Introduction

Until recently, it was assumed that the pathogenic fungus Batrachochytrium dendrobatidis (Bd) was not widely distributed in warm ecosystems such as lowland tropical rainforests because maximum environmental temperatures limit its growth (Ron, 2005; Rödder et al., 2009). This assumption received general support because laboratory studies showed that the optimal temperature for growth of Bd in the lab is 15°C to 25°C, whereas temperatures above 28°C resulted in spore death (Johnson, Berger, Phillips, & Speare, 2003; Piotrowski, Annis, & Longcore, 2004; Roznik, Sapsford, Pike, Schwarzkopf, & Alford, 2015; Stevenson et al., 2013; Stevenson, Roznik, Alford, & Pike, 2014). However, several studies have documented Bd infection in lowland rainforest amphibians over the past decade (Flechas et al., 2012; Kosch, Morales, & Summers, 2012; McCracken, Gaertner, Forstner, & Hahn, 2009; Whitfield, Kerby, Gentry, & Donnelly, 2012). These findings lent support to the idea that some lowland tropical forest habitats offer suitable environmental conditions for Bd growth, even though most lowland areas may contain suboptimal conditions limiting the pathogen spread and growth (Flechas, Vredenburg, & Amézquita, 2015). A recent study focusing on museum-stored specimens of species in the genus Leptodactylus (Anura: Leptodactylidae) showed that Bd has been present in the lowland Amazon for more than 80 years (Becker, Rodriguez, Lambertini, Toledo, & Haddad, 2016). In contrast, studies using species distribution models suggested that lowland Amazonian rainforest might not be suitable for Bd because high environmental temperatures prevent its growth (James et al., 2015).

Here, we surveyed four sites in southeast Peru to examine the prevalence and the intensity of infection of Bd in lowland Amazonian amphibians and to fill a gap between two areas where Bd has been present for more than a decade, namely, the montane forests in Manu National Park, Cusco (Catenazzi, Lehr, Rodriguez, & Vredenburg, 2011; Catenazzi, Lehr, & von May, 2013) and the lowland forests in Acre, Brazil (Becker et al., 2016). In one of these “hotspots” of Bd infection, the upper slopes of Manu National Park, more than 20 montane amphibian species have experienced population declines attributed to Bd epizootics over the past 15 years (Catenazzi et al., 2011; Catenazzi, Lehr, & Vredenburg, 2014). We also examined the thermal profile of the main microhabitats used by lowland Amazonian frogs to infer whether these microhabitats offer suitable thermal conditions for Bd growth.

Methods

We conducted this study at four lowland sites in the Madre de Dios region of southeastern Peru (Figure 1): Los Amigos Biological Station (CICRA is the Spanish acronym), 12°34′07″ S, 70°05′57″ W, 270 m elevation; Centro de Monitoreo 1, 12°34′17″ S, 70°04′29″ W, approximately 250 m elevation; Centro de Monitoreo 2, 12°26′57″ S, 70°15′06″ W, 260 m elevation; and Tambopata Research Center, 13°08′30″ S, 69°36′24″ W, 350 m elevation. The first three sites are relatively close to each other (3.5–25 km) and the fourth site (Tambopata Research Center) is 80 to 105 km away from the other three sites. A general overview of the amphibian fauna, the habitats, and the local climate at these sites was provided by von May et al. (2009), von May et al. (2010) and von May, Jacobs, Santa-Cruz et al. (2010).

Figure 1.

Map showing the location of the study sites (red triangles) and the main natural protected areas in southeast Peru. The inset shows the location of the study area in South America (orange polygon), and Peru highlighted in yellow. Map by Rudolf von May made with Quantum GIS ( https://qgis.org/en/site/). CICRA = Los Amigos Biological Station; CM1 = Centro de Monitoreo 1; CM2 = Centro de Monitoreo 2.

