High parasite diversity in the amphipod Gammarus lacustris in a subarctic lake

Abstract Amphipods are often key species in aquatic food webs due to their functional roles in the ecosystem and as intermediate hosts for trophically transmitted parasites. Amphipods can also host many parasite species, yet few studies address the entire parasite community of a gammarid population, precluding a more dynamic understanding of the food web. We set out to identify and quantify the parasite community of Gammarus lacustris to understand the contributions of the amphipod and its parasites to the Takvatn food web. We identified seven parasite taxa: a direct life cycle gregarine, Rotundula sp., and larval stages of two digenean trematode genera, two cestodes, one nematode, and one acanthocephalan. The larval parasites use either birds or fishes as final hosts. Bird parasites predominated, with trematode Plagiorchis sp. having the highest prevalence (69%) and mean abundance (2.7). Fish parasites were also common, including trematodes Crepidostomum spp., nematode Cystidicola farionis, and cestode Cyathocephalus truncatus (prevalences 13, 6, and 3%, respectively). Five parasites depend entirely on G. lacustris to complete their life cycle. At least 11.4% of the overall parasite diversity in the lake was dependent on G. lacustris, and 16% of the helminth diversity required or used the amphipod in their life cycles. These dependencies reveal that in addition to being a key prey item in subarctic lakes, G. lacustris is also an important host for maintaining parasite diversity in such ecosystems.

Gammarus species are omnivorous and function mainly as shredders, processing large organic matter (detritus) into smaller pieces that are consumed by other macroinvertebrates (Kelly et al., 2002;MacNeil et al., 1997). Gammarids are further significant prey items for fishes and are also eaten by birds and other vertebrates and invertebrates (reviewed in MacNeil et al., 1999), making them a highly connected species and thus a hub in the food web. Taken altogether, Gammarus spp. play an important role in aquatic ecosystems by contributing substantial biomass and through their roles as predator and prey (Kelly et al., 2002;MacNeil et al., 1997MacNeil et al., , 1999. Gammarus spp. are often intermediate hosts for diverse parasites (Bojko & Ovcharenko, 2019;Denny, 1969), which is likely due to their high connectance in food webs and functional roles in ecosystems (based on Locke et al., 2014). Parasites can alter food web dynamics by making intermediate hosts more susceptible to predation by final hosts (Lafferty & Shaw, 2013;Thomas et al., 1999). For example, acanthocephalans, cestodes, and trematodes can alter the phenotype and/or behavior of their gammarid intermediate host, often with a result of increased predation by final host birds or fishes (Bakker et al., 1997;Helluy & Thomas, 2003;Hindsbo, 1972;Knudsen et al., 2001). Parasites also affect the feeding ecology of gammarids by decreasing their shredding activity (Dianne et al., 2014;Labaude et al., 2017;Medoc & Beisel, 2011). Hence, parasitism of amphipods can alter food web dynamics through bottom-up (increased predation on these amphipods) and top-down (altered feeding ecology) mechanisms. A single host is typically infected by multiple parasite species that form a dynamic assemblage (Holmes & Price, 1986;Poulin, 2014), and gammarids are no exception. Coinfections with two manipulative parasites further complicate parasite-induced behavioral changes in gammarids (Cezilly et al., 2014;Haine et al., 2005). The ecosystem effects of parasitism may be particularly profound if the intermediate host is a key species or hub in the aquatic food web (Sures et al., 2017), and gammarids often play such a role in lacustrine ecosystems. To fully understand the ecological role of parasites at the ecosystem level in Takvatn, it is necessary to evaluate the entire parasite assemblage of a gammarid population.
Gammarus lacustris is one of the most widespread freshwater amphipod species, with a nearly circumpolar boreal distribution (Väinölä et al., 2008), yet its parasite assemblage has been examined in only a few studies (see the review by Bojko & Ovcharenko, 2019). Sokolov and Gordeev (2014) reported four larval helminths infecting G. lacustris in Kamchatka, Russia. Bojko (2017) found larval trematodes, acanthocephalans, and protozoans in populations of G. roeselii in Poland. Denny (1969) conducted a comprehensive study describing the metazoan parasite community of G. lacustris in a eutrophic Canadian lake. He found 12 species of helminths: eight cyclophyllidean cestodes, one nematode, and three acanthocephalans. The final hosts for all parasites were various bird species, as the lake harbors no permanent fish population. In the Holarctic however, G. lacustris is important prey for many salmonid fishes (MacNeil et al., 1999) and serves as intermediate host to several parasites that use salmonids as final hosts (Knudsen et al., 2008;Kuhn et al., 2016). In Lake Takvatn, northern Norway, G. lacustris is the only the amphipod in the system (Klemetsen & Elliott, 2010) and accounts for over 50% of the diet of Arctic charr (Salvelinus alpinus) between autumn and early winter (Prati et al., 2020), providing a key source of nutrition at a time when other prey items such as insects and zooplankton are less abundant. Hence, in subarctic lakes where fishes are common, like in Takvatn, such trophically transmitted parasites should be reflected in the assemblages of both G. lacustris and fish hosts.
Here, we investigate the diversity, prevalence, and abundance of parasites infecting G. lacustris in Lake Takvatn. Although arctic and subarctic ecosystems are generally species poor (Hoberg et al., 2012), which suggests a low diversity of parasites (Hechinger & Lafferty, 2005), recent studies show a surprisingly high taxonomic diversity of trematode parasites in Takvatn (Soldánová et al., 2017) and Icelandic lakes (Blasco-Costa et al., 2014). Three decades of comprehensive studies of the parasite communities of snails and fishes in Takvatn indicate that G. lacustris should harbor parasites that use both fishes and birds as final hosts, owing to the lack of other potential final host taxa in the ecosystem (Amundsen et al., 2009(Amundsen et al., , 2019Knudsen et al., 1999). The main objectives of the present study were to: i) describe the parasite community of G. lacustris in a subarctic lake, and ii) describe and contrast the abundance of parasites that use fishes and birds as final hosts.

