STATE OF THE LAKE SUPERIOR ECOSYSTEM
IN 201720
Bryan G. Matthias21, Thomas R. Hrabik, Joel C. Hoffman, Owen T.
Gorman, Michael J. Seider, Michael E. Sierszen, Mark R. Vinson, Dan
L. Yule, and Peder M. Yurista
The Lake Superior ecosystem, near pristine in comparison to the other
Laurentian Great Lakes, has seen major biological changes during the past
two decades. Starting in the late 1990s, pelagic prey-fish biomass has been
declining in both nearshore and offshore waters (Pratt et al. 2016; Vinson et
al. 2016). Declines have been observed in native coregonines, including
Cisco, Bloater, and Kiyi along with Deepwater Sculpin and non-native
Rainbow Smelt (Gorman 2012; Pratt et al. 2016; Vinson et al. 2016). These
species comprise a substantial proportion of the diets of native predators like
the lean and siscowet forms of Lake Trout (hereafter, siscowet) and Burbot
and introduced migratory salmonines (15-80% of total diets; Matthias and
20Complete publication including maps of place names, abstract, other chapters, scientific
fish	names,	and	references	is	available	at
http://www.glfc.org/pubs/SpecialPubs/Sp21 02.pdf.
B.G. Matthias. U.S. Fish and Wildlife Service, 850 South Guild Avenue, Suite 105,
Lodi, CA 95240, USA.
T.R. Hrabik. University of Minnesota Duluth, 207 Swenson Science Building, 1035
Kirby Drive, Duluth, MN 55812, USA.
J.C. Hoffman, M.E. Sierszen, and P.M Yurista. U.S. Environmental Protection
Agency, Center for Computational Toxicology and Ecology, Great Lakes Toxicology and
Ecology Division, 6201 Congdon Boulevard, Duluth, MN 55804, USA.
O.T. Gorman, M.R. Vinson, and D.L. Yule. U.S. Geological Survey, Great Lakes
Science Center, Lake Superior Biological Station, 2800 Lake Shore Drive East, Ashland,
WI 54806, USA.
M.J. Seider. U.S. Fish and Wildlife Service, Ashland Fish and Wildlife Conservation
Office, 2800 Lakeshore Drive East, Ashland, WI 54806, USA.
^Corresponding author (e-mail: brvan.matthias@fws. gov-!.
55

-------
Yule 2020; also see Kitchell et al. 2000; Negus et al. 2007; Gamble et al.
201 la, 201 lb; Isaac et al. 2012). The declines in prey resources are troubling
given lean and siscowet Lake Trout populations have remained relatively
stable during this time.
We built an EcoPath with EcoSim (EwE) model to quantify how the Lake
Superior ecosystem changed from 2005 to 2016 and to predict how the
ecosystem might change if 2016 commercial and recreational harvest levels
on all targeted species are sustained until 2055. The EcoPath model was
parameterized to the first lakewide Cooperative Science and Monitoring
Initiative (CSMI), a binational intensive monitoring and assessment program
conducted in 2005-2006 (https://www.epa.gov/great-lakes-
monitoring/coopcrativc-sciencc-and-monitoring-initiativc-csmi). The model
includes nearshore and offshore food webs, including 59 groups of
producers and consumers and 3 detrital groups (Fig. 25; for model details
and input data see Matthias and Yule 2020). The model represents a
significant increase in the breadth of lower trophic levels when compared to
past Lake Superior models (i.e., Kitchell et al. 2000; Cox and Kitchell 2004)
and recent models of Lakes Michigan and Huron (Langseth et al. 2012;
Rogers et al. 2014; but see Kao et al. 2016). We incorporated multiple levels
in the microbial loops representing significant sources of biomass and
carbon cycling from detrital sources. This model includes greater detail in
the offshore fish communities and all zooplankton communities than in prior
ecosystem models.
Fig. 25. Configuration of the Lake Superior EcoPath model representing both
nearshore (generally left side) and offshore zones (generally right side), benthic-
pelagic coupling, and nearshore-offshore coupling. Node size is proportional to
total biomass, and line thickness represents biomass flow between groups.
56