We took noninvasive skin swab samples from 282 individuals encountered in the field during the wet seasons of 2008 (N = 138), 2012 (N = 55), and 2014 (N = 89). Swabs collected in 2008 were analyzed using laboratory protocols described in Kosch et al. (2012). In 2012 and 2014, we used an MW113-Advantage Bundling sterile synthetic cotton swabs to sample the abdomen, thighs, and hind limbs from each specimen and stored dried in 1.5 ml tubes. We quantified Bd infection by quantitative polymerase chain reaction following the procedures of Boyle, Boyle, Olsen, Morgan, and Hyatt (2004) and Hyatt et al. (2007). We captured frogs by hand and stored them in separate plastic bags in the field. Occurrence of Bd was determined by estimating the quantity of zoospores on each animal (Zswab, zoospore equivalents; Briggs, Knapp, & Vredenburg, 2010; Vredenburg, Knapp, Tunstall, & Briggs, 2010), which in turn is calculated by multiplying the genomic equivalent obtained during quantitative polymerase chain reaction by the dilution factor (genomic equivalents × dilution factor = zoospore equivalents). We calibrated sample results with a qPCR standard curve. Swabs were categorized as Bd-positive when the equivalent of zoospores was >0 and Bd-negative when the equivalent of zoospores was =0 (Vredenburg et al., 2010). Bd prevalence was calculated by dividing the number of infected amphibians by the total number of individual swab samples. We considered two categories for the intensity of infection: low infection when Zswab was between 1 and 10,000 and high infection when Zswab was greater than 10,000 (Voyles et al., 2012). These threshold values are equivalent to those estimated as critical in the mortality (10,000 zoospores) of frog species reported in the previous studies (Vredenburg et al., 2010).

We used data loggers to obtain empirical data on microhabitat temperatures from lowland Amazonian habitats. First, we used iButton data loggers (Maxim Integrated Products, Sunnyvale, CA) placed in two forest microhabitats, leaf-litter (3 cm above the ground) and understory vegetation (180 cm above the ground), used by frogs across four forest types. Daily temperatures were recorded in floodplain forest, terra firme forest, bamboo forest, and palm swamp during part of the wet season of 2008 (44 days). In addition, we used HOBO data loggers (Onset Computer Corporation, Pocasset, MA) placed in the leaf litter of the floodplain and the terra firme forests to record the temperature from November 16, 2016 to December 5, 2017 (385 days). We compared these data with air temperatures measured at the local weather station over the same time period.

Results

We obtained Bd prevalence and infection intensity data from 51 species sampled in 2008, 27 species sampled in 2012, and 29 species sampled in 2014 (Appendix A). Nine species of frogs were infected by Bd (Table 1), and the prevalence of infection varied across years: Bd prevalence was 0.7% in 2008, 7.3% in 2012, and 4.5% in 2014. The infected species belong to five families that are found in different forest types and use different microhabitats (Table 1). Most infected species use aquatic microhabitats for breeding and larval development, as their tadpoles use either lotic or lentic bodies of water. These included a poison frog (Dendrobatidae), two tree frogs (Hylidae), one leptodactylid frog (Leptodactylidae), and two narrow-mouthed frogs (Microhylidae). In addition, one infected species (Adenomera andreae) lays foam nests on land and larvae complete their development inside the foam, while two species belong to the diverse terrestrial breeding frogs (Strabomantidae) and have direct development (i.e., there are no free-living tadpoles).

Table 1.

Species Infected by Batrachochytrium dendrobatidis (Bd) in Madre de Dios, Peru.

