Parasites, pathogens and commensals in the “low-impact” non-native amphipod host Gammarus roeselii

Background Whilst vastly understudied, pathogens of non-native species (NNS) are increasingly recognised as important threats to native wildlife. This study builds upon recent recommendations for improved screening for pathogens in NNS by focusing on populations of Gammarus roeselii in Chojna, north-western Poland. At this location, and in other parts of continental Europe, G. roeselii is considered a well-established and relatively ‘low-impact’ invader, with little understanding about its underlying pathogen profile and even less on potential spill-over of these pathogens to native species. Results Using a combination of histological, ultrastructural and phylogenetic approaches, we define a pathogen profile for non-native populations of G. roeselii in Poland. This profile comprised acanthocephalans (Polymorphus minutus Goese, 1782 and Pomphorhynchus sp.), digenean trematodes, commensal rotifers, commensal and parasitic ciliated protists, gregarines, microsporidia, a putative rickettsia-like organism, filamentous bacteria and two viral pathogens, the majority of which are previously unknown to science. To demonstrate potential for such pathogenic risks to be characterised from a taxonomic perspective, one of the pathogens, a novel microsporidian, is described based upon its pathology, developmental cycle and SSU rRNA gene phylogeny. The novel microsporidian Cucumispora roeselii n. sp. displayed closest morphological and phylogenetic similarity to two previously described taxa, Cucumispora dikerogammari (Ovcharenko & Kurandina, 1987), and Cucumispora ornata Bojko, Dunn, Stebbing, Ross, Kerr & Stentiford, 2015. Conclusions In addition to our discovery extending the host range for the genus Cucumispora Ovcharenko, Bacela, Wilkinson, Ironside, Rigaud & Wattier, 2010 outside of the amphipod host genus Dikerogammarus Stebbing, we reveal significant potential for the co-transfer of (previously unknown) pathogens alongside this host when invading novel locations. This study highlights the importance of pre-invasion screening of low-impact NNS and, provides a means to document and potentially mitigate the additional risks posed by previously unknown pathogens. Electronic supplementary material The online version of this article (doi:10.1186/s13071-017-2108-6) contains supplementary material, which is available to authorized users.


