Ecological Applications, 16(4), 2006, pp. 1487–1501
? 2006 by the the Ecological Society of America
GIZZARD SHAD PUT A FREEZE ON WINTER MORTALITY OF AGE-0
YELLOW PERCH BUT NOT WHITE PERCH
DEAN G. FITZGERALD,1,2JOHN L. FORNEY, LARS G. RUDSTAM, BRIAN J. IRWIN, AND ANTHONY J. VANDEVALK
Department of Natural Resources, Cornell Biological Field Station, Cornell University, Bridgeport, New York 13030, USA
York, USA, provided an opportunity to investigate causes of mortality during winter, a period
of resource scarcity for most juvenile fishes, in age-0 yellow perch (Perca flavescens) and age-0
white perch (Morone americana). This time series contains several environmental (e.g., winter
severity) and biological (e.g., predator abundance) signals that can be used to disentangle
multiple effects on overwinter mortality of these fishes. A multiple regression analysis
indicated that age-0 yellow perch winter mortality was inversely related to fish length in
autumn and to the abundance of alternative prey (gizzard shad [Dorosoma cepedianum] and
white perch). However, no length-selective predation of yellow perch by one of the main
predators, adult walleye (Sander vitreus), was detected. In contrast, white perch mortality was
directly associated with total predator biomass and abundance of white perch in autumn, and
inversely related to yellow perch abundance as a potential buffer species, but not to the
abundance of gizzard shad. Winter severity was not a significant predictor of mortality for
either perch species. Predicted winter starvation mortality, from a model described in the
literature, was much lower than observed mortality for yellow perch. Similar models for white
perch were correlated with observed mortality. These results collectively suggest that
predation is the main mechanism shaping winter mortality of yellow perch, while both
predation and starvation may be important for white perch. This analysis also revealed that
gizzard shad buffer winter mortality of yellow perch. Although winter duration determines the
northern limit of fish distributions, in mid-latitude Oneida Lake and for these species,
predator–prey interactions seem to exert a greater influence on winter mortality than
Four decades of observations on the limnology and fishes of Oneida Lake, New
nonnative species; Perca flavescens; Sander vitreus; starvation model; walleye; white perch; yellow perch.
cohort dynamics; Dorosoma cepediurum; gizzard shad; long-term; Morone americana;
Winter is a period of resource scarcity for most
temperate fishes, and variation in overwinter mortality
may be critical for year class formation (Sogard 1997,
Wootton 1998). For many fish species, the northern
distribution is set by the ability of first year or age-0 fish
to accumulate sufficient body reserves during the
summer to survive the first winter (e.g., Shuter and Post
1990). These energetic needs have led to selection for
faster growth in northern compared with southern
populations: the countergradient selection phenomenon
(Conover and Schultz 1995, Munch et al. 2003).
Predation on age-0 fish during winter may also increase
if they have to take higher risks to find prey (Wootton
1998, Walters 2000). In addition, winter is stressful for
fishes generally, as most species are poorly adapted to
prolonged periods of low temperatures (e.g., Hazel
1993). A better understanding of the causes of winter
mortality of age-0 fish will improve our ability to predict
recruitment rates after ecosystem change occurs follow-
ing nonnative species invasions or climate warming
(Rose 2000, Jackson et al. 2001, Campbell et al. 2005).
Multiple factors likely contribute to the magnitude
and variability of winter mortality in age-0 fishes. Field
and laboratory studies have frequently identified an
inverse relationship between fish length and rate of
winter mortality (Ricker 1969, Sogard 1997, Schultz et
al. 1998, Craig 2000). Length-dependent mortality can
be caused by starvation, predation, or interactions
between the two. The simplest explanation for length-
dependent winter mortality across habitats is starvation
of the smallest fish, because energy reserves (e.g., the
percentage of body mass as lipids) increase with length,
and small fish have higher mass-specific metabolic rates
than larger fish (Paloheimo and Dickie 1966, Cunjack
1988). Small age-0 fish are also likely to suffer higher
predation mortality, as large conspecifics have improved
abilities to avoid predators, because swimming speed
and visual acuity increase with length (Wootton 1998,
Craig 2000). Also, starving fish often exhibit altered
swimming patterns that may increase their visibility and
vulnerability to predators (Jonas and Wahl 1998,
Wootton 1998). Exceptions to the observation of higher
Manuscript received 6 December 2004; revised 13 December
2005; accepted 14 December 2005. Corresponding Editor: T. E.
