MORPHOLOGY OF THE FRANCISCANA ( PONTOPORIA BLAINVILLEI ) OFF SOUTHEASTERN BRAZIL : SEXUAL DIMORPHISM , GROWTH AND GEOGRAPHIC VARIATION

 Variations in body and skull morphology may exhibit geographic differences evidencing distinct population stocks. The objectives of the present study were to analyze such variation to test the hypothesis of a disjunct distribution of the franciscana (Pontoporia blainvillei) off the southeastern coast of Brazil. Body length and 39 cranial variables were measured from bycaught animals to considered sexual, ontogenetic and geographic variations. The areas studied were Espírito Santo (ES) (18o30S-19o40S), northern Rio de Janeiro (NRJ) (21o35S-22o25S) and São Paulo (SP) (23o30S-25o30S). Franciscanas from NRJ and SP presented significant sexual dimorphism, with the means for the metric characters larger for females than for males. Sexual dimorphism for franciscanas from ES was not examined due to limited sample size. The growth pattern for body and skull did not indicate clinal variation. The asymptotic values obtained for franciscanas from SP were smaller than the values obtained for franciscanas from NRJ and ES. Canonical discriminant analysis of the cranial metric characters indicated significant differences among the three geographic areas. Differences between areas ES and NRJ accounted for 85% of the variation (axis 1). The remaining 15% (axis 2) was due to difference between the area SP from the others. The geographic variation supports the hypothesis stock division in southeast Brazil; allopatry might be present. Therefore, three franciscana stocks from the southeastern coast of Brazil should be considered distinct for conservation and management actions. Resumo  Variações no padrão morfológico do corpo e do crânio podem apresentar diferenças geográficas evidenciando estoques populacionais distintos. O objetivo do presente estudo foi estudar tais variações para testar a hipótese de distribuição disjunta para a toninha (Pontoporia blainvillei) na costa sudeste do Brasil. Para tanto, o comprimento do corpo e 39 caracteres cranianos foram determinados, considerando-se as variações morfológicas sexual, ontogênica e geográfica. As áreas estudadas foram Espírito Santo (ES) (18o30S-19o40S), norte do Rio de Janeiro (NRJ) (21o35S-22o25S) e São Paulo (SP) (23o30S-25o30S). Toninhas do NRJ e SP, apresentaram dimorfismo sexual significativo, com as médias obtidas para os caracteres métricos maiores para as fêmeas do que para os machos. O dimorfismo sexual não pode ser testado para o ES devido a problemas amostrais. Os padrões de crescimento para o tamanho corpóreo e craniano indicaram que não há uma variação clinal para a espécie. Os valores assintóticos obtidos foram menores para os espécimens do SP em relação aos valores obtidos para os espécimens do NRJ e ES. A análise discriminante canônica para os caracteres métricos do crânio indicou diferença significativa entre as três áreas geográficas, sem sobreposição para os estoques analisados. Diferenças entre as áreas ES e NRJ foram explicadas por 85% da variação (eixo 1). O 15% restantes da variação (eixo 2) foram responsáveis pela diferença entre a área SP das demais áreas. A variação geográfica observada apoia a hipótese de distribuição disjunta no sudeste do Brasil; uma alopatria pode estar presente. Desta forma, os três estoques de toninhas na costa sudeste do Brasil devem ser considerados distintos para fins de conservação e manejo.


