Modest dispersal, low densities, and little change in water clarity characterize zebra mussel populations near the southern extent of their range

doi:10.21203/rs.3.rs-2733356/v1

Invasive, filter feeding zebra mussels (Dreissena polymorpha) typically cause an increase in water clarity shortly after their establishment. The purpose of this study is to evaluate whether this occurred in Midwest reservoirs, near the southern edge of their North American expansion, using a 40 + year dataset. We look for regime shifts and long-term trends in annual water clarity and compare these to the estimated zebra mussel invasion date for 7 invaded reservoirs in Missouri, USA. We also look at water clarity in 26 non-invaded, reference reservoirs to evaluate if zebra mussel impacts are being masked by changes in environmental factors. Collectively, our analyses provide a weight of evidence showing that zebra mussel establishment did not increase water clarity, likely because densities are too low to result in a noticeable impact. The highest zebra mussel density we observe is 65 mussels m^− 2, an order of magnitude less than in systems where they have had a sustained impact. Low densities could be due to a combination of sublethal environmental conditions. We identified common characteristics of invaded reservoirs, including reduced particulate inorganic material and water temperatures.

Dreissenids

Great Plains

Impoundment

Midcontinent

Reservoir

Zebra Mussels

Since their arrival in Lake Erie in 1986 (Carlton 2008), zebra mussels (Dreissena polymorpha) have spread as far as California (Benson et al. 2021a) and Texas (Churchill et al. 2017), with new water bodies being invaded every year (Benson et al. 2021a). Zebra mussels can reach peak densities 2–4 years after introduction (Karatayev et al. 2011) and can dramatically alter native ecosystem processes and function (Schloesser et al. 1991; Ward and Ricciardi 2007; Kim et al. 2015). Given their high filter-feeding capacity (Fanslow et al. 1995), zebra mussel invasions can lead to reductions in phytoplankton biovolume and other suspended particles throughout the water column (Higgins and Vander Zanden 2010; Kirsch and Dzialowski 2012), although their establishment has occasionally resulted in a marginal increase in phytoplankton biomass (Reynolds and Aldridge 2021). The magnitude of impact on water clarity depends on mussel density, which is influenced by several abiotic factors including water temperature (Jost et al. 2015), suspended inorganic solids (Madon et al. 1998), dissolved oxygen concentrations (McMahon 1996), and predation (Bartsch et al. 2005). Zebra mussel densities can fluctuate widely from year to year locally within a water body, but usually do not decline on a lake-wide scale over the long term unless replaced by quagga mussels (Dreissena rostriformis bugensis; Strayer et al. 2019). Several water body characteristics also influence zebra mussel’s impact on water clarity, including localized water flow (Xia et al. 2022) and hydrology (Jones et al. 2022a).

The impact of zebra mussels on Missouri reservoirs is poorly understood. Most North American studies investigating zebra mussel impacts have been conducted in northern, natural lakes (Fahnenstiel et al. 1995; Budd et al. 2001; Higgins and Vander Zanden 2010; Baranowska et al. 2013; North et al. 2013). In contrast, Missouri reservoirs are warmer, have higher concentrations of inorganic particles, are characterized by high intra- and inter-annual variation in nutrient and chlorophyll-a (chl-a) concentrations, often have longer stratification periods, and have higher flushing rates (Knowlton and Jones 2006; Jones et al. 2008; Jones et al. 2011). Reservoirs are characterized by longitudinal flow that impacts zebra mussel distributions compared to natural lakes where water movement between nearshore and offshore areas has a greater influence on phytoplankton (Locklin et al. 2020), and consequently sessile filter feeders like zebra mussels (Hecky et al. 2004). These abiotic factors affect zebra mussel survival and persistence (Schneider et al. 1998; Elderkin and Klerks 2005), so it is possible that zebra mussel impacts in reservoirs differ from natural lakes.

The purpose of this study is to determine whether zebra mussels impact water clarity in Missouri reservoirs. Our first objective is to identify whether shifts in water clarity in invaded reservoirs occur, and if so, whether they align with zebra mussel establishment. For this, we looked for regime shifts in long-term water clarity data to determine correspondence with the detection of zebra mussels. Our second objective is to evaluate whether environmental factors and predation might be limiting factors to the potential impact of zebra mussels on water clarity. To accomplish this, we looked for water clarity shifts in non-invaded reference reservoirs and compared them with water clarity in invaded reservoirs over the same period. We also measured zebra mussel density and biomass in several invaded reservoirs. We predict that zebra mussels will reduce water clarity and that impacts will be like those documented in northern latitude lakes. Our insights of the impact this aquatic invader has near the leading edge of its expansion will help inform the potential zebra mussel risk in other regions.

Study Reservoirs

This study was conducted in 33 reservoirs throughout Missouri, USA, using data collected between 1976 and 2019 (Table 1). We analyzed water clarity indicators over time in reservoirs invaded by zebra mussels (n = 7; Blue Springs, Bull Shoals, Jacomo, Lake of the Ozarks, Lotawana, Smithville, Truman) to assess whether there was a shift in water clarity post-invasion. Three reservoirs, Lake of the Ozarks, Lotawana, and Smithville have at least 11 years of data pre-, and 11 years post- zebra mussel invasion. We also measured these parameters over time in reference reservoirs (n = 26; Atkinson, Binder, Bowling Green, Brookfield, Capri, Clearwater, Council Bluff, Deer Ridge, Fellows, Forest, Higginsville, Hunnewell, Kraut Run, Lincoln, Long Branch, Mark Twain, McDaniel, North, Pomme de Terre, Shayne, Stockton, Sugar Creek, Table Rock, Viking, Wappapello, Watkins Mill), which had no known zebra mussel populations. Throughout the state, the Missouri Department of Conservation (MDC) records the first year in which adult zebra mussels are observed in invaded reservoirs. Private citizens report sightings, which are then verified by MDC personnel and reported to the United States Geological Survey (USGS; Benson et al. 2021a). To account for the latency period between when zebra mussels first reached each reservoir and when they were first observed, we report zebra mussel invasion ranges as 3 years before initial observation of zebra mussel adults (Burlakova et al. 2006). For this study, we analyzed invaded reservoirs with a minimum of 5 years of data both pre- and post- invasion. All reference reservoirs had at least 26 years of water clarity data spanning the same time period as the zebra mussel reservoirs, and no more than 4 years of non-continuous data for at least one parameter.

Table 1

Characterization for all invaded and reference reservoirs. Latitude and longitude refer to the coordinates near the dam. For all reservoirs except Bull Shoals, this is where water samples are collected. Maximum depth is based on the dam height of each reservoir and volume is the volume of each reservoir at full pool. Trophic status was determined from previous work in Missouri reservoirs (Jones et al. 2008). Flushing rate for Bull Shoals is unknown.
	Reservoir	Latitude	Longitude	Ecoregion	Trophic Status	Mean Depth (m)	Max. Depth (m)	Surface Area (km²)	Flushing Rate (times yr^− 1)	Watershed Area (km²)	Volume (km³)
Zebra Mussel Reservoirs	Blue Springs	39.0141	-94.3371	Osage Plains	Eutrophic	4.5	16.2	2.9	1.1	86.7	0.0133
	Bull Shoals	36.5308	-92.6047	Ozark Highlands	Mesotrophic	20.4	79.6	194.3		15633.2	7.4169
	Jacomo	38.9931	-94.3078	Osage Plains	Eutrophic	7.3	22.6	4.3	0.4	66.8	0.0317
	Lake of the Ozarks	38.2039	-92.6236	Ozark Highlands	Eutrophic	12.0	45.1	207.7	2.5	24993.3	2.4916
	Lotawana	38.9353	-94.2464	Osage Plains	Eutrophic	7.5	17.7	1.9	0.4	34.6	0.0143
	Smithville	39.3631	-94.5567	Glacial Plains	Eutrophic	6.2	32.0	28.7	0.4	543.2	0.1784
	Truman	38.2639	-93.4083	Osage Plains	Eutrophic	6.8	38.4	217.8	2.8	18394.7	1.4835
Reference Reservoirs	Atkinson	38.0064	-94.0461	Osage Plains	Eutrophic	1.7	8.5	1.9	1.2	17.0	0.0034
	Binder	38.6019	-92.3008	Ozark Boarder	Eutrophic	6.9	14.6	0.6	1.0	15.8	0.0041
	Bowling Green	39.3420	-91.1531	Glacial Plains	Mesotrophic	19.5	22.3	0.1	0.4	3.4	0.0017
	Brookfield	39.7961	-93.0283	Glacial Plains	Eutrophic	5.1	13.1	0.5	0.2	2.7	0.0023
	Capri	37.8978	-90.6275	Ozark Highlands	Oligotrophic	8.3	29.6	8.8	0.2	2337.7	0.0035
	Clearwater	37.1339	-90.7722	Ozark Highlands	Mesotrophic	30.9	46.9	1.7	3.5	34.6	0.2714
	Council Bluff	37.7324	-90.9145	Ozark Highlands	Oligotrophic	9.0	33.5	0.2	0.8	2.4	0.0156
	Deer Ridge	40.1803	-91.8269	Glacial Plains	Eutrophic	3.8	11.6	2.9	0.6	51.4	0.0007
	Fellows	37.3153	-93.2300	Ozark Highlands	Mesotrophic	11.3	30.5	2.3	0.5	38.3	0.0324
	Forest	40.1714	-92.6548	Glacial Plains	Mesotrophic	6.6	20.1	0.6	0.4	8.0	0.0153
	Higginsville	39.0617	-93.6700	Glacial Plains	Eutrophic	2.9	10.4	1.9	1.3	17.0	0.0020
	Hunnewell	39.7072	-91.8619	Glacial Plains	Eutrophic	3.5	11.0	0.8	0.7	10.3	0.0027
	Kraut Run	38.7353	-90.7619	Glacial Plains	Eutrophic	2.2	6.7	0.6	3.3	16.6	0.0013
	Lincoln	39.0203	-90.9217	Glacial Plains	Mesotrophic	14.8	21.0	0.1	0.5	4.0	0.0019
	Long Branch	39.7514	-92.5150	Glacial Plains	Eutrophic	4.3	21.6	9.9	1.1	270.7	0.0427
	Mark Twain	39.5223	-91.6467	Glacial Plains	Eutrophic	8.2	34.1	75.0	1.9	6086.3	0.6168
	McDaniel	37.2939	-93.3142	Ozark Highlands	Eutrophic	5.4	14.6	1.0	5.6	100.5	0.0056
	North	38.6861	-94.3583	Osage Plains	Eutrophic	1.9	8.5	0.2	3.3	5.1	0.0003
	Pomme de Terre	37.9017	-93.3200	Ozark Highlands	Mesotrophic	8.0	47.2	37.7	1.4	1514.8	0.2997
	Shayne	37.8976	-90.6412	Ozark Highlands	Oligotrophic	10.3	21.9	0.3	0.2	2.1	0.0031
	Stockton	37.6917	-93.7667	Ozark Highlands	Mesotrophic	9.8	49.1	112.3	0.8	2974.2	1.1003
	Sugar Creek	39.4736	-92.4792	Glacial Plains	Eutrophic	4.9	14.9	1.3	0.8	28.7	0.0065
	Table Rock	36.5950	-93.3108	Ozark Highlands	Mesotrophic	20.6	76.8	161.5	0.5	4852.7	3.3329
	Viking	39.9378	-94.0569	Glacial Plains	Mesotrophic	6.5	25.9	2.1	0.4	36.5	0.0136
	Wappapello	36.9311	-90.2806	Ozark Highlands	Eutrophic	2.3	12.8	32.5	19.2	3363.3	0.0756
	Watkins Mill	39.3933	-94.2631	Glacial Plains	Eutrophic	7.3	15.8	0.4	0.4	7.8	0.0032

Reservoir surface area, volume, and watershed area are from the Missouri Department of Natural Resources (MDNR) and were determined during reservoir impoundment (Missouri DNR 2019). The maximum depth of each reservoir is based on dam height. Mean reservoir depth was calculated from volume and surface area. Flushing rate was calculated by dividing total runoff by reservoir volume, where total runoff was calculated for each reservoir by multiplying watershed area (Missouri DNR 2019) by runoff (Missouri DNR 1986).