Background
Understanding and interpreting the role played by pathogens in the invasion mechanisms of their hosts is becoming increasingly important as legislative pressure is placed upon managers to prevent and control wildlife disease [1,2]. Often, the pathogens of invasive hosts are little known or cryptic, requiring dedicated screening efforts to elucidate underlying parasites and pathogens that may be vectored to new habitats by non-native species (NNS) [2,3].
The Amphipoda constitute a diverse crustacean group with many species displaying invasive characteristics that have spread throughout Europe via invasion corridors [4]. Poland is considered part of one such invasion corridor connecting the Ponto-Caspian region to western Europe [4,5], making it an important study site for both recipient and donor populations of amphipods destined to reach other parts of Europe. Most non-native amphipod taxa found in Poland originate from the Ponto-Caspian region; however some exceptions exist. One example includes Gammarus roeselii Linnaeus of Balkan origin and documented to have invaded western Europe (including Poland, Italy, France and Germany) over a century ago, with a relatively low impact [6][7][8][9][10]. This species continues to extend its non-native range, now encompassing the Apennine Peninsula [11]. Although the host per se is considered a low impact NNS [12], current risk assessments associated with its spread do not take account of its underlying pathogen profile, nor the effect of these pathogens on receiving hosts and habitats. Several parasites and pathogens of Gammarus roeselii are known, including the acanthocephalans Polymorphus minutus (Zeder, 1800) [13]; Pomphorhynchus laevis (Zoega in Müller, 1776) [14] and Pomphorhynchus tereticollis (Rudolphi, 1809) [15]; and the microsporidians Dictyocoela muelleri Terry, Smith, Sharpe, Rigaud, Timothy & Littlewood, 2004 (unofficial genus) [16]; Dictyocoela roeselii Terry, Smith, Sharpe, Rigaud, Timothy & Littlewood, 2004 (unofficial genus) [16]; Nosema granulosis Terry, Smith, Bouchon, Rigaud, Duncanson, Sharpe & Dunn, 1999 [16]; and several Microsporidium spp. [17,18] (see Table 1).
Acanthocephalan parasites have been observed to cause various behavioural [14], physiological [19] and biochemical changes [20] on their amphipod host, which could alter their host's invasive capability. Some of the microsporidians infecting G. roeselii (Table 1) are taxa previously associated with other invasive amphipod hosts [17,21,22]. Some unassigned 'Microsporidium' spp. infecting G. roeselii may in fact reside within the genus Cucumispora Ovcharenko, Bacela, Wilkinson, Ironside, Rigaud & Wattier, 2010 [23]. This genus currently contains two species isolated from invasive amphipods: Cucumispora dikerogammari (Ovcharenko & Kurandina, 1987) and Cucumispora ornata Bojko, Dunn, Stebbing, Ross, Kerr & Stentiford, 2015. Like their hosts, existing members of the genus Cucumispora may also be of Ponto-Caspian origin due to their identification within tissues of Dikerogammarus spp. native to that region [23]. However, the detection of Cucumispora-like sequences (based upon PCR diagnostics and sequencing) in non-native G. roeselii originating from the Balkans, suggests that microsporidia belonging to the Cucumispora may have a range extending further than the Ponto-Caspian region depending on whether G. roeselii is a co-evolved host [17]. Cucumispora spp. have been associated with a variable host range, inferring there is a possibility for transmission from Ponto-Caspian invaders; concluding that Cucumispora spp. are likely emerging diseases among amphipods [24]. Pomphorhynchus tereticollis Denmark DNA sequence and visual [15] Pomphorhynchus laevis France Visual [14] Microsporidia Dictyocoela muelleri France DNA sequence [16] Dictyocoela roeselii France DNA sequence [16] Nosema granulosis France DNA sequence [16] Microsporidium sp. G Germany DNA sequence [17] Microsporidium sp. 505 Germany DNA sequence [17] Microsporidium sp. nov. RR2 Germany DNA sequence [17] Microsporidium sp. nov. RR1 Germany DNA sequence [17] Microsporidium sp. group F Germany DNA sequence [18] Microsporidium sp. group E Germany DNA sequence [18] Microsporidium sp. 2 Germany DNA sequence [18] In order to understand the pathogen profile of a low-impact non-native species and assess the risk of pathogen introduction from such an invader, we surveyed a population of G. roeselii in north-western Poland with an aim to understand which pathogen groups were present, whether the pathogen profile of a low-impact invader was different from that of highimpact invaders, and whether these pathogens pose a significant threat to native wildlife. We present the outcome of this survey here as the first comprehensive pathogen survey of G. roeselii. Using a combination of field sampling, histology, transmission electron microscopy and molecular diagnostics, we define an array of novel pathogens associated with this host and taxonomically define a new member of the microsporidian genus Cucumispora infecting G. roeselii. We discuss these results relative to the impact of these pathogens on population success and impact in Poland, their potential risk of transfer with further spread of this host across Europe and the importance of screening low-impact, NNS for pathogens without simply focussing on screening high-impact invasive hosts.

Collection, dissection and fixation of Gammarus roeselii
Gammarus roeselii were sampled using standard hydrobiological nets and kick-sampling from the banks of a stream in Chojna, north-western Poland (Oder river catchment) (N52.966, E14.42906) on 23/06/2015. A total of 156 specimens were collected: 8 were fully dissected to remove muscle and hepatopancreas to fix for histology (Davidson's freshwater fixative), transmission electron microscopy (TEM) (2.5% glutaraldehyde) and molecular diagnostics (96% ethanol), and 148 were injected on site with fixative for histological screening. Carcasses in fixative, or live animals, were transported to Łόdź University, Poland for storage and/or dissection.

Histopathology and transmission electron microscopy
Specimens preserved in Davidson's freshwater fixative were transferred to 70% methylated spirit after 24-48 h and infiltrated with paraffin wax using an automated tissue processor (Peloris, Leica Microsystems, Milton Keynes, UK). Wax embedded tissues were then sectioned sagittally a single time on a Finesse E/NE rotary microtome (Thermofisher, Hemel Hempstead, UK) (3-4 μm thickness). Sections were glass mounted and stained using haematoxylin and alcoholic eosin (H&E) and examined using a Nikon Eclipse E800 light microscope. Images were captured using an integrated LEICA™ (Leica, Milton Keynes, UK) camera.
Sample preparation and observation via transmission electron microscopy (TEM) for muscle and hepatopancreas tissues dissected from G. roeselii followed that used by Bojko et al. [22].

