Table 3. Summary of the Random Forest (RF) and General Additive Model (GAM) for the diversity measures.

Diversity measures	% RF	%GAM	Local factors		Estuarine/catchment factors
			RF	GAM	RF	GAM
α-	30	51	shell	+shell-p***, −chl.a: mud*	ns	ns
γ-site	35	52	shell, med.sand	+shell-p***, +shell-r*	ns	ns
β-site	21	51	shell, med.sand, distance	+shell-p***, +shell-r**, s(chl.a)**	ns	+runoff*, +exotic*
β-conn	46	59	shell	−shell-p***, −shell-r**, −coarse*	inflow, urban	−inflow***, +intertidal** , −runoff.

Percentage of variance explained by each model and the significant effects of the variables at local and estuarine/catchment scales. All the GAMs were significant at p<0.001. Factor shell in the GAM is included as a categorical variable: “shell-p”, presence of shell, and “shell-r”, rare shell content.

^***p<0.001,

^**p<0.01,

^*p<0.05 and.: p<0.01.

“ns”: non-significant effects. +/− indicates the direction of the effects. s(factor) indicates smooth effects. “: ”crossed effects interaction.