Infectious wildlife diseases are one of the main drivers of biodiversity loss (Jones et al., 2008). At global scales, infectious diseases have been documented to cause massive declines in populations up to the depletion of native fauna communities (Daszak et al., 2000; Robinson et al., 2010; Woinarski et al., 2011). Climate drives species’ richness distribution, via niche-mediated processes, and ultimately affects the distribution of potential hosts and that of the pathogen itself (Rohr et al., 2011). In this sense, abiotic and biotic components interact synergistically to determine the set of conditions that allow pathogen survival, such as favourable climates and the existence of suitable hosts (Fisher and Garner, 2020). However, human activity also plays an important role in the emergence of infectious diseases by modifying natural environments and creating new opportunities for parasites and pathogens to evolve (Fisher et al., 2012). Diseases may, then, interact with anthropogenic drivers to disturb biological communities and contribute to the species extinctions (Wake and Vredenburg, 2009).

Amphibians are central in this discussion since they comprise the vertebrate group that is with extinction and have been decimated by infectious diseases in the last decades (Wake and Vredenburg, 2009). One specific fungus genus (Batrachochytrium) is particularly related to worldwide amphibian population declines through infection and disease (Scheele et al., 2019). Chytridiomycosis is caused by two congeneric fungal species: Batrachochytrium dendrobatidis (Bd), described in 1999 (Longcore et al., 1999), and B. salamandrivorans (Bsal), described in 2013 (Martel et al., 2014). The host range of Bsal is mostly restricted to salamanders and newts, with only transient infection in anurans (Stegen et al., 2017), while Bd is a parasitic chytrid fungus that can infect equally anurans, caecilians (Gymnophiona), salamanders and newts (Caudata) (Lambertini et al., 2017; Fisher and Garner, 2020). Spread across the globe, also with the aid of the global trade of amphibians, Bd is now considered one of the greatest threats to native amphibian across the globe (Scheele et al., 2019; Fisher and Garner, 2020).

The macroecological drivers of Bd infection on amphibians are still not well-established, but variation in extrinsic (e.g., climate and landscape configuration) and intrinsic (such as species-specific vulnerability as well as host community structure) are known to influence the dynamics of Bd infection within a community (James et al., 2015). Temperature and precipitation, for example, mediate Bd–amphibian dynamics by limiting vital physiological processes, such as pathogen growth and host evaporative water loss (Raffel et al., 2010). Amongst biotic factors, the distribution of Bd is generally related to features within amphibian host communities, such as species richness and trait composition (Ostfeld and Keesing, 2012, but see Becker et al., 2014; Liu et al., 2013; Searle et al., 2011). There are studies that point to a scenario in which that host species richness increases the pathogen’s occurrence, called the “amplification effect” (Becker et al., 2012); while others argue that species richness decreases pathogen’s occurrence, the “dilution effect” (Searle et al., 2011). However, the global trade of amphibians has led to the co-introduction of Bd fungus into pristine native faunas (O’Hanlon et al., 2018; Fisher and Garner, 2020), across gradients of integrity in human-dominated landscapes (Becker et al., 2012; Beyer et al., 2015), allowing new opportunities for Bd infection and evolution (Fisher et al., 2012).

Moreover, the controversy on assessments of the drivers of infectious diseases, including the infection by Bd fungus, may be an artifact of the scale-dependency of patterns and processes (Ricklefs, 1987; Wiens, 1989). That is so because apparent relationships (e.g. the relative importance of predictors and/or the direction of the relationship) may change according to the spatial scale of the study (Cohen et al., 2016; Halliday and Rohr, 2018; McGill, 2010). For example, biodiversity–disease relationships are expected to be stronger at local scales (Halliday and Rohr, 2018), where biotic interactions occur, and should weaken at regional larger scales, where abiotic factors, such as climate, are expected to predominate (Ricklefs, 1987). Therefore, scale-dependency can become a confounding effect on multiple systems, including the assessment of the drivers of pathogen occurrence and biological invasions (Catford et al., 2022).

