INTRODUCTION
The Atlantic coast of South America is included in the distribution area of several cetaceans, including resident species of dolphins such as Sotalia guianensis (Van Bénéden, 1864). This species occurs in tropical and subtropical waters from Honduras, in the Caribbean Sea, to Florianópolis, on the southern coast of Brazil (Flores & Da Silva, Reference Flores, da Silva, Perrin, Würsig and Thewissen2009), showing high site fidelity and coastal habits usually associated with estuary bays (Santos et al., Reference Santos, Acuña and Rosso2001; Flores & Bazzalo, Reference Flores and Bazzalo2004; Azevedo et al., Reference Azevedo, Viana, Oliveira and Sluys2005; Rossi-Santos et al., Reference Rossi-Santos, Wedekin and Monteiro-Filho2007; Nery et al., Reference Nery, Especie and Simão2008; Dias et al., Reference Dias, Herzinga and Flach2009). The other member of this genus, S. fluviatilis, also known as the tucuxi, is a riverine dolphin species completely adapted to fresh water and restricted to the Amazon region.
A recent study has shown that the morphological variation of S. guianensis follows a phylogeographic pattern (Cunha et al., Reference Cunha, da Silva, Solé-Cava, Ruiz-Garcia and Shostell2010) along the coast, and as geographic variation in morphology may result from distinct selective pressures in different environmental conditions (Gould & Johnston, Reference Gould and Johnston1972), it may be a great tool to define population substructure in this species. However, despite the recent taxonomic re-evaluation of the genus Sotalia (see Monteiro-Filho et al., Reference Monteiro-Filho, Monteiro and Reis2002; Cunha et al., Reference Cunha, da Silva, Lailson-Brito, Santos, Flores, Martin, Azevedo, Fragoso, Zanelatto and Sole-Cava2005; Caballero et al., Reference Caballero, Trujillo, Vianna, Barrios-Garrido, Montiel, Beltran-Pedreros, Marmontel, Santos, Rossi-Santos, Santos and Baker2007) there are few studies focused on S. guianensis geographic variation, especially the morphometric aspects of the periotic–timpanic bone. Among previous studies analysing morphometric variation in S. guianensis, Ramos (Reference Ramos2001) used morphometric skull variation, while other studies focused on interspecific comparisons between the two species of the genus (see Monteiro-Filho et al., Reference Monteiro-Filho, Monteiro and Reis2002; Fettuccia, Reference Fettuccia2006; Fettuccia et al., 2012).
However, acoustic adaptations to the aquatic environment determined several morphological changes in the auditory apparatus of cetaceans. Both auditory capsules, known as the periotic–timpanic bone complex, are probably the most divergent structures of the skull of cetaceans (Mead & Fordyce, Reference Mead and Fordyce2009) and are relevant in taxonomic studies of both Odontoceti (Kasuya, Reference Kasuya1973) and Mysticeti (Geisler & Luo, Reference Geisler and Luo1996). In this study, we evaluated the use of traditional morphometrics of the periotic–timpanic bone to identify geographic variation in S. guianensis populations and improve the taxonomic resolution and biogeographical aspects (see Perrin, Reference Perrin1975).
MATERIALS AND METHODS
We analysed 142 specimens of Sotalia guianensis from collections of the following institutions and research groups: Museu Paraense Emílio Goeldi (MPEG), Belém, Pará (N = 38); Instituto Ilha do Caju Ecodesenvolvimento e Pesquisa (PROCEMA/ICEP), Ilha do Caju, Maranhão (N = 10); Associação de Pesquisa e Preservação de Ecossistemas Aquáticos (AQUASIS), Caucaia, Ceará (N = 28); Universidade Estadual do Norte Fluminense (UENF), Campos dos Goytacazes (N = 43), Rio de Janeiro; and Grupo de Estudos de Mamíferos Marinhos da Região dos Lagos (GEMM-Lagos; N = 19), Rio de Janeiro.