Molecular diagnostics
Muscle tissue dissected from a single infected G. roeselii was confirmed positive for microsporidiosis via visual, histological and TEM diagnostics. Muscle tissue from the same individual was fixed in ethanol upon dissection and used for DNA extraction. DNA extraction was performed using a standard phenol-chloroform method. SSU rRNA gene amplification was performed using the primers MF1 (5′-CCG GAG AGG GAG CCT GAG A-3′) and MR1 (5′-GAC GGG CGG TGT GTA CAA A-3′) [25] and 2.5 μl of DNA template (~30 ng/μl) in a GoTaq flexi PCR reaction (per reaction: 10 pM of each primer; 0.25 M of each dNTP; 1.25 U Taq Polymerase; 2.5 mM MgCl 2 ) in a total volume of 50 μl. T c settings were: 94°C (5 min), 94°C -60°C -72°C (each 1 min; 35 cycles), 72°C (10 min). Amplicons were observed using gel electrophoresis on a 2% agarose gel (30 min/120 V) producing a microsporidian band at~800 bp. This band was excised and purified for forward and reverse sequencing via Eurofins genomics barcode-based sequencing service (Eurofinsgenomics, UK).

Phylogenetics and sequence analysis
The final SSU rRNA gene sequence for this microsporidian was 825 bp sequence length, which was placed into BLASTn (NCBI) to retrieve identical or close hits. The sequence was placed alongside several SSU rRNA gene sequences used by Ovcharenko et al. [23] to form the initial description of Cucumispora dikerogammari (GQ246188.1), as well as some closely linked, recently described microsporidian sequences , and all available partial or complete sequences from BLAST that link with close similarity to C. dikerogammari (GQ246188.1) and could potentially be candidates for the genus Cucumispora.
The sequences were aligned with MAFFT 7.017 [26] using default values, in Geneious 6.1.8 [27]. The phylogeny reconstruction was performed in MEGA 7 [28] using the maximum-likelihood [29] and Neighbour-Joining [30] methods. Clade credibility was assessed using bootstrap tests with 1,000 replicates [31]. The T92 model of evolution with gamma-distributed rate heterogeneity (G) was selected for the dataset using the complete deletion model selection algorithm implemented in MEGA 7. Clade IV microsporidian species were used as the outgroup to root the tree.
The carapace and appendages of G. roeselii were often coated with stalked ciliates and epibiotic rotifers (Fig. 1a), however the gills and brood pouch were commonly associated with all epibiotic commensals. None of the epibiotic commensals induced an immune response from the host and were common throughout the G. roeselii population ( Table 2).
A single animal was observed with a ciliated protist infection in the haemolymph, with accumulations of the parasite in the antennal gland, gills ( Fig. 1d), heart and appendages. No immune response toward the parasitic protist was noted throughout the histological screen.
Gregarines (Apicomplexa) were commonly associated with the gut (50% prevalence) (Fig. 1e) and less frequently, the hepatopancreatic tubules (< 1%). Gregarines were often seen in large numbers in the gut with both extracellular and intracellular developmental stages with occasional observation of syzygy. Gregarines elicited no apparent immune response from the host but were detected in significant numbers in the gut lumen.
A putative gut-epithelial virus was observed in 4 individuals where gut nuclei were present with an expanded, eosinophilic viroplasm, resulting in nuclear hypertrophy and marginated host chromatin (Fig. 1f ). No immune response was observed against this virus in the histology. In addition, a bacilliform virus was identified from the hepatopancreas of G. roeselii and is detailed below.
A putative RLO in the cytoplasm of hepatopancreatocytes was observed in a single individual (Fig. 1g). The cytoplasm of infected cells appeared dense, granular and purple in colour (H&E stain), a common feature of RLO infections in other hosts. Host nuclei were unaffected and no immune responses were observed in affected tissues.
Three metazoan parasites were observed infecting G. roeselii (see Table 2 for prevalence details). Digeneans were encysted in the gut, gonad and hepatopancreas (Fig. 1h). Large acanthocephalans such as Polymorphus minutus (Fig. 1i) and Pomphorhynchus sp. were often present in the same tissue types but never together in the same host. No helminth species elicited an immune response from the host.
Two microsporidian infections were observed during screening; the first in the hepatopancreas and the second in the muscle (the muscle infecting microsporidian is detailed below). The microsporidian from the hepatopancreas was observed in a single specimen fixed for histology, meaning that no ethanol or glutaraldehyde fixed materials were taken, resulting in a lack of information for full taxonomic analysis for this species. This microsporidian was present only in the hepatopancreas; specifically, in the cytoplasm of infected cells where several development stages could be seen in histological section (Fig. 1j). No immune response was observed against this microsporidian; however, disintegration of infected tubules was observed.

