Assessment of fish communities using eDNA: effect of spatial sampling design in lentic systems of different sizes

Freshwater fish biodiversity is quickly decreasing and requires effective
monitoring and conservation. Environmental DNA (eDNA)-based methods have
been shown to be highly sensitive and cost-efficient for aquatic
biodiversity surveys, but few studies have systematically investigated how
spatial sampling design affects eDNA-detected fish communities across
lentic systems of different sizes. We compared the spatial patterns of
fish diversity determined using eDNA in three lakes of small (SL; 3 ha),
medium (ML; 122 ha) and large (LL; 4,343 ha) size using a spatially
explicit grid sampling method. A total of 100 water samples (including 9,
17, and 18 shoreline samples and 6, 14, and 36 interior samples from SL,
ML, and LL, respectively) were collected, and fish communities were
analyzed using eDNA metabarcoding of the mitochondrial 12S region.
Together, 30, 35, and 41 fish taxa were detected in samples from SL, ML,
and LL, respectively. We observed that eDNA from shoreline samples
effectively captured the majority of the fish diversity of entire
waterbodies, and pooled samples recovered fewer species than individually
processed samples. Significant spatial autocorrelations between fish
communities within 250 m and 2 km of each other were detected in ML and
LL, respectively. Additionally, the relative sequence abundances of many
fish species exhibited spatial distribution patterns that correlated with
their typical habitat occupation. Overall, our results support the
validity of a shoreline sampling strategy for eDNA-based fish community
surveys in lentic systems but also suggest that a spatially comprehensive
sampling design can reveal finer distribution patterns of individual
species.