Monitoring a Norwegian freshwater crayfish tragedy: eDNA snapshots of invasion, infection and extinction

1. The European noble crayfish Astacus astacus is threatened by crayfish plague caused by the oomycete Aphanomyces astaci , which is spread by the invasive North American crayfish (e.g. signal crayfish Pacifastacus leniusculus ). Surveillance of crayfish plague status in Norway has traditionally relied on the monitoring survival of cage‐held noble crayfish, a method of ethical concern. Additionally, trapping is used in crayfish population surveillance. Here, we test whether en vironmental DNA (eDNA) monitoring could provide a suitable alternative to CPUE value 0.17. the


| INTRODUC TI ON
Environmental DNA (eDNA) monitoring of aquatic systems is a rapidly advancing research field that promises improvements, not only to aquatic species conservation, but also for early detection of invasive species and harmful pathogens at low densities and at any life stage or season (Bohmann et al., 2014;Kelly et al., 2014;Strand et al., 2014). Water can be screened for the presence of micro-and macroorganisms by either a broad approach such as metabarcoding (Shaw et al., 2016;Valentini et al., 2016), or a targeted approach using species-specific quantitative real-time PCR (qPCR) or droplet digital PCR (ddPCR) Strand et al., 2014;Thomsen & Willerslev, 2015). eDNA studies have been applied for detection of a wide range of aquatic macroorganisms including freshwater crayfish (Agersnap et al., 2017;Dougherty et al., 2016;Tréguier et al., 2014). Molecular detection and quantification of waterborne pathogens in environmental samples has been widely utilised for decades (Ramirez-Castillo et al., 2015).
Aphanomyces astaci infection is a notifiable disease both nationally in Norway (list 3, national disease;  and internationally (OiE, 2017). It causes a rapid decline in European crayfish populations, and is spread and maintained by invasive non-indigenous North American carrier crayfish that have rapidly established themselves in Europe (Holdich, Reynolds, Souty-Grosset, & Sibley, 2009). The pathogen invades the cuticle of all freshwater crayfish, but hyphal growth is inhibited by melanisation in resistant North American crayfish. In susceptible crayfish species, the hyphae grow deeper into tissues and organs, causing rapid death. The oomycete reproduces asexually via clonal flagellated zoospores that locate new crayfish hosts through weak chemotaxis. Zoospores can encyst and re-emerge several times, but both zoospores and cysts have a relatively short life span (2-8 weeks) dependent on water temperature (Söderhäll & Cerenius, 1999).
An A. astaci species-specific qPCR method is widely used for crayfish plague diagnostics and carrier status testing (Kozubikova, Vrålstad, Filipova, & Petrusek, 2011;OiE, 2017;Vrålstad, Knutsen, Tengs, & Holst-Jensen, 2009). The same method, which has been thoroughly tested and further developed (Makkonen, Strand, Kokko, Vrålstad, & Jussila, 2013;Strand et al., 2012), is used for eDNA monitoring for the presence of A. astaci zoospores and cysts in both small  and large water bodies Wittwer et al., 2018). These studies have established that clinically healthy American crayfish emit a low number of A. astaci zoospores to the water regardless of season (Strand et al., 2012Wittwer et al., 2018), while moribund infected susceptible crayfish emit huge numbers of infective zoospores (Makkonen et al., 2013). Lake Øymarksjøen in the Halden watercourse is one of a few lakes in Norway hosting a population of the non-indigenous signal crayfish Pacifastacus leniusculus, which were introduced illegally around two decades ago, but not discovered until 2008 (Vrålstad, Johnsen, Fristad, Edsman, & Strand, 2011). The unknown presence of signal crayfish partly ruined long-term attempts to restock the lake with indigenous noble crayfish (Astacus astacus), following the first outbreak of crayfish plague in 1989 (Taugbøl, 2004). When the restocked population increased in number, a new large outbreak of crayfish plague occurred in 2005 (Vrålstad et al., 2009). The Norwegian Food Safety Authorities (NFSA) enforced a permanent closure of the Ørje water locks between Lake Øymarksjøen and Lake Rødenessjøen in an attempt to prevent upstream spread of A. astaci and signal crayfish .
The noble crayfish population in Lake Rødenessjøen has been monitored every year since 2009 as a part of the national surveillance programme, using baited traps set at eight stations throughout the lake. During this period, the relative density of noble crayfish increased, and CPUE in 2014 ranged between 0.15 and 1.80 . In September 2014, both signal crayfish and noble crayfish were caught in the southern part of Lake Rødenessjøen just above the closed water locks.
The Norwegian Environmental Agency (NEA) regarded the event as another illegal introduction of signal crayfish, since long-distance migration over land or through the closed locks was highly unlikely (Norwegian Environmental Agency, 2014). The illegally status. This non-invasive, animal welfare friendly method excludes the need for cage-held susceptible crayfish in disease monitoring. Furthermore, eDNA monitoring is less likely to spread A. astaci than traditional methods. This study resulted in the implementation of eDNA monitoring for Norwegian crayfish plague and crayfish surveillance programmes, and we believe other countries could improve management strategies for freshwater crayfish using a similar approach.

