Turtles require specific habitat characteristics to achieve their basic needs for reproduction and survival, both in the wild and in captivity. If these requirements are not available, these reptiles generally do not do well in captivity (McKeown 1996). Captive management demands knowledge of several aspects of the biology of species to ensure that artificial conditions reach their biological requirements. Some basic aspects, such as housing, water quality, temperature, sunning area, diet, and care of hatchlings, should be considered for captive maintenance of aquatic turtles (Boyer and Boyer 2006). Furthermore, artificial conditions and captive husbandry may severely affect physiological processes (St. Aubin et al. 2001).

A health evaluation includes anamnesis, or clinical history; a physical examination; hematology; biochemistry assessments; and other clinical pathology investigations that may include fecal examination, urinalysis, cytology, histology, serology, microbiology, virus isolation, radiography, ultrasonography, endoscopy, and exploratory surgery (Barrows et al. 2004). Health assessment protocols in turtles have been proposed by Berry and Christopher (2001), Barrows et al. (2004), and Hernandez-Divers (2006).

Studies have been conducted to assess health in various chelonian species, such as the diamondback terrapin, Malaclemys terrapin (Werner et al. 2002); bog turtle, Clemmys muhlenbergii (Brenner et al. 2002); desert tortoise, Gopherus agassizii (Christopher et al. 2003); gopher tortoise, Gopherus polyphemus (Diaz-Figueroa 2005); alligator snapping turtle, Macroclemys temminkii (Chaffin et al. 2008); green turtle, Chelonia mydas (Hamann et al. 2006; Flint et al. 2010); and Central American river turtle, Dermatemys mawii (Rangel-Mendoza et al. 2009). In most of these investigations, signs of illness in evaluated individuals were not found, although some differences in health status were associated with reproductive status or seasonal effects. However, in the hematological evaluation of D. mawii (Rangel-Mendoza et al. 2009), manifestations of illness related to improper husbandry protocols were detected in captive populations.

The Central American river turtle, Dermatemys mawii Gray 1847, is a freshwater species naturally distributed in southeastern Mexico, Guatemala, and Belize (Vogt et al. 2011). Wild populations have been reduced considerably due to illegal hunting for consumption and habitat modification (Polisar and Horwich 1994; Convention on International Trade in Endangered Species of Wild Fauna and Flora [CITES] 2005). Dermatemys mawii is considered endangered under Mexican environmental laws (Secretaría del Medio Ambiente y Recursos Naturales [SEMARNAT] 2010); therefore, there is an official ban on its capture and collection from the natural environment. This turtle is included in the world's top 25 most endangered turtles according to the Turtle Conservation Fund (Turtle Conservation Coalition 2011). Since 1982, the species has been listed in Appendix II of CITES (2005) and as critically endangered on the Red List of the International Union for Conservation of Nature (IUCN 2013).

Reduction of natural populations of D. mawii, its economic importance for rural communities, and the need to generate conservation alternatives have prompted initiatives for its captive breeding. As of 2009, there were 14 D. mawii captive breeding centers officially recognized in Mexico (Comisión Nacional para el Conocimiento y Uso de la Biodiversidad/Dirección General de Vida Silvestre/Comisión Nacional de Áreas Naturales Protegidas [CONABIO-DGVS-CONANP] 2009). The husbandry techniques used to breed this turtle in captivity come from the traditional knowledge of rural consumers and the experience of local producers but none from formal scientific research. There is some evidence that current captive husbandry protocols may negatively affect the health of this species (Rangel-Mendoza et al. 2009).

The present study was designed to investigate these problems by determining the critical factors of management and husbandry that negatively affect the health of captive D. mawii. Health status was evaluated for 2 of the largest captive colonies of the species, including an anamnesis with managers and caretakers, a comprehensive physical examination, serum biochemistry analysis, microbiological analysis, and water quality monitoring. The results of these evaluations were then used to modify husbandry methods for captive breeding of this endangered species.

METHODS

RESULTS

Turtle colonies were managed in single rustic ponds at each site (outdoor soil excavation). Animals of different sizes were kept together (juveniles, nonreproductive adults, and reproductive adults) and coexisted with other wild species that inhabit farms (toads, frogs, iguanas, and snakes, among others) as well as fish (catfish or tilapia) that are bred temporarily in the ponds. The GOV pond was 40 m long, 20 m wide, and 3.5 m deep and held approximately 450 animals (density 0.56 individuals/m²), while the TAB pond was 25 m long, 18 m wide, and 3 m deep and held 41 animals (density 0.09 individuals/m²).

