public read Prof, Ronaldo Francini-Filho Departamento de Engenharia e Meio Ambiente, Universidade Federal da Paraíba, Rio Tinto, Paraíba, Brazil Assistant Professor The Abrolhos Bank (eastern Brazil) encompasses the largest and richest coral reefs of the South Atlantic. Coral reef benthic assemblages of the region were monitored from 2003 to 2008. Two habitats (pinnacles' tops and walls) were sampled per site with 3–10 sites sampled within different reef areas. Different methodologies were applied in two distinct sampling periods: 2003–2005 and 2006–2008. Spatial coverage and taxonomic resolution were lower in the former than in the latter period. Benthic assemblages differed markedly in the smallest spatial scale, with greater differences recorded between habitats. Management regimes and biomass of fish functional groups (roving and territorial herbivores) had minor influences on benthic assemblages. These results suggest that local environmental factors such as light, depth and substrate inclination exert a stronger influence on the structure of benthic assemblages than protection from fishing. Reef walls of unprotected coastal reefs showed highest coral cover values, with a major contribution of Montastraea cavernosa (a sediment resistant species that may benefit from low light levels). An overall negative relationship between fleshy macroalgae and slow-growing reef-building organisms (i.e. scleractinians and crustose calcareous algae) was recorded, suggesting competition between these organisms. The opposite trend (i.e. positive relationships) was recorded for turf algae and the two reef-building organisms, suggesting beneficial interactions and/or co-occurrence mediated by unexplored factors. Turf algae cover increased across the region between 2006 and 2008, while scleractinian cover showed no change. The need of a continued and standardized monitoring program, aimed at understanding drivers of change in community patterns, as well as to subsidize sound adaptive conservation and management measures, is highlighted. Abrolhos reefs -38.625 -36.5 -17.0 -19.0 2003 2005-03-27 2006 2008 MsC Pedro Meirelles Federal University of Rio de Janeiro PhD Student

Av. Carlos Chagas Fo. S/N CCS - IB - Laboratory of Microbiology e SAGE-COPPE - BLOCO A (Anexo) A3 - sl 102, Cidade Universitária Rio de Janeiro RJ 21941-599 Brazil

+55 21 2562-6567 +55 21 2562-6567 pedrommeirelles@gmail.com http://pedromeirelles.com.br

The long-term monitoring program of coral reef benthic assemblages of the Abrolhos Bank started in 2003, through engagement of scientists and members of governmental and non- governmental organizations related to coastal management. Surveys were always carried out in the summer (January–March), thus avoiding seasonal artifacts. Each site was about 300 m in diameter and composed by 1–3 interconnected reef pinnacles, except for the rocky reefs of the Abrolhos Archipelago (see below). Spatial coverage and sampling methodologies varied through time, with two main periods. From 2003 to 2005 point-intercept lines (10 m length and 100 points; n = 4 per site) [40] were haphazardly placed on the pinnacle’s tops, and groups of four quadrats (50650 cm; 25 intercepts) equally distributed within 10 m lines were haphazardly placed on the pinnacle’s walls. Each group of quadrats was considered as a single sample (n = 4 per site). Organisms immediately below each point were recorded in situ and classified as follows: turf algae, crustose calcareous algae, fire- corals (milleporids), fleshy macroalgae, live corals, octocorals and zoanthids. The ‘‘live coral’’ category includes only scleractinians, with no species distinction. During this first period, monitoring was performed in four areas (Fig. 1), as follows: Area 1) No-take reserve of Timbebas Reef (three sampling sites) – Located within the National Marine Park of Abrolhos (NMPA). Created by the Brazilian government in 1983, the NMPA comprises two discontinuous portions, one closer to the coast and poorly enforced (Timbebas Reef), and another farther from the coast and more intensively enforced (Abrolhos Archipelago and Parcel dos Abrolhos Reef). Areas 2 and 3) Multiple-use and no-take zones of Itacolomis Reef – Itacolomis Reef is the largest reef complex (,50 km2) within the Marine Extractive Reserve of Corumbau (MERC) [41], [42]. It is divided into two main zones: multiple-use (Area 2; seven sampling sites) and no-take (Area 3; three sampling sites). Area 4) Unprotected coastal reefs (five sampling sites) – It encompasses the Parcel das Paredes Reef and Sebastia ̃o Gomes Reef, both subjected to the highest fishing pressure in the region [4] (Fig. 1). Between 2006 and 2008 benthic assemblages were character- ized using fixed photo-quadrats [34] in both, reef tops and walls (n = 10 per site). Each sample was composed by a mosaic of 15 high-resolution digital images totaling 0.7 m2. Quadrats were permanently delimited by fixed metal pins and set at haphazardly distances along 20–50 m axes. Relative cover of different benthic organisms was estimated through the identification of organisms (lowest taxonomic level possible) below 300 randomly distributed points per quadrat (i.e., 20 points per photograph) using the Coral Point Count with Excel Extensions Software [43]. Besides sampling the same sites within the abovementioned areas, two additional areas were sampled between 2006 and 2008, the Abrolhos Archipelago (five sampling sites) and Parcel dos Abrolhos Reef (five sampling sites), both within the NMPA portion that is farther from the coast (Fig. 1). The Abrolhos Archipelago is a rocky reef with no clear distinction between reef tops and walls, thus a single habitat (the reef front) was sampled. In total, 27 sites were sampled and 448 photo-quadrats were obtained per year between 2006 and 2008. A summary of the environmental characteristics of each sampling site is shown in Table S1. Logistical support and research permits were provided by Parque Nacional Marinho de Abrolhos and Reserva Extrativista Marinha de Corumbau/ICMBio (through J.R.S. Neto, R. Jerolisky and R. Oliveira). Data from this work was made available for public access through the Dryad platform (http:// datadryad.org/).