While not directly measured in this study, previous reports of calcium concentrations and reservoir use help give context to Missouri reservoirs. Previous reported calcium concentrations in 80 out of 103 Missouri reservoirs are above 20 mg Ca L^− 1 (Wylie and Jones 1991), well above the 12 mg Ca L^− 1 threshold necessary for zebra mussel establishment (Cohen 2007). Zebra mussels are still spreading throughout Missouri and the reservoirs that are currently invaded may be related to the amount of boat traffic they receive. Based on a 2013 survey of Missouri anglers, zebra mussel reservoirs are visited 314% more often than reference reservoirs (Reitz 2015). Excluding Lake of the Ozarks, which receives almost twice as many anglers as any other reservoir in the state, invaded reservoirs received 179% more visitors than reference reservoirs (Reitz 2015).

Water Clarity Parameters

To assess whether zebra mussel introductions led to a shift in water clarity, we analyzed chl-a, total suspended solids (TSS), particulate organic matter (POM), particulate inorganic matter (PIM), and Secchi depth in both invaded and non-invaded reference reservoirs. Water samples are collected from the surface of each reservoir at the dam, except for Bull Shoals, which lacked a dam study site. Secchi depth and chl-a concentrations are measured at 2 locations in Bull Shoals. Bull Shoals site 1 (36.5169°N -92.6235°W) is 16.4 km up-reservoir of the dam in one of the main arms of the reservoir and has TSS and PIM data between 1978 and 2012. These parameters were never measured in more than 8 consecutive years, so while we included them in our habitat suitability index, we were not able to assess regime shifts nor trends. Site 2 (36.5013°N -92.9306°W) in Bull Shoals is 36.6 km up-reservoir of the dam, is located in a second main arm of the reservoir, and has data between 2003 and 2019. For all reservoirs, n = 1 to 37 water samples per reservoir per year are collected between May 15th and September 15th. We calculate the annual arithmetic mean for each parameter and use this metric in all statistical analyses.

To measure chl-a, water is filtered through 1.0 µm PALL AE glass fiber filter (GFF) prior to 2019, and through 0.7 µm Whatman GFF in 2019, and stored frozen until analysis. We quantified chl-a concentrations fluorometrically with a Turner Design 700 Fluorometer after ethanol extraction. We report total chl-a, which has not been acid corrected for pheophytin (Sartory and Grobbelaar 1984). The chl-a detection limit is 0.3 µg L^− 1. We measure TSS following standard methods (Method 2540 D and E; APHA 2017). Water is filtered through pre-weighed 1.5 µm, grade 934-AH Whatman filters, which are stored frozen until analysis. TSS filters are thawed and dried at 105°C for 30 minutes, weighed, and then incinerated at 550°C for 20 minutes to burn off organic material before being weighed again. This loss-on-ignition procedure allowed for the differentiation of TSS by subtracting the mass left after incineration, which is PIM, from the total filter mass before incineration, which is TSS. The difference is POM. TSS, PIM, and POM analyses have a detection limit of 0.1 mg L^− 1. Secchi depth is measured with a 20 cm black and white Secchi disk.

Zebra Mussel Density and Biomass

In summer 2019, we measured zebra mussel density and biomass in Bull Shoals (6/17/2019–6/21/2019), Lake of the Ozarks (5/29/2019, 6/12/2019, 7/11/2019, 7/29/2019, 8/28/2019), Lotawana (10/23/2019), and Smithville (5/22/2019, 6/10/2019, 7/17/2019, 8/19/2019). We did not measure zebra mussel density and biomass in Blue Springs, Jacomo, and Truman because at the time of sampling there were less than 5 years post-invasion water quality data. Zebra mussels were collected from transects placed perpendicular to the shore. The number of transects was based on reservoir size. Twelve transects were sampled in Bull Shoals, 12 in Lake of the Ozarks, 3 in Lotawana, and 10 in Smithville. The first transect was placed at the conjuncture of the dam and shoreline. Subsequent transects were evenly spaced around the shoreline, although some transects were moved to where previous observations of live mussels occurred (Benson et al. 2021a). To date, quagga mussels have only been observed in Missouri in the Mississippi River (Benson et al. 2021b) and we did not observe this invasive bivalve during our field sampling.

The number of sites sampled in each transect varied based on the slope of the littoral area, but typically 5 sites were sampled along each transect. Water levels are highly variable from year to year in several reservoirs, and we procured water level data from the Army Corps of Engineers for both Smithville¹ and Bull Shoals², while Lake of the Ozarks³ water levels were provided by Ameren Corporation. Daily water levels are not recorded for Lotawana but the reservoir was at full pool at the time of sampling. To avoid sampling sites that had been recently dry, we placed our shallowest site at 2 m deeper than the lowest water level recorded over the previous 6 years. Each subsequent site along the transect was placed 2 m deeper than the previous site. Sampling ceased when no mussels were collected at the previous 2 sites and the depth was 10 m deeper than the lowest water level recorded throughout the reservoir over the previous 6 years. The 6-year low in water levels for Lotawana is unknown, so we began sampling at 2 m.

At each site, 3 intact grab samples were collected with an Ekman dredge (0.023 m^− 2). Samples were rinsed through a 500 µm mesh and any live zebra mussels were retained. Each mussel’s length was measured (nearest 1.0 mm) from its umbo to its posterior margin and its weight measured (nearest 0.01 g) after opening its shell to remove water from the mantle cavity. All zebra mussels were counted, measured, and weighed within 48 hours of collection.

We calculated zebra mussel density and biomass for each sample site based on the area of the Ekman grab. For each reservoir, we report the average density across all sites as individuals per m² and the average live-weight biomass as g per m². These reservoir-wide density and biomass estimates reflect the nearshore area where our sampling occurred.

Temperature and Dissolved Oxygen Profiles

Temperature and dissolved oxygen profiles were measured one to 4 times each year between May 15th and September 15th. Profiles were taken near the dam at the deepest point in the reservoir using a Yellow Springs Instruments (YSI) multi-parameter sonde. No profiles were taken at Bull Shoals site 2. A YSI model 50B was used to take profiles prior to 2000, while a model 85 was used between 2000 and 2006, and a model 550A was used from 2006 to 2016. Prior to 2017, temperature (models 50B and 85: accuracy: ±0.1°C, resolution: 0.1°C; model 550A: accuracy: ±0.2°C, resolution: 0.1°C) and dissolved oxygen (all models accuracy: ±0.3 mg L^− 1, resolution: 0.01 mg L^− 1) measurements were made at discrete, vertical depths (0.5 m intervals). After 2016, dissolved oxygen (optical sensor; accuracy: ±0.1 mg L^− 1, resolution: 0.01 mg L^− 1) and temperature (accuracy: ±0.01°C, resolution: 0.001°C) profiles were continuous and collected using a YSI EXO3. Each profile was interpolated to 0.5 m using piecewise cubic hermite interpolating polynomial (PCHIP) with the pracma program in R (Borchers 2019). Mixing depth was calculated from the interpolated profile using rLakeAnalyzer (Winslow et al. 2019). If mixing depth was not identified in a profile, it was removed unless the profile extended to the reservoir bottom. Epilimnetic temperature and dissolved oxygen were determined for each profile by taking the mean of these parameters above the mixing depth. The depth where anoxia began, or anoxia depth, was considered to be the first depth where dissolved oxygen concentrations were ≤ 1 mg L^− 1 (Jones et al. 2011). Annual means of mixing depth, anoxia depth, epilimnetic temperature, and epilimnetic dissolved oxygen were used for statistical analyses.

Statistical Analyses

We looked for long-term trends in water clarity parameters for each reservoir. For each year of collection, the mean of each parameter was calculated and analyzed using a sequential t-test analysis of regime shifts (Rodionov 2004). We set the significance level to 0.05, the cut-off length to 10 years, and Huber’s Weight Parameter to one. The number of observations with this analysis are not fixed, meaning that a new test is run on each new data point (Rodionov 2004). This gives it an advantage over other regime shift tests because it can identify shifts as soon as they occur, not several years later (Rodionov and Overland 2005). Next, we used a Mann-Kendall trend test before and after each regime shift, or on the entire data series when no regime shift existed, to determine the presence of monotonic trends through time (Pohlert 2020). The year of the regime shift reported was considered after the shift. If a significant monotonic trend was detected, we determined its magnitude using a Sen’s slope calculation (Pohlert 2020). We report significant monotonic trends, as well as the magnitude of these trends for each reservoir and water clarity parameter. We also report the mean of each parameter before and after each regime shift, or on the entire data series if no regime shift existed. The year of the regime shift was included in the calculation of the mean for after the shift. If a regime shift created a period with fewer than 3 years of data, no statistical tests nor calculations were performed on that period.

We determined regime shifts in Secchi depth (Table 2), TSS (Table 3), PIM (Table 4), and chl-a (Table 5) in invaded reservoirs. A regime shift corresponding with the invasion date indicates that zebra mussels could be having an impact, but the direction and trend in water clarity after the shift is indicative of whether that impact is resulting in increased water clarity due to zebra mussel grazing. Increasing water clarity is associated with an increase in Secchi depth and a decrease in TSS, PIM, and chl-a. An increase in water clarity after the regime shift that corresponds with invasion date could indicate that zebra mussels are having an impact. Likewise, a declining trend in water clarity that is followed by no trend after a regime shift corresponding with invasion date suggests that zebra mussels could be having a moderating impact. We used a Mann-Kendall test and Sen’s slope calculation to determine whether there was an increasing trend in water clarity after the shift occurred, which would further suggest that the introduction of zebra mussels was leading to an increase in water clarity. Considering 4 indicators of water clarity in each invaded reservoir enabled us to establish a weight of evidence assessment about the impacts of zebra mussels on invaded reservoirs.

Table 2

Regime shifts in Secchi depth for invaded and reference reservoirs. The year that a regime shift occurs is identified. When no regime shift occurs in a reservoir, “none” is listed, and when more than one regime shift occurs in a reservoir, multiple years are listed. The number of years is the mean of all years within each period, and the Mann-Kendall p-value indicates whether a significant monotonic trend occurred within a period before and/or after a regime shift, or when no regime shift occurred, the entire data series. When sample size is less than 3, as indicated by “†”, no mean Secchi depth nor Mann-Kendall p-value are shown. As an example, Lake of the Ozarks displayed regime shifts in 1986 and 2007. The number of years (10), mean Secchi depth of 1.38, and Mann-Kendall p-value of 0.59 correspond with the period before 1986. The sample size of 21, mean Secchi depth of 1.65, and Mann-Kendall p-value of 0.65 correspond with the period from 1986 to 2006, and the sample size of 13, mean Secchi depth of 1.25, and the Mann-Kendal p-value of 0.08 correspond with the period from 2007 to 2019. When a significant monotonic trend is identified (Mann-Kendall < 0.05, “*”), the Sen’s slope value identifies the direction and magnitude of the trend.
	Reservoir	Invasion Year	Regime Shift Year(s)	Number of Years	Mean Secchi Depth (m)	Mann-Kendall p-value	Sen’s Slope (m yr^− 1)
Zebra Mussel Reservoirs	Blue Springs	2014	none	27	1.12	0.05*	0.01
	Bull Shoals site 1	2006	1996, 2012†	9, 10, 1	2.41, 4.07	0.40, 0.59	-
	Bull Shoals site 2	2006	2019†	16, 1	3.14	0.69	-
	Jacomo	2014	none	29	1.24	0.08	-
	Lake of the Ozarks	1999	1986, 2007	10, 21, 13	1.38, 1.65, 1.25	0.59, 0.65, 0.08	-
	Lotawana	2006	none	22	1.42	0.48	-
	Smithville	2007	2016	30, 4	1.01, 0.81	1, 0.09	-
	Truman	2011	none	34	1.21	0.64	-
Reference Reservoirs	Atkinson	n/a	2018†	29, 1	0.52	0.28	-
	Binder	n/a	1996	10, 23	1.11, 0.80	0.06, 0.05*	-0.01
	Bowling Green	n/a	2006	18, 14	1.95, 1.53	0.16, 0.51	-
	Brookfield	n/a	none	32	1.29	0.53	-
	Capri	n/a	2000	14, 20	4.23, 5.07	0.23, 0.45	-
	Clearwater	n/a	2008	22, 12	1.91, 1.49	0.23, 0.37	-
	Council Bluff	n/a	2019†	30, 1	3.42	0.09	-
	Deer Ridge	n/a	none	32	1.03	0.60	-
	Fellows	n/a	2019†	27, 1	2.77	0.56	-
	Forest	n/a	2008, 2019†	21, 11, 1	1.39, 1.02	1, 0.04*	0.04
	Higginsville	n/a	2015	26, 5	0.64, 0.46	0.08, 0.81	-
	Hunnewell	n/a	2003, 2011	15, 8, 9	0.90, 1.28, 0.96	0.20, 0.90, 0.47	-
	Kraut Run	n/a	none	26	0.48	0.71	-
	Lincoln	n/a	none	33	2.23	1	-
	Long Branch	n/a	1992	10, 27	0.85, 0.64	1, 0.36	-
	Mark Twain	n/a	none	32	1.00	0.60	-
	McDaniel	n/a	2013	22, 3	1.31, 1.75	0.27, 0.30	-
	North	n/a	2019†	30, 1	0.68	0.13	-
	Pomme de Terre	n/a	2008, 2019†	23, 11, 1	1.80, 1.23	0.56, 0.21	-
	Shayne	n/a	2009	19, 11	3.00, 4.09	0.75, 1	-
	Stockton	n/a	2008	23, 12	2.94, 1.96	0.40, 0.84	-
	Sugar Creek	n/a	none	34	0.82	0.43	-
	Table Rock	n/a	1991, 2019†	7, 28, 1	3.79, 2.19	0.23, 0.40	-
	Viking	n/a	2012	24, 4	1.35, 1.79	0.14, 0.09	-
	Wappapello	n/a	1999	13, 21	1.00, 0.78	0.67, 0.61	-
	Watkins Mill	n/a	2018†	28, 2	0.87	0.48	-
†Indicates that a regime shift was identified in the last 3 years of recorded data. No Mann-Kendall p-value is provided for these dates because this test requires a minimum of 3 values