General husbandry conditions of turtles differed between the study sites. Food availability was irregular at the farms and strongly dependent on government resources (GOV) as well as on donations or subventions (TAB), which, depending on stock, may be offered daily. Extruded floating food for tilapia fish were offered ad libitum to turtles, containing different levels of protein that varied from 25% up to 32% at GOV and was 30% at TAB. Food offered at TAB was from the Nutripec line for Tilapia by Purina®, while at GOV, animals were also fed with artificial food for tilapia by Silver Cup® and Api-tilapia and Api-bagre (catfish) lines by MaltaCleyton®. Turtle diets may be occasionally supplemented with vegetables (lettuce, spinach, and tomatoes), fruit (plantain and mangoes), or native vegetation, such as the “mazote” weed (butter daisy; Melampodium divaricatum), and introduced vegetation, such as water hyacinth (Eichhornia crassipes). These supplements were more frequent at GOV (once or twice weekly) and much more sporadic at TAB (once per month or less).

Climate conditions changed considerably between sample periods, according to the rainfall patterns of the areas, which dramatically affected water availability in the turtle ponds. At both sites, water levels decreased in May, which corresponds to the dry season in the region. Ponds were supplied with underground water (TAB) and/or water from the local municipal water supply (GOV) whenever determined necessary. Despite these actions, during the months of least rainfall (April and May), the ponds reached depths of < 0.5 m at GOV and < 1.0 m at TAB. During the rainy season (September–October), the ponds can reach depths up to 3.5 m at both farms and may even overflow. During the present study, there were no measures taken to control water quality (water change, filtration, aeration, recirculation, and so on) Ponds had little or no tree vegetation around their banks and were therefore exposed to direct sunlight.

Physical Examination

A total of 176 turtles were examined, including 24 males and 152 females. Individual size did not vary significantly among samples or between farms; animals from GOV showed BM and SCL of 5.2 ± 2.0 kg and 33.4 ± 4.5 cm, while at TAB, such measurements were 5.6 ± 1.6 kg and 34.4 ± 4.1 cm, respectively (F_1,175  =  1.67, p  =  0.20 for BM and F_1,175  =  0.61, p  =  0.44 for SCL).

The relationship of mass and size of the turtles varied among study sites (Fig. 1). Median and range of BM and SCL, the equations of regression lines of BM–SCL and their correlation coefficient, and median and range of BCI for each study site at each sampling period are shown in Table 1. Regression lines varied between farms for May sampling (F_1,1  =  7.67, p < 0.01) and August sampling (F_1,1  =  9.21, p < 0.01). For comparison of intercepts of regression lines between BM and SCL, there were differences between the slopes only in May (F_1,1  =  8.83, p < 0.01). At GOV, a difference was found in the body condition of animals among sampling periods (F_1,1  =  5.44, p < 0.01), with body condition being higher in February than in May or August. At TAB, the BM–SCL relationship was similar among samplings.

View Figure

• Download as Powerpoint
• Download Figure

Table 1. Relationship between body mass (BM, in g) and carapace length at the midline (SCL, in cm) of captive Dermatemys mawii from 2 study sites and sampling periods, including body condition index (BCI). GOV: Government of the State of Tabasco's turtle farm; TAB: Arroyo Tabasquillo Turtle farm.

View Fullscreen

Body condition index varied between study sites (F_1,168  =  322.28, p < 0.01) with GOV exhibiting lower values (Table 1). Sampling period also affected BCI (F_2,168  =  977.75, p < 0.01). Indices from all 3 assessments were significantly different; animals showed the highest BCI values in February and the lowest in August. However, BCI values were contrasting among sites in May: from all calculated indexes in the entire study, the lowest BCI values were registered in GOV and the highest ones in TAB; these marked extreme values are related to the significant difference among the slopes of the BM–SCL regression lines from each farm in May (Table 1). Also, BCI was affected by the interaction of site and month (F_2,168  =  815.43, p < 0.01).

Superficial shell lesions were the most frequent clinical finding, mainly on the plastron and more so during February than other sampling periods (Table 2). This type of lesion varied in extent, but most injuries covered less than 50% of the area of the shell. Lesions often appeared to be caused by trauma due to conspecific interactions and ranged from scratches (minor recent wounds) to severe bites. In some cases, trauma resulted in limb or facial mutilation (e.g., severe ocular or nasal injuries). Few cases of ulcers and chronic open wounds were also found in the skin of the neck, which usually showed caseous secretions or loss of adjacent tissue.

