Benthic habitat is an integral part of freshwater Mysis ecology

Abstract Diel vertical migration (DVM) is common in aquatic organisms. The trade‐off between reduced predation risk in deeper, darker waters during the day and increased foraging opportunities closer to the surface at night is a leading hypothesis for DVM behaviour. Diel vertical migration behaviour has dominated research and assessment frameworks for Mysis, an omnivorous mid‐trophic level macroinvertebrate that exhibits strong DVM between benthic and pelagic habitats and plays key roles in many deep lake ecosystems. However, some historical literature and more recent evidence indicate that mysids also remain on the bottom at night, counter to expectations of DVM. We surveyed the freshwater Mysis literature using Web of Science (WoS; 1945–2019) to quantify the frequency of studies on demographics, diets, and feeding experiments that considered, assessed, or included Mysis that did not migrate vertically but remained in benthic habitats. We supplemented our WoS survey with literature searches for relevant papers published prior to 1945, journal articles and theses not listed in WoS, and additional references known to the authors but missing from WoS (e.g. only 47% of the papers used to evaluate in situ diets were identified by WoS). Results from the survey suggest that relatively little attention has been paid to the benthic components of Mysis ecology. Moreover, the literature suggests that reliance on Mysis sampling protocols using pelagic gear at night provides an incomplete picture of Mysis populations and their role in ecosystem structure and function. We summarise current knowledge of Mysis DVM and provide an expanded framework that more fully considers the role of benthic habitat. Acknowledging benthic habitat as an integral part of Mysis ecology will enable research to better understand the role of Mysis in food web processes.

Although partial DVM in Mysis has been observed in a number of systems including the Laurentian Great Lakes (Bowers, 1988;Johannsson, Rudstam, Gal, & Mills, 2003) and the Baltic Sea (Rudstam et al., 1989), the magnitude of the population that remains benthic at night is not well understood and this portion of the population is typically excluded from population assessments and evaluations of the importance of Mysis in the ecosystem. We surveyed the literature to evaluate the extent to which benthic Mysis have been included in research. We restricted our analyses to freshwater Mysis (Porter, Meland, & Price, 2008) except for the marine Mysis mixta, as this species has been extensively studied in the Baltic Sea where it coexists with freshwater congenerics Mysis relicta and Mysis salemaai and has similar ecological roles (Salemaa, Tyystjarvimuuronen, & Aro, 1986;Salemaa, Vuorinen, & Välipakka, 1990). Our survey results suggest that benthic habitat is more important to Mysis ecology than previously assumed, and is necessary to consider this to better understand Mysis' production and its roles in food webs. We conclude with a set of questions and hypotheses to help frame research directions to fill in knowledge gaps in Mysis ecology.