K E Y W O R D S
crayfish plague, disease surveillance, environmental DNA, host-pathogen, invasive species, noble crayfish, signal crayfish, species-specific detection introduced signal crayfish were confirmed A. astaci carriers, indicating the probable onset of a new crayfish plague outbreak in the local noble crayfish population. A crayfish plague surveillance programme commissioned by the NFSA was therefore conducted using live noble crayfish in cages to monitor the spread of the disease. Traditional cage experiments using noble crayfish as 'canaries in a coalmine' had been the sole method utilised for field monitoring of crayfish plague since it is introduction to Norway in the 1970s (Håstein & Unestam, 1972;Vrålstad et al., 2014).
Decapod crustaceans are now covered by the Animal Welfare Act in Europe and the Law on Animal Welfare (LOV-2009-06-19-97) in Norway. Thus, the use of live crayfish for monitoring a lethal disease is of strong ethical concern. In addition to fatal infection with crayfish plague, cage-held crayfish are also subject to other causes of mortality such as moulting-associated cannibalism.
Furthermore, cage-held crayfish commonly escape due to illegal human interference . Previous studies have shown that eDNA monitoring of crayfish plague in large water systems is possible , but a direct comparison with traditional cage surveillance has not yet been performed.
In the present study, we took advantage of an emerging crayfish plague outbreak and compared traditional cage surveillance with eDNA monitoring using species-specific qPCR assays for targeted detection and quantification of A. astaci , noble crayfish and signal crayfish (Agersnap et al., 2017), from the same water samples. In addition, we used trapping data from 2014 and 2015 to compare and verify crayfish presence. We show that eDNA monitoring can reveal the presence of A. astaci in the water earlier than cages with live crayfish, and that the simultaneous monitoring of noble-and signal crayfish eDNA provides additional information on habitat status that otherwise must be obtained from separate CPUE surveys. Consequently, we propose that eDNA monitoring of the three species will prove a suitable, non-invasive and animal welfare friendly alternative to the traditional cage method.

