Annual bottom trawl surveys were initiated in the 1970s in Laurentian Great Lakes Superior, Huron, Michigan and Ontario and in 1990 in Erie to provide annual assessments of the status and trends of prey fish communities. Native Cisco Coregonus artedi and Bloater C. hoyi dominated the prey fish community of Lake Superior. Prey fish communities in lakes Huron and Michigan were dominated by nonnative Rainbow Smelt Osmerus mordax and Alewife Alosa pseudoharengus for much of 1978-2016, but Bloater was an important species during the 1980-1990s and more recently has become the dominant prey species in these lakes. Alewife dominated the prey fish community of Lake Ontario during all 1978-2016. While nonnatives dominated the prey fish community in Lake Erie, native Emerald Shiner Notropis atherinoides was an important species and occasionally the dominant prey fish after the establishment of Round Goby Neogobius melanostomus in the late 1990s. During the 1980s-1990s Bythotrephes cederstroemi, Dreissena polymorpha, and Dreissena bugensis caused profound changes in Laurentian Great Lakes ecosystems and likely contributed to declines in fish community biomass in lakes Michigan and Huron. The impacts of these invaders were more muted in lakes Erie and Ontario. Lake Superior stands out as the Laurentian Great Lakes success story: Lake Trout Salvelinus namaycush was restored, and native prey fishes dominate and support a viable fishery. Although the abundance of Bloater has increased recently in lakes Huron and Michigan, recovery of native prey fishes remains uncertain. The absence of native species among the principal prey fish in Lake Ontario indicates a lack of progress in native fish recovery. Recovery of native prey fishes remains unclear in Lake Erie. The ever-changing state of the Laurentian Great Lakes caused by the impacts of invasive species and ongoing climate and ecosystem change will continue to challenge restoration of native fish communities in the 21st Century.
The ecosystems of Laurentian Great Lakes (LGL) and their fish communities underwent massive changes in the 19th and 20th Centuries as Europeans settled the basin, cut down forests, tilled the soil, built cities, polluted waters, harvested fish, and introduced numerous nonnative species (Smith, 1968, 1972a, 1995; Christie, 1974; Mills et al., 1993; Eshenroder and Burnham-Curtis, 1999). As a result, native fish communities began collapsing sequentially, beginning with lakes Ontario and Erie during the late 19th-early 20th Centuries (Christie, 1973; Kerr and LeTendre, 1991; Smith, 1995; Hartman, 1973), and lakes Michigan, Huron, and Superior during the mid-20th Century (Wells and McLain, 1973; Berst and Spangler, 1973; Lawrie and Rahrer, 1973). Smith (1968, 1972a) and Christie (1974) provide a historical review of the changes in the LGL fish communities in the 20th Century, wherein overfishing and depredation by nonnative Sea Lamprey (Petromyzon marinus) in the 1940s-1950s on the apex predator, Lake Trout (Salvelinus namaycush) led to the collapse of native fish communities. By the 1960s, nonnative Rainbow Smelt (Osmerus mordax) and Alewife (Alosa pseudoharengus) populations had expanded exponentially in the absence of predators and largely supplanted native prey fishes, primarily ciscoes (Coregonus spp.) (Smith, 1968, 1972a; Christie, 1974). These nonnative fishes have been implicated in suppressing native fishes, primarily through predation on larvae and competition for zooplankton food resources (Evans and Loftus, 1987; Lantry and Stewart, 1993; Madenjian et al., 2008; Gorman, 2012; O’Gorman and Stewart, 1999; O’Gorman et al., 2013). Smith (1972b) proposed that recovery of native fish communities in the late 20th Century and beyond would depend on intensive and coordinated management and research focused on controlling invasive species, restoring native species, and improving water quality across the basin.