Table 2. Prevalence of main physical findings of captive Dermatemys mawii from 2 study sites (GOV and TAB) and 3 sample periods (February, May, and August) at Tabasco, Mexico (see Table 1 for definition of abbreviations).

View Fullscreen

Physical findings of turtles varied over time (Table 2). In general, integumentary lesions were most common in February and least common in August. During February, cutaneous lesions of the shell, pale oral mucosae, and sunken and swollen eyes were more frequently observed at both farms. Turtles exhibited signs of dehydration (decreased skin elasticity, partial opacity of the cornea, and sunken eyes) mainly during February and May. Ocular discharges, consisting of tear-like, odorless, and colorless secretions, were found only at TAB during February. At GOV, turtles with pale oral mucosae as well as a high prevalence of carapacial algae were most common in August.

Serum Biochemistry

Ninety samples of serum were collected, and 89 were analyzed (1 sample was considered unusable). Biochemical results are shown in Table 3.

Table 3. Median and range of serum biochemistry values in captive Dermatemys mawii from 2 study sites (GOV and TAB) and 3 sample periods (February, May, and August) at Tabasco, Mexico (see Table 1 for definition of abbreviations).^a

View Fullscreen

Blood parameters varied between study sites. Turtles at GOV had significantly lower calcium (F_1,83  =  21.26, p < 0.01), cholesterol (F_1,83  =  78.22, p < 0.01), creatinine (F_1,83  =  8.21, p < 0.01), glucose (F_1,83  =  7.25, p  =  0.013), phosphorus (F_1,83  =  74.43, p < 0.01), total protein (F_1,83  =  228.97, p < 0.01), triglycerides (F_1,83  =  24.34, p < 0.01), urea (F_1,83  =  156.14, p < 0.01), uric acid (F_1,83  =  70.28, p < 0.01), AP (F_1,83  =  74.20, p < 0.01), AST (F_1,83  =  7.01, p < 0.01), and LDH (F_1,83  =  39. 93, p < 0.01) than animals of TAB. Only CK did not exhibit differences between farms (F_1,83  =  2. 74, p  =  0.10).

Blood biochemistry varied in relation to the time of year when the evaluation was conducted (Table 3). In February, animals showed lower levels of cholesterol (F_2,83  =  14.34, p < 0.01) and glucose (F_2,83  =  6.45, p < 0.01) as well as higher values of triglycerides (F_2,83  =  8.20, p < 0.01), urea (F_2,83  =  111.32, p < 0.01), and AST (F_2,83  =  7.40, p < 0.01) compared with other sampling periods. Total protein reached higher levels in May (F_2,83  =  6.21, p < 0.01), whereas calcium dropped to its lowest value in August (F_2,83  =  14.13, p < 0.01) in relation to other assessment periods. Determinations of creatinine (F_2,83  =  74.19, p < 0.01), phosphorus (F_2,83  =  18.50, p < 0.01), and CK (F_2,83  =  57.92, p < 0.01) were different between all pairs of sampling periods; creatinine and phosphorus were lower in May and higher in August, whereas CK was lower in August and highest in February. There were no differences among blood assessments for AP (F_2,83  =  1.76, p  =  0.18) or LDH (F_2,83  =  7.00, p  =  0.56).

The interaction between study site and sampling period showed significant effects on creatinine (F_2,168  =  8.08, p < 0.01), phosphorus (F_2,168  =  6.65, p < 0.01), urea (F_2,168  =  12.65, p < 0.01), uric acid (F_2,168  =  12.63, p < 0.01), and LDH (F_2,168  =  7.00, p < 0.01). Those combined factors did not have a significant effect on calcium (F_2,168  =  0.14, p  =  0.87), cholesterol (F_2,168  =  2.79, p  =  0.07), glucose (F_2,168  =  1.65, p  =  0.20), total protein (F_2,168  =  2.13, p  =  0.13), triglycerides (F_2,168  =  2.87, p  =  0.06), AP (F_2,168  =  2.17, p  =  0.12), AST (F_2,168  =  2.97, p  =  0.06), or CK (F_2,168  =  0.71, p  =  0.50).

Bacteriology

One hundred bacteriological samples were obtained and analyzed from 35 turtles. Seven different microorganisms were isolated: Staphylococcus aureus, Escherichia coli, Proteus vulgaris, Serratia spp., Klebsiella spp., Candida spp., and 1 undetermined fungus species (Table 4). At GOV, nonpathogenic microorganisms were found during the May sampling; only saprophytic bacteria grew in the diverse culture media for that sampling period, and they were not considered in the results. The sample collection site with the greatest diversity of microorganisms detected was the cloaca, from which 6 of the 7 microorganisms were isolated (Table 5). At TAB, only Klebsiella spp. were found in turtles' shell lesions.

