LIMNOLOGY and Limnol. Oceanogr. 62, 2017, 1096–1110 OCEANOGRAPHY VC 2017 Association for the Sciences of Limnology and Oceanography doi: 10.1002/lno.10488

Colonization dynamics of the invasive predatory cladoceran, Bythotrephes longimanus, inferred from sediment records

Donn K. Branstrator,1* Ashley E. Beranek,1 Meghan E. Brown,2 Leif K. Hembre,3 Daniel R. Engstrom4 1Department of Biology, University of Minnesota Duluth, Duluth, Minnesota 2Department of Biology, Hobart and William Smith Colleges, Geneva, New York 3Department of Biology, Hamline University, St. Paul, Minnesota 4St. Croix Watershed Research Station, Science Museum of Minnesota, Marine on St. Croix, Minnesota Abstract Despite the central role of colonization to invasion biology, little is known about the early dynamics of founder populations because this time period is often unobserved and short-lived. This study documents with high resolution the early population ecology of an invading freshwater zooplankton, Bythotrephes longi- manus (Cladocera : Cercopagidae), by measuring accumulation rates of its caudal spine in sediments. Using dated (210Pb and 137Cs) sediment cores from four spatially distinct sites in Island Lake Reservoir (Minnesota, USA), we describe its first presence, early distribution, growth trajectory, and early impacts on prey. The sedi- ment record shows that B. longimanus first appeared and was widely distributed in the lake in 1982 (6 2 yr, standard deviation), 8 yr before its first detection in the water, making it one of the earliest documented invasions in North America and suggesting that ecosystems may serve as dispersal hubs for years prior to detection. Logistic growth models describing spine accumulation rates show that B. longimanus required about two decades to achieve an annual equilibrium (K). Prolonged buildup to K may owe to several factors, including accumulation of a sufficiently large bank of resting eggs, the obligate overwintering life stage. Early exponential growth was incongruent with the presence of a lag phase. Post invasion, Daphnia mendotae became proportionally the most abundant daphniid in the lake, but the timing of the switch in prey species composition coincided more with the proliferation of B. longimanus density and its attainment of K than with its arrival to the lake.

Range expansions by aquatic invasive species are increas- Zanden et al. 2010; Hoffman et al. 2011), and to evaluate ingly commonplace (Ruiz et al. 2000; Garcıa-Berthou et al. adaptive responses of resident species to the invader (Bou- 2005) prompting ecologists to investigate how and when dreau et al. 2013; Nunes et al. 2014). invaders overcome barriers to dispersal and establishment Despite the importance of colonization, it is challenging (Mack et al. 2000). One critical, yet understudied area of to study because invasions are typically unplanned, and research involves elucidating the population dynamics of invaders often arrive in small and isolated propagules that aquatic invaders during colonization (Crooks 2005). During go undetected until they have attained sizable densities and this stage, propagules first encounter a constellation of abiotic the stage is over (Crooks 2005). Lag phases, in particular, and biotic barriers to survival and reproduction that deter- prove difficult to characterize because long periods of data mine establishment success and the timing of environmental collection are often needed at a time when invaders are impacts. This transient period is characterized by a variety of themselves rare and possibly widely dispersed (Crooks 2005; time-sensitive events including (1) when the first propagules Aikio et al. 2010). In cases where lag phases have been quan- arrive, (2) the shape and magnitude of early population tified, substantial variation in duration exists and can be growth trajectories, and (3) the time between first arrival and more than a century (Rilov et al. 2004). proliferation. Knowledge of colonization dynamics can The majority of studies on invasive species are short-term improve the ability to identify when invader densities and and offer no temporal context (Strayer et al. 2006). impacts may stabilize (Crooks 2005; Strayer et al. 2006), to Although public records (e.g., the U.S. Geological Survey construct timelines for containment and eradication (Vander Nonindigenous Aquatic Species database) can offer valuable spatial and temporal data relevant to understanding this *Correspondence: [email protected] early timeframe of propagation, these resources often lack

1096 Branstrator et al. Bythotrephes colonization dynamics the time-continuous information needed to quantify growth easily distinguishable from native zooplankton subfossil trajectories. Detection and study of early temporal dynamics remains, and consequently their accumulation in sediments of colonizing plankton and nekton are particularly difficult can be used to establish an invasion year (Keilty 1988; Brans- owing to organism dilution, mobility, size, and phenology trator et al. 2006) and to track abundances over time (Hall (Edlund et al. 2000; Hoffman et al. 2011). and Yan 1997). A well-recognized approach to the limitations of direct Our study focused on the invasion of B. longimanus into observation is to reconstruct environmental histories from dat- Island Lake Reservoir, Minnesota (USA). Island Lake Reser- ed (e.g., via 210Pb) lake-sediment cores. This approach, termed voir was first reported to be infested with B. longimanus in paleolimnology, has enabled study of time-continuous records 1990 (based on its presence in the water column), making it of a wide variety of past events (Smol 1992) including lake the earliest confirmed inland lake invasion in the U.S.A. eutrophication, acidification, species invasions, and climate (Gravelle 1990). We addressed the following questions change (Hairston et al. 1999; Jeppesen et al. 2001; Leavitt et al. regarding the invasion of B. longimanus into Island Lake Res- 2009). Cladocera are among the best preserved zooplankton in ervoir: (1) Was there temporal correspondence between its lake sediments (Korhola and Rautio 2001), and numerous stud- year of first detection in lake sediments and the year of first ies have used their subfossils to reconstruct past food-web detection in the water column?, (2) Was its initial appear- dynamics, including invasion events (Kerfoot 1981; Hairston ance in the ecosystem lake wide or spatially limited?, (3) et al. 1999; Jeppesen et al. 2001). What was the shape and magnitude of its annual population In this study, we used paleolimnology to investigate the growth during colonization?, (4) Did the population demon- colonization dynamics of Bythotrephes longimanus (spiny strate a lag phase during colonization?, and (5) Were there water flea). Native to Eurasia, B. longimanus is a predatory evident changes in the abundance of its prey that were cladoceran that was first documented in North America in synchronous with particular periods of colonization? the Laurentian Great Lakes during the early 1980s (Sprules et al. 1990; Johannsson et al. 1991). By 2010, approximately Methods 200 inland lakes in Canada and the United States had been Study site invaded (Yan et al. 2011). As a carnivore, B. longimanus Island Lake Reservoir (Fig. 1; hereafter Island Lake), is a reshapes the abundance and diversity of native crustacean multibasin impoundment of the Cloquet River with surface zooplankton assemblages through its consumptive and non- area 5 30 km2, maximum depth 5 29 m, average summer consumptive effects on prey that consists largely of daphniid Secchi transparency 5 2 m, and average summer chlorophyll 21 and bosminid cladocerans (Pangle et al. 2007; Azan et al. concentration 5 8.5 lgL (Branstrator et al. 2006). The lake 2015). Among inland Canadian lakes, consumption by inver- was created in the 1890s when a low-rise dam, built to retain tebrate predators as a guild was 300% higher when spiny water for logging purposes, resulted in the coalescence of water flea was present than when it was absent (Foster and several pre-existing small lakes and wetlands. Island Lake has Sprules 2009). The high caloric demand of B. longimanus is some of the largest reported B. longimanus densities in North 23 explained in part by its fast growth and maturation rates America, eclipsing 100 individuals m (Brown et al. 2012). compared to native predatory zooplankton (Branstrator Sediments 2005). Although a vast amount of research has been con- To evaluate potential temporal and spatial variation in ducted thus far on this invader’s ecology (Yan et al. 2011), the colonization dynamics of B. longimanus, sediment cores little is known about its early population dynamics during were collected from four locations that included two sam- the colonization stage of lake invasions. pling sites in each of the lake’s two main (East and West) B. longimanus represents an excellent model organism to basins (Fig. 1). Sediments were collected using a piston corer study a species’ invasion pattern using lake sediment records. equipped with a 7-cm diameter polycarbonate core barrel First, it produces a long caudal spine composed of durable and operated from the lake surface by rigid drive rods. Sites chitin that preserves well in lake sediments (Keilty 1988; 1, 2, and 3 were cored in March 2009, through ice. Due to Hall and Yan 1997). In Gulspettvann Lake, Norway, B. longi- logistic constraints, site 4 was cored 16 months later, in June manus spines were recovered from 22 cm deep in a sediment 2010, during open water from an anchored boat. A portable core (Nilssen and Sandoy 1990). Second, during ecdysis the extruder (Glaser and Griffith 2007) was used to section the exoskeleton covering the caudal spine is not shed, even cores in the field which was typically in 0.5-cm increments though the spine lengthens by addition of intercalary seg- for the upper 10–15 cm and in 1.0-cm increments thereafter. ments during the first two molts (Branstrator 2005). Accord- Sections were packaged in airtight plastic containers and ingly, numbers of caudal spines can be used to track relative stored at 38C until analysis. At sites 1 and 2, two cores were densities of whole organisms in a straightforward fashion, collected within 2–3 m of one another. One core was ana- because a single spine in sediments reflects the historical lyzed for loss-on-ignition (LOI) and dating, described below, presence of a single individual. Third, caudal spines are and the other was analyzed for subfossils and Daphnia