Introduction
The franciscana, Pontoporia blainvillei (Pontoporiidae), is restricted to the coastal Atlantic waters of South America from Itaúnas (18º30S), Espírito Santo State, southeastern Brazil to Nuevo Gulf (42º35S), Argentina (Siciliano, 1994;Crespo et al., 1998).Pinedo (1991) proposed at least two different forms of the franciscana based on osteological differences: one found to the north (smaller animals) and the other to the south (larger animals) of Santa Catarina State (~29ºS).Ramos et al. (2000a) presented data on age and growth for franciscanas from Rio de Janeiro State (~22ºS) and found asymptotic body lengths smaller than those found by Kasuya and Brownell (1979) for franciscanas from Uruguay (~34ºS).This corroborates the study of Pinedo (1991) regarding the existence of at least two different phenotypic forms.
Analyses of the mtDNA from franciscanas of each of the geographic forms provided evidence of the existence of one genetic population involving dolphins from Rio Grande do Sul State (~33ºS), Uruguay and Argentina (~42ºS).No haplotypes from these locations were shared with dolphins from Rio de Janeiro State, supporting the hypothesis of at least two genetically distinct populations of franciscana (Secchi et al., 1998;Lázaro and Lessa, 2000;Hamilton et al., 2000).Siciliano et al. (in press) proposed two gaps in the distribution of the northern population.The first gap is located between Regência (19º40S), Espírito Santo State and Atafona (21º35S), northern Rio de Janeiro State and the second between Macaé (22º25S), Rio de Janeiro State and Ubatuba (23º30S), northern São Paulo State.Growth data suggest differences in morphology within northern franciscanas.The asymptotic lengths obtained by Rosas (2000) for franciscanas from São Paulo and northern Paraná States were smaller than the growth 1 Universidade Estadual do Norte Fluminense, Laboratório de Ciências Ambientais, Av.Alberto Lamego, 2000, Campos dos Goytacazes, RJ, 28015-620, Brazil.
Considerable effort has been given globally to managing impacted populations of small cetaceans.The first step in this management is to define the populations involved.One tool for defining mammal population is the analysis of geographic variation in morphology (Perrin, 1984).Three relevant aspects should be considered.First, quantification of the degree of sexual dimorphism found within populations is advisable before studying geographic variation in order to avoid attributing significant morphometric differences to inappropriate factors (Hersh et al., 1990).Second, knowledge of the differences in postnatal growth between distinct geographic areas helps in the identification of population plasticity (Calzada et al., 1997).Finally, geographic variation provides a basis for the description of stock units used to assess and manage dolphins (Schnell et al., 1986).
The objectives of the present study were to analyze variations in cranial morphology to test the hypothesis of stock division along the southeastern coast of Brazil.Body length and 39 cranial variables were analysed to consider sexual, ontogenetic and geographic variation.

Study area and sample
The sampling areas were defined according to known occurrence of the franciscana off the southeastern coast of Brazil.1).
The study specimens were accidentally caught in gillnet fisheries or found stranded.A total of 262 franciscanas were examined from different collections; 14 from area ES, 143 from area NRJ and 105 from area SP.

Age estimation and morphometric characters
Age was estimated in 205 franciscanas by counting the number of growth layer groups (GLGs) in the dentine and cement.We adopted the GLG pattern described for the species by Pinedo and Hohn (2000)   stained layer demarcated the boundary with unstained layer of the subsequent GLG.In the cementum, a complete GLG was composed of one narrow stained layer and a wide weakly stained layer.The method of preparation of decalcified thin and stained sections of the teeth for examination under an optical microscopy was used following the recommendations of Perrin and Myrick (1980) and Hohn et al. (1989).Body length (BL) of 189 specimens was measured along the longitudinal axis of the body from the tip of the upper jaw to the notch of the flukes (Norris, 1961).A total of 190 skulls from franciscanas were examined by one of the authors (RMAR) for cranial metric characters following Perrin (1975) and Schnell et al. (1985) with some modifications (Table 1).Due to the small number of skulls for which widths of the rostrum could be measured, four variables were deleted from the analysis (WR1/4, WR1/2, WPMx1/2 and WR3/4).A total of 35 variables remained from the original 39.Acronyms are shown in Table 1.The franciscanas were classified as immature or mature according to the relationship between age and body length.
For franciscanas from northern Rio de Janeiro State, males were considered mature if age was ≥2 years and body length ≥113.0 cm; for females, the criteria were ≥3 years and ≥130.0 cm (Ramos et al., 2000a) length was ≥112.0 cm and females, ≥122.0 cm, both with age ≥4 years (Rosas, 2000).

Sexual dimorphism
The franciscana has been considered a dimorphic species in body length and skull size, with females larger than males (Pinedo, 1991).Ramos et al. (2000a)

Geographic variation
Geographic variation in cranial morphology was examined for adult specimens only.Missing values in the data matrix were estimated with a non-linear Gompertz model or as the mean value by age class and sex to variables not fitted by non-linear model.An analysis of variance (ANOVA) was carried out to test interaction between sex and geographic area for each metric character using Statistic Program 5.5 for Windows.The Bonferroni correction was applied and only P < 0.001 was significant.The method of adjusting for sex differences described by Schenell et al. (1985) was applied.Correction terms were obtained to adjust measurements of the larger sex downward and the smaller sex upward, thus producing sex-adjusted or zwitter measurements.As a result, we were able to combine specimens for both sexes in an overall analysis of geographic variation.
A canonical discriminant analysis was applied to identify the subset of variables that, taken in linear combination, show the greatest degree of geographic variation (Afifi and Clarck, 1990).The method was applied with Statistic Program 5.5 for Windows.

