Application of species-specific primers to estimate the in situ diet of Bythotrephes [Cladocera, Onychopoda] in its native European range via molecular gut content analysis

Abstract The study of invasive species often focuses on regions of recent introduction rather than native habitats. Understanding an invasive species in its natural environment, however, can provide important insights regarding the long-term outcome of invasions. In this study we investigated the diet of the invasive spiny water flea, Bythotrephes longimanus, in two Austrian perialpine lakes, where it is native. The gut contents of wild-caught Bythotrephes individuals were estimated by quantitative polymerase chain reaction, targeting species-specific fragments of the barcoding region of the cytochrome c oxidase I gene of potential prey. The observed prey spectrum of Bythotrephes in the study lakes consisted primarily of Eudiaptomus gracilis and Diaphanosoma brachyurum. The Daphnia longispina complex, Leptodora kindtii and Mesocyclops leuckarti also contributed to the diet. Results indicate that Bythotrephes is a generalist feeder with a preference for epilimnetic prey.


INTRODUCTION
Biological invasions are recognized as important driver of global environmental change (Carpenter et al., 2011). Freshwater ecosystems are especially vulnerable to invasions and its effects (Sala et al., 2000). Although at least conspicuous invasive species with implicit impacts in their new environment are relatively well studied, detailed knowledge of the same species in their native areas is often lacking. Thus, there has been a continued bias toward studying invaders primarily in their introduced ranges (Parker et al., 2013). Among other impacts, invasives may affect existing trophic relationships, which, as key mediators of ecosystem functioning, determine community dynamics (Cardinale et al., 2012). Understanding the diet of invasives is a crucial factor in understanding biological invasions. Diet studies of both invasive and native populations are imperative to understand dietary requirements, dietary flexibility and thus potential impacts of invasive species (Courant et al., 2017). For instance, a broad dietary spectrum is a frequently cited characteristic of invasives, yet successful invaders do not necessarily exhibit a broad diet (Vázques, 2006).
The spiny water flea, Bythotrephes Leydig 1860, is a genus of predatory freshwater cladocerans with an extensive Eurasian native distribution. It occurs around the Central European Alps, in Northern Europe and throughout countries of the former Soviet Union into China (Therriault et al., 2002). Bythotrephes gained considerable attention after invading the Laurentian Great Lakes in North America in the 1980s (Sprules et al., 1990). Since then it has established in numerous lakes in Canada and the USA (e.g. Yan et al., 1992;Kerfoot et al., 2016), subsequently followed by decreases in species richness (Yan et al., 2002;Barbiero and Tuchman, 2004;Kelly et al., 2013) and biomass of zooplankton (Kerfoot et al., 2016). In contrast, other zooplankton species increased to previously unrecorded levels (Yan and Pawson, 1997). In European lakes with Bythotrephes, the observed species diversity of zooplankton was unaffected or even higher than in similar lakes where no Bythotrephes were found (Hessen et al., 2011;Kelly et al., 2013;Walseng et al., 2015;Horváth et al., 2017). However, the predation pressure exerted by Bythotrephes in Lago Maggiore, Italy, was deemed high enough to seasonally reduce Daphnia densities to low values (Manca et al., 2008). Additionally, the observed long-term rise in water temperature and changes in the thermal stratification of Lago Maggiore were assumed to positively affect reproduction and population density of native Bythotrephes and thus be associated with dramatic changes in the pelagic food web (Manca et al., 2007). These results further strengthen the need to study Bythotrephes in its native habitat.
Bythotrephes is largely considered a generalist predator. In some experimental feeding studies, cladocerans, including Daphnia, Ceriodaphnia and (Eu)Bosmina, were strongly preferred prey for Bythotrephes, whereas in other studies calanoid and cyclopoid copepodites, nauplii and even adult copepods (Tropocyclops extensus) were common prey (Vanderploeg et al., 1993;Dumitru et al., 2001;Wahlström and Westman, 2011;Jokela et al., 2013). Some feeding studies observed a more specialized predation. Despite offering several prey species (Daphnia mendotae, B. longirostris, Diaptomus sp. and Diacyclops thomasi), Pangle and Peacor (2009) only observed the consumption of D. mendotae. In a similar experiment, Bythotrephes reduced the abundance of Daphnia, Bosmina and Ceriodaphnia, whereas copepods and Diaphanosoma were not preyed upon (Jokela et al., 2013). However, laboratory feeding studies can be difficult to interpret and it is recognized that experimentally measured responses to prey, e.g. prey choice, may not accurately reflect responses in the field (Van Lenteren and Bakker, 1975;Symondson et al., 2002;McKemey et al., 2003). Allozyme analyses of wild-caught Bythotrephes from Lake Michigan  indicated a diverse diet consisting of cladocerans, cyclopoid copepods and calanoid copepods.
