Cetaceans along the southeastern Brazilian coast: occurrence, distribution and niche inference at local scale

Study area

The study area includes approximately 600 km of shoreline from 23°20′S to 25°50′S. In this region the continental shelf can be more than 230 km wide (Castro & Miranda, 1998) and reach depths between 120 m and 180 m (Mahiques et al., 2010). The SBCS is influenced by three different water masses: Coastal Water (CW), Tropical Water (TW) and South Atlantic Central Water (SACW) (Campos, 1995; Castro & Miranda, 1998). With a NE-SW orientation, the coastline is very heterogeneous with wide and long beaches and the presence of estuaries south from 24°S, a transition area until 23°48′S and the presence of small bays, coves and islands in the northern section (Tessler et al., 2006).

Six MPAs with distinct regulations and management plans cover almost the whole extension of coastal waters up to 40 m deep: three state MPAs known as “Área de Proteção Ambiental Marinha Sul, Central, and Norte”; another two state units known as “Parque Estadual Marinho da Laje de Santos” and “Parque Estadual Marinho da Ilha Anchieta”, and two federal protected areas known as “Estação Ecológica Tupiniquins” and “Estação Ecológica Tupinambás”. In the latter, the Alcatrazes Archipelago has great relevance based on its historical relation to cetacean records.

Oceanographic surveys

The presence of cetaceans in the area was assessed during a series of oceanographic surveys from two different projects. The first project “Occurrence, distribution and movement of cetaceans at São Paulo state coast”, hereinafter referred to as FAP, covered the entire extension of the study area between 2012 and 2015, and was divided into north and south transects (Fig. 1) due to logistic reasons. For the second project, in this study referred as PEMLS (Fig. 2), the focus was to evaluate the presence of cetaceans in the area of the “Parque Estadual Marinho da Laje de Santos” between 2013 and 2015. Transects were based in Buckland et al. (2001) and Sutherland (2006) with a few modifications to cover the larger number of islands along the coast.

The same methodology was used in both projects. A 15 m motor powered boat traveling at 15 knots was used. Information regarding the GPS position of the boat, time, sea state (Beaufort scale), waves, visibility and glare were taken every 30 min, in the beginning and the end of transects, and when cetaceans were detected. In this later situation, the boat would approach in order to identify the species, estimate group size, and gather photographs of the dorsal fin of dolphins and balaenopterid whales, flukes of humpbacks and head callosities of right whales for individual identification. Digital reflex cameras with 400 mm lenses were used and photographs were then analyzed according to the criteria described by Santos & Rosso (2008) to compose a catalog for each reported species. This information was further used to assess individual movements in the study area (see Santos et al., 2019).

Instituto Chico Mendes de Conservação da Biodiversidade provided the permits to work with cetaceans in the research area, including fieldwork (permit number: 37206-1).

Data analysis

The environmental data was obtained from global high-resolution rasters from MARSPEC (Sbrocco & Barber, 2013). The used variables were: depth of the seafloor (m), bathymetric slope (degrees), distance to shore (km), mean annual sea surface temperature (SST) (°C), annual variance in SST (°C), mean annual sea surface salinity (SSS) (psu) and annual variance in SSS (psu). These set of environmental variables are commonly used in other studies performed with cetaceans and are widely recognized to influence its distributions (see do Amaral et al., 2015, do Amaral et al., 2018; Barragán-Barrera et al., 2019) and references therein). ArcMapTM (Esri, http://www.esri.com) was used to extract the data for each sighting. Sightings without values regarding all environmental variables were excluded from the analysis.

For the species with 5+ sightings, univariate and multivariate analysis were performed to compare them in relation to the environmental variables measured at each sighting. Shapiro–Wilk test (shapiro.test, from stats package version 3.2.4) was used to assess if the data presented normal distribution. Kruskal-Wallis (kruskal.test from the same package) was used to test the null hypothesis that the distribution functions are the same among species (Zar, 1996). Dunn test (dunn.test from package dunn.test version 1.3.2) was used to determine which pair of species had a different distribution function (Dunn, 1964). The significance level α = 0.05 was used to all the previous analysis. Interspecies differences were further investigated through canonical discriminant analysis, which is a dimension-reduction technique related to principal component analysis (PCA) and canonical correlation. The methodology used in deriving the canonical coefficients parallels that of a one-way MANOVA. MANOVA tests for equality of the mean vector across class levels. Canonical discriminant analysis finds linear combinations of the quantitative variables that provide maximal separation among classes or groups. Given a classification variable and several quantitative variables, the CANDISC procedure derives canonical variables, linear combinations of the quantitative variables that summarize between-class variation in much the same way that principal components summarize total variation (Sas Institute Inc, 2017). The analyses were performed using standardized values of environmental variables through function scale (R Core Team, 2018) and canonical discriminant analysis was performed through candisc package (Friendly & Fox, 2020). Tests were performed in R v. 3.5.2 (R Core Team, 2018).

