Study area

Field work was conducted at Nhumirim Ranch (18°59’S, 56°39’W), a 4400 ha research station of The Brazilian Agricultural Research Corporation (Embrapa) located in the Pantanal wetlands. Pantanal is characterized by sandy soil with a mosaic vegetation of semi-deciduous forest, dispersed shrub vegetation, and seasonally flooded fields. Human population density is low (< 2 people per km²) and the main economic activity is cattle ranching [50]. The climate is tropical, with two marked seasons: a wet season (October to March) and a dry season (April to September). In the Nhumirim Ranch, the reproductive season of coatis starts by early August and extends to late November [51].

Capture procedures

From May 2006 to February 2009 we captured coatis up to four times per year, using wire box traps (1m x 0.40m x 0.50m) placed in a trapping grid of 7.2 Km² and with occasional traps placed outside of the grid. Traps were baited with bacon, set late in the afternoon and checked in the following morning. We anesthetized individuals with an intramuscular injection of Zoletil^@50 (Virbac^®; tiletamine hydrochloride and zolazepan hydrochloride, 10 mg/Kg), ear-tagged them with numbered colored tags (Nasco Rototags^®) and marked them with subcutaneous transponders (AnimalTag^®). Body size measures, body mass, as well as tooth eruption and wear were recorded and used to age individuals [52]. Animals were released at the capture site.

We collected and stored blood samples (up to 5mL) in Vacutainer^® tubes with EDTA. In addition, we prepared eight blood smears from captured animals and collected fecal samples from beneath traps or via fecal loop. Handling and sample collection procedures were standardized to minimize biases in the hematological profiles caused by an alarm response due to capture stress [53].

Ethics Statement

Field permit to capture coatis in the study area and to sample collection was granted by the Brazilian Government Institute for Wildlife and Natural Resources Care (IBAMA, first license #183/2005 –CGFAU/LIC; last license #11772–2). Procedures adopted in this study were approved by the University of Missouri Animal Care and Use Committee (protocol #4459) and are in accordance with the recommendations in the Guidelines of the American Society of Mammalogists for the use of wild mammals in research [54]. All animals were monitored until recovery from anesthesia and all efforts were made to minimize stress. Appropriate biosafety techniques and individual protection equipment were used during animal handling and sample manipulation.

Hematology

Packed cell volume (PCV), total red blood cell (RBC) counts and total white blood cell (WBC) counts were quantified within 8 hours of blood collection [13]. A subsample of whole blood was diluted 1:200 in distilled water and then placed in a Neubauer chamber where RBCs were immediately counted; total WBC counts were done in the Neubauer chamber after diluting whole blood at 1:20 in a 4% acid acetic solution for 10 minutes [13].

Blood smears were air-dried and fixed with methanol (Merk^®). Slides were stained with Giemsa’s azure eosin methylene blue solution (Merk^®) for differential leukocyte counts (that is, count of band neutrophils, mature neutrophils, eosinophils, lymphocytes, monocytes and basophils) and observation of blood cell morphology [13]. We diluted 10 mL Giemsa’s azure eosin methylene blue solution with 190 mL buffer solution (pH 6.8–7.0) and stained the smears for 15−20 minutes. We counted and classified 100 leukocytes across the entire cell monolayer using the crossing technique [13]. For animals showing leukocytosis (>25,000 WBC/ mm³), we counted 200 leukocytes to improve the accuracy of estimates. In addition, we characterized RBC morphology (cell shape and color), and recorded the number of metarubricytes (young, nucleated red blood cells). Red blood cells in the individual blood smears were characterized as altered or normal. They were qualified as altered when polycromasia (cells with distinct colors) and poikilocytosis (distinct sizes) were predominant or when more than 3 metarubricytes/100 leukocytes were found in blood smears. We also recorded the number of platelets (averaged number in 10 microscope fields) found during leukocyte counts. Blood smears for which we observed platelet aggregation were not included in the analysis as platelet aggregation may be an artifact of smear preparation and can influence total platelet count.

Parasite identification and quantification

We used the microhematocrit centrifuge technique (MHCT) [55] with blood subsamples from the captured animals to assess high parasitemia infections by T. evansi [40] and microfilariae; and hemoculture (HC) to assess high parasitemia by T. cruzi. Microfilariae are a blood-borne stage in the life cycle of filarid worms. We estimated the average number of microfilariae in 10 randomly chosen 100× microscope fields during MHCT.

