The reproductive biology of a species constitutes an important component of its overall life history strategy (Gibbons and Greene 1990; Stearns 1992); thus, obtaining information on reproductive ecology is important for turtle management and conservation (Dodd 1997; Passmore and Brooks 1997; Tucker and Moll 1997; Horne et al. 2003). Studies of turtle nesting ecology have revealed important direct and indirect effects on fitness and demography (Weisrock and Janzen 1999; Valenzuela 2001; Spencer and Thompson 2003; Moll and Moll 2004; Janzen et al. 2007; Páez et al. 2009a). The process of selecting a nesting site is important to females in an immediate sense because they are more vulnerable to terrestrial predators at this time (Spencer 2002) and because terrestrial predators also are often the major causes of nest mortality (Yerli et al. 1997; Burke et al. 1998; Escalona and Fa 1998). Nest site selection may also have indirect influences on a female's fitness. First, in many species, incubation temperature determines the sex ratio of the clutch (Valenzuela and Lance 2004). Incubation temperature and humidity also have been shown to influence embryo metabolism, development, and survivorship, as well as incubation period and hatching success, neonate size, residual yolk mass, and hatchling performance attributes such as growth and survivorship (Congdon and Gibbons 1990; Packard et al. 1993; Tucker and Paukstis 2000). For these reasons, many studies have attempted to identify the microhabitat variables that influence thermal and hydric conditions of nests, such as soil composition, soil grain size (Remor de Souza and Vogt 1994; Milton et al. 1997; Ferreira-J. and Castro 2003; Ferreira-J. et al. 2003), nest depth (Valenzuela 2001), and vegetative cover (Vogt and Bull 1984; Janzen and Morjan 2002). The distance of a nest from the shoreline is another important variable in nest site selection that may influence the probability that a nest will be depredated (Drummond 1983; Spencer and Thompson 2003), parasitized (McGowan et al. 2001; Hall and Parmenter 2008) or flooded (Plummer 1976; Escalona and Fa 1998), as well as influence subsequent hatchling survival (Weisrock and Janzen 1999; Kolbe and Janzen 2002).

The Magdalena River turtle (Podocnemis lewyana) is an endangered species endemic to the Magdalena, San Jorge, and Sinú river drainages of northern Colombia (Ernst and Barbour 1989; Páez et al. 2009b). Given the levels of intraspecific variation in reproductive ecology that have been documented in other species of Podocnemis, both among sites (Thorbjarnarson et al. 1993; Fachín-T. and von Mülhen 2003) and within or among nesting seasons (Foote 1978; Gibbons et al. 1982), detailed studies of the nesting ecology of P. lewyana at more than one site and for more than one nesting season are warranted but have yet to be undertaken. For this reason, we conducted a study of the nesting ecology of P. lewyana over 2 years at 2 contrasting sites within the Mompos Depression of the Magdalena River drainage in Colombia. This is the first detailed report on the nesting ecology of this endangered species (Restrepo et al. 2008; Páez et al. 2009b) that provides data necessary to develop management plans in this area.

METHODS

In 2005, we conducted field work on La Pava Island, a small island within the La Rinconada wetland (lat 9°13′50″N, long 74°14′35″W) in the northeastern region of the Mompos Depression of the Magdalena River drainage in northern Colombia (Restrepo et al. 2008). In 2006, we worked at 2 sites in the Chicagua River, which flows north through Isla Margarita, an island formed by the split of the Magdalena River into the Mompos and Loba branches (Fig. 1). The Chicagua River data obtained in 2006 were collected specifically in the Jorobado Channel (lat 9°09′59″N, long 74°40′43″W) and the nearby La Bella Island (lat 9°10′10″N, long 74°70′42″W).

View Figure

• Download as Powerpoint
• Download Figure

In both years, we conducted field work from February to May, the peak months of nesting activity in this region. Study beaches were searched daily in the company of a local inhabitant who located nests visually or with the aid of dogs. We also obtained oviductal eggs from females that had been harvested by local people. These oviductal clutches were incubated in artificial nests constructed high on a beach on La Pava Island to minimize the risk of nest flooding. The artificial nests were constructed to resemble natural nests in terms of their size, shape, and substrate type. Hereafter we refer to these as “oviductal” nests (Table 1).

Table 1 Numbers of different nest types by year and location.

