Study area

The study was performed from October to December 2013 in three areas located in the Araguaia River Basin, state of Tocantins, Brazil (Figure 1). As the three study sites have different management processes to P. expansa are expected different parasitism levels. The first study area was located in the Javaés River (9º53’22.20”S, 50º8’55.70”W), municipality of Pium, near the Araguaia National Park and north of Bananal Island. The area consists of preserved vegetation formed by typical fragments of Amazon forest and Cerrado (tropical savannah) vegetation. The second study area was a commercial breeding facility of P. expansa at the Praia Alta Ranch (10º44’05.33”S, 49º51’28.92”W), located in the municipality of Lagoa da Confusão, at the banks of the Formoso River. This facility holds approximately 30,000 animals distributed in four tanks for growth and fattening. The third study area was a breeding facility in the Xambioá Indigenous Land (7°05'10.09"S, 49°11'02.99"W) in the municipality of Santa Fé do Araguaia, an indigenous reservation located on the banks of the Araguaia River in the middle of the Amazon forest and inhabited by the Karajá indigenous people. That facility has only one tank with approximately 1,500 specimens of P. expansa.

Figure 1 Map of state of Tocantins showing locations of study areas. 

Animal and blood sampling

To estimate the sample size was used an equation for a finite population (^{DANIEL & CROSS, 2013}), and data from a previous study were used to perform the calculation (^{ARMOND, 2008}). In this study the prevalence was 99% of Haemogregarina sp., the size of the P. expansa wild population (N) was estimated at 2,100 individuals (Malvasio A, unpublished data), level of confidence 95%, and standard error 5%. Thus, the estimated sample size was 76 individuals of P. expansa. However, in the present study only 25 turtles were sampled in each of the three selected locations, totaling 75 animals. Considering the high prevalence rates in the study of ^{Armond (2008)}, which sampled 80 individuals, we believe that our sample size represents the universe of the host population.

Samplings were restricted to animals with a carapace width between 15 and 60 cm. In Brazil, this carapace width range is required for the slaughter of chelonians kept in captivity (^{BRASIL, 2008}). In the Javaés River, wild animals were captured either by fishing with barbless fish hooks or by trawling (^{PORTELINHA et al., 2013}) and were marked by drilling the tenth marginal scute of the carapace (^{MALVASIO et al., 2002}). In the breeding facilities, chelonians were manually captured with the aid of a fishing hand net with a 40 cm diameter and 50 cm bag depth. The carapace width and weight of each captured specimen were measured, and then they were visually inspected for the presence of ectoparasites, such as leeches, and for possible infections.

Blood samples were collected by tail venipuncture, and then the chelonians were released back to their collection sites. Blood smears were fixed with absolute methanol and stained with 10% Giemsa (^{EISEN & SCHALL, 2000}). The blood smears were examined for up to 20 minutes under a microscope (400x). The level of parasitemia was estimated by counting the parasite forms in 20 replicates of 100 erythrocytes (RBC) per field, totaling approximately 2,000 erythrocytes (GODFREY et al., ¹⁹⁸⁷, ¹⁹⁹⁰). This research was approved by IBAMA and the Brazilian Ministry of the Environment (approval number 38193-2/2013).

Statistical analysis

ANOVA was used to evaluate the differences in parasitemia among the animals collected in the different study areas. However, in the Xambioá Indigenous Land there was not positive animals, thus not entered in the analyzes. The correlation between parasitemia and body condition index (body weight in kilograms divided by the carapace width in cm) was verified by a simple linear regression. The BioEstat 5.0 software (^{AYRES et al., 2007}) was used for the statistical analysis. Values were considered statistically significant at p < 0.05.

Physical examination

The deformities and ulcerations observed in the animals of the Praia Alta Ranch were most likely related to management practices and to the tanks’ infrastructure. In the Xambioá Indigenous Land, where the Amazonian turtles were healthy, the bottom and the sides of their only tank were coated with cement. The tank water came from the Araguaia River, and the tank was cleaned by changing the water every two days. However, there was no screen or filter in the water supply line to prevent the entry of leeches and microorganisms. Food was supplied every day and consisted of commercial fish food, various fruits, leaves from the riverbanks’ trees, cassava and corn. In the breeding facility at the Praia Alta Ranch, where the animals were physically debilitated, the bottoms of the tanks were soil and only some sides were cemented. The diet consisted only of rice straw and crushed corn, supplied twice a day, every day. Cleaning was performed by changing the water every four days. Similar physical conditions to the ones observed in the P. expansa specimens kept in the Praia Alta Ranch were also observed in specimens of Dermatemys mawii in a study performed by ^{Rangel-Mendoza et al., (2009)}. According to these authors, the D. mawii specimens were kept in captivity under inadequate management conditions, which led to relatively poor physical condition and lesions. Wild chelonians also examined in that study were healthy (^{RANGEL-MENDOZA et al., 2009}).

Prevalence

This is the first report on the prevalence and parasitemia of Haemogregarina sp. in P. expansa chelonians on a population scale in Brazil (Table 2). The only previous records were in four animals collected in the Araguaia River, state of Goiás, Brazil (^{MUNDIM et al., 1994}) and in 75 of 96 (78%) specimens examined in Peru (^{PINEDA-CATALAN et al., 2013}).