Table 3

Regime shifts in total suspended solids (TSS) for invaded and reference reservoirs. The year that a regime shift occurs is identified. When no regime shift occurs in a reservoir, “none” is listed, and when more than one regime shift occurs in a reservoir, multiple years are listed. The number of years is the mean of all years within each period, and the Mann-Kendall p-value indicates whether a significant monotonic trend occurred within a period before and/or after a regime shift, or when no regime shift occurred, the entire data series. When sample size is less than 3, as indicated by “†”, no mean TSS nor Mann-Kendall p-value are. As an example, Lake of the Ozarks displayed regime shifts in 2007 and 2018. The number of years (28), mean TSS of 5.89, and Mann-Kendall p-value of 0.02 correspond with the period before 2007. The sample size of 11, mean TSS of 7.74, and Mann-Kendall p-value of 1 correspond with the period from 2007 to 2017. No mean TSS and Mann-Kendall p-value are reported for the period after 2017 because the sample size for this period is less than 3. When a significant monotonic trend is identified (Mann-Kendall < 0.05, “*”), the Sen’s slope value identifies the direction and magnitude of the trend. In the previous Lake of the Ozarks example, a Sen’s slope calculation was performed on the period before 2007, and a -0.09 mg L-1 year-1 slope was identified.
	Reservoir	Invasion Year	Regime Shift Year(s)	Number of Years	Mean TSS (mg L^− 1)	Mann-Kendall p-value	Sen’s Slope (mg L^− 1 yr^− 1)
Zebra Mussel Reservoirs	Blue Springs	2014	none	26	6.49	0.89	-
	Jacomo	2014	none	27	5.55	0.29	-
	Lake of the Ozarks	1999	2007, 2018†	28, 11, 2	5.89, 7.74	0.02*, 1	-0.09
	Lotawana	2006	2010, 2019†	11, 9, 1	5.79, 3.81	0.76, 0.25	-
	Smithville	2007	none	34	8.13	0.03*	0.07
	Truman	2011	2011	27, 7	6.60, 7.58	0.44, 0.55	-
Reference Reservoirs	Atkinson	n/a	2017†	28, 2	16.09	0.57	-
	Binder	n/a	2007, 2018†	20, 10, 2	7.92, 10.74	0.38, 0.86	-
	Bowling Green	n/a	1998, 2019†	10, 20, 1	4.86, 2.91	0.79, 0.87	-
	Brookfield	n/a	none	31	5.72	0.06	-
	Capri	n/a	2000	13, 20	1.60, 1.20	0.95, 0.47	-
	Clearwater	n/a	2005, 2018†	18, 11, 2	3.35, 7.02	0.82, 0.64	-
	Council Bluff	n/a	2018†	29, 2	1.56	0.42	-
	Deer Ridge	n/a	2010	21, 10	6.71, 9.90	0.01*, 0.15	-0.30
	Fellows	n/a	2019†	27, 1	2.29	0.82	-
	Forest	n/a	2008, 2016	20, 8, 4	5.57, 11.49, 6.91	0.97, 0.39, 0.31	-
	Higginsville	n/a	none	30	14.88	0.01*	0.24
	Hunnewell	n/a	none	29	6.81	1	-
	Kraut Run	n/a	none	26	16.66	0.52	-
	Lincoln	n/a	none	32	3.43	0.05*	-0.03
	Long Branch	n/a	none	36	10.58	0.61	-
	Mark Twain	n/a	2019†	31, 1	6.94	0.68	-
	McDaniel	n/a	2004, 2014†	13, 10, 2	4.64, 5.40	0.33, 0.47	-
	North	n/a	2013	24, 7	11.87, 16.90	0.98, 1	-
	Pomme de Terre	n/a	2008, 2014	22, 6, 6	3.79, 6.58, 5.04	0.45, 0.06, 1	-
	Shayne	n/a	2019†	23, 1	2.03	0.26	-
	Stockton	n/a	2008, 2015	22, 7, 5	2.83, 5.97, 9.34	0.35, 0.23, 0.09	-
	Sugar Creek	n/a	none	31	9.84	0.93	-
	Table Rock	n/a	none	23	2.24	0.63	-
	Viking	n/a	none	27	5.73	0.02*	-0.10
	Wappapello	n/a	2010	23, 9	9.05, 11.52	0.03*, 0.25	0.15
	Watkins Mill	n/a	2018†	28, 2	8.25	0.40	-
†Indicates that a regime shift was identified in the last 3 years of recorded data. No Mann-Kendall p-value is provided for these dates because this test requires a minimum of 3 values

Table 4

Regime shifts in particulate inorganic matter (PIM) for invaded and reference reservoirs. The year that a regime shift occurs is identified. When no regime shift occurs in a reservoir, “none” is listed, and when more than one regime shift occurs in a reservoir, multiple years are listed. The number of years is the mean of all years within each period, and the Mann-Kendall p-value indicates whether a significant monotonic trend occurred within a period before and/or after a regime shift, or when no regime shift occurred, the entire data series. When sample size is less than 3, as indicated by “†”, no mean POM nor Mann-Kendall p-value are shown. As an example, Lake of the Ozarks does not display a regime shift in PIM. The number of years (41), mean PIM of 3.63, and Mann-Kendall p-value of 0.16 correspond with the entire period. When a significant monotonic trend is identified (Mann-Kendall < 0.05, “*”), the Sen’s slope value identifies the direction and magnitude of the trend.
	Reservoir	Invasion Year	Regime Shift Year(s)	Number of Years	Mean PIM (mg L^− 1)	Mann-Kendall p-value	Sen’s Slope (mg L^− 1 yr^− 1)
Zebra Mussel Reservoirs	Blue Springs	2014	2014	22, 5	3.05, 2.33	0.16, 0.46	-
	Jacomo	2014	2001, 2019†	11, 16, 1	2.84, 1.66	0.88, 0.21	-
	Lake of the Ozarks	1999	none	41	3.63	0.16	-
	Lotawana	2006	2009	10, 11	3.02, 1.15	1, 0.01*	-0.06
	Smithville	2007	none	34	4.42	0.84	-
	Truman	2011	2011	27, 7	3.99, 4.15	0.32, 0.76	-
Reference Reservoirs	Atkinson	n/a	2017†	28, 2	9.50	0.02*	-0.23
	Binder	n/a	2019†	31, 1	3.20	0.03*	-0.06
	Bowling Green	n/a	none	32	1.68	0.01*	-0.04
	Brookfield	n/a	2001, 2018†	13, 16, 2	3.99, 3.23	0.62, 0.30	-
	Capri	n/a	none	33	0.60	0.01*	-0.01
	Clearwater	n/a	2005, 2014	18, 9, 6	2.01, 4.25, 2.42	0.94, 0.18, 0.71	-
	Council Bluff	n/a	none	31	0.68	0.45	-
	Deer Ridge	n/a	2002	13, 18	4.83, 2.55	0.50, 0.26	-
	Fellows	n/a	none	28	0.85	0.06	-
	Forest	n/a	2008	20, 12	4.11, 6.71	0.72, 0.15	-
	Higginsville	n/a	none	30	10.08	0.46	-
	Hunnewell	n/a	2003, 2019†	14, 16, 1	3.82, 2.38	0.38, 0.26	-
	Kraut Run	n/a	none	26	6.57	0.35	-
	Lincoln	n/a	none	32	1.70	0.01*	-0.03
	Long Branch	n/a	none	36	7.43	1	-
	Mark Twain	n/a	none	32	4.22	0.59	-
	McDaniel	n/a	2004, 2014†	13, 10, 2	1.56, 2.11	0.27, 0.37	-
	North	n/a	2019†	30, 1	5.74	0.75	-
	Pomme de Terre	n/a	2008, 2018†	22, 10, 2	1.20, 2.14	0.01, 0.01	-0.04, -0.22
	Shayne	n/a	2008, 2018†	18, 10, 2	1.44, 0.71	0.45, 0.03*	-0.05
	Stockton	n/a	2009, 2015	23, 6, 5	1.30, 3.00, 4.93	0.75, 1, 0.09	-
	Sugar Creek	n/a	2014	26, 5	5.80, 4.59	0.19, 1	-
	Table Rock	n/a	none	29	0.94	0.46	-
	Viking	n/a	none	27	3.50	0.01*	-0.07
	Wappapello	n/a	2009, 2019†	22, 10, 1	4.19, 6.01	0.40, 0.37	-
	Watkins Mill	n/a	1997, 2018†	9, 19, 2	5.67, 4.15	0.60, 0.02*	0.10
†Indicates that a regime shift was identified in the last 3 years of recorded data. No Mann-Kendall p-value is provided for these dates because this test requires a minimum of 3 values