| LITER ATURE SURVE Y ON MYSIS ECO LO GY
We conducted literature surveys for three aspects of freshwater Mysis ecology-demographics, in situ diets, and experiments that examined Mysis feeding behaviour. Each survey was initiated using Web of Science (WoS) with varying sets of search terms, spanning 1945-2019. For demographics, we surveyed WoS using the terms "Mysis" AND "abundance OR densit* OR biomass* OR growth OR size OR length OR distribution" under the field "Topic". We used the terms "Mysis" AND "diet OR gut OR stomach" under the field "Topic" to survey WoS for in situ diets. For experiments on Mysis feeding behaviour, we used two sets of terms under the field "Topic". The first included "Mysis" AND "feeding OR predation" AND "experiment" under the field "Topic", and the second included "Mysis" AND "contaminant* OR toxi*" under the field "Topic".
We examined the resultant titles and abstracts for relevance to each topic, and extracted the relevant information from each study.
However, several relevant papers were not identified by the WoS search. For example, at the conclusion of the literature survey, only 47% of the papers used to evaluate in situ diets were identified by WoS. Therefore, during the survey, we examined the cited references in all relevant papers that were published prior to 1945 (the earliest year for WoS search), journal articles and theses that were not listed in WoS, and additional references known to the authors. We did not include papers that developed new sampling devices (Lasenby & Sherman, 1991) or relied solely on acoustics without net samples (Levy, 1991;Miller, 2003). Papers that relied on previously published values (Morgan, 1985;Morgan, Threlkeld, & Goldman, 1978;Sell, 1982;Sullivan & Rudstam, 2016) were not included. Theses and dissertations were included unless data were published in peer-reviewed literature. In those cases (e.g. Morgan, 1985), information from the published papers was used. Studies that reported metrics but not methods were excluded (Adare & Lasenby, 1994;Lasenby & Langford, 1972). More details are provided in Tables S1-S3, and the results are summarised below. Also note that Mysis taxonomy was revised in the mid-2000s, with new names given to four sibling species formerly known as M. relicta (M. salemaai, Mysis segerstraeli, Mysis diluviana, and M. relicta;Audzijonytė & Väinölä, 2005). With the exception of the high Arctic, the only mysid species in North America is M. diluviana. Even so, wherever we report a particular species, we refer to the name used in the cited publication.
The results of the literature survey clearly show that freshwater Mysis research has largely ignored benthic-caught individuals and their benthic environment (Figures 1 and 2). Mysis demographics (n = 142 total studies) included benthic-caught individuals in a minority of studies, ranging from 15% of the studies that estimated biomass (n = 40) to 40% of studies that assessed life-stages (n = 76; Figure 1; Table S1). Conversely, such estimates were based solely on pelagic-caught individuals in 60-85% of demographic studies.
However, in a handful of studies which sampled pelagic and benthic habitats at night, variable but potentially very high proportions of Mysis have been reported on the bottom at night (3-84%, Table 1).
Detritus was reported in Mysis stomachs in 19 of the 32 studies, including nine of 10 studies that included both benthic-and pelagic-caught Mysis, and somewhat surprisingly, in eight of 18 studies that only examined pelagic-caught Mysis. If mysids feed continuously when on bottom, then pelagic-caught Mysis sampled early in the night, soon after ascent, may still contain benthic resources in their guts. The remaining 10 pelagic-only studies did not include detritus as a possible prey category (Table S2). In a recent study, benthic-caught Mysis during the night and day had similar amounts of detritus in their stomachs, suggesting that Mysis feed continuously when on the bottom (O'Malley & Stockwell, 2019) and benthic resources are likely to contribute significantly to Mysis growth and production (Lehtiniemi et al., 2002;Sierszen, Kelly, Corry, Scharold, & Yurista, 2011;Whall & Lasenby, 2009). We also found that Mysis feeding ecology experiments (n = 58 total studies) used benthic-or combined benthic-and pelagic-caught individuals in only 21% of the studies, whereas 43% of the studies used only pelagic-caught individuals and 36% did not report the habitat from which Mysis were captured for the experiments (Table S3).
Pelagic zooplankton was by far the most used prey item in feeding experiments (>90% of experimental studies), followed by brine shrimp Artemia, which is not a natural prey of Mysis (22% of studies; Figure 2, Table S3). We found few studies where detritus, benthic invertebrates, or algae were used in experiments (n ≤ 7 each, Figure 2, Table S3) despite their presence in field diets (see Role of benthic food resources below).
The lack of historical focus on the benthic environment may bias our basic understanding of Mysis ecology and the role of Mysis in ecosystems. As evidenced by findings from studies that have examined benthic and pelagic Mysis at night (Table 1), only a fraction of a population may inhabit the pelagic environment at night (Naesje, 1995).
The ecology of Mysis is more complex than that of an animal simply hiding on the bottom during the day waiting for the cover of darkness to migrate up the water column at night to feed on plankton.
Researchers have probably focused on the pelagic phase of Mysis for several reasons. First, Mysis DVM has been assumed to be a population-level phenomenon and consequently, night pelagic samples should provide representative estimates of abundance and other population demographics. Second, pelagic habitat is easier to sample than benthic habitat. Sampling devices towed along the bottom can fill with substrate (e.g. mud, silt, detritus, dreissenid mussel shells), get caught on obstructions, or not hold tight to bottom with sharp bathymetric relief or complex substrate. Towing a net through the water column is far easier and safer, despite the necessity of working at night. A number of early studies concluded that pelagic vertical net tows at night were the best method for quantitative estimates of Mysis because they yielded the highest areal density estimates among methods (Grossnickle & Morgan, 1979;Morgan & Threlkeld, 1982;Nero & Davies, 1982;Sell, 1982;Shea & Makarewicz, 1989), which probably influenced following generations of Mysis researchers (but see Reynolds & DeGraeve, 1972).
Third, Mysis introductions into non-native lakes resulted in negative impacts on pelagic zooplankton in many cases (Goldman et al., 1979; F I G U R E 1 Percent of studies that reported freshwater Mysis spp. and Mysis mixta demographic data using individuals collected from benthic, pelagic, or both habitats. Number of studies reporting each type of demographic data is in parentheses. Total number of studies with at least one demographic metric was 142. Density (number of individuals per unit volume or area) includes abundance (total individuals) F I G U R E 2 Different prey types used in freshwater Mysis spp. or Mysis mixta feeding experiments, presented as percent of studies (out of 58) that used each prey type. In some cases, multiple prey types were used in individual studies so sum of percentages exceeds 100% Kinsten & Olsén, 1981;Lasenby et al., 1986), which may have in-