| Study area and collection
Takvatn (69°07′N, 19°05′E) is a subarctic, oligotrophic, and dimictic lake in northern Norway that has been the focus of intensive ecological and food web studies for more than 30 years (details in Amundsen et al., 2009Amundsen et al., , 2013Amundsen et al., , 2019. The lake is situated 214 m above sea level with a surface area of 15 km 2 and a maximum depth of ca 80 m. There is little macrovegetation in the lake, but the littoral zone (3-10 m depth) has dense beds of the grass-like macroalgae Nitella sp., which contain the highest abundances of G. lacustris .
We sampled gammarids in the littoral zone (0-8 m depth) by dragging a benthic sled along Nitella sp. beds as described in Knudsen et al. (2001). Gammarids were collected from each haul, placed in buckets with lake water and vegetation, and brought back to the lab for dissection within 48 hr. Individuals not dissected within 24 hr were kept cool overnight in the refrigerator or outdoors (at approx. 4-8°C). To obtain a broader range of parasite diversity in G.
lacustris in the lake, gammarids were collected from five sites (L1-L5), including two in the vicinity of an important nesting area for birds (L4 and L5; . Sampling was carried out over three years during different times of the ice-free period (August and October 2012, June and September 2013, and August 2015).
Not all sites were sampled every year.

| Dissection and parasite identification
We blotted 474 amphipods on paper towels, measured length (eye to end of telson; mm) and wet weight (g). Due to potential variation in length measurements, we generated a length-weight regression from a subsample (y = 132.17x + 5.62, R 2 = 0.95) and used weight-based estimates of length for all analyses. We compressed whole gammarids between glass plates (150 mm × 100 mm × 3.5 mm) and examined them under a stereo microscope (Leica Wild M3, maximum magnification of 40×). Parasites were counted and transferred for further inspection under a compound microscope if needed. Parasites were identified to the nearest taxonomic level based on morphology, and select specimens were preserved in 95% ethanol for genetic analysis in a separate study (trematodes only; details in Soldánová et al., 2017) or formalin for further identification (all other parasites).