1Present address: EcoMetrix Incorporated, 14 Abacus
Road, Brampton, Ontario L6T 5B7 Canada.
winter mortality in small fish exist, and indicate that
winter mortality can be independent of body length,
possibly due to temperature intolerance or disease (Eddy
1981, Lankford and Targett 2001, McCollum et al.
2003). In addition, the diversity of prey species can
influence predator–prey interactions during any season,
and abundant prey species can buffer other species from
predation (Forney 1971, 1974, Sogard 1997). Thus,
winter mortality reflects the species-specific and length-
dependent temperature tolerance and energy depletion
rates, and is also affected by the suite of potential
predator and prey species present (Sogard 1997, Schultz
and Conover 1999, Craig 2000).
Long-term data series are ideal for investigating the
importance of different factors affecting winter mortal-
ity and recruitment rates of fishes. One such data series
is available for Oneida Lake, New York, USA. Studies
of the fishes in Oneida Lake were initiated in 1956 due to
the irruption of nonnative white perch, Morone amer-
icana, and gizzard shad, Dorosoma cepedianum; consis-
tent surveys of age-0 fish have been completed since 1962
(Forney 1980, Mills and Forney 1988). Oneida Lake is
naturally productive and located on the Lake Ontario
plain of central New York. With a surface area of ;207
km2, a maximum depth of ;16 m, a mean depth of ;6.5
m, and long fetch, it is well mixed through most of the
year (Mills et al. 1978). This lake supports one of New
York’s most valuable sport fisheries for walleye, Sander
vitreus, and yellow perch, Perca flavescens (Connelly and
Brown 1991). In the 1950s, white perch were viewed as
potential competitors with yellow perch, but recruitment
of white perch to the adult stage was sporadic and the
resident population remained relatively small (Aslop
and Forney 1962). Age-0 gizzard shad were abundant in
1954 but were rarely collected after 1956 (Fig. 1; Mills
and Forney 1988). The gizzard shad reappeared in the
early 1970s, and age-0 gizzard shad were often abundant
after 1983; the likely routes of entry to the lake were the
canal and locks that connect the lake to adjacent
catch for gizzard shad during 1962–2002.
Total catch of yellow perch, white perch, and gizzard shad in the 15-week summer gill net survey and in the otter trawl
DEAN G. FITZGERALD ET AL. 1488
Vol. 16, No. 4
watersheds (Mills and Forney 1988, Roseman et al.
Several studies have assessed the interactions involv-
ing yellow perch, white perch, and gizzard shad in
Oneida Lake between spring and autumn, and none
have documented negative effects of age-0 white perch
or gizzard shad on the growth or survival of age-0
yellow perch (Prout et al. 1990, Roseman et al. 1996,
Hall and Rudstam 1999). White perch and gizzard shad
spawn after yellow perch in Oneida Lake, and age-0
yellow perch can switch to benthic invertebrates if
zooplankton density is reduced by the other species in
summer (Prout et al. 1990, Roseman et al. 1996). If
direct competition is minimized among these age-0
cohorts, interactions involving these species may be
mostly positive because the presence of an abundant,
alternate prey can reduce short-term predation rates on
each species (Mills and Forney 1988, Prout et al. 1990,
Roseman et al. 1996). Over the long term, an alternate
prey may lead to greater predator biomass and thereby
increase the mortality rate of the primary prey species
(apparent competition; Holt and Lawton 1994, Polis
and Strong 1996). Assessment of possible long-term
effects on population from processes like apparent
competition requires an analysis with time lags that
may be best achieved through simulation models (e.g.,
Rose et al. 1999), and is beyond the scope of this paper.