Table 5

Regime shifts in chlorophyll-a uncorrected for pheophytin (chl-a) for invaded and reference reservoirs. The year that a regime shift occurs is identified. When no regime shift occurs in a reservoir, “none” is listed, and when more than 1 regime shift occurs in a reservoir, multiple years are listed. The number of years is the mean of all years within each period, and the Mann-Kendall p-value indicates whether a significant monotonic trend occurred within a period before and/or after a regime shift, or when no regime shift occurred, the entire data series. When sample size is less than 3, as indicated by “†”, no mean chl-a nor Mann-Kendall p-value are shown. As an example, Lake of the Ozarks displayed regime shifts in 2007 and 2018. The number of years (31), mean chl-a of 15.58, and Mann-Kendall p-value of 0.03 correspond with the period before 2007. The sample size of 11, mean chl-a of 23.34, and Mann-Kendall p-value of 0.88 correspond with the period from 2007 to 2017. No mean chl-a and Mann-Kendall p-value are reported for the period after 2017 because the sample size for this period is less than 3. When a significant monotonic trend is identified (Mann-Kendall < 0.05, “*”), the Sen’s slope value identifies the direction and magnitude of the trend. In the previous Lake of the Ozarks example, a Sen’s slope calculation was performed on the period before 2007, and a 0.16 µg L-1 year-1 slope was identified.
	Reservoir	Invasion Year	Regime Shift Year(s)	Number of Years	Mean Chl-a (µg L^− 1)	Mann-Kendall p-value	Sen’s Slope (µg L^− 1 yr^− 1)
Zebra Mussel Reservoirs	Blue Springs	2014	none	28	19.76	0.14	-
	Bull Shoals site 1	2006	1996	9, 11	6.20, 2.96	0.34, 0.06	-
	Bull Shoals site 2	2006	none	17	4.47	0.80	-
	Jacomo	2014	none	29	18.22	0.46	-
	Lake of the Ozarks	1999	2007, 2018†	31, 11, 2	15.58, 23.34	0.03*, 0.88	0.16
	Lotawana	2006	none	22	18.16	0.96	-
	Smithville	2007	2012	26, 8	18.26, 29.43	0.05*, 0.54	0.28
	Truman	2011	2005, 2015	21, 10, 3	15.74, 22.13, 19.94	0.16, 0.37, 1	-
Reference Reservoirs	Atkinson	n/a	none	30	41.78	0.18	-
	Binder	n/a	1998	12, 21	21.90, 42.81	0.95, 0.25	-
	Bowling Green	n/a	none	32	8.07	0.09	-
	Brookfield	n/a	none	32	7.94	0.24	-
	Capri	n/a	2018†	32, 2	1.58	0.09	-
	Clearwater	n/a	2017	31, 3	6.01, 8.15	0.01*, 0.30	0.10
	Council Bluff	n/a	2017	28, 3	2.24, 1.73	0.41, 0.30	-
	Deer Ridge	n/a	2009, 2017	21, 8, 3	14.23, 44.77, 28.03	0.38, 0.90, 1	-
	Fellows	n/a	none	28	5.10	0.66	-
	Forest	n/a	2005, 2016	18, 11, 4	5.05, 9.78, 6.76	0.24, 1, 0.31	-
	Higginsville	n/a	2004, 2019†	15, 15, 1	23.28, 53.30	0.17, 0.14
	Hunnewell	n/a	2012	24, 8	19.75, 29.35	1, 0.06	-
	Kraut Run	n/a	2004	15, 11	59.96, 87.84	0.10, 0.76	-
	Lincoln	n/a	none	32	4.94	0.80	-
	Long Branch	n/a	2004	21, 16	12.60, 23.14	0.35, 0.22	-
	Mark Twain	n/a	none	32	18.09	0.37	-
	McDaniel	n/a	2019†	27, 1	16.41	0.80	-
	North	n/a	2001, 2016	12, 15, 4	39.19, 58.81, 97.42	0.11, 1, 0.73	-
	Pomme de Terre	n/a	2008, 2019†	23, 11, 1	14.52, 19.78	0.83, 0.12	-
	Shayne	n/a	2014	24, 6	1.25, 1.89	0.11, 0.13	-
	Stockton	n/a	2008, 2019†	23, 11, 1	6.95, 21.52	0.11, 0.35	-
	Sugar Creek	n/a	2019†	33, 1	23.97	0.20	-
	Table Rock	n/a	1992, 2019†	8, 27	4.48, 13.66, 9.59	0.02*, 0.45	0.55
	Viking	n/a	none	28	9.36	1	-
	Wappapello	n/a	2004	18, 16	24.71, 34.90	0.20, 0.96	-
	Watkins Mill	n/a	none	30	19.31	0.35	-
†Indicates that a regime shift was identified in the last 3 years of recorded data. No Mann-Kendall p-value is provided for these dates because this test requires a minimum of 3 values

We assessed habitat suitability of invaded and reference reservoirs using a correlation (scaling 2) principal component analysis (PCA) with no rotation. Vectors for the PCA included mixing depth, anoxia depth, epilimnetic temperature, epilimnetic dissolved oxygen, TSS, PIM, and the ratio of PIM:POM. PIM:POM is reported as a natural log to reduce bias associated with taking the average of ratios (Isles 2020). Bull Shoals site 2 was excluded from the PCA because no epilimnetic temperature nor dissolved oxygen data exists for this site. For each reservoir, the mean for each of these parameters was calculated from the entire long-term dataset and then converted to a z-score to standardize PCA inputs. The PCA was calculated using Program R (R Core Team 2019) and was plotted using the factoextra package (Kassambara and Mundt 2020).

We looked at the relationship between POM and chl-a to assess whether photoacclimation was influencing our findings. Phytoplankton can produce increased chl-a in low light conditions and, if this occurs, chl-a may over represent phytoplankton biomass. If light availability has decreased over time in Missouri reservoirs, chl-a may be increasing even though actual biomass is not. This could influence our assessment of temporal trends in chl-a. Neither POM nor chl-a were normally distributed, even after transformation, so we ran a Spearman’s rank correlation test on these parameters in both invaded and reference reservoirs.

Study Reservoirs

Study reservoirs were located throughout the state and ranged in size, depth, volume, trophic status, and watershed size (Table 1). Four invaded and 2 references reservoirs were in the Osage Plains, while 2 invaded and 10 reference reservoirs were in the Ozark Highlands (Jones et al. 2008). Both regions are unglaciated, but the Osage Plains overlay limestone, sandstone, and shale, while the Ozark Highlands are characterized by cherty soils (Jones and Knowlton 1993). One invaded and 13 reference reservoirs were in the loess and glacial till dominated Glacial Plains, and one reference reservoir was in the Ozark Boarder (Jones and Knowlton 1993; Jones et al. 2008; Table 1). Photoacclimation (Falkowski and LaRoche 1991) did not affect our findings, as POM and chl-a were substantially correlated for the entire long-term dataset in both reference (p < 0.0001, rho = 0.91) and invaded reservoirs (p < 0.0001, rho = 0.78) and in the latter, both before (p < 0.0001, rho = 0.77) and after (p < 0.0001, rho = 0.78) invasion.

Regime Shifts in Invaded Reservoirs

In Blue Springs, a PIM regime shift occurred during the zebra mussel invasion range. Mean PIM concentrations were 24% lower after the shift than before, and there was no trend before nor after (Fig. 1). No regime shifts were identified during nor after the zebra mussel invasion range for Secchi depth nor TSS. A chl-a regime shift occurred 2 years after the zebra mussel invasion range, but there was only one year post-invasion to detect a trend. Of the 4 parameters considered, PIM was the only one consistent with what would be expected if zebra mussels were having an impact.

In Bull Shoals, both Secchi depth and chl-a concentrations were measured from both sites. At site 1, a Secchi regime shift occurred 3 years after the zebra mussel invasion range (Fig. 2). After this regime shift, a trend was not identified because there was only one year post shift (Table 2). At Bull Shoals site 2, a Secchi regime shift occurred 10 years after the zebra mussel invasion range (Fig. 2). After this regime shift, a trend was not identified because there was only one year post shift (Table 2). No chl-a regime shift was identified at site 2 during or after the zebra mussel invasion range, and there was no trend in this parameter during the study period. Neither water clarity parameter in Bull Shoals increased after the zebra mussel invasion, suggesting that mussels did not have an impact.

In Jacomo, a PIM regime shift occurred 2 years after the zebra mussel invasion range (Fig. 3). After this regime shift, a trend was not identified because there was only one year post shift (Table 4). No regime shifts nor trends were identified during or after the zebra mussel invasion range for Secchi depth, TSS, nor chl-a. Of the 4 water clarity parameters, none were consistent with what would be expected if zebra mussels were having an impact.

In Lake of the Ozarks, a Secchi regime shift occurred 8 years after the zebra mussel invasion range (Fig. 4). Mean Secchi depth decreased after the shift and there was no trend pre- or post- shift (Table 2). A regime shift in TSS and chl-a occurred 8 years after the zebra mussel invasion range (Fig. 4). Before the TSS shift, a decreasing trend of 0.09 mg L^− 1 yr^− 1 was identified (Table 3). After the shift, mean TSS increased but there was no trend (Table 3). A second TSS shift occurred 19 years after the zebra mussel invasion range, but there were only 2 years post-shift. For chl-a, there was an increasing trend of 0.16 g^− 1 L^− 1 yr^− 1 until the regime shift (Table 5). Mean chl-a increased after this shift, but no trend was identified. No regime shift nor trend was identified during or after the zebra mussel invasion range for PIM. Of the 4 water clarity parameters, none were consistent with what would be expected if zebra mussels were having an impact.

In Lotawana, a PIM regime shift occurred during the zebra mussel invasion range. After the regime shift, mean PIM decreased with a declining trend of 0.06 mg L^− 1 yr^− 1 (Table 4). Regime shifts in TSS occurred one and 10 years after the zebra mussel invasion range (Fig. 5). Mean TSS decreased after the first shift, but no trend was identified pre- nor post- regime shift (Table 3). There were only 2 years after the second regime shift so we were unable to determine a trend. No regime shifts nor trends were identified during or after the zebra mussel invasion range for Secchi depth nor chl-a. We examined 4 water clarity parameters for Lotawana, 2 of which, TSS and PIM, decreased after the zebra mussel invasion, as would be expected if zebra mussels were having an impact.

In Smithville, no regime shifts occurred during the zebra mussel invasion range. Regime shifts in Secchi depth and chl-a occurred 6 and 2 years, respectively, after the zebra mussel invasion range (Fig. 6), although there was not a trend after the regime shift for either parameter (Tables 2, 5). Chl-a increased after the shift, which was preceded by an increasing trend of 0.28 µg L^− 1 yr^− 1, or 1% yr^− 1, and followed by no trend. Despite no identified regime shift, an increasing trend of 0.09 mg L^− 1 yr^− 1 in TSS was identified throughout the study period. No regime shifts nor trends were identified during or after the zebra mussel invasion range for Secchi depth, TSS and PIM. None of the 4 water clarity parameters we examined in Smithville were consistent with what would be expected if zebra mussels were having an impact.

In Truman, a regime shift occurred during the zebra mussel invasion range for TSS and PIM (Fig. 7), but the mean of both parameters increased and there was no trend after the shift occurred. A chl-a regime shift occurred 4 years after the zebra mussel invasion range, but there was only one-year post-shift (Table 5). No Secchi regime shift nor trend was identified during or after the zebra mussel invasion. Of the 4 water clarity parameters we examined in Truman, none provided evidence that zebra mussels were having an effect.

To summarize, an increase in water clarity consistent with what we would expect if zebra mussels were having an impact occurred in 2 out of the 4 water quality parameters in Lotawana, and only one parameter in Blue Springs and Truman. No parameters in Bull Shoals, Jacomo, Smithville, nor Lake of the Ozarks displayed a pattern that was consistent with the expected change in water clarity if zebra mussels were having an impact.

Regime Shifts in Reference Reservoirs

We looked for regime shifts in Secchi depth, TSS, PIM, and chl-a in 26 reference reservoirs to determine statewide trends in water clarity that should be accounted for in our analysis of invaded reservoirs and might be indicative of a larger pattern. If water clarity did not increase in invaded reservoirs but decreased in reference reservoirs, it could suggest that zebra mussels were having a moderating effect on water clarity. In other words, water clarity might look static in invaded reservoirs if impacts from zebra mussels were offset by external factors causing water clarity to decrease throughout the state and over time.

A Secchi regime shift occurred in 77% of reference reservoirs. Mean Secchi depth increased post shift in 50% and decreased in 4% of reservoirs. In 8% of reservoirs, there were 2 regime shifts. Mean Secchi depth increased after the first shift but decreased after the second. Half of the reservoirs saw a significant trend after the regime shift occurred. This trend was positive in Forest, indicating a yearly increase in water clarity, but was negative in Binder.

For TSS, 65% of reference reservoirs experienced a regime shift, but mean TSS increased in 19% and decreased in 8% of reservoirs. In 8% of reservoirs, there were 2 regime shifts where mean TSS increased after the first shift and decreased after the second shift. Nineteen percent of reservoirs saw a trend after the regime shift occurred. This trend was positive in Higginsville and Wappapello, indicating a decreasing trend in water clarity, but negative in Deer Ridge, Lincoln, and Viking. Like Secchi depth, there was only one reservoir where a regime shift was followed by a trend showing an increase in water clarity.

Fifty-eight percent of reference reservoirs experienced a PIM regime shift. Mean PIM increased in 19% of those reservoirs and decreased in 23%. After the shift, a trend occurred in 35% of reservoirs. This trend was positive only in Watkins Mill, showing a decrease in water clarity. A negative trend, and increase in water clarity, occurred after a regime shift in Atkinson, Binder, Bowling Green, Capri, Lincoln, Pomme de Terre, Shayne, and Viking.

For chl-a, a regime shift occurred in 69% of reference reservoirs. Mean chl-a increased in 46% of reservoirs and decreased in 4%. No trends were observed after a regime shift in any reference reservoirs, but we identified positive trends in chl-a before a regime shift in both Clearwater and Table Rock.

