The distribution and abundance of deepwater ciscoes
in Canadian waters of Lake Superior
Thomas C. Pratt
with 8 ﬁ gures and 5 tables
Abstract: Deepwater ciscoes have declined precipitously in the Laurentian Great Lakes, and only
Lake Superior contains a fauna resembling that of the early 20th century. The discovery of areas
of high abundance and the identiﬁ cation of habitat preferences in Lake Superior would provide
important information for the basin-wide recovery of these species. This research completed the
most comprehensive deepwater cisco survey in the nearshore Canadian waters of Lake Superior to
(1) determine and compare the current distribution and relative abundance of ciscoes, (2) compare
coregonid abundance and community structure at speciﬁ c sites to historic data, and (3) assess habitat
preferences of Lake Superior ciscoes. Deepwater ciscoes remain the dominant prey ﬁ shes in Lake
Superior, and all four species—Coregonus artedi, C. hoyi, C. kiyi and C. zenithicus—remain widely
distributed. In general, embayment areas contain the highest densities of all species except C. kiyi,
which are found offshore, and there are lower densities in the eastern part of the lake. There have been
tremendous changes in the cisco community since the historic surveys, with the formerly dominant
C. zenithicus being replaced by C. artedi and C. hoyi. Spatial segregation primarily occurred with
depth; C. kiyi is most abundant in deep water (>130 m), C. hoyi and C. zenithicus are most abundant at
mid-depths (80–110 m), while C. artedi is most abundant at depths of <60 m.
Keywords: Lake Superior, Ciscoes, distribution, abundance, habitat.
The continuing loss of cisco diversity in the North American Laurentian Great Lakes since
the landmark coregonid surveys of KOELZ (1929) is one of the least discussed, but potentially
most important, conservation issues facing Great Lakes ﬁ shery managers. Ciscoes are now
recognized as being important ecological integrators in deep lakes as a trophic connection
between benthic invertebrates and piscine predators, and there is interest in restoring cis-
coes in areas where they have been extirpated to help stabilize and recover deepwater food
webs (FAVÉ & TURGEON 2008). There were once seven cisco species identiﬁ ed from the Great
Fisheries and Oceans Canada, Great Lakes Laboratory for Fisheries and Aquatic Sciences, 1219 Queen
St. E., Sault Ste Marie, Ontario P6A 2E5, Canada.
* Corresponding author email: email@example.com
1612-166X/12/0063-0025 $ 4.25
© 2012 E. Schweizerbart’sche Verlagsbuchhandlung, 70176 Stuttgart, Germany
Advanc. Limnol. 63, p. 25–41
Biology and Management of Coregonid Fishes – 2008
26 T.C. Pratt
Lakes, but three now are believed to be extinct and the remaining communities are greatly
diminished (SMITH 1964, SMITH 1972). Lake Superior has the most intact cisco fauna, with
four species remaining: Coregonus artedi, considered a shallow water species, and C. hoyi,
C. kiyi, and C. zenithicus, which collectively make up the deepwater ciscoes. C. kiyi is only
extant in Lake Superior; Lake Superior also contains one of only two taxonomically accepted
populations of C. zenithicus (COSEWIC 2009).
Deepwater ciscoes were once part of a sizeable targeted ﬁ shery in Lake Superior, with
almost 11 million metric tonnes having been harvested from 1894–1950 (HOFF & TODD
2004, CHIARAPPA 2005). C. zenithicus constituted more than 90% of the deepwater ciscoes
in lakewide surveys in the 1920s (KOELZ 1929), but the most recent surveys in United States
waters have documented a tremendous shift in the deepwater community. C. zenithicus
appears to have been replaced by C. hoyi and C. kiyi and make up, on average, <5% of the
catch in areas where they remain (HOFF & TODD 2004). The C. zenithicus decline has pri-
marily been attributed to commercial over-harvest (LAWRIE 1978), although the introduction
of invasive species, habitat degradation and inter-speciﬁ c competition or predation are sug-
gested as being factors that may be limiting recovery (BRONTE et al. 2010). The large decline
in deepwater ciscoes has resulted in the listing of C. kiyi as a species of Special Concern
and the recommendation that C. zenithicus be listed as Threatened under Canadian federal
Species-at-Risk legislation. Despite the obvious need for a comprehensive deepwater cisco
survey in the Canadian waters of Lake Superior, it has never been completed.
Assessing deepwater cisco populations is challenging due to the spatial overlap and mor-
phological plasticity of these species. The three Lake Superior deepwater cisco species and
C. artedi , which can also be found at similar depths, share physical characteristics and there
is no single diagnostic character that separates these species, although together the combina-
tion of body shape, head and eye morphology and gill raker counts allow identiﬁ cation of
each species (TODD & SMITH 1980, COSEWIC 2009). SELGEBY & HOFF (1996) presented evi-
dence for differences in inter-speciﬁ c depth distribution, with most C. zenithicus and C. kiyi
found at depths of 105–145 m, most C. hoyi at 65–105 m, and C. artedi primarily at <65 m,
but there remains considerable overlap in the depths used by the three deepwater forms, and
very little else is known about the habitat preferences of these three species.
The discovery of areas of high relative abundance and the identiﬁ cation of habitat prefer-
ences in Lake Superior would provide important baseline information for planning the basin-
wide recovery of these cisco species. The objectives of this study were to (1) determine and
compare the current distribution and relative abundance of cisco species across the Canadian
waters of Lake Superior, (2) compare coregonid abundance and community structure at spe-
ciﬁ c sites to the analogous historic data from the KOELZ (1929) assessment, and (3) assess
habitat (slope, temperature, substrate, and depth) preferences of Lake Superior ciscoes.