The present study analyzes winter mortality as an
outcome of the specific conditions present each year
between autumn and spring.
Other physical and biological changes, in addition to
the irruption of white perch and gizzard shad, have been
observed in Oneida Lake since 1956. The large mayfly
Hexagenia limbata was extirpated during the late 1960s
due to summer anoxia (Mills et al. 1978). Prior to
extirpation, the abundance of mayfly was high (;200
individuals/m2), and these invertebrates likely buffered
age-0 yellow perch from predation during their mass
emergence in June and July (Forney 1980). Other
changes included the reduction of nutrient levels
through the 1980s and the invasion by the filter-feeding
zebra mussel Dreissena polymorpha in 1991 (Idrisi et al.
2001). Nutrient reductions in combination with the large
zebra mussel population have improved water clarity in
the lake, led to an expansion of submerged macrophytes,
and an increase in shallow water benthic invertebrates
(Mayer et al. 2002). In addition, the increase in
piscivorous cormorants (Phalacrocorax auritus) that
began during the 1980s led to higher mortality rates of
subadult (age-1 to age-3) walleye and yellow perch
(Rudstam et al. 2004). Not all of these changes are
important during the winter, however, as cormorants
migrate south and macrophyte coverage declines during
The effective management of ecologically or econom-
ically important fishes requires an understanding of how
natural and anthropogenic factors affect populations
(Koonce et al. 1977, Rose 2000, Fitzgerald et al. 2004).
For example, climate warming is likely to influence both
recruitment dynamics in general and winter mortality in
particular for temperate fishes (Casselman 2002, Shuter
et al. 2002). The increase of gizzard shad in Oneida Lake
particularly after 1983 represents a natural experiment
(Carpenter 1990), and provides an opportunity to study
how winter mortality patterns of age-0 yellow and white
perch are affected by the fluctuating density of an
alternate prey species. This analysis is timely, as it
considers the possible effects of a southern species like
gizzard shad that may become more abundant in Oneida
Lake due to climate warming (e.g., Vanni et al. 2005).
Given the importance of predator–prey dynamics
structuring this food web, and the role of walleye as
the main piscivore (Mills and Forney 1988, Rose et al.
1999), we expect that abundant age-0 gizzard shad
cohorts will buffer other age-0 fishes from predation
during winter. Elsewhere, walleye select gizzard shad
and other soft-rayed fish, like minnows, over spiny-
rayed prey, like yellow or white perch (e.g., Lake Erie;
Knight and Vondracek 1993).
In this paper, we use long-term fish and limnological
data from Oneida Lake to investigate the role of
biological and physical factors shaping the rates of
winter mortality in age-0 cohorts of yellow and white
perch. If the cause of winter mortality is starvation and/
or intolerance of low water temperatures, we expect
mortality rates to be negatively correlated with body
length before the winter and positively correlated with
the duration of low temperature during winter. If
starvation is an important factor in winter mortality,
we also expect that observed mortality should be
correlated with and similar to predictions from models
for length-dependent winter starvation based on dura-
tion of low temperature and age-0 length in the autumn
(yellow perch [Post and Evans 1989] and white perch
[Johnson and Evans 1991]). If the cause of winter
mortality is predation, mortality rates should be
positively correlated with the abundance of predators
and negatively correlated with the abundance of
alternate prey (age-0 fish) that could buffer predation.
Predation can also be length-dependent, as small fish are
expected to be more vulnerable to predation than large
fish. We therefore assessed the diets of walleye in
October for length selectivity just prior to winter.
Analysis of observations from just one lake allows for
the quantification of causative mechanisms without the
confounding influence of genotypic or latitudinal
factors. Use of identical methods for the entire study
simplifies the interpretation of our analyses (Beard et al.
