Size composition , monthly condition factor and morphometrics for fishery-dependent samples of Rioraja agassizi ( Chondrichthyes : Rajidae ) , off Santos , Southeast Brazil

Specimens of Rioraja agassizi were collected monthly off Santos, Southeast Brazil, between latitudes 23°37’S and 27°40’S, from March, 2005 to March, 2006. A total number of 278 males and 1023 females were captured. Observed depth range was 10120 m, being the species absent above 120 m. Sex ratio significantly favored females. The entire length range of R. agassizi was observed. Females ranged in length from 16.0 to 59.4 cm and males from 13.0 to 47.2 cm. Mean female length was significantly higher than that of males. For grouped length distributions, asymmetry patterns were observed in both sexes. Length-width functions were sexually dimorphic. Males with lengths < 24.5 cm were relatively wider than females. Length-total weight curves were significantly different than length-gutted weight ones, in both sexes. Females were heavier than males for a given TL class. The angular coefficient b was significantly <3 (negative allometry) in the males, and also for the females with exception of spring when b was >3 (positive allometry). The monthly variation of the condition factor showed significant differences in both sexes.


Introduction
Skates (Rajidae) are widely distributed from Artic to Antarctic waters and from coastal to ~3000 m deep, being rare in tropical waters especially in coral reefs (McEachran & Notabartolo di Sciara, 1995) and absent from the Pacific Plate and Red Sea (Springer 1982apud Last &Yearsley, 2002).The genus Rioraja Whitley, 1939, has a single species, the 'rio skate' Rioraja agassizi (Müller & Henle, 1841), commonly known in Brazil as 'raia-santa'.The absence of a caudal fin is a distinct character of this species (Figueiredo, 1977).
From a biogeographic perspective, R. agassizi belongs to the pattern of species endemic to the austral region of South America, being, along with the genus Atlantoraja, limited to the south-western region of the Atlantic Ocean (McEachran & Miyake, 1990;McEachran & Aschliman, 2004;Compagno, 2005).The species ranges from Espírito Santo State (Brazil) to Argentina, inhabiting coastal waters from the shore down to 130 m (Figueiredo, 1977).
Despite the uniform-brown coloration, there are two darker oval ocelli with a light or dark center, in the inner area of the pectoral fins, being highly variable in shape and with a lighter ring around them.In some individuals, this circle is surrounded by a single row of minute white points.Throughout the dorsal face of the skate, little, circular; white and symmetrically placed spots can occur, even in the pelvic fins area, in adult as well as young (Figueiredo, 1977;Barbosa & Gomes, 1998;Oddone, pers. obs.).The white ventral surface with white and light pink tones allows it to be distinguished from young Atlantoraja platana with which at first sight it could be mistaken.Light ventral colouration in R. agassizi is typical of species that inhabit shallow waters (McEachran & Notabartolo di Sciara, 1995).
Skates, as other elasmobranches, are long-lived, attain sexual maturity at large size and have low fecundity and long development period.These features make them susceptible to overexploitation (Walker, 1999).Casarini (2006) estimated the longevity of R. agassizi to be as 20 and 23 years respectively for males and females.It is important for management purposes to know where the center of the stock is and if it is subject to seasonal variation (Walker, 1999).In Southeast Brazil, only the large individuals of R. agassizi are landed and exported (Casarini, 2006).
The aim of this work was to describe the latitudinal and bathymetric distribution of R. agassizi in the fishing area of Santos and Guarujá's fishing fleet (southeast Brazil); to analyze the frequency composition of total length for both sexes and the relationships between total length and total width, and total and gutted weight; to establish the type of growth in weight with total length (allometry/isometry); and finally to determine the condition factor throughout a year.

