Results
We found 152 species, 93 of which still undescribed (61.2%), in 348 samples collected across 116 beaches in the Western Mediterranean (Figure 2B, GBIF dataset: https://doi.org/10.15468/64xtt9, Curini-Galletti et al., 2023). Most of the species exhibited very limited distribution ranges, with 58 species limited to a single beach, 22 to two and 12 to three beaches. In contrast, only three species were found in more than 50 beaches: Coelogynopora gynocotylaSteinböck, 1924 (65 beaches), Otoplana bosporana Ax, 1959 (63 beaches), Postbursoplana fibulata Ax, 1956 (56 beaches). Species richness per sample ranged up to 10 species (totalling up to 17 species per beach) although 59.8% of the samples ranged between 1 and 3 co-occurring species (Supplementary Table S2).
Taxonomic richness did not differ significantly across beach levels (Figure 2C), but was significantly explained by environmental variables: negatively affected by mean grain size, having fewer species with coarser grain size, and marginally affected by eastward horizontal surface velocity, having more species with stronger winds (Table 1). Considering each beach level separately (Supplementary Table S5), species richness in the swash level was significantly and negatively affected by northward horizontal surface velocity and precipitation of the coldest quarter; none of the analysed environmental variables affected richness in the shoaling level; richness in the subtidal level was negatively affected by mean grain size and positively by eastward horizontal velocity.
Species traits calculated after the Gower distance matrix were significantly affected by the beach level, as well as by mean grain size (Figure 3A). The effects of the beach level on trait dissimilarity had different direction in the swash than in the subtidal level.Despite the fact that we detected significant effects of environmental variables, most of the variance in the community matrix remained unexplained (Variance partitioning analyses, residuals = 87%), with geography, environment and their interactions explaining only 4%, 7% and 2% of the variance respectively (Figure 3B). Geographical distances were weakly correlated to beta diversity (Mantel test: r = 0.095, p = 0.001); the correlation was stronger in the shoaling and subtidal levels than in the swash levels, where it was not significant (Figure 3C-E). Differences in community composition of proseriates were explained by beach level, beach identity, and mean grain size, but not eastward horizontal surface velocity (Table 2). The effect of these variables on each species is summarized in Supplementary Table S6. The species that was most strongly affected by beach level was Otoplana bosporana(LRT = 111.6, p-adjusted = 0.001), occurring significantly more often in the swash than in the other two levels.
Properties of the functional space differed across beach levels. Functional richness was significantly lower than the null modelling only in the swash level (observed = 0.040, estimated = 0.057, ses = -2.8, p = 0.0049) (Figure 4) (H3); but not in shoaling (observed = 0.043, estimated = 0 .059, ses= -1.6, p=0.1002) and subtidal levels (observed = 0.047, estimated = 0.065, ses = -1.8, p = 0.0778) (Figure 4A-C). Species contributions were mostly negative, but higher in the swash level (Contribution average = -12.20 ± 0. 23), followed by the shoaling (Contribution average = -9.29 ± 0.17) and the subtidal (Contribution average = -4.75 ± 0. 12) (Figure 4D).
Finally, our analyses confirmed that those differences were due to the different proportion of traits adapted to hydrodynamics in each beach levels. Specifically, the proportion of species with cephalic sensory area (GLMM: LR Chisq = 16.5, p = 0.0003), brain capsule (LR Chisq = 50.3, p < 0.0001), ventral sole (LR Chisq = 30.5, p < 0.0001), and flat body (LR Chisq = 52.5, p < 0.0001), all exhibit higher proportions in the swash level (Figure 5, Supplementary Table S7) (H4). The proportion of species with well-developed adhesive glands (LR Chisq = 20.9, p < 0.0001) was higher in the shoaling, whereas no differences were found regarding species with eversible pharynx (LR Chisq = 2.5, p = 0.2873) (Figure 5).