In lotic ecosystems, the River Continuum Concept (RCC) provides a framework for understanding changes in environmental factors and free-living communities, yet how parasite populations shift along river continua remains less clear. We quantified infections by a pathogenic trematode parasite (Nanophyetus salmincola) in >14,000 host snails across 130 stream reaches spanning 165 km in the Willamette River Basin in western Oregon, USA. Environmental factors – including flow volume, temperature, benthic algae, canopy cover, woody debris, and land cover – changed predictably with stream order, consistent with the RCC. From first- to eighth-order reaches, infection prevalence decreased by ~42-fold and infected snail density, a measure of disease risk to fish hosts, decreased ~3-fold. Infected snail density, but not prevalence, was positively associated with snail biomass density, and individual infection probability increased strongly with host size. Shifts in snail population characteristics across stream orders, however, did not explain the observed changes in N. salmincola populations, suggesting that environmental variables and corresponding changes in non-snail hosts explain the downstream decrease in infections. Our findings show predictable spatial variation in disease risk to vertebrate hosts from N. salmincola and indicate the RCC can help explain shifts in parasite populations in lotic ecosystems.

Study location – Our survey design included 130 stream reaches each measuring 5 m in length, with 18 to 20 reaches in the mainstem Willamette River and each of six focal sub-basins. Three focal sub-basins drained from the western foothills of the Cascade Range (the Middle Fork Willamette, McKenzie, and Santiam Rivers) and three drained from the eastern foothills of the Coast Range (the Coast Fork Willamette, Mary’s, and Luckiamute Rivers). 

Field surveys – Surveys of snail population characteristics (density, sizes, biomass density) and environmental factors were conducted from 10-Jul to 01-Oct-2019. Snail densities were measured using quadrats spaced at 1-m intervals (0.25 m2; n=5 quadrats per reach) within a maximum depth of ~2 m in each reach. Biomass densities were estimated by measuring all snails (shell length) and using length-to-dry mass conversion equations (see "Snail surveys" for details). For each reach, Strahler stream order was determined using Google Earth Pro satellite imagery, elevation was recorded using GPS, water temperature was measured using a YSI instrument, and % canopy cover was measured using a spherical densiometer. Relative levels of riparian vegetation density, benthic algal abundance, and in-stream woody debris were classified on a numerical scale of 1 to 5 (low-high) based on observations by the first author. We obtained mean annual discharge (m3 s-1), and quantitative measures of channel gradient and confinement (rescaled to values from 1 to 5) using an existing online stream classification system (McManamay and DeRolph 2018). We also obtained measures of percent land cover within a 1-km buffer around each reach using the National Land Cover Database (MRLC 2019). Land cover was grouped into three categories: forested (deciduous, evergreen, and mixed), agricultural (cultivated crops, hay/pasture), and urbanized (developed; high to low intensity).
    
Parasite quantification – We opportunistically collected >100 Juga snails measuring >10 mm in shell length from each reach for dissection. All snails were measured, cracked with pliers, and then examined under a dissecting microscope (8X–35X). N. salmincola was identified based on unique morphological features of the microcercous cercariae, including a bluntly conical tail and presence of a stylet. For immature infections, wet mounts were examined under a compound microscope (4X–100X) to aid identifications. Infections by other trematode species were recorded, categorized by morphological features, and opportunistically vouchered for future molecular identifications. In the current study, we observed at least 12 trematode morphotypes in the dissected Juga snails, but we focus on N. salmincola due to its relatively well understood ecology and its importance in causing disease in species of conservation concern. Information on other trematode taxa in Juga from three of our 130 sites that were characterized with molecular methods is also presented in Preston et al. (2021).
    
Snail surveys – We used randomized transect-quadrat sampling to quantify snail biomass within each reach. Quadrats were spaced at 1-m intervals (0.25 m2; n=5 quadrats per reach) within a maximum depth of ~2 m. The distance from the wetted edge for each quadrat was randomized using a random number generator. If stream current made sampling impractical at a specific location, a maximum safe distance from the wetted edge was used as an upper limit for the random quadrat locations. All snails within each quadrat were hand-collected (using snorkeling in deeper locations), counted, and measured for total shell length (mm). Snail density and size measurements were converted to biomass densities (g m-2) using a length-to-dry biomass conversion equation obtained from measuring, drying and weighing 172 locally collected Juga snails from 3 to 34 mm in shell length (DM = 0.00002*L2.65). All biomass densities in this paper refer to dry tissue biomass excluding the snail shell. 
    
Estimating trematode densities – To estimate infected snail density at each stream reach, we first used reach-specific binomial logistic regressions on N. salmincola infection status from dissection data, with snail size as the only predictor. The reach- and size-specific N. salmincola infection probabilities from the logistic regressions were then applied to each snail recorded in the quadrats to assign an estimated infection status. The densities of infected snails were summed within quadrats to extrapolate the reach-level values of infected snail density.

Please see FALKE_and_Preston_2021_Ecosphere_readme.txt for usage notes, metadata, and additional information, including citations of literature and databases. Additional inquiries about the dataset and usage may be directed via email to Landon Falke.