The use of Molecular Gut Content Analysis (MGCA) approaches has increased the precision of diet estimations due to high taxonomic resolution and sensitivity to macerated or degraded prey tissue (King et al., 2008;Nielsen et al., 2018). DNA-based methods of diet estimation compare well to known (fed) diet when there is a small number of potential resources, but similarity between different methods of diet estimation generally declines with increasing number of potential resources (Nielsen et al., 2018). Quantitative polymerase chain reaction (qPCR)based techniques have successfully been implemented in zooplankton diet studies: from laboratory feeding experiments revealing the first molecular detection of a specific species of algae in calanoid copepod guts and fecal matter (Nejstgaard et al., 2003) and the detection of predation of Acartia nauplii in calanoid copepods (Durbin et al., 2008), to in situ diet studies of wild-caught doliolids (Frischer et al., 2014;Walters et al., 2019). In their studies on copepods both Nejstgaard et al. (2008) and Durbin et al. (2008) noted that the amount of DNA estimated via qPCR was lower than expected. The possibility of rapid digestion resulting in an underestimation of stomach contents was later confirmed Troedsson et al. 2009;Durbin et al. 2012). The high sensitivity of DNA-based diet analyses also comes with the disadvantage of possible detection of false positives due to contamination, which is an ubiquitous problem, especially challenging in small aquatic invertebrates due to the surrounding water teaming with dietary components (Passmore et al., 2006;King et al., 2008). In qPCR assays, the cycle threshold (ct) value is defined as the number of cycles required for the fluorescent signal to exceed background levels. Hence, samples with high amounts of target DNA require fewer cycles than samples with little target DNA. It is a common approach, especially in clinical studies and zoonotic disease research, to apply a cutoff of late ct values and only regard results with fewer cycles as true positives (e.g. Caraguel et al., 2011;Baerwald et al., 2012;Bolotin et al., 2015;Aminu et al., 2020).
In this study, we estimated the in situ diet of Bythotrephes using qPCR. We designed primer pairs targeting short sequences of the mitochondrial cytochrome c oxidase subunit I (COI) gene of potential crustacean prey species, occurring in two study lakes of the native region of Bythotrephes longimanus, Erlaufsee and Mondsee, located in the Austrian perialpine region. The tested prey species included common taxa of oligo-and mesotrophic European lakes: the cladocerans Daphnia longispina complex, (Eu)bosmina sp., Leptodora kindtii and Diaphanosoma brachyurum, and the copepods Eudiaptomus gracilis, Mesocyclops leuckarti, Cyclops vicinus and Cyclops sp. X (see Method). Our aim was to design and apply species-specific primers for potential crustacean prey species in the study lakes and to provide an estimation of the in situ diet of Bythotrephes in its native range to elucidate its choice of prey in the field and improve the understanding of the trophic role of this predator in lake food webs.  (Horváth et al., 2017). The lakes were sampled biweekly between May and November 2018, using closable nets to sample epilimnion, metalimnion and hypolimnion, and net mesh sizes of 40 μm, 100 μm and 285 μm. For abundance estimates zooplankton was subsampled prior to counting. Entire samples were examined for Bythotrephes and Leptodora. Bythotrephes used in qPCR were collected across the season (Table I) to minimize bias of one-time sampling. Samples were fixed and stored in 80% ethanol. Zooplankton specimens were separated and rinsed in ethanol before DNA extraction.

DNA extraction
Total DNA was extracted from whole individuals of Bythotrephes and potential prey species using the QIAGEN DNeasy Blood & Tissue Kit (Valencia, CA, USA) following manufacturer protocols. DNA was eluted in 100 μL of AE buffer and stored at −20 • C. Following genomic DNA (gDNA) extraction, purified DNA was quantified using a Qubit 2.0 Fluorometer (Invitrogen, Carlsbad, CA, USA) and a Qubit dsDNA High Sensitivity Assay Kit (Life Technologies, Eugene, OR, USA). Amenability of the extracts for PCR amplification was determined by amplifying 18S rRNA using universal primers Univ-18SF-577F (5 CAG CAG CCG CGG TAA TTC C 3 ) and Univ 18S-1180R (5 CCC GTG TTG AGT CAA AAG C 3 ) as previously described (Frischer et al., 2017). Amplification was accomplished using a GeneAmp 9 700 Thermocycler (Applied Biosystems, Foster City, CA) and included a 3 min initial denaturation at 95 • C followed by 30 amplification cycles [94 • C (30 s), annealing temperature of 60 • C for 30 s, 72 • C (1 min)] followed by a 10 min final extension step at 72 • C.

