Winter swarming behavior by the exotic cladoceran Daphnia lumholtzi Sars , 1885 in a Kentucky ( USA ) reservoir

We describe swarming behavior in the invasive cladoceran Daphnia lumholtzi Sars, 1885 in a Kentucky, USA, reservoir during winter 2017. The taxon is a highly successful tropical invader and has spread throughout the lower latitude systems in the USA since its discovery in 1991. Other than a few isolated reports, the abundance of D. lumholtzi is often <1 organism L. Previous studies indicate that D. lumholtzi is a largely thermophilic species often peaking in abundance in late summer after native daphnids are gone from the water column of lakes and reservoirs. Prior to our study, there have been no published reports of swarming behavior by this species. We document the occurrence of massive swarms (>10,000 organisms L) of sexually reproducing females of this exotic cladoceran at water column temperatures <10 °C.


Introduction
The large cladoceran, Daphnia lumholtzi (Sars, 1885), native to Africa, Asia and Australia, was discovered in the USA in the early 1990s and documented from several locations in the southeastern USA (Havel and Hebert 1993), eventually spreading across much of the southern portion of the US and into the Laurentian Great Lakes (Tudorancea et al. 2009).Daphnia lumholtzi is a highly successful invader given its wide distribution in the continental US (Havel and Shurin 2004;Havel et al. 2002Havel et al. , 2005;;Walker et al. 2013).Rapid expansion in habitats outside of its native range in South America have also recently been documented (Kotov and Taylor 2014;Sousa et al. 2016).Highest density and biomass of D. lumholtzi are usually observed in mid to late summer (Work and Gophen 1999a;Havens et al. 2000;Lennon et al. 2001;Havel and Graham 2006) when native daphnids are rare or absent from the water column (Havens et al. 2011).Despite its wide distribution, there is little evidence that this taxon has displaced endemic cladocerans (Work and Gophen 1999a;Havens et al. 2011;Sousa et al. 2016).Typically D. lumholtzi only achieves abundances of ca. 1 individual L -1 and is rarely found at concentrations higher than endemic daphnids (Havens et al. 2011).Daphnia lumholtzi exhibits long carapace spines that are believed to deter predation by juvenile fish (Swaffar and O'Brien 1996).Both head spines and tail spines have been observed to increase in response to predator kairomones (Sorensen and Sterner 1992;Dzialowski et al. 2003) and increasing temperatures (Yurista 2000).Large, monospecific aggregations of zooplankton (i.e., swarms) have been reported for a number of freshwater and marine species (Young 1978;Ambler 2002 and references therein).Concentrations of organisms are typically several orders of magnitude higher in these swarms compared to other locations and are confined to the top strata of the water column.Swarms are often located in the littoral region of the water bodies and may be several meters across.Typically, zooplankton swarms involve masses of organisms swimming in the same direction, at the same speed.Such swarms are believed to be a defense against predation, by acting to confuse predators and to decrease the probability of being captured (Jensen et al. 1998;Ambler 2002).Swarms have been recorded for a number of species of cladocerans (Ratzlaff 1974;Jakobsen and Johnsen 1988;Young and Taylor 1990), particularly daphnids (Young 1978;Mitchell et al. 1995).However, to date there are no published reports in the scientific literature of swarming behavior by D. lumholtzi in either its native or its invaded range.
This study reports an incidence of extraordinarily large D. lumholtzi swarms during winter in a Kentucky, USA reservoir.This event was unique from historic reports of D. lumholtzi in that a) it is the first scientific report, to our knowledge, of swarming behavior in D. lumholtzi b) the swarms displayed the highest densities of D. lumholtzi reported thus far in the scientific literature and c) the swarms occurred during winter, at temperatures far below the expected range for D. lumholtzi.