Materials and methods
A total of 140 nets were set in the Canadian waters of Lake Superior during four years: 2004, 2006,
2007, and 2008 (Fig. 1). Two types of nets were used in the sampling: a ‘traditional’ net and an ‘experi-
mental’ net. The traditional net was composed of four 92 m nylon mesh panels, alternating 64 mm and
70 mm stretch mesh (for a total gang length of 366 m); mesh size and material for the traditional nets
were chosen to as closely replicate the nets used in the historic KOELZ (1929) surveys, though KOELZ
The distribution and abundance of deepwater ciscoes 27
(1929) used cotton or linen nets instead of nylon. The experimental net consisted of eight randomly
assigned 46 m monoﬁ lament panels, with stretched mesh sizes of 38, 45, 51, 57, 64, 70, 76, and 89 mm.
The experimental nets were designed to capture a wider size range of ﬁ shes so that the smaller C. hoyi
and C. kiyi could be better enumerated. Sampling dates and locations, and the number of each net type
set in a given year are presented in Table 1. The sampling covered 17 of the Lake Superior ﬁ sheries
management units. Management units are modiﬁ ed statistical districts previously used for reporting
commercial ﬁ shery statistics (SMITH et al. 1961; HANSEN et al. 1995).
Gillnets were deployed on the bottom, and minimum and maximum depth data were collected from
every net set. Starting in 2007, an Ekman dredge was used to collect a sediment sample from the
immediate area where each net gang was set to better understand the habitat preferences of deepwater
ciscoes. In addition, a temperature data logger was afﬁ xed to each net, and water temperature was
recorded at 10 s intervals.
Gillnet catches were tallied on-board by mesh size, and a total weight of ﬁ sh caught in each net was
taken for each species. Ciscoes captured in 2004 and 2006 were tentatively identiﬁ ed to species based
on external morphological characteristics (primarily mouth and ﬁ n position, gill raker characteristics,
and colour) and ﬁ ve individuals from each of the four species were bagged, labelled, and frozen for
conﬁ rmation of tentative identiﬁ cation, and additional morphometric and biological analyses. Ciscoes
were identiﬁ ed with the aid of a variety of unpublished keys. In 2007 and 2008, all ciscoes were frozen
and returned to the laboratory for later analyses.
In the laboratory, frozen ﬁ sh were thawed, photographed (full body, head, and gill rakers), weighed,
and total length recorded. Fish were identiﬁ ed by three biologists, and any potential changes in clas-
Table 1. The sampling locations, dates, and number of net sets for each of the four years of deepwater
cisco sampling in Lake Superior.
Year Sampling location Dates # experimental
2004 Rossport September 14–23 0 27
2006 Thunder Bay
Rossport to Marathon June 12–21
August 9–17 5
2007 Sault Ste Marie to Wawa June 19–July 18 17 10
2008 Wawa to Rossport June 8–July 7 24 10
Fig. 1. Map displaying the key sites and net locations from the deepwater cisco survey in Lake Superior.
28 T.C. Pratt
siﬁ cation were noted. A number of specimens had characteristics from more than one species; these
ﬁ sh were classiﬁ ed as hybrids with the species for which they most closely ﬁ t the criteria listed ﬁ rst
(e.g., a shortjaw/bloater would have more shortjaw cisco characteristics than bloater characteristics). In
2006, a chest freezer failed and most of the collected specimens spoiled, and identiﬁ cations could not
be revisited. Gill rakers and a tissue sample for future genetic analysis were removed and stored in 70%
ethanol. Aging structures (scales and otoliths) were also collected, and individuals were examined for
sex determination and state of maturity for an expected life history examination.
The relative abundance of ciscoes was assessed using the combined mesh size capture data from over-
night gillnet sets, except for three nets that were left for two nights due to adverse weather conditions.
For those three sets, the catch was divided by two, and those data were used in the analyses. Only nets
that were set 60 m or deeper were used in the following analyses. The relative abundance of C. artedi,
C. hoyi, C. kiyi, and C. zenithicus was compared from the Canadian waters of Lake Superior from three
areas: western (ﬁ sheries management units 5 and 6), northern (ﬁ sheries management units 11, 12 17,
18, 19, 20, and 21) and eastern (ﬁ sheries management units 23, 24, 26, 27, 28, 31, 33, and 34) waters via
one-way analyses of variance. Data were ln (x +1) transformed to meet the assumptions of homogene-
ity of variance and normality of the residuals. When signiﬁ cant differences were detected, Tukey HSD
tests were used to distinguish among areas (ZAR 1999). Bonferroni corrections were applied to ensure
that the question-wise probability remained at α = 0.05. Only data from experimental gillnets were used
in this analysis as the mesh sizes of the traditional nets do not effectively capture C. hoyi or C. kiyi. The
relative abundance data for both the traditional and experimental nets were converted to ﬁ sh per gillnet
kilometre, and the results displayed by ﬁ sheries management unit.
