fronds were an abundant and taxonomically diversified group, dominated by peracarid crustaceans, but with a large proportion of molluscs (Costa, 2003). Prior to the recent invasion of the Azorean shores by the now dominant alga Rugulopteryx okamurae (Dictyotales, Ochrophyta), H. scoparia was a perennial and common alga in São Miguel Island, being very important for the local epiphytic fauna, since its biomass presented a low seasonal variation (Neto, 1997). Thus, this alga constituted a stable habitat for the establishment of assemblages of invertebrates and its dense ramification enabled the proliferation of a large epiphytic community that profited from all the microhabitats available in the algal fronds (Sánchez-Moyano et al., 2000a) and on which other invertebrates feed.

Most of the mollusc species found in algal fronds in the Azores belong to the family Rissoidae (e.g. Bullock et al., 1990; Gofas, 1990; Azevedo, 1991; Ávila, 1998; 2000a, 2003, 2005; Costa and Ávila, 2001; Ávila et al., 2005; Cordeiro and Ávila, 2015; Baptista et al., 2021). However, Bittium nanum (Mayer, 1864) [formerly misidentified by authors as Bittium latreillii (Payraudeau, 1826)] usually dominates these communities, with very high densities (Ávila, 2000b, 2003). Tricolia azorica (Dautzenberg, 1889) (see Baptista et al., 2023) is another very common gastropod, as well as Jujubinus pseudogravinae Nordsieck, 1973 (cf. Ávila et al., 2011), Manzonia unifasciata Dautzenberg, 1889 and Alvania sleursi (Amati, 1987). Besides T. azorica, endemics as Rissoa guernei Dautzenberg, 1889, Setia subvaricosa Gofas, 1990 and Setia ermelindoi S.P. Ávila & Cordeiro, 2015, are also found among algae (Costa and Ávila, 2001; Cordeiro and Ávila, 2015). Among the bivalves, Parvicardium vroomi van Aartsen, Menkhorst & Gittenberger, 1984 seems to be the most important species in these epifaunistic communities (Ávila, 2000b; Costa and Ávila, 2001), but Talochlamys multistriata (Poli, 1795) is also quite common.

The densities of molluscs found by us in this work are relatively high and some of them are the highest ever reported to the Azores: Bittium nanum [59,474 individuals/100 g ADW (Algal Dry Weigh)]; Parvicardium vroomi (4,867 individuals/100 g ADW); Setia subvaricosa (2,949 individuals/100 g ADW) and S. ermelindoi (762 individuals/100 g ADW) ( Table 5 ). Our findings also indicate that within the malacofauna associated with H. scoparia in the Azores, the gastropod family Rissoidae was the most prevalent, as evidenced by the identification of 11 species. The endemic Setia subvaricosa was generally the most abundant rissoid species in the samples, and the also endemics Alvania angioyi, Rissoa guernei and Setia ermelindoi were relatively well represented ( Supplementary Table 2 ).

In the malacological communities of H. scoparia in the Mediterranean (South of Spain), Rissoa guerinii Récluz, 1843 was very abundant (Sánchez-Moyano et al., 2000a) and in the coast of the Basque country, Rissoa parva (da Costa, 1778) was the dominant species (Borja, 1986a, b, c). In spite of being well represented in H. scoparia in São Miguel, Rissoa guernei was not very abundant. Barleeia unifasciata (Montagu, 1803) a highly abundant species in H. scoparia in the Mediterranean (Borja, 1986a, b, c) does not occur in the Azores (Ávila, 2005; Freitas et al., 2019). The most frequent species of molluscs found on this alga in the Mediterranean by Chemello and Milazzo (2002) were the rissoids Alvania lineata Risso, 1826, Alvania scabra (Phillippi, 1844) [reported as A. oranica (Philippi, 1844)], Rissoa similis Scacchi, 1836 and Setia ambigua (Brugnone, 1873), although some species belonging to other families, such as Cerithiidae (Bittium spp.), Tricoliidae (Tricolia spp.) and Columbellidae (Collumbela sp.) were also relatively abundant and frequent. The Azorean congenerics Bittium nanum and Tricolia azorica were also very abundant in the samples analysed by us and indeed, B. nanum is the dominant species in almost all the samples, making almost 50% of all the counted specimens ( Figure 3 ).

Bittium reticulatum (da Costa, 1778), Rissoa guerinii and Alvania discors (Allan, 1818) [reported as Alvania montagui Payraudeau, 1826] were found to be dominant in southern Spain locations studied by Sánchez-Moyano et al. (2000b), in conditions of high sedimentation, in communities where the diversity of molluscs was also high. In the same locations, Jujubinus ruscurianus (Weinkauff, 1868) and Skeneopsis planorbis (Fabricius O., 1780) were also abundant. The latter species and the Azorean endemism Jujubinus pseudogravinae were also frequent in H. scoparia samples from São Miguel Island. All these species and their congeneric seem to be characteristic from these algal communities (Borja, 1986a, b, c; Russo, 1997; Sánchez-Moyano et al., 2000a; Chemello and Milazzo, 2002). It is important to stress the endemic character of many mollusc epiphytic species in H. scoparia at São Miguel Island, especially the gastropods A. angioyi, R. guernei, Setia ermelindoi and S. subvaricosa.