View Fullscreen

In the Jorobado Channel we also located nests by daily searches with local inhabitants and dogs. Nests that were deemed particularly vulnerable due to their proximity to the shoreline were transferred to artificial nest chambers constructed in a beach on La Bella Island to protect them from flooding and the high levels of human and natural nest predation at the Jorobado Channel site. Natural nests also were found on the La Bella Island beach, and rising river levels prompted us to transfer some of them to the artificial nest area located higher up on the same beach. We hereafter refer to these as “transferred” nests (Table 1). In both years, all intact natural, transferred, and oviductal nests were covered with 1 × 1 cm mesh galvanized chicken-wire cylinders measuring 40 cm in diameter and 50 cm in height that supported a 3 × 3 cm plastic mesh to afford some protection from predators and permit capture of emerging hatchlings.

We collected the following data for each nest: oviposition date; distance to the nearest body of water measured perpendicularly to the shoreline; soil type classified according to its texture and humidity (Jaramillo 2002); vegetative cover (grass, shrub, leaf litter, or open) quantified using a 1 m² plot divided into 10 cm² divisions (Escalona and Fa 1998; Restrepo et al. 2006); exposure to sunlight categorized as full shade (nests that received no more than 3 hours of direct sunlight each day), half shade (nests that received up to 6 hours direct sunlight each day), and full sun (nests completely exposed to full solar radiation all day); depth to the first egg; and maximum depth of the nest chamber.

For most nests, we quantified clutch size, measured the length (x) and width (y) of each egg to the nearest 0.1 mm with digital calipers, and weighed egg mass to the nearest 0.1 g with a digital balance. We calculated egg volume using the formula V  =  πxy²/6 (Vanzolini 1977). Eggs were replaced (natural nests) or transferred to artificial nest chambers (transferred nests) by burying them in the same approximate position as they were found, without rotation, to avoid embryo mortality (Limpus et al. 1979; Soini and de Soini 1995a). Eggs were manipulated only during the first 24 hours following oviposition. For a subset of unopened natural nests, we estimated clutch size at the end of the incubation period by summing the number of live hatchlings, dead hatchlings, and eggs found in the nest chamber.

Nests were inspected daily and opened at the first signs of hatching to quantify the number of empty egg shells, unhatched eggs, dead hatchlings, and live hatchlings still present. Nests that failed to show signs of hatching after 75 days of incubation were also opened. Invertebrates encountered in the eggs or nest chambers were collected and preserved in 70% ethanol for later identification. Samples were collected from eggs with fungal infections for microscopic identification.

For each nest, we calculated the following: incubation period (duration) as the interval between the day of oviposition or removal of the eggs from the oviducts of harvested females to the day of first evidence of hatching activity; hatching success as the percentage of the oviposited eggs that hatched; predation rate as the percentage of oviposited eggs to suffer depredation by vertebrates; and infestation rate as the percentage of the oviposited eggs to suffer invertebrate or fungal infestation.

To investigate the effect of incubation temperature on the hatchlings, we monitored nest temperatures in 21 nests (6 natural nests in 2005 and 4 natural and 11 transferred nests in 2006) by burying HOBO data loggers at nest depth, 3 cm to one side of the nest chamber (so as to not disturb the clutch by placing the relatively large data loggers directly into the nest chamber). We measured the following on each hatchling: straight-line carapace length (SCL), carapace width, plastron length, and body mass (BM) using the same calipers and balance as for the egg measurements described earlier. Hatchlings were then transported to the Universidad de Antioquia, where they were sacrificed with an injection of 2% xylocaine in the axial region. One gonad was removed from each hatchling and fixed in 10% buffered formalin for histological preparation and staining with hematoxylin and eosin. Sexual determinations were made based upon the criteria of Danni and Alho (1985) and Estrada and Uribe (2002). Voucher specimens were deposited in the Museo de Herpetolología of the Universidad de Antioquia (MHUA 17239–49; 17264–5; 17289–651; 17653–702; and 17713–27).

Statistical analyses were conducted using the JMP software package (SAS 1995). Mean values are presented as ±1 SD. When data were normally distributed and variances were homogeneous, Student t-tests were used; otherwise, comparisons were made using nonparametric Kruskal-Wallis or Mann-Whitney tests. Frequency data were compared using χ² tests. Results were considered significant at p < 0.05.

