*2.4. Statistical Analysis*

Both abiotic and biotic data were used for the data analysis. Water variables were temperature, salinity, and dissolved oxygen; sediment variables included water content, OM, chlorophyll-*<sup>a</sup>*, and phaeopigments. Biotic data consisted in the abundance of the meiofauna and were used to construct a taxa-by-site and period matrix. The environmental data variation was represented by means of box-plots for each variable and each site. The biotic parameters computed were the number of taxa (S, taxon richness), the number of individuals per taxa (A, abundance), and the Shannon (H', diversity) and Pielou (J, evenness) indices, as well as Ne/Co ratio. These biotic variables were computed for the three sampling sites C1, C2, and C3, for each replicate and date.

As for multivariate analysis, the non-parametric permutational analysis of variance (two way-PERMANOVA), based on Bray-Curtis (dis)similarity measures [53] was carried out to test significant di fferences of the structure of community among sites (three levels: C1, C2, and C3), periods (two levels: July and February), and site × period interactions as fixed factors. The data were log(x+1) transformed before the analysis. The PERMANOVA, based on Euclidean distance, was also used to test the significant di fferences of all the biotic univariate measures (i.e., total meiofaunal abundance, number of taxa, Shannon-diversity, Pielou evenness, and Ne/Co ratio). A log (x+1) transformation of data was applied only for the total meiofaunal abundance. The significance was computed by permutation with 9999 replicates. The pairwise comparisons between all pairs of sites were computed as post-hoc test and the Bonferroni correction procedure was followed to account for multiple simultaneous correlations [54]. The principal component analysis (PCA), based on the correlation matrix, was used to explore the faunal variations within the lagoon and periods. The environmental variables were used to understand the key environmental variables accounting for the much % of variance a ffecting the meiofaunal distribution. The multivariate procedure non-metric multidimensional scaling (nMDS) was used to investigate the di fferences between the sites; the more informative environmental variables were added in the analysis to best explain the meiobenthos structure and they were superimposed in the graph [55]. The meiofaunal major taxa contributing most to (dis)similarities among the sites were identified using the similarity percentages (SIMPER) test.