| Statistical analyses
We characterized the parasitism in G. lacustris samples by calculating prevalence, mean intensity, and mean abundance (defined in Bush et al., 1997) and assessed parasite infracommunity composition using the 7-set Venn diagram "Adelaide" (Dusa, 2020). We investigated whether the infections of parasites with indirect life cycles varied between sampling locations and sampling period, using two analyses. To analyze if the abundance of Plagiorchis sp. differed between sampling locations and periods, we used a mixture model (zero-inflated negative bionomial generalized linear model (ZINB GLM); R (version 3.5.1; R Core Team, 2018), with G. lacustris size (length) as a covariate (Zuur et al., 2009). The ZINB GLM contains two parts; a negative binomial GLM that models parasite counts and a binomial GLM that models the probability of observing excess zeros above those of the count process (Zuur et al., 2009). Other parasite species were low in intensity so we used infection status (infected vs. uninfected) rather than abundance as the binomial response variable in logistic regressions with the same predictor variables (sampling location, period, and G. lacustris size).

| Parasitism of G. lacustris
We found seven parasite taxa in 474 G. lacustris, where 77% (N = 364) of the amphipods were infected with at least one parasite (Table 1)

| Abundances of bird and fish parasites
Plagiorchis sp. had the highest overall prevalence among bird parasites (Table 1). For fish parasites, Crepidostomum spp. were the most prevalent (Table 1) , and higher at a bird nesting site (L5; Table 2). The probability of infections with Crepidostomum spp. was higher at site L4 but also increased with host size and was higher in fall compared with late summer (Figure 3; Table 3). The probability of being infected with C. farionis also increased with host size (data not shown). For the other parasite species, however, there were no significant patterns between probability of infection and the predictors, probably due to their low prevalence in the G. lacustris population.

| D ISCUSS I ON
We found seven parasite taxa infecting G. lacustris in Takvatn   with other sites, the estimate of 7 species has 95% confidence limits between 6 and 8 species (Hsieh et al., 2016). Further, some "species" might represent species complexes, due to the potential for cryptic species-an increasingly common discovery among helminth para- lacustris, and other invertebrates in Takvatn, using samples that were collected during the present study, and revealed four genetically different species of Crepidostomum: C. metoecus, C. farionis, plus two new cryptic species (Crepidostomum sp. 1 and 2 in Soldánová et al. (2017)).

However, Crepidostomum farionis is reported in Gammarus pulex from
other northern systems (Awachie, 1968) and is likely able to infect G.
lacustris in Takvatn, as are the two other species (M. Soldánová, personal communication). Therefore, we decided to use "Crepidostomum spp." in the present paper rather than limit the identification to C.

| Bird parasites
The persistently high abundance and prevalence of Plagiorchis sp.
( Table 1) are likely related to the relatively rich and stable aquatic bird community , the putative final hosts for Plagiorchis sp. in Takvatn, and also the high abundance of its first intermediate snail host R. balthica (Klemetsen & Elliott, 2010) (Amundsen et al., 2009(Amundsen et al., , 2019Knudsen et al., 1999).
The acanthocephalan cystacanths were likely Polymorphus sp., based on morphological identification and published records in the region (Tomáš Scholz, Czech Academy of Sciences, personal communication). Prevalence in G. lacustris remained low across sites and years, which is in contrast to the Canadian study reporting 12.7% prevalence for Polymorphus marilis (Denny, 1969). However, since many acanthocephalan parasites alter the behavior of their Gammarus host by increasing their vulnerability to predation by bird final hosts (Bakker et al., 2017;Helluy & Thomas, 2010;Jacquin et al., 2014;Lagrue et al., 2013), their ecological significance can be high despite a low prevalence. The potential for behavior modification in the system may also mean that the low prevalence we observed could result from the preferential selection of infected amphipods by nonhost fish or bird predators (Knudsen et al., 2001;Lafferty, 1999;Song & Proctor, 2020;Ubeda et al., 1994).
The cyclophyllidean cysticercoid occurred in only two of the 474 G. lacustris specimens. Cyclophyllidean cysticercoids have been reported from gammarid amphipods, including Lateriporus teres and Microsomacanthus microsoma in G. lacustris (Nikolov et al., 2008), (Dezfuli et al., 2002), and two species of unidentified cysticercoids in Hyallela patagonica (Rauque & Semenas, 2013). The size and shape of the Takvatn cysticercoids and their rostellar hooks resemble M. microsoma more than L. teres, but their species identity has not been established.