Prey species and primer design
A set of 8 primer pairs (Table II)  Cyclops sp. X (Krajíček et al., 2016) is a, formally undescribed, European cyclopoid copepod, often misidentified as Cyclops abyssorum. C. vicinus, while present in Erlaufsee, was not included in the final qPCR analyses, because it was extremely rare in our zooplankton samples.

qPCR assay and data analysis
Quantitative standards for qPCR were prepared by amplifying the mtCOI gene from each of the targeted prey species using the Folmer primer set, cloned into a plasmid and subsequently used as quantitative standards for the qPCR assay.  Hercules, CA, USA). qPCR reactions were conducted in 20 μL reaction volumes containing a final concentration of 1X SsoFast™ EvaGreen ® Supermix, 0.3 μmol of each primer, and template concentrations of plasmid DNA (pDNA) containing a cloned copy of the target COI gene ranging from 10 1 to 10 8 copies per reaction. Prey DNA concentrations derived from 123 female Bythotrephes specimens (n Erlaufsee = 53; n Mondsee = 70) were estimated by qPCR using each of the 7 prey-specific primer sets designed in this study (Table II). qPCR reactions were conducted in 20 μL reactions essentially as described above, except that template concentrations ranged from 0.23 ng to 8.14 ng target gDNA per reaction. Empirically optimized amplification annealing temperatures for each of the assays are reported in Table II. qPCR reaction conditions included an initial enzyme activation step at 95 • C for 30 s followed by 40 cycles of denaturation (95 • C, 5 s) and annealing/extension (Table II, 5 s). After cycling, product melt-temperatures were evaluated from 55 • C to 95 • C at 0.5 • C increments for 5 s each.
The abundance of COI genes was estimated relative to standard curves prepared with quantified pDNA containing an insert of the target COI gene from the respective prey species (Supplementary Fig. 1). All qPCR reactions utilized a Bio-Rad CFX96 Real-Time PCR System. Samples, standards and no template controls were assayed in triplicate. For each triplicate, mean ct values were calculated. Samples were dismissed if two out of three replicates were below detection limits. Copy numbers were estimated from the ct values for each species, each qPCR run and associated standard curve. Mean efficiency (E) of all valid qPCR reactions was 105.5 (SD = 5.3), mean coefficient of determination (R 2 ) was 0.98 (SD = 0.01), mean E and R 2 for each prey species are summarized in Table III. Four qPCR runs were excluded from further analyses due to high (126.0%, 137.6%) or low (87.9%) amplification efficiencies and mean ct values of 37.1 and 36.5 for the nontemplate control.
Because M. leuckarti and D. brachyurum do not occur in Erlaufsee, specific primers for these species (Table II) were chosen to test for contamination/false positives in the MGCA by applying the primers to Bythotrephes sampled from this lake. Tests for M. leuckarti resulted in ten negatives and two signals (ct = 39.46 and ct = 36.77). Tests for D. brachyurum resulted in five negative specimens and five specimens with signals between ct 38.3 and 39.72. We therefore considered a cutoff at ct = 36 as an adequate condition for this data set. Subsequent results refer to counts with ct < 36.
Statistical analyses were conducted using R (R Core Team, 2019). Significant differences in diet between the lakes and between respective prey species were tested using Fisher's exact test, based on counts of positive (ct < 36) versus the sum of inconclusive and negative signals (Table I), and Kruskal-Wallis test for differences in starting quantity (SQ) of positive counts (base R 3.6.1). Figures were plotted using the R packages ggplot2 (Wickham, 2016), scales (Wickham and Seidel, 2020), ggpubr (Kassambara, 2020), mdthemes (Neitmann, 2020) and dplyr (Wickham et al., 2020).
E. gracilis was present in all lake layers of both study lakes throughout the sampling period (Fig. 3). D.
brachyurum was the dominant cladoceran in Mondsee, peaking in mid-July and decreasing steeply in August. Daphnia was common in Erlaufsee, but comparatively rare in Mondsee. Similarly, Bosmina was common in early summer in Erlaufsee, but rarely present in Mondsee. L. kindtii was very abundant in Mondsee, but rare in Erlaufsee. In contrast to Erlaufsee, cyclopoid copepods were dominant in Mondsee with mostly M. leuckarti occupying the epilimnion.

