Serum samples collected from 547 equids in the Pantanal region of Brazil were evaluated for antibodies to Equine Infectious Anemia Virus (EIAV) by the agar gel immunodiffusion test. Risk factors associated with EIAV seropositivity were evaluated and spatial dependence investigated using a Spatial Lag Model. EIAV prevalence on farms in the Pantanal was 52.0% (13/25) with adjusted prevalence between equids of 31.5% (17.4-48.8% 95% CI). Intra-herd prevalence ranged from 5.0 to 77.0%. Statistical analysis demonstrated that farms and animals in regularly flooded areas had respectively 60 and 146 fold higher chance to be sero-positive than farms and animals located in non-flooded areas. Spatial Lag Model results were generally consistent with this conclusion although there was a negative spatial correlation between farms located within in regularly inundated regions, suggesting that other factors, such as management practices, probably play a significant role in transmission of EIAV. Equids with clinical signs were 3.74-fold more likely to be sero-positive than those without clinical signs. The results of this work reveal a high prevalence of EIAV in the Pantanal area of Brazil demonstrating that equids reared in this region are at great risk of infection.

Summary 1. The objective of this study was to identify longitudinal distribution patterns of large migratory fish species in the Uruguay River basin, southern Brazil, and construct statistical distribution models for Salminus brasiliensis, Prochilodus lineatus, Leporinus obtusidens and Pseudoplatystoma corruscans. 2. The sampling programme resulted in 202 interviews with old residents and fishermen at 167 different sampling sites in the major rivers of the basin. Interviewees were asked whether these fish species were present in the nearby river segment, recently or in the past. The average age of respondents was 56 years, and the great majority had resided for more than 30 years in the area where they were interviewed. 3. Fish distribution patterns were modelled using altitude and basin area as environmental descriptors. Distributional patterns were estimated using logistic equations (LOGIT): , where Al is the altitude and BA is the basin area. 4. Accuracies of the distribution models were between 77 and 85%. These accuracies are similar to those of published distribution models for other fishes based on larger numbers of environmental descriptors. 5. The historical distributional ranges make clear that all operating or planned large hydropower dams in the Uruguay River basin, with one exception, limit or will limit the distribution of migratory fishes.