Table 4. Bacterial isolates of captive Dermatemys mawii from 2 study sites and 3 sample periods at Tabasco, Mexico (see Table 1 for definition of abbreviations).

View Fullscreen

Table 5. Bacterial isolates of captive Dermatemys mawii from 2 sites at Tabasco, Mexico, differentiated by body collection site (see Table 1 for definition of abbreviations).

View Fullscreen

Water Quality Assessment

Some values of water quality were significantly different between study sites (Table 6). TAB presented lower values of water transparency and DO than GOV (p < 0.01), whereas TAB presented higher values of TAN, nitrates, and fecal coliforms than GOV (p < 0.05). All other values of water quality were not statistically different between farms.

Table 6. Physicochemical and biological water parameters in the ponds of captive Dermatemys mawii from 2 sites (GOV and TAB), registered monthly from February to August. Includes Kruskal-Wallis statistic for differences related to study site (see Table 1 for definition of abbreviations).

View Fullscreen

Water quality differed over time at each study site (Fig. 2). Maximum TAN levels were registered in May at GOV (1.638 mg/l) but in March at TAB (2.239 mg/l); the lowest TAN level was found in March at GOV (0.033 mg/l) but in August at TAB (0.374 mg/l; Fig. 2a). Lowest DO levels were registered in April at GOV (2.85 mg/l) but in August at TAB (0.4 mg/l; Fig. 2b). Minimum values of BOD₅ were found in February at both farms (3.5 mg/l at GOV and 13.6 mg/l at TAB), but the highest BOD₅ was registered in July at GOV (35.1 mg/l) and in August at TAB (66.9 mg/l; Fig. 2c). Values of pH were highest in May and June at GOV (8.6) but in August at TAB (9.5), while the lowest pH values were registered in February at GOV (7.7) and in May at TAB (7.4; Fig. 2d).

View Figure

• Download as Powerpoint
• Download Figure

DISCUSSION

Differences in the physical condition of turtles between study sites, as inferred from the comparison of mass and length relationships, are probably related to the husbandry conditions of the turtles in each farm. However, the mass/length relationship may not be a reliable indicator of the health condition of the turtles because mass may vary in relation to season, diet, exercise, care methods, and health condition (Barrows et al. 2004). Nevertheless, the present analysis was used to compare the physical condition among the evaluated groups, not to find differences between healthy and nonhealthy animals. The BCI analysis clearly shows that differences in the body condition of turtles existed between farms.

At both study sites, the most frequent abnormal physical findings were superficial carapacial and plastral lesions, which may be caused by bacterial, viral, or fungal infections, chemical irritation, nutritional diseases, and trauma (McArthur 2004). Erosion of the keratinized scutes of the carapace is not rare in aquatic turtles; superficial lesions of the shell are commonly related to poor water quality (lack of filtration, irregular water changes, or lack of water conditioners), abrasive substrates, inadequate temperature, malnutrition, stress related to overcrowding, and inappropriate husbandry, among other factors (Barten 1996). In our study sites, poor water quality and high population density were probably the factors responsible for scute erosion.

In a previous study of D. mawii, wild turtles showed no shell lesions that might compromise their health, while all evaluated captive turtles showed diverse types of trauma on their bodies that were thought to be caused by poor husbandry protocols (Rangel-Mendoza et al. 2009). The high frequencies of shell lesions at GOV and TAB are clear evidence of the negative effects of the poor husbandry conditions in captivity on the health of the turtles. Lesions might be indicative of aggressive behavior (biting and scratching) as a result of overcrowding in the current captive conditions.

Aquatic turtles are very susceptible to bacterial infections when malnourished or when living with inefficient hygiene systems. In addition, common lesions in turtles tend to be easily infected due to high quantities of bacteria found in their aquatic environment (Rosskopf and Shindo 2003). Besides, bacteria are rarely primary pathogens in reptile disease, although they are present secondarily as a result of compromised immune function associated with deficient care and husbandry in captivity (Chinnadurai and DeVoe 2009).

Ulcerative lesions of the carapace could appear simultaneously with anorexia, lethargy, and septicemia and can be a cause of death. These conditions have been described as part of the syndrome in aquatic turtles called septicemic cutaneous ulcerative disease (SCUD; Barten 1996). Bacteria such as Citrobacter, Serratia, Proteus, and Pseudomonas, as well as other Gram-negative bacteria, have been associated with this condition (Boyer 1996; Wilkinson 2004b). Two of the bacterial genera considered to be related to SCUD were found in the present study (Serratia spp. and Proteus vulgaris).