Daily leaf litter temperatures of the two main forest types at Los Amigos Biological Station were very similar throughout the year (Figure 2), with greater fluctuations in the dry season (between May and October). During this period, low temperatures were associated with cold fronts. These cold surges occur during the Austral winter and are characterized by a rapid drop in air temperature associated with incursions of cold air masses originating in the Antarctic region and southern South America (Marengo, 1984; Marengo, Cornejo, Satyamurty, Nobre, & Sea, 1997). Animal activity in lowland Amazonia is affected during these cold surges. The ranges of daily minimum temperatures in both habitats were similar in the floodplain (13.4°C–25.7°C) and the terra firme (13.9°C–25.3°C). Likewise, the ranges of daily mean temperatures in both habitats were similar in the floodplain (15.1°C–26.9°C) and the terra firme (15.2°C–26.7°C). We also observed similar daily maximum temperatures throughout most of the year, with the exception of some periods of higher maximum temperature in the floodplain in August, September, and October 2017 (Figure 2). Furthermore, during most of the year, the maximum temperature measured in the leaf litter rarely exceeded 28°C (red line in Figure 2).

Figure 2.

Daily temperatures recorded from November 16, 2016 to December 5, 2017 (385 days) in two forest types at Los Amigos Biological Station, Peru. Weather station data are shown in gray; temperature data collected in the leaf litter are shown in green (floodplain forest) and blue (terra firme forest). In each case, the maximum and minimum temperatures delimited the polygons, and the line in the middle represents the mean temperature. The maximum temperature measured in the leaf litter rarely exceeds 28°C (red line).

Daily temperatures measured in the leaf litter and the understory vegetation of the four main forest types at Los Amigos show that two habitats, floodplain forest and terra firme forest, experience lower temperature fluctuations than the bamboo forest and the palm swamp (Figure 3). In addition to the distribution of Bd-positive records provided by Becker et al. (2016), we updated the known distribution of Bd-positive records in the lowland Amazon (<300 m elevation) to include records by McCracken et al. (2009), Kosch et al. (2012), and this study (Figure 4).

Figure 3.

Daily maximum temperatures recorded during the wet season in four habitats around Los Amigos Biological Station, Peru. These data were collected using iButton data loggers between February 14, 2008, and March 30, 2008. (a) Terra firme, (b) bamboo, (c) floodplain, and (d) palm swamp.

Figure 4.

Map showing the location of previous Bd-positive records (violet circles) in lowland sites (<300 m elevation; Becker et al., 2016; Kosch et al., 2012; McCracken et al., 2009) and new records from Madre de Dios, Peru (red triangle). The area outlined in orange indicates the boundary of the Amazon Basin, and the major river network is indicated in blue. Bd = Batrachochytrium dendrobatidis.

Discussion

Our findings suggest that the environmental temperatures in the leaf litter and understory vegetation of some lowland Amazonian forests offer suitable conditions for Bd growth. In particular, the leaf litter and understory vegetation of the floodplain forest and terra firme forest experience lower temperature fluctuations and maximum temperatures than bamboo and palm swamp habitats (Figure 3). Considering these findings, we hypothesize that Bd prevalence and infection will be less common in bamboo and palm swamp habitats because the leaf litter and understory vegetation of these forest types experience higher temperatures and broader temperature fluctuations. These conditions are known to inhibit growth of Bd in the lab and likely will inhibit its growth on hosts that primarily live in these forest types.

Many studies have predicted that lowland Amazonian forests do not offer suitable conditions for Bd growth (Ron, 2005, Rödder et al., 2009), and some researchers have argued that including lowland records of Bd infection in nonendangered amphibians will not be useful for understanding the climatic conditions of the pathogen (Menéndez-Guerrero & Graham, 2013). Although these studies recognized that Bd has a broad environmental niche, they suggested that lowland amphibian populations are not experiencing declines caused by chytridiomycosis. Nevertheless, given that a number of amphibian species are known to be infected by Bd in lowland Amazonia (<300 m elevation; Becker et al., 2016; McCracken et al., 2009; this study), and considering that at least one lowland species (Pristimantis toftae, Strabomantidae) is highly susceptible to chytridiomycosis (Catenazzi et al., 2017), we think that all Bd-positive records should be considered when making predictions about the pathogen’s distribution and potential impact on lowland rainforest species.