Hemoculture (HC) was used to assay for epimastigote forms of T. cruzi and for isolating them for subsequent molecular characterization, as described elsewhere [40]. Trypanosoma evansi was directly identified in the stained blood smears after detection of trypomastigote forms with the MHCT technique. Trypanosoma cruzi and T. evansi can be distinguished on Giemsa-stained smears, as T. evansi does not show a visible kinetoplast while T. cruzi has a terminal/subterminal kinetoplast and shows a characteristic “C-shape”. Moreover, T. evansi appear in high numbers in the MHCT but does not grow in HC. Positive HC and/or MHCT were used as indicators of high parasitemias and infectivity competence [43], since detection of trypomastigote forms in the bloodstream means the vectors have relatively higher chances of becoming infected (and therefore potentially transmitting) the parasite when taking a blood meal; negative HC/MHCT, in turn, are equivalent to low/no parasitemia [40]

Beginning in August 2007, we quantified eggs of helminths and oocysts of gastrointestinal protozoans found in fecal samples as the number of eggs/oocyst per gram of feces. A subsample of 1−3 g of feces was weighed to the nearest 0.01 g, suspended in 20 mL of 10% formalin and filtered through a gauze mesh. The diluted sediment was then divided in two 15mL tubes, one of which was analyzed by sugar solution flotation (sugar density 1.27) and sedimentation techniques [56]. After sedimentation, the pellet was suspended with 1mL of 10% formalin and a subsample of 80 μL from this solution was placed on a slide for analysis [57]. Slides from the sugar flotation and sedimentation techniques were analyzed at 100× and all egg and oocyst types were photographed under 400×. Eggs and oocyst were measured and classified into morphotypes. The total number of each egg and oocyst found with the sedimentation technique was extrapolated to the total weight of feces present in the 15 mL tube (usually 1.5 grams). To this total, we added the eggs found during the sugar flotation done with the same material.

Identification at the species level was not possible due to lack of information on helminths of coatis. We therefore used an aggregate-scale approach [11], that is, we identified and grouped eggs/oocysts at the most specific taxa possible. Therefore, in addition to the filarids (microfilariae) quantified with the MHCT technique, we considered four groups of gastrointestinal parasites in the aggregate-scale analyses: Eimeriidae spp (coccidians), Oligacanthorhynchidae sp. (acanthocephalans), Strongylida spp. (hookworms) and Cestoda spp. Other groups had low prevalences and were not analyzed. Aggregate-scale analyses as well as the egg/oocyst and filarid quantifications employed are not refined methods to quantify parasite abundance due to factors such as sampling heterogeneity and resource competition between parasitic species leading to density-dependent suppression of egg production [58,59]. However, physiological effects, if strong, can nonetheless be detected using such aggregate-scale approaches [11].

Statistical analyses

There is no clear consensus regarding the most suitable condition index [60,61]. We opted to assess variation in coati body condition based on the standardized residuals from an ordinary linear regression between body mass (g) and head-body length (mm) of individuals, while accounting for age and sex effects. This should circumvent the effects of animal growth on the condition index. Therefore, the residuals were calculated for males and females separately because coatis are dimorphic in size and after residual calculations, only greater than 2 years old coatis were considered. We excluded pregnant/lactating females from the analyses, as we know this condition affect both their hematological profile and body condition, regardless of their health condition and the presence of parasites. Seasonality was broadly categorized as reproductive (August to November) or non-reproductive (January to early July). We combined Log₁₀RBC and PCV values on subsequent analyses by using the factor scores of the first component of a principal component analysis (PCA), since those variables are highly correlated. Total WBC and leukocyte differential numbers were log_e-transformed for normalization prior to analyses, except for band cells, which showed a negative binomial distribution and were analyzed accordingly; and platelet numbers, which were squared root transformed before analysis. The RBC morphology was compared using one-tailed Fisher’s Exact tests between categories of qualitative variables (host sex, season and T. cruzi and T. evansi high parasitemias). In these analyses, recaptures were not considered to avoid pseudo-replication (see below).