1097 Branstrator et al. Bythotrephes colonization dynamics

47.05 East basin West basin A C 1 3 B N) o D E 2

4 F Latitude ( 47.00 N 1 km

92.20 92.15 92.10 o Minnesota° Longitude ( W)

Wisconsin

Duluth-Superior Harbor

Fig. 1. Regional location and map of Island Lake Reservoir, St. Louis County, Minnesota. Letters indicate plankton net tow collection sites and circled numbers indicate sediment core collection sites. Sediment core collection site coordinates (depth) are: site 1, 47.038 N, 92.118 W (14 m); site 2, 47.028 N, 92.178 W (14 m); site 3, 47.038 N, 92.148 W (14 m); site 4, 47.018 N, 92.188 W (8 m). resting eggs (ephippia). The remaining material from the total activity measured on each interval, with supported cores used for LOI and dating analyses at sites 1 and 2 was activity estimated from the asymptotic activity at depth. used secondarily to verify the first occurrences of B. longima- Dates and sediment accumulation rates were determined by nus. Our experience working with the cores at sites 1 and 2 the constant rate of supply (c.r.s.) model with dating preci- indicated that after the material had been removed for LOI sion (6 1 standard deviation) calculated by first-order propa- and dating, there would have been cladocerans remaining gation of counting errors (Appleby 2001). These error for analysis; hence, only single cores were analyzed for LOI, calculations do not account for violations of dating-model dating, and cladocerans at sites 3 and 4. At site 3, generally assumptions (conformable deposition, constant 210Pb flux to 30–50% of the material per section was searched for clado- the core site), and hence represent a minimum estimate of cerans. At site 4, generally 70–80% of the material per sec- dating uncertainty. tion was searched for cladocerans. The difference was due to Cesium-137 was measured as a supplemental dating mark- lower sediment density (higher water content) at site 3 er on two cores to identify sediments deposited during the which necessitated more volume for LOI and dating 1963–1964 peak in atmospheric nuclear testing. Freeze-dried analyses. subsamples were sealed in polypropylene sample tubes and Dry-density (dry mass per volume of fresh sediment), measured for 1–3 d using an Ortec high-resolution germani- water content, and organic content were determined by um well detector and multichannel analyzer. At sites 1 and standard LOI methods (Dean 1974). Sediment subsamples 2, where separate cores were collected for dating and cladoc- were dried overnight at 1058C and then burned at 5508C for eran analysis, dates in the cores used for cladoceran recon- 4 h. Mass measurements were made of the wet samples and struction were derived from equivalent depths in the after each heating on an electronic analytical balance. corresponding dated cores. Cores from all sites were dated with 210Pb, and to verify Cladocerans were enumerated on all core sections dating the resulting chronologies, one core from each basin was from 1970 to present. This included all of the material in also dated with 137Cs (sites 1 and 2). For 210Pb dating, freeze- cores designated for zooplankton analyses from sites 1 and 2 dried subsamples from select intervals (18–20 per core) were and the fractions remaining after other analyses (noted analyzed for 210Po, a decay product of 210Pb, by isotope- above) for sites 3 and 4. To reduce damage to B. longimanus dilution (209Po), alpha spectrometry. Excess (unsupported) caudal spines, sediments were not sieved, except for site 3 210Pb was calculated by subtracting supported activity from where they were pre-screened on 53-lm mesh to expedite

1098 Branstrator et al. Bythotrephes colonization dynamics searching. Samples were diluted with distilled water and The numbers of B. longimanus spines and Daphnia ephip- manually searched for caudal spines under a dissecting pia in core sections were corrected for sediment focusing microscope (Leica MZ125). To reconstruct the number of B. and converted to a flux to the sediments, referred to here as longimanus individuals, we followed Hall and Yan (1997). an accumulation rate (number m22 y21), as: The caudal spine of B. longimanus ranges from about 5– ðD3S310; 000Þ=ðÞCF=AF ; (1) 9 mm in length, but is often broken in sediment material (Hall and Yan 1997). This reduces the chance that spines where D 5 total subsample spine or ephippium density in a bridge cuts in core sections and bias results. When we sec- core section (number cm23), S 5 core section-specific linear tioned the cores, we never observed spine material sticking sedimentation rate (cm y21), CF 5 core-specific 210Pb flux vertically out of the sediment suggesting that spines (pCi cm22 y21), and AF 5 mean atmospheric 210Pb flux (pCi and spine fragments were generally lying flat which is how cm22 y21) which was assumed to be 0.5 pCi cm22 y21 based spines settle on the bottom of aquaria in the laboratory on 210Pb deposition monitoring at two Minnesota National (D.K. Branstrator pers. observ.). Atmospheric Deposition Program sites (Lamborg et al. 2012). Daphnia are a preferred prey of B. longimanus, and shifts We used accumulation rates of B. longimanus caudal in their abundance and species composition are commonly spines and Daphnia ephippia as proxies for their relative noted post invasion (Azan et al. 2015). We reconstructed the annual population sizes (sum of total number of animals liv- number of daphniids in sediments at sites 1 and 2 by num- ing during a year). Interannual trajectories in B. longimanus ber of ephippia. Diagnostic characteristics of ephippia were accumulation rates in sediments were modeled with the used to assign them as Daphnia mendotae, Daphnia pulex,or logistic equation: Daphnia retrocurva (Kerfoot et al. 2004). Although the abun- ÀÁ 2rt dance of daphniid ephippia in sediments is a reasonable Nt5K= 1 1be ; (2) proxy of their relative abundance in the water column, it is not a 1 : 1 correspondence because it represents only the using a nonlinear least squares approach and Gauss-Newton individuals that invested in sexual reproduction (Jankowski algorithm where Nt 5 population size (annual accumulation 22 21 and Straile 2003; Manca et al. 2007). rate, as spines m y ) at a given time t, K 5 carrying capac- 22 To make comparisons between the sediment record and ity (equilibrium annual accumulation rate, as spines m 21 contemporary data regarding the spatial distribution and rel- y ), r 5 growth rate (increase in annual spine accumulation 21 ative abundance of B. longimanus in Island Lake, we report rate, as y ), and b 5 a constant. Both Nt and t were known. previously collected but unpublished estimates of water col- K was assigned for each site as the average of the yearly accu- umn densities of B. longimanus based on open-water net mulation rates during 2002–2009, which is the period when tows. Collections were made with plankton nets (0.5-m accumulation rates began to stabilize (see Fig. 4). The b and diameter round mouth, 1.5-m length, 500-lm aperture) r parameters were estimated by the best-fit model. For model drawn from 1 m above the sediment to the surface. Collec- input, the first year was 1980 when all but one sediment tions were done at night to ensure that B. longimanus, which core indicated that no B. longimanus were present in the nocturnally migrates up from its diurnal position near the lake. The exception was at site 1 where B. longimanus was bottom in Island Lake (Brown et al. 2012), would be cap- noted by a single caudal spine in a core section dated to tured. Collections were made in 2002, 2003, and 2004 from 1978. For modeling purposes, this isolated spine was May to November on a monthly (sometimes weekly) basis; ignored. Site 4 was not modeled with Eq. 2 because temporal resolution of the sediment strata was poor due to high inor- and in 2010 once each during June, July, and September. ganic content and dry density that reduced the linear accu- Collections were preserved (70% ethanol) and B. longimanus mulation rate and years per section (see below). were counted under a dissecting microscope (Leica MZ125). The residual between the observed accumulation rates Analysis based on subfossil evidence and the predicted accumulation B. longimanus spines that are damaged may be more diffi- rates based on Eq. 2 may be interpreted to represent excur- cult to reconstruct into whole individuals if diagnostic por- sions from a long-term theoretical growth trajectory for a tions of the spine are missing (e.g., the mid-spine kink), and colonizing population. Excursions may be caused by any this could bias interpretation of the subfossil record. We number of internal and external factors. We used correlation therefore tested for changes in the proportions of whole analysis to assess synchrony (timing and magnitude) of these spines through time for sites 1, 2, and 4; site 3 was excluded excursions, without lags, between sites. To do this, a Pearson because the core sections had been processed with sieving product-moment correlation coefficient (r) was estimated for (see above). Trends were assessed by simple linear regression sets of residuals matched to the same year for pairs of sites. of percentage whole spines as a function of estimated year To test the significance of r being different from zero, we (210Pb), and tested for slope 5 0 with analysis of variance used t-tests with Bonferroni adjusted p values for multiple (ANOVA) and the F-ratio using SYSTAT 13.0. comparisons as we compared the data sets in pairs from all