RESULTS

During the study we measured 16 P. lewyana adult females and weighed 11, yielding a mean SCL of 40.1 ± 2.2 cm (range 37.4 to 44.3 cm) and a mean BM of 5.6 ± 1.8 kg (range 4.6 to 7.8 kg). At La Pava Island, we located 42 nests, 28 of which had already been depredated, and at the Jorobado Channel we located 39 nests, 16 of which had already been depredated.

We evaluated nest site characteristics for 81 nests in the 2 nesting sites (42 in 2005 and 39 in 2006), although we were unable to quantify the distance to the nearest water body for 1 nest in 2006. Combining data for both years, nest site locations ranged from 0.6 to 39.2 m distance from the nearest water, with a mean of 10.3 ± 8.4 m. When we classified nests into 5 m interval groups, the frequency of nests in each interval was not homogeneous (χ²  =  56.4; df  =  7, p < 0.001; n  =  80), with only 11 nests (13.7%) located more than 15 m from the nearest water body.

Nests were not distributed homogeneously among the 3 available soil types, with no nests located in clay substrates and significantly more nests located in sandy soils (80.2%) than in loam substrates (χ²  =  29.6; df  =  1, p < 0.001). No nests were located under shrubs, but there was a significant difference in the proportion of nests in the remaining 3 categories, with 56.2% of the nests in open areas, 36.2% in grassy areas, and only 7.5% covered by leaf litter (χ²  =  29.85; df  =  2, p < 0.001). Open and grassy areas did not differ in their frequency of use but were used significantly more than areas with leaf litter (χ²_grass  =  15.11; χ²_open  =  30.77; df  =  1; p < 0.001). A significantly greater proportion of the nests were constructed in areas that received direct sunlight (87.6%) than in areas of half-shade (6.1%) or full shade (6.1%; χ²  =  107.55; df  =  3, p < 0.001).

Nest chambers were asymmetrically bottle-shaped, with oval entrances leading to chambers that sloped to one side. The mean depth to first egg was 9.4 ± 3.9 cm (n  =  35; range 3 to 18 cm), while mean depth of the nest chamber was 19.4 ± 4.6 cm (n  =  37; range 12 to 35 cm).

Mean clutch size was 20.5 ± 4.6 eggs (n  =  64 clutches, range 10 to 31 eggs). We measured 596 eggs and weighed 274 (Table 2). Eggs varied from roundish to ellipsoid with a slightly flexible parchment texture and pallid rosy gray color that turned to chalk white a few days after oviposition. Hatching success rates were quantified for 41 nests over the 2 years (Table 3). The 3 nest types (natural, transferred, and oviductal) did not differ in hatching success rates (Kruskal-Wallis test, H  =  5.48; p  =  0.065). There were no differences among years/sites in hatching success rates (Mann-Whitney test, U  =  1.13; p > 0.10), with a mean hatching success rate of 36% at La Pava Island in 2005 (n  =  21, range 0% to 100%) and 50.5% at the Chicagua River in 2006 (n  =  20, range 0% to 90.91%). Of the 44 nests lost to predators during the 2 seasons (28 in 2005 and 16 in 2006), we observed or inferred (based on tracks) that 72.7% were consumed by the teiid lizard, Tupinambis teguixin (Teiidae), 25% were lost to humans, and one nest (2.2%) apparently was trampled by cattle.

Table 2 Descriptive statistics for Podocnemis lewyana eggs from the Mompos Depression.

View Fullscreen

Table 3 Hatching success rates (%) in natural, transferred, and oviductal Podocnemis lewyana nests in the Mompos Depression.

View Fullscreen

In 2005, we documented that P. lewyana eggs were infested by invertebrates, particularly ants (Hymenoptera). The most common genera found in nests were Solenopsis and Crematogaster, although we also documented infestations by species of Dorymyrmex, Camponotus, and Paratrechina. Nests also were infested by hemipterans of the family Cydnidae, and larval and adult coleopterans of the families Tenebrionidae and Carabidae, as well as larvae of the dipteran family Sarcophagidae. Many nests were infested with fungi, although we do not know whether the fungi were responsible for the eggs dying or whether they opportunistically infested eggs that were unviable for other reasons. We were able to identify 5 fungi to genus or species: Trichocladium sp., Scedosporium apiospermum, Fusarium sp., Penicilium sp., and Chrysonilia sitophila. Mean nest infestation rate (invertebrates and fungi pooled) in natural nests was 35.3%, with a comparable rate for artificial nests (34.2%; Mann-Whitney test, U  =  0.05; p > 0.10; n  =  21). Logistic regression showed that the probability of nest infestation was not related to the distance of the nest from the nearest water body (χ²  =  0.08; p > 0.10; n  =  19). Nest infestation incidence was not related to the type of vegetative cover (Kruskal-Wallis test, H  =  2.06; p  =  0.356; n  =  21), substrate type (Kruskal-Wallis test, H  =  6.21; p > 0.10; n  =  21), or exposure to sunlight (Mann-Whitney test, U  =  0.39; p > 0.10; n  =  21).