Gammarus roeselii bacilliform virus: histopathology and TEM
A novel virus infecting the nuclei of hepatopancreatocytes was observed using histology and TEM. Histologically, the virus was present only in the nuclei of infected hepatopancreatocytes and caused host chromatin margination and nuclear hypertrophy due to an expanded viroplasm. Uninfected cell nuclei showed normal chromatin configuration without expanded viroplasm (Fig. 2a  inset). This viral pathology was present in 12.2% of specimens (Fig. 2a).
TEM of an infected hepatopancreas tubule and associated cells revealed a viroplasm consisting of large bacilliform virus particles in the host cell nucleus (Fig. 2b). Virions were rod-shaped and consisted of an electron dense, cylindrical core (length 177.4 ± 18 nm, width 35.9 ± 6 nm) and, were surrounded by a single membrane (length 224.0 ± 17 nm, width 70.0 ± 13 nm) (Fig. 2c). Currently no genetic data are available for this virus. This novel virus is termed Gammarus roeselii Bacilliform Virus (GrBV) until further data can be acquired, to allow for taxonomic identification.

Microsporidian life-cycle and ultrastructure
Ultrastructurally, the developmental cycle of the microsporidian in G. roeselii resembled that of C. dikerogammari described by Ovcharenko et al. [23] and C. ornata described by Bojko et al. [22]. Infected muscle fibres contained tightly packed merogonial and sporogonial life stages, which developed in direct contact with the host muscle cytoplasm; often in the sarcolemmal space. The microsporidian development began with a diplokaryotic meront (2n) bound by a thin cell membrane (Fig. 4a). Nuclear division of the diplokaryotic meront formed a tetranucleate merogonal plasmodium (2 × 2n) present with a string of four nuclei separated by a thin membrane (Fig. 4b). The tetranucleate meront plasmodium can show early thickening of the cell membrane (Fig. 4b) prior to its division to form two diplokaryotic sporonts (2n), which show further thickening of the cell membrane prior to any formation of spore extrusion apparatus (Fig. 4c-d). Later stage sporonts developed an electron dense cytoplasm prior to formation of early spore extrusion apparatus (Fig. 4e). The maturing sporoblast became electron dense and cucumiform in shape, with an early anchoring disk and coiled, irregularshaped, polar filament in cross-section (Fig. 4f ). The condensed sporoblast displayed the earliest development of an electron lucent endospore (Fig. 4f ) and became increasingly turgid during spore maturation (to presume an oval shape) (Fig. 5a-b). Further thickening of the electron-lucent endospore, circularisation of the polar filament cross-sections and development of spore organelles, such as the polaroplast and polar vacuole, occurred in the late sporoblast ( Fig. 5a-b). At this stage, the exospores resumed an irregular surface (most clearly seen in the image of the final spore, Fig. 5c). The final diplokaryotic spore was 2.2 ± 0.1 μm in length (n = 30) and 1.5 ± 0.1 μm in width (n = 30), contained an anchoring disk, bi-laminar polaroplast, 9-10 turns of the polar filament [cross-sectional diameter: 92 ± 13 nm (n = 30)] with rings of proteins at varying electron density, thickened spore wall (plasmalemma, endospore, exospore) and a ribosome-rich, electrondense cytoplasm (Fig. 5c). The spore wall was of variable thickness according to location; thinnest at the terminal point of the anchoring disk (40 ± 6 nm) and thicker elsewhere (up to 185 ± 50 nm).