Site description
Nolin Reservoir (commonly known as Nolin River Lake) is a highly dendritic reservoir located in Edmonson, Grayson and Hart counties, Kentucky, USA (Figure 1).Surface area of the reservoir is approximately 23.4 km 2 and maximum reservoir depth is approximately 28.3 m and 20.7 m during summer and winter pool, respectively.The drainage area of Nolin Reservoir is approximately 1,820 km 2 and is fed primarily by inflow from the Nolin River in the northeast corner of the reservoir.The reservoir dam regulates outflow in the southwest corner of the reservoir and is operated by the U.S. Army Corps of Engineers for the primary purpose of flood control.Water residence time for Nolin Reservoir varies seasonally.The Kentucky Division of Water has classified Nolin Reservoir as a stable eutrophic system in the latest lake assessment.

Sample collection
On January 3, 2017 swarms in calm, littoral areas of Nolin Reservoir were observed from the shore (Figure 2), along with audible water turbulence.Whole water samples were collected mid-day on January 4, 2017 at two swarm sites (swarm 1: 37.280 °N, -86.229 °W, swarm 2: 37.279 °N, -86.247 °W) for the analysis of both zooplankton and phytoplankton.Concurrently, triplicate whole water samples were collected at a site outside of the swarms, near Nolin Reservoir dam (USGS station 03310900), for additional phytoplankton analyses.Samples from USGS station 03310900 were preserved with Lugol's solution immediately following collection.A multiparameter water-quality sonde, also stationed at USGS site 03310900, recorded water temperature, chlorophyll-a, pH, dissolved oxygen and turbidity at 15-minute intervals both before and after the swarm events took place, according to guidelines described in Wagner et al. (2006).Phytoplankton and zooplankton samples were shipped overnight to BSA Environmental Services, Inc. in Beachwood, Ohio, USA.The zooplankton, phytoplankton, and multiparameter water-quality data are reported in Cherry et al. (2017) at https://doi.org/10.5066/F7BZ657V.

Identification and enumeration
Upon arrival at the laboratory, separate aliquots were taken from the whole water swarm samples for zooplankton and phytoplankton analyses.Phytoplankton from swarm samples were preserved with Lugol's solution, gently homogenized, and subsampled into Utermöhl chambers.Counts and taxonomic identifications were then performed using an inverted light microscope (Leica DMIL) at 800X.Phytoplankton from USGS station 03310900 (non-swarm samples) were analyzed using the membrane filtration technique (McNabb 1960) at 630X magnification (Leica DMLB) with a count threshold of 400 natural units (colonies, filaments and unicells).Phytoplankton biovolume for both swarm and non-swarm samples was estimated based on geometric shapes (Hillebrand et al. 1999).
100 mL from each zooplankton aliquot was filtered through 35 µm mesh in order to transfer organisms into ethanol.Zooplankton samples were then homogenized using a magnetic spinner at low speed and subsampled into Utermöhl chambers.Samples were counted using inverted light microscopes (Wilovert) at 100X to determine total abundance (organisms L -1 ).After total abundance was determined, 100 individuals were examined from each sample and were classified into four groups: males, females without eggs, females with eggs and females with ephippia.For females with eggs, the numbers of eggs were counted, and a mean was computed for number of eggs per female.Ten individual specimens from each sample were measured for body length (excluding head and tail spines), and biomass estimates were calculated based on established length/width relationships for Daphnia ambigua Scourfield, 1947found in McCauley (1984).Length was compared between the two swarms using a t-test.