The bacteria found in the present study have frequently been isolated in sick reptiles (Jacobson 2007). Nevertheless, these microorganisms have also been reported in the normal intestinal flora in clinically healthy turtles and other reptiles (McArthur et al. 2004; Paré et al. 2006); thus, their presence by itself cannot be considered as a disease indicator. Serratia spp. have been reported to be responsible for SCUD pathogenesis in freshwater turtles and may be able to introduce itself subcutaneously through integumentary trauma (Jacobson 2007). Nevertheless, in the present study, Serratia spp. were present only on oral mucosa but not in any carapacial lesions. Klebsiella spp. were isolated from carapacial lesions and have been associated with respiratory infections in turtles (McArthur 2004; Jacobson 2007; Chinnadurai and DeVoe 2009). Staphylococcus spp., E. coli, and Klebsiella spp. have been found in turtles in cases of stomatitis, gingivitis, or pharyngitis (Jacobson 2007). Proteus vulgaris and E. coli have been reported for subcutaneous masses and abscesses (Jacobson 2007). Candida spp. are opportunistic fungi and may be present in intestinal, integumentary, and respiratory infections (Jacobson 2007). Escherichia coli is a normal component of the bacterial flora in reptile intestines (Paré et al. 2006); its isolation in the present study cannot be interpreted as pathogenic, and therefore additional studies would be required to determinate pathological serotypes in the E. coli population present in the captive environment of D. mawii.

Surprisingly few bacterial species were found in the present study. The microbiological methods used may have limitations, such as the transportation of samples in tubes without culture media (such as Stuart media) and incubation at mammalian body temperature (37°C) that could reduce the viability of the bacteria present in the samples. However, bacterial findings from the present study are a preliminary first report for D. mawii, and more research is needed, using more advanced methodologies.

Some physical findings may have been related to dehydration (such as sunken opaque eyes and low skin elasticity; McArthur 2004), which is common in sick reptiles and may be associated with disease (Donoghue 2006). Poor water quality and inappropriate diet may also lead to dehydration of aquatic reptiles (Mitchell 2006). Dehydration may result in clinical signs of anorexia, inactivity or lethargy, generalized weakness, weight loss, metabolic disease, failure to urinate, and gout (McArthur 2004). Other physical findings were related to irritation of the eyes, such as ocular discharge, swollen eyelids, and partial opacity of the corneas, situations that might be determined by a poor aquatic environment, as discussed below.

Serum biochemical values in the groups of turtles studied here are, in some cases, far from reference values for healthy individuals of this species. Uric acid levels at TAB, during the last assessment, moderately exceeded normal values (0.1–3.2 mg/dl; Rangel-Mendoza et al. 2009). To maintain water balance, reptiles conserve water by eliminating nitrogenous wastes as uric acid and urate salts (Campbell 2006). Serum uric acid is recommended as a good indicator for renal compromise in reptiles (Campbell 2006). Hyperuricemia may be associated with dehydration, severe renal disease, gout, severe bacteremia and septicemia, diets high in protein, and urea (Campbell 1996; Klingerberg 1996).

Another nitrogen compound analyzed in the present study, urea, was found reaching levels 10 times higher at TAB than at GOV in May. A normal urea level for this species varies from 1 to 5 mg/dl (Rangel-Mendoza et al. 2009). At TAB, turtles show remarkably higher urea values than the reference values at all sampling times, whereas at GOV, this was true only during the first assessment. High levels of urea in turtles may indicate kidney disease (Campbell 1996), dehydration, or an excessively high protein diet (Brenner et al. 2002). Urea is highly soluble in water and may easily cross biological membranes (McArthur et al. 2004). High serum levels of urea could be related to high concentrations of nitrogen waste products in the aquatic environment, although this remains speculative at present. It is important to design further studies that address the relationship between concentrations of nitrogen waste products in the water and the turtles' blood. Excessive nitrogen excretion could also be a result of low-quality dietary protein (Donoghue 2006). Dietary studies should be performed in this species to evaluate the effect of dietary protein on serum nitrogen compounds. Based on the observed increased serum uric acid and urea concentrations, renal function may have been compromised, resulting in the observed dehydrated condition of many turtles. The turtle kidney functions in the processes of excretion, osmoregulation, fluid balance, production of hormones, and the metabolism of vitamin D (McArthur et al. 2004).