1099 Branstrator et al. Bythotrephes colonization dynamics three sites (1 vs. 3, 1 vs. 2, and 2 vs. 3). One outlier associat- flatten, suggesting the possibility of sediment mixing. How- ed with site 2 was not included in the statistical analysis. ever, the flattening is more likely a consequence of 210Pb We also evaluated temperature as an environmental cor- dilution by accelerating sedimentation, which is supported relate of the residual differences in B. longimanus accumula- by the greater depths over which 210Pb was relatively uni- tion rate between subfossil observations and model form (10 cm in core 2 and 6 cm in core 3) compared to predictions. Long-term air temperatures were obtained from mixed depths in other area lakes—typically 1–2 cm, as indi- public records of the U.S. Department of Commerce, cated by 7Be measurements (D.R. Engstrom unpubl.). Fur- National Oceanic & Atmospheric Administration, for the ther, sharp changes in the concentrations of cladocerans Duluth International Airport station (Latitude 46.848 N provide additional evidence that these upper sediments had 92.188 W). May, June, July, and August values were averaged not been homogenized. Finally, the match between 137Cs yearly for the 30-yr period from 1980 to 2009. This 4-month dating and the 210Pb chronology in core 2 (see below) also window captures the most important period of annual supports the interpretation of accelerating sedimentation recruitment from resting eggs and parthenogenetic growth and not sediment mixing. of B. longimanus in Island Lake (Brown et al. 2012). The 30- Age-depth profiles for the two East basin cores (1 and 3) yr average temperature was assigned as the baseline, and dif- were almost identical, with dating differences on the order ferences in yearly temperature from the baseline were com- of 2–3 yr for the same core depth (Fig. 2). The oldest dates puted as residuals. Simple linear regression was used to extended back to the 1830s, well before reservoir construc- assess the relationship between the residuals in B. longima- tion and land-use change by European settlers in northeast- nus accumulation rate (see above) as a function of residuals ern Minnesota. The period of interest for this study (1980– in air temperature over the 30-yr period at each site. We 2010) was contained in the upper 7.5 cm of both cores, pro- tested for slope 5 0 with ANOVA and the F-ratio. We viding an average temporal resolution of 2 yr per core sec- assessed linear trends in residuals of B. longimanus and tem- tion. Dry mass accumulation rates (DMAR) were also similar perature paired to the same year, and after lags of 1 yr between the two East basin cores and relatively constant dur- 22 21 imposed in each direction, to account for uncertainty in ing this 30-yr time window (0.02–0.03 g cm yr ) (Fig. 2). 210Pb dating. Sediment accumulation rates in the two West basin cores Owing to the small sample sizes, we performed nonpara- were substantially different from those in the East basin— metric tests (Kruskall-Wallis H-test, Mann-Whitney U-test) to much faster at site 2 (19 cm accumulation since 1980) and evaluate among-site differences in B. longimanus sediment slower at site 4 (3.5 cm since 1980), with DMAR doubling accumulation rates and among-site differences in B. longima- during this period in core 2 but remaining relatively steady nus densities estimated from the net tow collections. in core 4 (Fig. 2). Precision (6 1 standard deviation) of the 210Pb dating decreases with depth from about 6 1.3 yr at the Results tops of the cores to 6 2.0 yr at the 1980 horizons. These errors vary somewhat among cores and are minimal esti- Core lithology and dating mates of dating uncertainty, propagated from analytical pre- Notable variation in sediment lithology was evident with- cision. Actual reliability of dates is best assessed by in and among cores. At all sites, the uppermost sediment confirmation from an independent dating method, in this had the highest organic content (22–28%) which decreased case 137Cs. down-core by 2–4% by 1980 at sites 1–3, and to 15% by The 137Cs results generally confirmed the 210Pb chronolo- 1980 at site 4 (Fig. 2). Based on the full length of the LOI gies. Peak atmospheric deposition of 137Cs was nominally profiles, it appears that cores from the East basin sites (1 and 1963–1964, just before the global treaty banning atmospher- 3) were collected in pre-existing lake basins, as there was no ic nuclear testing. At site 1, 137Cs produced a sharp peak at obvious discontinuity in sediment lithology that would oth- 9.5–10 cm, an interval corresponding to a 210Pb date of erwise indicate contact with a terrestrial surface or river 1964–1967, indicating excellent correspondence between the channel deposits (Fig. 2). This was not the case for the West two dating isotopes (Fig. 2). At site 2, a somewhat wider basin sites (2 and 4), where organic lake sediments transition 137Cs peak was evident between 25 and 28 cm, which to highly inorganic material around 48 and 12 cm depth, according to 210Pb is between 1953 and 1963. This suggests respectively. the 210Pb dates are slightly biased old at site 2, but they are Total 210Pb activities decreased monotonically with depth still within the analytical precision of 210Pb chronology, in all cores from surface values of 11–27 pCi g21 to relative- which at that depth horizon was about 6 3 yr (1 standard ly constant background (supported) values of 0.42–0.77 pCi deviation). g21 (Fig. 2). The trends in the upper portions of cores 1 and 4 were near-exponential, indicating minimal disturbance B. longimanus subfossils from bioturbation or core collection, whereas in core 2— The percentages of B. longimanus caudal spines recovered and to some extent, core 3—the activity profiles tended to whole vs. those reconstructed from fragments varied

1100 Branstrator et al. Bythotrephes colonization dynamics

Site 1 Site 2iteSite 3 S 4

Dry Density (g cm-3) Dry Density (g cm-3) Dry Density (g cm-3) Dry Density (gm cm-3) 0 0.2 0.4 0.6 0.8 1 1.2 1.4 0 0.2 0.4 0.6 0.8 1 1.2 1.4 0 0.2 0.4 0.6 0.8 1 1.2 1.4 0 0.2 0.4 0.6 0.8 1 1.2 1.4 A 0 0 0 % Organic 20 Dry Density 20 20 20

40 40 40 40

60 60 60 60 Core Depth (cm) 80 80 80 80

100 100 100 100 0 1020304050 0 1020304050 0 1020304050 0 1020304050 % Organic (LOI) % Organic (LOI) % Organic (LOI) % Organic (LOI) B 0 0 0 5 10 5 10 10 5 Pb 20

15 210 15 10 20 30 20 25 40 25 15 supported 30 50 30 35 35 20 Core Depth (cm) 60 40 40 25 45 70 45 50 80 50 30 0.3 1 10 30 0.3 1 10 30 0.3 1 10 30 0.3 1 10 30 Total 210Pb Activity (pCi g-1) Total 210Pb Activity (pCi g-1) Total 210Pb Activity (pCi g-1) Total 210Pb Activity (pCi g-1) 137Cs (pCi g-1) 137Cs (pCi g-1) 012345012345 C 0 0 0 5 5 5 5 137Cs Peak 10 (1963-64) 10 10 10 15 15 15 15 20 20 20 20 25 25 137Cs Peak 25 25 (1963-64) 30 30 30 30 35 35 35 Core Depth (cm) 35 Age 40 40 40 40 137Cs 45 45 45 45 50 50 50 50 0 25 50 75 100 125 150 175 200 0 25 50 75 100 125 150 175 200 0 25 50 75 100 125 150 175 200 0 25 50 75 100 125 150 175 200 210Pb Age (yr) 210Pb Age (yr) 210Pb Age (yr) 210Pb Age (yr) D 2020 2020 2020 2000 2000 2000 2000 1980 1980 1980 1980 1960 1960 1960 1960 1940 1940 1940 1940 1920 1920 1920 1920 1900 1900 1900

Pb Date 1900

210 1880 1880 1880 1880 1860 1860 1860 1860 1840 1840 1840 1840 1820 1820 1820 1820 1800 1800 1800 1800 0 0.02 0.04 0.06 0.08 0.1 0 0.05 0.1 0.15 0.2 0 0.02 0.04 0.06 0.08 0.1 0 0.05 0.1 0.15 0.2 Sediment Accumulation (g cm-2 yr-1) Sediment Accumulation (g cm-2 yr-1) Sediment Accumulation (g cm-2 yr-1) Sediment Accumulation (g cm-2 yr-1)

Fig. 2. Sediment lithology and dating profiles for cores: (A) percent organic matter and dry density; (B) total 210Pb activity; (C) 210Pb age and 137Cs activity; and (D) DMAR vs. 210Pb date. Error bars represent 6 1 standard deviation.