Speciﬁ c locations in Canadian waters initially ﬁ shed by KOELZ (1929) were ta rgeted during the sur-
veys to allow a contrast in the coregonid community over the past 80 years. Traditional nets ﬁ shed
from all depths were used in this analysis. In addition to the four cisco species, the lake whiteﬁ sh C.
clupeaformis was included in this examination. Nets of various lengths were often ﬁ shed for more than
one night in the KOELZ (1929) survey, so the number of ﬁ sh captured in these sets was adjusted to reﬂ ect
the relative abundance captured per net kilometre over a single night of effort. Two species identiﬁ ed by
KOELZ (1929), C. reighardi and C. nigrippinus, were later synonymized with C. zenithicus and included
in the totals for that species (TODD & SMITH 1980, TODD et al. 1981). Differences in the composition of
the coregonid community at speciﬁ c sites surveyed by KOELZ (1929) and the contemporary surveys was
assessed using log-linear analysis (ZAR 1999). Bonferroni corrections were applied to ensure that the
question-wise probability remained at α = 0.05. The use of log-linear analyses was not possible for the
Alona Bay, Quebec Harbour, Silver Islet, and Thunder Cape sites because the communities were too
simple and there were too many empty cells in the analysis. The total number of coregonids captured
at the sites was compared between the KOELZ (1929) and contemporary surveys using a paired t-test to
determine if observed changes in community composition reﬂ ected losses of coregonids or a potential
redistribution of ﬁ shes within the community. Data were ln (x+1) transformed to meet the assumption
of homogeneity of variances required for the use of parametric statistics.
A standard multiple regression approach was used to evaluate whether any of four habitat para-
meters—slope, depth, temperature and substrate type—were able to account for signiﬁ cant variation in
the number of each cisco species that was captured at each site. Catch data were ln (x+1) transformed to
meet the assumptions of homogeneity of variance and normality of residuals. The analysis was limited
to experimental nets from 2007–2008 because temperature and substrate data were only available from
that timeframe, and the traditional nets sampled C. hoyi and C. kiyi poorly. Slope was calculated as the
percent change in depth from the shallowest to deepest part of a net set. Depth was the mean depth of
the set, temperature was the mean bottom temperature during the net set, and substrate was classiﬁ ed
as the dominant substrate type (sand, silt or clay) from the bottom grab. Rough weather that prevented
substrate sampling and a broken temperature logger resulted in a few instances where these data were
not available for a given net; after excluding these sites from the analysis, 39 sites remained. To allow
The distribution and abundance of deepwater ciscoes 29
a better understanding of the inﬂ uence of depth on the distribution of the four cisco species, all experi-
mental net sets (n = 72) were divided into 20 m depth bins and the mean catch per depth bin was plotted
for each species. Depth data were further divided into 10 m depth bins, and the depths at which more
than 50% of the cumulative catch for that species occurred were highlighted as the abundance median
as per SELGEBY & HOFF (1996). Similarly, substrate data (n = 41 sites) were plotted with mean catch data
to further elucidate habitat preferences.
A total of 4249 ciscoes were collected from the 140 net sets. C. hoyi and C. artedi made up
the majority of the cisco catch in both the experimental (n = 61) and traditional (n = 79) nets
(Table 2). The patterns of relative abundance for the four cisco species were similar, with
lower abundance generally observed in the eastern basin. The relative abundance of C. artedi
signiﬁ cantly differed among areas (F2,58 = 7.5; P = 0.001), with fewer individuals found in the
eastern waters of Lake Superior than in either the western or northern areas (eastern vs. west-
ern Tukey HSD P = 0.009; eastern vs. northern Tukey HSD P = 0.007; Fig. 2). This pattern
is readily apparent in C. artedi abundance by management unit from both the experimental
and traditional nets. High relative abundance was noted in the management units nearest the
shore close to Thunder Bay and along the north shore, but with the exception of a couple of
the Michipicoten Harbour and Whiteﬁ sh Bay management units from the experimental nets,
relative abundance along the eastern shore was low (Fig. 3a, b).
Table 2. The total number of ciscoes captured in experimental (n = 61) and traditional (n = 79) gillnets
set in the Canadian waters of Lake Superior.
Net type C. artedi C. hoyi C. kiyi C. zenithicus
Experimental 542 1294 313 345
Traditional 889 613 51 202
Fig. 2. The relative abundance of
Coregonus artedi, C. hoyi, C. kiyi, and
C. zenithicus collected in experimental
gillnets from three broad areas
(western, northern, and eastern) in the
Canadian waters of Lake Superior.
30 T.C. Pratt
Fig. 3. The mean number of Coregonus artedi captured per gillnet km from the Canadian management
units of Lake Superior in 2004–2008 using a) experimental and b) traditional gill nets ﬁ shed at depth
60 m or greater.
The distribution and abundance of deepwater ciscoes 31
The relative abundance of C. hoyi differed among areas (F2,58 = 5.5; P = 0.006, w ith lower
abundance in eastern areas than northern areas (eastern vs. northern Tukey HSD P = 0.004;
Fig. 2). Traditional nets were ineffective at sampling the relatively small C. hoyi, as only the
management units near the village of Rossport and in Whiteﬁ sh Bay had reasonable catches
(Fig. 4b). C. hoyi was captured in high abundance along the north shore management unit in
experimental nets, and there were pockets of high abundance in Whiteﬁ sh Bay, Michipicoten
Harbour and near Black Bay (Fig. 4a).