By the 1970s, mounting ecological issues and concerns about LGL ecosystems prompted Smith (1973) to call for greater coordination of fishery research and management among state, provincial, and federal agencies. Under the guidance of the Great Lakes Fishery Commission, the first Joint Strategic Plan for Management of Great Lakes Fisheries (Strategic Plan) was adopted in 1981 and subsequently updated in 1988 and 1997 (GLFC, 2007). During this period, the Great Lakes Fishery Laboratory (now the Great Lakes Science Center - GLSC) initiated annual bottom trawl surveys to assess the status and trends of LGL prey fish communities in support of the Strategic Plan. In this paper, the following questions were addressed regarding the response of prey fish communities to changes in LGL ecosystems during 1978-2016: (1) how similar are prey fish communities among lakes? (2) How have prey fish communities changed in response to management actions, nonnative species, and continued ecosystem change? and (3) What do recent trends suggest about LGL fish communities in the 21st Century?
Bottom trawls were used to assess relative abundance of prey fishes in the LGL. Like all sampling methodologies, bottom trawls have inherent sampling biases; catchability varies among species and smaller-sized individuals are more vulnerable to capture (Stockwell et al., 2006, 2007; Yule et al., 2008). Nevertheless, bottom trawls serve as a synoptic instrument through which the composition and relative abundances of a broad array of species can be compared over time and among lakes. Annual bottom trawl surveys were initiated in 1973 in Lake Michigan, 1976 in Lake Huron, 1978 in lakes Superior and Ontario, and 1990 in Lake Erie. To facilitate cross-basin comparisons, results of surveys during 1978-2016 were examined for lakes Superior, Huron, Michigan, and Ontario and during 1990-2016 for Lake Erie. Although each survey was conducted using bottom trawls, they differed among lakes in proportion of the lake covered (Figure 1), seasonal timing, gear used, and the way trawls were towed (across or along bottom contours). The surveys largely targeted nearshore waters and sampled a range of depths: (1) 10-120 m in Lake Superior, (2) 9-110 m in lakes Huron and Michigan, (3) 8-150 m in Lake Ontario, and (4) 5-30 m in Lake Erie. Surveys were conducted in spring in lakes Superior and Ontario, and in fall in lakes Michigan, Huron, and Erie. More detailed methodologies are described in Stockwell et al. (2007) for Lake Superior, Argyle (1982) for Lake Huron, Madenjian et al. (2003, 2014) for Lake Michigan, Owens et al. (2003) and Weidel et al. (2017c) for Lake Ontario, and Tyson et al. (2006) for Lake Erie.
Comparison of trends in relative abundance of individual prey fish species among lakes where they occurred were made with population indices standardized to the maximum observed mean annual biomass (kg ha−1) or density (number ha−1) for each prey species. Essentially, this standardization rescaled each species’ abundance distributions from 0.0 to 1.0, which facilitated comparison of species trends and resolved potential biases associated with different sampling methodologies among lakes. Abundance distributions were examined for the following principal prey fish species in each lake: (1) age-1 and older Cisco (Coregonus artedi) in Lake Superior; (2) age-1 and older Bloater (Coregonus hoyi) in lakes Superior, Huron, and Michigan; (3) age-1 and older nonnative Rainbow Smelt in lakes Superior, Huron, Michigan, Erie, and Ontario; (4) age-1 and older nonnative Alewife in lakes Huron, Michigan, and Ontario; (5) age-0 and older nonnative Round Goby (Neogobius melanostomus) in lakes Huron, Michigan, Ontario, and Erie; and (6) age-1 and older Emerald Shiner (Notropis atherinoides) in Lake Erie. Population indices were based on biomass for lakes Superior, Huron, Michigan and Ontario and on relative density for Lake Erie. Comparison of trends in relative abundances of prey fish communities among lakes were based on the summed population indices of the principal prey species standardized to the maximum observed mean annual biomass or density. The summed abundances of these principal prey fish species represented >85% of the long-term total biomass or density of all prey fish species in each lake, and thus serve as an indicator for status and trends of the prey fish communities
Comparison of trends in year-class strength of principal prey species was based on standardized relative density of juvenile age classes that best predicted year-class strength in each lake. Different juvenile age classes were used to estimate year-class strengths because of differences in methodologies and timing of surveys among lakes. Year-class strengths were estimated from densities (number ha−1) of the following juvenile age classes in each lake: (1) age-1 Cisco, Bloater, and Rainbow Smelt in Lake Superior; (2) age-0 Bloater, Alewife and Rainbow Smelt in Lake Huron; (3) age-0 Bloater and Rainbow Smelt and age-3 Alewife in Lake Michigan; (4) age-0 Rainbow Smelt in Lake Erie; and (5) age-1 Alewife and Rainbow Smelt in Lake Ontario. For all species, juveniles were separated from older individuals based on age-correlated size classes or length cut-offs identified using gaps in length-frequency distributions (Vinson et al., 2014; Madenjian et al., 2014; Riley et al., 2019; Weidel et al., 2017b, FTG, 2018).