Bivalves as Cardita calyculata (Linnaeus, 1758), Gregariella semigranata (Reeve, 1858) and Parvicardium vroomi had already been reported as algal turf inhabitants in São Miguel by Morton et al. (1998) and Ávila et al. (2005). Cardita seems to be a frequent genus in this type of community in the Mediterranean (Russo, 1997), as well as P. vroomi that was found there by Sánchez-Moyano et al. (2000b) on H. scoparia. P. vroomi was the most abundant bivalve in the samples of H. scoparia from São Miguel, similarly to what was reported by Ávila (2000b) for the multispecific algal samples collected on the north shore of this island at São Vicente Ferreira.

Borja (1986b) claimed that B. reticulatum is more abundant in November and that may be the reason why our Autumn samples are much richer in B. nanum than the Spring samples. However, and at least for the infralittoral communities studied by Azevedo (1991), the seasonal variation of B. nanum abundance was not very high, in spite of the higher abundance observed in the winter, especially in the most exposed locations on the north coast of São Miguel Island.

The choice of a single substrate such as H. scoparia for spatio-temporal comparisons of the mollusc community minimizes structural variability. However, it is possible that the environmental parameters may contribute to changes in the epifaunal populations by affecting the structure of biogenic habitats themselves. For example, hydrodynamics can influence the size and morphological structure of algae (Neto, 1997, 2001; Neto et al., 2000), which in turn may contribute to variation in the distribution of associated biota (Boaden et al., 1975).

Despite some variation among models, there were four environmental predictors that were common to all the four analyses (richness and assemblage structure using both AIC and BIC; cf. Tables 2 , 3 ): algal volume (that correlates with algal dry weight), seawater temperature, coastal orientation and depth.

The observation by Edgar (1983a), and many other subsequent papers (e.g., Rueda and Salas, 2003; Rosenfeld et al., 2017), that the abundance of molluscs is related to the algal biomass was corroborated in São Miguel by Azevedo (1992), who also noted that the algal biomass had a negative influence on diversity indices, due to increases in dominance. As such, it is not surprising that in the present study, there was also an important and consistent association between richness and assemblage structure with algal biomass ( Figure 4B ). This is most likely the result of a simple species-area relationship (cf. Ávila et al., 2018 and references therein), with greater amounts of habitat (the alga) supporting a richer assemblage of molluscs, although the influence of other factors acting on host size or morphologies, cannot be discarded without further examination.

Along with algal volume, seawater temperature (or seasons, which was discarded from the model due to collinearity) was also one of the most important factors associated with variation in richness and structure of mollusc assemblages ( Figure 4A ). Seasonality of abiotic factors (e.g., temperature and photoperiod) is mentioned by Edgar (1983b) as having a marked influence in the epifaunistic community and may be directly related to the life cycle of epifaunal species. Moreover, the great number of ovigerous females and juveniles of several invertebrate taxa other than the molluscs that were observed in the Spring samples by Costa (2003), and the fact that, in the malacological analysis, Bittium nanum was the species that most contributed for the observed separation due to high abundances related to reproduction, in the Autumn samples, reinforces the idea of a seasonal rather than annual pattern. The photoperiod is related to the onset of reproduction of many invertebrate species in Spring. The increased photoperiod in Spring will also result in an increased productivity of epiphytic filamentous algae. The number of animals may increase as a consequence of the increased heterogeneity of the habitat (provided by the growth of the host plant) and of the increase in epiphytes’ density. These fluctuations are particularly expected in temperate environments such as the Azorean islands, sheltered enough to allow a considerable growth of epiphytic algae (Edgar, 1983b). Also, the greatest values of algal biomass occur in Spring and Summer (Neto, 1997), which is the period when our second sampling campaign was carried out. Thus, the higher abundance of animals on Halopteris scoparia observed in the samples collected in the Spring may be related to an increased biomass of that alga in that period and also to an increase in the density of epiphytes. Similar seasonal fluctuations have been reported for epiphytic communities of Bryozoa (Conradi et al., 2000) and for epiphytic fauna of macrophytes all over the world (Edgar and Aoki, 1993). These fluctuations are related to variations in reproduction and recruiting associated with variations in food and substrate availability, factors which are not independent. Seasonal fluctuations in abundance related to population cycles of many species may be important for the detected variation. Finally, the space availability in the host plant also limits the number of organisms (Robertson and Mann, 1982). Therefore the decrease in biomass that H. scoparia suffers in winter (Neto, 1997) may result in reduced epifaunistic populations due both to a decreased microhabitat (Schneider and Mann, 1991) and to a decrease in food availability, which is not so easily retained in the plant ramifications (Borja, 1986b).

