Disentangling the links between habitat complexity and biodiversity in a kelp‐dominated subantarctic community

Study sites and sampling design The Magellan region is located in the South East Pacific, encompassing a diversity of habitats that include fjords, inland seas, glaciers, gulfs, and channels—a complex landscape resulting from the combined effect of tectonic processes and glaciation. The oceanographic features and diverse environmental conditions determine a particular marine biogeographic unit (Camus 2001) dominated by benthic invertebrates and extensive kelp forests (Friedlander et al. 2018). To analyse the potential spatial variation in the effects of kelp’s structural complexity, the study was conducted at two sites (4 km apart) in the Strait of Magellan (Fig. 2), named “Bahía Buzos” (-53.627°S, -70.919°W) and “Carrera” (-53.587°S, -70.921°W). Both sites harbour L. flavicans forests of around 400 m2, located between 0.5 and 4 m depth (C. Velasco-Charpentier pers. obs.). Both sites are moderately wave-exposed, with a hard substratum consisting of boulders and large rocks covered by crustose coralline algae. The patchy rocky reefs extend from the intertidal to ~ 8 m depth. An observational sampling program of kelp habitat complexity and associated communities was established in both sites between November 2018 and August 2019. Observations were conducted every three months in both sites. Punta Carrera was sampled only during spring and summer due to logistics limitations and harsh weather. At each sampling time and site, SCUBA divers deployed a 100 m transect parallel to the coastline. The transects were placed on substrata of similar rock composition (boulders), depth (0.5 – 2 meters), and inclination (~ 0°). In each transect, ten 0.25 m2 quadrants separated by ~ 3 meters were photographed with a Nikon D7200 camera equipped with an Ikelite housing and a rectilinear Tokina 11-16 lens. We selected quadrants with at least one L. flavicans specimen. So, the unit of replication was each kelp specimen (N = 10 per site and season). Each thallus was wrapped with a mesh bag (1 mm2 pore size). Then, the holdfast was detached from the rock with a knife and the bag was swiftly closed to minimise the escape of mobile macrobenthic organisms. Within one hour after collection, samples were transported to the Laboratorio de Ecofisiología y Biotecnología de Algas, Universidad de Magallanes to measure the thallus morphology attributes and species sorting. Estimation of kelp structural complexity and associated biodiversity Once in the laboratory, we measured five kelp structural complexity attributes. The number of stipes, maximum length (i.e. thallus length, from holdfast base to the apex in cm), holdfast diameter (i.e. maximum width of the holdfast base in cm), and total wet weight in g (0.01 g precision) were directly measured from the collected kelps. Foliage cover was estimated from each digital photography. This variable was categorized as 1 (fronds cover between < 10 - 30% of the quadrant), 2 (fronds cover between 31-50%), 3 (fronds cover between 51 - 80%), and 4 (frond cover between 81-100%). For each L. flavicans thallus, macrobenthic (> 1 mm) mobile and semi-sessile organisms (e.g. mussels), in addition to tube dwelling sessile organisms (e.g. spirorbid polychaetes) were collected from thallus surface with aid of forceps and scalpels. In addition, the stipes and discs were dissected to collect organisms living in crevices and galleries. We used specialized literature and field guides to identify each individual to the lowest taxonomic level possible, usually species (Osorio et al. 1979; McLean 1984; Pastorino 2005; Häussermann and Försterra 2009; González-Wevar et al. 2018). Species-specific abundance was estimated as counts of individuals. These data were used to estimate species richness, defined as the number of taxonomic identities; total abundance, defined as the total number of individuals per taxonomic identities; and community structure, defined as the combination of species identities and abundances. Before the analyses, Pearson-product moment correlations were calculated among the explanatory variables in order to account for collinearity.