After applying the Bonferroni correction, no differences in relative abundance were
observed in C. kiyi from the western, northern and eastern areas of Lake Superior (F2,58 =
3.2; P = 0.048; Fig. 2). Likely because of its small size, C. kiyi was not often captured in
traditional nets (Fig. 5b), and, with the exception of Michipicoten Harbour C. kiyi was almost
absent from eastern waters in the experimental nets (Fig. 5a). C. kiyi was the most abundant
in offshore zones along the north shore, and was detected in moderate abundance in western
waters (Fig. 5a).
Patterns of C. zenithicus relative abundance were similar to those of C. hoyi, with sig-
niﬁ cant differences detected among areas (F2,58 = 5.9; P = 0.004), and lower abundance in
eastern areas than northern areas (eastern vs. northern Tukey HSD P = 0.003; Fig. 2). Outside
of the north shore, only the Michipicoten Harbour management unit had high relative abun-
dance in experimental nets (Fig. 6a), while C. zenithicus was consistently captured in low
abundance in the majority of management units that were sampled (Fig. 6b).
The historic Lake Superior coregonid community was quite different than the community
that exists today. C. zenithicus was the dominant coregonid in the KOELZ (1929) surveys,
making up more than 90% of the catch. In contrast, C. artedi and C. hoyi were the dominant
ciscoes at most sites today (Table 3). The coregonid community composition has signiﬁ -
cantly changed in every area of the lake (Table 4), primarily due to the decline in C. zenithi-
cus. The total number of coregonids that were captured has not signiﬁ cantly changed over
that time period (t = 1.81, df = 9, P = 0.11), although there may be patterns among different
areas of the lake. Catches from the majority of sites in eastern and western Lake Superior had
declined, while those along the north shore had not (Table 3).
The four habitat attributes (slope, temperature, substrate type and depth) signiﬁ cantly
accounted for part of the variation in the capture of the four Lake Superior ciscoes, ranging
from 30% of the C. artedi capture variation to 72% of the C. kiyi capture variation (Table 5).
Depth signiﬁ cantly contributed to the multiple regression models for all four species, while
substrate signiﬁ cantly contributed to all but the C. kiyi model (Table 5). Temperature never
signiﬁ cantly added to any model, and slope was only identiﬁ ed as important for C. hoyi
(Table 5). C. artedi was captured at all depths, with a general trend of decreasing abundance
with increasing depth (Fig. 7a), and the abundance median was 50–70 m for this species (Fig.
7b). C. kiyi displayed the opposite trend, increasing in abundance with increasing depth (Fig.
7a). C. kiyi was not captured at depths <70 m, and had the deepest abundance median (Fig.
7b). Both C. hoyi and C. zenithicus were absent at the shallowest depths, most abundant at
intermediate depths and declined in abundance at the deepest depths, although the pattern for
C. zenithicus was slightly deeper than C. hoyi (Fig. 7a, 7b). Sand-dominated substrates were
the least utilized for all species, with catches of C. artedi, C. hoyi and C. zenithicus being
highest on silt-dominated substrates (Table 5, Fig. 8).
32 T.C. Pratt
Fig. 4. The mean number of Coregonus hoyi captured per gill net km from the Canadian management
units of Lake Superior in 2004-2008 using a) experimental and b) traditional gillnets ﬁ shed at depth
60 m or greater.
The distribution and abundance of deepwater ciscoes 33
Fig. 5. The mean number of Coregonus kiyi captured per gillnet km from the Canadian management
units of Lake Superior in 2004–2008 using a) experimental and b) traditional gill nets ﬁ shed at depth
60 m or greater.
34 T.C. Pratt
Fig. 6. The mean number of Coregonus zenithicus captured per gill net km from the Canadian
management units of Lake Superior in 2004-2008 using a) experimental and b) traditional gillnets
ﬁ shed at depth 60 m or greater.
The distribution and abundance of deepwater ciscoes 35
Table 3. A comparison of Coregonus spp. captured in the same location by KOELZ (1929) and the contemporary DFO surveys. Numbers in parentheses
represent standard errors.