In this study, we explored the influence of spatial scale on the importance of biotic, abiotic and anthropogenic influences, using proxies of such components Bd occurrence. We hypothesize that 1) biotic factors should have the highest relative importance at local scales, whereas 2) abiotic climatic variables should have the highest relative importance at regional-continental scales, and that 3) anthropogenic influence should be stronger at mid-range scales, because of the expression of land-use change dynamics at landscape spatial scale (Cohen et al. 2016).

Human footprint was found to be stronger in the state of São Paulo and other metropolitan areas, main cities in the countryside and coastal regions across the Atlantic Forest (Fig. 2A). Amphibian species richness varies throughout the Atlantic Forest (Fig. 2C). It is higher on the coastal regions, mainly at latitudes between 20 and 25 °South (∼80 to 130 species). Minimum monthly evapotranspiration is primarily higher on the countryside and Semideciduous portion of the Atlantic Forest, whereas it is lower on the coastal regions where Ombrophilous Forest remains (Fig. 2B).

Fig. 2.

Predictor variables given in 0.0186 degree of spatial resolution: human footprint (A); minimum monthly evapotranspiration (mm/month) (B); and amphibian richness (species count per pixel) (C).

(0.34MB).

Human footprint was more important to predict Bd prevalence than the other two predictors for both smaller (104  km2) and larger scales (>106  km2) used in this study (Fig. 3). However, its relative importance (e.g., average weight) decreased and remained similar to that of the other predictors in intermediate scales (∼105  km2) and then increased again. At intermediate scales, minimum monthly evapotranspiration increased in importance and showed higher relative importance than the two others. At the same scale, relative importance of amphibian richness also increased, but remained lower than the other two variables. MMPE had the highest contribution to explaining Bd prevalence in intermediate scales and was slightly more important than amphibian richness for all scales, however the weights for all three predictors were not significantly different at that scale (Fig. 3, Table 1). Throughout the Atlantic Forest, amphibian species richness was the factor that had the least important contribution to Bd prevalence prediction. In intermediate sized regions, there was an increase in importance of amphibian richness in explaining Bd prevalence, but it remains less important than the others.

Fig. 3.

Average weights of each variable for each location, by area: amphibian richness (green), human footprint (red) and minimum monthly evapotranspiration (blue), for each spatial scale. Each dot accounts for one variable at a certain data point. Each datapoint has three components: average weight for each variable for each location. Areas are converted to log10, so 105 km² is equivalent to 5 in this graph. Values for each location can be found in Table 1.

(0.12MB).

Infectious diseases act synergistically with other factors to decimate native fauna populations worldwide, yet the main drivers of disease occurrence and prevalence remain poorly known (Heard et al., 2013; Stephens et al., 2016). Humans have profoundly altered the Atlantic Forest and dramatically reduced its pristine areas, mainly by habitat alteration, fragmentation and degradation (Marques and Grelle, 2021). In this study, we tested for the relative influence of biotic (host richness), abiotic (minimum monthly potential evapotranspiration) and anthropogenic (Human Footprint Index) factors in predicting the prevalence of the chytrid fungus on amphibian communities across a range of environmental gradients in the Atlantic Forest, while controlling for autocorrelation and scale-dependency.

Surprisingly, we found that human footprint index had the highest relative importance to explain Bd prevalence in most spatial scales, where Bd and HFI were positively correlated in different ecoregions and subregions of the Atlantic Forest of South America. Compared to HFI, host richness and minimum monthly potential evapotranspiration had smaller relative importance on explaining Bd distribution at most spatial scales. Our findings also suggest the notion that anthropogenic-driven modulation of Bd infection is scale-dependent, once the effect of HFI was indistinguishable from other drivers at intermediate scales. This may be a suggestion for future studies to explore the effect of anthropogenic threats in intermediary scales, since not much evidence is available on the scale at which anthropogenic threats affect biodiversity (Cohen et al., 2016).