To a lesser extent, coastal orientation ( Figure 4D ) and depth ( Figure 4C ) were also important factors associated with variation of the structure of mollusc assemblages by all models. Coastal orientation can determine a number of environmental conditions, which in turn may influence the distribution of organisms. For instance, coastal orientation can determine the exposure to predominant winds and oceanic swells (e.g. leeward versus windward coasts of islands). Wave-action has profound effects on nearly all aspects of an organism’s life (Denny, 1988), such as recruitment and dislodgment of organisms (e.g. Vadas et al., 1990; Blanchette, 1997), supply of food and nutrients (e.g. McQuaid and Lindsay, 2007) and foraging activities of consumers (e.g. Vergés et al., 2009; Taylor and Schiel, 2010). It is thus not surprising that differences in community structure have been found between the leeward and windward coasts of islands (e.g. Hassett and Boehlert, 1999; Tuya and Haroun, 2006; Wernberg and Connell, 2008), even though Martins et al. (2013) found no consistent differences in the structure of macroalgae between the northern and southern coasts of São Miguel. In this study, the high number of sites in the North (30) and South (58) and only 4 in the East and 12 in the West prevents a more accurate analysis regarding the influence of geographical location in mollusc diversity. Still, it is obvious that both the bivalve Parvicardium vroomi and the gastropod Rissoa guernei are rare in these sites (E, W), when compared with North and South sites ( Supplementary Table 2 ). In spite of the scarcity of sheltered sites used in our analysis (only 3 sites) when compared with medium or highly exposed sites (43 and 58 sites, respectively) some considerations can be made. Tricolia azorica and Setia subvaricosa were more abundant in sheltered sites in contrast with Parvicardium vroomi that was rare in these locations. Sánchez-Moyano et al. (2000a) observed, in Algeciras bay (Spain), that when wave exposure increased, there was a tendency to an increase in diversity and a decrease in the abundance of organisms living in H. scoparia. This fact was also detected in the same bay by Conradi et al. (2000) in peracarids inhabiting the bryozoan Bugula neritina, as a consequence of a higher number of taxa represented in the more exposed localities. Kluijver (1997) reported that the distribution of vagile organisms in infralittoral communities of the North Sea was influenced by water movements, with a lower abundance of organisms in the more sheltered zones. In contrast, Tararam et al. (1981) in a study on Sargassum algae in Brazil did not find a relation between different wave exposure levels and diversity even though density would be higher in sheltered conditions. These apparently contradictory findings could only show different patterns in sedimentation, a factor that is strongly dependent on hydrodynamics (Moore, 1972). In very exposed sites there is no sediment deposition; in moderately exposed sites deposition of coarser sediments is allowed whereas in sheltered areas fine particulated matter will settle down. More detritus usually promote diversity or at least density of organisms, possibly by raising the number of detritivorous (Soughtgate, 1982). However, the deposition of fine sediments that occur at less exposed sites covers epifauna (Gibbons, 1988) and can decrease abundance and diversity by collapsing the spaces between branches and interfering with the feeding structures of the organisms (Hicks, 1980).

Similarly to coastal orientation, depth is another key environmental factor in subtidal habitats, which influence a number of environmental conditions. For instance, light intensity and water motion decline with depth affecting photosynthetic rate and nutrient uptake (see Hurd (2000) for a review). Even though the range of depth sampled was relatively small (5.5 to 15 m; Table 1 ), the first few metres below the surface are also where, proportionally, the largest changes in environmental conditions associated with depth occur, which may explain the observed association of the biota with depth ( Figure 4C ; Tables 2 , 3 ). Moreover, the relationship between depth and sample richness seems to be hump-shaped rather than linear. This could in theory be associated with disturbance caused by water motion, which decreases with depth.

Despite its significant effect when examined in isolation (cf. Figure 4E ), disturbance was seldom selected by models when considering all predictor variables together. This suggests that variation associated with disturbance is proportionally small (see Tables 2 , 3 ) and/or is already accounted for in terms of intensity and direction by previous predictor variables in the model (Anderson et al., 2008). We note that in some of the disturbed sites, Parvicardium vroomi and Setia subvaricosa, along with Alvania angioyi, are the most abundant species, in contrast with non-poluted and polluted sites, where usually Bittium nanum is, by far, the most abundant species.

Analysis of species abundance demonstrated that the endemic rissoid Setia subvaricosa, notably scarce on the north shore, was significantly more abundant in the south, especially in areas considered as naturally disturbed or polluted. This pattern suggest that S. subvaricosa could be used as an indicator species for disturbances within the Azores. Thus, the application in the Azores of this methodology aimed to detect disturbance in marine communities induced by effluents, favours a sampling program in Spring-Summer (when disturbance seems to be more susceptible to detection), and the use of H. scoparia in the subtidal zone, as the target alga is recommended due to its large covering of rocky shore substrates.