Cholesterol levels were within the reference range of the species (62–215 mg/dl; Rangel-Mendoza et al. 2009), except at GOV, where lower values were found during February, which could be a sign of insufficient caloric intake (Wilkinson 2004a). Blood cholesterol showed seasonal variation, being lower in February and higher in August, which could be related to reproductive events (Raphael 2003). On the other hand, medians of total protein levels found in TAB turtles exceeded reference values for the species (1.2–3.0 g/dl; Rangel-Mendoza et al. 2009) during all samplings. Hyperproteinemia occurs during dehydration or hyperglobulinemia associated with chronic inflammatory disease (Campbell 2006).

Normal levels of creatinine in reptiles are typically less than 1 mg/dl, and high levels are related to dehydration and kidney disease in reptiles; however, creatinine is not a strong diagnostic parameter in reptiles (Campbell 1996). In the present study, all evaluated animals showed lower levels of creatinine than the reference values. A significant decrease in calcium was found during the last assessment (August), shortly before oviposition, which could be due to a temporary physiological variation related to reproductive activity (Raphael 2003). Normal values of phosphorus in reptiles vary between 1 and 5 mg/dl, the range found during the present study. On average, the values for glucose found in the present study were within the normal range for this species (25–98 mg/dl; Rangel-Mendoza et al. 2009).

The AP enzyme is found in many tissues in turtles (Anderson et al. 2011). Levels over 450 UI/l have been reported in immature animals and in females in preovulatory follicular stasis (Wilkinson 2004a). In our study, most AP levels were under 450 UI/l. Activity of AST seems to not be organ specific, and significant levels of this enzyme have been detected in the kidney, liver, and heart muscle of turtles (Anderson et al. 2011). In general, normal plasma AST activity is less than 250 IU/l, and an increase suggests hepatic or muscle injury (Campbell 2006). In our study, all AST levels were < 250 IU/l. Creatine kinase is considered a muscle-specific enzyme and is used to assess muscle cell damage. Levels of CK in the present study did not surpass those reported for juvenile C. mydas (Anderson et al. 2011) but exceeded maximum values found in captive head-started red-bellied cooters (Pseudemys rubriventris; Innis et al. 2007). LDH activity over 1000 IU/l may be associated with damage to the liver, kidney, and skeletal and heart muscle in reptiles, and levels are often elevated in sick animals with tissue damage (Wilkinson 2004a; Campbell 2006). This enzyme exhibited elevated values in both study sites that were higher at TAB, which may suggest compromise of any organs where this enzyme has activity.

Poor water conditions in ponds where D. mawii are reared could be one of the reasons explaining the variation in body condition. There was an excessive presence of organic matter at both study sites, estimated from high rates of BOD₅. This parameter in aquaculture ponds usually has values between 5 and 20 mg/l (Boyd and Tucker 1998), conditions that were exceeded in 5 of the 7 evaluations performed at TAB and 3 of 7 at GOV. The high values of BOD₅ likely explain the low values of dissolved oxygen. Organic matter pollution is higher at TAB, resulting in low measurements of DO, because the higher the amounts of organic matter, the lower quantity of DO in the water due to the oxygen required for the oxidation process (Navarrete-Salgado et al. 2004). In the wild, river turtles prefer well-oxygenated water and have highly vascularized pharyngeal papillae (Winokur 1988) that presumably act as a complementary respiration mechanism to the pulmonary system. This species can stay underwater for prolonged periods without surfacing to breathe (Ernst et al. 1997). Poor aeration in ponds where this species is reared could interfere with this physiological process and might lead to stress.

The evaluated ponds showed excessive enrichment of nitrogen and phosphorus. In freshwater systems, nitrogen can be found in nitrates, nitrites, and ammonia (Navarrete-Salgado et al. 2004). Low concentrations of this element limit primary production of the pond (phytoplankton growth), while in excess it will diminish water quality (Hargreaves 1998). Nitrogen enters aquaculture systems via food and pond fertilization as well as atmospheric nitrogen fixed by bacteria (Hargreaves 1998). The accumulation of nitrogenous wastes is a frequent consequence of inefficient periodicity in water change, overpopulation, overfeeding, and accumulation of organic matter or failures in water filtration (Roberts and Palmeiro 2008), situations present at both study sites.