To summarize, of the 4 indicators of water clarity that we examined, we did not observe a consistent pattern throughout the reference reservoirs. Water clarity increased in 46, 8, 23, and 4% for Secchi depth, TSS, PIM and chl-a, respectively, but decreased in 4, 27, 19, and 46% of these same parameters. Consistent water clarity decreases in reference reservoirs could suggest that zebra mussels are having a moderating effect in invaded reservoirs, where their impacts are being masked by broader trends. Our weight of evidence shows that no overarching trends exist in reference reservoirs.

Zebra Mussel Density and Biomass

Zebra mussel density, biomass, and length were measured in Bull Shoals, Lake of the Ozarks, Lotawana, and Smithville (Table 6). Lake of the Ozarks had the highest mean density per site (64.72 mussels m^− 2), followed by Smithville (25.76 mussels m^− 2), Lotawana (2.34 mussels m^− 2), and Bull Shoals (1.75 mussels m^− 2). Smithville had the greatest mean live weight mussel biomass per site (10.10 g m^− 2), followed by Lake of the Ozarks (6.69 g m^− 2), Bull Shoals (2.28 g m^− 2), and Lotawana (0.20 g m^− 2). Mean zebra mussel length was highest in Bull Shoals (22.6 mm), followed by Smithville (10.3 mm), Lotawana (9.5 mm), and Lake of the Ozarks (9.5 mm).

Table 6

Zebra mussel density and biomass from 4 invaded Missouri reservoirs. Density (individuals m-2) and biomass (live weight, g m-2) are means of all sampling sites within each reservoir. Standard error is also reported. Mussels were collected in the summer of 2019.
Invaded Reservoir	Number of Sampling Sites	Density (mussels m^− 2)	Biomass (g m^− 2)
Bull Shoals	58	1.75 ± 1.07	2.28 ± 1.54
Lake of the Ozarks	58	64.72 ± 32.42	6.69 ± 2.83
Lotawana	11	2.34 ± 1.77	0.20 ± 0.14
Smithville	36	25.76 ± 14.92	10.10 ± 10.10

Habitat Suitability

We initially included 7 environmental variables in our PCA (Table 7) but ultimately removed TSS, PIM:POM, and mixing depth because they explained little variance. When TSS was included, the first 2 components only explained an additional 0.4% of the variation in the study reservoirs. The inclusion of PIM:POM and mixing depth reduced the amount of variation explained by 9.8 and 5.9%, respectively. When PIM, dissolved oxygen, anoxia depth, and epilimnetic temperature were included in the PCA, the first component explained 50.1% of the variation in all study reservoirs and was characterized predominantly by PIM, anoxia depth, and epilimnetic temperature (Fig. 8). The second component, which explained 33.7% of the variation in the dataset, was characterized by epilimnetic dissolved oxygen. Of the 7 invaded reservoirs we were able to include in the PCA, only one was characterized by higher PIM concentrations and epilimnetic temperatures. Four invaded reservoirs were characterized by lower epilimnetic temperatures, because they were located opposite the temperature vector, while 2 of these were also characterized by lower PIM concentrations and 3 by deeper depths for the onset of anoxia because of their location opposite these vectors. Epilimnetic dissolved oxygen was the primary driver of variation in 2 of the invaded reservoirs. Reference reservoirs overlapped with invaded reservoirs but displayed much greater variability in the parameters they were characterized by, as indicated by their spread throughout the PCA biplot. Of the 26 reference reservoirs in this study, 7 were located opposite the epilimnetic temperature vector compared to 3 located along it, indicating that reference reservoirs were also more often characterized by lower epilimnetic temperatures.

Table 7

Parameters considered for habitat suitability using a principal component analysis (PCA). The top number is the mean for each reservoir, while the bottom number in parentheses is the range (min-max). Temperature and dissolved oxygen (DO) are means of the epilimnion, while total suspended solids (TSS) and particulate inorganic matter (PIM) were measured from surface water samples. We report the ratio of PIM to POM as a natural log, ln(PIM):ln(POM), to reduce bias associate with calculating the mean from ratios. Anoxia depth is the depth at which DO concentrations are below 1 mg L-1. Ultimately, we did not include TSS, PIM:POM, and mixing depth in the PCA because they explained little variance. For Bull Shoals, only site 1 was included in the habitat suitability analysis because no temperature nor dissolved oxygen data exists for Bull Shoals site 2.
	Reservoir	Mixing Depth (m)	Epilimnetic Temperature (°C)	Epilimnetic DO (mg L^− 1)	Anoxia Depth (m)	TSS (mg L^− 1)	PIM (mg L^− 1)	PIM:POM
Zebra Mussel Reservoirs	Blue Springs	2.07 (0.77–3.25)	26.41 (25.22–28.33)	7.9 (6.6–9.3)	4.67 (2.50–7.00)	6.49 (4.80–7.68)	2.89 (1.86–4.47)	-0.25 (-1.21–0.58)
	Bull Shoals site 1	5.18 (4.13–6.13)	27.30 (24.05–31.07)	8.3 (7.3–9.4)	8.44 (8.00–9.33)	3.32 (2.77–4.13)	1.45 (0.63–2.13)	-0.44 (-1.41–0.21)
	Jacomo	3.81 (2.89–5.05)	26.11 (23.24–29.55)	8.1 (7.4–8.8)	6.15 (5.00–7.00)	5.55 (2.50–9.87)	2.15 (1.10–4.48)	-0.51 (-1.40–0.08)
	Lotawana	4.23 (3.11–6.15)	25.96 (22.72–29.33)	7.7 (5.5–8.7)	6.43 (6.00–7.125)	4.72 (1.19–8.40)	2.04 (0.66–6.30)	-0.46 (-1.54–1.03)
	Lake of the Ozarks	7.75 (2.00–14.61)	25.88 (21.61–29.68)	7.4 (3.8–13.2)	11.79 (6.00–23.50)	6.37 (3.50–11.89)	3.63 (1.35–8.55)	-0.68 (-1.77–0.50)
	Smithville	5.35 (3.64–10.30)	25.73 (22.92–28.13)	7.7 (4.8–10.5)	7.71 (5.00–11.00)	8.13 (4.65–20.40)	4.42 (2.20–6.99)	0.25 (-0.48–1.12)
	Truman	5.41 (1.01–8.21)	25.98 (21.27–29.31)	7.4 (4.7–11.5)	7.73 (5.00–9.67)	6.80 (3.60–13.68)	4.02 (1.43–10.13)	0.18 (-0.96–1.49)
Reference Reservoirs	Atkinson	1.82 (0.96–2.86)	27.85 (23.67–33.00)	7.4 (4.1–10.1)	3.57 (2.50–5.80)	15.71 (7.47–30.93)	9.14 (3.55–19.37)	0.26 (-0.84–1.07)
	Binder	2.65 (1.59–3.80)	26.14 (23.79–28.81)	7.4 (4.1–9.5)	3.91 (3.13–5.00)	8.85 (4.00–15.70)	3.14 (1.05–10.47)	-0.70 (-2.37–1.07)
	Bowling Green	2.35 (1.48–5.51)	25.16 (22.8–28.9)	7.5 (4.8–10.2)	6.35 (3.38–14.50)	3.55 (1.55–12.20)	1.68 (0.55–8.43)	-0.37 (-1.33–0.81)
	Brookfield	3.07 (2.14–4.27)	25.30 (21.56–27.66)	7.6 (5.3–10.2)	5.46 (4.33–7.50)	5.66 (2.77–9.07)	3.59 (1.53–5.83)	0.56 (-0.03–1.16)
	Capri	4.57 (2.75–5.95)	25.58 (23.26–28.09)	7.9 (5.2–10.8)	15.01 (8.00–20.00)	1.36 (0.80–2.90)	0.60 (0.20–1.80)	-0.38 (-1.40–1.30)
	Clearwater	3.73 (1.50–5.79)	26.82 (21.44–30.12)	8.2 (4.9–11.9)	6.87 (4.00–9.00)	4.32 (2.27–9.10)	2.74 (1.13–6.77)	0.38 (-0.55–0.85)
	Council Bluff	3.43 (1.38–5.39)	25.73 (21.21–30.94)	7.5 (4.9–10.0)	10.02 (4.50–20.00)	1.47 (0.88–2.15)	0.68 (0.33–1.20)	-0.31 (-1.12–0.57)
	Deer Ridge	1.57 (1.10–2.30)	26.43 (22.67–31.20)	7.4 (4.9–10.6)	3.28 (2.25–5.63)	7.74 (2.57–12.73)	3.51 (074–8.67)	-0.40 (-1.51–0.80)
	Fellows	3.99 (2.50–5.51)	26.04 (22.79–29.58)	8.1 (5.1–10.8)	9.44 (5.00–14.00)	2.37 (1.63–3.27)	0.91 (0.33–1.60)	-0.60 (-1.81–0.00)
	Forest	2.91 (1.97–4.16)	25.50 (21.63–29.72)	7.7 (5.1–9.7)	5.96 (4.50–7.33)	6.89 (2.90–14.40)	5.08 (1.80–11.63)	0.94 (0.15–1.98)
	Higginsville	2.34 (1.08–3.75)	26.40 (20.52–30.67)	6.9 (2.2–11.4)	3.83 (2.50–6.00)	15.44 (7.83–50.60)	10.08 (4.63–41.67)	0.54 (-0.56–1.77)
	Hunnewell	2.24 (1.54–3.44)	25.88 (22.43–29.08)	8.0 (4.9–11.1)	3.95 (3.00–4.75)	6.77 (3.34–11.38)	3.07 (0.93–5.25)	-0.21 (-1.09–0.83)
	Kraut Run	1.76 (0.50–2.44)	27.56 (24.18–30.54)	7.9 (4.4–11.3)	3.13 (2.50–4.00)	16.85 (10.80–28.57)	6.57 (3.43–19.63)	-0.53 (-1.23–0.41)
	Lincoln	2.05 (1.24–3.44)	26.29 (22.41–29.03)	7.6 (4.7–10.4)	5.63 (2.17–10.33)	3.28 (1.58–17.83)	1.70 (0.50–13.17)	-0.23 (-1.05–1.02)
	Long Branch	3.77 (0.77–6.37)	24.98 (19.96–29.79)	7.1 (5.0–10.8)	6.59 (4.83–9.25)	10.58 (4.05–32.62)	7.42 (2.60–22.60)	0.76 (-0.36–1.73)
	Mark Twain	4.09 (1.35–6.45)	25.45 (22.0–28.77)	7.6 (4.8–11.5)	8.89 (4.00–14.50)	6.99 (3.10–15.80)	4.22 (1.00–13.20)	0.20 (-0.93–1.60)
	McDaniel	2.54 (1.13–3.50)	26.42 (22.72–29.26)	8.0 (4.6–12.9)	5.45 (3.00–28.00)	4.96 (2.58–7.50)	1.81 (0.45–3.63)	-0.67 (-1.75–-0.09)
	North	1.90 (1.27–2.87)	26.42 (22.49–29.04)	7.4 (4.1–11.0)	3.27 (2.50–4.50)	13.01 (4.87–21.73)	5.84 (2.37–14.55)	-0.33 (-1.46–0.73)
	Pomme de Terre	5.00 (2.77–7.70)	26.79 (23.54–29.75)	8.0 (4.9–11.9)	8.15 (5.33–12.00)	4.50 (2.70–8.13)	1.48 (0.59–3.82)	-0.86 (-1.57–0.02)
	Shayne	3.41 (2.20–4.77)	26.38 (23.58–29.25)	7.7 (5.0–10.2)	13.00 (9.50–16.00)	1.92 (0.98–3.13)	1.18 (0.35–2.30)	0.33 (-1.57–1.15)
	Stockton	6.98 (4.43–9.39)	26.21 (22.66–29.50)	8.4 (5.6–10.8)	11.18 (7.25–15.75)	4.43 (1.55–11.89)	2.13 (0.53–6.61)	-0.35 (-1.26–0.81)
	Sugar Creek	2.38 (1.14–3.95)	26.11 (22.59–31.58)	8.6 (7.9–10.4)	5.14 (3.70–7.00)	9.84 (7.23–13.80)	5.57 (3.90–7.62)	0.25 (-0.22–0.95)
	Table Rock	5.52 (2.50–7.05)	26.64 (20.78–30.59)	8.9 (6.4–11.4)	9.30 (6.75–12.00)	2.31 (0.96–4.70)	0.94 (0.23–2.93)	-0.51 (-2.46–1.01)
	Viking	6.13 (1.85–27.35)	24.77 (8.05–29.00)	7.2 (5.0–9.2)	6.85 (4.50–10.00)	5.41 (3.03–10.20)	3.50 (1.90–7.60)	0.56 (-0.30–1.10)
	Wappapello	2.26 (0.75–4.83)	27.52 (24.00–31.20)	8.3 (4.4–12.8)	4.60 (2.50–8.50)	9.80 (3.77–15.80)	4.70 (1.73–8.70)	-0.09 (-0.75–0.67)
	Watkins Mill	2.59 (1.03–3.99)	26.09 (20.78–29.47)	7.5 (5.5–9.8)	4.35 (3.00–8.00)	8.02 (4.40–12.28)	4.47 (1.50–8.63)	0.17 (-0.61–0.89)