Mean incubation period for all nests that hatched was 59.1 ± 8.4 days (range 50 to 89 days; n  =  32). In 2005, mean incubation period was significantly longer than in 2006 (Table 4; Mann-Whitney test, U  =  9.70; p  =  0.002; n  =  32). In 2005, mean incubation period of the natural nests was not different from that of the oviductal nests (Table 4; Mann-Whitney test; U  =  1.05; p > 0.10; n  =  12), but in 2006, mean incubation period for the natural nests was significantly longer than for the transferred nests (Table 4; Mann-Whitney test; U  =  4.5; p < 0.001; n  =  20).

Table 4 Mean incubation period (days) for Podocnemis lewyana nests in the Mompos Depression.

View Fullscreen

We measured 350 neonates, finding all 4 of the measured variables positively correlated (Spearman correlations, p < 0.001), so we here only report results in terms of SCL and BM. The neonates in 2005 were significantly larger and heavier (mean SCL: 43.2 ± 2.9 mm; mean BM: 17.4 ± 3.3 g; n  =  110) than those obtained in 2006 (mean SCL: 42.2 ± 2.9 mm; mean BM: 15.7 ± 2.4 g; n  =  240; Mann-Whitney test, SCL: U  =  2.60, p  =  0.009; BM: U  =  3.71, p < 0.001). However, in neither year did neonates from the different nest types differ in BM (2005, natural vs. oviductal, n  =  110, Mann-Whitney test, U  =  0.745, p > 0.10; 2006, natural vs. transferred nests, n  =  240, U  =  −1.26, p > 0.10), but they differed in size in 2006 (2005, natural vs. oviductal, n  =  110, SCL: Student t-test, t  =  1.045, p > 0.10; 2006, natural vs. transferred, Mann-Whitney test, SCL: U  =  7.01, p < 0.001).

Mean incubation temperature for all monitored nests was 32.8°C ± 1.7°C (range 28.7°C to 34.6°C; n  =  21). Mean incubation temperature for the 6 nests monitored in La Pava Island was 30.8°C, while in La Bella Island at the Chicagua River site, the mean incubation temperature recorded in the 15 nests was 33.8°C (Table 5). Considering only natural nests, mean incubation temperatures differed significantly among years/sites (Table 5; Student t-test, t  =  11.052; p  =  0.006). In 2006, the incubation temperatures of natural nests were significantly lower than temperatures of transferred nests (Table 5; Mann-Whitney test, U  =  −2.663; p  =  0.008).

Table 5 Descriptive statistics for incubation temperatures measured in Podocnemis lewyana nests in the Mompos Depression.

View Fullscreen

We calculated mean incubation temperatures during the thermosensitive portion of the incubation period (second third of the incubation period; Valenzuela et al. 2003), obtaining a mean for all monitored nests of 33.6°C ± 1.3°C (n  =  21). Based upon inspection of only data from the second third of incubation, mean incubation temperatures of natural nests were significantly lower in 2005 than in 2006 (Table 5; Student t-test, t  =  23.44; p < 0.001; n  =  21). In 2006, mean incubation temperatures during the second third did not differ significantly among natural nests and transferred nests (Table 5; Mann-Whitney test, U  =  −0.196; p > 0.10; n  =  15). Logistic regression showed that the probability of a nest hatching successfully was related to its incubation temperature, with lower hatching rates in cooler nests (χ²  =  4.67; p  =  0.03; n  =  21), although incubation temperature did not influence hatchling size (SCL: R²  =  0.023, p > 0.10; BM: R²  =  0.071, p > 0.10, n  =  21). Incubation period (duration) was positively correlated with mean incubation temperature (R²  =  0.63; p < 0.001; n  =  21).