We used an information theoretic approach to contrast the relative performance of models that predict measures of host health parameters. We created linear models accounting for seasonal dynamics (reproductive vs. non-reproductive season), host sex, and parasite abundance to investigate the relative importance of these factors for body condition and hematological parameters of individual coatis. The standardized residuals of body size (body condition), the PCA factor representing PCV and RBC, total WBCs and each leukocyte type count, as well as platelet numbers were the response variables in the models. Basophils were not analyzed because they were relatively less encountered. The number of band cells was modeled using generalized linear models with a negative binomial distribution and log link. Data from recaptured individuals (obtained about every 3 months) were eliminated from data modeling to avoid pseudo-replication, but were included in mean/median and 95% confidence interval estimations. For seven recaptured animals showing low and high T. cruzi parasitemias in different field excursions, we compared the numbers of distinct blood cells between periods of low and high parasitemias using paired t-test. A similar approach was not performed for T. evansi because of sample size limitations.

Because combining all independent variables in a single set of models would result in an overly complex analysis with many models and a reduced sample size of egg/oocyst counts, which were initiated in 2007, we ran two separate sets of analyses. Sample sizes based on blood collection were larger than fecal samples, so we first contrasted a set of models with T. cruzi, T. evansi, microfilariae (hemoparasites) and extrinsic variables (host sex and seasonality) and then proceeded by adding each gastrointestinal parasite to the best-fitting model of hemoparasites and extrinsic variables. In this set of models, high parasitemias by T. cruzi and T. evansi, host sex, and seasonality were qualitative descriptors and average microfilariae/field was a quantitative variable. We created model sets including all possible additive combinations of independent predictors of hemoparasites and extrinsic variables (32 models in total); furthermore, interaction terms were included in additional models whenever investigation of predictor vs. response variable plots revealed possible interactions between these variables.

Candidate models represented independent hypotheses, and model fit was compared using Akaike Information Criterion corrected for small sample size (AICc) and for overdispersion when needed (QAICc). Models were ranked based on the difference between the best approximating model (model with the lowest AICc or QAICc) and all others in the set of candidate models (ΔAICc, ΔQAICc). Models with differences within two units of the top model were considered to have strong empirical support and were considered competitive models [62]. However, larger models that differ from the best fitting model by one parameter and had essentially the same maximized log-likelihood as the best model were not considered to be competitive; when there was no clear best-fitting model, the relative importance of each predictor was quantified by adding the Akaike weights across all the models that included that predictor (variable weight) [62]. Variables weights lower than 0.40 were not discussed due to the relatively low variable importance. In addition, variable weights for interaction terms were not calculated because those were included in a relatively small proportion of models [62]. When the best-fitting model was the intercept-only model, we did not proceed with model analysis and interpretation.

Since sample size for gastrointestinal parasites was small, we did not run models with all possible combination of gastrointestinal parasites, hemoparasites and extrinsic variables. Instead, we added each gastrointestinal parasite predictor to the best-fitting hemoparasite and extrinsic variables model (see above) and observed whether the model fit was improved. In this way, we accessed the relative importance of hemoparasites, gastrointestinal parasites and extrinsic variables on coati health. The global model containing all gastrointestinal parasites was included in the set of models in order to calculate c-hat and verify the need of using QAICc rather than AICc for model comparison. A total of 63 fecal samples were analyzed for gastrointestinal parasites, but sample size used to run the models varied from 16 (for models in which platelet numbers were the dependent variable) to 31 (for models of most hematological variables); this is so because a) we considered only one capture for each coati individual in the model analyses and b) not all samples analyzed for gastrointestinal parasites had the other independent variables available.

Possible effects of Trypanosoma evansi

The detection of T. evansi presence was a predictor of lower numbers of RBCs and PCV (Table 1 and Fig 1), and of alterations in the overall morphology of RBCs and number of metarubricytes. This result confirms a preliminary investigation in which the impact of T. cruzi and T. evansi parasitemias on PCV values of the coatis was evaluated [40].

RBC and PCV values of coatis with high T. evansi parasitemias were similar to those found in other studies; this parasite causes a macrocytic normochromic anemia in coatis that with time may become normocytic normochromic, i.e., anemia without morphological changes in RBCs [37,45]. Many species of domestic and wild mammals show anemia and RBC morphological changes when infected by T. evansi [63–66]. Yet, despite the hematological alterations, RBC and PCV values were not much below values reported for captive uninfected animals [45,67]. A possible explanation for why most of the captured animals with high T. evansi parasitemias did not show very low RBC and PCV values is that although free-ranging coatis may develop anemia, severely affected individuals might not survive long in the wild. Alterations in RBC and PCV values can also be accompanied by a decrease in the number of platelets [68,69], as it was also found for coatis in this study. There is an array of possible reasons for a decrease in platelet numbers associated with T. evansi infection, including reduction of the platelet life span, splenic and hepatic sequestration, self-immune response and/or disseminated intravascular coagulation, characterized by systemic activation of blood coagulation, due to constant contact with trypomastigote forms and the vascular endothelium [12,69,70].