Despite of our hypothesis that MMPE would have highest influence in large scales, we found MMPE to be less important than HFI in most of the studied spatial scales, in contrast to what has been found for the Americas (James et al., 2015). However, there was an expressive increase on MMPE relative importance at intermediate scales of analysis, and sudden decrease in importance in larger scales, the same pattern observed by Cohen et al. (2016). Those are interesting and contradict our hypothesis of climate being the main driver in large scales. They might suggest that in a highly altered biome such as the Atlantic Forest, human alterations might surpass climate variables in importance to explain pathogen occurrence.

Unfortunately, as it is common for tropical areas, we did not have enough prevalence records in the smallest ecoregions analysed and therefore we could not consider scales smaller than 104 km2. Thus, given the scale of analyses, it would be unlikely to detect the effect of biotic factors, even if they were highly correlated with chytrid prevalence, since species interactions are expected to be important in local scales (Willis and Whittaker, 2002). Further sampling effort in less studied regions of the Atlantic Forest such as the Northeast of Brazil can be of great value to fill the knowledge gap on the distribution of Bd in the region.

However, one key finding was that in larger scales HFI was more important to explain occurrence of Bd. We hypothesize two possible and non-exclusive mechanisms for the influence of HFI on Bd prevalence: 1) human altered habitats contain higher Bd prevalence because humans transport it to new areas through contaminated water and individuals species; and 2) human altered habitats affect amphibians’ immunological response, causing them to be unable to cope with the infection and spreading it. Humans can influence Bd prevalence via direct transportation and increasing propagule pressure in invaded regions (Fisher and Garner, 2020). In fact, the evidence therefore suggests that the global expansion of the distribution of the fungus happened through transcontinental commercial trade, highways and through the introduction of exotic amphibians in new sites when they are exported for food, research or collections (Fisher and Garner, 2020; Liu et al., 2013; O’Hanlon et al., 2018; Schloegel et al., 2012), which corroborates the anthropogenic mediation of its recent introduction elsewhere (Lips et al., 2003; Rödder et al., 2009).

On the other hand, human altered habitats can affect amphibians’ immunological response, causing them to be unable to cope with the infection and spreading it (Davidson et al., 2007). For example, the presence of pesticides in aquatic environments is related to immunosuppression and increased parasitism following pesticide exposure in amphibians (Christin et al., 2004; Rohr and Palmer, 2013). Moreover, human footprint has been associated with higher invasibility (Beans et al., 2012), suggesting that there may be biological mechanisms that facilitate the spread of invasive species in human-dominated landscapes. We argue that the two processes may act together, as transportation can bring the pathogen and human alterations in the environment can make hosts more susceptible to infection, both leading to a perceived increase in importance for HFI in explaining pathogen occurrence. This might be exacerbated in highly modified environments such as the Atlantic Forest, especially in south-eastern Brazil, where most of our data collections points are, there are many bullfrog farms (Ribeiro and Toledo, 2022) which are likely to impact susceptible native species through the release of Bd zoospores into the environment (Ribeiro et al., 2019).

Our findings challenge the expected effect of factors influencing species distribution, because climatic factors supposedly should prevail at larger spatial scales (Ricklefs, 1987; Willis and Whittaker, 2002), even though the scale in which anthropogenic effects influences biodiversity has not been widely tested (Cohen et al., 2016). After testing that, we found that in this study anthropogenic features of landscapes were more important than an important climate variable itself to explain the prevalence of the chytrid fungus across a wide range of landscapes from a biodiversity hotspot. The results found here, therefore, add to the mounting evidence that, at least for this particular fungus, anthropogenic factors can be even more important than classically analysed factors concerning the distribution of organisms such as biotic and climatic variables. It is, thus, urgent that the effects of human pressures are evaluated in time and space so we can mitigate the environmental damage posed to native faunas worldwide (Steffen et al., 2015).

JAS, LFT and LPS conceived and designed the study; LFT provided chytrid data; JAS and LPS gathered and compiled the data; JAS and LPS performed analysis; JAS, LFT and LPS drafted the manuscript. All authors provided input, approved the final version of this manuscript, and agree to be accountable for all aspects of the work.

Authors declare no competing interests.