MATERIALS AND METHODS
Age-0 fish abundance and growth
Age-0 fishes were collected weekly with a 5.5-m otter
trawl (6.4-mm stretched-mesh cod end) at 10 standard
stations in Oneida Lake from the end of July through
October from 1962–2002. Trawling at each station
August 20061489 ANALYSIS OF FISH MORTALITY OVER WINTER
consisted of a single haul; each haul lasted five minutes
and swept an area of ;0.1 ha. Trawl stations had limited
aquatic plants and favorable bottom topography. Trawl
depths varied from 4 to 12 m, and substrates shifted
from sandy mud at shallow sites to mud and detritus at
deeper sites. All age-0 fishes were sorted, counted,
weighed, and a random sub-sample (up to 75 fish) or the
whole catch from a sampling date measured for total
length to the nearest mm. The catch of age-0 fishes from
30–40 trawl hauls each October were pooled and the
mean catch per effort (CPE in fish per hectare swept)
was used as an index of autumn density (Forney 1980).
Total length of age-0 fish in October was indicative of
end of season lengths; no significant increase in length
was observed when trawling continued into November
(1962–1964, 1968; Fitzgerald 2000).
To quantify winter mortality, trawling was conducted
at the standard stations over two or three days (10 hauls
per day) between late April and early May, using the
same gear and sample processing. No spring trawling
was completed during 1968, 1979, 1980, or 1982. For
age-0 yellow perch, the instantaneous winter mortality,
(Zyp; per winter) was calculated from the natural
logarithm of the ratio between catch per ha in autumn
(CPEYPAUT) and in the spring (CPEYPSPR):
Spring trawls are ineffective for quantifying the
density of age-1 white perch cohorts (Forney 1980).
Thus, we base our index of winter mortality of age-0
white perch (Zwp) on the subsequent catch of age-2 and
age-3 fish from that cohort in the standard summer gill
net survey two and three years later (CGNtþ2 and
CGNtþ3). Gill nets of 38–102 mm stretch mesh were set
at 15 locations around Oneida Lake, one location per
week from June through the middle of September from
1964 to 2004; this strategy allows for consideration of
the 1962–2001 white perch cohorts. To be consistent
with yellow perch, this index was calculated as the
natural logarithm of the ratio of the catch of age-0 fish
per hectare in autumn trawls (CPEWPAUT) and gill net
Zwp¼ ln½CPEWPAUT=ðCGNtþ2þ CGNtþ3Þ?:
We did not include the catch of age-1 white perch
because this age group is not fully vulnerable to the
mesh sizes used in the summer gill net survey. This index
is proportional to overwinter mortality, if mortality of
age-1 and older white perch is low compared to winter
mortality or if it is constant over time. This assumption
is reasonable, given that yearling and older white perch
are rare in walleye and cormorant diets, and that this
species is not actively sought by anglers (Forney 1980,
Connelly and Brown 1991, VanDeValk et al. 2002,
Rudstam et al. 2004).
We will refer to both the observed winter mortality of
yellow perch and the winter mortality index for white
perch as winter mortality. However, it should be noted
that the white perch winter mortality value is an index,
not an instantaneous rate, as we are comparing the catch
per area from the trawl with the subsequent catch of a
cohort in the summer gill net survey over two years.
Only the relative value of the winter mortality index is
important for white perch and it should not be
compared directly with the observed winter mortality
for yellow perch.
Adult fish abundance
In this analysis, we consider adult walleye and yellow
perch to be fish age-3 and older. These fish are
considered the primary predators on age-0 fishes in
Oneida Lake (Forney 1980, Mills and Forney 1988,
Rose et al. 1999). The abundance of walleye ?age-4 and
yellow perch ?age-3 was estimated with mark–recapture
in 17 (for walleye) and 11 (for yellow perch) years during
the study period. For other years, abundance of these
fish was estimated from catches in standard gill nets and
trap nets; details are in Rudstam et al. (2004). June
sampling of these fish allowed for estimates of mass-at-
age and biomass. For this study, the biomass of age-3
walleye was estimated from the catch of age-4 walleye in
the following year, assuming a 20% annual mortality
rate. Spring biomass of walleye and yellow perch were
assumed to represent fish present during the preceding
Walleye diet analysis
To investigate the potential for length selectivity in the
walleye diet during winter, we compared prey length in
walleye diets in October with age-0 yellow and white
perch lengths from the trawls described previously.
