were close to 1 (i.e. a more uniform structural complexity at the
scale considered, Fig. 5). Coastal breakwaters sampled in Arica
(tetrapods) and Antofagasta (granite boulders) showed the highest
median values of structural complexity (2.8 and 2.5, respectively),
with the former having a maximum index value of 8.7 (see outliers
in Fig. 5).
Our generalized additive model (GAM) indicated that total AL
density, estimated across seven sampling localities and habitats,
had a significant non-linear relationship with the structural
complexity (Fig. 6a, Table 5). At intermediate structural complexity,
with index values ranging from 1.9 to 2.5, typical for most urban
breakwaters (see Fig. 5), total AL density was highest (Fig. 6a).
Similarly, total plastic material (i.e. bags, bottles, small pieces) and
paper (including cardboards) showed a relatively similar pattern,
increasing at intermediate structural complexity (Fig. 6b and c).
Despite the fact that the model revealed significant relationships
for these AL categories (Table 5), deviance explained by the model
was lower for plastics and paper (16.3, 14.5%, respectively) than for
total AL (i.e. 35%; Table 5). This result arose from the poor fit of the
model at high structural complexity (typical for tetrapods) for these
AL categories, which is confirmed by the wide confidence intervals
observed at the complexity levels (see Fig. 6b and c). Anyways, the
smooth function of the model was significant for these AL categories
(Table 5). No significant estimates were obtained for glass
(bottles and small pieces), and the model only explained 6.5% of
total deviance in glass density (Table 5).