Unveiling biogeographical patterns of the ichthyofauna in the Tuichi basin, a biodiversity hotspot in the Bolivian Amazon, using environmental DNA

degrees of success. Jackman et al. (2021) identified 84 molecular taxonomic units (MOTU), of which only 4 to the species level (1.6%), out of the 258 fish species detected using traditional sampling methods in the Adolpho Ducke Forest Reserve near Manaus in the Brazilian Ama- zon. Sales et al. (2021) recovered 34 out of the 111 species (30.1%) described in the Jequitinho- nha River basin, Eastern Brazil; de Santana et al. (2021) recovered 58 out of 443 species (13.1%) in the Javari River, Western Brazil; whereas up to 65% (132 out of 203 species) were detected in several watercourses of French Guiana [26]. Since the beginning of the 19th cen- tury until recently, 444 species of fish have been identified in the Beni basin [1]. Our eDNA approach, although carried out in a small portion of the same basin (10% of the area of the Beni basin) and during a single hydrological period, has resulted in a list of 207 species (47%). Among these species, 57 are newly identified for the Beni watershed as compared to the cur- rent most exhaustive census compiled by [1] among which 16 were named “affinis” because they were molecularly close to species not described in the Amazon basin. These 16 species could potentially be new species for science. All these results confirm our hypothesis that eDNA would be more efficient than traditional assessment methods to inventory species diver- sity and to unravel rare or cryptic species. They also emphasize the need to fill the knowledge gaps in the Tuichi sub-basin [2], where species potentially new for science were detected,

which should foster further sampling efforts using both traditional and eDNA approaches. In accordance with recent recommendations [58], which identified 87% of expected fish fauna in small streams with a single 34 l sample, we filtered an average of 38 l of water per site which allowed us to collect sufficient eDNA and achieve an mean coverage of 79.8% of the esti- mated number of species under the limits of the available databases (Bold, GenBank). Although increasing the sequencing effort could significantly increase the observed species diversity as shown by [59], sequences rarefaction tests performed on our samples show that this effort was sufficient (S1A Fig). It is worth noting that the results presented here were achieved in a 4-week sampling time- frame, for an average analysis and sequencing cost of approximately $450 per site (year base 2020), further emphasizing the effectiveness of the eDNA approach to inventory species. This Table 1. Correlations and significance levels between environmental variables and NMDS ordinations. These statistics are reported for the 3 NMDS ordinations car- ried out on i) rivers only, ii) lakes only and iii) all sites. Rivers Lakes Rivers and Lakes r2 Pr(>r) r2 Pr(>r) r2 Pr(>r) Hydro-ecoregion HER 0.59 ��� 0.81 �� 0.73 ��� Altitude ALT 0.78 ��� 0.93 ��� 0.53 ��� Longitude LONG 0.73 ��� 0.66 � 0.69 ��� Latitude LAT ns 0.66 . 0.28 � Geology 0.49 ��� ns 0.61 ��� Pedology 0.31 � 0.76 �� 0.59 ��� Ecotype na na 0.52 ��� Length of the hydrographic network HNL na 0.67 � na

Area of the basins ns 0.67 � na Age of the lakes na ns na Area of the lakes na ns na Significance codes: 0 ‘���’ 0.001 ‘��’ 0.01 ‘�’ 0.05 ‘.’ 0.1 ‘ ‘ 1 https://doi.org/10.1371/journal.pone.0262357.t001 PLOS ONE Fish diversity and distribution in the Tuichi basin PLOS ONE | https://doi.org/10.1371/journal.pone.0262357 January 4, 2022 15 / 25