1. Understanding how ecological communities are assembled remains a key goal of ecosystem ecology. Because communities are hierarchical, factors acting at multiple scales can contribute to patterns of community structure. Parasites provide a natural system to explore this idea, as they exist as discrete communities within host individuals, which are themselves part of a community and metacommunity.

2. We aimed to understand the relative contribution of multi-scale drivers in parasite community assembly and assess how patterns at one level may mask those occurring at another. Specifically, we wanted to disentangle patterns caused by passive sampling from those determined by ecological drivers, and how these vary with scale.

3. We applied a Markov Random Fields model and assessed measures of β-diversity and nestedness for 420 replicate parasite infracommunities (parasite assemblages in host individuals) across two freshwater mussel host species, three sites and two time periods, comparing our results to simulations from four different ecologically relevant null models.

4. We showed that β-diversity between sites (explaining 25% of variation in parasite distribution) and host species (41%) is greater than expected, and β-diversity between individual hosts is smaller than expected, even after accounting for parasite prevalence and characteristics of host individuals. Further, parasite communities were significantly less nested than expected once parasite prevalence and host characteristics were both accounted for, but more nested than expected otherwise, suggesting a degree of modularity at the within-host level that is masked if underlying host and parasite characteristics are not taken into account. The Markov Random Fields model provided evidence for possible competitive within-host parasite interactions, providing a mechanism for the observed infracommunity modularity.

5. An integrative approach that examines factors at multiple scales is necessary to understand the composition of ecological communities. Further, patterns at one level can alter the interpretation of ecologically important drivers at another if variation at higher scales is not accounted for. 

We collected mussels from three sites in Cambridgeshire, UK: Brandon Creek (henceforth BC), King’s Dyke (KD) and the Old West River at Stretham (OW), all of which are part of the Great Ouse river system (Fig. 1). Sampling incorporated two species: the duck mussel Anodonta anatina (Linnaeus 1758) and the painter’s mussel Unio pictorum (Linnaeus 1758), both non-endangered unionid bivalves common throughout Europe (Lopes-Lima et al. 2017) that possess a broad range of parasites (Brian & Aldridge 2019). We sampled on two occasions: 7th May 2019 (“Visit 1”), and 7th November 2019 (“Visit 2”).

 At all sites, mussels were sampled by hand from the river margin. In total across both visits, we collected 420 mussels (240 A. anatina, 180 U. pictorum). Because extended storage of live mussels in the laboratory could cause parasites to leave the host, or move between hosts, we immediately placed mussels in 75% ethanol upon sampling before transporting them back to the laboratory. The two species were not mixed at any point once being removed from the river. Exploratory dissection on both ethanol-stored and live mussels showed that the storage of mussels in ethanol prior to analysis did not affect the detection of any parasites in the study, and it has previously been shown as an effective way of sampling mussel parasites (Conn et al. 2008). As we only sampled non-threatened non-cephalised invertebrates, no ethical approval was required.

In the laboratory, we sliced the anterior and posterior adductor muscles to open the mussel. We inspected all parts of the mussel in systematic fashion, and identified all parasites to the finest possible taxonomic resolution using keys (we did not use molecular identification; see S1: Supplementary Methods). We inspected samples of mantle fluid (1 mL) under 40 magnification using a GXM-L3200 compound microscope to identify the presence of ciliates and nematodes. The mantle, gills and pericardial cavity of the mussel were inspected under a GXMMZS0745-T stereomicroscope at 16 magnification to identify further ciliates, mites, chironomids, bitterling (Rhodeus amarus) embryos and aspidogastrean trematodes. Finally, we squashed samples of gonad tissue between two glass microscope slides and studied them at 40 magnification (following Brian & Aldridge 2020) to identify digenean trematodes. For ciliates and digenean trematodes we only noted presence or absence, while for mites, chironomids, bitterling embryos, aspidogastrean trematodes, and nematodes we also counted the numbers of individuals. If the same parasite appeared in multiple life-history stages or in multiple host tissues, we treated them separately (following Brian & Aldridge 2021a). We measured the maximum length of all mussels (nearest 0.5 mm) with Vernier callipers, dried them to constant mass (nearest 0.001 g), and identified them as either male, non-gravid female or gravid female via inspection of their gill tissue (where gravidity refers to the phenomenon of female unionid mussels harbouring larval mussels in specialised water-tubes in their gills).

The files are provided as either .csv files (data files) or as .R files (accompaning code). Please see the 'README_file.docx' for full usage notes and a complete description of all files.