Location Site Survey nMean
C. artedi C. clupea-
C. hoyi C. kiyi C. zenithi-
Marie Whiteﬁ sh
Bay Koelz 1 69.5 3.6 (-) 1.8 (-) 0.0 (-) 0.0 (-) 182.3 (-) 187.7 (-)
DFO 4 71.0 8.9 (4.1) 0.7 (0.7) 47.8 (10.6) 0.0 (0.0) 4.8 (2.3) 62.2 (9.5)
Point Off Alona
Bay Koelz 1 109.7 0.0 (-) 0.0 (-) 0.0 (-) 0.0 (-) 72.9 (-) 72.9 (-)
DFO 1 110.9 0.0 (-) 0.0 (-) 8.2 (-) 0.0 (-) 0.0 (-) 8.2 (-)
Island 3 miles SE
Koelz 1 146.3 0.0 (-) 0.0 (-) 0.0 (-) 0.0 (-) 32.8 (-) 32.8 (-)
DFO 1 150.3 0.0 (-) 0.0 (-) 5.5 (-) 0.0 (-) 0.0 (-) 5.5 (-)
Rossport Off Bread
Rock Koelz 1 155.5 3.3 (-) 0.0 (-) 6.6 (-) 0.0 (-) 172.2 (-) 182.1 (-)
DFO 2 152.9 13.7 (13.7) 0.0 (-) 10.7 (2.7) 0.0 (-) 8.2 (8.2) 32.8 (24.6)
Channel Koelz 1 135.3 0.0 (-) 0.0 (-) 0.0 (-) 0.0 (-) 13.1 (-) 13.1 (-)
DFO 3 133.2 6.4 (5.1) 13.7 (6.3) 2.7 (2.7) 0.0 (-) 9.1 (5.1) 31.9 (5.6)
Island Koelz 1 76.8 0.0 (-) 0.0 (-) 6.6 (-) 0.0 (-) 22.1 (-) 28.7 (-)
DFO 13 78.3 69.2 (13.6) 24.5 (2.7) 54.3 (12.3) 3.8 (2.9) 16.0 (4.8) 167.7 (27.4)
Moffat Strait Koelz 1 24.7 23.0 (-) 26.2 (-) 0.0 (-) 0.0 (-) 16.4 (-) 65.6 (-)
DFO 1 28.3 16.4 (-) 68.4 (-) 0.0 (-) 0.0 (-) 0.0 (-) 84.8 (-)
Thunder Bay North Silver
Islet Koelz 1 25.6 0.0 (-) 0.0 (-) 0.0 (-) 0.0 (-) 144.4 (-) 144.4 (-)
DFO 1 18.0 19.1 (-) 0.0 (-) 0.0 (-) 0.0 (-) 0.0 (-) 19.1 (-)
Cape Koelz 1 56.7 0.0 (-) 0.0 (-) 0.0 (-) 0.0 (-) 45.9 (-) 45.9 (-)
DFO 2 51.2 34.2 (17.8) 0.0 (-) 0.0 (-) 0.0 (-) 0.0 (-) 34.2 (17.8)
Welcome Island Koelz 1 42.1 0.0 (-) 0.0 (-) 23.0 (-) 0.0 (-) 193.6 (-) 216.5 (-)
DFO 2 49.8 37.6 (6.6) 2.1 (1.1) 2.1 (0.7) 0.0 (-) 0.2 (0.2) 41.8 (7.1)
36 T.C. Pratt
Table 4. Log-linear analysis results of coregonid species composition from sites ﬁ shed by KOELZ (1929)
and revisited by this study.
Location Site Maximum
Sault Ste Marie Whiteﬁ sh Bay 212.5 <0.001
Rossport Off Bread Rock 77.4 <0.001
Simpson Channel 18.3 0.001
Off Salter Island 62.3 <0.001
Moffat Strait 37.6 <0.001
Thunder Bay South Welcome Island 203.7 <0.001
Fig. 7. a) The mean capture of Coregonus artedi, C. hoyi, C. kiyi, C. zenithicus from the experimental
gillnets by depth bin. Error bars represent standard deviation. b) The range of depths of capture (thin
line) and the depth at which more than 50% of the cumulative catch for that species occurred (thick line)
for Coregonus artedi, C. hoyi, C. kiyi, and C. zenithicus.
The distribution and abundance of deepwater ciscoes 37
Fig. 8. The mean number of Coregonus artedi, C. hoyi, C. kiyi, and
C. zenithicus by dominant substrate type caught in experimental
gillnets in Lake Superior.
Table 5. Multiple regression results and estimates for the intercept (β0) and predictor variables (βi’s) used to evaluate habitat variables in the cisco
habitat preference models. The predictor variables included slope (SLOPE), temperature (TEMP, in ºC), dominant substrate type (SUBS; sand, silt or
clay) and depth (DEPTH, in m). Signiﬁ cant predictor variables are marked with an asterisk.
SLOPE TEMP SUBS DEPTH R2Fdf P
Species β0ΒSLOPE PβTEMP PΒSUBS PβDEPTH P
C. artedi 40.59 0.129 0.380 0.202 0.221 -0.320 0.049* 0.422 0.009* 0.30 3.64 4,34 0.014
C. hoyi 74.77 0.250 0.044* 0.236 0.088 -0.420 0.002* 0.531 <0.001* 0.52 9.21 4,34 <0.001
C. kiyi -2.77 -0.032 0.723 0.197 0.062 -0.009 0.924 0.888 <0.001* 0.72 22.26 4,34 <0.001
C. zenithicus 34.5 0.133 0.286 0.241 0.088 -0.336 0.159* 0.633 <0.001* 0.50 8.48 4,34 <0.001
Coregonids remain the dominant prey for piscivorous ﬁ shes
in the Canadian waters of Lake Superior, despite signiﬁ -
cant changes in the community since the surveys of KOELZ
(1929). Overexploitation and the introduction of invasive
species decimated the nearshore and offshore ﬁ sh commu-
nities in Lake Superior by the mid-20th century (LAWRIE &
RAHRER 1972, BRONTE et al. 2003), and the ﬁ sh community
has been slowly evolving since that time. C. artedi popula-
tions began recovering in the 1970s and are now considered
to be at their former historic abundance levels (EBENER et
al. 2009). C. hoyi populations exploded in the United States
waters of Lake Superior in the mid-1980s (BRONTE et al.