Sexual Dimorphism
Only one mature female was obtained for the area ES (Espírito Santo State), making inferences of sexual dimorphism impossible for franciscana in this area (Table 2).
In area NRJ (northern Rio de Janeiro State), females were larger than males for all the 36 variables (Table 3).The discriminant separation was significant (Wilks Lambda = 0.04738; ~F36,16 = 8.9357; P < 0.001).Four metric variables (BL, MaDLTF, WPS, GHFM) best showed the discriminant between the sexes.Of the 53 individuals in the analysis, 96.2% were correctly classified a posteriori using the classification function (Table 4).Because of this high reliability, we sexed two franciscanas of unknown sex with the classification function.
In area SP (São Paulo State), females were larger than males in 92.5% of the variables (Table 5).The discrimination was significant (Wilks Lambda = 0.20333; ~F3.21 = 27.427;P < 0.001).Three metric variables (BL, CBL, VEHB) best showed the discriminant between the sexes.All the 25 franciscanas were correctly classified a posteriori by the discriminant function (Table 4).Of the 20 franciscanas of unknown sex, 14 (70%) were classified by the discriminant function.Six franciscanas were not classified by the function due to absence of measured cranial variable used in the function (i.e.BL, CBL, VEHB).

Growth
About 55% of the mean values for cranial variables were smaller for franciscanas from SP than franciscanas from ES and NRJ (Tables 2, 3 and 5).The coefficient of variation was smaller than 20% for all variables analysed.Therefore, the mean values of body length and cranial metric characters were considered suitable for growth pattern analysis.The growth parameters are presented in Tables 6, 7 and 8.The growth curve for franciscanas from area ES was adjusted for males because only one female was obtained for this area (Table 6).The asymptotic value for body length (BL) was reached at about three years of age.One cranial variable of the feeding apparatus (WRB) reached asymptotic value earlier, at about four years.The other cranial variables reached asymptotic values at between five and six years (see Table 6).Only one variable of the feeding apparatus (MiDLTF) attained the asymptotic value after seven years.
In area NRJ, the asymptotic value for females was larger than for males for all the variables analysed (Table 7).Males attained the asymptotic value earlier than females for 69% of the variables.Body length (BL) attained the asymptotic value at four years for males and five years for females.Most of the cranial variables of the braincase, breathing-sound and feeding apparatuses in males and females reached the asymptote at about six years (see Table 7).However, some variables of the breathing-sound and feeding apparatuses of the females reached the asymptote one year later, at seven years.In contrast, some variables of the breathing-sound and feeding apparatuses in males reached the asymptote one year earlier, at five years.The hearing apparatus (LLTC, LRTC and WPS) was the last to reach asymptotic values, at seven years in males and eight in females.
In area SP, the asymptotic values for females were larger than males for all the variables analysed (Table 8).Males attained the asymptote earlier than females in 65.4% of the variables.Body length (BL) reached the asymptotic value at four years for males and five years for females.Most of the cranial variables of the braincase, breathingsound, feeding and hearing apparatuses in males and females attained the asymptote at four years or at six years of age (see Table 8).Some variables of the breathing-sound and feeding apparatuses of females attained the asymptote at about seven years.
The ANOVA for the interaction between geographic variation and sex was significant for three variables only (8.6%): VEHB, GLLP and DRIN (Table 9).The interaction indicated that sexual dimorphism for these characters varied geographically and for this reason, these variables were excluded from analysis of geographic variation.Sexual dimorphism was independent of geographic variation for 91.4% of the variables (Table 9).The correction factor presented in Table 10 was subtracted from the values for males and added to the values for females.Therefore, males and females were analysed simultaneously.
The canonical discriminant analysis of the 32 cranial variables uncovered a significant difference among geographic areas (Wilks Lambda = 0.0428917; F64.100 ≅ 5.982051; P < 0.001).The discriminant function classified correctly a posteriori one hundred percent of the 84 specimens analysed (Table 11).Evaluation of the standard coefficients of the canonical discriminant analysis indicated that five variables best represented the difference among the geographic areas.These variables were of the breathing-sound (DREN), braincase (CBL) and feeding (GPostOW, GLLPTF, LLLTR) apparatuses (Table 11).Discriminant axis 1 explained 85% of the variance among the geographic areas, representing mainly the difference between areas ES and NRJ.Discriminant axis 2 explained the remaining 15% of variance, representing the difference between area SP and the other areas.Figure 2 presents the projection of the 84 specimens in the canonical axes.No overlap occurred among the three geographic areas.