Kendall’s coefficient of concordance W was used following Siegel and Castellan (1988) and Zar (1999) to determine if trends in standardized mean annual relative abundances of two or more prey fish communities or prey fish species were statistically “concordant” among lakes. In our application, W reflects the degree of agreement in the rankings of mean annual abundances of communities or species. Values of W range from 0.0 (complete discordance or disagreement) to 1.0 (complete concordance or agreement). To guide interpretation of concordance (agreement) in rankings of prey fish communities or prey fish species, values of W < 0.3 were considered discordant (no agreement), 0.3-0.5 as weakly concordant (weak agreement), >0.5-0.7 as moderately concordant (moderate agreement), and >0.7 as strongly concordant (strong agreement). Calculation of W and degrees of freedom were corrected for ties in ranks; ties occurred in ≤ 10% of cases. To avoid excessive discrimination of negligible differences in relative abundances, precision was set to three decimal places. To test for significance, W was expressed as Chi-Square X 2 = k(N-1)/W with N-1 degrees of freedom where k = number of community/species rankings being compared and N = number of cases (years) included. Significance for X2 values was set at P = 0.05. When making comparisons of communities or species among lakes, data were restricted to years when all lakes were sampled. Thus, in comparisons among all lakes, 1992, 1993, 1998, and 2000 were omitted, and comparisons that included Lake Erie excluded years before 1990. Comparisons of trends in Round Goby among lakes were limited to years after 1993.
Laurentian Great Lakes prey fish communities
Among the LGL, Rainbow Smelt was a principal prey species in all lakes and Emerald Shiner was a principal prey species only in Lake Erie. Bloater were absent in surveys in lakes Erie and Ontario; Cisco were rarely encountered or absent in lakes Huron, Michigan, Erie, and Ontario; Alewife were in low abundance in lakes Superior and Erie; and Round Goby were absent in surveys in Lake Superior. Trends in relative biomass of principal prey fish species in lakes Superior, Huron, Michigan, and Ontario were weakly concordant during 1978-2016 (Figure 2; W = 0.41; P = 0.01). During the same period, strong concordance in trends was observed among lakes Superior, Huron and Michigan (W = 0.78; P < 0.001), whereas pairwise comparisons between Lake Ontario and other lakes were not significant (W = 0.33–0.42; P = 0.73–0.95). When Lake Erie was included and data were restricted to 1990–2016, no concordance was observed among lakes (W = 0.22; P = 0.31).
The composition of prey fish communities in each lake varied over time (Figure 2). In Lake Superior, Rainbow Smelt dominated early years (1978–1983), followed by coregonids Cisco and Bloater (1984–2016). In Lake Huron, Alewife and Rainbow Smelt dominated the community up until 2002. Following resurgence during the early 1980s, Bloater was an important component of the community throughout the 1990s. Prey fish abundance declined sharply after 2002, primarily due to a near disappearance of Alewife. Following a low point in 2008, prey fish abundance increased and was dominated by Bloater, the result of recruitment from a succession of new cohorts in 2005, 2007, and 2013-2015. By 2013, Alewife were rare in surveys and Rainbow Smelt abundance was low. Round Goby appeared in surveys in 1997 and remained a minor component of the community. In Lake Michigan, Alewife dominated the community in early (1978–1981) and later years (2002–2013), whereas Bloater dominated middle (1983–2001) and most recent years (2015–2016). After 2013, Alewife and Bloater abundance declined to very low levels. Abundance of Rainbow Smelt, a minor component of the community, peaked in the early 1980s, declined afterwards, reached a smaller peak in 2002, and declined to very low levels after 2007. Round Goby appeared in surveys in 2003 and became a major component of the community after 2007. The Lake Ontario prey fish community was consistently dominated by Alewife (1978–2016). Abundance of Rainbow Smelt, a minor component of the community, peaked in 1978, declined afterwards, reached smaller peaks in 1987 and 1998, and declined to low levels after 2000. Round Goby appeared in surveys in 2002 and was a major component of the community in 2006 and 2016. In Lake Erie, Rainbow Smelt dominated the community during the early years (1990–1997). Round Goby appeared in surveys in 1994 and rapidly increased in abundance and dominated middle years (2000–2001) and in 2016. Emerald Shiner abundance increased during the late 1990s following the decline in Rainbow Smelt and the increase in Round Goby, and was dominant in later years (2006, 2008, 2012, and 2014).