2003), and this species is now the lake’s dominant deep-
water cisco form. This research conﬁ rms that C. artedi and
C. hoyi now dominate the cisco community in Canadian
waters as well. While still widely dispersed in Lake Supe-
rior, C. zenithicus persists at a much lower abundance than
it did during historic surveys. The replacement of C. zenithi-
cus has been a gradual process; it constituted over 90% of
the KOELZ (1929) surveys, 34% of the cisco community in
the late 1950s (E.H. BROWN JR., unpublished data, cited in
HOFF & TODD 2004), around 11% of the PECK (1977) sam-
38 T.C. Pratt
ples from the early 1970s, and only 5% of the community along the south shore of Lake
Superior in 1999–2001 (HOFF & TODD 2004). C. zenithicus made up a higher proportion of
the catch (13%) in this survey, which might be interpreted as evidence for limited recovery
in the past decade, or that abundance in the Canadian waters of Lake Superior is higher than
those in the United States waters. SMITH (1970) hypothesized that the introduction and popu-
lation growth of alewife (Alosa pseudoharengus) populations resulted in the loss of coregon-
ids from the other Great Lakes, and the inability of alewife to establish in Lake Superior may
be why ciscoes have persisted in Lake Superior but not the remaining Great Lakes.
Ciscoes are not evenly distributed around the Canadian waters of Lake Superior. In gen-
eral, eastern management areas had lower relative abundance for all species than northern
and western management areas. These patterns may reﬂ ect the nearshore habitat available in
those broad areas, the location of spawning areas and the location of current ﬁ shing effort.
The western management areas contain Black Bay and Thunder Bay, two embayments which
contain the largest C. artedi ﬁ sheries on the lake (EBENER et al. 2009). Embayments are the
most productive habitats in Lake Superior (DEVINE et al. 2005), and the locations in the
eastern management area that did contain relatively high cisco abundance are embayments,
including Whiteﬁ sh Bay and Michipicoten Harbour. Little is known about the spawning are-
as for deepwater ciscoes, but almost all the known spawning areas for C. artedi are in the
western part of the lake (EBENER et al. 2009).
Commercial harvest and subsistence harvest is nearly absent along the north shore of
Lake Superior, but a large commercial effort exists in the western management areas and
an unknown, but presumably signiﬁ cant, First Nation ﬁ shery exists in the eastern manage-
ment areas (EBENER et al. 2009). It is possible that the combination of limited embayment
and spawning habitat, and the First Nation ﬁ shery are limiting populations in the eastern
part of the lake. Regardless of the reason or reasons, the combination of relatively high
abundance, good habitat and limited ﬁ shery along the north shore may provide opportuni-
ties for preserving the ciscoe species of conservation concern (C. zenithicus and C. kiyi)
in Lake Superior. The recent establishment of the Lake Superior National Marine Con-
servation Area (PARKS CANADA 2007), which encompasses approximately 10,000 km2 of
lakebed along the north shore, will bring additional resources and attention to conservation
concerns in this area.
While there has been no signiﬁ cant decrease in the numbers of coregonid species captured
from the contemporary and KOELZ ( 1929) surveys, there are reductions in the majority of the
KOELZ (1929) locations that were revisited. It is important to note that differences in catch-
ability between the gillnets ﬁ shed by KOELZ (1929) and those in this study likely resulted in
an underestimate of the already high historic catches, which would have further increased
these differences. The replica nets used in this study were constructed of nylon, which has
approximately three times the capture efﬁ ciency of the cotton gillnets used by KOELZ (1929)
(MCCOMBIE & FRY 1960, PYCHA 1962).
Depth is an important factor for segregating Lake Superior ciscoes, though there is sub-
stantial among-species overlap. C. artedi was once considered a shallow water species, but
there is increasing evidence, including this study, that they can be abundant at depths >100
m (SELGEBY & HOFF 1996, EBENER et al. 2009). However, C. artedi remains the dominant
shallow water cisco form in the lake and it is not known whether this species spawns in
deep water (EBENER et al. 2009). The presence of C. artedi outside of its historical depth
The distribution and abundance of deepwater ciscoes 39
range may indicate movement into a feeding niche vacated with the decline of deepwater
ciscoes in the lake; this is the proposed mechanism for the apparent sympatric evolution
of ciscoes within the Great Lakes (TURGEON & BERNATCHEZ 2003). The depth distribu-
tions of C. hoyi and C. zenithicus overlapped considerably, with C. zenithicus found at
only slightly deeper depths than its congener. These depth preferences have not changed
in Lake Superior over the past few decades (SELGEBY & HOFF 1996, HOFF & TODD 2004),
although once C. hoyi was thought to have a shallower distribution (KOELZ 1929). C. kiyi
inhabited the deepest depths, and its abundance appeared to be increasing up to the limits
of the depths sampled in this study, suggesting that it may be even more abundant in depths
deeper than 140 m. Similar observations were made by SELGEBY & HOFF (1996). Diet is
likely the mechanism for maintaining depth segregation, as adult C. artedi in Lake Supe-
rior remain primarily zooplanktivorous (LINK et al. 1995, JOHNSON et al. 2004), while the
remaining species are mostly benthivores, primarily consuming Mysis relicta and Diporeia
hoyi (KOELZ 1929, ANDERSON & SMITH 1971, HOFF & TODD 2004). C. kiyi has been shown
to undertake extensive daily vertical migrations following M. relicta (HRABIK et al. 2006).
Given the contrasting abundance trends over the past century, and the spatial and diet
overlap between C. hoyi and C. zenithicus, it is possible that competition between these
congeners is responsible for the observed decline in C. zenithicus. In general, the depth
distributions of all cisco species were deeper than those observed by KOELZ (1929), but this
may be due to gear limitations of the early 20th century surveys more than a shift in depth
over the past century (HOFF & TODD 2004).