| Study site
The study site ( Figure 1) is part of the large Halden watercourse, which is 149.5 km long and consists of several lakes and connecting rivers and channels. The watershed covers 1,584 km 2 and con-  Frozen crayfish were thawed, and tissue samples of eye, tail muscle and cuticle were subjected to DNA extraction using the QIAamp ® DNA mini kit on a QIAcube automated DNA extractor (Qiagen) following the manufacturers protocol. Crayfish plague diagnostics were performed using an A. astaci-specific qPCR (Vrålstad et al., 2009), with modifications in the annealing temperature (Kozubikova et al., 2011). If crayfish plague was confirmed, the corresponding cage was removed from the watercourse. Cage surveillance lasted from September 2014 to October 2015. and 2). For cost-efficiency reasons, only two replicate water samples were collected per station. In total, 55 and 57 water samples were collected with an average of 3.3 and 4.0 L/filter in 2016 and 2017 respectively. Generally, for all stations, the water samples were taken upstream and at some distance (>20 m in the river and >200 m in the lake) to the nearest caged noble crayfish to avoid detection of eDNA from those crayfish. Between 1 and 10 L were filtered per sample depending on the turbidity of the water. The water samples were collected above the bed (~7 cm), 2-5 m from the shore, and filtered directly onto glass fibre filters (47 mm, 2 μm pore size, AP2504700 Millipore, Billerica, MA, USA) using a peristaltic pump (Masterflex L/S or E/S, Cole-Parmer, Vermon Hills, IL, USA) with Tygon tubing (Cole-Parmer) and an in-line filter holder (47 mm, Millipore). Each filter was transferred to a 15-ml sterile falcon tube, stored on ice in a cooling box until transported to the laboratory within 12 hr, and frozen at −20°C. The volume of the filtered water was measured and discarded on the shore at each site. Water samples were always collected in an upstream to downstream direction to avoid transferring A. astaci spores upstream. Also, stations outside the infection zone (risk zone) were always sampled before stations within the infection zone ( Figure 1). Before filtration at each station, water was pumped through the hose and filter holder for a few minutes to rinse away remains of spores or eDNA from the previous upstream station, and F I G U R E 1 The study site includes parts of the large Halden watercourse in Norway with names for involved lakes, channels and rivers. Cage stations (green squares) and environmental DNA (eDNA) stations (blue circles) were established successively from 2014 to 2016 in a south-north direction, starting at the signal crayfish invasion site at Ørje locks (bold black line; station 1). Cage stations 1-6 and eDNA stations 1-7 and 12 are within the regulated infection zone, while the eDNA stations 8-11 and 13-15 are located in the risk zone, separated from the infection zone by migration barriers (bold black lines) such as dams and waterfalls to avoid filtering any disturbed sediments from the current station.

| eDNA water sampling
After sampling of all stations within a zone (risk zone or infection zone), the tubing and filter holder were disinfected with 10% bleach for 30 min, followed by rinsing with 10% sodium thiosulfate, to remove DNA traces.

| eDNA analyses
DNA was extracted from filters using the CTAB (cetyltrimethylammonium bromide) extraction protocol described by Strand et al. (2014) with minor modifications (full protocol in Appendix S1).
Briefly, the filters were freeze-dried, 4 ml of CTAB buffer was added and the filters were then fragmented using a pestle. The samples were frozen (−80°C) and thawed (65°), followed by addition of proteinase K and incubated at 65°C for 60 min. Chloroform was added, the sample was centrifuged and the supernatant (3 ml lysate) from each sample was divided into two 2-ml Eppendorf tubes for easier The DNA samples were analysed using three different probebased singleplex qPCR assays referred to as Aphast, Astast and Paclen (see Table 1 for a qPCR assay specifics). Aphast is the A. astaci qPCR assay adapted for detection and quantification in water If none or only one of the replicates was detected above limit of quantification (LOQ), further quantification was not performed and the result for the eDNA sample was reported as below LOQ (<LOQ) (see Table 1 for limit of detection (LOD) and LOQ specifics). A sample result was only regarded as positive if the overall detection (mean for all PCR replicates) was above LOD (Table 1).
Following Kozubikova et al. (2011) andAgersnap et al. (2017), a cut-off was set at Ct 41, defining positive signals with a Ct value ≥41 negative (i.e. not detected). Environmental DNA copy numbers per litre water were calculated from the eDNA copy number quantified in the qPCR reactions according to Agersnap et al.
(2017) using the equation: C L = (C rAB * (V e /V r ))/V w . Here, C L represents the copies of eDNA per litre lake water, C rAB represents the copies of eDNA in reaction volume summarised for subsample A and B, V e represents the total elution volume after extraction, V r represents the volume of eluded extract used in the qPCR reaction and V w represents the volume of filtered lake water. The Aphast qPCR assay targets the multicopy ITS nrDNA-region (see Table 1).
The spore concentrations for A. astaci (spores/L) were estimated according to Strand et al. (2011Strand et al. ( , 2014 using the equation: C L /138, based on the estimation that one spore contains ~138 copies of the target DNA.   Table 2). Table S1 provides details for eDNA copy numbers for all targets, and A. astaci spore estimates.