1101 Branstrator et al. Bythotrephes colonization dynamics

2010 throughout the sediment record until the top of the core where the rate increased and then dropped. At sites 1–3, where there was good temporal resolution of

Pb) the sediment core sections, we analyzed trends in accumula-

210 2000 tion rate residuals between sites as well as between accumu- lation rate residuals and air temperature residuals (Fig. 5). The Pearson product-moment correlation coefficients for the accumulation rate residuals were high between pairs of sites Site 1990 (Fig. 6): site 1 vs. site 3 (r 5 0.727, p 5 0.02), site 1 vs. site 2 1 (r 5 0.541, p 5 0.17), and site 2 vs. site 3 (r 5 0.630, p 5 0.06). 2 Estimated Year ( 4 The strongest relationship was between sites 1 and 3, where based on visual inspection of the data one can see that after 1980 2001 there was particularly tight correspondence in the tim- ing and magnitude of the excursions around K (Fig. 4). The 0204060 80 100 annual residuals around mean air temperature were not sig- Percentage of B. longimanus nificant linear predictors of B. longimanus accumulation rate caudal spines recovered whole residuals (Fig. 5) at any of the three sites for pairs of residuals matched to the same year or for pairs of residuals lagged by Fig. 3. Calculated percentages of B. longimanus reconstructed from the 1 yr in either direction (F-ratio, p 0.10 in every test). presence of caudal spines that were whole (unbroken) as a function of estimated year of deposition. B. longimanus spine accumulation rates, computed as aver- ages across all core sections from first occurrences to the considerably among sites 1, 2, and 4 (Fig. 3). The percentages youngest sections analyzed, were 1282 (site 1), 1260 (site 2), 22 21 of spines recovered whole were generally greater at the East 1498 (site 3), and 916 (site 4) individuals m y and were basin site (range 5 27–91%) than at either West basin site not statistically distinguishable (Kruskal–Wallis, H 5 2.00, (range 5 0–53%). The percentage of spines recovered whole p 5 0.57, df 5 3). Densities of B. longimanus from the plank- declined from shallow (younger) to deep (older) sediments at ton net tows for the 4 yr of data (2002–2004 and 2010) showed some degree of between-year and among-site varia- site 2 (ANOVA F1,26 5 23.37, p < 0.01), but not at site 1 tion (Table 2). However, there was no statistical difference in (ANOVA F1,13 5 0.04, p 5 0.86) or site 4 (ANOVA F1,5 < 0.01, p 5 0.97). the distributions of values grouped as East basin vs. West The first occurrences of B. longimanus in lake sediments basin (Mann–Whitney, U 5 22.0, n 5 14, p 5 0.75). dated to the late 1970s and early 1980s (Fig. 4). Specifically, The sensitivity of our approach to detect B. longimanus in in the replicate cores at site 1, the first occurrences were in the sediment record is important to consider. Sensitivity can 1978 and 1984 (mean 5 1981). In the replicate cores at site be estimated by the area of lake from which a sediment core 2, the first occurrences were both in 1982. At sites 3 and 4, collects depositional material, and then projecting this over where only one core was analyzed per site, the first occur- a relevant area and timeframe. Given that the sediment core rences were in 1986 and 1982, respectively. The tight and tube inside diameter was 6.7 cm, each core surface recruits 2 overlapping temporal correspondence between dates of first its material from an overlying lake area of 35.2 cm ,or 2 detection in each basin argues against an invasion that 0.0035 m . Solving the proportionality: was initially basin specific, especially given that the range 1m2=0:0035 m25xB: longimanusm22=1B: longimanusm22; of dates lies within the envelope of 210Pb dating precision (3) (6 4.0 yr, 2 standard deviations). Logistic growth models, set with a start date in 1980, for the unknown x yields x 5 284. This indicates that at least effectively described the trajectories of mean spine accumu- 284 B. longimanus must recruit annually to the sediment per lation rates at sites 1–3 (Fig. 5, Table 1). At these three sites, m2 of lake surface area, assuming an even spatial distribution the growth curves reached asymptotes near the year 2000, of animals above the lake bottom, as a necessary condition suggesting that the population required about 2 decades to to detect at minimum 1 B. longimanus in a sediment core reach K. Accordingly, we estimated K as the average spine section representative of a single year of material deposition. accumulation rate for years after 2001. Values ranged from 1817 to 2456 individuals m22 y21. Estimated r, determined Daphnia ephippia by the best-fit model, ranged from 0.24 to 0.34 y21. At site 4 Based on inferences from preserved ephippia, Daphnia (Fig. 4), where poor temporal resolution precluded modeling assemblages in both basins were dominated by D. mendotae with a logistic growth curve, the accumulation rate achieved and D. pulex (Fig. 7). D. retrocurva was present throughout in 1982, the year of first detection, was generally maintained the period of analysis but was far less abundant than the

1102 Branstrator et al. Bythotrephes colonization dynamics

4000 Site 1 Site 3

3000 ) -1 y

-2 2000

1000

0 4000 Site 2 Site 4 accumulation rate (spines m 3000

2000 B. longimanus

1000

0 1970 1980 1990 2000 2010 1970 1980 1990 2000 2010 Estimated Year (210Pb)

Fig. 4. B. longimanus spine accumulation rate as a function of estimated year of deposition. For sites 1 and 2, data for the nondated cores are shown, and arrows indicate dates of first detection of caudal spines in the corresponding dated cores. For site 1, the arrow is at 1984. For site 2, the arrow is at 1982, and the same year as first detection in the nondated core. For sites 3 and 4, data for the dated cores are shown. other two species, even prior to B. longimanus invasion. At distribution, early population growth trajectories, lag phases, site 1, the proportional compositions of D. mendotae and D. and early impacts on prey. pulex remained relatively constant over most of the period of B. longimanus invasion and proliferation, but D. mendotae did First presence and early spatial distribution increase near the top of the core. At site 2, the proportional Based on radioisotope dating, 1982 is the most parsimoni- compositions of the two daphniid species diverged coinci- ous year of first presence of B. longimanus in Island Lake. dent with a major pulse in B. longimanus accumulation rate This was the year of first detection in three of six sediment in the mid-1990s (Fig. 4). D. mendotae remained the propor- cores and at two of four sites with the other two sites point- tional dominant species thereafter. ing to 1981 and 1986 as year of first presence. Factoring in 6 4 yr (2 standard deviations) around 1982 for uncertainty in Discussion the 210Pb dating at this sediment horizon, the earliest pres- Invasive species pose major threats to natural resources ence is estimated as 1978 and the latest as 1986, or between and ecosystem services (Mack et al. 2000; Pejchar and 4 and 12 yr prior to 1990 when B. longimanus was first Mooney 2009), but their timelines of establishment, prolifer- detected in the water column. ation, and impacts remain poorly understood (Crooks 2005). We estimated that our method is sufficiently sensitive to We used sediment subfossil remains of the predatory crusta- detect at least 1 B. longimanus caudal spine per year in sedi- cean B. longimanus to study its temporal dynamics during ments in habitats where the overlying cumulative annual colonization including the date of first presence, early spatial areal density is 284 B. longimanus m22. This is considerably

1103 Branstrator et al. Bythotrephes colonization dynamics

3000 4000 A C Site 1 Site 3 ) ) -1 -1 3000 y y -2 -2 2000 accumulation accumulation accumulation 2000

1000 1000 rate (spines m rate (spines rate (spines m rate (spines B. longimanus B. longimanus 0 0 1980 1990 2000 2010 1980 1990 2000 2010 Estimated Year (210Pb) Estimated Year (210Pb) 4000 3 B D Site 2 2 )

-1 3000 y 1 -2

accumulation 2000 0

-1 1000 rate (spines m rate (spines -2 B. longimanus 30-yr2009) average (1980 to May to August average minus average August to May 0 -3 1980 1990 2000 2010 1980 1990 2000 2010 Estimated Year (210Pb) Year

Fig. 5. Panels A–C: B. longimanus spine accumulation rate as a function of estimated year of deposition (as in Fig. 4) with best-fit line superimposed defined by the logistic equation (see Eq. 2). Panel D: Long-term air temperature records in Duluth, Minnesota, presented as residual differences between annual means (May–August) minus 15.88C which is the May–August average from 1980 to 2009.