Ammonia and nitrite are particularly toxic in freshwater systems (Hargreaves 1998), and concentrations of both compounds tend to increase when DO concentrations in water are low (Navarrete-Salgado et al. 2004). These undesirable conditions (lack of oxygen and excess nitrogen) are present in both study sites. Ammonia is excreted as a final by-product of the catabolism of protein in aquatic organisms and can be toxic when accumulated (Hargreaves 1998). Ammonia is present in ionized () and nonionized (NH₃) forms, with the nonionized form being the most toxic (Roberts and Palmeiro 2008). The proportion of NH₃ in the environment depends on pH and temperature, and ammonia is more toxic in warm waters with high pH (Roberts and Palmeiro 2008). The highest NH₃ in the present study was 0.96 mg/l, which was registered at TAB in March, when TAN was 2.239 mg/l, pH was 9, and water temperature was 28.8°C. That value exceeded the maximum acceptable levels for long-term exposure to nonionized ammonia for fishes and crustaceans that are between about 0.05 and 0.2 mg/l of NH₃ form at pH values above 7 (Boyd and Tucker 1998).

In fish, ammonia toxicity manifests as hyperactivity, convulsions, loss of balance, lethargy, and coma, but in aquaculture tanks it is expressed as sublethal reduction of fish growth or suppression of immunocompetence rather than death (Hargreaves 1998). Elevated ammonia levels in the environment may result in an elevation in body ammonia levels in fishes (Randall and Tsui 2002). To maintain a healthy aquarium for fish, it is recommended that ammonia levels measure 0 mg/l (Roberts and Palmeiro 2008). In turtles, environmental ammonia toxicity has not been thoroughly studied. In the Chinese softshell turtle Pelodiscus sinensis, 100% mortality was seen within 24 hrs after an intraperitoneal injection of 12.5 µmol/g of NH₄Cl. This species shows several defense mechanisms to counter ammonia toxicity, including an increase in excretion of ammonia, reduction of amino acid catabolism, and suppression of endogenous ammonia production (Ip et al. 2008). However, the effects of high concentrations of environmental ammonia on the physiology of other aquatic chelonians are largely unknown.

Nitrite toxicity in fish results in methemoglobinemia and hypoxia due to hemoglobin oxidation, which reduces the transport of oxygen to tissues (Hargreaves 1998; Roberts and Palmeiro 2008). The optimal level of nitrite in aquarium water is 0 mg/l (Roberts and Palmeiro 2008). Nitrite levels registered in the present study were markedly lower than the values reported to affect embryonic development in the Atlantic salmon Salmon salar (14 mg/l; Williams and Eddy 1989).

Another factor limiting productivity in aquatic systems is phosphorus. Its presence as orthophosphates establishes the nutrition state of a system, and high levels are indicators of eutrophic water (Navarrete-Salgado et al. 2004). Phosphate levels greater than 0.12 mg/l cause decreased hatching and increased incidence of larval deformities in common carp (Toor et al. 1983). The maximum values registered at GOV, but particularly at TAB, exceeded the reference value cited above. Nevertheless, its presence reinforces conditions of system enrichment that promote primary productivity of the pond. Enhanced organic matter in a water body reduces its photic phase, increases oxygen consumption for decomposition, elevates the abundance of suspended solids, and, in general, promotes the degradation of water quality.

In water bodies, an increase in pH can promote the formation of ammonia (Hargreaves 1998; Roberts and Palmeiro 2008) and might, as previously described, compromise the health of aquatic organisms, such as fish and invertebrates. Normal pH levels in the majority of productive aquatic systems fluctuate between 6 and 9, with changes during daytime related to photosynthesis (Boyd and Tucker 1998). The effects of pH variations on fish and crustaceans vary from death in highly acidic water (pH < 4) or highly alkaline water (pH > 11) to reduced reproduction and decreased growth (Boyd and Tucker 1998). Effects on pH changes are not immediate; however, chronic exposition may lead to stress, which subsequently induces immunosuppression and disease susceptibility (Roberts and Palmeiro 2008).

There are no specific water quality standards for the rearing of D. mawii as there are for other species, such as fish and crustaceans. However, Mexican environmental laws describe certain standards for the protection of aquatic life and wetlands (CONAGUA 2009), including a pH of 6.5 to 8.5, TSS of 30 mg/l, and fecal coliforms of 1000 MPN/100 ml. These normative parameters were not met in either of our study sites. The highest value of total coliforms at TAB was > 110 times the Mexican norm (CONAGUA 2009), reaching a maximum of 110,000 MPN/100 ml (Table 6). When coliform count reaches an excessive level, the situation needs to be corrected immediately by changing or sterilizing the water (Stamper and Semmen 2012).