Walleye used for diet analysis were collected between
8:00 and 10:00 a.m. with a 12.2-m otter trawl (55-mm
stretched-mesh cod end that does not retain age-0 fish).
Water temperatures in October were typically ,158C,
and low enough to permit identification and length
measurement of most prey consumed the previous night
(Forney 1977). Mean lengths of prey in walleye diets and
in the trawls in October were compared with linear
regression using all years with at least three measure-
ments of prey in the diet. We then tested if the regression
of mean lengths of prey in walleye diets and in the trawls
had a slope different from 1 and an intercept different
from 0 (t test, Sokal and Rohlf 1995). Because a large
number of both age-0 yellow and white perch were
observed in the walleye diet during 1993, length
distributions of the prey were compared with trawl-
captured lengths with a Kolmogorov-Smirnov (KS)
two-sample test (Sokal and Rohlf 1995). The total length
of walleye and the lengths of the perch species in the diet
during 1993 were also tested for correlation (Sokal and
Because gizzard shad is a schooling species and poorly
represented in the otter trawls in Oneida Lake (Roseman
et al. 1996), walleye diet data were also used to derive an
index of abundance for this species. Diet data were
DEAN G. FITZGERALD ET AL. 1490
Vol. 16, No. 4
available from October for 1971–1978 and 1981–2002.
The index was based on the number of age-0 gizzard
shad consumed per kilogram of walleye. Because age-0
gizzard shad were absent or very rare in gillnets, trawls,
and larval fish samples before 1984, they were consid-
ered absent from the diet during years 1962–1970 and
1979–1980 (Fig. 1; Mills and Forney 1988).
The winter duration was represented by the number of
days between 1 December and the calendar day of the
ice breaking up on the lake (see Plate 1). For the one
winter without complete ice cover (2001–2002), the
earliest breakup date observed during the study period
was used as a surrogate ice breakup date. For each year,
1 December was used instead of ice-in day because ice-in
dates were only available after 1975, and because water
temperatures are generally ,68C, a temperature consid-
ered stressful for fishes (Hazel 1993). Ice-in dates after
1975 have varied from 3 December to 8 January (E. L.
Mills, unpublished data). Between November and April
2000–2003, daily water temperatures were collected with
a HOBO temperature recorder (Onset Computer Cor-
poration, Bourne, Massachusetts, USA) suspended at a
depth of 10 m at a station close to the Cornell Biological
Field Station (Bridgeport, New York, USA).
Winter mortality analyses
Analyses of winter mortality in age-0 yellow and white
perch used all available observations from the 1962–
2002 cohorts. First, the effect of length in autumn on
winter mortality was investigated by testing for corre-
lation between winter mortality with mean length in
October for age-0 yellow and white perch.
Second, the winter mortality of both species was
compared with predictions of winter starvation derived
from models for yellow perch (Post and Evans 1989) and
white perch (Johnson and Evans 1991). The starvation
models estimate probability of starvation mortality for
each species, given the length distributions observed in
October and the duration of the winter in days. The
model for yellow perch is based on total days with
temperatures ,68C, whereas the models for white perch
were derived from laboratory experiments at both 2.58C
and 48C. Because the 1 December to ice-off date period
more closely tracks days with water temperatures
,2.58C in Oneida Lake, we used the 2.58C model for
white perch. If starvation is important for one or both
species in Oneida Lake, predictions from these models
should be of similar magnitude and correlated with the
winter mortality estimates.