Table 1. Parameter estimates and mean corrected R2 for the Lake supports much larger B. longimanus densities than Harp logistic model: Nt 5 K/(1 1 be2rt) best fit to data (Fig. 5) where Nt Lake (Brown et al. 2012), it seems probable that our method is B. longimanus annual accumulation rate (spines m22 y21)ata was sufficiently sensitive to detect B. longimanus very early in given time t, K is carrying capacity (equilibrium annual accumula- its colonization of Island Lake, lending further support to 22 21 tion rate, as spines m y ), r is growth rate (increase in annual 1982 as a plausible year of invasion. spine accumulation rate, as y21), and b is a constant. Both Nt and The earliest published account of B. longimanus in an t were known, K was assigned as the average accumulation rate in inland North American lake is in 1989 in three lakes (Lake years after 2001, and b and r were estimated by the model. Joseph, Lake Muskoka, and Lake Rosseau) located in south- Site K r b Mean corrected R2 ern Ontario, Canada (Yan et al. 1992). Our results from sedi- ment reconstructions imply that inland lake invasions in 1 1817 0.24 11.19 0.67 North America began earlier, by as much as 7 yr. By infer- 2 2456 0.22 40.69 0.85 ence, the invasion timelines of B. longimanus in the Lauren- 3 2444 0.34 468.46 0.86 tian Great Lakes may also be earlier than currently estimated through plankton net tows, which is supported by previous lower than the annual accumulation of individuals (553 ani- speculations (Sprules et al. 1990; Johannsson et al. 1991) mals m22) that Hall and Yan (1997) detected in the second that B. longimanus was probably present in the Great Lakes year of invasion in Harp Lake, Ontario. Given that Island at low, cryptic densities for years prior to its first detection

1104 Branstrator et al. Bythotrephes colonization dynamics

1500 Confirmation or adjustment of these dates could be addressed with paleolimnology. A In Island Lake, B. longimanus was first documented in the water column in 1990 (Gravelle 1990), 8 yr after our sedi- 500 ment data indicate its initial appearance. This time gap underscores the likelihood of delays between presence and

Site 3 detection for invasive species (Edlund et al. 2000), a result -500 that has implications for reconstructing landscape-scale pat- terns of lake-invasion sequences (Havel et al. 2002; MacIsaac

) et al. 2004). This result also raises management concern that -1

y n = 15 ecosystems can act as dispersal hubs for years before threats -2 -1500 are recognized. Site 1 1,500 In Island Lake, there was no indication of a basin-specific chronological sequence to invasion. In fact, by 1982, spines B were already distributed in both East and West basins, indi- cating either rapid dispersal or multiple introductions 500 around the same period. Lack of statistical differences in average sediment accumulation rates among coring sites, and in water column densities of B. longimanus between Site 2 -500 basins, suggests that the open-water regions of the lake are broadly inhabitable by the invader, which could have facili- tated early dispersal.

spine accumulation rate (spines m rate (spines accumulationspine n = 14 Lake Superior, based on proximity and size, was the likely -1,500 propagule source for the invasion of Island Lake. If so, our Site 1 results cast doubt on the validity of 1987 as an accurate 1,500 invasion year for Lake Superior (Cullis and Johnson 1988). C An alternative historical propagule source is the Duluth-

B. longimanus Superior Harbor. The harbor is connected to Lake Superior

given as observed rate minus rate predicted by given as observed rate minus rate predicted by growth logistic 500 through two shipping channels (Fig. 1) and receives the larg- est discharge of interregional ballast water among all harbors in the Laurentian Great Lakes (Rup et al. 2010). It is recog- Site 3 nized as an invasion hotspot where several other transocean- -500 ic migrants established early or first population centers on the continent (Bronte et al. 2003; Grigorovich et al. 2003). A n = 14 third potential historical propagule source is one of the low- -1,500 er Great Lakes, however, the requisite distances over which -1,500 -500 500 1,500 propagules would have needed to survive to reach Island Site 2 Lake is incongruent with the species’ low tolerance to air B. longimanus spine accumulation rate (spines m-2 y-1) exposure (Branstrator et al. 2013). given as observed rate minus rate predicted by logistic growth Early population growth and lag phase Fig. 6. Residual differences between observed minus predicted (based Logistic models effectively describe the growth trajectories on the logistic equation, see Fig. 5) B. longimanus accumulation rate in of B. longimanus during colonization. The general pattern is the sediments. Each point represents the simultaneous residual values consistent with proliferation in an open niche where resour- 210 for the same year inferred from Pb dating: (A) site 1 vs. site 3, (B) ces are initially unrestrictive, followed by progressive site 1 vs. site 2, and (C) site 2 vs. site 3. Circled point indicates outlier density-dependent limitation around a zero net growth equi- not included in the statistical analysis. Lines represent a 1:1 relationship. librium. Brown et al. (2012) demonstrated that annual death rates and diel vertical migration of B. longimanus in Island in the water column. The currently accepted timeline of first Lake during 2002–2004 were under strong bottom-up control appearance of B. longimanus in the Great Lakes (based on by food abundance, consistent with resource limitation and water column sampling) is Lake Ontario in 1982, followed density dependence as internal regulators of annual growth. by Lake Huron in 1984, Lake Erie in 1985, Lake Michigan in Wide fluctuations around logistic growth support varia- 1986, and Lake Superior in 1987 (Bur et al. 1986; Lehman tion in interannual fitness of the population. Broad interan- 1987; Cullis and Johnson 1988; Johannsson et al. 1991). nual fluctuations (e.g., doubling) in B. longimanus density,

1105 Branstrator et al. Bythotrephes colonization dynamics

Table 2. Yearly average 6 1 standard error water column densities (number m 2 2)ofB. longimanus at site locations in the East and West basins of Island Lake during select years, site average, and bottom depth (Z) at site. Yearly averages are based on night-time monthly or weekly collections during May–November (2002, 2003, 2004), and monthly collections during June, July, and September (2010). Site average is the mean of all years. Approximate sampling locations (lettered A-F) are shown in Fig. 1.

Year

Basin (location) 2002 2003 2004 2010 Site avg. Z (m) East (A) - - - 312 6 121 312 9 East (B) - - 327 6 58 248 6 3 288 16 East (C) 60 6 15 283 6 36 169 6 58 371 6 165 221 13 West (D) - - - 204 6 97 204 10 West (E) 256 6 56 331 6 79 353 6 92 206 6 42 286 12 West (F) - - 492 6 201 127 6 31 310 9

6000

Site 1iteS 2D. mendotae D. pulex

) D. retrocurva -1

y 4000 -2 accumulation rate

2000 (ephippia m

Daphnia spp. 0 1970 1980 1990 2000 2010 1970 1980 1990 2000 2010 Estimated Year (210Pb)

Fig. 7. Daphnia ephippium accumulation rate as a function of estimated year of deposition at sites 1 and 2. even decades after its invasion, have been shown in other nonexistent, increase in population density or range exten- lakes where long-term density trends are available (Barbiero sion (Mack et al. 2000; Crooks 2005). This pronounced and and Tuchman 2004), indicating complex ecological interac- possibly prolonged window of time during which growth tions and difficulty in predicting year to year changes in and expansion of an invading species is exceedingly slow, or peak densities and impacts. The temporal synchronicity of nonexistent, can be statistically delineated from a second these oscillations in Island Lake, particularly at the two East stage when the population is growing markedly faster mea- basin sites, suggests the possibility of lake-wide forcing due sured by, for example, exponential or logistic trends (Aikio to external (e.g., temperature) or internal (e.g., fish year class et al. 2010). By the definition of Aikio et al. (2010), B. longi- strength, food density) factors. Deviations in air temperature manus bypassed a lag phase in Island Lake, as growth was from a long-term average were not significant predictors of already exponential from the year of first detection in 1982. these oscillations. However, lake water temperature and date This outcome may not be surprising given that reservoirs are of spring mixing warrant further investigation as environ- exceptionally conducive to growth of B. longimanus (Brown mental drivers, given their effects on B. longimanus produc- et al. 2012). In Harp Lake, Ontario, a natural glacial-scour tivity in the laboratory and field (Kim and Yan 2010; Young basin, B. longimanus also appeared to bypass a lag phase et al. 2011). (Hall and Yan 1997; Yan et al. 2001). Both lakes lie well A central question of our study was whether B. longimanus within the recognized range boundaries of the current North demonstrated a lag phase during colonization. The lag phase American invasion footprint (Kerfoot et al. 2011). For inva- is widely recognized as a period of initial slow, or sions occurring at range edges, we might expect increased

