Prevalence and Lineage Diversity of Avian Haemosporidians from Three Distinct Cerrado Habitats in Brazil

Habitat alteration can disrupt host–parasite interactions and lead to the emergence of new diseases in wild populations. The cerrado habitat of Brazil is being fragmented and degraded rapidly by agriculture and urbanization. We screened 676 wild birds from three habitats (intact cerrado, disturbed cerrado and transition area Amazonian rainforest-cerrado) for the presence of haemosporidian parasites (Plasmodium and Haemoproteus) to determine whether different habitats were associated with differences in the prevalence and diversity of infectious diseases in natural populations. Twenty one mitochondrial lineages, including 11 from Plasmodium and 10 from Haemoproteus were identified. Neither prevalence nor diversity of infections by Plasmodium spp. or Haemoproteus spp. differed significantly among the three habitats. However, 15 of the parasite lineages had not been previously described and might be restricted to these habitats or to the region. Six haemosporidian lineages previously known from other regions, particularly the Caribbean Basin, comprised 50–80% of the infections in each of the samples, indicating a regional relationship between parasite distribution and abundance.


Introduction
Understanding the influence of different landscapes on parasitehost interactions in wild populations is of particular relevance in a world undergoing dramatic change in climate. Such environmental change can influence parasite-host relationships in many ways [1], particularly by altering the transmission of endemic pathogens of wildlife, which can lead to the emergence of new diseases in wild populations, domesticated animals, and humans [2]. Habitat degradation and diversity loss due to increased conversion of natural habitats to agricultural uses and urbanization influence the distribution and abundance of wildlife species and thus may be a major driver of change in the ecology of pathogen transmission.
Increases in the occurrence of human malaria in degraded habitats have been related to habitat change altering the distribution and behavior of human malaria vectors [3,4]. As human populations continue to grow exponentially, and as globalization increases with more travel and trade, anthropogenic pressures on wildlife habitat and populations also will increase [5].
The prevalence of malaria in natural populations of birds and mammals exhibits complex associations with landscapes over a range of spatial scales [6][7][8], including the influence of habitat degradation on vectors and disease patterns [9].
Brazil stands out among countries with high biodiversity. Thus, its natural environments are considered a priority for conservation because of their high degree of endemism and advanced degradation [10]. Nine of the ten countries that account for more than 80 percent of the world's primary forest area lose at least 1 percent of their forest area per year; Brazil loses 4 percent each year [11]. Among non-forest environments, the Brazilian Cerrado, which is considered one of the world's biodiversity hotspots [10], is also one of the most endangered ecosystems in Brazil, being rapidly reduced in area due to agricultural expansion and urbanization. Brazil has lost 48.2% of its original Cerrado vegetation. Nowadays, deforestation destroys about 20,000 square kilometers of Cerrado each year. In September 2009, the Action Plan for Prevention and Control of Deforestation in the Cerrado Lands (PPCerrado) was launched to stop this forest devastation by 2011. [12] The impacts of human activities on the Amazon forest and Cerrado biomes are magnified in areas of transition between the two types of environment because of the importance of ecotones for the maintenance of wildlife populations [13]. The role of ecotones between major biomes for the maintenance of parasite populations is, however, poorly understood. Avian populations differ morphologically between the rainforest and ecotone, despite high gene flow; the morphological differences between habitats are as large as those found between related species [14]. In addition, because ecotone habitats may be a source of evolutionary novelty, greater attention should be paid to their conservation in order to preserve the processes that may be important to maintain rainforest diversity [15]. In the present study, we analyze the prevalence and lineage richness of haemosporidian parasites infecting Brazilian wild birds in three environments: natural cerrado habitat, urbanized cerrado habitat, and cerrado habitat that is intermixed with areas of Amazonian forest in a natural transition zone. Comparisons between these locations should provide a first indication of the influences of anthropogenic disturbance and habitat ecotones on the parasites of an assemblage of birds.