Third, multiple linear regression analyses were used to
investigate the importance of a suite of potential
explanatory variables for winter mortality in both
species. To decrease the confounding effects of multi-
collinearity, all variables were considered a priori for
correlation. Then, only the variables believed to be most
directly related to winter mortality were retained from
pairs of variables that were highly correlated (r ? 0.800;
Graham 2003, Gotelli and Ellison 2004). For example,
adult walleye and adult yellow perch biomass in spring
were highly correlated; therefore we chose to sum the
biomass of these two predators into a total predator
biomass variable. Also, although zebra mussel density
may affect habitat structure and possibly potential
predation rates on age-0 fishes during winter, this
variable was removed because it was highly correlated
with total predator biomass. Variables retained in the
multiple regression were (1) winter duration (days), (2)
age-0 yellow perch or age 0 white perch mean length in
October (millimeters), (3) the density of age-0 yellow
perch in October (fish per hectare), (4) the density of
age-0 white perch in October (fish per hectare), (5) the
number of age-0 gizzard shad in the walleye diet (fish per
kilogram walleye), and (6) the biomass of predators
during spring (kilograms per hectare). Ecologically
relevant two-way interactions were also identified a
priori for evaluation with the multiple regressions. These
interaction terms represented starvation (winter dura-
tion 3 age-0 yellow perch length in October and winter
duration 3 age-0 white perch length in October) and
predation (total predator biomass 3 age-0 yellow perch
density in October and total predator biomass 3 age-0
white perch density in October and total predator
biomass 3 age-0 gizzard shad number in walleye diet).
All analyses were done with the GLM procedure of JMP
4 (SAS Institute 2001). Prior to these analyses, the
estimates of age-0 yellow perch and white perch density
in October were ln(x þ 1)-transformed and the gizzard
shad number in the walleye diets were ln(x þ 0.1)-
transformed to stabilize variance. Because the variables
assessed with the regressions are measured with different
units, we also present standardized beta coefficients
since the magnitude is proportional to the strength of
the effect, given the range of the independent variable in
question. The residuals from the regressions were
graphically inspected for deviations from normality
(Graham 2003, Gotelli and Ellison 2004). Significance
levels for all tests were set a priori at a ¼ 0.05 and were
not adjusted for multiple comparisons, following the
recent arguments by Moran (2001) and Gotelli and
During 1962–2002, winter ice breakup occurred, in
three years, as early as day 76 (counting from 1 January)
and as late as day 110, with a mean of day 93 (SD¼11 d;
Table 1). In the winter of 2001–2002, the ice cover was
not complete, so we used day 76 as the breakup date.
This was the first year without complete ice cover since
records were started during the winter of 1845–1846
(Magnuson et al. 2000). Available winter water temper-
atures showed the water temperatures observed on 1
December were 3.68, 7.88, and 3.78C in 2000, 2001, and
2002, respectively (Fig. 2). Temperature was ?2.58C on
August 20061491 ANALYSIS OF FISH MORTALITY OVER WINTER
121 days in 2000–2001, on 92 days in 2001–2002, and on
110 days in 2002–2003. These temperatures correlate
with our winter duration index (134, 107, and 136 days
for the winters of 2000–2001, 2001–2002, and 2002–
2003, respectively). These observations reveal the lake
experiences extended periods of water temperatures
?2.58C during winter, independent of ice cover (Fig. 2).
Predator biomass and walleye diets
Predator biomass during 1962–2002 ranged from 14.3
to 78.0 kg/ha (Table 1). Diet of walleye (mean number
of walleye examined ¼ 132, range ¼ 64–194 fish) in
October included age-0 yellow perch, white perch,
gizzard shad, and other species (Table 1, Fig. 3). During
1962–1983, the walleye diet had a mean of ,0.001
gizzard shad/kg walleye (range¼0–0.1, SD¼0.002), and
during 1984–2002 had a mean of 1.2 gizzard shad/kg
walleye (range¼0–5.4, SD¼1.7; Fig. 3). Thus, although
gizzard shad was more abundant since 1984, it was also
variable (Fig. 3). Since 1984, gizzard shad was the
dominant diet item during 10 years (1984, 1987, 1989–
1992, 1994, 2000–2002), yellow perch during seven years
(1985–1986, 1988, 1993, 1995–1996, and 1998), emerald
shiner (Notropis atherinoides) during one year (1999),
and white perch during one year (1997).
The analysis of the diet of walleye caught in October
1962–2002 revealed no length selectivity for yellow
perch, but some selectivity for smaller white perch.
TABLE 1. Physical and biological observations from Oneida Lake between 1962 and 2002.
