Environmental DNA (eDNA) monitoring of two different freshwater host-pathogen complexes in the interface between nature and aquaculture

Abstract

Environmental DNA (eDNA) monitoring methods are increasingly used as a supplement or substitute to conventional monitoring. This rapidly advancing research field promises improvements for aquatic species conservation and the detection of invasive species and pathogens. The eDNA dynamics of some groups of organisms like fish have been extensively studied, in particular fish of commercial interest or where there is a high invasive potential. However, there are still many knowledge gaps on eDNA dynamics and monitoring potential for rare and elusive species, and for host-pathogen complexes. The overarching goal of this thesis was to explore, develop and evaluate the potential of targeted eDNA detection and quantification as surveillance and biosecurity tool both in nature and aquaculture. For this purpose, we chose two dissimilar host-pathogen complexes, which are of economic importance and relevance regarding species conservation: The Atlantic salmon and the salmonid parasite Gyrodactylus salaris and freshwater crayfish with their obligate parasite Aphanomyces astaci. The salmon fluke G. salaris has caused significant damage to indigenous Atlantic salmon populations in Norway, and the Norwegian Government is working towards the eradication of this parasite. The oomycete A. astaci, carried and transmitted by American freshwater crayfish species, causes crayfish plague and is the largest threat to endangered European crayfish species, and is registered as a list 3 disease (national disease) in Norway. The same applies for G. salaris. In these host-pathogen complexes, fish shed much larger amounts of eDNA than crayfish as they are covered with a mucus layer. Conversely, the sporulating oomycete A. astaci is readily detectable using the eDNA methodology while the flatworm G. salaris assumingly only shed minute amounts of eDNA. Three main research questions were asked: 1) Can the eDNA methodology work equally well or better than conventional methods for biomonitoring of the host-pathogen models, particularly at low prevalence? 2) Can eDNA copy numbers serve as a proxy for host density and pathogen intensity? 3) How will environmental factors and organism biology influence the emission and detectability of host-pathogen eDNA? We used both qPCR and ddPCR and drew upon already published species-specific assays or developed new ones where required (paper I, III, IV). For eDNA sampling, we adapted an already developed method but modified minor aspects like equipment (paper III) and storage of filter samples. Sampling of eDNA was conducted both under natural conditions in the field and under controlled conditions in an aquarium facility. We designed and conducted two mesocosm experiments to compare eDNA copy numbers with parasite intensity of G. salaris on Atlantic salmon (paper IV) and to examine the influence of temperature, density and food availability on the detectability of eDNA of A. astaci and signal crayfish (paper V). We showed that simultaneous eDNA monitoring of host-pathogen complexes is advantageous for biomonitoring purposes, but the outcome is highly dependent on the type of organism targeted and its biological traits (paper I-V). For the crayfish – A. astaci complex the eDNA methodology proved more sensitive and more animal welfare friendly than conventional methods, and simultaneous detection of crayfish provide information regarding noble crayfish population status (presence-absence) and potential threats from disease or non-indigenous crayfish. The method eliminates the need for live caged noble crayfish for disease monitoring, and detects the presence of crayfish down to very low population densities provided sufficient sampling effort (paper II, III, V). For the Atlantic salmon – G. salaris complex, results from our mesocosm experiment suggest that the eDNA methodology fails to detect parasite presence at low intensities with the same detection reliability as conventional methods (paper IV), but will nevertheless be a useful supplement to the work-intensive conventional methods (paper I). Field data also suggest a higher degree of detection success than we observed which is most likely due to experimental constraints in our study. We also developed assays for direct eDNA detection of specific mitochondrial haplotypes of G. salaris. These can differ in pathogenicity towards Atlantic salmon and may yield information on the origin of the infection. However, these assays targeting the mitochondrial COI gene are less sensitive than the nuclear ribosomal ITS-assay, which is better suited and more robust for presence-absence screening of G. salaris. Estimations of biomass or relative abundance inferred from eDNA copy numbers are not straightforward as the amount of shed and detectable eDNA is substantially influenced by a multitude of factors. This poses a particular challenge for the detection of organisms that – through their very nature – shed less eDNA than others such as G. salaris or crayfish, of which the latter additionally spend considerable time buried beneath the substrate in their habitat. Lifecycle events play a major role in the eDNA dynamics. As dead crayfish emit more eDNA than live ones, a mass mortality event could be mistaken for a high density population and likewise, the capture of a dead free floating G. salaris specimen could be mistaken for high parasite intensity on fish. Furthermore, environmental factors heavily influence eDNA detectability even when the presence of the organisms remains unchanged. Our results show that changes in host density and pathogen intensity can be concealed by many other factors, rendering estimations of relative abundance highly challenging and for most practical purposes impossible. Here, detection frequency and probability of positive detection stand out as a better indicators of crayfish population density or G. salaris parasite intensity. The shedding of eDNA does not happen in a uniform rate or manner and the source of eDNA also varies depending on the organism. The main source of eDNA of A. astaci are zoospores that are released at a relatively low rate from American carrier crayfish, but massproduced during crayfish plague outbreaks. For G. salaris, there appears to be minimal shedding of eDNA from live parasites, leaving the main source of eDNA to be specimens that have become detached from their hosts, floating in the water. Its biology, including the clonal reproduction and parasitic nature, leave very few eDNA traces in the water from live, attached parasites. However, the rapid reproduction rate and near exponential growth of parasite numbers aid eDNA detection as we observed an increase of probability of positive detection per sample with increasing numbers of parasites. For the host, abraded cells and mucus constitute the main source of eDNA, and we did observe a generally high and relatively stable amount of eDNA from Atlantic salmon (paper I, IV). Due to the hard exoskeleton, crayfish shed substantially less eDNA than fish. Further, both temperature and food influenced the eDNA detection rates of crayfish and A. astaci (paper V). For A. astaci 20 °C was close to the upper temperature limit for sporulation, leading to drastically reduced eDNA detectability. The presence of food probably led to faster eDNA degradation through increased microbial activity, which greatly reduced the eDNA amount from crayfish. Here, live A. astaci spores were able to withstand this, and eDNA detection was not affected. Life-cycle events can significantly influence the released amount of eDNA. Crayfish release more eDNA during reproduction, moulting and death, and infections with A. astaci lead to increased sporulation, particularly during crayfish mass mortalities resulting from crayfish plague. Environmental factors, such as dilution effects and inhibitors in the water impact negatively on the eDNA detectability. The differences in eDNA emission within and between the two host-pathogen models require special considerations for monitoring strategies with respect to water temperature and target organism biology. A consideration of life-cycle events may increase detection success. Sample numbers and volume required for high detection probability must be considered. For eDNA monitoring of host-pathogen complexes, we strongly recommend testing the samples for all relevant species, even if only one is of direct interest. This, and method considerations suitable for specific habitats, should guide the eDNA monitoring strategy. In conclusion, our results show that the amount of detectable eDNA can fluctuate in response to environmental or biological influences while the physical presence of the target organisms remains unchanged. Therefore, eDNA monitoring seems unsuitable for direct quantification of relative density or biomass, but is a powerful tool for presence-absence monitoring when the organism biology and ecology, along with environmental factors and habitat characteristics are taken into account.