Sex was successfully determined for 244 neonates both by examination of the external morphology of the gonads and with histology. External determination of sex was corroborated as correct in 94% of these cases, so we assigned sex to the remaining 38 neonates for which adequate histological sections were not obtain based only on their gonadal morphology. In 2005, the sex ratio documented was 1♀ : 1♂, while in 2006 the sex ratio was significantly skewed in favor of females (8.4♀ : 1♂; χ²  =  51.77; df  =  1; p < 0.001). In 2006, sex ratios also differed among natural nests (2.6♀ : 1♂) and transferred nests (6♀ : 1♂; χ²  =  30.14; df  =  1; p < 0.001).

DISCUSSION

The body sizes recorded for reproductive females of P. lewyana at La Pava Island were within the range reported for the species elsewhere (Dahl and Medem 1964; Rueda-A. et al. 2007; Gallego-G. and Castaño-M. 2008; Páez et al. 2009a, 2009b). The range in clutch size that we documented was also similar to previous reports for females from the Sinú River (Dahl and Medem 1964; Gallego-G. and Castaño-M. 2008) and the Ayapel wetland (Castaño-M. 1986).

The fact that most nests in the Mompos Depression were located within 15 m of the shoreline suggests that females reduce their exposure to terrestrial predators (including humans) by not venturing far inland to nest. In other species of turtles an inverse relationship has been documented between the distance of nests from the shoreline and local human population density; the latter being an indirect measure of predation pressures (Moll and Legler 1971; Temple 1987; Spencer 2002; Spencer and Thompson 2003).

Both our study and that of Gallego-G. and Castaño-M. (2008) found that a majority of the nests were constructed in sand substrates, similar to reports for other species of Podocnemis (Vanzolini 1977; Thorbjarnarson et al. 1993; Soini and de Soini 1995b). Sand substrates warm faster (decreasing the incubation period), are more porous (offering better drainage), and do not harden upon drying (facilitating neonate emergence; Moll and Moll 2004). In contrast, excessively humid soils could lead to fungal or bacterial infestations (Garduño-A. and Cervantes-H. 1996) and restrict gas exchange between the soil and nest chamber (affecting embryo development; Congdon and Gibbons 1990; Etchberger et al. 1991).

Our data showed P. lewyana to nest principally in areas without vegetative cover, although females occasionally used soils covered with leaf litter or low grass. The vast majority of nests were exposed to direct sunlight (87.6%), consistent with other reports of freshwater turtles that seem to prefer to place nests in open areas without vegetative cover (Moll and Legler 1971; Congdon and Gibbons 1990; Thorbjarnarson et al. 1993; Wilson 1998). These conditions presumably are optimal, as nests incubating at high temperatures have reduced incubation periods, minimizing their risks of being depredated or flooded by unpredictable, precipitous rises in water levels.

Our descriptions of the form, color, and texture of the egg shells, as well as our data on the length, width, and weight of the eggs of P. lewyana, agree with previous reports (Ramo 1982; Castaño-M. 1997; von Hildebrand et al. 1997; Pezzuti and Vogt 1999; Fachín-T. and von Mühlen 2003). Upon comparing our results to data published on other Podocnemis turtles, egg length is comparable in all species, while egg width and weight are more variable, with larger species having wider and heavier eggs (Ramo 1982; Castaño-M. 1997; von Hildebrand et al. 1997; Pezzuti and Vogt 1999; Fachín-T. and von Mühlen 2003).

The positive relationship we documented between incubation temperature and hatching success rate may have been an artifact, because nests that failed to develop for whatever reason would subsequently exhibit lower incubation temperatures than viable nests that were producing metabolic heat. Metabolic heat production has been shown to be important in other species that, like P. lewyana, lay relatively large clutches in deep nests (Remor de Souza and Vogt 1994; Páez and Bock 1998; Godley et al. 2001; Broderick et al. 2001; Restrepo et al. 2005). However, P. lewyana eggs incubated under constant but low temperatures (28°C) in the laboratory experienced higher mortality, even early in the incubation period, compared to eggs incubated at higher temperatures (Páez et al. 2009a), implying that the cool nests in the field we studied actually suffered higher mortality rates.