| Cage surveillance versus eDNA monitoring
We observed that presence/absence data, as well as fluctuation in eDNA concentrations, depicted to a large extent the TA B L E 1 Overview of the three species-specific assays used in the study, targeting Astacus astacus, Pacifastacus leniusculus (Agersnap et al., 2017) and the crayfish plague agent Aphanomyces astaci (Vrålstad et al., 2009). The target gene regions are mitochondrial genomic cytochrome oxidase 1 (CO1) and the nuclear genomic internal transcribed spacer (ITS) biological status of the crayfish and habitat in terms of freedom from disease, early infection, mortality and extinction. When the ice cover thawed in 2015, plague-induced mortality in the cage was observed at station 3 3 weeks prior to our first eDNA sampling event (24 April, Figure 3a, Table 2). Here, high levels of eDNA from A. astaci and noble crayfish were detected, with a further increase 2 weeks later, followed by a decline to trace amounts in the following weeks with no detection by August (Figure 3c). At station 4, only low levels of noble crayfish eDNA were detected on 24 April, while both noble crayfish and A. astaci were detected 2 weeks later (May 8th, Figure 3c).
One week later, crayfish plague-induced mortality was observed in the cage (Figure 3a, Table 2 (Table 2), and the eDNA concentrations of noble crayfish remained stable throughout the sample period. No eDNA from signal crayfish was detected at any station other than station 1 (Figure 3). The parallel increase and subsequent decrease in eDNA concentrations of A. astaci and noble crayfish correlated significantly (rho = 0.485; p = 0.0043, Figure 3c). Table S2 provides eDNA copy numbers for all targets, and spore estimates of A. astaci for 2015. Six samples from 2014 and another six from 2015 were excluded due to minor contamination detected in the laboratory work control or DNA blank control for these samples respectively (c.f. Figure 3).

| Trapping data versus eDNA
We found that trapping data and eDNA data are in agreement with regard to presence/absence results. At stations 2 and 3, noble crayfish eDNA was detected in 2014 (Table S1) (Tables S1 and S2). The trapping surveys suggest that signal crayfish were restricted to the southern part of the lake at low density.
Here, 110 signal crayfish were caught in 2015 using 960 trap nights F I G U R E 4 Triangular split circles indicate detection of eDNA from Astacus astacus (green), Pacifastacus leniusculus (yellow) and Aphanomyces astaci (red) per station in 2016 and 2017; these are not to be interpreted as pie charts. No detection is indicated with no colouring. Stations 1, 4, 6-7 and 12 are within the infection zone, while the stations 8-11 and 13-15 are located in the risk zone, separated from the infection zone by migration barriers (bold black lines) such as dams and waterfalls. The only change from 2016 to 2017 is found at station 6, where eDNA from Aphanomyces astaci was detected only in 2016 (CPUE = 0.12), and only large individuals were trapped (average 118.2 mm, N = 91), suggesting their recent release.