We grouped samples from different localities in four operational taxonomic units (OTUs; Vanzolini, Reference Vanzolini2002; Heyer, Reference Heyer2005) based on geographic proximity, and assuming ecological homogeneity (Peloso & Avila-Pires, Reference Peloso and Avila-Pires2010), in three Brazilian regions: north (Unit Amapá/Pará, AP/PA); north-east (Unit Maranhão/Piauí, MA/PI; and Ceará, CE); and south-east (Unit Rio de Janeiro, RJ) (Figure 1).
Fig. 1. Location of the five operational taxonomic units (OTUs) on the Brazilian coast by region. North: unit Amapá/Para (AP/PA); north-east: units Maranhão/Piauí (MA/PI) and Ceará (CE); south-east: unit Rio de Janeiro (RJ). Map: D.L. Arcoverde.
We measured 21 metric characteristics of each periotic–timpanic bone complex. Linear measures were taken, according to Kasuya (Reference Kasuya1973), with a 200 mm caliper (0.1 mm precision), except the measures 12, 14 and 21 that were first reported in this study (Figure 2; Tables 1 and 2). All measurements were taken by DLA to minimize the variation attributed to different collectors (Perrin et al., Reference Perrin, Yablokov, Barlow and Mina1994).
Fig. 2. Five views of left side periotic–timpanic of Sotalia guianensis (MPEG 39451) showing the measures used: A, ventral; B, lateral; C, dorsomedial; D, anterior; E, posterior. CW, cochlear window; EF, elliptical foramen). Photographs: D.L. Arcoverde.
Table 1. List of measurements of periotic–timpanic complex bone taken of Sotalia guianensis specimens from Brazil according to Kasuya (Reference Kasuya1973). The measures 12, 14 and 21 were first presented in this study.
Table 2. Descriptive statistic of periotic–timpanic bone complex of Sotalia guianensis from north: Ampá/Pará (AP/PA), north-east: Maranhão/Piauí (MA/PI) and Ceará (CE), south-east regions of Brazil: Rio de Janeiro (RJ). N, total number; X, average; Min, minimum; Max, maximum; SD, standard deviation; CV, coefficient of variation in %. Measurements are in millimetres.
Assuming the absence of sexual dimorphism in Sotalia (Borobia, Reference Borobia1989; Monteiro-Filho et al., Reference Monteiro-Filho, Monteiro and Reis2002) as well as the absence of ontogenetic variation in the periotic–timpanic bone complex of Odontoceti (Kasuya, Reference Kasuya1973), both sexes and different stages of development were grouped together in the analysis. According to Kasuya (Reference Kasuya1973) the periotic–timpanic bone complex had no significant asymmetry, which was later confirmed for S. guianensis (Parente et al., Reference Parente, Alves, Furtado-Neto and Monteiro-Neto1999). Therefore, for this study we used the data from the right periotic–timpanic complex, and from the left complex when the right was not available. The definition of the periotic–timpanic bone complex followed Simões-Lopes (Reference Simões-Lopes2006), while the osteological terminology and anatomic orientation followed Mead & Fordyce (Reference Mead and Fordyce2009).
We evaluated the morphometrics using a discriminant analyses function (DAF) for all the OTUs in order to identify the more powerful measurements in discriminating groups (Tabachnick & Fidell, Reference Tabachnick and Fidell2001). The DAF analyses were conducted in STATISTICA® v.7.1 (StatSoft Inc., USA) (StatSoft, 2005) considering a level of significance of P ≤ 0.05.
RESULTS
Discriminant analysis of the periotic–timpanic bone complex showed significant geographic variation among OTUs (Wilks' λ: 0.04292, F (48.295) = 11.578; P <0.01). Root 1 showed 75% of the variation while root 2 showed 20% of the variation. The classification matrix showed 96.6% of correct classification, and the RJ unit had the best correct classification (98.3%) followed by the AP/PA and the CE units (both with 95.8%), while the MA/PI unit showed 87.5% of correct classification.