Age-0 yellow perchAge-0 white perch
Index§ AutumnSpring AutumnSpring
73.5 (9.7, 50)
62.2 (6.7, 50)
70.4 (6.9, 75)
60.9 (5.0, 50)
70.7 (7.6, 50)
72.9 (8.2, 75)
70.1 (8.0, 96)
63.2 (8.1, 119)
76.3 (10.5, 220)
58.4 (6.3, 99)
69.2 (7.8, 76)
84.7 (11.4, 203)
70.7 (7.0, 50)
63.2 (7.7, 73)
72.3 (9.6, 166)
71.7 (7.0, 146)
73.0 (10.1, 67)
75.4 (8.3, 70)
81.0 (8.8, 103)
56.5 (5.1, 125)
62.6 (8.6, 150)
82.7 (91, 94)
70.9 (5.8, 75)
73.9 (10.1, 84)
81.3 (10.4, 75)
78.6 (8.6, 75)
81.9 (10.2, 79)
80.7 (6.8, 56)
70.8 (9.0, 75)
80.1 (9.6, 75)
73.8 (9.8, 66)
84.8 (9.2, 64)
82.5 (10.7, 75)
92.0 (8.9, 75)
80.0 (12.2, 55)
79.3 (6.2, 55)
83.5 (7.4, 75)
82.5 (10.0, 75)
74.7 (9.0, 75)
85.6 (9.1, 75)
75.8 (8.1, 75)
94.1 (7.3, 99)
67.2 (10.6, 107)
79.2 (14.5, 86)
76.1 (6.4, 100)
77.8 (8.4, 100)
84.2 (7.8, 75)
85.4 (12.5, 47)
81.1 (9.8, 88)
79.0 (12.4, 100)
79.4 (7.1, 100)
84.2 (8.4, 34)
87.9 (9, 75)
69.6 (10.6, 160)
86.8 (7.5, 75)
64.2 (9.8, 75)
76.8 (13.1, 125)
85.2 (8.7, 57)
78.2 (10.9, 134)
81.7 (8.3, 165)
76.4 (7.4, 61)
74.3 (6.9, 75)
83.3 (8.9, 75)
71.5 (7.14, 75)
68.2 (8.5, 75)
72.3 (15.7, 38)
63.2 (6.2, 75)
83.1 (12.8, 22)
78.1 (8.2, 75)
69.8 (8.5, 75)
90.8 (11, 75)
70.5 (15.1, 68)
76.5 (8.4, 75)
80.1 (11.5, 75)
97.2 (6.5, 75)
85.9 (12.8, 65)
77.7 (7.7, 75)
86.6 (7.8, 75)
97.0 (10.6, 20)
78.2 (10.4, 75)
87.9 (9.5, 75)
82.9 (7.6, 75)
Notes: ND indicates ‘‘not determined.’’
? Predators are walleye and yellow perch ?age-3.
? Measurements are, respectively, mean (SD; n).
§ Index of recruitment for age-0 white perch based on subsequent catch of age-2 and age-3 fish in summer gill net survey.
DEAN G. FITZGERALD ET AL.1492
Vol. 16, No. 4
The length of yellow perch (mean¼74.1 mm, range 63–
92.5 mm) in the diet of walleye was positively correlated
(P ¼ 0.005, N ¼ 13, r2¼ 0.52) with the length (mean ¼
75.8 mm, range ¼ 65.4–85.6 mm) in the trawl (Fig. 4).
This regression had a slope coefficient of 0.82 6 0.47
(mean 6 2 SE) that was not significantly different from 1
(t¼0.74, P¼0.434), and an intercept coefficient¼13 6
36 (mean 6 2 SE) that was not different from 0 (t¼0.72,
P ¼ 0.484). Comparison of the 1993 lengths of yellow
perch in the walleye diet with the sample from the trawl
revealed no significant difference (KS two-sample test, P
¼ 0.489), and the lengths in the diet were not correlated
(P . 0.500) with walleye length (Fig. 5).