| Implementing eDNA monitoring
The comparative data obtained with eDNA monitoring and traditional methods (cages and trapping) convinced the authorities to officially include eDNA as a monitoring method. Thus, in 2016, eDNA was officially integrated into the national crayfish plague monitoring programme commissioned by NFSA. Cages were only used in the risk zone (data not shown), and cage surveillance was discontinued from 2017. The eDNA monitoring focus shifted to the River Hølandselva (station 6-7), and upstream locations (station 8-15) in addition to stations 1 and 4 (Figure 1). Several new stations (8-10, 13-15) were established in the risk zone to monitor potential spread.
Noble crayfish eDNA was detected at all stations in the risk zone ( Figure 4, Table S3), while no signal crayfish or A. astaci eDNA was detected here. In the River Hølandselva, eDNA from A. astaci and noble crayfish was detected at the outlet of the river in 2016 (station 6), while only eDNA from noble crayfish was detected further upstream in the river (station 7) (Figure 4). At station 4, eDNA of A. astaci and noble crayfish was no longer detected, and in 2017, all signs of A. astaci had disappeared from all stations with the exception of station 1 (Figure 4). At station 1, eDNA from signal crayfish and A. astaci was still detected (Figure 4). Table S3 provides details for eDNA detection frequency for all targets for 2016-2017.