We found no differences in hatching success rates among natural, transferred, and oviductal nests, despite the fact that in other species like P. unifilis and Chelonoidis carbonaria, there is evidence that manipulating clutches affects hatching rates (Marcellini and Davis 1982; Soini and Cóppula 1995; Fachín-T. and von Mülhen 2003; but see Ortega et al. 2000). Our observations indicate that the teiid lizard is the principal predator of P. lewyana nests in the Mompos Depression, as has been shown elsewhere for other species of Podocnemis (Soini and Cóppula 1995; Soini and de Soini 1995b). We also detected nest infestation by a number of species of ants that voraciously attacked both eggs and recently emerged hatchlings, similar to reports in other turtles (Fowler 1979; Ramo 1982; Fachín-T. 1994; Moulis 1997; Buhlmann and Coffman 2001; Parris et al. 2002). We do not know whether the other invertebrates we found infesting nests (cydnid hemipterans, sarcophagid and phorid flies, and coleopterans) actually killed the embryos or were merely opportunistic scavengers (Fowler 1979; Iverson and Perry 1994; Broderick and Hancock 1997; McGowan et al. 2001; Rosano-H. and Deloya 2002). More likely to cause embryo mortality were the fungi we found with hyphae growing on the external and internal surfaces of intact egg shells, which presumably affect embryo development by impeding gas exchange and calcium uptake.

We found comparable fungal infection rates in natural and oviductal nests, reaffirming that interring eggs obtained from the oviducts of sacrificed females is a viable management method to obtain hatchlings. We found no relationship between the incidence of nest infestation and environmental variables, despite our a priori expectation that nests located under vegetative cover, in the shade, and/or in moister soil substrates would be more vulnerable to ants and fungal attack than nests located in dry, sandy substrates (Fachín-T. 1994; Soini and de Soini 1995b; Garduño-A. and Cervantes-H. 1996).

Mean nest incubation temperatures documented for P. lewyana in the Mompos Depression were comparable to those reported for 2 transferred nests in the Sinú drainage (Gallego-G. and Castaño-M. 2008). Incubation temperatures are influenced by the interaction of climatic conditions and characteristics of the nest micro-environment (Vogt and Bull 1982; Schwarzkopf and Brooks 1987). For example, in our study nest temperatures at La Pava Island in 2005 were lower than those in the Chicagua River site in 2006, presumably because at the latter site, all nests were located in sand substrates with no vegetative cover, and most were in the full sunlight.

These differences in incubation temperature among the sites were also reflected in differences in mean incubation periods (durations), with La Pava Island nests incubating significantly longer. At La Pava Island in 2005, eggs obtained from recently sacrificed females and buried in artificial nest chambers had incubation periods comparable to those of natural nests. In contrast, at the Río Chicagua site, natural and transferred nests differed in incubation periods in 2006, presumably due to differences in environmental conditions at the site where the artificial nests were constructed.

The sex ratio differences we found between the 2 sites and years are consistent with differences in mean incubation temperatures, with more females being produced at higher incubation temperatures. While mean incubation temperature or duration of the incubation period have been shown to not be the best predictors of nest sex ratios (Georges 1989; Georges et al. 1994), this result is still consistent with laboratory studies that documented temperature-dependent sex determination in this species (Páez et al. 2009a).

The mean SCL we documented in hatchling P. lewyana from the Mompos Depression was lower than previously reported for the species (Dahl and Medem 1964), and hatchlings from the Río Chicagua site were smaller than those from La Pava Island. Given that we did not monitor water potentials or humidities of the nesting substrates at either site, it is impossible to make inferences concerning differences in body size since these variables have been shown to influence hatchling size in other species (Valenzuela and Lance 2004). In 2006 the natural nests had both longer incubation periods and produced larger hatchlings, suggesting temperature also may affect hatchling size in this species (Tucker and Paukstis 2000; Páez et al. 2009a).

Studies with other freshwater turtle species (Trachemys scripta and T. callirostris) have shown that adult female size, and hence egg size and hatchling size, are influenced by the level of human exploitation a population experiences (Close and Seigel 1997; Bernal et al. 2004; Daza and Páez 2007). It is possible that sustained exploitation of P. lewyana has led to comparable changes in mean female, egg, and hatchling size in the Mompos Depression over historical times (Dahl and Medem 1964; this study) or currently is maintaining differences among more heavily (Río Chicagua) vs. less intensely (La Pava Island) hunted sites.

Our results indicate that nest transfer or careful interment of eggs obtained from sacrificed female P. lewyana did not negatively affect embryo viability, suggesting that these represent viable management methods for this threatened species (see Tucker 1995). However, such management efforts should be careful to insure that nests do not incubate at cool temperatures (less than 29°C) because such nests may show reduced hatching success rates (Páez et al. 2009a).