Root 1 completely discriminated the AP/PA unit from the RJ unit. The MA/PI unit greatly overlapped with the AP/PA and partially overlapped with the CE unit, but showed almost no overlap with the RJ unit. The CE unit overlapped with all other units in our analysis (Figure 3).
Fig. 3. Projection of root 1 and root 2 of canonical analyses based on periotic–timpanic complex bone of Sotalia guianensis in four OTUs analysed: (Ampá/Pará (AP/PA), Maranhão/Piauí (MA/PI), Ceará (CE) and Rio de Janeiro (RJ)). Ellipses confidence: 95%.
According to the DFA, eight measurements showed significant differences among the OTUs: 1, 2, 3, 5, 7, 13, 14 and 19 (see Tables 1 & 3). The tympanic bone concentrated the variation identified and showed a pattern of growth from north to south along the geographical distribution on the Brazilian coast. The smallest specimens belonged to the AP/PA unit, on the north coast, and the largest specimens were found in the RJ unit, on the south-east coast.
Table 3. Coefficients of canonical analyses of 21 morphometric measurements of periotic–timpanic bone complex of Sotalia guianensis. Values in bold indicates the measures which best demonstrated the differences between OTUs (Measurements 1, 2, 3, 5, 7, 13, 14, 19).
DISCUSSION
Previous morphology studies focused on the comparison between the two currently known species of the genus Sotalia (Casinos et al., Reference Casinos, Bisbal and Boher1981; Borobia, Reference Borobia1989; Silva & Best, Reference Silva and Best1996; Ramos, Reference Ramos2001; Monteiro-Filho et al., Reference Monteiro-Filho, Monteiro and Reis2002, Fettuccia, Reference Fettuccia2006; Fettuccia et al., Reference Fettuccia, da Silva, Rocha and Simões-Lopes2012). Few have analysed variation within S. guianensis or its riverine ecotype S. fluviatilis. According to Borobia (Reference Borobia1989) and Monteiro-Filho et al. (Reference Monteiro-Filho, Monteiro and Reis2002), the skull of S. guianensis showed no morphometric variation, although the former author expected to find a growth pattern from north to south along the Brazilian coast. Casinos et al. (Reference Casinos, Bisbal and Boher1981) suggested that marine populations would show the variability found by Ramos (Reference Ramos2001), who demonstrated a decrease in skull length with the increase of latitude, comparing specimens only from the south-eastern coast of Brazil.
The growth of the tympanic bone in this study follows the growth of the skull in higher latitudes, as previously shown by molecular genetic markers (Cunha et al., Reference Cunha, da Silva, Lailson-Brito, Santos, Flores, Martin, Azevedo, Fragoso, Zanelatto and Sole-Cava2005; Cunha et al., Reference Cunha, da Silva, Solé-Cava, Ruiz-Garcia and Shostell2010). This variation may be attributed to Bergmann's rule. Data from more comprehensive studies corroborate the expectations of Borobia (Reference Borobia1989); that is, on average the skull of Sotalia guianensis populations increases with increasing latitudes. Casinos et al. (Reference Casinos, Bisbal and Boher1981), in the Macaibo Lake, Venezuela, recorded an average length of cranium of 335.5 mm, while Fettuccia (Reference Fettuccia2006), along the Brazilian coast, observed three different areas on the coast, and found cranium lengths of 351 mm in Amapá (northern region), 392 mm in Ceará (north-eastern region) and 387.5 mm in Santa Catarina (southern region). This is in disagreement with Ramos (Reference Ramos2001), probably because of the low sampling of some geographic areas, (see Parente et al., Reference Parente, Alves, Furtado-Neto and Monteiro-Neto1999) or differences in classification criteria of adults used by each author (Borobia, Reference Borobia1989; Fettuccia, Reference Fettuccia2006).