Microsporidian phylogeny
The amplicon derived from the microsporidian infecting the musculature of G. roeselii provided an 825 bp sequence of the SSU rRNA gene. This sequence showed closest similarity to Microsporidium sp. 1049 (FN434092.1: 98% similarity; query cover: 99%; e-value = 0.0) a microsporidian isolated from Gammarus duebeni duebeni from Dunstaffnage Castle (Scotland, UK), and Microsporidium sp. MSCLHCY01 (HM800853.2: 96% similarity; query cover: 96%; e-value = 0.0) a microsporidian isolated from the copepod Lepeophtheirus hospitalis, parasitizing the starry flounder, Platichthys stellatus, from British Colombia, Canada. The closest named species were Cucumispora ornata (KR190602.1: 95% similarity; query cover: 99%; evalue = 0.0), a microsporidian pathogen isolated from the invasive demon shrimp, Dikerogammarus haemobaphes Eichwald, from the Carlton Brook invasion site, UK, and Cucumispora dikerogammari (GQ246188.1: 93% similarity; query cover: 96%; e-value = 0.0), a microsporidian isolated from the killer shrimp, Dikerogammarus villosus Sowinsky, from an invasion site in France. Several microsporidian SSU sequences show high similarity (~90-100%) to those Fig. 4 Transmission electron micrograph of early spore development for Cucumispora roeselii n. sp. a Diplokaryotic meront displaying attached nuclei (N; white arrow). Note the thin cell membrane (black arrow). b Tetranucleate cell displaying four attached nuclei (N; white arrows) with a thickening cell wall (black arrow). c After division, two early diplokaryotic (N; white arrow) sporoblasts are produced with further cell membrane thickening (black arrow). d Early diplokaryotic (N; white arrow) sporoblast displaying further thickening of the cell membrane (black arrow). e The early sporoblast begins to become electron dense and condense with some early development of spore organelles such as the polar filament (black arrow). f Fully condensed sporoblast development stage present with electron dense cytoplasm and coiled polar filament (PF) and anchoring disk (AD). At this stage the formation of the early endospore is visible (white arrow). Scale-bars: 500 nm corresponding to the genus Cucumispora and are included in Additional file 1: Table S1, depicting their host and geographical origin.

Dikerogammarus villosus
DvBV [3] Gammarus roeselii GrBV Present study closely associated with H. aquatica at 95-99%) on the tree, despite the overall sequence similarity (96%) (Fig. 6).  Etymology: The specific epithet "roeselii" is derived from the host species, which refers to the thorns down the back of the animal that resemble those of a rose (Rosoideae). It also holds an additional meaning, referring to the "thorned" appearance of the spore wall in this new microsporidian species.

Description
Ultrastructurally, spores appear oval (length 2.2 ± 0.1 μm; width 1.5 ± 0.1 μm), with a "thorned" spore wall consisting of an electron lucent endospore and electron dense exospore at varying thicknesses either around the spore (138 ± 27 nm), at the point of the anchoring disk (40 ± 6 nm), or at the periphery of the anchoring disk (185 ± 50 nm). The polar filament turns between 9-10 times around the centre and posterior of the spore. This parasite is diplokaryotic throughout its life-cycle. Similarity of the SSU rDNA sequence to the type species C. dikerogammari was 93%.
Transmission information is currently unavailable but predicted to be horizontal as derived from the pathology; no infection of the gonad was observed.