Results
Microscopic analysis confirmed the presence of D. lumholtzi in the swarm samples.No other zooplankton species (crustaceans or rotifers) were observed within the swarm.Swarm specimens exhibited relatively reduced head and tail spines generally not exceeding body length (Figure 3), unlike D. lumholtzi specimens typically observed in summer samples where head and tail spines can greatly exceed body length.Densities of D. lumholtzi in the two wholewater swarm samples were 13,724 L -1 and 12,400 L -1 , for swarm 1 and swarm 2 respectively.Swarm populations were dominated primarily by female specimens (Table 1).Within the population of females, females without eggs (11,665 L -1 and 9,920 L -1 ) were more abundant than females with eggs (1,510 L -1 and 2,480 L -1 ).Females with eggs averaged approximately 7-8 eggs per clutch.Females with resting eggs (ephippia) (412 L -1 ) and male specimens (137 L -1 ) were only present in swarm 1. Mean length of D. lumholtzi was similar between the two swarm samples, with swarm 1 being slightly smaller on average (1.70 mm) than swarm 2 (1.80 mm) (t(42)=0.76,p > 0.05), mostly due to smaller egg-bearing females.Phytoplankton populations inside the swarms averaged nearly an order of magnitude less than outside of the swarms (Table 2) and displayed limited diversity, consisting almost entirely of the Cyanobacterium Chroococcus microscopicus.The diversity of phytoplankton in the non-swarm samples was high, averaging 18 species per sample.Non-swarm phytoplankton populations included C. microscopicus, but also contained several species of diatoms (Bacillariophyta) and green algae (Chlorophyta), as well as representatives from Euglenophyta, Cryptophyta and Chrysophyta.
Beginning in mid-December, 2016, chlorophyll-a concentrations began increasing and peaked at approximately 6 µg L -1 in late December (Figure 4A), indicating that an increase in phytoplankton biomass preceded the D. lumholtzi swarm.On January 3, 2017 (concurrent with the initial visual observation of D. lumholtzi swarms) chlorophyll-a concentrations rapidly decreased to about 1.5 µg L -1 .Turbidity rapidly increased (indicative of swarm formation) from approximately 12 FNU on December 31, 2016 to a peak of approximately 25 FNU on January 6, 2017 (Figure 4B).Temperatures in the top 5 meters of the water column were approximately 8 °C during the swarm event.Sonde data also indicated that the surface of the water column during the swarm was circumneutral (pH ~7.7) and well oxygenated (~9.2 mg L -1 ).