Microsomacanthus pachycephala in Echinogammarus stammeri
The two gammarid specimens containing cysticeroids were both collected at a site with high bird abundances, which are the final hosts for most hymenolepidid cestodes, including those previously reported in gammarid amphipods (Dezfuli et al., 2002;Nikolov et al., 2008).

| Fish parasites
Crepidostomum spp. were the most common of the three fish parasite species (Table 1) and more prevalent at Takvatn than the 2% prevalence observed in G. lacustris from another subarctic lake (Sokolov & Gordeev, 2014). Adult Crepidostomum spp. frequently infect Arctic charr in Takvatn (Kuhn et al., 2016) and are common in other lakes with salmonids (Arctic charr and brown trout, Salmo trutta) in the region (Knudsen, 1995;Knudsen et al., 2008;Siwertsson et al., 2016). Takvatn, as the other potential final host taxa are absent (Amundsen et al., 2009(Amundsen et al., , 2019Knudsen et al., 1999). Interestingly, 38% of the Crepidostomum spp. metacercariae were progenetic. Progenesis would be an advantageous strategy for Crepidostomum spp. if predation by appropriate final hosts (fishes) was uncommon (as might occur at lakes without fishes). The pattern of Crepidostomum spp.
progenesis at Takvatn has yet to be explored, but Lagrue and Poulin (2009) report that Coitocaecum parvum, another allocreadiid trematode, becomes progenetic in the absence of chemical cues from the definitive host fish.
Prevalence of the nematode C. farionis was slightly higher than the 1%-4% reported previously (Knudsen et al., 1999). Other studies have observed equal (Awachie, 1973) or higher (Sokolov & Gordeev, 2014) prevalences of C. farionis in G. lacustris. However, in the nearby lake, Fjellfrøsvatn, within the same drainage as Takvatn, C. farionis is rare in G. lacustris (0.2%; Knudsen et al., 2001). Despite the relatively low prevalence in G. lacustris in both lakes, C. farionis is prevalent in Arctic charr in Takvatn and Fjellfrøsvatn (Knudsen et al., 2002(Knudsen et al., , 2004 perhaps because these worms are long-lived and accumulate over time in fish hosts (Moravec, 1994 Awachie, 1966;Dezfuli et al., 2000). Despite the moderate infection levels of C. truncatus in G. lacustris, the infection in the final salmonid hosts in this region can be very high (50%-70%; Amundsen et al., 2003), which has been shown to be a result of selective predation on infected amphipods (Franceschi et al., 2007;Knudsen et al., 2001).

Possible explanations for the variability in length could be that
Rotundula sp. at Takvatn is a different species than those previously described, or the small sample sizes in the published studies do not represent the full range of sizes found in those populations from immature trophs to mature gamonts (Tamara Cook, Sam Houston State University, personal communication); our study certainly did not comprehensively sample enough individuals to obtain a full picture of the Rotundula sp. population at Takvatn.

| CON CLUS ION
We found a high parasite diversity and high prevalences of several

DATA ACCE SS I B I LIT Y S TATE M E NT
All data supporting this study can be accessed from Dryad: https:// doi.org/10.25349/ D9B89T.

ACK N OWLED G M ENTS
We thank Laina Dalsbø and Karin Strand Johannessen for practical support during field collections. The study was supported by the Norwegian Research Council (NFR 213610) and UiT The Arctic University of Norway. Any use of trade, product, or firm names in this publication is for descriptive purposes only and does not imply endorsement by the U.S. Government.

CO N FLI C T O F I NTE R E S T
The authors declare that they have no competing interests. Per-Arne Amundsen https://orcid.org/0000-0002-2203-8216