We looked at 4 water clarity parameters in 6 zebra mussel-invaded reservoirs, and 2 water clarity parameters in Bull Shoals, also invaded by zebra mussels. Our assessment is especially robust in Lake of the Ozarks, Lotawana, and Smithville, where we have at least 11 years of data pre- and post- zebra mussel invasion date. In 5 out of 7 reservoirs, water clarity did not increase after zebra mussels invaded. Water clarity increased for a single parameter in Blue Springs, and for half the parameters in Lotawana. Our assessment of 26 reference reservoirs shows that water clarity is not changing ubiquitously across the state and suggests that the lack of zebra mussel impact was not a result of the mussels having a moderating effect on trends caused by external factors. Taken collectively, our analyses amass a weight of evidence showing that zebra mussel establishment did not increase water clarity in Missouri reservoirs, and that any potential impact the mussels might have does not stand out against the other sources of variation in water clarity (Knowlton and Jones 2006). These findings are supported by 2019 sampling, which revealed that zebra mussel density and biomass were roughly 10 times lower in Missouri reservoirs compared to northern, natural lakes where zebra mussels have previously been associated with an increase in water clarity (Ozersky et al. 2011; North et al. 2013; Knight et al. 2018; Rudstam and Gandino 2020). One explanation for the low densities could be that one or more habitat suitability parameters are near the tolerable threshold(s) for zebra mussels. It is possible that conditions in Missouri are suboptimal, allowing zebra mussels to survive while preventing them from forming the high densities that cause dramatic increases to water clarity. Invaded reservoirs tend to have cooler epilimnetic temperatures, lower PIM concentrations, and deeper anoxia depths than reference reservoirs. While zebra mussels can survive in Missouri reservoirs, these conditions could prevent the mussels from reaching densities great enough to impact water clarity. Noteworthy is that the exotic cladoceran Daphnia luholtzi dispersed among Missouri reservoirs (Havel et al. 2002) to more impoundments than zebra mussels have currently invaded. This comparison suggests that dispersal mechanisms are in place but that habitat and predation limit zebra mussel colonization.

Zebra Mussel Populations

Zebra mussel biomass and density is lower in Missouri than in northern, natural lakes. The highest mean biomass we observed here was 10 g m^− 2, which is two orders of magnitude lower than maximum densities in northern lakes (1,270 g m^− 2, Custer and Custer 1997). The highest mean density of zebra mussels we found (65 individuals m^− 2, in Lake of the Ozarks), is over an order of magnitude lower than densities shown to cause increased water clarity (Reed-Andersen et al. 2000). In Lake Simcoe (ON, Canada) for example, zebra mussel densities reached as high as 30,000 individuals m^− 2 but were more commonly found at densities of 3,300 individuals m^− 2 (Ozersky et al. 2011; North et al. 2013). It is unsurprising that we did not detect increased water clarity as estimates suggest that a minimum of 1,000 mussels m^− 2 must be present before a noticeable decline in chl-a concentrations are observed (Reed-Andersen et al. 2000).

Possibly the low zebra mussel density and biomass measurements in 2019 reflected a low year for zebra mussel populations. Zebra mussel populations display high levels of temporal variability and, in extreme cases, can change by more than 10-fold from year to year (Strayer et al. 2019). Even a 10-fold increase to 2019 measurements would be considerably below zebra mussel density and biomass in northern lakes. This could explain why we did not observe a universal increase in water clarity post zebra mussel invasion. Zebra mussels may not be able to form large populations in Missouri because they are living in tolerable but stressful conditions. This is consistent with experimental studies that show zebra mussel mortality is higher when 2 environmental parameters are stressful but sublethal compared to just one (Mathai et al. 2020).

Water Temperature and Water Levels

Warm water temperatures could limit zebra mussels from reaching high densities in Missouri reservoirs. Between May and August, surface water temperatures in most Missouri reservoirs are greater than 25°C, while mean surface water temperatures in a quarter of reservoirs is ≥ 30°C (Jones et al. 2011). Zebra mussel filtration rates begin to decline above 22°C; above 25°C, zebra mussel growth rates decline (Thorp et al. 1998), respiration rates increase (Alexander et al. 1994), and byssus thread production and foot activity is reduced (Rajagopal et al. 1997). Between 31 and 32°C, 100% zebra mussel mortality can occur in as little as 2 days or as long as 47 days, depending on the tolerance of individual mussels (McMahon et al. 1994; Elderkin and Klerks 2005). Based on water temperatures and zebra mussel thermal tolerance, it is possible that zebra mussels that settle in the littoral zone of Missouri reservoirs are eliminated during summer, while surviving mussels are physiologically stressed, reducing the likelihood for dense populations to form.

Fluctuating water levels may also reduce zebra mussel densities by exposing mussels at deeper depths to suboptimal water temperatures. In Bull Shoals and Truman Reservoir for example, water levels can fluctuate widely from year to year because these reservoirs are managed by the Army Corps of Engineers to prevent downstream flooding. August water levels since 2016 have fluctuated by 8.5 m in Bull Shoals⁴, while water level fluctuation in Truman Reservoir⁵ have been as high as 5.2 m. Such wide fluctuations could expose zebra mussels that were in deeper, cooler waters the previous year to water temperatures that are suboptimal or lethal. Water level drops could also kill zebra mussels by exposing them to air, and in fact, we observed empty zebra mussel shells above the water line that were attached to permanent boat launch structures in Bull Shoals.

Suspended Inorganic Particles

Suspended particles can negatively impact zebra mussel fitness by clogging inhalant siphons and gills, and reducing rates of respiration, clearance, and ingestion (Madon et al. 1998; Tuttle-Raycraft et al. 2017). Reduced zebra mussel clearance and ingestion rates have been observed in as little as 1 mg L^− 1 PIM, especially when food quality is low (Madon et al. 1998). Others have not observed negative effects of suspended particles until PIM concentrations are 27 mg L^− 1, above which zebra mussels increase their pseudofeces production (Lei et al. 1996), and 31 mg L^− 1, which is the concentration above which respiration rates begin to decline (Alexander et al. 1994). The natural log of the ratio of inorganic to organic particles may also influence zebra mussel physiology, with 1.61 as a cutoff under which zebra mussel ingestion, assimilation, water processing, and clearance rates are reduced (Madon et al. 1998; Schneider et al. 1998). PIM concentrations in Missouri are generally higher than lakes at more northern latitudes as median summer concentrations are 3.1 mg L^− 1 (Jones et al. 2008). Median Missouri PIM concentrations are above the lowest threshold for negative zebra mussel effects (1 mg L^− 1, Madon et al. 1998), but below others (31 mg L^− 1, Alexander et al. 1994; 27 mg L^− 1, Lei et al. 1996). The natural log of PIM to POM was below the 1.6 threshold (Madon et al. 1998; Schneider et al. 1998 in 98% of the sampling events, suggesting that zebra mussels in Missouri reservoirs may be frequently experiencing physiological stress due to suboptimal PIM to POM ratios. Storm events, and the associated increase in PIM from runoff (Hou et al. 2017) and resuspension (Jin and Ji 2001), could increase PIM concentrations and create intermittent periods of increased stress. Increased PIM from runoff would be especially pronounced up-reservoir closer to the inflow, and less evident near the dam where our samples were collected (Kimmel and Groeger 1984). The PIM to POM ratio and PIM concentrations by themselves could be a contributing factor to a suboptimal environment for zebra mussels.

There is evidence that zebra mussels can adapt to chronic levels of high PIM (Ouellette-Plante et al. 2017). Mussels previously exposed to high suspended PIM concentrations are less likely to be affected by future exposures (Summers et al. 1996), possibly because of behavioral techniques such as increasing their palp to gill surface area to facilitate food and non-food particle sorting (Payne et al. 1995), closing their inhalant siphons during periods of high suspended particles (Wiśniewski 1990), and increasing pseudofeces production (Chapman et al. 2017). These adaptations could enable zebra mussels to survive in Missouri reservoirs, despite sub-optimal concentrations of suspended PIM, but they do incur an energetic cost that might reduce overall fitness (Goldsmith et al. 2021). The PCA we used to assess habitat suitability showed that 2 of 7 invaded reservoirs were characterized by low PIM, and only one invaded reservoir was characterized by elevated PIM concentrations (Fig. 8). This suggests that PIM could be an important influence on zebra mussel densities in Missouri, and the chronic levels of relatively high suspended PIM could be one of multiple factors that ultimately prevents zebra mussels from reaching high densities.

Dissolved Oxygen Concentrations

Hypolimnetic oxygen concentrations are probably suboptimal for zebra mussels in Missouri reservoirs throughout most of the summer. Zebra mussels are one of the least tolerant bivalves to low oxygen levels. Their dissolved oxygen survival thresholds depend on water temperature because of increased metabolic demands associated with warmer temperatures (McMahon 1996; Alexander and McMahon 2004). Under anoxic conditions, 100% zebra mussel mortality can occur in as little as 4 days when temperatures are 25°C, but as many as 42 days when temperatures are 5°C (Mathews and McMahon 1999). By August, temperatures in Missouri hypolimnions often exceed 20°C (Jones et al. 2020) and approximately 80% of reservoirs throughout the state experience hypolimnetic anoxia throughout 65% of the water column (Jones et al. 2011; Petty et al. 2020). Most remain stratified between May and September when the mean summer thermocline depth is ≤ 5.2 m in 90% of reservoirs (Jones et al. 2011). Zebra mussels typically colonize substrate within the upper 9 m when oxygen levels are favorable (Burlakova et al. 2006; Hetherington et al. 2019), but in Missouri, where the thermocline is often near 5 m and hypolimnetic oxygen levels are reduced throughout much of the summer, they may be limited to shallower depths. Drops in water levels could leave the mussels dry or expose them to warmer surface waters. Rising water levels may also further reduce habitat if the mixing depth forms at a shallower depth than where zebra mussels have settled in the littoral zone. It is possible that during some years, when water levels are high and the depth of the hypolimnion is greater, dissolved oxygen plays an important role in zebra mussel survival. During years when water levels are low, oxygen may not be as important because zebra mussels are in the oxygenated epilimnion. Deeper anoxia depths, which were characteristic of 4 invaded reservoirs, may be an indication of the importance of hypolimnetic dissolved oxygen on zebra mussels. Deeper anoxia depths could expand zebra mussel habitat availability by increasing the depth at which water remains oxygenated throughout the summer.

Hydrology

Hydrology is likely an important factor in our ability to observe zebra mussel impacts on water clarity. A defining feature of many midcontinent reservoirs is their high flushing rates, which range from 0.2 to 19.2 times per year (Table 1), causing extreme variability in many parameters. For example, minimum seasonal means in Lake of the Ozarks for POM and Secchi averaged half of the historic mean, while seasonal maximums were twice as high (Jones et al. 2022a). Variability in Lake of the Ozarks is also high for PIM and chl-a, especially for individual samples, where it can be 105 and 30 times the minimum, respectively (Jones et al. 2022a). It may be difficult to observe zebra mussel effects on water clarity when variability from hydrology is so high, especially because the zebra mussel densities we observed were so low.

In addition to diminishing zebra mussel impacts on water clarity, hydrology and the high variability associated with it, may also act as another stressor for zebra mussels living in reservoirs. High reservoir flushing rates mean that inflow is an important driver of water quality. During summer, when many Missouri reservoirs are stratified, rain events can lead to plunging inflows that deliver inorganic particles to the benthos (Jones et al. 2022b). This could be an additional stressor for zebra mussels and may explain why in some reservoirs, zebra mussel densities are higher as distance from the inflow increases (Locklin et al. 2020).