The relative proportion of native species in prey fish communities varied among lakes (Figure 2). In Lake Superior, Cisco and Bloater dominated after 1980 and averaged 79% of total prey fish abundance (1978-2016). In Lake Huron, Bloater dominated after 2008 and averaged 41% of total abundance (1978-2016). In Lake Michigan, Bloater dominated middle years (1983-2001) and after 2014 and averaged 57% of total abundance (1978-2016). There were no native species among the principal prey fish species in Lake Ontario. In Lake Erie, Emerald Shiner dominated some middle (2006, 2008) and later (2012, 2014) years and averaged 32% of total abundance (1990-2016).
Coregonids Cisco and Bloater
Trends in relative abundance of age-1 and older coregonids Cisco and Bloater were moderately concordant among lakes Superior, Huron, and Michigan (Figure 3; W = 0.67; P < 0.001). Similarly, trends in Bloater relative abundance were moderately concordant among the same lakes (W = 0.65; P < 0.001). Trends in Cisco and Bloater relative abundance in Lake Superior showed strong agreement (W = 0.78, P < 0.001). In lakes Superior, Huron, and Michigan, relative abundance of age-1 and older coregonids peaked during the late-1980s through the early-1990s. Low relative abundance was observed in lakes Superior and Michigan during 2007-2011. In contrast, relative abundance of Bloater rebounded in Lake Huron following a record low in 2008 to 75% of peak abundance in 2012, but since then has trended downward (Figure 3).
Trends in relative year-class strength of coregonids Cisco and Bloater were weakly concordant (W = 0.47; P = 0.001) among lakes Superior, Huron, and Michigan (Figure 3). In contrast, trends in relative year-class strength of Cisco and Bloater within Lake Superior were strongly concordant (W =0.83; P < 0.01) as were Bloater in lakes Superior and Michigan (W = 0.80; P = 0.02). No other comparisons showed significant concordance among lakes. Lakes Superior and Michigan shared a common pattern of stronger year-classes of coregonids during the 1980s and early-1990s and weaker year-classes thereafter. Lake Superior was unique in having moderate year-classes in 1998 and 2003, whereas Lake Huron was unique in having strong year-classes in 2005, 2007, and during 2013-2015. Trends in relative year-class strength of coregonids among lakes Superior, Huron and Michigan were moderately concordant prior to 2003 (W = 0.57; P < 0.001), before the appearance of strong year-classes in Lake Huron.
Trends in relative abundance of age-1 and older Rainbow Smelt were strongly concordant among lakes Superior, Huron, Michigan, and Ontario (Figure 4; W = 0.80; P < 0.001). In all four lakes, relative abundance began at record high levels followed by fluctuating but declining trends thereafter, reaching record lows after 2000 and only exceeded 20% of maximum abundance in 2005 in lakes Michigan and Ontario and 2010 in Lake Ontario. When Lake Erie was included and data were restricted to 1990-2016, trends in relative abundance were moderately concordant (W = 0.61; P < 0.001), but no pairwise comparisons with Lake Erie were significant (W = 0.59-0.68; P > 0.12).