Discussion
Since D. lumholtzi became established in North America in the early 1990s, its seasonal dynamics and distribution patterns have been in accordance with the extensive literature and expectations for a tropical invader in cooler, higher latitude ecosystems.
Experimental studies indicate that the physiological condition of D. lumholtzi at temperatures greater than 25 °C is more robust than native daphnids (e.g.Work and Gophen 1999b;Lennon et al. 2001).Although there have been previous observations of D. lumholtzi in a North American reservoir during winter months (Sorensen and Sterner 1992, temperatures ranging from 11-20 °C), peaks in D. lumholtzi abundance are more commonly reported to occur during late summer when water column temperatures are greatest (Work and Gophen 1995;East et al. 1999;Havens et al. 2000;Havel and Graham 2006).The late summer period in North American lakes and reservoirs corresponds to a time of low algal food quality and an otherwise poorly occupied niche by native daphnids (Lennon 1999;Beaver et al. 2014).Considering the broad agreement between previous reports on the timing of maximum abundance of D. lumholtzi, the swarms observed in this study were highly unusual in the timing of their occurrence.Environmental factors (both biological and physical) contributing to zooplankton swarm development are poorly understood.Several drivers of swarming behavior in aquatic invertebrates have been examined (Folt and Burns 1999), including predator avoidance (Pijanowska and Kowalczewski 2007), food availability (Jakobsen and Johnsen 1988) and opportunity for sexual reproduction (Young 1978).Experimental studies show that swarming behavior by daphnids can occur in response to kairomones (chemical signals emitted by predators and received by prey) from fish and invertebrate predators such as Chaoborus (Pijanowska 1994;Jensen et al. 1998;Kvam and Kleiven 1995).Fish kairomone production is also associated with increased spine length in D. lumholtzi when organisms are exposed in a pre-embryonic state (Dzialowski et al. 2003;Yurista 2000).
Like most freshwater zooplankton, D. lumholtzi produces resting eggs prior to diapause.Resting eggs are encased into a hard ephippium (containing two fertilized eggs) that is often deposited into the sediments to hatch once more favorable conditions are encountered.Daphnia lumholtzi produces an order of magnitude more ephippia than native daphnids (Acharya et al. 2006).Environmental cues that can trigger production of ephippia can include photoperiod, or temperature.Fish kairomones are also associated with higher production of resting eggs in mature females (Alekseev 2004;Slusarczyket al. 2005).Ephippia frequently hatch en masse based on favorable environmental cues (Hairston et al. 1990;Hairston and Cáceres 1996), which results in increases in water column abundance.Given the unusual seasonal timing of the observed events, it is possible that a strong and sudden environmental cue (or suite of cues) induced D. lumholtzi ephippia to emerge (possibly from ephippia deposited during a previous, undocumented swarm event).Temperature increase may be an environmental cue breaking diapause; limited distribution of D. lumholtzi in northern lakes and reservoirs where eggs in sediments are not subjected to elevated water temperatures supports this notion (Lennon 1999).However, in this case it is unlikely that temperature played a role.The increase in chlorophyll-a, indicative of increased phytoplankton population growth, in the weeks preceding the swarm events is one possible environmental trigger that could explain the strange timing of the swarms.
Daphnia lumholtzi has had some invasion success in reservoirs at higher latitudes in North America, despite lower water temperatures (Havel and Shurin 2004).Kentucky Reservoir (commonly known as Kentucky Lake), located only about 250 km from Nolin Reservoir, experiences annually recurring populations of D. lumholtzi (Levine and White 2009), albeit at relatively low concentrations.A swarm of D. lumholtzi of similar composition to the one in this study was described from Lake Cumberland (also a reservoir), located approximately 145 km from Nolin Reservoir, in October 2009 (Steinitz-Kannon and Lenca 2013), however this event was poorly documented.Frisch et al. (2013) found that several genetic strains of D. lumholtzi are present in the United States, which may be adapted to a range of temperatures along a latitudinal gradient.It is likely that the same strain of D. lumholtzi occurs in reservoirs nearby to Nolin, as it has been shown to survive short-term human-mediated transfer between water bodies (Havel and Stelzleni-Schwent 2000).This implies that similar events to the one observed in this study are possible elsewhere in the region.If dispersal of D. lumholtzi continues to grow in scope, it is likely that similar swarm events could occur elsewhere in the country.

Conclusions
The abundance of D. lumholtzi reported in the current study far exceeds any published values.The occurrence of such large swarms of D. lumholtzi under the observed environmental conditions has implications for future invasions.The strategy of wintertime emergence and dropping of genetically nuanced ephippia-during a period of low predation risk and limited competition from other daphnidscould represent a previously undetected route for successful invasion of new habitats.However, it is difficult to draw conclusions based on a single, isolated incident.Further monitoring of similar sites in the future is needed to place this event in a broader ecological context.

Figure 1 .
Figure 1.Map of Nolin Reservoir showing locations of dam (bold black line), littoral swarm sample sites (swarm 1 and swarm 2) and non-swarm site (03310900, closest to dam).Inset shows location of Nolin Reservoir within the state of Kentucky and within the continental USA.Scale bar refers to reservoir map.

Figure 2 .
Figure 2.Macroscopic view of D. lumholtzi swarm 2 in Nolin Reservoir (A) on January 3, 2017, upstream of the dam tower, photo by D. Rodgers (B) on January 5, 2017, downstream of the dam tower, photo by D. Rodgers and (C) on January 4, 2017 in a collected whole water swarm sample, photo by J. Thomason.

Figure 3 .
Figure 3. Photomicrographs of D. lumholtzi from swarm 1 including (A) female without eggs (B) female with eggs (C) female with ephippium and (D) male.Photos by T. Renicker.