Zebra Mussels Along the Southern Extent of Their Expansion

Our habitat suitability analysis suggests invaded reservoirs are more often characterized by cooler epilimnetic water temperatures, lower PIM concentrations, and deeper oxygenated epilimnions. We found that many reference and invaded reservoirs shared these conditions, but this overlap is unsurprising. The absence of an invader does not necessarily mean that conditions prevent that invader from becoming established, especially along the expanding edge of its range because it is possible the invader has not yet arrived. In Missouri, Lake of the Ozarks was the first reservoir where zebra mussels were observed and is also where recreational activity is highest (Reitz 2015). It could be that the reference reservoirs in this area might be more susceptible to zebra mussel establishment if they are invaded, while reference reservoirs with warmer water temperatures and/or higher PIM might be less likely to support dense zebra mussel populations if they are invaded.

Texas represents the southern extent of the zebra mussel invasion in North America and Europe (Locklin et al. 2020). There, zebra mussels have been observed in 39 lakes and reservoirs and are expected to continue their expansion (Robertson et al. 2020). Zebra mussel densities are higher down-reservoir compared to up-reservoir, which could reflect the longitudinal flow of these systems (Locklin et al. 2020). The importance of longitudinal flow on zebra mussel impacts to lentic processes in reservoirs differs from natural lakes, where zebra mussels cause nutrients to move from the pelagic offshore to the littoral nearshore (Hecky et al. 2004). Mortality in Texas reservoirs is high due to frequent summer temperatures above 30°C (Churchill et al. 2017; Arterburn 2020), and, due to extreme climate events and high mortality rates associated with water temperatures near their thermal tolerance, zebra mussels along the southern edge of their range may experience large population swings (Churchill et al. 2017). If these “boom and bust” cycles are common in zebra mussel population densities at southern latitudes, it would explain why we did not observe a consistent increase in water clarity post invasion. It is possible that our 2019 sampling may have occurred on a down, or “bust” year. Even if this was the case, zebra mussel densities were over an order of magnitude less in the 4 reservoirs we sampled than in any other invaded lake previously reported, suggesting that not all areas are equally impacted by this invader. In Belton Lake, Texas for example, zebra mussel densities from the spring cohort were as high as 21,160 mussels m^− 2 at a single site (Locklin et al. 2020), much greater than the maximum zebra mussel density of 1,682 mussels m^− 2 that we measured at a site in Lake of the Ozarks, Missouri. The reason why densities vary so dramatically in reservoirs at or near the southern zebra mussel expansion is still unclear, but our study suggests that differences in PIM concentrations and epilimnetic temperatures could play a role.

Predation from fish may also be impacting zebra mussel populations along the southern extent of their expansion more than in established populations to the north. Many fish species found in Missouri are known consumers of zebra mussels (Pflieger et al. 1997; Bartsch et al. 2005) which can make up over half of blue catfish and pumpkinseed sunfish diets (Magoulick and Lewis 2002; Colborne et al. 2015). Zebra mussels are also a food item for common carp (Tucker et al. 1996). While predation occurs in lakes with high zebra mussel densities, in Missouri predation may have a greater impact on zebra mussel populations because zebra mussel densities are low. Fish prefer zebra mussels over 5 mm (Magoulick and Lewis 2002), which may further impact mussel populations as zebra mussels do not reproduce until they have reached 5–7 mm (Delmott and Edds 2014).

Using 40 + years of historical water clarity data, we looked for regime shifts in water clarity corresponding with zebra mussel establishment. Considering the 4 water clarity parameters and 7 Missouri reservoirs with zebra mussel populations, we amounted a weight of evidence showing that zebra mussel establishment is not consistently associated with an increase in water clarity. In Missouri reservoirs, epilimnetic temperatures, TSS, PIM, and hypolimnetic dissolved oxygen concentrations are near, or occasionally exceed, zebra mussel survival thresholds, and hydrology causes highly variable conditions. It is possible that living in sublethal conditions prevents zebra mussels from establishing consistently high densities in Missouri reservoirs because dispersal mechanisms are present (Havel et al. 2002). This theory was supported by 2019 sampling of 4 invaded reservoirs. We observed zebra mussel densities no greater than 1,682 mussels m^− 2 at a single site, over 10 times less than densities in more northern lakes (Ozersky et al. 2011; North et al. 2013; Knight et al. 2018; Rudstam and Gandino 2020). Our habitat suitability assessment of Missouri reservoirs shows that invaded reservoirs were characterized by lower PIM concentrations and cooler epilimnetic temperatures, which could provide some indication about the parameters limiting zebra mussel densities. Additional research is needed to confirm whether this is the case, but our findings provide important insights into the effects of invasive zebra mussels on water quality that will help us better understand their impacts as they continue to expand throughout North America.

Acknowledgements

Thank you to Chris Brunet, Kyra Florea, Matt Sauer, and Clare Vanderwerken for all your incredible help in the field. Thank you also to the hundreds of volunteers and technicians over the past 40+ years who collected, processed, and analyzed all the data that made this study possible. We would especially like to thank Carol Pollard, whose assistance and suggestions were especially helpful for this study, and Dan Obrecht, whose advice, insight, and knowledge were instrumental to the design of this study, as well as the interpretation of its findings. Your support has been greatly appreciated. Thank you also to the Missouri Department of Natural Resources for supporting these long-term monitoring programs throughout their existence