| THE NEED FOR AN E XPANDED FR AME WORK ON MYSIS B ENTHI C RE SOURCE S AND HAB ITAT
The focus on pelagic Mysis restricts our understanding of its general ecology, and poses several potentially large knowledge gaps with implications for population assessments, food web ecology, and fisheries management (Kitchell et al., 2000;Pothoven & Madenjian, 2008;Pothoven, Nalepa, Schneeberger, & Brandt, 2001).

| Mysis demographics and benthic habitat sampling
Mysis occupy benthic habitat night and day (Table 1). Evidence suggests that the individuals caught on the bottom both night and day are represented by disproportionately more adults, including gravid females, compared to pelagic individuals caught during the night ( Figure 3; Euclide et al., 2017;McWilliam, 1970;Reynolds & DeGraeve, 1972). Consequently, assessments based on night pelagic samples may underestimate population density and biomass-the latter to a greater extent because of the exponential increase in mass with length. Population size-structure and life-stages from night pelagic samples may also bias our inferences of population demographics.
Research using cameras (Bergersen & Maiolie, 1981) and video (O'Malley, Dillon, et al., 2018) indicate that benthic Mysis densities may be 2-10× greater than estimates based on traditional benthic sled collections, suggesting that estimates of the proportion of Mysis that are benthic at night based on benthic sleds (Table 1)  found average CV of 23 (day) and 68% (night) for pelagic sampling, and 41 (day) and 32% (night) for benthic sampling. Understanding sampling gear bias and precision is critical for measuring potential demographic differences among habitats. Adoption and refinement of digital recording systems, in concert with benthic sleds, are promising methods to Overall, the literature suggests that assessments of Mysis at night using pelagic gear, and their application to ecological processes (e.g. production, zooplanktivory), provide an incomplete picture of Mysis populations and their role in ecosystem structure and function.

| Role of benthic food resources
Benthic food resources probably play a significant role in Mysis energy dynamics. Mysis consume and grow on a variety of benthic foods including detritus, zooplankton eggs, amphipods, and benthic zooplankton (Albertsson, 2004;Johannsson et al., 2001;Karlson & Viitasalo-Frosen, 2009;Parker, 1980;Sierszen et al., 2011;Viitasalo & Viitasalo, 2004). For example, no differences in growth rates or survival were observed for M. mixta fed Artemia nauplii versus dried and ground plant material (Enteromorpha spp.), with growth rates similar to those observed in situ (Gorokhova & Hansson, 1999). Neomysis americana also grew well on a diet of cordgrass (Spartina alterniflora) detritus (Zagursky & Feller, 1985). Similar results were reported for mysids fed fresh and decaying plant material and detritus (Irvine, Moss, Bales, & Snook, 1993;Lasenby & Van Duyn, 1992;Lesutiene, Gorokhova, Gasiunaite, & Razinkovas, 2008;Speirs, Lawrie, Raffaelli, Gurney, & Emes, 2002). Ingestion rates of detritus were approximately double those of phytoplankton in laboratory feeding experiments, on a caloric basis, and suggested that detritus played a significant role in growth of Lake Tahoe Mysis (Morgan, 1979). Large Mysis captured in the pelagia of Lake Superior at night in September relied on benthic sources for 27-58% of their diet (Sierszen et al., 2011), Mysis from Okanagan Lake assimilated 4-59% of their carbon from benthic sources (Whall & Lasenby, 2009), and pelagic zooplankton could not solely support Mysis growth in Lake Ontario (Johannsson, Rudstam, & Lasenby, 1994), providing further evidence that benthic food resources contribute to assimilated energy. Detritus has been found in the stomachs of almost all species of mysids from freshwater to marine habitats (Mauchline, 1980;Takahashi, 2004). Such results are not surprising, as mysids are efficient at digesting detritus and contain gut enzymes needed to breakdown plant material (Foulds & Mann, 1978;Friesen, Mann, & Novitsky, 1986;Wainwright & Mann, 1982). Consequently, more consideration and better estimates of daytime foraging are needed to fully realise the contribution of benthic resources to mysid growth and survival, and by extension, the influence of mysids on ecosystems.
Production estimates need to take into account the demographic distribution of mysids in both pelagic and benthic habitats, and the resources in those habitats including detritus and possibly benthic macro-and meio-fauna (Karlson & Viitasalo-Frosen, 2009). Growth and mortality estimates based solely on pelagic samples may underand overestimate rates, respectively, because larger (and gravid) individuals of a population may disproportionately occupy benthic habitat at night (Figure 3). Furthermore, the lower temperature in bottom waters may increase energy conversion efficiency and lipid content and hence energy density of mysids (Chess & Stanford, 1999).