Sexual dimorphism
The sexual dimorphism observed in the present study, with values for females larger than for males in all variables analysed for three geographic areas, corroborates other studies that considered the species highly dimorphic (Pinedo, 1991;Ramos et al., 2000a).
Females larger than males are also observed in other species of small cetaceans, such as the harbour porpoise (Phocoena phocoena) and the vaquita (P.sinus) (Hohn et al., 1996;Read and Tolley, 1997).Sexual dimorphism can be manifested in a variety of ways, from the possession of secondary sexual characters to differences in size (Ralls, 1977).
Sexual dimorphism has been associated with a differential investment of energy in growth, reflecting divergent reproductive strategies in the two sexes (Calzada et al., 1997).Other possible selective factors are female dominance over males, the reduction of inter-sexual competition for food, more intense competition for some resource, such as food, by females than by males, and the fact that a big mother is often a better mother.The factors are clearly not mutually exclusive: more than one of them may affect a single species (Ralls, 1976).The sexual dimorphism observed in franciscanas might be related to reproductive strategies or availability of resources.

Growth
No differentiation of growth pattern among the functional apparatuses was observed, suggesting that the braincase, breathing-sound and feeding apparatuses present a similar development with age.An exception was observed for the hearing apparatus development in franciscanas from northern Rio de Janeiro State, which presented a late physical LAJAM 1(1): 129-144, Special Issue 1, 2002  maturity in comparison to the other functional apparatuses.The early development of the hearing and visual apparatuses observed in the franciscana could be related to the preference of the species for estuarine areas with high turbidity.
The results on growth pattern of body size and cranial characters for franciscanas in the present study do not indicate that there is a clinal variation for the species.
Initially, the hypothesis of clinal variation was suggested by Pinedo (1991) and supported by Ramos et al. (2000a) based on data obtained for the extremes of the species geographic distribution.Rosas (2000) and Ramos et al. (2000b) presented the first evidence that the variation is not clinal.The asymptotic values for body size in franciscanas from São Paulo (113.3cm for males and 128.9cm for females; Rosas, 2000) were smaller than values obtained for franciscanas from northern Rio de Janeiro (117.1cm for males and 144.7cm for females; Ramos et al., LAJAM 1(1): 129-144, Special Issue 1, 2002  Kasuya and Brownell, 1979).The differences observed in the asymptotic values and size at birth between the geographic areas in the present study are consistent with these conclusions.Franciscanas may present different characteristics within population stocks independent of latitudinal distribution.
The franciscana also exhibits geographic variation in reproductive strategy.The northern franciscanas reproduce year round, while southern franciscanas have seasonal reproduction (Harrison et al., 1981;Pinedo et al., 1989;Danilewicz, et al., 2000;Ramos et al., 2000a).The results of the present study suggest that there is a difference in reproductive investment in the northern franciscanas.The variation in resource allocation between somatic and reproductive investment might cause alteration in reproductive periodicity, as well as morphological differentiation (Stearns, 1992).