-------
Nearshore Nearshore Offshore	Offshore
benthic	pelagic	pelagic	benthic
The EcoSim model was fit to 2005-2016 data from the CSMI (e.g., Yurista
et al. 2009; Isaac 2010; Yule et al. 2013; Pratt et al. 2016), U.S.
Environmental Protection Agency's Great Lakes National Program Office
(e.g., Barbiero et al. 2019), U.S. Geological Survey trawl survey (see Vinson
et al. 2016), coordinated siscowet surveys (see Status of Siscowet Lake
Trout in Lake Superior in 2017 chapter), Minnesota and Wisconsin DNR
gillnet and acoustics surveys (C. Goldsworthy, unpublished data; B. Ray,
unpublished data), acoustics and fish community surveys in Ontario (Fisch
et al. 2019a; E. Berglund, unpublished data), and statistical catch-at-age
models from the Michigan, Minnesota, and Wisconsin DNRs (see Modeling
Subcommittee, Technical Fisheries Committee 2018). This model
represented data from across Lake Superior (Fig. 26) and encompassed all
trophic levels. Agency surveys indicate declines in biomass across the prey-
fish community (Cisco, Bloater, Kiyi, and Deepwater Sculpin); relatively
stable populations of Lake Whitefish, Rainbow Smelt, and nearshore Slimy
57

-------
and Spoonhead Sculpins; lean and siscowet fonns of Lake Trout; and
Burbot. The EcoSim model for 2005-2016 estimated large declines (>20%
since 2005) for siscowet, Bloater, Kiyi, and sculpins. Unlike the survey
trends, Cisco biomass was estimated to remain stable along with that of Lake
White fish and lean Lake Trout (Fig. 27). Biomass of Burbot and Rainbow
Smelt was estimated to increase (see Fig. 27).
Fig. 26. Map showing political jurisdictions and spatial areas where EcoSim
model fitting procedures occurred.
92°0'0"W	90°0'0"W	88D0'0"W	86°0'0"W	84°0'0"W
WI-2
0
50
100
200
O
O
to
*3-
Kilometers
92°0'0"W
90°0'0"W
88°0'0"W
86°0'0"W
58

-------
Fig. 27. Relative biomass trends for the Lake Superior fish community (points
and thin lines) and trends averaged over all surveys (thick dashed black line),
2005 to 2016 for management units or surveys. Model predicted biomass from
EcoSim (thick solid black line) for major fish species or groups. Nearshore
sculpin includes Slimy and Spoonhead Sculpins (USGS = U.S. Geological
Survey; SCAA = statistical catch-at-age analysis; OMNRF = Ontario Ministry
of Natural Resources and Forestry).
ro
QJ
Q.
OC
V)
430 mm)
Smelt (> 130 mm)
4.0
1.0
0.5
0.0
2006 2(ni 2016
Cisco { > 300 mm)
A
Nearshore Sculpin
EwE predicted
Average trend
CSMI acoustic survey
USGS trawl
Siscowet survey
SCAA assessment
OMNRF index
Thunder Bay acoustic survey
MN summer acoustic survey
MN fall acoustic survey
MN small mesh
Wl-l spring survey
Wl-l summer survey
WI-2 spring survey
WI-2 summer survey
WI-2 fall survey
Deepwater Sculpin
Lean Lake Trout ( > 500 mm)
4.0
2.0
0.0
1.5
1.0
0.5
0.0
Siscowet ( > 500 mm)

2006 2011 2016
Year
2006 2011 2016
59

-------
Predicted long-term trends (i.e., 2005-2055) appear consistent with the
estimated 2005-2016 EwE trends, provided harvest remains at the 2016
level. Biomass is predicted to decline for most coregonines, sculpins, and
Burbot; increase for Rainbow Smelt; and remain stable for Cisco and both
forms of Lake Trout. The prediction that Cisco will remain stable is contrary
to data provided for this species in this reporting period (see Status of
Prey Fish in Lake Superior in 2017 chapter). Cisco appears to decline in
most surveys (Fig. 27), but there is high variability within these trends. In
addition, we have not yet been able to account for the high variability
observed in Cisco recruitment, which influenced biomass trends. Future
work should seek to assess the drivers of Cisco recruitment and incorporate
recruitment variability in EcoSim. Model development will continue into
future reporting periods, and we will be able to utilize future CSMI efforts
and agency surveys to better inform the model and test predictions of
population trajectories over time. For example, there are concerns by Lake
Superior biologists that the biomass of large-sized (>500 mm total length)
siscowets and leans generated from the bottom-trawl surveys are misleading
because the trawl itself does a poor job of capturing these large-sized fish.
The bottom-trawl biomass inputs to the EwE were downweighted relative to
other data sources because of these concerns. Studies of the selectivity and
catchability of large-sized Lake Trout forms and other species to the bottom
trawls would better inform future EwE simulations. The long-term goal is to
provide a reliable forecasting tool that can be used to predict outcomes of
management actions on various fish community objectives, given observed
trends in the Lake Superior ecosystem (sensu Kitchell et al. 2000).
60

-------