Detailed analyses were performed for the period between 2006 and 2008 (‘‘short-term comparisons’’), in which data was obtained with a higher taxonomic resolution and a greater spatial coverage (see above). Inferences for the entire sampling period (2003–2008, ‘‘long-term comparisons’’) were performed by making separate analyses for the two sampling periods: 2003–2005 and 2006–2008, and by considering only the same sampling sites and benthic categories (i.e. by standardizing data obtained in the two sampling periods). Long-term changes were taken into account only when similar trends were recorded for both sampling periods. Some metal pins marking the fixed photo-quadrats were lost during the sampling period. These samples were excluded from the analyses in order to assure that exactly the same photo- quadrats were used for the temporal comparisons. Final sample size ranged between 7–10 quadrats per habitat per site per year. Three common genera of fleshy macroalgae (Canistrocarpus spp., Dictyota spp. and Dictyopteris spp.) were difficult to distinguish in the images, thus being pooled into a single category (hereafter called ‘‘other fleshy macroalgae’’). All scleractinians were identified to the species level, except for Siderastrea spp., a genus for which three morphologically similar species are recorded for Brazil (S. stellata, S. siderea and S. radians) [44]. Data was also pooled for two morphologically similar fire-coral species (Millepora brasiliensis and M. alcicornis), but treated separately for the small-sized and conspicuous Millepora nitida. Analysis of variance (ANOVA) was used to evaluate spatial and temporal variations in benthic cover. Two separate groups of ANOVA were calculated, the first one focusing on differences between tops and walls (considering reef pinnacles only) and the second one focusing on differences between reefs while ignoring between-habitat variability, this latter including the shallow rocky reefs of the Abrolhos Archipelago (which has no distinction between tops and walls). Because data could not be collected in the tops of three reefs (see Table S1), between-site variability was ignored in the ANOVA models, thus avoiding missing observa- tions and the need of application of a less robust ANOVA model. In order to satisfy ANOVA assumptions of normality and homocedasticity, benthic cover percentages were converted to arcsin !x. Student-Newman-Keuls (SNK) multiple comparisons of means were performed as a post-hoc test [45]. Non-metric multidimensional scaling (MDS) ordination was used to summarize spatial and temporal similarities (Bray-Curtis) on the structure of benthic assemblages, and separate one-way analyses of similarities (ANOSIM) were used to evaluate significant differences according to reef areas, habitats and years [46]. Canonical correspondence analysis [47] was used to evaluate the influence of ecological and environmental explanatory variables on the structure (i.e. composition and relative cover) of benthic assemblages. Three fish functional groups are likely to exert strong influence on the benthos: 1) Large-bodied scrapers and grazers (Labridae: Scarinae), 2) Large-bodied browsers (Labridae: Sparisomatinae) and 3) Small-bodied territorial dam- selfish (Pomacentridae) [24], [48]–[50]. Biomass estimates for these three functional groups, together with depth, latitude, distance offshore and levels of protection were used as explanatory variables in the canonical correspondence analysis. Data on fish biomass was obtained from previous surveys [4], [41]. A forward selection procedure was used to include only the most important independent variables in the model, i.e. those contributing to increase the explanatory power of the model. Only significant variables, as defined by a Monte Carlo permutation test (999 permutations), were included in the final model. Reef areas were dummy-coded for levels of protection from fishing, as follows: 1) open-access reefs, 2) Itacolomis Reef (multiple-use portion), 3) Itacolomis Reefs (young no-take reserve), 4) Timbebas Reef (old and poorly enforced no-take reserve) and 5) Abrolhos Archipelago and Parcel dos Abrolhos Reef (old and well enforced no-take reserve) (see [4] for detailed information on protection levels of these areas; see Table S1). Multiple linear regression analyses [45] were used to evaluate the relative influence of major non-building organisms (i.e. turf algae, fleshy macroalgae and Palythoa caribaeorum) on the abun- dance of key reef-building organisms (scleractinians and crustose calcareous algae). Percentage cover data are compositional and thus subjected to constant sum constraint. Because this may mask true relationships among variables, analyses were performed using the centered log-ratio transformation [51].