Introduction
Human-facilitated pathogen dispersal (often termed pathogen pollution) has led to an increase in emerging infectious diseases over the last few decades (Cunningham et al. 2017) and is a major threat to global biodiversity and food security. Introduced pathogens can cause catastrophic animal (e.g. crayfish plague, avian malaria) and plant (e.g. chestnut blight, citrus greening) declines and economic losses (Daszak et al. 2000; Anderson et al. 2004). Pathogen pollution is also a primary threat to public health, as it has contributed to several recent deadly epidemics and pandemics (e.g. SARS CoV2) (Daszak et al. 2000; Morens et al. 2020).
Understanding how hosts respond to novel pathogens is necessary to predict and prepare for outbreaks of emerging pathogens, which are on the rise globally (Cunningham et al. 2017). However, there is a set of seemingly contradictory hypotheses on novel pathogens in the literature. The naïve host syndrome hypothesis asserts that hosts are especially vulnerable to novel pathogens because hosts have lower evolved immunological defenses against these pathogens, resulting in high host mortality and host population suppression (Carey et al.1999; Anderson et al. 2004; Taraschewski 2006; McKenzie & Peterson 2012; Lymbery et al. 2014). In contrast, there is also a very rich literature indicating that pathogens are generally better adapted to infect and replicate in local rather than novel host species (Lively & Jokela 1996; Gandon & Van Zandt 1998; Kaltz & Shykoff 1998; Lively & Dybdahl 2000; Torchin et al. 2003; Torchin & Mitchell 2004; Morran et al. 2011; Strauss et al. 2012; Lymberyet al. 2014; Parker et al. 2015; Bolnick & Stutz 2017; Johnson et al. 2021). Hence, the naïve host syndrome suggests that pathogens are able to invade novel hosts because of a lack of co-evolutionary history, whereas local adaptation suggests that pathogens are better able to invade local hosts because of their co-evolutionary history (Lively & Jokela 1996; Gandon & Van Zandt 1998; Kaltz & Shykoff 1998; Lively & Dybdahl 2000; Torchin et al. 2003; Torchin & Mitchell 2004; Morran et al. 2011; Strausset al. 2012; Lymbery et al. 2014; Parker et al.2015; Bolnick & Stutz 2017; Johnson et al. 2021) (Figure 1). Rarely do local adaptation and naïve host syndrome studies cite one another or acknowledge their ostensibly mixed messages.
Here, we hypothesize that pathogens are generally better adapted to infect and replicate in local hosts, resulting in deadlier host-pathogen outcomes. However, we also hypothesize that enough variance exists in novel host-pathogen outcomes to pose substantial risk that an especially virulent host-pathogen combination will occur given sufficient pathogen introduction events (Torchin & Mitchell 2004; Lloyd-Smith et al.2005; Reeder et al. 2012; Cohen et al. 2018; Golaset al. 2021). Thus, we postulate that pathogen pollution is dangerous because, as pathogen introductions occur with increasing frequency, the probability increases that (i ) a particularly deadly strain of a pathogen will devastate a naïve host population, (ii ) a particularly vulnerable host population will be exposed to a new pathogen, and (iii ) especially virulent host-pathogen combinations will occur (Fig. 1). Researchers might be biased towards the assumption that novel pathogens are often devastating because they predominantly only observe pathogen introductions that establish and are problematic, even though most introductions might fail because of a lack of co-evolutionary history (Torchin et al. 2003; Torchin & Mitchell 2004).
Amphibian-Batrachochytrium dendrobatidis (Bd) interactions are ideal to address these hypotheses for several reasons. Bd spread globally in the early 20th century, possibly with the expansion of trade (O’Hanlon et al. 2018), and thus is an invasive pathogen in much of its range with host-parasite outcomes that vary greatly in virulence. Bd represents one of the most urgent ecological disasters on the planet as it is implicated in the declines and extinctions of over 500 amphibian species around the globe (Scheeleet al. 2019b) and can even adversely affect co-occurring non-amphibian species (Brannelly et al. 2012; McMahon et al. 2013; Nordheim et al. 2021). Further, Bd is considered endemic in many regions (Venesky et al. 2014), with some evidence that this endemicity is driven by amphibian hosts adapting to local Bd strains (Voyles et al. 2018; Waddle et al. 2019; Fisher & Garner 2020; McDonald et al. 2020). While many host-pathogen systems exhibit a trade-off between virulence and transmission, field and laboratory evidence suggest transmission is not significantly limited by virulence, at least in part due to non-amphibian reservoir hosts (Fisher et al. 2012; McMahon et al. 2013). Further, high Bd loads lead to greater host mortality, suggesting that mortality is likely a suitable measure for pathogen performance in this system (Greischar & Koskella 2007; Fisher et al. 2012; Fu & Waldman 2019; Scheele et al. 2019b).
To test the hypotheses described above, we identified six populations of toads from across North America (Arizona, California, Louisiana, Ohio, Tennessee, and Quebec, Canada; Table S1) and, in a “common-garden” experiment, measured host mortality and infection prevalence and abundance when the toads were exposed to their local strain of a chytrid fungus, Batrachochytrium dendrobatidis (Bd), five non-local strains, and a sham control (Table S2). We define a novel strain as one that is not from the same host population. To complement the common-garden experiment, we assembled a host mortality dataset of 84 experiments from 26 Bd studies that included 23 amphibian species, 22 unique Bd strains, and wide variability in local and novel host-parasite interactions. Using this dataset, we conducted a global-scale meta-analysis to test for evidence of local adaptation, particularly susceptible host populations, and especially deadly Bd strains and host-Bd strain combinations.