Our results suggest that water quality at TAB was lower than at GOV, although in general water quality was poor at both sites. Excess feeding and wastes (feces and urine) have been gradually accumulating in the ponds since the beginning of the farms. There is evidence that this condition is due to excessive accumulation of organic matter and high densities of coliform bacteria in rearing ponds. This unsanitary scenario is exacerbated by the lack of filtration (physical, chemical, or biological), water changes, and control and regulation of food supplementation as well as the high density of turtles and their interaction with other species in the ponds.

In its natural habitat, D. mawii shows a preference for deep permanent water bodies (large rivers and their basins and lagoons), although it may occupy temporary shallow water bodies, which it inhabits during the rainy season and may not be able to abandon before the drought season begins (Vogt et al. 2011). Zenteno-Ruiz et al. (2010) reported a higher number of turtles in the Tabasquillo River, a habitat with water depths ranging from 0.7 to 1.56 m, 0.91-m transparency, and DO of 8.02–0.65 mg/l. This water body is considered to be a high-quality habitat for D. mawii (Rangel-Mendoza 2007). However, in the turtle farms studied, water conditions of transparence and DO were very different from those reported in the natural habitat of D. mawii and are associated with a poor-quality environment that may have a negative effect on turtle condition and health.

Dermatemys mawii preferentially inhabits environments with abundant vegetative cover (branches, logs, or roots) over the surface of the water (Zenteno-Ruiz et al. 2010). These are found in areas of slow water currents associated with submerged vegetation fragments (Vogt et al. 2011). The confinement design for the turtles in the present study does not consider shelter areas, be they natural or artificial or submerged or exposed, a situation that can lead to stress in the turtles.

Little is known regarding the effect of water quality on the physiology or health of turtle species. As far as water quality management concerns, recommendations exist to ensure their well-being, focusing mainly on frequent water changes of the pond containing the animals (Mautino and Page 1993; Boyer and Boyer 2006).

Stocking density seems to be a crucial aspect that has repercussions for turtle health and water quality. Captive juveniles of the soft-shelled turtle Pelodiscus sinensis managed at lower density exhibited a higher survival rate, higher growth rate, and greater transfer of consumed energy to growth. Furthermore, the excretion of nitrogenous wastes to the environment was relatively lower with reduced stocking density (Jing and Niu 2008). In the studied farms, to avoid the high prevalence of external lesions, animal density per pond must be reduced. Low turtle density also decreases the accumulation of wastes in the water. If stocking density is reduced and water conditioning measures are implemented, the quality of the aquatic environment will improve, and the turtles' captive breeding may show better results.

It is imperative to improve water quality in the ponds where river turtles are currently maintained. There are several ways to ensure this condition, including frequent substantial water changes and installation of mechanical, chemical, and biological filtration. Although filtration may not eliminate the need for water changes, it may reduce the water change frequency (Mautino and Page 1993; Boyer and Boyer 2006). Water changes also can reduce excessive fecal coliform counts since water sterilization via ultraviolet radiation and ozone is not viable given the rural and austere conditions of studied farms. Improving water quality of the ponds will translate into more favorable conditions for the turtles' homeostasis as well as less favorable environmental conditions for opportunistic microorganisms to grow.

Another management factor that is important for the health of captive turtles is their diet. It is necessary to determine the dietary protein requirement of this species since the protein levels of the current diets may be higher than the species' requirements. In light of its herbivorous natural diet, it is likely that captive turtles may show better health if diets with lower protein levels are offered. Furthermore, the type of dietary protein, animal or vegetable, that the turtles are being fed needs to be evaluated. Offering animal protein to herbivorous turtles may predispose them to hyperuricemia and consequently to gout (McArthur and Barrows 2004). Routinely, the turtles from the present study have been fed a food for omnivorous fish containing different types of animal protein. Experiments based on enzymatic digestive activity of D. mawii are currently being conducted to determine the species' nutritional requirements and to formulate an artificial diet to use in their captive breeding.

In conclusion, the present study demonstrates that captive Central American river turtles in the 2 study sites are not in adequate general health. The factors that appear to be most important in relation to health include water quality management, overcrowding, and the diet of the turtles. Future studies are needed to address specific etiologic agents, causing some of the health problems detected here. This should be addressed throughout properly designed pathological, virological, and microbiological studies. To achieve this, both ante- and postmortem examinations of turtles are mandatory.

Despite all the efforts made in recent years by federal and state governments in Mexico for the conservation of D. mawii through captive breeding, the results have not been entirely successful due in part to the poor health found in the captive colonies (Rangel-Mendoza et al. 2009; Vogt et al. 2011). It is imperative to incorporate the results and recommendations derived from this and other recent studies to ensure the feasibility of captive rearing as an alternative to the recovery of this highly endangered species.