The second important habitat factor inﬂ uencing abundance for all ciscoes except C. kiyi
was dominant substrate type. In general, species were more likely to be captured over silt,
then clay and ﬁ nally sand substrates. This pattern is likely related to prey availability; benthic
invertebrates such as D. hoyi are more abundant over silt than clay-dominated substrates, and
sand contains the lowest densities (SLY & CHRISTIE 1992, LOZANO et al. 2001, NALEPA et al.
2003). This is because ﬁ ner substrates contain more organic material as food for the benthic
invertebrates (MARZOLF 1965, SLY & CHRISTIE 1992, LOZANO et al. 2001). Benthivores such
as C. hoyi and C. zenithicus would be expected to congregate in areas of greatest prey pro-
duction, which would be the ﬁ nest substrates. Interestingly, in Lake Ontario, there is a rela-
tionship between depth and substrate, with shallower depths containing coarser (more sand-
dominated) substrates (LOZANO et al. 2001), and there are correspondingly higher densities of
invertebrates such as M. relicta in deeper depths (LOZANO et al. 2001, RUDSTAM et al. 2008).
In summary, this research provides the ﬁ rst comprehensive survey of coregonids in the
Canadian waters of Lake Superior since the KOELZ (1929) survey in the early 20th century.
While there have been signiﬁ cant changes in the community since that time, Lake Superior
remains the only Great Lake where coregonids remain the dominant prey ﬁ shes. All four
Lake Superior ciscoes are widely distributed, with lower overall densities in the eastern part
of Lake Superior, but higher relative abundance in embayment areas regardless of their posi-
tion on the lake. Ciscoes are spatially segregated by depth, with C. artedi dominating water
<60 m, C. hoyi and C. zenithicus at intermediate depths (80–110 m), and C. kiyi dominant
in deeper waters (>120 m). The two species of conservation concern, C. kiyi and C. zenithi-
cus, were most abundant along the north shore of Lake Superior, and the establishment of
a National Marine Conservation Area along the north shore should provide momentum for
preserving coregonid diversity at these sites.
40 T.C. Pratt
A number of key people contributed to this study, including Lisa O’Connor, Bill Gardner, Marla Thi-
bodeau, Lisa Voigt, Jennifer Allemang, Jon Chicoine and Jim Dyson in the ﬁ eld collections, and Jan-
ice McKee, Kim Caldwell, Greg Elliott, Cheryl Widdiﬁ eld and Sharon Templeton in the laboratory
processing. Ken Mills and an anonymous reviewer provided valuable advice on an earlier draft of this
manuscript. Funding was provided by the Department of Fisheries and Oceans Species at Risk Coor-
dination Espèces en Péril (SARCEP) and Environment Canada’s Intergovernmental Recovery Fund.
ANDERSON, E.D. & SMITH, L.L. (1971): A synoptic study of food habits of 30 ﬁ sh species from western
Lake Superior. – Minn. Agric. Exp. Stn. Rep. No. 279.
BRONTE, C.R., EBENER, M.P., SCHREINER, D.R., DEVAULT, D.S., PETZOLD, M.M., JENSEN, D.A., RICHARDS,
C. & LOZANO, S.J. (2003): Fish community change in Lake Superior, 1970–2000. – Can. J. Fish.
Aquat. Sci. 60: 1552–1574.
BRONTE, C.R., HOFF, M.H., GORMAN, O.T., THOGMARTIN, W.E., SCHNEEBERGER, P.J. & TODD, T.N. (2010):
Decline of the shortjaw cisco in Lake Superior: the role of overﬁ shing and risk of extinction. – Trans.
Am. Fish. Soc. 139: 735–748
CHIARAPPA, M.J. (2005): Overseeing the family of whiteﬁ shes: The priorities and debates of coregonid
management on America’s Great Lakes, 1870–2000. – Environ. Hist. 11: 163–194.
COSEWIC (2009): COSEWIC assessment and update status report on shortjaw cisco Coregonus
zenithicus in Canada. Committee on the Status of Endangered Wildlife in Canada. Ottawa, ON.
DEVINE, J.A., SCHRAM, S.T. & HANSEN, M.J. (2005). A food web analysis of the ﬁ shery in Chequamegon
Bay, Lake Superior. – Wisc. Dept. Nat. Resour. Fish. Manage. Rep. 150. Madison, WI.
EBENER, M.P., STOCKWELL, J.D., YULE, D.L., GORMAN, O.T., HRABIK, T.R., KINNUNEN, R.E., MATTES, W.P,
OYADOMARI, J.K., SCHREINER, D.R., GEVING, S., SCRIBNER, K., SCHRAM, S.T., SEIDER, M.J. & SITAR, S.P.
(2009): Status of cisco (Coregonus artedi) in Lake Superior during 1970–2006 and management and
research considerations. – Great Lakes Fish. Comm. Spec. Publ. Ann Arbor, MI.
FAVÉ, M.J. & TURGEON, J. (2008): Patterns of genetic diversity in Great Lakes bloaters (Coregonus hoyi)
with a view to future reintroduction in Lake Ontario. – Conserv. Genet. 9: 281–293.
HANSEN, M.J., SCHORFHAAR, R.G., PECK, J.W., SELGEBY, J.H. & TAYLOR. W.W. (1995): Abundance
indices for determining the status of lake trout restoration in Lake Superior. – North Am. J. Fish.
Manage. 15: 830–837.
HOFF, M.H. & TODD, T.N. (2004): Status of the shortjaw cisco (Coregonus zenithicus) in Lake Superior.
