Diseases are ubiquitous in natural systems, with broad effects across populations, communities, and ecosystems. However, the drivers of many diseases remain poorly understood, particularly in marine environments, inhibiting effective conservation and management measures. We examined biogeographic patterns of infection in the foundational seagrass Zostera marina by the parasitic protist Labyrinthula zosterae, the causative agent of seagrass wasting disease, across >20° of latitude in two ocean basins. We then identified and characterized relationships among wasting disease prevalence and a suite of host traits and environmental variables. Host characteristics and transmission dynamics explained most of the variance in prevalence across our survey, yet the particular host traits underlying these relationships varied between oceans, with host size and nitrogen content important in the Pacific and host size and density most important in the Atlantic. Temperature was also a key predictor of prevalence, particularly in the Pacific Ocean. The strength and shape of the relationships between prevalence and some predictors differed in our large-scale survey vs previous experimental and site-specific work. These results show that both host characteristics and environment influence host-parasite interactions, and that some such effects scale up predictably, whereas others appear to depend on regional or local context.

Study sites: In the summer of 2015, we surveyed 17 discrete, monospecific eelgrass beds across the northern hemisphere, mostly along the coasts of North America (9 in the Atlantic Ocean, 8 in the Pacific Ocean). All of the surveyed sites were in protected, shallow waters (0 - 2 m depth at low tide).

Wasting disease survey: At each site, we collected a total of 20 individual vegetative eelgrass shoots at least two meters apart along two 10 meter transects running parallel to shore. We clipped shoots at the base in the field and placed them in individual plastic bags. We timed sampling to target peak wasting disease prevalence based on local knowledge of system dynamics, which resulted in most sites being surveyed in mid-summer near peak eelgrass biomass.

Within five hours of collection in the field, we scored the third youngest leaf of each shoot for signs of wasting disease (i.e., lesions). We focused on this leaf because it has been shown to harbor the highest intensity of wasting disease (Bockelmann et al. 2013) and remains sufficiently free of epibionts, making lesions clearly distinguishable. At each site, we used a standardized protocol including published descriptions and photographs of wasting disease to guide scoring (Burdick et al. 1993, Groner et al. 2014, Groner et al. 2016). Leaf lesions typically indicate intense wasting disease infections, and thus we interpret lesion distribution as a reliable reflection of L. zosterae distribution (Dawkins et al. 2018, Brakel et al. 2019). However, leaf lesions could also reflect processes other than parasite abundance. For example, though black and brown lesions are established visual signs of wasting disease, other stressors can also cause lesions, and low levels of L. zosterae parasite may be present in asymptomatic leaves (Bockelmann et al. 2013, Groner et al. 2016). Because we had local experts familiar with the signs of other stressors specific to each site (e.g., UV damage or grazing), we minimized mis-classification between lesions associated with disease and lesions caused by other stressors. We also collected notes and representative photographs of putative wasting disease lesions as necessary to ensure consistency of scoring.

Host and environmental surveys: In separate surveys at these same sites in 2014, we collected data on host and environmental factors that could mediate wasting disease prevalence. These data were collected from 20, 1-m² plots spaced roughly 2 m apart and included eelgrass density (measured as shoots per m²), blade length (measured from the meristem to the tip of the longest leaf), leaf N content (from young leaf material of 5 pooled shoots in each plot run on a CHN analyser), allelic richness (calculated from 5 shoots per plot at 24 DNA microsatellite loci), periphyton (grams of total dried algae, microbes, and detritus present on the leaf surface per grams of dried eelgrass), and mesograzer abundance (number of mesograzers per grams of eelgrass) following methods described in greater detail in Duffy et al. (2015, 2022). Many of these variables are generally consistent at a given site from year to year when sampled in the same season (Douglass et al. 2010, Bertelli et al. 2021), and sometimes consistent across many years (Reynolds et al. 2017). In addition, previous studies have demonstrated that variables measured months prior can be as or more predictive of wasting disease prevalence than those measured coincident with sampling of wasting disease (Groner et al. 2021, Graham et al. 2023). However, we may be underestimating the predictive power of host and environmental factors on wasting disease by sampling them in consecutive years rather than the same year.

To quantify temperature and salinity at each site, we extracted estimates of the maximum mean monthly sea surface temperature and minimum mean monthly salinity from the Bio-ORACLE data set for 2000-2014 (Tyberghein et al. 2012, Assis et al. 2018; 9.6-km² resolution). We chose maximum temperature to capture summer conditions when past epidemics have been observed (Rasmussen 1977, Burdick et al. 1993) and minimum salinity because lower salinities inhibit the parasite (McKone & Tanner 2009). We used the raster package in R (Hijmans 2019) to extract these data from all cells within 10 km of each site, and we averaged these estimates to generate site-level predictors. As a potential complement to these long-term data, we also measured in situ water temperature and salinity simultaneous with the wasting disease survey.