Temperature-sex determination in Podocnemis expansa ( Testudines , Podocnemididae )

1. Instituto Chico Mendes de Conservação da Biodiversidade (ICMBio), Rua 229, 95, Setor Universitário, 74605-090, Goiânia, GO, Brazil. (kbonach@yahoo.com.br) 2. Universidade Federal de Tocantins, UFT, ALCNO 14, Av. NS 15, s /n, Bloco II, Colegiado de Engenharia Ambiental, 77010-970, Palmas, TO, Brazil. (malvasio@uft.edu.br) 3. Departamento de Patologia, FMVZ, Universidade de São Paulo, Av. Prof. Dr. Orlando Marques de Paiva, 87, Butantã, 05508-900, São Paulo, SP, Brazil. (ermatush@usp.br) 4. Laboratório de Ecologia Isotópica, Universidade de São Paulo, Caixa Postal 96, 13416-000, Piracicaba, SP, Brazil. (lmverdade@usp.br)

The giant South American river turtle -Podocnemis expansa (Schweigger, 1812) -is one of the largest freshwater turtles of the world.It occurs in rivers and lakes of the Amazon and Orinoco river basins (Smith, 1979).The females reach 80 cm of carapace length and 60 kg of body-mass, whereas the males are smaller (Pritchard & trebbau, 1984).
Decades of overexploitation resulted in dramatic population declines in many rivers of the Amazon basin (Smith, 1979).For this reason, the Brazilian government established a "ranching" program (sensu hutton & Webb, 1992) based on the protection of nesting habitats, translocation of eggs, commercial use of captive-reared turtles, and release of hatchlings in the wild near their nesting areas of origin (Fao, 1988;ibama, 1989).
The manipulation of nests and eggs can affect the sex of embryos, since the species presents temperature-sex determination (TSD) (alho et al., 1985).For conservation purposes, in theory, a 1:1 sex ratio would be more convenient (davenPort, 1997).However, for commercial purposes ranchers could desire a female-biased population because females are larger than males, produce eggs, and a few males can inseminate a great number of females (vogt, 1994).
Changes on the sex ratio of a population could in theory affect its dynamics within a few generations (caughley, 1977).In addition, temperature of incubation can also affect other embryonic aspects in reptiles such ABSTRACT.This study has been carried out at the central region of the Araguaia river on the border between the states of Goiás and Mato Grosso in the Brazilian Amazon Basin from September to December 2000.We recorded temperature fluctuation, clutch-size, incubation period and hatching success rate and hatchlings' sex ratio of five nests of Podocnemis expansa (Schweigger, 1812).Despite the relatively small sample size we infer that: a) nests of P. expansa in the central Araguaia river have a lower incubation temperature than nests located further south; however, incubation period is shorter, hatching success rate is lower and clutch-size is larger; b) Podocnemis expansa may present a female-malefemale (FMF) pattern of temperature sex-determination (TSD); c) thermosensitive period of sex determination apparently occur at the last third of the incubation period; and, d) future studies should prioritize the relationship between temperature variation (i.e., range and cycle) and embryos development, survivorship and sex determination.
as incubation period, integument pigmentation and post-hatching growth rate (Joanen & mcneaSe, 1987;deeming & FerguSon, 1989;Webb & cooPer-PreSton, 1989).For these reasons, this study aims to determine the temperature of incubation in natural nests of the species and its pattern for the sex determination of embryos.

MATERIAL AND METHODS
We conducted this study in the central region of the Araguaia river basin on the border between the states of Goiás and Mato Grosso in the Brazilian Amazon Basin from September to December 2000.We sampled five nests distributed in four sand beaches at both margins of the Araguaia river (13°21'57.6"S,50°39'05.7"W;13°30'12.4"S,50°44'12.4"W;13°23'32.3"S,50°40'12.8"W;13°25'11.5"S,50°39'55.6"W).The following nest characteristics were taken: egg-chamber depth and width, distance from the water and sediment type (according to bonach et al., 2003, 2006, 2007).Fieldwork was undertaken 7:00 to 9:30 AM or 4:00 to 6:30 PM in order to avoid egg desiccation.We implanted a data-logger Hobo TBI32-20+50 Onset (precision ± 0.1 C) in the middle of the top layer of the egg-chamber of each nest, approximately 25 to 49 cm beneath the surface (according to bonach et al., 2006), at day 1 of the incubation period.The data-loggers were programmed to store temperature every other hour during the whole incubation period.
This interval was considered short enough to prevent undetected peaks in temperature.They were tied by a thin rope to a numbered stick fixed in the ground above the nest.All nests were fenced for protection against predators and checked at day 45 of incubation in order to evaluate the development of embryos.
Besides temperature fluctuation we recorded clutch-size, incubation period and hatching success rate of all nests.Sex ratio of all hatchlings (n=386) was determined for the five clutches.Regression equations were established between the percentage of females per clutch and the average temperature for the three thirds of the incubation period (i.e., days 1 to 20, 21 to 40, 41 to hatching), both treated as continuous variables (Sokal & rholF, 1995).In order to be sexed, hatchlings were sacrificed in sulphuric ether chambers at the age of 30 days after hatching.Gonads were extracted, fixed in formaldehyde 10%, included in paraffin and transversally cut in order to prepare histological slides.Sex identification followed the description of malvaSio et al. (1999) (Figs 1, 2).At the time of the analyses, nonlethal sexing techniques such as hormone level (lance et al., 1992), geometric morphometrics (valenzuela et al., 2004) and laparoscopy (kuchling, 2006) were either considerably less accurate or not efficiently applicable to hatchlings this small (< 70g).