Discussion
This study presents the first comprehensive pathogen screen of the non-native gammarid, G. roeselii, outside of its native range and includes a taxonomic description of a novel species of microsporidian belonging to the genus Cucumispora. The novel microsporidian is named herein as Cucumispora roeselii n. sp. Studies such as this one are important to advise risk assessment criteria for invasive and non-native species, specifically in the light of absent information on the pathogens and parasites of invasive and non-native species [2]. While G. roeselii has previously been considered as a low-impact invader, in this case we identify G. roeselii as a potentially highprofile invader because of its status as a pathogen carrier, transferring pathogens along its route of introduction and spread. It is important to consider if these pathogens could transmit to native wildlife, if they act as a regulator for the host species; limiting its potential impact when present, or if they could be used against the invader in a targeted biological control approach.
Gammarus roeselii is not of Ponto-Caspian origin or part of the genus Dikerogammarus, as the hosts of both previously described Cucumispora spp. [22,23]. Cucumispora dikerogammari and C. ornata are both thought to originate in the native range of their hosts. However the inclusion of C. roeselii n. sp. in this genus requires reconsideration of the origins and range of Cucumispora spp. Were this parasite to have originated from the hosts native range (The Balkans) it could indicate an interesting phylogeographic spread of microsporidia within this genus. There is a possibility that this parasite has been acquired from the Polish environment, and/or from other invaders.
Several genetic isolates provide strong sequence similarity to members of the Cucumispora [17,[21][22][23][33][34][35][36], Unpublished works through BLASTn] (Additional file 1: Table S1; Fig. 6). The ranges of these sequenced parasite isolates belong mainly to European territories, but some studies demonstrate isolates from Caribbean and Canadian waters [34,36]. This information suggests that members of the genus Cucumispora may be present around the globe, and their recent identification further suggests their role as emergent pathogens, not only in gammarids but in copepods as well [36]. However, recently published information suggests that hyperparasitic microsporidia with the capability to infect protists appear to have similar SSU sequences to the Cucumispora and have been placed into the recently erected genus Hyperspora [37]. Until further information is provided in the form of legitimate taxonomic descriptions from more of the SSU isolates in Fig. 6, the native/invasive range and host range of many potential Cucumispora spp. remains an interesting phenomenon.
Some isolates show close relatedness to taxonomically described Cucumispora spp. (Fig. 6). Microsporidium sp. G (haplotypes 1, 2, 3 and 4) isolated from D. haemobaphes (Germany) is 99% similar to Cucumispora ornata and clades closely in the tree presented in Fig. 6. It is likely these are the same parasite and should be synonymised [17]. However, determining a taxonomic basis on a single gene does not propagate a strong scientific standing and histological and TEM evidence for Microsporidium sp. G from both D. haemobaphes and G. roeselii should be confirmed in each host before amalgamating.

Microbial associations and invasion biology of Gammarus roeselii
Several pathogens, parasites and commensals were identified histologically as part of this study. Polymorphus minutus and Pomphorhynchus sp. represent two known acanthocephalan parasites of G. roeselii (Table 1) also observed in this sample from Chojna. Epibiotic rotifers, ciliated protists and filamentous bacteria are commonly associated with aquatic species [3,38] as are gut dwelling gregarines in amphipod hosts [3,39].
Digenean associations with amphipods are also common and several are known to utilise amphipods as intermediate hosts before entering further hosts where they can reach sexual maturity [40]. Digeneans detected in this study were of an undetermined species (possibly multiple species) and its/their life-cycle and reason for parasitizing G. roeselii is currently unknown.
The parasitic ciliated protist (Fig. 1d) has not been noted from G. roeselii in the past and is likely a novel association for this species. Without DNA sequence data it is uncertain whether this parasite is taxonomically novel or not. Parasitic ciliates have been noted in amphipods in the past, such as Fusiforma themisticola Chantangsi, Lynn, Rueckert, Prokopowicz, Panha & Leander, 2013, which parasitizes Themisto libellula (Lichtenstein in Mandt) [41].
A second microsporidian association in this study was of a rare parasite (<1% prevalence) targeting the hepatopancreas of G. roeselii. Most microsporidia that target the hepatopancreas of crustaceans fall into the 'Clade IV' of microsporidian taxonomy (Terresporidia) [42] and further, into the Hepatosporidae [43,44]. Obtaining TEM and SSU sequence data would help to taxonomically identify this species. A recent study by Grabner et al. [17] revealed two microsporidian SSU sequences, isolated from G. roeselii, that correspond to microsporidia from Group IV (Terresporidia); the histopathology presented by this study may link to one of these isolates and further tests should be carried out to confirm this and identify the species taxonomically.
A single observation of a putative RLO in the cytoplasm of infected hepatopancreatocytes is an interesting association as few RLOs have been noted from amphipods in the past. To date, the only examples include putative Rickettsiella-like SSU rDNA sequences available from BLASTn (NCBI) and systemic haemolymph infections caused by RLOs in Gammarus pulex (L.) [45] and Crangonyx floridanus Bousfield [46].