The growth pattern shown by the tympanic bone in this study follows the growth of the skull with increasing latitudes, which may be related to water temperature, a distribution pattern known as Bergmann's rule; that is, a homoeothermic animal that lives in cold waters is larger than one of the same species that lives in warmer waters (see Rensch, Reference Rensch1938). In fact, Schnell et al. (Reference Schnell, Douglas and Hough1986), studying patterns of geographic variations of Stenella attenuata, found strong correlations between skull measurements and environmental variables.
Cunha et al. (Reference Cunha, da Silva, Solé-Cava, Ruiz-Garcia and Shostell2010) suggested the existence of at least six distinct populations of Sotalia guianensis along the Brazilian coast using mtDNA analyses: Pará, Ceará, Rio Grande do Norte, Bahia, Espírito Santo and a southern/south-eastern area, from Rio de Janeiro to Santa Catarina. Their classification partially matches the three units suggested here: AP/PA = Pará, CE = Ceará and RJ = Rio de Janeiro. The differentiation between these OTUs suggests possible restrictions to gene flow, corroborating the hypothesis that S. guianensis had a series of allopatric expansions southwards along the Brazilian coast, where the distance acted as a geographic barrier restricting the gene flow between subsequent populations and favouring the emergence of distinct mtDNA haplotypes (Cunha et al., Reference Cunha, da Silva, Lailson-Brito, Santos, Flores, Martin, Azevedo, Fragoso, Zanelatto and Sole-Cava2005; Cunha et al., Reference Cunha, da Silva, Solé-Cava, Ruiz-Garcia and Shostell2010). According to Möller et al. (Reference Möller, Wiszniewski, Allen and Beheregaray2007), some physical characteristics of coastal areas (bays and estuaries), as well as site fidelity patterns and behaviour specializations, may cause genetic differences among dolphin populations.
The variation in the characters (see coefficients of variation, Table 2) increased from the AP/PA unit to the CE unit, after which it decreased substantially up to the RJ unit. Such phenotypic variation was also observed in molecular data (Cunha et al., Reference Cunha, da Silva, Lailson-Brito, Santos, Flores, Martin, Azevedo, Fragoso, Zanelatto and Sole-Cava2005; Cunha et al., Reference Cunha, da Silva, Solé-Cava, Ruiz-Garcia and Shostell2010), which indicated the populations in northern and north-eastern regions to be more variable, both genetically and morphologically, unlike the more homogenous south-eastern and southern populations. Ramos (Reference Ramos2001) explained that pattern as the result of gene flow between those populations, although Cunha et al. (Reference Cunha, da Silva, Solé-Cava, Ruiz-Garcia and Shostell2010) had a different explanation, assuming it was caused by a ‘founder effect’. In our study it was not possible to access specimens of S. guianensis from the southern region to evaluate their variability in the periotic–timpanic bone complex to reveal whether there was a similar pattern.
The traditional morphometrics of the periotic–timpanic bone complex was revealed to be an efficient tool to identify geographic variations of S. guianensis. The variation found between the OTUs corroborated previous studies in the literature, involving skull morphometrics and molecular data, confirming the existence of distinct population stocks in the species distribution along the Brazilian coast. Future research might analyse samples from the Brazilian southern region, and from other localities of the Atlantic coast of Central and South America to broaden the understanding of the species stocks in those areas.
ACKNOWLEDGEMENTS
The authors thank Instituto Ilha do Caju Ecodesenvolvimento e Pesquisa/Projeto Cetáceos do Maranhão (PROCEMA/ICEP) and Associação de Pesquisa e Preservação de Ecossistemas Aquáticos (AQUASIS).
FINANCIAL SUPPORT
This study was supported by grants from Programa PROF/Programa de Pós Graduação em Zoologia do Museu Paraense Emílio Goeldi/Universidade Federal do Pará and the William Rossiter – Cetacean Society International, granted to the first author.
