Introduction
Tropical freshwater ecosystems harbour the most taxonomically diverse fish communities on the planet (Pelicice et al., 2017), which fulfill a wide range of critical functions such as nutrient recycling and seed dispersal (Lévêque et al., 2008; Toussaint et al., 2016; Vitule et al., 2017), while providing essential services like food provision and water quality regulation (Collen et al., 2014; Hoeinghaus et al., 2009; Longin et al., 2021; Pelicice et al., 2023). Despite their great ecological and societal value, neotropical freshwater fish are experiencing alarming biodiversity loss over recent decades due to multiple threats induced by human activities, including deforestation, flow regulation, overexploitation, non-native species introduction, and pollution (Antunes et al. 2016; Barlow et al. 2018). In many countries, the lack of adequate monitoring tools and reliable data hampers efforts to develop effective conservation stategies, contributing to an increasing homogenization of fish communities and modifications of species assemblages through local extinctions and species introductions (Barlow et al., 2018; Pelicice et al., 2017; Su et al., 2021).
To understand and quantify the impact of human pressures on freshwater ecosystems, it has become urgent to accuratly assess fish assemblages across various spatial and temporal scales. However, traditional methods for conducting fish inventories in tropical freshwaters (e.g. using gillnets, longlines or biocides) are costly and destructive (Allard et al., 2016; Araújo et al., 2009). They are also heavily reliant on the presence of taxonomic experts, particularly in highly diversified areas such as tropical rivers, characterized by complexes of closely related species, where morphological identification of specimens can be extremely challenging. Another strong limitation of traditional inventory net-based methods is their selectivity and limited capacity to detect small or rare species that may have a strong contribution to ecosystem functioning (Mouillot et al., 2013, 2014).
In recent years, environmental DNA (eDNA) metabarcoding has allowed the emergence of promising non-invasive and cost-efficient tools for biodiversity monitoring (Takahashi et al., 2023). While the analysis of water, soil, or sediment eDNA is now widely adopted, researchers are increasingly investigating the potential of « natural samplers » for local scale biomonitoring, based on living organisms that, through feeding, aggregate the DNA of species in their immediate environment (Siegenthaler et al. 2019). Ideal natural samplers should be abundant, widely distributed, easy to collect, and feed opportunistically on a wide taxonomic range of prey. (Boyer et al. 2015). For example, several studies have used the blood DNA of hematophagous insects (Kocher et al., 2017; Massey et al., 2022) or the feces of generalist predators (Nørgaard et al., 2021) to inventory flora and fauna in terrestrial environments. In aquatic environments, filter-feeding organisms such as sponges or mussels have been investigated as natural samplers for marine biodiversity to capture a wide range of eDNA from supsended particules in the water colum ( Weber et al. 2023; Mariani et al. 2019; Gallego et al. 2024). Similarly, the dietary DNA of small detritivorous invertebrates have shown promise for providing insight into local fish assemblages (Cordone et al., 2022; Siegenthaler et al., 2019).
Here, we investigated whether freshwater detritivorous shrimp can serve as “natural samplers” for assessing the local composition of highly diverse fish assemblages in neotropical riverine ecosystems, comparing their effectiveness with traditional monitoring methods. Our study focused on the fish communities of large rivers in French Guiana, a territory almost entirely covered by primary rainforest and characterized by a very dense hydrographic network that supports over 400 fish species, equivalent to the entirety of Western Europe (Le Bail, 2012). These ecosystems face increasing anthropogenic pressures, particularly from small-scale gold mining and logging activities, which have been shown to significantly alter local fish assemblages (Allard et al., 2016; Cantera, Coutant, et al., 2022; Coutant et al., 2023). In this context, French Guiana must comply with European regulations aiming at developing surveillance programs on water quality (Water Framework Directive 2000/60/EC, hereafter WFD) and requires the development of effective and non-invasive tools to complement or replace current inventory methods.
We developed a multi-marker metabarcoding approach to analyze the dietary DNA (dDNA, hereafter) of several abundant and widely distributed species of shrimp, which can be easily captured using traps along the riverbanks. These small scavenger crustaceans have restricted home ranges (Hirt et al., 2017) and exhibit a highly versatile feeding regime, ranging from feces to fish carcasses (da Cruz et al., 2021; Silveira De Melo & Nakagaki, 2013). As a result, they concentrate surrounding DNA in their digestive tracts, making them well suited as samplers for assessing local biodiversity. To obtain the most comprehensive and robust picture of fish assemblages, the dDNA was analysed using three mitochondrial markers (12S rRNA and cytochrome c oxidase 1 [COI] gene) that differ in taxonomic resolution and coverage (Polanco F. et al., 2021; Quéméré et al., 2021). We compared the taxonomic diversity and inferred size range of fish assemblages recovered through dDNA with those recorded after 10 days of intensive fishing with gillnets, cast nets and traps, as well as with data obtained from the standardized WFD fish monitoring protocol traditionally used in French Guiana.
Given the opportunistic feeding strategy of shrimp and their diverse microhabitats, we expected that dDNA analysis would reveal a greater number of fish species compared to traditional WFD monitoring methods, likely being more effective at identifying rare or cryptic species. Additionally, dDNA analysis is expected to be less biased toward larger species than gillnet-based fishing methods. From a methodological perspective, we also anticipated that the use of multiple markers would help overcome PCR biases and provide a more complete picture of fish diversity compared to a single-marker approach.