– Ann. Zool. Fenn. 41: 147–154.
HRABIK, T.R., JENSEN, O.P., MARTELL, S.J.D., WALTERS, C.J. & KITCHELL, J.F. (2006): Diel vertical
migration in the Lake Superior pelagic community. I. Changes in vertical migration of coregonids in
response to varying predation risk. – Can. J. Fish. Aquat. Sci. 63: 2286–2295.
JOHNSON, T.B., BROWN, W.P., CORRY, T.D., HOFF, M.H., SCHAROLD, J.V. & TREBITZ, A.S. (2004): Lake
herring (Coregonus artedi) and rainbow smelt (Osmerus mordax) diets from western Lake Superior.
– J. Great Lakes Res. 30 (Suppl 1): 407–1413.
KOELZ, W. (1929): Coregonid ﬁ shes of the Great Lakes. – Bull. Bur. Fish. 43: 297–643.
LAWRIE, A.H. (1978): The ﬁ sh community of Lake Superior. – J. Great Lakes Res. 4: 513–549.
LAWRIE, A.H. & RAHRER, J.F. (1972): Lake Superior: effects of exploitation and introductions on the
salmonid community. – J. Fish. Res. Bd. Can. 29: 765–776.
LINK, J., SELGEBY, J.H., HOFF, M.H. & HASKELL, C. (1995): Winter diet of lake herring (Coregonus
artedi) in western Lake Superior. – J. Great Lakes Res. 21: 395–399.
LOZANO, S.J., SCHAROLD, J.V. & NALEPA, T.F. (2001): Recent declines in benthic macroinvertebrate
densities in Lake Ontario. – Can. J. Fish. Aquat. Sci. 58: 518–529.
The distribution and abundance of deepwater ciscoes 41
MARZOLF, G. (1965): Vertical migration of Pontoporeia afﬁ nis (Amphipoda) in Lake Michigan. – Proc.
Conf. Great Lakes Res. 8: 133–140.
MCCOMBIE, A.M. & FRY, F.E. (1960): Selectivity of gill nets for lake whiteﬁ sh, Coregonus clupeaformis.
– Trans. Am. Fish. Soc. 90: 176–184.
Nalepa, T.F., Fanslow, D.L., Lansing, M.B. & Lang, G.A. (2003): Trends in the benthic macroinvertebrate
community of Saginaw Bay, Lake Huron, 1987 to 1996: responses to phosphorus abatement and the
zebra mussel, Dreissena polymorpha. – J. Great Lakes Res. 29: 14–33.
PARKS CANADA (2007): 2007–2008 estimates: A report on plans and priorities. – Parks Canada Agency,
PECK, J.W. (1977): Species composition of deep-water ciscoes (chubs) in commercial catches from
Michigan waters of Lake Superior. – Res. Rep. No. 1849, Mich. Dept. Nat. Resour., Fish. Div. Ann
PYCHA, R.L. (1962): The relative efﬁ ciency of nylon and cotton gill nets for taking lake trout in Lake
Superior. – J. Fish. Res. Bd, Can. 19: 1085–1094.
RUDSTAM, L.G., SCHANER, T. GAL, G., BOSCARINO, B.T., O’GORMAN, R., WARNER , D.M., JOHANNSSON ,
O.E. & BOWEN, K. (2008): Hydroacoustic measures of Mysis relicta abundance and distribution in
Lake Ontario. – Aquat. Ecosyst. Health Manage. 11: 355–367.
SELGEBY, J.H.,& HOFF, M.H. (1996): Seasonal bathymetric distributions of 16 ﬁ shes in Lake Superior,
1958–75. – Biol. Sci. Rep. 7, Natl. Biol. Serv, Washington, D.C.
SLY, P.G. & CHRISTIE, W.J. (1992): Factors inﬂ uencing densities and distributions of Pontoporeia hoyi
in Lake Ontario. – Hydrobiologia 235/236: 321–352.
SMITH, S.H., BUETTNER, H.J. & HILE, R. (1961): Fishery statistical districts of the Great Lakes. – Great
Lakes Fish. Comm.Tech. Rep. No. 2, Ann Arbor, MI.
SMITH, S.H. (1964): Status of the deepwater cisco population of Lake Michigan. – Trans. Am. Fish. Soc.
SMITH, S.H. (1970): Species interactions of the alewife in the Great Lakes. – Trans. Am. Fish. Soc. 4:
SMITH, S.H. (1972): Factors of ecologic succession in oligotrophic ﬁ sh communities of the Laurentian
Great Lakes. – J. Fish. Res. Bd. Can. 29: 717–730.
TODD, T.N., & SMITH, G.R. (1980): Differentiation in Coregonus zenithicus in Lake Superior. – Can. J.
Fish. Aquat. Sci. 37: 2228–2235.
TODD, T., SMITH, G. & CABLE, L. (1981): Environmental and genetic contributions to morphological
differentiation in ciscoes (Coregoninae) of the Great Lakes. – Can. J. Fish. Aquat. Sci. 38: 59–67.
TURGEON, J. & BERNATCHEZ, L. (2003): Reticulate evolution and phenotypic diversity in North American
ciscoes, Coregonus spp. (Teleostei:Salmonidae): implication for the conservation of an evolutionary
legacy. – Conserv. Genet. 4: 67–81
ZAR, J.E. (1999): Biostatistical analysis. Prentice-Hall, Englewood Cliffs, N.J.