Screning microscopy and PCR
We examined samples from 676 wild birds (29 families and 122 species) for haemosporidian parasites. Of these, 49% were infected with Plasmodium spp. or Haemoproteus spp., using both microscopy and PCR as diagnostic methods.
The proportion of infected birds detected by PCR and by microscopy differed significantly (Table 1). Considering the microscopy results as the reference point, 254 PCR-positive samples and 165 microscopy-positive results yielded a sensitivity of 72% (95% CI: 65-79%); 422 PCR-negative and 511 microscopynegative birds yielded a specificity of 74% (95% CI: 69-76%). Some blood parasite infections amplified by PCR were not detected by microscopy (6.6%), and 19% of samples were negative by microscopy and positive by PCR. Thus, the overall agreement (95% CI) of the PCR with the microscopy standard estimated by Youden's index was 0.46 (0.38-0.54).
The prevalence of haemosporidian infections varied among the three areas studied (Table 1). Prevalence in the urban area (disturbed cerrado, UA: 56.2%; n = 108) exceeded that of the intact cerrado (LSP: 42.2%; n = 76; p = 0.007). Prevalence in the intact cerrado and transition area did not differ significantly (CSP: 58.3%) (n = 147; p = 0.2). Parasite prevalence in the urban and transition areas did not differ significantly using both microscopy and PCR as diagnostic methods (p = 0.70). The analysis of a single family of wild birds (Pipridae), which had similar representation in our sample among the three areas, revealed similar prevalences (p = 0.5); haemosporidian prevalence also was not heterogeneous among areas in the sample of the single most abundant species in our sample, Pipra fasciicauda (Pipridae) (p = 0.5). Parasite prevalence in three additional common species (Turdus leucomelas, Coereba flaveola and Volatinia jacarina) did not differ between intact cerrado and the urbanized area (p.0.05).

