*2.4. Data Analyses*

In order to evaluate environmental di fferences among sites, multivariate analyses were carried out on the environmental data using the principal component analysis (PCA) ordination method and correlation matrices [38]. The components loadings were considered to quantify the correlation between the variables and the principal components scores. Water and sediment variables were analyzed separately in order to investigate the trophic features of the Yundang Lagoon from two di fferent perspectives and to evaluate the consistency of results between the two ecosystem components (water and sediment). Near-bottom water variables included salinity, ammonium, nitrate, nitrite, dissolved inorganic nitrogen (DIN = sum of ammonium, nitrate, and nitrite), RP, N/P ratio, DO, BOD5, CODMn, chlorophyll-*a* (Chl-*a*), and suspended solids (SS). Sediment variables included the sand (>63 μm), silt (63-8 μm) and clay (<8 μm) fractions, Md, Wc, and TOC (TN was excluded from the multivariate analyses as it correlated to TOC at *R* > 0.90).

PERMANOVA was used to tests for di fferences in the arrangemen<sup>t</sup> of the sites in terms of biotic parameters (i.e., abundance, number of species, diversity, biomass) and the abundance of individual species [39]. The pairwise post hoc test was also performed for all pairs of site comparisons. The Bonferroni correction was used and 9999 permutations were performed [40]. Di fferences among sites related to macrozoobenthos data were examined by means of non-metric multidimensional scaling (nMDS) based on the Bray–Curtis similarity matrix. Dominance curves [41] were plotted for each site, and species were ranked in order of importance in terms of percent cumulative abundance. Canonical correspondence analysis (CCA) was performed on macrozoobenthos abundance data (log transformed) and environmental variables (nitrate, ammonium, RP, SS, sand, Md, TOC) at each site, in order to test the response of the macrozoobenthic community to the main environmental (trophic and confinement) gradients to determine what provided the best combination of variables supporting the ordination model [42]. In particular, nitrate, ammonium, and RP for the water and TOC for the sediments were used as trophic variables, while SS for the water and sand and Md for the sediments were used as variables related to the hydrodynamics and confinement of the lagoon.

All statistical analyses were performed by means on the PAST statistics program (version 3.12).