Materials and Methods
Samples of Rioraja agassizi were obtained through a partnership system with fishermen as it is mostly a non commercialized species (Casarini, 1999).A total of eight fishing vessels (trawler and paired trawler), took part in the survey, bringing regularly (once or twice a month) samples of this species.Vessels' length over all ranged from 19.5 to 22.5 m and beam from 5.3 to 13 m m.The stretched mesh of the trawlnet varied from 45 to 90 mm and the groundrope ranged from 22 to 70 m.
Whenever possible, skippers identified the sample bag with a number corresponding with a trawling position, along with it characteristics, recorded on a board map.Data collected for each trawl were: date, depth, latitude/longitude, number of hauls and number of hauls with occurrence of the species.As the presence/absence of the species per haul was not regularly recorded, and since when recorded the information was imprecise, it was assumed that such data is unknown.
Biometric measurements were: total length (TL, cm), total width (DW, cm), total weight (T W , g) and gutted weight (body weight without digestive tract, liver and gonads and oviduct for the females, Gw, g).Linear measurements were made with 0.1 cm precision.Electronic scales used had 1 and 5 g precision respectively.
The Condition Factor (CF) was calculated monthly according to King (1995): CF T =Tw/TL b and CF G =Gw/TL b , where the sub-indexes T and G are the total and gutted weight, respectively, b is the angular coefficient of the potential curve adjusted to the relationship TL-Tw and TL-Gw.
The use of parametric/non-parametric tests was chosen by testing the normality and homogeneity of the variables' variance by Lilliefors' and Levene's tests respectively.To compare TL-Tw/Gw curves, data were log-transformed to become linear and the F-test was used (Souza, 1998).Sex ratios were tested using 2 .To assess the nature of the growth in weight with TL (isometry, allometry, etc.), the value of the angular coefficient b was compared with the theoretical value of 3.For this purpose, monthly samples of males and females were grouped by season, that is, summer (January, February, March), autumn (April, May, June), winter (July, August, September) and spring (October, November, December).Comparisons among CF obtained for all individuals by month, were performed using Kruskal-Wallis' H-test and post-hoc Mann- Whitney's U-test Bonferroni corrected (Sokal & Rohlf, 1995), using the freeware packet PAST (Hammer et al., 2001).In all cases the significance level considered was 0.05.

Results
A total number of 278 males and 1023 females of R. agassizi were collected in 25 fishing trips, carried out from March, 2005 to March, 2006 by 8 fishing vessels of Guaruja's fishing fleet.The species was captured monthly, during the entire period.Ob-served depth range was 10-120, being absent below 120 m (Table 1).Descriptive statistics for the monthly R. agassizi samples by sex are presented in Table 2.In the area covered by the vessels, the species occurred throughout the year (Fig. 2).
Considering grouped samples, the entire TL range of R. agassizi was observed.Females ranged in TL from 16.0 to   59.4 cm (n=1019; mean=46.9;SD=4.4) and males from 13.0 to 47.2 cm (n=278; mean=38.8;SD=3.4).Females were significantly larger than males (t=-28.5;d. f. =1295; p<0), and such pattern was maintained monthly.The difference in mean TL for males and females in presented in Table 2.For grouped TL distributions, asymmetry patterns were observed in both sexes, with asymmetry values different from 0 (Table 2), varying from -1.8 to -1.3 for males and females, respectively (Fig. 3).The asymmetry pattern remained when monthly samples were considered, though varying between positive and negative values.In the males, positive asymmetry was observed in May, June, October, November, December and March of 2006, whereas only in December for females.Mean TL varied between 43.4 (February) and 53.0 (October) for the females and between 36.7 (March) and 41.2 (May) for the males (Table 2).Total length/TW functions for R. agassizi were significantly different for both sexes (F (2,1291) =75.2; p<0.005) (Table 3).Males were wider than females for TL< 24.5 cm (TW=17.4cm).From that size up, females become wider than males for all TL classes (Fig. 4).Total length-Tw potential curves were significantly different than TL-Gw curves (Table 3, Fig. 5) in females (F (2,1002) =53.6; p=0.000) and males (F (2,533) =80.3; p=0.000).Females were heavier than males for a given TL class, considering both Tw and Gw (F (2,1279) =75.2; p=0.000;F (2,1132) =580; p=0.000, respectively).In Table 4, mean values for Tw and Gw by month and sex along with the respective TL range are presented.
Males grew in weight in a negative allometric manner with TL (b<3), when considering Tw as well as Gw, in the four seasons.The females, however, followed the same pattern of negative allometry, with exception of spring when their growth was positively allometric (b>3) (Table 5).
For the males, July was not included in the calculation of CF when considering Tw, and when considering Gw, June and July were not considered, because in those months the small sample led to outlier b.This was due to the fact that only males with large TL values occurred, and because of the absence of juveniles in the samples, the potential curve was not properly adjusted.Kruskal-Wallis' test detected significant differences in the CF throughout the year, when considering Tw (H (11, N= 277) =267.66,p=0.0000)) and Gw (H (10, N=260) =245.For the females, the CF significantly varied throughout the year, when considering both Tw and Gw (H (12, N=1007) =954.6, p=0.0000 and H (12, N=939) = 840.0,p=0.0000, respectively).The overall median CF value for Tw was 0.0074.Post hoc pairwise comparisons were significant (P<0.05) between all the months with exception of: February-March/2006; Ocotber-March/ 2006; October-February; May-December and March-November.Overall CF with Gw median's value was 0.016, with significant differences between all the month pairs (P<0.05), with exception of: February-March/2006; April-July; March-March/2006 and March-February (Fig. 6, Table 4).