Alexander JE, Thorp JH, Fell RD (1994) Turbidity and temperature effect on oxygen consumption in the zebra mussel (Dreissena polymorpha). Can J Fish Aquat Sci 51:179–184
Alexander JE, McMahon RF (2004) Respiratory response to temperature and hypoxia in the zebra mussel Dreissena polymorpha. Comp Biochem Phys A 137:425–434
APHA (2017) Standard Methods for the Examination of Water and Wastewater 23rd edn. Baird RB, Eaton AD, Rice EW (eds) Washington DC: American Public Health Association, Denver: American Water Works Association, Alexandria: Water Environment Federation
Arterburn H (2020) Zebra mussel populations in warmer waters. Dissertation, University of Texas at Arlington
Baranowska KA, North RL, Winter JG, Dillon PJ (2013) Long-term seasonal effects of dreissenid mussels on phytoplankton in Lake Simcoe, Ontario, Canada. Inland Waters 3:285–296
Bartsch MR, Bartsch LA, Gutreuter S (2005) Strong effects of predation by fishes on an invasive macroinvertebrate in a large floodplain river. J North Am Benthol Soc 24:168–177
Benson AJ, Raikow D, Larson J, Fusaro A, Bogdanoff AK, Elgin A (2021a) Dreissena polymorpha (Pallas, 1771). United States Geological Survey. http://nas.er.usgs.gov/taxgroup/mollusks/zebramussel/zebramusseldistribution.aspx. Accessed 5 April 2021
Benson AJ, Richardson MM, Maynard E, Larson J, Fusaro A, Bogdanoff AK, Neilson ME, Elgin A (2021b). Dreissena rostriformis bugensis (Andrusov, 1897): U.S. Geological Survey. https://nas.er.usgs.gov/queries/SpeciesAnimatedMap.aspx?speciesID=95. Accessed 4 June 2021
Borchers HW (2019) pracma: practical numerical math functions. https://CRAN.R-project.org/package=pracma. Accessed 5 April 2021
Budd JW, Drummer TD, Nalepa TF, Fahnenstiel GL (2001) Remote sensing of biotic effects: zebra mussels (Dreissena polymorpha) influence on water clarity in Saginaw Bay, Lake Huron. Limnol Oceanogr 46:213–223
Burlakova LE, Karatayev AY, Padilla DK (2006) Changes in the distribution and abundance of Dreissena polymorpha within lakes through time. Hydrobiologia 571:133–146
Carlton JT (2008) The zebra mussel Dreissena polymorpha found in North America in 1986 and 1987. J Great Lakes Res 34:770–773
Chapman PM, Hayward A, Faithful J (2017) Total suspended solids effects on freshwater lake biota other than fish. B Environ Contam Tox 99:423–427
Churchill CJ, Hoeinghaus DJ, La Point TW (2017) Environmental conditions increase growth rates and mortality of zebra mussels (Dreissena polymorpha) along the southern invasion front in North America. Biol Invasions 19:2355–2373
Cohen AN (2007) Potential distribution of zebra mussels (Dreissena polymorpha) and quagga mussels (Dreissena bugensis) in California: Phase 1 Report. San Francisco Estuary Institute, Oakland
Colborne SF, Clapp ADM, Longstaff FJ, Neff BD (2015) Foraging ecology of native pumpkinseed (Lepomis gibbosus) following the invasion of zebra mussels (Dreissena polymorpha). Can J Fish Aquat Sci 72:983–990
Custer CM, Custer TW (1997) Occurrence of zebra mussels in near-shore areas of western Lake Erie. J Great Lakes Res 23:180–115
Delmott SE, Edds R (2014) Zebra mussel maturation and seasonal gametogenesis in Marion Reservoir, Kansas, USA. BioInvasions Rec 3:247–260
Elderkin CL, Klerks PL (2005) Variation in thermal tolerance among three Mississippi River populations of the zebra mussel, Dreissena polymorpha. J Shellfish Res 24:221–226
Fahnenstiel GL, Lang GA, Nalepa TF, Johengen TH (1995) Effects of zebra mussel (Dreissena polymorpha) colonization on water quality parameters in Saginaw Bay, Lake Huron. J of Great Lakes Res 21:435–448
Falkowski PG, LaRoche J (1991) Acclimation to spectral irradiance in algae. J Phycol 27:8–14
Fanslow DL, Nalepa TF, Lang GA (1995) Filtration rates of the zebra mussel (Dreissena polymorpha) on natural seston from Saginaw Bay, Lake Huron. J Great Lakes Res 21:489–500
Goldsmith AM, Jaber F, Ahmari H, Randklev CR (2021) Clearing up cloudy waters: a review of sediment impacts to unionid freshwater mussels. Environ Rev 29:100–108
Havel JE, Shurin JB, Jones JR (2002) Estimating dispersal from patterns of spread: spatial and local control of lake invasions. Ecology 83:3306–3318
Hecky RE, Smith REH, Barton DR, Guildford SJ, Taylor WD, Charlton MN, Howell T (2004) The nearshore phosphorus shunt: a consequence of ecosystem engineering by dreissenids in the Laurentian Great Lakes. Can J Fish Aquat Sci 61:1285–1293
Hetherington AL, Rudstam LG, Schneider RL, Schneider RL, Holeck KT, Hotaling CW, Cooper JE, Jackson JR (2019) Invader invaded: population dynamics of zebra mussels (Dreissena polymorpha) and quagga mussels (Dreissena rostriformis bugensis) in polymictic Oneida Lake, NY, USA (1992–2013). Biol Invasions 21:1529–1544
Higgins SN, Vander Zanden MJ (2010) What a difference a species makes: a meta-analysis of dreissenid mussel impacts on freshwater ecosystems. Ecol Monogr 80:179–196
Hou X, Feng L, Duan H, Chen X, Sun D, Shi K (2017) Fifteen-year monitoring of the turbidity dynamics in large lakes and reservoirs in the middle and lower basin of the Yangtze River, China. Remote Sens Environ 190:107–121
Isles PDF (2020) The misuse of ratios in ecological stoichiometry. Ecology 101:e03153
Jin KR, Ji ZG (2001) Calibration and verification of a spectral wind-wave model for Lake Okeechobee. Ocean Eng 28:571–584
Jones JR, Obrecht DV, Perkins BD, Knowlton MF, Thorpe AP, Watanabe S, Bacon RR (2008) Nutrients, seston, and transparency of Missouri reservoirs and oxbow lakes: an analysis of regional limnology. Lake Reservoir Manag 24:155–180
Jones JR, Knowlton MF (1993) Limnology of Missouri reservoirs: an analysis of regional patterns. Lake Reservoir Manage 8:17–30
Jones JR, Knowlton MF, Obrecht DV, Graham JL (2011) Temperature and oxygen in Missouri reservoirs. Lake Reservoir Manag 27:173–182
Jones JR, Thorpe AP, Obrecht DV (2020) Limnological characteristics of Missouri reservoirs: synthesis of a long-term assessment. Lake Reservoir Manag 36:412–422
Jones JR, Thorpe AP, Obrecht DV (2022a) Limnological characteristics of Lake of the Ozarks (Missouri, USA): long-term assessment following formation of a reservoir series. Lake Reservoir Manag 38:288–303
Jones JR, Pollard C, Obrecht DV (2022b) Factors influencing phosphorus in midcontinent impoundments (USA) and challenges for detecting abatement. Inland Waters 12:154–162
Jost JA, Soltis EN, Moyer MR, Keshwani SS (2015) Linking zebra mussel growth and survival with two cellular stress indicators during chronic temperature stress. Invertebr Biol 134:189–202
Karatayev AY, Burlakova LE, Mastitsky SE, Padilla DK, Mills EL (2011) Contrasting rates of spread of two congeners, Dreissena polymorpha and Dreissena rostriformis bugensis, at different spatial scales. J Shellfish Res 30:1–9
Kassambara A, Mundt F (2020). factoextra: extract and visualize the results of multivariate data analyses. https://CRAN.R-project.org/package=factoextra
Kim TY, North RL, Guildford SJ, Dillon P, Smith REH (2015) Phytoplankton productivity and size composition in Lake Simcoe: the nearshore shunt and the importance of autumnal production. J Great Lakes Res 41:1075–1086
Kimmel BL, Groeger AW (1984) Factors controlling primary production in lakes and reservoirs: a perspective. Lake Reserv Manag 1:277–281
Kirsch KM, Dzialowski AR (2012) Effects of invasive zebra mussels on phytoplankton, turbidity, and dissolved nutrients in reservoirs. Hydrobiologia 686:169–179
Knight JC, O’Malley BP, Stockwell JD (2018) Lake Champlain offshore benthic invertebrate community before and after zebra mussel invasion. J Great Lakes Res 44:283–288
Knowlton MF, Jones JR (2006) Temporal variation and assessment of trophic state indicators in Missouri reservoirs: implications for lake monitoring and management. Lake Reservoir Manage 22:261–271
Lei J, Payne BS, Wang SY (1996) Filtration dynamics of the zebra mussel, Dreissena polymorpha. Can J Fish Aquat Sci 53:29–37
Locklin JL, Corbitt DN, McMahon RF (2020) Settlement, density, survival and shell growth of zebra mussels, Dreissena polymorpha, in a recently invaded low latitude, warm water Texas reservoir. Aquat Invasions 15:408–434
MacIsaac HJ, Rocha R (1995) Effects of suspended clay on zebra mussel (Dreissena polymorpha) faeces and pseudofaeces production. Arch Hydrobiol 135:53–64
Madon SP, Schneider DW, Stoeckel JA, Sparks RE (1998) Effects of inorganic sediment and food concentrations on energetic processes of the zebra mussel, Dreissena polymorpha: implications for growth in turbid rivers. Can J Fish Aquat Sci 55:401–413
Magoulick DD, Lewis LC (2002) Predation on exotic zebra mussels by native fishes: effects on predator and prey. Freshw Biol 47:1908–1918
Mathai PP, Bertram JH, Padhi SK, Singh V, Tolo IE, Primus A, Mor SK, Phelps NBD, Sadowsky MJ (2020) Influence of environmental stressors on the microbiota of zebra mussels (Dreissena polymorpha). Microb Ecol 81:1042–1053
Mathews MA, McMahon RF (1999) Effects of temperature and temperature acclimation on survival of zebra mussels (Dreissena polymorpha) and Asian clams (Corbicula fluminea) under extreme hypoxia. J Molluscan Stud 65:317–325
McMahon RF (1996) The physiological ecology of the zebra mussel, Dreissena polymorpha, in North America and Europe. Amer Zool 36:339–363
McMahon RF, Matthews MA, Ussery TH, Chase R, Clarke M (1994) Further studies of heat tolerance in zebra mussels: effects of temperature acclimation and chronic exposure to lethal temperatures. In Proceedings: Fourth international zebra mussel conference '94, pp. 251–272. Wisconsin Sea Grant Institute, Madison
Missouri DNR (2019) MO 2019 Dams. Rolla, MO: Missouri Geological Survey, Dam and Reservoir Safety Program
Missouri DNR (1986) Missouri Water Atlas. Rolla, MO: Missouri Geological Survey, Division of Geology and Land Survey
North RL, Barton D, Crowe AS, Dillon PJ, Dolson RML, Evans DO, Ginn BK, Håkanson L, Hawryshyn J, Jarjanazi H, King JW, La Rosa JKL, Leόn L, Lewis CFM, Liddle GE, Lin ZH, Longstaffe FJ, Macdonald RA, Molot L, Ozersky T, Palmer ME, Quinlan R, Rennie MD, Robillard MM, Rodé D, Rühland KM, Schwalb A, Smol JP, Stainsby E, Trumpickas JJ, Winter JG, Young JD (2013) The state of Lake Simcoe (Ontario, Canada): the effects of multiple stressors on phosphorus and oxygen dynamics. Inland Waters 3:51–74
Ouellette-Plante J, Morden AL, Johnson LE, Martel AL Ricciardi A (2017) Acclimation by invasive mussels: spatiotemporal variation in phenotypic response to turbidity. Freshw Sci 36:325–337
Ozersky T, Barton DR, Evans DO (2011) Fourteen years of dreissenid presence in the rocky littoral zone of a large lake: effects on macroinvertebrate abundance and diversity. J N Am Benthol Soc 30:913–922
Payne BS, Miller AC, Hubertz ED, Lei J (1995) Adaptive variation in palp and gill size of the zebra mussel (Dreissena polymorpha) and Asian clam (Corbicula fluminea). Can J Fish Aquat Sci 52:1130–1134
Petty EL, Obrecht DV, North RL (2020) Filling in the flyover zone: high phosphorus in Midwestern (USA) reservoirs results in high phytoplankton biomass but not high primary productivity. Front Environ Sci 8:11
Pflieger WL, SullivanM, Taylor L (1997) The fishes of Missouri. Missouri Department of Conservation, Jefferson City, MO
Pohlert T (2020) trend: Non-Parametric Trend Tests and Change-Point Detection. https://CRAN.R-project.org/package=trend. Accessed 5 April 2021
R Core Team, 2019. R: a language and environment for statistical computing. R Foundation for Statistical Computing, Vienna, Austria. URL: https://www.R-project.org/
Rajagopal S, Van der Velde G, Jenner HA (1997) Response of zebra mussel, Dreissena polymorpha, to elevated temperatures in the Netherlands. In: FM D’Itri (ed) Zebra mussels and aquatic nuisance species. Ann Arbor Press, Chelsea. pp 257–273
Reed-Andersen T, Carpenter SR, Padilla DK, Lathrop RC (2000) Predicted impact of zebra mussel (Dreissena polymorpha) invasion on water clarity in Lake Mendota. Can J Fish Aquat Sci 57:1617–1626
Reitz RA (2015) Results of the 2013 Missouri statewide angler survey. Resource Science Division, Missouri Department of Conservation.
Reynolds SA, Aldridge DC (2021) Impacts of invasive quagga mussels (Dreissena rostriformis bugensis) on reservoir water quality, as revealed by progressive-change BACIPS analysis. Water Res 197:117105
Robertson JJ, Swannack TM, McGarrity M, Schwalb AN (2020) Zebra mussel invasion of Texas lakes: estimating dispersal potential via boats. Biol Invasions 22:3425–3455
Rodionov SN (2004) A sequential algorithm for testing climate regime shifts. Geophys Res Lett 31:L09204.
Rodionov S, Overland JE (2005) Application of a sequential regime shift detection method to the Bering Sea ecosystem. ICES J of Mar Sci 52:328–332
Rudstam LG, Gandino CJ (2020) Zebra or quagga mussel dominance depends on trade-offs between growth and defense–Field support from Onondaga Lake, NY. PLoS ONE 15:e0235387.
Sartory DP, Grobbelaar JU (1984) Extraction of chlorophyll a from freshwater phytoplankton for spectrophotometric analysis. Hydrobiologia 114:117–187
Schloesser DW, Kovalak WP (1991) Infestation of native unionids by Dreissena polymorpha in a power plant intake canal in Lake Erie. J Shellfish Res 10:355–359
Schneider DW, Madon SP, Stoeckel JA, Sparks RE (1998) Seston quality controls zebra mussel (Dreissena polymorpha) energetics in turbid rivers. Oecologia 117:331–341
Strayer DL, Adamovich BV, Adrian R, Aldridge DC, Balogh C, Burlakova LE, Fried-Petersen HB, -Tóth LG, Hetherington AL, Jones TS, Karatayev AY, Madill JB, Makarevich OA, Marsden JE, Martel AL, Minchin D, Nalepa TF, Noordhuis R, Robinson TJ, Rudstam LG, Schwalb AN, Smith DR, Steinman AD, Jeschke JM (2019) Long-term population dynamics of dreissenid mussels (Dreissena polymorpha and D. rostriformis): a cross-system analysis. Ecosphere 10:e02701
Summers RB, Thorp JH, Alexander JE Jr, Fell RD (1996) Respiratory adjustment of dreissenid mussels (Dreissena polymorpha and Dreissena bugensis) in response to chronic turbidity. Can J Fish Aquat Sci 53:1626–1631
Thorp JH, Alexander JE Jr, Bukaveckas BL, Cobbs GA, Bresko KL (1998) Responses of Ohio River and Lake Erie dreissenid molluscs to changes in temperature and turbidity Can J Fish Aquat Sci 55:220–229
Tucker JK, Cronin FA, Soergel DW, Theiling (1996) Predation on zebra mussels (Dreissena polymorpha) by common carp (Cyprinus carpio). J Freshw Ecol 11:363–372
Tuttle-Raycraft S, Morris TJ, Ackerman JD (2017) Suspended solid concentration reduces feeding in freshwater mussels. Sci Total Environ 598:1160–1168
Ward JM, Ricciardi A (2007) Impacts of Dreissena invasions on benthic macroinvertebrate communities: a meta-analysis. Divers Distrib 13:155–65
Winslow L, Read J, Woolway R, Brentrup J, Leach T, Zwart J, Albers S, Collinge D (2019) rLakeAnalyzer: lake physics tools. https://CRAN.R project.org/package=rLakeAnalyzer. Accessed 5 April 2021
Wiśniewski R (1990) Shoals of Dreissena polymorpha as bio-processor of seston. In: Gulati RD, Lammens EHRR, Meijer ML, van Donk E (eds) Biomanipulation Tool for Water Management. Developments in Hydrobolgy, vol. 61. Springer, Dordrecht https://doi.org/10.1007/978-94-017-0924-8_39. Accessed 5 April 2021
Wylie GD, Jones JR (1991) Assessment of the sensitivity of Missouri reservoirs to acidification. J Freshwater Ecol 6:431–437
Xia Z, Depew DC, Valipour R, MacIsaac HJ, Weidman RP (2022) Impacts of invasive mussels on a large lake: direct evidence from in situ control-volume experiments. Sci Total Environ 849:157924

https://waterdata.usgs.gov/mo/nwis/uv?
https://www.swl-wc.usace.army.mil/pages/data/tabular/htm/bulsdam.htm
http://apps.ameren.com/HydroElectric/Reports/Osage/HSTBagnellDaily.aspx
https://www.swl-wc.usace.army.mil/pages/data/tabular/htm/bulsdam.htm
https://waterdata.usgs.gov/mo/nwis/uv?site_no=06922440

Modest dispersal, low densities, and little change in water clarity characterize zebra mussel populations near the southern extent of their range

Status:

Version 1

Abstract

Figures

Introduction

Methods

Study Reservoirs

Water Clarity Parameters

Zebra Mussel Density and Biomass

Temperature and Dissolved Oxygen Profiles

Statistical Analyses

Results

Study Reservoirs

Regime Shifts in Invaded Reservoirs

Regime Shifts in Reference Reservoirs

Zebra Mussel Density and Biomass

Habitat Suitability

Discussion

Zebra Mussel Populations

Water Temperature and Water Levels

Suspended Inorganic Particles

Dissolved Oxygen Concentrations

Hydrology

Zebra Mussels Along the Southern Extent of Their Expansion

Conclusions

Declarations

References

Footnotes

Status:

Version 1