Discussion
This study was carried out to design and evaluate the use of an eDNA assay for the detection of S. nitida in ditches at Stodmarsh NNR and to compare to manual survey data taken shortly after water samples were collected. S. nitida DNA was detected at 62.5% (5 of 8) of the ditches where S. nitida presence was detected by manual survey and in two additional ditches where S. nitida was not detected by manual survey (7 ditches in total). A further 22 ditches sampled within Stodmarsh NNR and outside the known distribution ofS. nitida were negative for S. nitida. This resulted in an observed percentage agreement of 84% and a kappa coefficient of 0.56 which shows a moderate agreement between the results. When taken together, manual survey in combination with eDNA analysis has led to improved S. nitida detection rates, that is, 10 of 32 ditches rather than the 8 reported by field survey. However, it still should be noted that where eDNA was detected but no S. nitida found by manual survey (ditches 98 and 115) it is unknown whether the animals were present but not detected by manual survey (present in very low numbers), or whether they are absent and the eDNA drifted into the ditch e.g. through hydrological connections. Ditches 98 and 115 are not directly adjacent to occupied ditches and the direction of flow would not ‘normally’ facilitate drift between the ditches concerned in normal flow conditions, but ultimately they are connected nevertheless. Site occupancy models can be used to account for imperfect detection and were used by Schmidt et al. (Schmidt et al. 2013) to demonstrate their applicability to eDNA surveys. When applied to the data within this study, site occupancy estimates were greater than the actual observed proportion though not significantly when combining field survey with eDNA assay. This matches the observed increase in positive detections from 8/32 to 10/32 ditches when both techniques were combined.
Where rare or threatened species are concerned, it is likely that their detection by either manual survey or eDNA will be imperfect leading to an underestimation of its distribution.
During particular time periods or developmental stages, some species can be difficult to detect potentially biasing survey outcomes (Gotelli and Colwell 2001; Mackenzie
et al. 2006). This may have been the case here for the two ditches whereS. nitida was found by manual survey but not via qPCR of eDNA. A study in Poland showed that S. nitida breed during April-May (Książkiewicz and Goldyn, 2008) and although conditions may be slightly different in the UK as these ditch water samples were taken outside the likely breeding season their detection will be more difficult as there is likely to be less S. nitida DNA in the water. The sampling strategy of taking 20x 30 mL samples along the length of each ditch should have overcome the fact that eDNA can be highly localised in space and time (Li et al, 2019). To achieve a higher level of coverage (especially for longer ditches), more samples may need to be taken. This could allow for targeting more locations within the ditches whereS. nitida would most likely be present, that is, where the ditch is thickly vegetated, thus improving the probability of detection.
A further cause of false negative results can be PCR inhibition (Jane et al. 2015, McKee et al. 2015). Upon testing of the DNA extracts only two samples, from ditches 136 and 146, were found to cause complete inhibition of the qPCR inhibition assay. These were not from ditches that were positive for S. nitida on manual survey so are unlikely to be false negatives. Therefore, it is unlikely that inhibition is the reason for the qPCR of the samples from ditches 62, 70 and 108 being negative for S. nitida .
It is also likely that the volume of water filtered will play an important role in the detection of S. nitida . The volume of water filtered was between 200 mL and 500 mL for all samples - the volume determined by how soon the filter clogged. A larger pore sized filter eg. 0.45µM or 0.8µM may have allowed a larger volume of water to be filtered which could enable more S. nitida eDNA to be recovered at sites where it was found by manual survey. It is common for volumes of water between 500 mL and 5L to be filtered although there is little consensus on the minimum volume (Bruce et al., 2021). Small volumes (0.25 L) have been shown to contain detectable eDNA from macroinvertebrate species when a range of volumes up to 2 L were sampled and analysed (Machler et al., 2016) and increasing the volume of filtered water has been shown to have a positive effect on eDNA capture and PCR amplification efficiency (Muha et al., 2019).
Finally, the density of snails in the ditch system will play an important role in its detectability. The relative abundance of S. nitida was recorded during the manual survey (Supplemental file 1), and at sites where the eDNA assay was negative for S. nitida but it was found during manual survey, the abundance was recorded as occasional or rare. Studies on the New Zealand mud snail have shown that this snail’s eDNA can be detected when snails are present at low densities (Goldberg et al, 2013), this species is similar in size to S. nitida , however, far larger volumes of water (4 L) were filtered (using multiple filters if necessary) prior to analysis which could account for its detectability at low density although eDNA transport would need to be considered as these samples were taken from a river system.
Since this study was carried out a study (Hobbs et al., 2021) investigating the population structure of S. nitida individuals from Poland, Germany, Sweden and the UK to identify differences both within and between populations has been published. The study found that there are two distinct genetic lineages (also distinct in shape), one in western Europe (UK, Germany – Lineage 1) and one in eastern Europe (Poland, Sweden – Lineage 2). Although only a UK population was tested during the present study it likely that the assay designed herein would also detect S. nitida populations in eastern Europe as during the primer design phase the three S. nitida sequences retrieved from Genbank were from specimens collected in Poland, Denmark and Germany. Future work could involve testing our primer/probe combination on individuals from these other populations.
To confirm the results of the eDNA assay designed herein further manual survey is required to corroborate the two additional eDNA assay positive ditches found. For those manual survey samples where S. nitidaspecimens were found but the eDNA assay did not detect S. nitidaDNA a larger pore sized filter/filtration of more water may be required to enable eDNA assay corroboration. This study has shown that eDNA assays can be used for the detection of S. nitida and if used in the future could have time and cost savings and could inform manual survey and therefore management of the ditches.