1106 Branstrator et al. Bythotrephes colonization dynamics resistance by temperature or other niche factors to promote explain the long-term build-up to an equilibrium density lag phases during colonization. (Hairston et al. 2000). Despite the apparent absence of a lag phase (Aikio et al. As a cautionary note, our reconstructions of B. longima- 2010) in Island Lake, B. longimanus took about two decades, nus spine accumulation rates in the sediments at site 2 may from 1982 to about 2000, to reach density equilibrium. This be biased by increasing frequencies of spine fragmentation is somewhat surprising given its r-selected life-history. Few in older strata. Increased fragmentation, however, does not studies have examined colonization timelines of freshwater necessarily imply that the material was subject to longer, cladocerans at whole-lake scales, but the limited evidence or more severe, deterioration in the sediments. In Gulspett- suggests that years may be needed for populations to reach vann Lake, Norway, B. longimanus spines were recovered stability. For example, Sarnelle and Knapp (2004) found that from 22 cm depth indicating their longevity as subfossils after removal of zooplanktivorous fish from small California (Nilssen and Sandoy 1990). Hall and Yan (1997) found that alpine lakes (1.7–3.4 ha surface area), intensive water column only 5.3% of B. longimanus caudal spines broke and no sampling with plankton nets required up to 4 yr to detect spine dissolution occurred after 4.5 months in a spike- native Daphnia middendorffiana that were re-establishing recovery experiment in the laboratory. Fragmentation from resting eggs. Similarly, where paleolimnology has been could alternatively reflect predation and gut passage by fish used to reconstruct invasion timelines of Daphnia similis and or other predators (Hall and Yan 1997; Jarnagin et al. Eubosmina coregoni, years were required to achieve major 2004). The fact that percentages of broken spines recovered bursts in population sizes (Hairston et al. 1999; Suchy et al. here were apparently random over time in cores 1 and 4 2010). suggests that breakage of the spine may be more complicat- Several alternative, but complimentary mechanisms may ed than a simple function of time-based deterioration explain the decadal timescale that B. longimanus required to would predict. reach its presumed annual equilibrium. First, it is known that Eurasian populations with distinct genotypes periodical- Early impacts on prey ly seeded the North American Great Lakes over the past sev- After the invasion of B. longimanus, a notable change eral decades (Berg et al. 2002; Colautti et al. 2005). This occurred to the daphniid assemblage as inferred through the staged introduction provides a template for the possibility of presence of ephippia. Specifically, D. mendotae either became ongoing genetic subsidy of Island Lake that could progres- dominant (site 2) or remained the dominant daphniid (site sively increase the population’s evolutionary potential. Sec- 1). This reflects a post invasion pattern widely documented ond, there is high genetic variation and heritability in the Laurentian Great Lakes and small inland lakes where associated with the length of the caudal spine of B. longima- daphniids commonly shift from multiple species assemblages nus (Miehls et al. 2012), which is a key defense against fish including D. mendotae, Daphnia pulicaria, and D. retrocurva to predators (Barnhisel 1991). Accordingly, the decadal timeline dominance by D. mendotae after B. longimanus invades (Azan to reach equilibrium could reflect a degree of improved local et al. 2015). The success of D. mendotae in the presence of B. fitness through rapid evolution of longer spines. Third, envi- longimanus is ascribed to its quick escape response (Pichlova- ronmental switches owing to external forcing (e.g., tempera- Ptacˇnıkova and Vanderploeg 2011). The apparent lack of a ture) or internal erosion of biotic resistance (e.g., noticeable negative response by D. retrocurva in Island Lake competitive release) could progressively favor B. longimanus is noteworthy since it is considered to be highly vulnerable in the lake. Fourth, in order to successfully overwinter, oth- to B. longimanus. The result may be related to its already low erwise clonally reproducing female B. longimanus must birth relative abundance pre-invasion. males and mate to produce resting eggs (Branstrator 2005). Our results not only confirm previous patterns of prey This life-cycle switch to resting eggs generally occurs in the declines but offer insight on the timing of community shifts. fall, and modeling and mesocosm studies show that resting Specifically, the transition in the daphniid assemblage in the egg production by B. longimanus is sensitive to Allee effects West basin was relatively unnoticeable for a decade or more as a result of mate limitation (Drake 2004; Gertzen et al. after invasion. Not until B. longimanus attained densities 2011). Hatching studies in Island Lake indicate that approxi- approaching its longer-term equilibrium (middle to late mately half of the egg bank turns over annually and approxi- 1990s) did concomitant changes occur to the daphniid mately 85% of the resting eggs that do hatch fail to mature assemblage. Thereafter, the changes that favored D. mendotae (Brown and Branstrator 2011). These high rates imply that appeared to be relatively stable despite fluctuations in B. strong investment is required in the first year and annually longimanus density. These temporal dynamics have ramifica- thereafter to replenish and grow the egg bank. Hence, if it tions for understanding the timing of indirect, cascading takes years to build the egg bank during colonization, and effects on other trophic levels, and underscore the value of there is demographic feedback between the size of the active long-term perspectives on ecosystem response to species resting egg pool and the summer population, it could help invasion (Strayer et al. 2006).