Fig. 4.
Relationship of disc width (cm) as a dependent variable of total length (cm) for males (black symbols) and females (grey symbols) of Rioraja agassizi.3. Relationships analyzed for Rioraja agassizi: TL=total length (cm); DW=total width (cm); Tw=total weight (g); and Gw=gutted weight (g) for males (M) and females (F), along with the respective regression equation and correlation coefficient (R 2 ), sample number (n); pair of relationships compared (comparisons) by F-test (F), along with degrees of freedom of the numerator (n.d. f.) and denominator (d.d. f), probability (p).

Discussion
Information on habitat use in elasmobranchs is important not only with specific objectives such as determination of nursery area, but also for management purposes and species conservation.During the past decade, studies on habitat use have increased specially for commercially fished species and extracted by-catch, as is the case for the rajids (Simpfendorfer & Heupel, 2004).
Fisheries have negatively affected rajid fishes, in part because of their mostly sedentary nature even to the point of extinction, rendering many species in a critical situation (Brander, 1981;Casey & Myers, 1998;Dulvy & Reynolds, 2002).
For fishery-dependent samples, it is not possible to obtain a reliable estimation of CPUE because we have only access to a 'fleeting image' of the real population (Walker, 1999), which is especially true for this work.This is due to the contradictory and low quality and reliability of the information provided by the skippers regarding the magnitude of total catch and discard.However, because of the monthly presence of R. agassizi in the samples, it is assumed that it is present in the area throughout the year without variation in its abundance.
Reinforcing this hypothesis, Ponz Louro (1995) observed that after Psammobatis glandissimilis, R. agassizi is the most abundant rajid in Ubatuba, São Paulo State, and Figueiredo (1977) had already considered the species as 'common' in the same area.For the Uruguayan inner shelf, R. agassizi together with Atlantoraja castelanui and A. cyclophora were distributed in 30-60% of the area with abundances between 1000 and 5000 kg/ square nautical mile (Paesch, 2006).
Most skate species have a limited, well-defined distribution, and there are very few ranging species (Walker, 1999).It is known that R. agassizi ranges from Argentina to Rio de Janeiro (Brazil) (Figueiredo, 1977).Though precise limits of its distribution remain unknown to date, R. agassizi is probably distributed continuously along the cited range, because it is recorded in southeast and south Brazil, Uruguay and Argentina (Figueiredo, 1977;Menni & Lopez, 1984;Vooren, 1998;Meneses & Paesch, 2003).
With regard to depth of occurrence, Menni & Lopez (1984) classified R. agassizi as belonging to the 'inner shelf mixed fauna' fish group (Menni & Gosztonyi, 1982).McEachran & Miyake (1990) stated that R. agassizi occurred at depths down to ~50 m.However, in this study, R. agassizi was recorded at maximum depths of 120 m and Menni & Lopez (1984) stated its depth range was from 22 to 89 m.For the south Brazilian area, Vooren (1998) confirmed its presence at depth of 20-50 m, in agreement with Compagno (2005) who considers this species as demersal in the shelf of it distribution range.For Vooren op.cit., R. agassizi is a species of constant presence in south Brazil, which completes it cycle entirely in these waters without seasonal variations in its abundance, being one of the 27 elasmobranch species recorded in the South Brazilian shelf (Vooren & Klippel., 2005).Menni et al. (1981) recorded the southernmost limit for this species in the Golfo de San Matías (Argentina, at 41°30'), where it is fished at up to 181 m deep, while Figueiredo (1977) considered that the species occurred at a maximum depth of 130 m in southwest Brazil.In Ubatuba, São Paulo's State's littoral area, R. agassizi occurs only in the coastal area at down to 50 m, with similar proportions of males and females during the whole year (Ponz Louro, 1995).Casarini (2006) observed that R. agassizi was the most abundant species in the depth range of 10-50 m in southeast Brazil.
Sexual segregation was not observed in this work and sex ratio favored the females.The same was observed by Casarini (2006) for the same species in the same area.However, unisexual aggregations of Raja clavata were recorded off the coast of East Anglia by Holden (1975), and could be common for rajids (Steven, 1933).For the southwestern Atlantic, Sympterygia acuta, S. bonapartii and Atlantoraja cyclophora do not segregate by sexes (De Queiroz, 1986;Oddone & Vooren, 2004, 2005).Ebert (2005) studied eight Bathyraja species, recording no sexual segregation in six of them and a sex ratio favoring males in only two, along the eastern Bering Sea continental slope.
Data on the distribution of sizes in the catch can indicates a population's status or the risk of overfishing, as too Fig. 5. Relationships between total (black symbols) and gutted weight (g) (grey symbols) for male (above) and females (below) of Rioraja agassizi.small fish may mean not enough are managing to spawn (Hilborn & Walters, 1992).The same authors stated that young fish are less vulnerable to fishing gears than older ones.However, Oddone (unpublished data) recorded the presence of young and even neonates of R. agassizi and Atlantoraja spp., a fact that to date seemed to be unusual in rajids and that reflects the vulnerability of these life stages to bottom trawl fisheries.For the South Brazilian shelf Vooren & Klippel (2005) did not record neonates of R. agassizi, at depths of 7-20 m.Patterns of negative asymmetry in TL distributions, indicate that both for males and females, more specimens occurred with high values of TL and a few with low values of TL.A positive pattern of asymmetry would indicate just the opposite.