The species niche was estimated with the computational program NicheA, which uses scenopoetics variables to estimate Grinnellian niches (Qiao et al., 2016). Scenopoetic variables are non-interactive and, therefore, not affected by the species. NicheA uses the occurrence spots of the species and the three main components of the PCA of the environmental data to determine for each species the realized niche and fundamental niche as convex polyhedrons and ellipsoids in the environmental space, respectively. The fundamental niche was then transferred to the geographical space, indicating the closest areas to the centroid of the niche as the ones with adequate condition to the occurrence of the species. The volume and overlap of the fundamental and realized niche were calculated for every species pair, both in the environmental and geographic space.

S. guianensis

S. guianensis has a continuous distribution along the coast from Honduras (14°N) to Florianópolis, southern Brazilian coast (27°S) (Crespo et al., 2010). They are usually reported in estuaries, bays and coastal sheltered areas, with most sightings reported up to 25 m deep, and apparently limited to 50 m of depth (da Silva et al., 2010). Along the coast of São Paulo State, knowledge of the species has been focused on long-term studies conducted with a resident population in a protected estuary in the southern sector, the Cananeia estuary (Santos et al., 2019). These were the first efforts directed towards gathering data on S. guianensis along local coastal waters. The potential distribution area described in this study shows a clear preference for shallow waters and the 20 m isobath as an average limit for the species distribution along the southeastern Brazilian coast.

P. blainvillei

P. blainvillei also has a coastal distribution (do Amaral et al., 2018), with most of the sightings reported between 8 and 15 m deep, with limits at the 30 m isobath for sightings, and 35 m for incidental captures (Bordino et al., 2002). Most of the sightings reported here were made inside those limits, except for two groups sighted at 35 m and 44 m deep. The mean depth of the presented sightings was also higher than the described limits for the species. The absence of sightings along the central and the southern coasts of the surveyed coast could be explained by the water transparency –darker in the quoted subareas -, the cryptic behavior of the species, and its preference to use shallower waters close to the shore which were not completely surveyed due safety reasons.

S. frontalis

Throughout its range, S. frontalis is mainly reported in the continental shelf up to 1,000 m of depth across tropical and subtropical waters in the Atlantic Ocean (Herzing & Perrin, 2017). Along the Brazilian coast, the species is recorded mainly from 18 to 34°S and northern of 6°S (Moreno et al., 2005; do Amaral et al., 2015). The central region of São Paulo State coast showed the highest number of sightings of S. frontalis, especially in the area surrounding the PEMLS. This MPA is located in the central part of the potential distribution area estimated by the model for this species. The present and a previous study (Santos, Figueiredo & Van Bressen, 2017) show a year-round presence of Atlantic spotted dolphins along the coast of São Paulo, rendering an excellent opportunity for longitudinal investigations.

T. truncatus

The bottlenose dolphin was the third most common species in this study and had the second largest potential distribution area, with sightings occurring throughout the entire surveyed area. The species is known to have a continuous distribution along the Brazilian coastal waters and estuaries (Lodi et al., 2016) in depths ranging from 1.6 to 50 m (Laporta et al., 2016). Particularly since 1990, longitudinal studies have been conducted with common bottlenoses in estuaries and coastal waters of southern Brazil (Simões-Lopes & Fábian, 1999; Fruet et al., 2015). From Santa Catarina State (∼27°S) to lower latitudes, S. guianensis occupies shallower waters (da Silva et al., 2010), meanwhile T. truncatus are found in slightly deeper waters (up to 50 m). The distribution patterns described by the models in the present study clearly show the Guiana dolphin potential distribution area closer to the shore, the bottlenose dolphin found in deeper waters, and the overlap between them found close to the 20 m isobath.

B. edeni

Sightings of B. edeni in the study area are in agreement with literature, as they are reported year round along the southeastern Brazilian coast, with a higher number of sightings among spring and summer months when they are in search of preys in shallower waters (Zerbini et al., 1997; Siciliano et al., 2004; Santos et al., 2010; Gonçalves, Augustowski & Andriolo, 2016; Santos et al., 2019). The geographical areas indicated as closer to the centroid of the species niche, like São Sebastião Island and the Alcatrazes archipelago, could be related to high biological productivity. Jefferson, Webber & Pitman (2008) and Kato & Perrin (2009) suggested that Bryde’s whales have a preference for areas with higher productivity in tropical and subtropical basins. Tardin et al. (2019) showed a positive correlation between B. edeni and high concentrations of chl-a in depths between 30 and 60m in southeast Brazil.