| D ISCUSS I ON
eDNA monitoring provides a reliable, non-invasive, ethical and animal welfare friendly alternative to cage monitoring for early detection of crayfish plague. During the predicted freshwater crayfish disaster in the Norwegian Halden watercourse, we demonstrated that eDNA monitoring can reveal the invasion of signal crayfish at low densities, as well as low numbers of waterborne infectious A. astaci spores 2-3 weeks prior to observation of mortality in cage-held susceptible crayfish. Furthermore, eDNA monitoring is less likely to spread A. astaci than traditional methods. As a direct consequence of the present study, eDNA monitoring has been adopted in crayfish plague disease management in Norway (Vrålstad, Rusch, Johnsen, Tarpai, & Strand, 2018;. We also confirmed the efficacy of simultaneous eDNA monitoring of three target organisms, represented in this study by a Red list species, an invasive species and a harmful pathogen, which has recently been demonstrated for invasive signal crayfish, endangered white-clawed crayfish and the crayfish plague pathogen in the UK (Robinson, Webster, Cable, James, & Consuegra, 2018).
eDNA monitoring provides a snapshot of the crayfish and habitat status, such as invasion, infection and extinction. After the discovery of low signal crayfish eDNA levels (early invasion state), the repeatedly observed and significantly correlated increase and subsequent decline of eDNA from A. astaci and noble crayfish spanning only a few weeks at each station depict the acute disease situation (infection outbreak) followed by local noble crayfish extinction. Increased levels of noble crayfish eDNA during the crayfish plague outbreak could be caused by decay of dead noble crayfish, resulting in increased eDNA release to the ambient water. However, behavioural changes, such as uncoordinated spasmodic limb tremors (Alderman et al., 1987), loss of nocturnality (Westman, Ackefors, & Nylund, 1992), reduced escape reflex and progressive paralysis (OiE, 2017) make noble crayfish easier prey. Increased feeding on crayfish by predators may also contribute to increased eDNA shedding. The rapid decline and disappearance of A. astaci eDNA also supports previous studies showing that A. astaci has a short life span outside its host (Svensson & Unestam, 1975;Unestam, 1966). The rapid transmission of crayfish plague and the subsequent loss of noble crayfish throughout Lake Rødenessjøen (15.95 km 2 ), Lake Skulerudsjøen (1.82 km 2 ) and River Hølandselva from September 2014 to August 2015, demonstrates the devastating effect of crayfish plague on indigenous European crayfish populations (Holdich et al., 2009;Söderhäll & Cerenius, 1999;Svoboda, Mrugala, Kozubikova-Balcarova, & Petrusek, 2017). The rapid spread of A. astaci throughout the lakes can be facilitated by several factors, including an enormous bloom of infectious swimming zoospores produced from each dying crayfish individual (Makkonen et al., 2013), and wind driven currents leading to rapid spread from crayfish to crayfish in the population. Furthermore, fish feeding on diseased and dying crayfish act as long-distance vectors since A. astaci survive the passage through the fish gut (Oidtmann, Heitz, Rogers, & Hoffmann, 2002). However, despite the rapid spread throughout the two lakes, the outbreak was still active in River Hølandselva 1 year after initial infection. Advancement of spread then slowed, most likely due to slower upstream spread in a flowing river combined with the absence or very low density of noble crayfish, working as barriers for further spread. In fact, the crayfish plague seemingly burnt out, as it is no longer detectable in terms of eDNA in 2017.
Our study indicates that trapping data and eDNA data are comparable when used to measure the presence/absence, but do not always agree for measuring biomass. Relatively low CPUE mea- False negatives resulting from PCR inhibition are always a risk with environmental samples. The water in Halden watercourse is relatively turbid (e.g. Lake Skulerudsjøen and Lake Rødenessjøen had average secci depths of 1.2 and 1.6 m, respectively, in 2016). Filtering larger volumes of water might increase the risk of inhibition during PCR, due to the presence of PCR inhibitors such as humic acids. All our samples were run both undiluted and 10-fold diluted in order to account for PCR inhibition, and several samples showed signs of inhibition (difference in Ct values of <2.85). This may in some cases have led to underestimation of the actual eDNA concentration of some samples in this study. Additionally, the presence of low levels of eDNA from crayfish may be masked in some samples due to inhibition of the PCR reaction. Recent studies suggest that the use of ddPCR increases the detection rate of eDNA compared to qPCR, especially at low DNA concentrations, and is more robust against inhibition . ddPCR also offers absolute quantification and precise multiplexing (two or more targets in the same reaction) (Whale, Huggett, & Tzonev, 2016). Adopting the existing assays to develop a multiplex assay for eDNA detection of all three species in a single reaction would thus be beneficial. Additionally, future eDNA studies should also be designed to incorporate occupancy modelling to estimate the detection sensitivity using traditional surveillance and eDNA monitoring (Schmelzle & Kinziger, 2016).
An important goal of this study was to contribute to the reduction or replacement of live crayfish in crayfish plague monitoring. As a direct result, NFSA replaced cage surveillance of crayfish plague with eDNA monitoring, contributing to the 3Rs (replacement, reduction, refinement; https ://www.nc3rs.org.uk/the-3rs) and improved animal welfare. From 2018, NEA has also implemented eDNA monitoring of noble crayfish and signal crayfish as a supplement to the traditional CPUE surveillance, which also increases the number of surveyed watercourses. As there is no cure for crayfish plague, it is essential to minimise the risk of spreading the pathogen to new areas. Since A. astaci is a notifiable disease in Norway, national legislation demands monitoring measures and control strategies to reduce the risk of further spread. Other countries in Europe may also choose to monitor crayfish plague, since this is also an OiE-listed, notifiable disease (OiE, 2017). Mitigation strategies in Norway include area restrictions, prohibiting crayfish trapping, increasing public awareness and mandatory disinfection of equipment. We advocate the use of the presented approach for early warning and targeted surveillance of non-indigenous crayfish species and crayfish plague in natural habitats, and for determination of the magnitude of an outbreak. It can also be used for improved conservation of indigenous crayfish, for example for assessing habitat status for crayfish restocking purposes or selection of Ark sites (Nightingale et al., 2017).
One of the primary benefits of eDNA monitoring in aquatic environments is the possibility for temporal and spatial monitoring of several organisms from the same eDNA samples. This approach is highly relevant for the study of other host-carrier-pathogen groups in marine and freshwater environments (Bass, Stentiford, Littlewood, & Hartikainen, 2015;. Additionally, recurrent sampling and long-time storage (e.g. biobank) of eDNA samples gives the possibility for retrospective analysis for other species of interest or even whole communities using environmental metabarcoding (Deiner et al., 2017). Environmental metabarcoding might even reveal emerging pathogens and/or invasive species that would go undetected unless specifically screened for, and could identify the causative agents for declines in other indigenous species. In the near future, technological advances will propel the eDNA monitoring concept forward, maturing from manually sampled eDNA snapshots to automated and continuous eDNA monitoring in real time.

DATA ACC E S S I B I L I T Y
Data available via the Dryad Digital Repository https ://doi. org/10.5061/dryad.vf86jb2 .