Trends in relative year-class strength of Rainbow Smelt were weakly concordant (Figure 4; W = 0.38; P < 0.03) among lakes Superior, Huron, and Michigan, and Ontario. Omitting Lake Huron, the comparison was moderately concordant (W = 0.66; P < 0.001). Pairwise comparisons showed strong concordance between lakes Superior and Michigan (W = 0.79; P < 0.02) and lakes Superior and Ontario (W = 0.77; P < 0.03) but no other comparisons were concordant. When Lake Erie was included and data were restricted to 1990-2015 year-classes, concordance among lakes was weak (W = 0.31; P = 0.05). Pairwise comparisons between Lake Erie and other lakes showed no significant concordance.
In Lake Superior, relative year-class strength varied from moderate to strong during 1977–1998, weak during 1999–2004, and varied from weak to moderate during 2005–2015 (Figure 4). In Lake Huron, relative year-class strength varied from weak to moderate in 1977–1997, followed by an alternating series of moderate to strong and weak year-classes in 1998–2016. The strongest year-class on record occurred in 2005, whereas the weakest year-classes on record occurred in 2010 and 2016. In Lake Michigan, relative year-class strength declined 99% from the peak in 1980 to very weak in 2001. Moderate year-classes were produced in 2005 and 2008, followed by near-record lows during 2007–2014. In Lake Ontario, alternating strong and weak year-classes occurred during 1977–1998 followed by weak to moderate year-classes during 1999–2015. In Lake Erie, the strongest year class appeared in 1992 and subsequent strong year-classes in 2003, 2008, and 2014 were progressively weaker, declining to 40% of the peak value.
Trends in relative abundance of age-1 and older Alewife among lakes Huron, Michigan, and Ontario were not significantly concordant (Figure 5; W = 0.41; P > 0.16) but were strongly concordant between lakes Huron and Michigan (W = 0.82; P < 0.01). Pairwise comparisons between Lake Ontario and lakes Huron and Michigan were not significant (W < 0.5; P > 0.5). In Lake Huron, relative abundance was high during 1978-1981, declined to low levels during 1982-1986, rebounded with peaks in 1987 and 1994, and declined to near-zero levels in 2004-2016. In Lake Michigan, relative abundance was high during 1978-1981 and declined to lower levels until a moderate peak appeared in 2002, a result of recruitment of the large 1998 year-class. Afterwards, relative abundance continued to decline except for a minor peak in 2013, which was due to one large catch at the 9-m station off Saugatuck, Michigan (Madenjian et al. 2014), and was followed by near-zero levels in 2015 and 2016. Relative abundance of Alewife in Lake Ontario was highly variable with no apparent trend.
Trends in relative year-class strength of Alewife during 1977-2013 showed no concordance among lakes (W = 0.38; P = 0.56; Figure 5). In Lake Huron, year class strength was variable until 2003 when a record year-class appeared. Values fluctuated near zero thereafter and can be explained by low relative abundance of adults after 2004 (Figure 5). In Lake Michigan, year-class strength was variable, with peaks in 1998 and 2010, and three successive years of low values during 2011-2013. In Lake Ontario, year-class strength was low before 1998 and was followed by a series of peaks in 1998, 2005, 2009, 2012.
Trends in relative abundance of age-0 and older Round Goby was weakly concordant (W = 0.46; P = 0.01) among lakes where this species has become established (Huron, Michigan, Erie, and Ontario; Figure 6). However, comparisons among lakes were hindered by the desynchronized expansion of Round Goby populations during the 1994–2016. The first trawl records occurred in Lake Erie (1994), followed by Lake Huron (1997), Lake Ontario (2002), and Lake Michigan (2003). After appearing in Lake Erie, relative abundance fluctuated with an increasing trend, peaking in 2001, and declining thereafter. A similar pattern of expansion and decline was observed in Lake Huron, where relative abundance peaked in 2011 and 2012 and declined >90% thereafter. Likewise, relative abundance in Lake Michigan peaked in 2013 and was followed by >90% decline. In Lake Ontario, relative abundance was highly variable; increased rapidly after initial detection in 2002 to a peak in 2006, declined 98% in 2008, reached a record high in 2012, declined >70% during 2013-2014 and recovered during 2015-2016.