Negative asymmetric distribution of frequencies of TL for both sexes were observed in Atlantoraja cyclophora (Oddone & Vooren, 2004).
The largest specimen of R. agassizi recorded by Figueiredo (1977) was a 49 cm adult male.In this work the largest sizes observed were 59.4 cm for the females and 47.2 cm for the males, confirming the sexual dimorphism in size.In sharks, females attain maximum size and older age than males (Cortés, 2000).Several Rajidae are sexually dimorphic too, regarding size (Jardas, 1973;Nottage & Perkins, 1983;Oddone & Vooren, 2004).Exceptionally, Ebert (2005) observed that in Bathyraja interrupta, B. minispinosa and B. parmifera, males attained a larger size than females, while in B. lindbergi, B. maculata and B. trachura both sexes grew to approximately the same size, observing that only in B. aleutica and B. taranetzi did females achieve noticeably larger sizes than males.Cousseau et al. (2000) in Argentinean waters recorded contrasting maximum TLs of 61.8 cm for a male and 70.8 cm for a female of R. agassizi.This suggests a size gradient in the TL for R. agassizi, which increases southward, and which was already noted in teleost fishes (Wootton, 1996).
Sexually dimorphic length-weight relationships, were observed in A. cyclophora (total and gutted weight), Psammobatis normani, P. extenta and S. bonapartii (Braccini & Chiaramonte, 2002;Mabragaña et al., 2002;Oddone & Vooren, 2004;Mabragaña & Cousseau, 2004), though Raja miraletus showed an opposite pattern (Ungaro, 2004).In this work, it was observed that throughout the year, males and females grew with negative allometry with TL, with exception of spring when females followed a pattern of positive allometry.This could be due more to food availability in spring in the area than to reproduction, as in this species vitellogenesis, ovulation and egg-laying occur throughout the year (Oddone et al., 2007).Possible effect of stomach contents on the weight of the skates were avoided by calculating b with gutted weight, so differences in body weight may well reflect differences in body condition.
The correlation coefficient of the ratio TL-Gw was particularly lower in the males (0.767).This would indicate that, in spite of the fact that such correlation was significant for the potential curve adjusted, a trend in the Gw to increase more slowly with total length exists, when compared with Tw, for a given TL value.This fact was not evident in the females and could then be related to the sexual dimorphism in the body shape that is observed in rajid species, i.e., males present a striking concavity in the anterior part of the disc, from the snout to the tip of the pectoral fins, which is not observed in the females.When considering Tw, such trend is likewise masked by the effect of the viscera and especially stomach content, reflecting a more typical potential curve pattern because of the higher value of the angular coefficient b.
In summer, b was estimated to be 2.984 and 2.888 for total and gutted weight respectively.Although these values were statistically different from 3, the sample number (71) may be considered, because the larger the sample size, the more the significance of the differences, and therefore for this season growth could have been isometric as well.Braccini & Chiaramonte (2002) recorded isometric growth for males and positive allometric growth in females of P. extenta.
The condition factor can provide information on the general condition of fishes in the habitat they live, as well as indicate phenomena such as spawning period, alterations in population density and feeding conditions (Braga, 1986).In R. agassizi, CF varied significantly throughout the year, reflecting variations in the availability of food and mainly in the reproductive events, such as egg-laying, vitellogenesis and production of the egg cases.Seasonal variation of the diet seems to be associated with the availability of the prey, whose distribution and abundance are related to the dynamics of the water masses of the region (Muto et al., 2001).Loefer & Sedberry (2003) observed that in Rhizoprionodon terraenovae the condition factor in both sexes was low during the mating season, an indication of probable stress.
The CF may also be influenced by the liver weight's variation throughout the year, as the liver is strongly correlated with reproduction in rajids (Oddone & Velasco, 2006) and elasmobranchs in general.Sexual dimorphism in the liver weight was noted already in several rajid species, and this is related to the energy expenditure of females during vitellogenesis, oocyte maturation and gestation.Females store great quantities of lipids in the liver during the pre-vitellogenic phase of the reproductive cycle (Lucifora et al., 2002), and the production of vitellogenin (the precursor to egg yolk proteins), takes place in the liver (Gelsleichter, 2004).In Raja asterias, for instance, an increase in the gonad weight is accompanied by the storage of substances in the liver (Capapé, 1980).In males of this species, two well defined annual peaks in the gonadosomatic index exist, while in the females of R. agassizi and others such as A. cyclophora (Oddone & Vooren, 2005), such peaks are not that evident, confirming the absence of seasonality in the reproductive variables, such as gonad weight, egg-laying frequency throughout the year, etc.This observation may well explain the rather greater variation for males in the CF (with regard to Gw), by the end of the year (December) when compared with females.
A growth pattern where males are wider than females in the first life stages, turning the females wider than males from a given TL, may be typical of at least some rajids and was also observed in A. cyclophora (Oddone, 2003;Oddone & Vooren, 2004).