DISCUSSION
This study provides the first molecular in situ diet estimation of the predatory cladoceran Bythotrephes in its native range by MGCA using qPCR. Both cladoceran and copepod prey were consumed, confirming that Bythotrephes is a generalist predator. Main differences in the diet between the study lakes concern the occurrence of D. brachyurum and M. leuckarti in Mondsee and the absence of said species in Erlaufsee. E. gracilis and D. brachyurum were detected at a much higher frequency in the MGCA than other species (Fig. 2), suggesting some level of feeding selectivity.
To evade visual predators like fish, some zooplankton taxa are able to migrate into deep lake layers (Ringelberg, 2010). This behavior has also been observed for North American species as a response to the invasion of Bythotrephes (e.g. Pangle et al., 2007;Bourdeau et al., 2011). E. gracilis nauplii in Mondsee are reportedly distributed in the epilimnion, in stark contrast to nauplii of Cyclops (Nauwerck, 1988). Strict epilimnetic distribution has also been reported for D. brachyurum (Nauwerck, 1988;Gaviria-Melo et al., 2005;Błe˛dzki and Rybak, 2016), presumably benefitting the visual predator Bythotrephes (Muirhead and Sprules, 2003;Pangle and Peacor, 2009). Our data generally support these observations, especially D. brachyurum was very abundant in the epilimnion of Mondsee and E. gracilis was present in the upper layers of both study lakes throughout the sampling period (Fig. 3). Yet, vertical distribution alone might not explain choice of prey in all cases: In North America, the hypolimnetic calanoid Senecella calanoides decreased after Bythotrephes invasion, whereas Limnocalanus macrurus did not (Kerfoot et al., 2016). A closely related North American species to D. brachyurum, D. birgei, was observed to decline in invaded lakes, but lab experiments could not observe evidence of predation impact on the genus Diaphanosoma (Jokela et al., 2013). A slightly smaller proportion of observed Daphnia signals may imply more successful defense strategies. However, Daphnia were rare in Mondsee and quite abundant in Erlaufsee, which might explain the different proportion of positive signals between the lakes. It is unclear whether Bythotrephes in Mondsee prefer D. longispina, which is generally capable of vertical migration (Ringelberg, 2010), or D. cucullata, which exhibit epilimnetic distribution, but also conspicuous defense structures (Laforsch and Tollrian, 2004). In North America, different species of Daphnia were repeatedly described as preferred prey of Bythotrephes (e.g. Vanderploeg et al., 1993;Pangle and Peacor, 2009;Kerfoot et al., 2016). M. leuckarti also exhibits a primarily epilimnetic lifestyle (Nauwerck, 1988;Błe˛dzki and Rybak, 2016;Nilssen and Waervågen, 2000). In North America, Mesocyclops edax is a rare case of cyclopoid copepod reported to decline after Bythotrephes introduction (Barbiero and Tuchman, 2004;Kerfoot et al., 2016).
Similar to its close relative in Europe, M. edax is typically distributed above the thermocline (Marcogliese and Esch, 1992). L. kindtii was occasionally detected in Bythotrephes guts. Predation of North American L. kindtii has been observed in laboratory experiments (Branstrator, 1995). Interestingly, L. kindtii's response to Bythotrephes has been described as neutral in Norway but negative in Canada (Kelly et al., 2013).
Previous studies demonstrated noteworthy differences in the diet of Bythotrephes (see Introduction), which could only partially be explained by the chosen methods. Vanderploeg et al. (1993) showed spatial variations in the diet of Bythotrephes in Lake Huron. Whereas cladocerans were favored prey in one location and copepods were hardly eaten, in another location without its preferred prey, Bythotrephes exhibited a moderate clearance rate on nauplii. Similarly, we would expect seasonal diet variations in our study lakes, dependent on the abundance of different prey species. It is, e.g. conspicuous that the highest number of positives for D. brachyurum (Table I) was observed around the time of its peak abundance (Fig. 3). D. brachyurum was seasonally present in large numbers, constituting a notable source of prey. Future studies may focus on seasonal diet variations by substantially increasing the number of analyzed specimens and reporting how broad, specialized or dependent on prey abundances the diet is across the season. Here it would be advisable to sample both predators and prey from the whole water column to allow for correlating zooplankton abundances with observed diet proportions.