Changes in white blood cells due to infection by T. evansi in coatis have not been examined as closely as have changes in RBC and PCV [37,40]. A lower total WBCs in coatis with high parasitemias by T. evansi was previously detected [40], and was mainly caused by a lower number of neutrophils (including band cells) in males with high parasitemias of T. evansi. Species in the order Carnivora show varied responses to T. evansi infection regarding the number of circulating neutrophils [65,71,72], and reasons underpinning this variation are not known. Coatis also showed a reduced number of eosinophils, which may be a consequence of acute infection [12]. Waves of high parasitemia caused by variable surface glycoproteins are characteristic of this salivarian trypanosome and may be responsible for acute episodes during the course of infection. Eosinopenia (low number of eosinophils) was reported in cats experimentally infected with T. evansi [72], but the same was not observed in dogs [65,71]; indeed, interpretation of such results is troublesome, as eosinophil reference values for many species may be zero cells per mm³ of blood.

Finally, we observed a lower body condition of coatis with high parasitemias of T. evansi especially during the reproductive season. The lower body weight reflects a broader energetic challenge that is likely biologically meaningful for reproduction and survival of coatis, as it is for other species [6,8,73,74]. Trade-offs on resource allocation between host immune response and life history traits are expected and can play a significant role on host survival and/or reproduction [75], In this context, we speculate that adult coati males showing high T. evansi parasitemias during the reproductive season may be at a disadvantage when competing for access to groups of females.

Possible effects of Trypanosoma cruzi

High parasitemias of T. cruzi were not associated with changes in RBC or PCV values of coatis, except perhaps for the three animals also infected with T. evansi (TC+TE+; Fig 1). However, total WBCs and neutrophils were slightly lower in coatis with high parasitemias of T. cruzi (Table 1 and Fig 2), whereas the decrease in the average number of monocytes was more pronounced. In mice, acute T. cruzi infection leads to alterations of the hematopoietic system, resulting in leukopenia amongst other alterations [76], but infection by T. cruzi has variable effects on host hematological parameters of different species [77–80]. Monocytes and lymphocytes, however, are believed to play an important role in halting T. cruzi during the early, acute stage of infection [79][81], with peripheral monocyte numbers increasing due to subacute and chronic inflammatory responses usually caused by bacterial and protozoan infections [12]. Therefore, the lower number of mature neutrophils and the decrease in monocyte numbers of animals with high parasitemias of T. cruzi suggest that coatis may not respond well to high parasitemias of T. cruzi. Nevertheless, some coatis kept high T. cruzi parasitemias for 7.5 months on average (N = 8) [40], which may mean that the effects of T. cruzi on coati health are more related to morbidity than mortality.

Possible effects of other parasites

We do not know the microfilariae species occurring in coatis from the Pantanal and the impact of most of these filarids in wild species is unknown. Coatis are wild hosts for Dirofilaria immitis in Argentina [82] and Brugia guyanensis in French Guiana [83]; in Brazil, they are infected with D. repens and D. incrassata [84]. Infection by D. immitis cause moderate normocytic normochromic anemia in dogs by reducing RBC, hemoglobin concentration and PCV values [85][86]. We found a contribution of abundance of the microfilariae to reduced body condition, RBC and PCV values, as well as band neutrophil numbers in coatis. Given the high prevalence of microfilaria in adult coatis and the absence of animals with clinical signs of infection, we suggest this parasite by itself has a minor effect on coati health.

Overall, gastrointestinal parasites were not associated with hematological parameters or decreased body condition, even though other studies have found an association between the presence or abundance of such parasites and a decrease in host body condition [5,6,9,87]. However, it is important to consider that sample size for gastrointestinal analyses was small and the statistical power for these analyses is therefore low. Despite that, we found that intensity of cestode eggs was positively associated with number of eosinophils. Eosinophils are involved in the regulation of allergic and inflammatory responses, playing an active role in almost all types of inflammatory processes. Eosinophilia (increased number of circulating eosinophils) can be found as a response to several parasitic infections [12,14] and may occur as a response to cestode infections. The large size of helminths precludes them to be phagocytized and as a consequence, eosinophils release toxin molecules that combat helminths extracellularly. The abundance of coccidian oocysts was also slightly associated with increased numbers of circulating neutrophils. Coccidian infection can cause diverse alterations in the blood parameters of experimentally infected hosts [88,89] and their effect on coatis has never been investigated.