| Benthic behaviour
In addition to mysids' ability to consume benthic resources, Mysis probably use benthic habitat to further reduce their vulnerability to visual predators. For example, bloater (Coregonus hoyi) capture success rate for Mysis off bottom was nearly double that when Mysis were on bottom (Crowder & Binkowski, 1983). The sediment surface is used by Mysis to propel away from predators with greater acceleration and maximum speed than possible in pelagic habitat and to maintain their position on top of sediment in strong currents, and they can also burrow into the sediment (Bowers, 1988 Robertson, Powers, & Anderson, 1968;Sellers, 1995 Hondorp, Pothoven, & Brandt, 2011;Weidel et al., 2017), consume prey on or in the sediment (Kraft & Kitchell, 1986;Selgeby, 1988) using vibration to detect prey (Hoekstra & Janssen, 1985;Janssen, 1990).
Such a foraging strategy, however, may be relatively ineffective on Mysis (Bowers, 1988) given sculpins' capture success rates of only c.
10% for Mysis compared to 45 and 78% for less agile amphipod and chironomid prey, respectively (Hondorp, 2006). In systems where dreissenid mussels have colonised and expanded into Mysis benthic habitat, the additional complex shell habitat and organic matter could influence Mysis interaction with predators and sediments (Stewart, Miner, & Lowe, 1998), but this remains to be tested. Similarly, in systems where nonnative dreissenid mussels have increased sedimentation rates and sediment organic matter content (e.g. Klerks, Fraleigh, & Lawniczak, 1996;Stewart et al., 1998), the importance of Mysis-sediment dynamics will probably be greater than in systems without deep, filter-feeding mussels.

| Food web and ecosystem effects
Mysis use of benthic food resources probably has compounding effects. The total energy consumption by Mysis populations may be underestimated if based on pelagic feeding rates. When energy consumption is inferred from growth rates and bioenergetics models, part of that energy intake will be from benthic resources and, if not accounted for, will bias high the inferred effects on zooplankton.
Benthic Mysis and benthic resources used by Mysis may explain why mass-balance food web models often estimate greater fish predation on Mysis than observed Mysis production or biomass can support. Mysis biomass and production had to be increased 2-3fold over observed values to meet the estimated consumption demands of planktivorous fishes in a Lake Ontario model (Stewart & Sprules, 2011). Lake Michigan mass-balance food web models also estimated fish predation of Mysis to be greater than Mysis biomass observations (Rogers, Bunnell, Madenjian, & Warner, 2014). Similarly, in the Baltic Sea, zooplankton production was insufficient to support the quantity of Mysis consumed by fishes (Harvey, Cox, Essington, Hansson, & Kitchell, 2003) when pelagic invertebrates were assumed to consume 100% zooplankton (Sandberg, Elmgren, & Wulff, 2000).
To account for the imbalance, and based on results from stable isotope analyses (Hansson et al., 1997), a 50/50 diet balance of zooplankton and plant material for Mysis was needed in the food web model (Harvey et al., 2003). The evidence from the literature suggests such holes in mass balance food web models could be filled by accounting for Mysis in benthic habitats and their use of benthic resources.
One consequence of the pelagic focus of Mysis research in freshwater systems is the implicit inference that pelagic production is the dominant energy pathway for Mysis. Our literature survey, however, suggests that Mysis may spend more time on the bottom than assumed. In winter, when pelagic production is low and Mysis fecundity is high, the motivation to migrate into the water column at night is also presumably low and a greater proportion of the population probably spends more time on bottom than during the other seasons (e.g. Johannsson et al., 2001;Salemaa et al., 1986; but see Lehtiniemi et al., 2009; Figure 4). Conversely, in summer when pelagic production is high, the motivation to migrate is also presumably high but short nights limit access to pelagic resources. Additionally, the abundance and quality of benthic resources is influenced by season and water column depth (Auer, Cannon, & Auer (2009);Eadie, Chambers, Gardner, & Bell, 1984;Ostrom, Long, Bell, & Beals, 1998;Scharold, Lozano, & Corry, 2004), suggesting dynamic cost:benefit tradeoffs to migration over space and time, which probably influences the amount of benthic feeding by mysids (Johannsson et al., 2001;Sierszen et al., 2011Sierszen et al., , 2014. However, when integrated over a year, the proportion of time and the proportion of mysids that occupy benthic habitat is probably greater than pelagic habitat. Because An underlying theme in Mysis research, either implicitly implied or explicitly stated by the focus on the pelagia, is that benthic resources act as a resource subsidy to support Mysis populations through periods of low pelagic production (Chipps & Bennett, 2000;Patwa, Christensen, Lasenby, Webster, & MacKay, 2007). An alternative theme is that pelagic production serves as a seasonal resource subsidy or provides a limiting nutrient for a foundational benthic energy pathway. The potential for benthic resources to be the dominant energy source for mysids, and the role Mysis may play in moving, distributing, and cycling benthic-derived energy and nutrients at the sediment-water boundary and between benthic and pelagic habitats in lakes could be far greater than we thought, and remains a hypothesis to be tested.

