Environmental DNA characterization of amphibian communities in the Brazilian Atlantic forest: Potential application for conservation of a rich and threatened fauna

Highlights

• We examine the use of eDNA metabarcoding to assess amphibian community composition.

• eDNA metabarcoding results were supported by a 5-year visual and acoustic survey.

• All species detected using eDNA have at least one life stage associated to streams.

• Survey constancy index was positively correlated to the proportion of positive PCR in eDNA analysis.

Abstract

The Brazilian Atlantic forest is a highly threatened biodiversity hotspot, harboring one of the highest levels of amphibian species richness in the world. Amphibian conservation in Neotropical biomes is critical because freshwater ecosystems typically experience sharp declines in biodiversity before much is known about species that depend on those environments. Environmental DNA (eDNA) analysis is a new approach for monitoring aquatic organisms and provides valuable information on species occurrence in freshwater ecosystems. Here, we assess community diversity in Neotropical streams using eDNA metabarcoding. We compare data from a five-year traditional field survey with results from a short-term eDNA analysis in four streams of the Atlantic forest of southeastern Brazil. We recorded 19 species over 5 years using visual-acoustic methods, of which 10 species are associated with the streams during at least one life stage (egg, tadpole or post-metamorphic). We were able to detect nine of the latter species using eDNA metabarcoding from water samples collected over 4 days. Amphibian community composition as measured by both methods showed high similarity in three streams, but was not perfect, as eDNA failed to detect known species in a fourth stream. Furthermore, in one stream we detected through eDNA metabarcoding a species (Aplastodiscus eugenioi) found only once during the 5-year traditional survey in that stream. Also, three species (Cycloramphus boraceiensis, Hylodes asper, and Hylodes phyllodes) with the highest dependence on aquatic habitat showed the highest number of positive PCR detections on eDNA samples. Our results showed that eDNA metabarcoding can be a useful tool to assess community diversity in tropical streams, and although not perfect in detection, this method can potentially improve our ability to conserve Neotropical amphibians.

Introduction

The Brazilian Atlantic forest is home to a high diversity of amphibians, harboring over 500 species (approximately 7% of all known species) of which 88% are endemic to this biome (Haddad et al., 2013). Distributed along the coast of eastern Brazil, the Atlantic forest covers mountainous terrains with a large number of high-gradient streams (Morellato and Haddad, 2000). These freshwater environments harbor a high diversity of amphibians, including habitat specialists with most or all life cycle stages depending on lotic waters (Haddad et al., 2013). The Atlantic forest is also one of the most threatened tropical ecosystems in the world (Myers et al., 2000) with only 16% of its original vegetation cover remaining (Ribeiro et al., 2009), posing a high risk to habitat specialists endemic to this biome. Several other factors, such as invasive species, and emergent infectious diseases (Both et al., 2011, Carvalho et al., 2017, Eterovick et al., 2005) also potentially contribute to observed amphibian declines in Atlantic forest. Not surprisingly, approximately 88% of the threatened anuran species in Brazil are from the Atlantic forest (ICMBio, 2016a) and most of the documented frog declines in Brazil include species closely associated with streams (Eterovick et al., 2005). Furthermore, over one fifth of Atlantic forest amphibians that are stream specialists lack baseline data and are listed as Data-Deficient according to the International Union for Conservation of Nature (IUCN, 2015, ICMBio, Instituto de Conservação Chico Mendes da Biodiversidade, 2016b). Thus, it is critical to describe the occurrence and distribution of stream frog species in the Atlantic forest and any method that makes this task easier and more precise will be of crucial importance for amphibian conservation.

Field techniques traditionally used to assess amphibian occurrence require a large investment in fieldwork and come with their own sources of error (Heyer et al., 1993). For instance, frog detectability through acoustic survey are highly variable among species, some species call only a few days per year making their detection less likely, and all types of traps used to sample amphibians (e.g., pitfall, funnel, pipe) provide biased results (Dodd, 2010, Petitot et al., 2014). Recently, the analysis of environmental DNA (eDNA) has emerged as a promising alternative approach for detecting aquatic species. The term eDNA refers to DNA shed by an organism in the environment, such as water, soil, or even air (Taberlet et al., 2012a), which can be then sampled for sequencing and species identification (Bohmann et al., 2014). The use of eDNA has recently gained widespread attention because it allows researchers to detect species even at low abundances (Dejean et al., 2012) and during short term field samplings (Lopes et al., 2017), making eDNA ideal for applications in conservation biology.

The number of ecological studies applying eDNA analysis to monitor species is increasing, yet few studies have sampled eDNA from aquatic environments to perform broad community surveys (but see Shaw et al., 2016, Valentini et al., 2016). Most eDNA efforts have focused on detection of a particular species using species-specific molecular markers (Thomsen and Willerslev, 2015). These efforts, while valuable, do not assess species diversity. If the goal is to survey the biodiversity of particular taxa and assess community composition, one can use DNA-based identification of a group of species at a locality with universal primers, an approach known as “DNA metabarcoding” (Taberlet et al., 2012b).

Here we report the results of a short-term eDNA metabarcoding study applied to a system of Atlantic forest streams where amphibian community composition was well known through a long-term traditional survey. By comparing our eDNA results with our traditional survey, we assess the utility of the eDNA metabarcoding for the characterization of tropical frog communities from a megadiverse region. If one or a few eDNA samples can accurately describe amphibian community composition in tropical streams, instead of long-term, time and labor-expensive traditional surveys, this method has high potential for use in conservation studies of Atlantic forest frogs.