Viruses in the Amphipoda
A variety of viruses have been identified from Crustacea either morphologically, via DNA sequence data or through searching for endogenous viral elements in the genome of crustacean hosts [47][48][49]. Few have ever been identified from hosts belonging to the Order Amphipoda. To date only three published viral associations have been made from amphipods: the first is in the form of histology and TEM images of a bacilliform virus from the hepatopancreas of Dikerogammarus villosus and referred to as Dikerogammarus villosus Bacilliform Virus (DvBV) [3]; the second, an unassigned circovirus from a Gammarus sp. [50]; and the third includes various circular-virus associations to Diporeia spp. [51].
Although DvBV was, previous to this study, the only visually confirmed virus from an amphipod, bacilliform viruses from the hepatopancreas of crustaceans are common and several have been identified morphologically (Table 3). GrBV, isolated from the hepatopancreas of G. roeselii in this study, fits morphologically and pathologically alongside the viruses in Table 3. Penaeus monodon nudivirus (PmNV) has been the focus of genome sequencing efforts, revealing that this group of morphologically-similar viruses are likely nudiviruses (Nudiviridae) [52]. Further genome sequencing and generalised primer-designs for nudivirus genes would benefit this area greatly and allow further taxonomic insight into the viral life history.
The viral pathology in the gut of G. roeselii remains putative due to a lack of appropriately fixed material to observe virions via TEM. Pathologically, the presence of the infection (nuclei of gut epithelia) suggests a DNA virus. It is uncertain at this point whether this infection is caused by GrBV simply infecting a separate tissue type; this cannot be diagnosed using our current data and materials. Re-sampling and TEM processing should provide informative data, however genetic data would be most beneficial; a valid point for many of the viruses in Table 3.
Cucumispora roeselii n. sp.: invasion threat or beneficial for control?
Although the prospect of invaders carrying pathogens pose a potential problem [1,53], in some instances parasites can act as controlling agents [54]. This phenomenon may be taking place with the D. haemobaphes invasion of the UK, where the microsporidian pathogen, C. ornata, may be limiting the health of the invasive population [22]. Amphipod populations without their microsporidian pathogens are not regulated as they would be in their native range, and loss of their "enemies" may result in greater fitness and a higher impact on the environment; such as that observed with the killer shrimp at invasion sites in the UK [3,55].
Gammarus roeselii is considered to be a low impact non-native species [12] in freshwater systems across Europe [6,[8][9][10]12]. However, this non-native host may not be the main issue but instead its pathogens could act as "biological weapons" to facilitate invasion and harm wildlife [1,2,53]. The concept of being a pathogen carrier is often ignored in risk assessment, often due to a lack of information around the capability to accurately assess the risk invasive pathogens pose [2]. Possible parasite transmission from G. roeselii to native fauna is high; this is based on the large diversity of parasites and pathogens observed by this study. Due to limited records, it is difficult to be certain which pathogens and parasites are from the native range of G. roeselii and which have been acquired during its introduction and spread. Assessment of co-evolved pathogens in the native range of G. roeselii would increase our understanding of the origins of C. roeselii n. sp. and the other pathogens observed during this study. Examples of enemy release in gammarids are available, including: the loss of pathogens during the introduction process [3] and of gammarids carrying pathogens into novel invasion sites [22,35].
It may be possible that the pathogens identified as part of this study regulate the host species, and escape from these regulators could increase the impact and risk of G. roeselii. Understanding the associated mortality rate, host range, behavioural alterations and physiological changes these pathogens impose upon their host would allow further assessment of whether these pathogens are regulating non-native G. roeselii populations in Chojna and elsewhere within Europe. Information gleaned from such studies could define whether C. roeselii n. sp., and other pathogens associated with G. roeselii, could be useful as biocontrol agents, or if they are emerging diseases and detrimental for vulnerable wildlife.

Conclusions
This study has identified several pathogens and parasites, which utilise G. roeselii as their host; including a novel species description of a microsporidian parasite. These pathogens could pose a significant threat to native wildlife. This example study displays the importance of screening non-native, low impact invaders for pathogens to identify their potential to carry and transmit wildlife disease to native fauna and flora. Disease profiling should be factored into the risk assessment of invasive and non-native species and current assessment should not rely on host-focussed studies alone.

Additional file
Additional file 1: Table S1. Geographical and host data for those microsporidian gene isolates that clade within the "Cucumispora candidates" group in Fig. 6. (DOCX 20 kb) Abbreviations NNS: Non-native species; SSU: Small subunit ribosomal RNA