Geographic variation
The geographic variation in cranial morphology observed in the present study confirms the hypothesis that the species has more than two population stocks.The following management subdivision has been proposed by The geographic variation revealed here is consistent with the hypothesis of disjunct distribution in southeastern Brazil.The areas may not overlap (see Figure 2).The hypothesis of a disjunct distribution for the northern population was proposed by Siciliano et al. (in press) based on two gaps in the distribution in southeastern Brazil.The authors related these gaps mainly to two factors: lack of fluvial drainage, which would increase the abundance of trophic resources and narrowing of the continental shelf decreasing habitat areas.
The preference of the franciscana for estuarine areas with high turbidity can be related to optimal conditions for feeding (Siciliano andSantos, 1994, Crespo et al., 1998).
The franciscana can be considered specialists in trophic terms, obtaining greater success in areas that present favorable conditions for its feeding (Di Beneditto, 2000).
The trophic specialization can act to limit the habitat occupied (Crespo et al., 1998).
The franciscana is characteristically coastal, living within 30 nautical miles of the coast and up to depths of 30m (Praderi et al., 1989).Water depth might also be considered a factor limiting its distribution (Secchi and Ott, 2000).Thus, in areas where the continental shelf is wider, the species will have more extensive habitat and, possibly, trophic resource.In contrast, in the area of Arraial do Cabo and southern Espírito Santo State, a narrowing of the continental shelf and an increase in water clarity are present (Szpilman, 1992;Muehe and Valentini, 1998).These factors may cause the gaps in the occurrence pattern of franciscana off the southeastern coast of Brazil (Siciliano et al., in press).
The two gaps observed in franciscana distribution may prevent gene flow, favoring the isolation of populations and perceptible morphologic variation.Variation in body size in cetaceans separated by little geographic distance has been observed in several species, such as the harbour porpoise and striped dolphin (Stenella coeruleoalba) (Perrin, 1984;Calzada and Aguilar, 1995;Gao and Gaskin, 1996).
The factors that cause this morphological variation may include reproductive isolation due to a fragmentation of the species distribution.
The geographic variation observed in the present study supports the hypothesis of disjunct distribution of the franciscana in southeastearn Brazil.Understanding of phenotypic variation is important in the formulation of conservation strategies, particularly in relation to coastal species, which present limitations in the preferred habitat and greater vulnerability to activities such as fishing operations and to pollution (Borobia et al., 1991;Secchi et al., in press).The results of the present study suggest that the franciscana stocks off the southeastern coast of Brazil should be considered distinct for conservation and management actions.

Figure 1 .
Figure 1.Areas of occurrence of Pontoporia blainvillei in southeastern Brazil.The arrows indicate the three geographic locations considered in the present study: area ES -Espírito Santo State, from Itaúnas to Regência (18º30S-19º40S); area NRJ -Northern Rio de Janeiro State, from Atafona to Macaé (21º35S-22º25S); and area SP -São Paulo State, from Ubatuba to Cananéia, including the Baía de Paranaguá, in the northernmost portion of Paraná State (23º30S-25º30S).

Table 2 .
Mean, standard deviation (SD) and coefficient of variation (CV) of metric variables for mature male and female franciscanas (Pontoporia blainvillei) from Espírito Santo State.
1 Acronyms in Table1.Body size (BL) was measured in cm and the other variables in mm.

Table 3 .
Mean, standard deviation (SD) and coefficient of variation (CV) of metric variables for mature male and female franciscanas (Pontoporia blainvillei) from northern Rio de Janeiro State.Body length at zero age predicted by the Gompertz model for male franciscanas was 71.0cm for area ES, 70.3cm for area NRJ and 68.5cm for area SP.For females, body length at zero age was predicted at 70.2cm for area NRJ and 67.3cm for area SP.

Table 5 .
Mean, standard deviation (SD) and coefficient of variation (CV) of metric variables for mature male and female franciscanas (Pontoporia blainvillei) from São Paulo State.
1 Acronyms in Table1.Body size (BL) was measured in cm and the other variables in mm.

Table 4 .
Classification function from discriminant analysis of mature male and female franciscanas (Pontoporia blainvillei) from northern Rio de Janeiro (NRJ) and São Paulo (SP) states.

Table 6 .
Growth parameters obtained by Gompertz model fitted to metric variables-at-age of male franciscanas (Pontoporia blainvillei) from Espírito Santo State.

Table 7 .
Growth parameter values obtained by Gompertz model fitted to metric variables-at-age of male and female franciscanas (Pontoporia blainvillei) from northern Rio de Janeiro State.

Table 8 .
Growth parameter values obtained by Gompertz model fitted to metric variables-at-age of male and female franciscanas (Pontoporia blainvillei) from São Paulo State.

Table 9 .
Analysis of variance of the metric variables for areas, for sexes and for interaction between area and sex in franciscanas (Pontoporia blainvillei) from southeastern Brazil.

Table 11 .
Classification function and standard coefficients of the canonical discriminant analysis of franciscanas (Pontoporia blainvillei) for three geographic areas, Espírito Santo (ES), northern Rio de Janeiro (NRJ) and São Paulo (SP) states, southeastern Brazil.The bold values indicate the variables that best showed the differences between geographic areas.