1107 Branstrator et al. Bythotrephes colonization dynamics

References Brown, M. E., D. K. Branstrator, and L. J. Shannon. 2012. Pop- ulation regulation of the spiny water flea (Bythotrephes long- Aikio,S.,R.P.Duncan,andP.Hulme.2010.Lag-phasesinalien imanus) in a reservoir: Implications for invasion. Limnol. plant invasions: separating the facts from the artefacts. Oikos. Oceanogr. 57: 251–271. doi:10.4319/lo.2012.57.1.0251 119: 370–378. doi:10.1111/j.1600-0706.2009.17963.x Bur, M. T., D. M. Klarer, and K. A. Krieger. 1986. First records Appleby, P. G. 2001. Chronostratigraphic techniques in of a European cladoceran, Bythotrephes cederstroemi,in recent sediments, pp. 171–203. In W. M. Last and J. P. Lakes Erie and Huron. J. Great Lakes Res. 12: 144–146. Smol [eds.], Tracking environmental change using lake doi:10.1016/S0380-1330(86)71711-5 sediments. Volume 1: basin analysis, coring, and chrono- Colautti, R. I., M. Manca, M. Viljanen, H. A. M. Ketelaars, H. logical techniques. Kluwer. Burgi,€ H. J. MacIsaac, and D. D. Heath. 2005. Invasion Azan,S.S.E.,S.E.Arnott,andN.D.Yan.2015.Areviewofthe genetics of the Eurasian spiny waterflea: Evidence for bot- effects of Bythotrephes longimanus and calcium decline on zoo- tlenecks and gene flow using microsatellites. Mol. Ecol. plankton communities—can interactive effects be predicted?. 14: 1869–1879. doi:10.1111/j.1365-294X.2005.02565.x Environ. Rev. 23: 395–413. doi:10.1139/er-2015-0027 Crooks, J. A. 2005. Lag times and exotic species: The ecology Barbiero, R. P., and M. L. Tuchman. 2004. Changes in the and management of biological invasions in slow-motion. crustacean communities of Lakes Michigan, Huron, and Ecoscience. 12: 316–329. doi:10.2980/i1195-6860-12-3- Erie following the invasion of the predatory cladoceran 316.1 Bythotrephes longimanus. Can. J. Fish. Aquat. Sci. 61: Cullis, K. I., and G. E. Johnson. 1988. First evidence of the 2111–2125. doi:10.1139/f04-149 cladoceran Bythotrephes cederstroemi Schoedler in Lake Barnhisel, D. R. 1991. The caudal appendage of the cladocer- Superior. J. Great Lakes Res. 14: 524–525. doi:10.1016/ an Bythotrephes cederstroemi as defense against young fish. S0380-1330(88)71584-1 J. Plankton Res. 13: 529–537. doi:10.1093/plankt/13.3.529 Dean, W. E. Jr. 1974. Determination of carbonate and organ- Berg, D. J., D. W. Garton, H. J. MacIsaac, V. E. Panov, and I. ic matter in calcareous sediments and sedimentary rocks V. Telesh. 2002. Changes in genetic structure of North by loss on ignition: Comparison with other methods. J. American Bythotrephes populations following invasion Sediment. Res. 44: 242–248. from Lake Ladoga, Russia. Freshw. Biol. 47: 275–282. doi: Drake, J. M. 2004. Allee effects and the risk of biological 10.1046/j.1365-2427.2002.00805.x invasion. Risk Anal. 24: 795–802. doi:10.1111/j.0272- Boudreau, P. E., K. L. Pangle, E. M. Reed, and S. D. Peacor. 4332.2004.00479.x 2013. Finely tuned response of native prey to an invasive Edlund, M. B., C. M. Taylor, C. L. Schleske, and E. F. predator in a freshwater system. Ecology. 94: 1449–1455. Stoermer. 2000. Thalassiosira baltica (Grunow) Ostenfeld Branstrator, D. K. 2005. Contrasting life histories of the pred- (Bacillariophyta), a new exotic species in the Great Lakes. atory cladocerans Leptodora kindtii and Bythotrephes longi- Can. J. Fish. Aquat. Sci. 57: 610–615. doi:10.1139/f99-284 manus. J. Plankton Res. 27: 569–585. doi:10.1093/plankt/ Foster, S. E., and W. G. Sprules. 2009. Effects of the Bythotre- fbi033 phes invasion on native predatory invertebrates. Limnol. Branstrator, D. K., M. E. Brown, L. J. Shannon, M. T. Thabes, Oceanogr. 54: 757–769. doi:10.4319/lo.2009.54.3.0757 and K. Heimgartner. 2006. Range expansion of Bythotre- Garcıa-Berthou, E., C. Alcaraz, Q. Pou-Rovira, L. Zamora, G. phes longimanus in North America: Evaluating habitat Coenders, and C. Feo. 2005. Introduction pathways and characteristics in the spread of an exotic zooplankter. establishment rates of invasive aquatic species in Europe. Biol. Invasions. 8: 1367–1379. doi:10.1007/s10530-005- Can. J. Fish. Aquat. Sci. 62: 453–463. doi:10.1139/f05-017 5278-7 Gertzen, E. L., Leung, B. B., and N. D. Yan. 2011. Propagule Branstrator, D. K., L. J. Shannon, M. E. Brown, and M. T. pressure, Allee effects and the probability of establishment Kitson. 2013. Effects of chemical and physical conditions of an invasive species (Bythotrephes longimanus). Eco- on hatching success of Bythotrephes longimanus resting sphere. 2: art30. eggs. Limnol. Oceanogr. 58: 2171–2184. doi:10.4319/ Glaser, P. H., and M. Griffith. 2007. A field extruder for rap- lo.2013.58.6.2171 idly sectioning near-surface cores from lakes and wet- Bronte, C. R., M. P. Ebener, D. R. Schreiner, D. S. DeVault, M. lands. J. Paleolimnol. 38: 459–466. doi:10.1007/s10933- M. Petzold, D. A. Jensen, C. Richards, and S. J. Lozano. 006-9078-6 2003. Fish community change in Lake Superior, 1970-2000. Gravelle, J. 1990. Slimy species sneaks into inland lake. Can. J. Fish. Aquat. Sci. 60: 1552–1574. doi:10.1139/f03-136 Duluth News Tribune, Duluth, MN, September 13. Brown, M. E., and D. K. Branstrator. 2011. Patterns in the Grigorovich, I. A., R. I. Colautti, E. L. Mills, K. Holeck, A. G. abundance, phenology, and hatching of the resting egg Ballert, and H. J. MacIsaac. 2003. Ballast-mediated animal stage of the invasive zooplankter Bythotrephes longimanus: introductions in the Laurentian Great Lakes: Retrospective Implications for establishment. Biol. Invasions. 13: 2547– and prospective analysis. Can. J. Fish. Aquat. Sci. 60: 740– 2559. doi:10.1007/s10530-011-0080-1 756. doi:10.1139/f03-053

1108 Branstrator et al. Bythotrephes colonization dynamics

Hairston, N. G., Jr., L. J. Perry, A. J. Bohonak, M. Q. Fellows, spiny water fleas (Bythotrephes longimanus). Biol. Invasions. C. M. Kearns, and D. R. Engstrom. 1999. Population biol- 13: 2513–2531. doi:10.1007/s10530-011-0078-8 ogy of a failed invasion: Paleolimnology of Daphnia exilis Kim, N., and N. D. Yan. 2010. Methods for rearing the invasive in upstate New York. Limnol. Oceanogr. 44: 477–486. zooplankter Bythotrephes in the laboratory. Limnol. Ocean- doi:10.4319/lo.1999.44.3.0477 ogr.: Methods. 8: 552–561. doi:10.4319/lom.2010.8.552 Hairston, H. G., Jr, A. M. Hansen, and W. R. Schaffner. 2000. Korhola, A., and M. Rautio. 2001. Cladocera and Other Bran- The effect of diapause emergence on the seasonal dynam- chiopod Crustaceans, pp. 5–41. In J. P. Smol, H. J. B. Birks, ics of a zooplankton assemblage. Freshw. Biol. 45: 133– and W. M. Last [eds.], Tracking environmental change using 145. doi:10.1046/j.1365-2427.2000.00386.x lake sediments. Volume 4: zoological indicators. Kluwer. Hall, R. I., and N. D. Yan. 1997. Comparing annual popula- Lamborg, C. H., D. R. Engstrom, W. F. Fitzgerald, and P. H. tion growth estimates of the exotic invader Bythotrephes Balcom. 2012. Apportioning global and non-global com- by using sediment and plankton records. Limnol. Ocean- ponents of mercury deposition through 210Pb indexing. ogr. 42: 112–120. doi:10.4319/lo.1997.42.1.0112 Sci. Total Environ. 448: 132–140. Havel, J. E., J. B. Shurin, and J. R. Jones. 2002. Estimating Leavitt, P. R., and others 2009. Paleolimnological evidence dispersal from patterns of spread: Spatial and local control of the effects on lakes of energy and mass transfer from of lake invasions. Ecology. 83: 3306–3318. doi:10.1890/ climate and humans. Limnol. Oceanogr. 54: 2330–2348. 0012-9658(2002)083[3306:EDFPOS]2.0.CO;2 doi:10.4319/lo.2009.54.6_part_2.2330 Hoffman, J. C., J. R. Kelly, A. S. Trebitz, G. S. Peterson, and Lehman, J. T. 1987. Palearctic predator invades North Ameri- C. W. West. 2011. Effort and potential efficiencies for can Great Lakes. Oecologia. 74: 478–480. doi:10.1007/ aquatic non-native species early detection. Can. J. Fish. BF00378947 Aquat. Sci. 68: 2064–2079. doi:10.1139/f2011-117 MacIsaac, H. J., J. V. M. Borbely, J. R. Muirhead, and P. A. Jankowski, T., and D. Straile. 2003. A comparison of egg- Graniero. 2004. Backcasting and forecasting biological bank and long-term plankton dynamics of two Daphnia invasions of inland lakes. Ecol. Appl. 14: 773–783. doi: species, D. hyalina and D. galeata: Potentials and limits of 10.1890/02-5377 reconstruction. Limnol. Oceanogr. 48: 1948–1955. doi: Mack,R.N.,D.Simberloff,W.M.Lonsdale,H.Evans,M.Clout, 10.4319/lo.2003.48.5.1948 and F. A. Bazzaz. 2000. Biotic invasions: Causes, epidemiology, Jarnagin, S. T., W. C. Kerfoot, and B. K. Swan. 2004. Zoo- global consequences, and control. Ecol. Appl. 10:689–710. plankton life cycles: Direct documentation of pelagic doi:10.1890/1051-0761(2000)010[0689:BICEGC]2.0.CO;2 births and deaths relative to diapausing egg production. Manca, M., B. Torretta, P. Comoli, S. L. Amsinck, and E. Limnol. Oceanogr. 49: 1317–1332. doi:10.4319/lo.2004. Jeppesen. 2007. Major changes in trophic dynamics in 49.4_part_2.1317 large, deep sub-alpine Lake Maggiore from 1940s to 2002: Jeppesen, E., P. Leavitt, L. De Meester, and J. P. Jensen. 2001. A high resolution comparative palaeo-neolimnological Functional ecology and palaeolimnology: Using cladoceran study. Freshw. Biol. 52: 2256–2269. doi:10.1111/j.1365- remains to reconstruct anthropogenic impact. Trends Ecol. 2427.2007.01827.x Evol. 16: 191–198. doi:10.1016/S0169-5347(01)02100-0 Miehls, A. L. J., S. D. Peacor, and A. G. McAdam. 2012. Johannsson, O. E., E. L. Mills, and R. O’Gorman. 1991. Genetic and maternal effects on tail spine and body Changes in the nearshore and offshore zooplankton com- length in the invasive spiny water flea (Bythotrephes longi- munities in Lake Ontario: 1981-88. Can. J. Fish. Aquat. manus). Evol. Appl. 5: 306–316. doi:10.1111/j.1752- Sci. 48: 1546–1557. doi:10.1139/f91-183 4571.2011.00221.x Keilty, T. J. 1988. A new biological marker layer in the sedi- Nilssen, J. P., and S. Sandoy. 1990. Recent lake acidification and ments of the Great Lakes: Bythotrephes cederstroemi cladoceran dynamics: Surface sediment and core analysis (Schodler)€ spines. J. Great Lakes Res. 14: 369–371. doi: from lakes in Norway, Scotland and Sweden. Philos. Trans. R. 10.1016/S0380-1330(88)71566-X Soc. London B. 327:299–309.doi:10.1098/rstb.1990.0066 Kerfoot, W. C. 1981. Long-term replacement cycles in cla- Nunes, A. L., G. Orizaola, A. Laurila, and R. Rebelo. 2014. doceran communities: A history of predation. Ecology. Rapid evolution of constitutive and inducible defences 62: 216–233. doi:10.2307/1936683 against an invasive predator. Ecology. 95: 1520–1530. doi: Kerfoot, W. C., X. Ma, C. S. Lorence, and L. J. Weider. 2004. 10.1890/13-1380.1 Towards resurrection ecology: Daphnia mendotae and D. Pangle, K. L., S. D. Peacor, and O. E. Johannsson. 2007. Large retrocurva in the coastal region of Lake Superior, among nonlethal effects of an invasive invertebrate predator on the first successful outside invaders?. J. Great Lakes Res. zooplankton population growth rate. Ecology. 88: 402– 30: 285–299. doi:10.1016/S0380-1330(04)70392-5 412. doi:10.1890/06-0768 Kerfoot, W. C., F. Yousef, M. M. Hobmeier, R. P. Maki, S. T. Pejchar, L., and H. A. Mooney. 2009. Invasive species, eco- Jarnagin, and J. H. Churchill. 2011. Temperature, recrea- system services and human well-being. Trends Ecol. Evol. tional fishing and diapause egg connections: Dispersal of 24: 497–504. doi:10.1016/j.tree.2009.03.016