Identifying parasite community composition and relative species abundance from fecal samples is a challenge and not always feasible [11]. Even with the recent advances in molecular techniques for identification of helminth eggs and larvae [90][91], identification may be too costly or inefficient, especially if the host species is poorly known and the chances of finding new species of parasites is high. Despite such methodological drawbacks, analyzing aggregate-scale parasite data should not be discarded because strong physiological effects can still be detected [11] and the obtained results may guide future research on specific parasite groups. In this study, even though we had a small sample size for analyses involving gastrointestinal parasites and identification of helminths was not done at the species-level, we observed that parasites such as filarids, cestodes and coccidians are common in coatis from the Pantanal and that they may affect the health of coatis. Future research should therefore focus on the species-level identification of these species as well as on more refined methods of parasite quantification.

Reproductive seasonality and sex differences

Overall, coati body condition was lower during the reproductive season, and the relation between high parasitemias by hemoparasites and changes in the hematological parameters was stronger during that season (Figs 1 and 2b). A reduction in body fat during the reproductive season has been observed for N. narica from Tikal National Park, Guatemala and probably reflects reproductive stress, perhaps associated with reduced food consumption [92]. During the reproductive season, poor nutrition and investment in reproduction weaken the immune system in ways that could affect the ability of hosts to defend against parasites [20–22].

Female coatis apparently cope better with infection by T. evansi than do males. The lower WBCs and neutrophils related to T. evansi detection were usually greater in males, especially during the reproductive season. Estrogen seems to play an important role in resistance to T. cruzi [93,94] and many other parasites, whereas testosterone depresses the immune system of males [23,25]. In addition, coatis are polygynous, show sexual dimorphism, and males strongly compete for females during the reproductive season, which sometimes result in wounds [51]. Given these behavioral and morphological characteristics, hematological differences between coati males and females are expected to be pronounced [95]. This would also contribute to an even stronger decrease in immune response during the mating season, when testosterone levels in coati males peak [96]. However, it is important to notice that sample sizes for positive T. cruzi or T. evansi females were small.

Finally, we observed that neutrophil numbers were higher mainly in males during the reproductive season. An increase in the number of circulating young neutrophils (shift to left) occurs when demand from the bone marrow increases and a large percentage of band cells are released in the bloodstream [12]. This finding may indicate that the challenge to the immune system of males is greater during that season, in agreement with the other findings described above.

Concomitant infections

There is an increasing need for investigation of the concurrent impact of multiple parasites on host health [8,49,97]. The role of simultaneous infection by multiple parasites on host health is usually overlooked even though coinfection is the rule [49,98,99]. In this study we detected effects of concomitant infections on host health. For instance, filarids along with T. evansi were present in the best models explaining variation in body condition and RBC/PCV (Table 2). Likewise, WBC and neutrophils were lower in coatis infected with T. cruzi or T. evansi, and we found some evidence that such numbers may be even lower in double infected males (Fig 2a), a pattern suggested by Alves and co-workers [40].

The order of infection by multiple parasites and its effect on host health should be an important focus for future investigations [100], although it may be difficult to control for in free-ranging animals. In this context, experimental studies may be useful to disentangle the effect of each parasite on host health as well as to clarify the mechanisms by which parasites may affect host health. For instance, Rodriguez and co-workers [101] verified that the number of blood-circulating T. cruzi trypomastigotes in mice varied as a function of the development of early experimental infection with the cestode Taenia crassiceps, which would alter host susceptibility to T. cruzi infection and the magnitude of peaks in T. cruzi parasitemia. Therefore, patterns of high T. cruzi parasitemias in coatis may be at least partially influenced by co-infection with other parasites. Indeed, concurrent micro- and macroparasite infections may down-regulate different immunologic pathways. In general, the immune response to intracellular parasites involves T helper 1 (Th1) cells, while response against extracellular parasites are mediated by T helper 2 (Th2) cells. Cytokines produced by Th1 and Th2 cells suppress one another’s immune function [102–104]. Notwithstanding, field studies such as this work with coatis and Trypanosoma allow a more realistic approach when investigating the effect of parasites on host health in the wild. If a parasite has economic, veterinary or public health importance and the host is thought to be a reservoir, in situ studies become essential to understand and predict parasite persistence and impacts in reservoir populations.