Trends in LGL prey fish communities were assessed by examining trends in the combined relative abundances of principal prey fish species. Trends among lakes Superior, Huron, and Michigan were concordant, but these trends were not concordant with those in lakes Ontario or Erie. When trends for individual species were compared among lakes, concordance was greatest for coregonids Cisco and Bloater and Rainbow Smelt and least for Alewife and Round Goby. Highest relative abundance of coregonids occurred during the late-1980s through the early-1990s, with relatively low abundance after 2000 (Lake Huron as the exception). Rainbow Smelt abundance declined slowly and erratically over the last quarter century and reached record low levels during the 2000s. In general, Alewife abundance in lakes Huron and Michigan was substantially higher during the 1980s and 1990s compared to the 2000s and 2010s, when abundance declined to record low levels. However, the peaks of the 1980s and 1990s were substantially lower (76-83%) than those in the mid-1960s (Madenjian et al., 2005). In contrast, Alewife abundance in Lake Ontario was highly variable with no evident trend. This result for Lake Ontario contrasts with previous population analyses, such as Madenjian et al. (2008), and is a result of applying a correction to trawl samples of prey fish data collected in Lake Ontario after 1996 (Weidel et al. 2017a, b).
General concordance in population trends of Bloater and Cisco among lakes Superior, Huron, and Michigan supports hypotheses for large-scale environmental factors influencing recruitment synchrony in LGL coregonids (Bunnell et al., 2010; Rook et al., 2012; Myers et al., 2015). However, trends in relative abundance of Bloater in Lake Huron after 2008 contrasted with those in lakes Superior and Michigan. After 2008, Bloater abundance in Lake Huron reached 75% of its maximum record in 2012, primarily due to recruitment of multiple strong and moderate year-classes produced during 2005-2011. Although a record year-class was produced in 2013 and was followed by two successive strong year-classes, these later year-classes did not translate as expected into higher levels of relative abundance of age-1 and older fish (Riley et al., 2019).
In general, trends in relative year-class strength of prey fish species were less concordant than trends in relative abundance of principal prey fish species across the basin, and only coregonids showed significant concordance among lakes Superior, Huron, and Michigan, the only lakes for which coregonid year-class data were available. The appearance of strong and moderate year-classes of Bloater in Lake Huron during 2005-2015 contradicted a general trend of weak year-classes in lakes Michigan and Superior. However, the moderate 2016 year-class of Bloater in Lake Michigan was the largest since 1990. Discordance in the recent appearance of strong year-classes of coregonids among lakes Superior, Michigan, and Huron suggests that environmental or ecological differences among lakes may now be contributing to recruitment asynchrony despite increasingly similar trophic status, primary production, and zooplankton communities in these lakes (Barbiero et al., 2012, 2018; Bunnell et al., 2014). There was no concordance in year-class strength for Alewife among lakes Huron, Michigan, and Ontario. However, methodological differences in estimating year-class strength among lakes may be contributing to the disagreement. The use of different ages of juvenile Alewife among lakes to assess year-class strength is problematic because abundances at age-0, age-1, and age-3 are affected by differential survivorship. For example, the timing of sampling of age-0 fish in the fall can greatly affect their vulnerability to bottom trawls, and by age-3, Alewife are fully recruited and vulnerable to bottom trawls (Madenjian et al., 2005, 2015). There was weak to moderate concordance in cross-basin trends in year-class strength for Rainbow Smelt and strong concordance among pairs of lakes, i.e. Superior-Michigan and Superior-Ontario.
Trends in relative abundance of Round Goby varied among lakes where this species has successfully invaded (Huron, Michigan, Erie, and Ontario). Recent trends in relative abundance remain variable in lakes Huron, Michigan and Ontario, which suggests future population equilibrium remains uncertain. The pattern of increase, peak, and decline of Round Goby in Lake Erie suggests the population may be approaching equilibrium. Although the recent trends in relative abundance among lakes are mixed, there is mounting evidence that Round Goby has become increasingly important in the diets of Lake Whitefish (Coregonus clupeaformis), Yellow Perch (Perca flavescens), and apex predators, such as Lake Trout, Walleye (Sander vitreus), Smallmouth Bass (Micropterus dolomieui), and Burbot (Lota lota) (He et al. 2015). Round Goby were absent from spring bottom trawl surveys in Lake Superior, but their presence in the harbors and embayments of Duluth and Thunder Bay (Bergstrom et al., 2008; Ontario Ministry of Natural Resources, unpublished data) suggests the potential for future population expansion.
