Estimating prevalence of avian haemosporidians in natural populations : A comparative study on screening protocols

Background: Birds harbour an astonishing diversity of haemosporidian parasites. Renewed interest in avian haemosporidians as a model system has placed a greater emphasis on the development of screening protocols to estimate parasite prevalence and diversity. Prevalence estimates are often based on the molecular or blood-smear microscopy techniques. However, variation in diagnostic sensitivity among screening methodologies represents a potential source of bias that may lead to erroneous inference in comparisons of prevalence across studies. Here, we analyzed a suite of blood samples for the presence of parasites using four diagnostic tools and compared method-specific estimates of detection probability to assess the relative performance of screening strategies. Methods: We screened a total of 394 bird blood samples collected in India (n = 203) and Sweden (n = 191) for the combined presence of Plasmodium, Haemoproteus and Leucocytozoon with three PCR assays: (i) qPCR; (ii) restriction enzyme-based assay; and (iii) nested protocol. In addition, we examined blood smears for estimates of parasite intensity which was further screened using qPCR method to evaluate if parasite intensity shows a relationship with qPCR (Ct values). Furthermore, we used single infected samples from parasite intensities: low, medium, high, very high to establish the reproducibility in qPCR. Results: For the combined data sets from India and Sweden, detection probability for submicroscopic and low intensity infections was highest for the qPCR method, followed by the nested protocol and the restriction enzyme-based assay. For high parasite intensities, the qPCR had high PCR reproducibility, with three out of three PCR replicates being positive and with consistent Ct values across all tenfold dilution series. For parasite intensities at very low and submicroscopic samples, the qPCR was reproducible in one out of the three replicates. The intensity of parasitemia estimated from smears showed inverse relationship with Ct values in both the Indian and Swedish data sets. Conclusions: Our study highlights the importance of accounting for methodological issues to better estimate infection in parasitological studies and illustrates how a wider deployment of diagnostic tools combined with statistical approaches is needed for each study, in order to provide adequate insight into the most appropriate approach to avoid erroneous inferences