S. bredanensis

The rough-toothed dolphin is known for its presence in deep and oceanic tropical and subtropical waters, with common records in shallow waters of the southeast coast of Brazil (Lodi, 1992; Jefferson, 2017; Santos et al., 2019). Evidences of site and group fidelity were previously observed in the surveyed area (Santos et al., 2019). The small geographical area for the potential niche indicates a possible restriction in the use of area and also for the environmental requirements for the investigated stock. As species with restricted distribution are more vulnerable to human impact (Gaston & Fuller, 2009), it is important to further understand their life history and environmental requirements in the coast of São Paulo state.

Niche overlap

The ecological niche models (ENM) presented in this study indicated a clear overlap between species niche and their potential distribution area. With largest niche volume and potential distribution area, B. edeni and S. frontalis are known to occupy areas that expand further than the SBCS. They also presented the largest niche overlap, not only between them, but also with other species (Figs. 7–9). The exception is the overlap with S. guianensis. The Guiana dolphin occupies areas closer to the shore and showed overlaps with P. blainvillei, and T. truncatus, and no overlap with S. bredanensis.

The use of ENMs is relevant when considering species with wide distribution like S. frontalis and T. truncatus, as it allows the evaluation of precise areas of occurrence. This information can be used to further understand the biogeography of those species, as well as to investigate their trophic ecology, like prey distribution according to predators.

It was expected that the franciscana dolphin would have a similar geographical area to the Guiana dolphin, as they are both reported to use coastal and turbid waters (Santos et al., 2002). However, the species was only sighted in surveys in the northern sector and in higher depths than they are usually reported. The clearer waters of this region could have raised the probabilities of sightings. Therefore, the niche estimated and higher values of mean annual SST and SSS obtained here may be biased for the entire study area, since individuals from the Southern sector were not portrayed in the analysis due to lesser probabilities of sightings. Nonetheless, it is important to better evaluate the way distinct populations use the area for conservation purposes, as a genetic distinction in stocks was pointed out by Cunha et al. (2014). Franciscana dolphins are vulnerable to extinction with a high mortality provoked by incidental captures throughout its whole restricted distribution (Zerbini et al., 2018). Thus, any information on the biogeography of distinct stocks is relevant to establish management plans.

Besides predation of prey, the only species interaction observed was a mixed group of 12 bottlenose dolphins and one Bryde’s whale in December 2012. do Amaral et al. (2015) and Di Tullio et al. (2016) reported the presence of mixed groups of cetaceans along the Brazilian coast. The bottlenose dolphin was present in more than half of the groups reported by Di Tullio et al. (2016) in deeper waters off Southern Brazil. Aggressive behavior was reported between males of Atlantic spotted and bottlenose dolphins in the Bahamas (Herzing, Moewe & Brunnick, 2003). Bottlenose and Guiana dolphins had previously been observed in aggressive interactions ignited by the larger bottlenoses (Acevedo-Gutierez et al., 2005; Wedekin, Daura-Jorge & Simões-Lopes, 2004) in both limits of the distribution of S. guianensis. In a general sense, toothed whales tend to avoid direct competition by adopting different behaviors, diets and physiologic habits (Bearzi, 2005). These characteristics might explain the observed species co-occurrences in the surveyed area.

Niche overlap is also related to geographical overlap (Warren, Glor & Turelli, 2008). Partitioning resources can reduce competition among species that occupy the same area (Roughgarden, 1976). This could explain the considerable niche and geographical overlap estimated for S. frontalis, B. edeni and T. truncatus (see Figs. 6–8; Tables 2 and 3). Another alternative is that the environment used by these species has abundant resources, therefore allowing them to coexist (Bearzi, 2005).

The presence of cetaceans close to several islands in the studied area might be related to the “island effect”, which means that oceanic islands present higher productivity than surrounding areas (Doty & Oguri, 1956). The areas including the main islands (e.g., Ilha Anchieta, Ilhabela, Alcatrazes, Laje de Santos) in this survey are close to the center of the niche centroid for B. edeni, S. frontalis, P. blainvillei and S. bredanensis. This result highlights the importance of these areas, which, in most occasions, are all included inside the limits of established MPAs. However, none of the published management plans considered the patterns of occurrence and distribution of cetaceans. Thus, we recommend that the results presented here should be considered in the next revision of these tools for nature conservation.