James et al. (2015) emphasized that questions relating to prevalence and distribution of Bd should be reframed, given that Bd might have been present historically at “cold spots”—areas where the pathogen may be present but offer suboptimal conditions for Bd growth—such as lowland tropical forests. Our findings support the idea that these “cold spots” might be widespread, especially in old-growth habitats that contain environmental conditions within the pathogen's tolerance window. Another important consideration is that amphibian immunity (especially adaptive) generally improves with temperature (Raffel, Rohr, Kiesecker, & Hudson, 2006; Raffel et al. 2012) so at warm temperatures, the frog’s immune system might be more effective at deterring Bd infection. A possible test of this idea would be to track the progress of Bd infections throughout the year, and examine if prevalence of infection decreases at warmer temperatures (e.g., Kriger & Hero 2007).

We hypothesized that 28°C is the critical maximum temperature limiting Bd growth in lowland Amazonia, assuming it has the same thermal niche of other Bd lineages. However, this assumption needs to be tested through experimental studies and special attention should be placed on measuring microhabitat characteristics that may differ from general habitat characteristics. In the meantime, it is reasonable to assume that many lowland Amazonian rainforest habitats offer suitable conditions for Bd growth. In turn, Bd growth is limited by occasional pulses of high temperatures that exceed 30°C as suggested by our iButton temperature data, and we propose that some Bd lineages may be able to adapt to the local conditions. Recent experimental studies focusing on Australian rainforest frogs (Litoria spenceri) support this idea, given that frequent exposure to temperatures exceeding optimum Bd growth values result in lower infection (Greenspan et al., 2017). Furthermore, Bd infections were cleared in animals that were exposed to daily heat pulses of 29°C lasting 4 h (Greenspan et al., 2017). At our main study site in Peru (Los Amigos Biological Station), our iButton data showed that some microhabitats experience heat pulses exceeding the hypothesized critical temperature of 28°C. Thus, these daily temperature fluctuations and occasional heat pulses might protect lowland Amazonian frogs from more severe infections caused by Bd.

Two lineages of Bd have been reported in South America, Bd-Brazil and the globally dispersed pandemic lineage (O’Hanlon et al., 2018). Further experimental work is required to test whether critical maximum temperature of these lineages is similar to that reported for other tropical isolates (Stevenson et al., 2013). If the Bd lineage from Madre de Dios is more similar to the Bd-Brazil lineage, then studies on critical temperature might require controlled laboratory experiments that include both Bd-Brazil and Bd-globally dispersed pandemic lineage.

Implications for Conservation

Over the past decade, many studies have examined the prevalence of Bd on amphibian populations in montane areas such as the eastern slopes of the Andes (e.g., Catenazzi et al., 2011; Seimon et al., 2017), and researchers highlighted that protecting the continuity and connectivity of natural habitats is essential for species' long-term survival in the face of pathogen infections and environmental change (Seimon et al., 2017; von May et al., 2008). Fewer studies examined the incidence of Bd in lowland tropical amphibians because none of these sites appeared to have suffered species loss attributable to chytridiomycosis. Nevertheless, it is essential to continue monitoring the prevalence of chytridiomycosis in lowland Amazonia and other lowland tropical rainforests (e.g., Indonesia, the Philippines, Equatorial Africa) to better understand what limits the pathogen’s distribution. Field studies in Amazonia will need to track the progress of Bd infections throughout the year, and especially during the dry season when cold fronts cause sudden drops in temperature that could potentially make frogs more susceptible to disease. Further studies focusing on lowland rainforest frogs are needed to assess whether Bd is enzootic in Amazonian amphibians and the role of “cold spots” for long-term disease dynamics. These studies will allow researchers to examine how many species have evolved defenses (Catenazzi et al., 2017) and identify those that are more vulnerable in order to prioritize conservation actions.