In the current study, all of the wild chelonians from the Javaés River and the chelonians from the breeding facility of the Praia Alta Ranch were infected by hemogregarines. It is important to highlight that the specimens kept at the Xambioá Indigenous Land and the Praia Alta Ranch were captured newly hatched from the wild and soon transferred to the captive facilities. Thus, they were not exposed to blood feeding by leeches in the wild. In addition, in both locations, the water in the tanks was obtained from nearby rivers and transported by pipes that did not have any physical barriers to prevent leech entry into the tanks. In the Xambioá Indigenous Land, the absence of chelonians infected by hemogregarines was most likely due to more efficient management of the animals in terms of cleaning and feeding, a lower density of animals in the tank or even an absence of leeches in the surrounding area. In the Xambioá Indigenous Land, there has been no report of leeches occurring in animals or in the breeding tank. In contrast, in the Praia Alta Ranch, the presence of leeches has been reported (^{ARMOND, 2008}). In the latter location, the tanks’ water is obtained from the Formoso River, and leeches may have entered the tanks in parasitized fishes or even with floating vegetation when they were not on their hosts (^{SAWYER, 1986}).

The absence of vectors in wild chelonians of the Javaés River may be related to the period in which samplings were performed. ^{Koffler et al. (1978)} reported lower ectoparasitism by leeches during the summer months and during the reproductive period of Clemmys insculpta. In the region where the current study was performed, the climate, according to Köppen, is classified as tropical wet and dry (Aw), and summer is from November to April with 95% of the rainfall occurring during this period (^{FERREIRA & CASTRO, 2003}).

Parasitemia

The current study also recorded the second-highest parasitemia by hemogregarines in the Podocnemis genus and the highest parasitemia for the P. expansa species (Table 2). In P. unifilis of the Tapajós River, state of Pará, Brazil, the recorded level of parasitemia was 6% (n = 72) (^{SOARES et al., 2014}). Until recently, the highest levels of parasitemia recorded were 1.7% (n = 7) in Chelydra serpentina (^{MCAULIFFE, 1977}) and 1.6% (n = 45) in Testudo graeca ibera (^{MIHALCA et al., 2008}).

Captive-bred reptiles under inadequate management, under confinement stress and with nutritional deficiencies are more susceptible to parasites (^{TELFORD, 1971}; ^{WILSON & CARPENTER, 1996}). Therefore, a higher prevalence and parasitemia of hemogregarines were expected in the animals kept in captivity. Although the prevalence levels were the same in the wild animals of the Javaés River and the specimens of the breeding facility of the Praia Alta Ranch, the mean level of parasitemia was significantly different in these locations, 4.2% vs. 1.2%, respectively. This variation is most likely due to two aspects of the hemogregarines’ transmission cycle in chelonians: leeches and the environment. It is likely that when conditions are amenable for the existence of leeches in the environment (wild or captive), infections by hemogregarines may occur continuously, maintaining high levels of parasitemia.

Studies have shown that the exposure of leeches to pesticides such as atrazine has negative effects on their growth, egg production, feed intake and survival rate (^{STREIT & PETER, 1978}). In the tanks of the Praia Alta Ranch, the leeches are most likely exposed to environmental pressure from pesticides because the Formoso River that supplies the tanks traverses large tracts of agricultural areas with soybean, corn and rice crops. Therefore, the presence of pesticides could affect the leech population, leading to a decrease in the reinfection and parasitemia levels of the chelonians.

In a study of the parasitism of Geoffroy's side-necked turtle, Phrynops geoffroanus, by hemogregarines and leeches, ^{Campos Brites & Rantin (2004)} recorded a prevalence of hemogregarines of 15.4% (n = 26) in an agricultural area and 37.5% (n = 32) in an urban area. Leeches were found only in the urban area. According to ^{Campos Brites & Rantin (2004)}, this variation is most likely related to the water quality of the Uberabinha River. Although there were pesticide residues in both sampled locations, the concentration of atrazine was 18 times higher in the rural area and was suggested as the main cause of the absence of leeches in the rural area.

The observed lack of a relationship between P. expansa body condition index and parasitemia suggests that parasites do not adversely affect the health of chelonians. Similarly, ^{Rossow et al. (2013)} did not find a significant correlation between parasitemia and body condition index in the aquatic chelonians Rhinoclemmys funerea, Kinosternon leucostumum and K. scorpioides from Costa Rica. However, hemogregarines can be harmful to their hosts when associated with other diseases despite the fact that they are not considered pathogenic (^{OPPLIGER et al., 1996}; ^{PEIRCE & ADLARD, 2004}). In wild snakes with high levels of Hepatozoon sp. parasitemia, decreases in the growth, body condition and survival of juveniles were observed. In these reptiles, low parasitemia levels were only observed in animals of advanced age (^{MADSEN et al., 2005}; ^{UJVARI & MADSEN, 2005}).

Amazonian chelonian ranching aims to prevent hunting and illegal meat trade. For the Praia Alta Ranch, it is necessary to change the tanks’ infrastructure and management protocol to achieve more efficient and reliable production. The current study reported the occurrence of hemogregarines at the population scale of its host, the side-necked turtle P. expansa, and contributed to expanding the knowledge about hemoparasites in chelonians in Brazil.