Cyclops sp. X and Bosmina could not be detected by qPCR given the applied cutoff. It should be noted that the detection of Cyclops sp. X in Mondsee represents its first Austrian record. Although predation of cyclopoid copepods by Bythotrephes has been previously observed (Vanderploeg et al., 1993;Schulz and Yurista, 1995), it is generally considered uncommon, with some exceptions. For example, Kerfoot et al. (2016) observed the significant reduction of naupliar stages of copepods in invaded lakes. The case of Bosmina is particularly interesting, because B.(E.) coregoni itself has been introduced to the Great Lakes region (Mills et al., 1993), where it has been reported to decline significantly in lakes where Bythotrephes has been introduced ( Barbiero and Tuchman, 2004). One possible explanation for the lack of detection in the guts might be that in Mondsee, Bosmina was very rare and in Erlaufsee it was only common early in the season (Fig. 3). Further, Schulz and Yurista (1999) speculated that Bosmina, as well as copepod nauplii, could constitute a greater proportion of the diet of juvenile (i.e. instar I) Bythotrephes. We did not include the developmental status of Bythotrephes in this study, but instar II specimens were certainly most abundant in our samples (data not shown). Thus, samples analyzed may not reflect the full range of Bythotrephes diet across all instars. Future dietary studies should include, and distinguish between, all life stages of Bythotrephes. Whether Bythotrephes is the main reason for the observed changes in abundance and the pronounced vertical migration of some species in Mondsee and Erlaufsee is beyond the scope of this study and has to be analyzed carefully, with the inclusion of environmental lake data and study lakes without this predator (Pichler et al, in prep.).
There are several possible sources of bias that may affect the interpretation of MGCA results. Most notably, environmental contamination (Passmore et al., 2006;King et al., 2008), cannibalism (Pompanon et al., 2012) and secondary predation (Sheppard et al., 2005;King et al., 2008) are relevant sources of error. Within this study, secondary predation might apply to potentially cannibalized conspecifics, copepods and L. kindtii in particular, while the remaining cladocerans are known to primarily feed on algae and bacteria. Zooplankton is sampled by net tows, concentrating many thousands of specimens, which might get entangled and leave traces of DNA on predators of interest as well. Besides separating and rinsing individual specimens, we attempted to diminish false positives by discarding late signals, knowingly accepting possible type II errors in the final diet estimation. Applying a cutoff and analyzing early qPCR signals is a common method to decrease the possible effect of contamination (see Introduction).
Detected signals differed regarding signal strength. In copepods, rapid digestion allows the detection of prey DNA up to 6 h after ingestion (Nejstgaard et al., 2003;Durbin et al., 2008). The rate of disappearance of PCR signals in these studies was correlated with estimates of gut evacuation rates, suggesting similar circumstances for Bythotrephes, which has an estimated gut passage time of 6-12 h . Detected signals most likely represent daytime feeding, because predation by Bythotrephes is light-dependent (Pangle and Peacor, 2009). Different sizes of prey species were not considered, because Bythotrephes is a sloppy feeder and does not ingest whole prey (Yurista et al., 2010). Additionally, a positive signal does not indicate whether one or more specimens of the same prey species, small larvae or large adults have been captured, especially considering rapid digestion and that the time of ingestion is unknown. Particular cases of conspicuously high starting quantities (Fig. 1) could perhaps be explained by very recent consumption and/or predation of ovigerous females. The quantity of target detected is expected to scale with the number of eggs consumed (Weber and Lundgren, 2009). Cotterill et al. (2013) demonstrated that even a single developing oocyte may increase mitochondrial copy number on the order of a thousand-fold. Ovigerous females of E. gracilis had higher encounter and mortality rates in laboratory experiments with predatory fish than nonovigerous females, most likely due to their highly visible egg-clutch (Svensson, 1992). The same may be true for encounters with Bythotrephes. Although it is impossible to exclude methodological artifacts, it seems unlikely these would affect the major conclusions of this study.

CONCLUSION
In this study, we estimated the diet of Bythotrephes in its native range. Based on 123 individuals collected from two Austrian lakes, the diet of Bythotrephes consisted of E. gracilis, D. brachyurum, D. longispina complex, M. leuckarti and L. kindtii, with the most commonly detected prey species being E. gracilis. The diet differs between the two studied lakes, likely due to differences in prey species composition. These observations support earlier conclusions that the spiny water flea is a generalist predator in both native and invaded ranges. Prey selection is therefore likely the result of differences in prey distribution and migration behavior in the water column. Future studies should aim to incorporate species-specific developmental status and corrections for breakdown of DNA in predator guts.