The relative proportion of native species in prey fish communities varied among lakes. The decline of Rainbow Smelt in Lake Superior by the early-1980s and subsequent expansion and continuous dominance of native Cisco and Bloater shows recovery of the native prey fish community. In Lake Michigan, Bloater followed a parallel trend of expansion until the late-1990s, when continued recruitment failure resulted in a decline in the proportion of native species. The recovery of Bloater in Lake Huron during the 1980s and 1990s was not as pronounced as in Lake Michigan, but the appearance of strong year-classes during the 2000s, coupled with sharp declines in Alewife and Rainbow Smelt abundance, resulted in an increased proportion of native species, averaging 82% during 2010-2016, matching that in Lake Superior. As a result of recruitment during 2013-2016, proportion of Bloater in the prey fish communities of lakes Huron and Michigan currently range 72-85% (2015-2016). In Lake Erie, Emerald Shiner became more abundant following the expansion of Round Goby and reduced dominance of Rainbow Smelt during the late 1990s. Lake Ontario differs from other LGL in not having a native principal prey fish species and one nonnative species, Alewife, consistently dominated the prey fish community, averaging 93.5% of the abundance during 1978–2016.
Prey fish communities of lakes Superior, Huron and Michigan became much smaller in relative abundance (rarified) during the 2000s. Structure of the rarified prey fish community of Lake Superior remained relatively unchanged compared to the period of maximum abundance in the 1990s, suggesting that the foodweb structure has remained relatively stable (Ives et al., 2019). The structure of the rarified prey fish communities of lakes Huron and Michigan was variable in the 2000s. Following the expansion of Round Goby after 2007, the proportion of Rainbow Smelt in Lake Michigan declined sharply. Alewife became rare in both lakes, after 2005 in Huron and after 2013 in Michigan. Bloater became the dominant prey fish in Lake Huron after 2006 and in Lake Michigan after 2014. These changes suggest that rarified communities can have very different and variable compositions and may reflect changing ecological conditions and foodweb structures that became apparent in lakes Huron and Michigan after the expansion of dreissenids during the 2000s (Riley et al., 2008; Kao et al., 2016; Madenjian et al., 2015; Ives et al. 2019).
Smith’s (1972b) assessment of the future of LGL fish communities was conditioned on appropriate responses to needs for controlling invasive species, restoring native species, and improving water quality across the basin. Trends in prey fish communities over the past 39 years allow us to evaluate how prey fish communities have responded to changes in LGL ecosystems. So far, efforts to control invasive species in the LGL have produced mixed results. Great strides have been made to control nonnative Sea Lamprey, which allowed recovery of wild Lake Trout stocks in Lake Superior (Schreiner and Schram 1997; Hansen 1999; Muir et al. 2013) and greatly reduced their impacts on large-bodied fishes in other lakes (McLaughlin et al. 2003). Lake Superior stands out as a success story, where the reduction of nonnative Rainbow Smelt following the recovery of Lake Trout likely contributed to the recovery of the native fish community (Gorman and Hoff, 2009; Gorman, 2007, 2012). The native prey fish community remains intact, and while nonnative Rainbow Smelt remains as a major component of the fish community, it no longer dominates. In lakes Huron and Michigan, introduced Chinook and Coho salmon and stocked Lake Trout have greatly reduced Alewife populations and may have contributed indirectly to successful year-classes of Bloater in more recent years (Madenjian et al. 2008; Dettmers et al. 2012; O’Gorman et al. 2013), and as we have seen, Bloater dominated the prey fish community in Lake Huron during 2006–2016 and in Lake Michigan during 2015–2016. Although fish communities in the LGL remain dominated by nonnative species as a whole, continued predation pressure on nonnative prey fish species may provide a window for sustained recovery of Bloater and Cisco (Eshenroder and Burnham-Curtis, 1999; Zimmerman and Krueger, 2009; Ives et al., 2019). Although nonnative species continue to dominate the Lake Erie prey fish community in most years, the distinctive characteristics of Lake Erie challenges comparisons with other lakes. The Lake Ontario prey fish community stands out as being farthest from a natural state. The dominance of Alewife and rarity of native species suggest little progress in controlling nonnatives and restoring native fish species. Unlike lakes Michigan and Huron, predator populations in Lake Ontario have not reduced the abundance of Alewife and its dominance of the prey fish community (Madenjian et al., 2008; Stewart et al., 2010; O’Gorman et al., 2013).