Parasite lineages and habitats
Sequencing of the cytochrome b gene revealed 11 Plasmodium lineages and 10 Haemoproteus lineages overall. We were unable to determine sequences of 19 individuals that showed multiple infections and of some additional samples with degraded DNA. Of 11 Plasmodium lineages, 6 were restricted to a single study area: TOC-11, TOC-21, TOC-28 and TOC-32 in the urban area; TOC-19 in LSP; and TOC-14 in CSP ( Figure 1). Five of 10 lineages of Haemoproteus were recovered from single areas: TOC-5, TOC-7, TOC-26 and TOC-29 in LSP; and TOC-22 in CSP. The lineages of Plasmodium TOC-4, TOC-9, TOC-16, TOC-24 and TOC-15 and lineages of Haemoproteus TOC-1, TOC-2, TOC-3, TOC13 and TOC-20 were observed in more than one habitat (LSP, CSP or/and UA) ( Figure 1). Six of the lineages found in this study had previously been described in other studies ( Table 2). Fifteen lineages are described for the first time in this study: 8 Plasmodium and 7 Haemoproteus. The widespread lineages were recovered from significantly more host species (6.565.2 sd, range 2-14, n = 6) than the local lineages (1.761.1 sd, range 1-5, n = 15) (Kruskal-Wallis x 2 = 6.8, df = 1, p = 0.01). The Venn diagram in Figure 2 summarizing the distribution of the parasite lineages among the three areas shows that the mixed habitat area (CSP) has the fewest unique lineages.
Parasite diversity was higher in the urban area (78.7% of the combined diversity) than in the intact (61.4%) and transition (56.6%) areas, but the differences were not statistically significant (p.0.05) ( Table 3). Table 4 lists the parasite lineages and host species and families parasitized. The number of host species increased with the number of each parasite lineage recovered ( Figure 3). The proportion of parasite lineages sharing identical sequences in more than one host species occurs more in Haemoproteus than in Plasmodium (Figure 3), however these proportions did not differ significantly (p = 0.68).
In the present study, we confirm the weak correspondence between PCR and microscopy results, probably due to the lowintensity of peripheral parasitemia and variation in diagnostic sensitivity [21][22][23]. In this study, the low levels of parasitemia detected by microscopic examination (ranging from 1 to 5 parasites/200 microscopic fields), with few plasmodia life stages (mainly young trophozoites), also makes morphological identification of the parasite difficult. Certainly, to obtain realistic estimates of parasite prevalence, it is important to combine both microscopy and PCR.
In our study, overall prevalence (Haemoproteus/Plasmodium) was higher in the urban area (urbanized cerrado, UA) and the highest number of unique Plasmodium lineages also was observed among birds sampled in the UA (Figure 1). We observed significantly lower prevalence of haemosporidians in the natural cerrado (LSP) compared to the UA. In contrast, we did not observe a significant difference between the diversity of parasites lineages in the natural cerrado and urbanized cerrado. However, when we analyze a single family (Pipridae) and a single bird species (Pipra fasciicauda) sampled in all three habitats, we observed no statistically significant difference in prevalence or parasite lineage diversity between the habitats. The presence or absence of certain host species is likely the most important factor influencing the presence of parasite lineages and overall parasite prevalence, as shown in several previous studies [24,25]. The similarity in parasite diversity among the three areas might be related to the wide variety of wild birds distributed among these areas. Thus, it is likely that the presence or absence of a parasite lineage depends on the presence of suitable host species rather than particular attributes of the habitat independently of the host. In contrast, Chasar [8] showed that opening forest habitats in Cameroon, Africa, increased both the prevalence and diversity of parasite lineages in a particular host species. However, this type of habitat alteration likely results in more dramatic changes in avian and vector communities than the urbanization of cerrado habitat. The high prevalence of blood parasites found in the transition area (CSP) might be related to periods of flooding, allowing reproduction of vectors and consequently increasing the number of infections. Studies have demonstrated an increased risk of malaria in humans in proximity to bodies of water, which provide abundant mosquito breeding sites [26][27][28]. Future studies of mosquito ecology may reveal the vector-parasite competence relationships in transition areas (Amazonian and Cerrado) to be particularly useful in explaining observed patterns of heterogeneity in avian malaria infection in terms of vector abundance and diversity. We have little knowledge of the vectors that transmit the parasites of these Brazil Cerrado birds. There are reports of Culex quinquefasciatus and Aedes albopictus [29,30], which are known to be vectors of haemosporidians elsewhere, in the north of Brazil, but we do not know if these are vectors of avian malaria in the studied areas.
Spatial variation in parasite prevalence has been demonstrated [31,32], and it is likely that different parasite lineages are associated with particular vector communities. This factor could explain the differences in the geographic distribution of parasite lineages at different sites. In this study, however, we did not  observe significant differences in lineage diversity among the three habitats investigated suggesting that vectors assemblages were sufficient to support similar parasite assemblages independently of the environmental setting. Plasmodium and Haemoproteus are widely distributed blood parasites that have been reported in many families of birds. Some studies have suggested that Haemoproteus parasites are relatively more host specific than Plasmodium [33]. However, in our study, lineages of Haemoproteus were shared among more host species than lineages of Plasmodium, although this difference was not significant. Certainly, parasites of both genera are broadly competent to infect multiple host species, which underlies the phenomenon of frequent host switching over the evolutionary history of haemosporidian parasites [16][17][18][19].
Habitat can influence parasite load in host species, altering the frequency of contact with vectors of blood parasites [34]. However, in this study, although we observed a greater diversity of parasites in the urban area, there was no statistical difference in the diversity of lineages among the three areas studied, including the mixed environment, which might be expected to have a high abundance of vectors following periods of flooding. Some studies at local scales [35,36] have shown that malaria parasites may differ in prevalence and lineage diversity within small areas. However, this contrasts with the widespread distribution of most parasites in this study.
We observed broad variation in the number of hosts from which we recovered individual parasite lineages, ranging from one to 15 avian host species. The lineage with the greatest number of host species (TOC-2 Haemoproteus sp.) is also known from Venezuela and the state of Alabama in the United States [24]. Another lineage with wide host distribution (TOC-15, 10 hosts) is also known from Missouri in the United States and from the Lesser Antilles [24,37]. In general, the six parasite lineages identified in this study that are also known from outside the region exhibited the greatest local abundance and host diversity in this study.
We observed individual lineages of Haemoproteus parasitizing many host species in as many as eight families of birds. Other surveys have detected similar host diversity in some lineages of Haemoproteus [38], indicating that some Haemoproteus lineages can   infect a broad variety of hosts. In contrast to these generalist lineages, we also recovered several individual lineages from single host species. Many of these were present in low abundance in our sample, and so it is difficult to assess host distribution [24]. Additional studies in northern region of Brazil will undoubtedly expand the host ranges of many of the apparent specialist haematozoan lineages.