1109 Branstrator et al. Bythotrephes colonization dynamics

Pichlova-Pt acˇnıkova, R., and H. A. Vanderploeg. 2011. The Yan, N. D., W. I. Dunlop, T. W. Pawson, and L. E. MacKay. quick and the dead: Might differences in escape rates 1992. Bythotrephes cederstroemi (Schoedler) in Muskoka explain the changes in the zooplankton community com- Lakes: First records of the European invader in inland position of Lake Michigan after invasion by Bythotrephes?. lakes in Canada. Can. J. Fish. Aquat. Sci. 49: 422–426. Biol. Invasions. 13: 2595–2604. doi:10.1007/s10530-011- doi:10.1139/f92-048 0076-x Yan, N. D., A. Blukacz, W. G. Sprules, P. K. Kindy, D. Rilov, G., Y. Benayahu, and A. Gasith. 2004. Prolonged lag Hackett, R. E. Girard, and B. J. Clark. 2001. Changes in in population outbreak of an invasive mussel: A shifting- zooplankton and the phenology of the spiny water flea, habitat model. Biol. Invasions. 6: 347–364. doi:10.1023/B: Bythotrephes, following its invasion of Harp Lake, Ontario, BINV.0000034614.07427.96 Canada. Can. J. Fish. Aquat. Sci. 58: 2341–2350. doi: Ruiz, G. M., P. W. Fofonoff, J. T. Carlton, M. J. Wonham, 10.1139/f01-171 and A. H. Hines. 2000. Invasion of coastal marine com- Yan, N. D., B. Leung, M. A. Lewis, and S. D. Peacor. 2011. munities in North America: Apparent patterns, processes, The spread, establishment and impacts of the spiny water and biases. Annu. Rev. Ecol. Syst. 31: 481–531. doi: flea, Bythotrephes longimanus, in temperate North America: 10.1146/annurev.ecolsys.31.1.481 A synopsis of the special issue. Biol. Invasions. 13: 2423– Rup, M. P., S. A. Bailey, C. J. Wiley, M. S. Minton, A. 2432. doi:10.1007/s10530-011-0069-9 Whitman Miller, G. M. Ruiz, and H. J. MacIsaac. 2010. Young, J. D., A. L. Strecker, and N. D. Yan. 2011. Increased Domestic ballast operations on the Great Lakes: Potential abundance of the non-indigenous zooplanktivore, Bytho- importance of Lakers as a vector for introduction and trephes longimanus, is strongly correlated with greater spread of nonindigenous species. Can. J. Fish. Aquat. Sci. spring prey availability in Canadian Shield lakes. Biol. 67: 256–268. Invasions. 13: 2605–2619. doi:10.1007/s10530-011-0077-9 Sarnelle, O., and R. A. Knapp. 2004. Zooplankton recovery after fish removal: Limitations of the egg bank. Limnol. Ocean- Acknowledgments ogr. 49: 1382–1392. doi:10.4319/lo.2004.49.4_part_2.1382 This work was financially supported by a grant from the University of Smol, J. P. 1992. Paleolimnology: An important tool for Minnesota Duluth Office of the Vice Chancellor for Academic Affairs, Hamline University, and the Minnesota Sea Grant College Program sup- effective ecosystem management. J. Aquat. Ecosyst. ported by the National Oceanic and Atmospheric Administration office Health. 1: 49–58. doi:10.1007/BF00044408 of Sea Grant, United States Department of Commerce, under grant Sprules, W. G., H. P. Riessen, and E. H. Jin. 1990. Dynamics NA16-RG1046. The U.S. Government is authorized to reproduce and of the Bythotrephes invasion of the St. Lawrence Great distribute reprints for government purposes, notwithstanding any copy- Lakes. J. Great Lakes Res. 16: 346–351. doi:10.1016/S0380- right notation that may appear hereon. This paper is journal reprint JR 635 of the Minnesota Sea Grant College Program. We thank Sarah Gros- 1330(90)71429-3 shuesch, Aaron Lingwall, Erin Mortenson, Amy Myrbo, Samantha Oliver, Strayer, D. L., V. T. Eviner, J. M. Jeschke, and M. L. Pace. Doug Ricketts, Amelia Sancilio, Shawn Schottler, and Matt TenEyck for 2006. Understanding the long-term effects of species inva- field and lab assistance. We thank two anonymous reviewers who sions. Trends Ecol. Evol. 21: 645–651. doi:10.1016/ improved the manuscript. Cores are archived at LacCore (National j.tree.2006.07.007 Lacustrine Core Facility), Department of Earth Sciences, University of Minnesota-Twin Cities. Suchy, K. D., A. Salki, and B. J. Hann. 2010. Investigating the invasion of the nonindigenous zooplankter, Eubosmina cor- Conflict of Interest egoni, in Lake Winnipeg, Manitoba, Canada. J. Great Lakes None declared. Res. 36: 159–166. doi:10.1016/j.jglr.2009.12.004 Vander Zanden, M. J., G. J. A. Hansen, S. N. Higgins, and M. Submitted 05 May 2016 S. Kornis. 2010. A pound of prevention, plus a pound of Revised 31 August 2016 cure: Early detection and eradication of invasive species Accepted 17 October 2016 in the Laurentian Great Lakes. J. Great Lakes Res. 36: 199–205. doi:10.1016/j.jglr.2009.11.002 Associate editor: Takehito Yoshida

1110