Fig. 1 .
Fig. 1.Map of the study area, southeast Brazil, south-western Atlantic Ocean.Symbols represent all the fishing hauls from where samples of Rioraja agassizi were collected.

Fig. 2 .
Fig. 2. Maps of Southeast Brazil, showing the fishing hauls by month, conducted by a total of eight fishing vessels.Symbols represent the fishing hauls from where samples of Rioraja agassizi were collected (dotted circles), or when this information was not available, the total hauls recorded by the skippers (crosses).

Fig. 3 .
Fig. 3. Absolute frequency of total length (cm) for males (full line) and females (dotted line) for grouped samples of Rioraja agassizi.

Fig. 6 .
Fig. 6.Median condition factor by month (from March, 2005 to March, 2006) considering total (black line) and gutted weight (g) (grey line) for males (above) and females (below) of Rioraja agassizi.Vertical bars represent the data interval.

Table 1 .
Fishing trips carried out by eight fishery vessels of Guaruja´s (Southeast Brazil) fishing fleet conducted from March of 2005 to March of 2006 from which samples of Rioraja agassizi were obtained; name of the fishing vessels, date (days, month, year), number of hauls and number of hauls with presence of the species (*), depth of the hauls and depth with presence of the species (*).Days marked with (*) represent single days when Rioraja agassizi were collected, when this information was available.*dates where samples of Rioraja agassizi were collected; total trip days are unknown.

Table 2 .
Number of males and females of Rioraja agassizi captured by month; total length range (cm), mean total length (mean), standard deviation (SD) and sex ratio males:females (M:F), along with 2 test's value, degrees of freedom (d.f.) and probability (p).

Table 5 .
Seasonal variation of the angular coefficient estimate (b) for males and females of Rioraja agassizi; standard error of the regression total length-total/gutted weight, t 1 =t-test value of the regression and p 1 =probability of the regression regarding b, correlation coefficient of the regression, d. f. =degrees of freedom and probability (p 1 ), t 2 =t-test value of the comparison with the theoretical value 3 and probability (p 2 ) and the nature of growth according to b.

Table 4 .
Monthly range, standard deviation (SD), and sample size (n) for total weight (Tw), gutted weight (Gw), angular coefficient (b), and CF, calculated with both total and gutted weight (g) by total length range for males and females of Rioraja agassizi.