DISCUSSION
Besides daily fluctuations, temperature variation is possibly also related with sand physical characteristics such as grain size distribution and thermal conductivity in P. expansa nests (Souza & vogt, 1994).The optimum thermal range for the incubation of chelonian eggs is between 29ºC and 35ºC (eWert, 1985).In this study, the average temperature of the nests is in this range.However, we found mean, minimum and maximum temperatures below those reported for the species at Araguaia National Park in the southern Araguaia river 33ºC to 35ºC, 25.8ºC and 39.1ºC, respectively (A.Malvasio, pers. observ.).von hildebrand et al. (1988) report an average temperature of 28ºC 15 cm deep for P. expansa nests in Colombia.However, SPotila et al. (1987) did not find significant thermal differences according to different depths in Chelonia mydas (Linnaeus, 1758) nests.SPotila et al. (1987) reported a much smaller range of daily thermal fluctuation (1.5ºC) in C. mydas nests than we found for P. expansa in this study (28.5ºC).This adds some complexity to the study of the role of temperature of incubation on sex determination of freshwater turtles, in such a way that not only mean temperatures should be considered but also extreme temperatures eventually experienced by the embryos and -possibly even more important -the range and frequency of daily thermal variation as stated by georgeS et al., (1994) for Caretta caretta (Linnaeus, 1758).However, the relationship between incubation temperature daily fluctuation and embryos' sex determination in turtles is still unclear (valenzuela, 2001).(Schweigger, 1812) in central Araguaia river, border between the states of Goiás and Mato Grosso in the Brazilian Amazon Basin, from September to December 2000 (n = 5) (standard error of mean < 0.15°C for all average temperatures below).

Tab. I. TSD in Podocnemis expansa
bonach et al.The analysis of variance of the coefficients generated by the Fourier Analysis (bloomField, 1976) can be useful in future studies, helping detect for instance eventual intraspecific differences of the pivotal temperature, as observed by chevalier et al. (1999) for Atlantic and Pacific nesting populations of Dermochelys coriacea (Linnaeus, 1766).However, this approach needs specific experimental design where both frequency and amplitude of temperature variation can be tested as distinct dependent variables.
The hatching success rate found in this study was lower than previous reports for the species in northern Amazon (88.9 %) (zWink & young, 1990) and southern Araguaia river (92.5%) (A.Malvasio, pers. observ.).However, the great variation of the species nesting habitats mentioned above and the small sample size of this study prevent us to take definite conclusions about this.Podocnemis expansa occurs on a broad geographic area with a considerable heterogeneity of nesting habitats in space and time.Therefore, we can expect a great variation of nesting success from year to year on a micro-as well as on a macrogeographic scale (verdade et al., 2002; villela et al., 2008).
Although we had lower temperatures of incubation than those recorded by A. Malvasio (unpubl. data), the average period of incubation in both studies were quite similar (58.2 days in this study and 59 days, A. Malvasio, unpubl. data) and intermediate in relation to others (alho et al., 1979: 48 days; von hildebrand et al., 1988: 67.2
Hatchling turtles do not present an evident sexual dimorphism on their secondary sexual characteristics.Therefore, to sex hatchlings it is necessary to use more invasive techniques such as the determination of the testosterone seric level after FSH injection (valenzuela et al., 1997(valenzuela et al., ), gonads anatomy (malvaSio et al., 1999(valenzuela et al., , 2002)), histological identification of the ovary and testis (danni & alho, 1985;malvaSio et al., 2002) and laparoscopy (vogt, 1994).In this study we used the histological identification of the gonads with more than 99% of secure classification, by far more accurate than the other methods.
TSD has been previously described for P. expansa with males produced at lower and females at higher temperatures (MF pattern) (alho et al., 1985;lance et al., 1992;Souza & vogt, 1994).However, valenzuela (2001) suggests the species can also present a FMF pattern, with females produced at the extreme temperature, what is corroborated by the present study.
The thermosensitive period for sex determination in chelonians is considered to be the second third of the incubation period (bull, 1983).However, in the present study the thermosensitive period apparently occurred after the day 40 of incubation.Considering the average incubation period of 58 days, the thermosensitive period for sex determination surprisingly seemed to occur at the last third.Some intraspecific variation can in theory occur, but its evolutionary implications are still undetermined and should be pursued in future studies.On the other hand, the implications of a late sex determination for ranching management programs are clear: sex ratio of hatchlings will be affected by early egg translocation.This should be a concern for managers, since in some conservation turtle projects, there are eggs transferences.At last, in this study we considered only mean temperatures, whereas extreme temperatures may eventually be biologically more relevant.For this reason, future studies should prioritize the relationship between temperature variation (i.e., range and cycle) and embryos development, survivorship and sex determination.

Fig. 3 .
Fig. 3. Relationship between the temperature of incubation [mean (dots) ± standard deviation (lines)] between day 41 of incubation and hatching and the ratio of females of Podocnemis expansa (Schweigger, 1812) at the central region of the Araguaia river on border between the states of Goiás and Mato Grosso, from September to December 2000.