Study sites
The study was carried out between 2007 and 2009 in three different areas in the Brazilian North region (5u169N, 33u449S) (

Microscopic analysis
Blood samples were collected from the brachial vein on the underside of the wing and two or three smears were prepared using approximately 5 mL of blood. The smears were air-dried, fixed in absolute methanol, and stained for 20 min in 10% Giemsa (Sigma Chemical Co., St. Louis, Missouri, USA), pH 7.4 [39]. The slides were exhaustively examined with a light microscope, and parasite density was quantified after examination of 200 microscopic fields (approximately 150 erythrocytes/field = 30,000 erythrocytes total) at 10006 magnification under oil-immersion [39].

Molecular analysis
DNA extraction. The remaining approximately 20 mL of blood sample was stored at room temperature (22-25uC)    . Thermal cycling conditions were as follows: initial denaturation of 2 min at 94uC followed by 35 cycles with 1 min denaturation at 94uC, 1 min annealing at 62uC, and extension at 72uC for 1 min 10 sec. This was followed by a final extension of 3 min at 72uC. The amplified products were visualized in 6% polyacrylamide gels stained with silver nitrate [39,40]. Microscopically negative samples showing a positive DNA amplification were exhaustively re-examined in order to confirm the presence of parasites.
Cytochrome b amplification. From samples in which we detected positive results (microscopy and/or mitochondrial DNA amplification), we amplified a fragment of 591 bp of the cyt b gene under the following conditions: an outer reaction using primers 3932F (59-GGG TTA TGT ATT ACC TTG GGG TC-39) and DW4R (59-TGT TTG CTT GGG AGC TGT AAT CAT AAT GTG-39) [41] with 1 ml of genomic DNA was subjected to an initial denaturation of 4 min at 94uC, followed by 35 cycles of 94uC for 20 sec, 49uC for 10 sec, and 68uC for 45 sec, and a final extension at 68uC for 3 min. For most samples, 0.5-ml aliquot of this product was used as a template for a nested reaction with primers 413F (59-TCA ACA ATG ACT TTA TTT GG-39) and 926R (59-GGG AGC TGT AAT CAT AAT GTG-39) [25] under initial denaturation of 94uC for 1 min and 28 cycles of 94uC for 20 sec, 52uC for 10 sec, and 68uC for 50 sec and then 68uC for 7 min. PCR products were screened on 1% agarose gels, stained with ethidium bromide, and visualized with a UV light source.
Phylogenetic analysis. Positive PCR products were purified for cycle sequence reactions using ExoSAP-IT (USB Corporation) following manufacturer's instructions. Bi-directional sequencing with dye-terminator fluorescent labeling was performed in an ABI Prism 3100 automated sequencer (Applied Biosystems, Inc.). We sequenced 591 base pairs of the cyt b gene for Plasmodium spp. and Haemoproteus spp.
DNA sequences were aligned using CLUSTALX [42] and edited using Seq ManII version 4 (DNASTAR Inc.) and are available through GenBank (accession numbers HQ287536-HQ287556). Sequences were compared for identification to their closest matches in GenBank using the NCBI nucleotide Blast search, and to unpublished sequences using a local blast search in the laboratory of R. E. Ricklefs. We produced a maximum likelihood phylogenetic tree for the parasite sequences using RaxML [43] with the GTR + gamma model of nucleotide evolution and 100 bootstrap replications.

Statistics
We used the nonparametric estimator Chao1 [44] calculated with scores of 200 randomizations by EstimateS 8.2 [45] to estimate the host species richness and parasite lineage richness in each study site.
Because sample sizes of hosts were unbalanced between areas, we used contingency table analyses to detect interactions between location and parasite prevalence. Statistics were carried out using Prism 5.0 for Windows (GraphPad Software, Inc.). We used Stata 9.07 to construct tables and to calculate the agreement using Youden's index (Y = Sensitivity+Specificity21, which reflects the likelihood of a positive result among truly positive subjects versus that for negative subjects; it ranges from 0 to 1) [46] between molecular and microscopic assessments of parasite prevalence.