| RE S E ARCH QUE S TI ON S AND HYP OTHE S E S
Below, we identify a series of important questions and hypotheses that we believe will advance our understanding of the role of benthic habitat to Mysis ecology. We organised the questions under three themes: (1) what drives decisions to migrate or not; (2) the importance of benthic habitat for Mysis assessment; and (3) how benthic resources may impact Mysis dynamics in a broader ecosystem context.

| (2) How much does the presence of benthic Mysis affect estimates of abundance and production?
Fundamental questions remain about the extent to which pelagic-only sampling biases our inferences about Mysis population dynamics. phic systems, where benthic production in the littoral zone is the dominant source of system primary production (Ask et al., 2009;Sierszen, McDonald, & Jensen, 2003;Vadeboncoeur et al., 2003)?
(2b) What are the demographic differences between benthic-and pelagic-caught mysids, and how do such differences influence population production estimates? Assessing demographic differences between benthic and pelagic habitats across bathymetric depths, time of day, and seasons will provide more accurate population assessments, as well as contribute to questions and hypotheses about the mechanisms driving partial DVM (see questions 1a-c above) and to fill in missing biomass and production in food web models (see Food web and ecosystem effects above).

| (3) How does mysids' use of benthic resources affect their ecological resistance to system change and their ecosystem function?
Zooplanktivory by Mysis can be intense and alter zooplankton community structure and function (Lasenby et al., 1986;Nesler & Bergersen, 1991). However, decreases in pelagic zooplankton may not necessarily induce a negative feedback in Mysis abundance or growth because they can exploit benthic resources (Chipps & Bennett, 2000) and thus exhibit a high degree of adaptive capacity with changing conditions (McMeans et al., 2016). For example, in addition to withstanding intense competition for pelagic zooplankton (Bunnell et al., 2011), Mysis may also be able to resist declines in system productivity as a result of oligotrophication (Barbiero, Lesht, & Warren, 2012) and shifts in energy flow from green (pelagic) to brown (benthic) pathways induced by invasive species (Vanderploeg, Liebig, Nalepa, Fahnenstiel, & Pothoven, 2010). The ability of mysids to use both pelagic and benthic resources probably serves as a buffer against declines in either resource, such that they can maintain their role as an energy conduit across habitats and trophic levels during periods of system change (Johannsson et al., 2001).

| CON CLUS ION
The disproportionate focus on the pelagic phase of Mysis DVM in the published literature and the likely, but perhaps under-appreciated, role that benthic habitat plays in Mysis ecology suggests a complementary lens through which we should approach Mysis research. Partial DVM in freshwater mysids appears to be the norm rather than the exception. Thus, instead of the seemingly implicit assumption that the pelagic habitat is the most important habitat for Mysis and the benthic environment is simply a hiding place during daylight hours, we propose an alternative and perhaps provocative perspective that benthic habitat is equally, if not more, important than pelagic habitat to understanding Mysis ecology. Such a shift in perspective requires testing that focuses research more equally on the two habitats. A more-balanced perspective will result in a better understanding of the drivers of Mysis DVM behaviour and yield new insights into the ecosystem effects of animals, such as mysids, that rely on both benthic and pelagic habitats.

DATA AVA I L A B I L I T Y S TAT E M E N T
As this is a review paper there are no data.