Nonnative species continue to be introduced into the LGL and some have had profound impacts on ecosystems throughout the basin. The large cladoceran Bythotrephes cederstroemi was discovered in 1985 (Bur et al., 1986), followed by Zebra Mussel in 1988 (Hebert et al., 1989), Round Goby in 1990 (Jude et al., 1992), and Quagga Mussel in 1991 (May and Marsden, 1992). Collectively, these species have changed the character of LGL foodwebs and ecosystems (Hecky et al., 2004; Nalepa et al., 2005; Barbiero et al 2011, 2012, 2018). Bunnell et al. (2014) assessed changes in LGL foodwebs as a result of “bottom up” effects caused by Dreissenid Mussels and concluded that these species have led to “oligotrophication” of lakes Michigan and Huron, which has resulted in reduced energy and nutrient availability for prey fish communities. The combination of “bottom up” effects and large predator populations (“top down” effects) have resulted in reduced prey fish populations in lakes Huron and Michigan (Bunnell et al., 2014; Madenjian et al., 2015, 2018; Kao et al., 2016). So far, dreissenids have had little success in Lake Superior, likely because of colder temperatures and already low productivity (Bronte et al., 2003, Bunnell et al., 2014). Although dreissenids have successfully invaded lakes Erie and Ontario, bottom up impacts on the fish communities are muted compared to lakes Michigan and Huron (Bunnell et al. 2014; Ives et al. 2019).
Smith’s (1972b) hope for a return to stability in LGL ecosystems can only happen if managers prepare for a future of changing eco-scapes, which include ongoing climate change, proliferation of new invasive species, and ever-changing foodweb interactions. Unfortunately, the existing knowledgebase needed to address rapidly changing LGL ecosystems is insufficient to address current and emerging challenges (Sterner et al. 2017). For example, increased light and temperature regimes from climate change have already resulted in altered lower foodwebs in the LGL (O’Beirne et al., 2017; Reavie et al., 2017). Moreover, the numerous invasive species present in the LGL complicate the management and restoration of native fish communities envisioned by Smith (1972b). While the LGL may never be restored to their former natural state, they can be restored to a level of functionality, whereby their ecosystems maintain fish communities largely dominated by native species and provide essential services, such as healthy food and clean water for surrounding human populations (Zimmerman and Krueger, 2009; Ives et al. (2019). Committed efforts to restore native fish species, such as Lake Trout across the LGL since the 1950s, and recent efforts to restore Bloater in Lake Ontario (OMNRF, 2015; Klinard, 2019), can provide essential knowledge of LGL ecosystems in an era of changing eco-scapes. The science and management actions developed to restore native fish species can provide essential tools to facilitate achievement of ecosystem functionality (Zimmerman and Krueger, 2009; Dettmers et al., 2012; Ives et al., 2019).
L. Evrard, E. Roseman, S. Riley, D. Hondorp, C. Madenjian, D. Bunnell, and B. Weidel (USGS GLSC), provided data from lakes Superior, Huron, Michigan, and Ontario. J. Deller, E. Weimer and Z. Slagle (Ohio Dept. Natural Resources); J. Markham (New York State Dept. Environmental Conservation); C. Murray and M. Hosack (Pennsylvania Fish and Boat Commission), and A. Cook and P. Penton (Ontario Ministry of Natural Resources and Forestry) provided data from Lake Erie. Constructive reviews by B. Rook and an anonymous referee improved the manuscript. All GLSC sampling and handling of fish during research were carried out in accordance with guidelines for the care and use of fishes by the American Fisheries Society (http://fisheries.org/docs/wp/Guidelines-for-Use-of-Fishes.pdf).