For white perch, the length (mean ¼ 75.2, range ¼
61.7–84.4 mm) in the diet of walleye caught in October
1962–2002 was not correlated (P ¼ 0.264, N ¼ 13, r2¼
0.11) with the length of fish (mean¼75.9, range¼63.2–
87.9 mm) in the trawl (Fig. 4). However, the slope of this
relationship was significantly less than 1 (coefficient ¼
0.24 6 0.47, t¼3.70, P¼0.005), indicating that walleye
selected smaller white perch when the mean length was
large (Fig. 4). Comparison of the 1993 lengths of white
perch in the walleye diet with lengths in the trawl catch
April for a winter with no complete ice cover (2001–2002), and for a winter with complete ice cover (2002–2003).
Oneida Lake water temperatures observed at a sampling station at a 10-m water depth between 1 November and 30
FIG. 3. Prey fish observed in walleye diet in October during 1984–2002.
August 20061493 ANALYSIS OF FISH MORTALITY OVER WINTER
revealed no difference (KS two-sample test, P ¼ 0.413),
and the white perch lengths were not correlated (P .
0.500) with walleye length (Fig. 5).
Growth, density, and winter mortality of age-0 yellow
perch and white perch
Large interannual differences were observed in the
mean length and density of age-0 yellow and white perch
in October during 1962–2002 (Table 1). Age-0 yellow
perch density for the entire period ranged from 32–6611
fish/ha (Table 1). The mean density of age-0 yellow
perch during 1962–1983 was 1836 fish/ha (range ¼ 203-
6611, SD¼1913 fish/ha) that was significantly larger (t¼
2.53, P¼0.016) than the mean density during 1984–2002
of 472 (range ¼ 32–3289, SD ¼ 811 fish/ha). Age-0 white
perch density for the entire period ranged from 6–2476
fish/ha (Table 1). The mean density of age-0 white perch
during 1962–1983 was 554 fish/ha (range¼16–2476, SD¼
792 fish/ha) that was not different (t ¼ 1.34, P ¼ 0.187)
from the mean density during 1984–2002 of 374 6 373
fish/ha with a range of 6–1545 fish/ha).
Yellow perch were collected during all spring sample
periods and winter mortality was estimated as an
instantaneous rate (per winter) from the observed
change in density between autumn and spring (Table
1). Age-0 yellow perch had a mean winter mortality rate
during 1962–1983 of 2.91/winter (range¼0.17–4.79, SD¼
1.29) that was significantly larger (t ¼ 3.28, P ¼ 0.002)
than the mean during 1984–2002 of 1.40 (range: ?0.73–
4.38, SD ¼ 1.51). White perch cohorts either experienced
high overwinter mortality or their vulnerability to trawls
decreased between autumn and spring. White perch were
collected during only 13 of 36 spring periods whereas all
cohorts except one contributed ?1 age-2 or age-3 fish to
the summer gill net survey (range ¼ 0–551; Table 1). To
include all cohorts in subsequent analyses, one fish was
added to all catches of age-2 and age-3 from each cohort
(i.e., x þ1). The white perch winter mortality index had
a mean during 1962–1983 of 1.83 (range: ?1.10–4.34,
SD ¼ 1.63) that was not different (t ¼ 1.16, P ¼ 0.252)
from the mean for 1984–2001 of 1.19 (range:?1.20–4.45,
SD ¼ 1.84).
Length-dependent winter mortality
Winter mortality was negatively correlated with mean
length in October for both yellow perch and white perch
during 1962–2002 (Fig. 6; for yellow perch, P , 0.001,
N¼37, r2¼0.39; for white perch, P , 0.001, N¼40, r2¼
0.26). For yellow perch, this relationship was stronger
during 1962–1983 (P ¼ 0.003, N ¼18, r2¼ 0.44)
compared with 1984–2002 (P ¼ 0.211, N ¼ 19, r2¼
0.09) but there was no significant difference in slope
between the two time periods (ANCOVA, P ¼ 0.69).
This pattern may be partly due to smaller fish present in
the early years, although there was a substantial overlap
of mean lengths observed during the entire period (Fig.
6). For white perch, the relationship was stronger in the
later years (1984–2002, P¼0.003, N¼18, r2¼0.44) than
in the earlier years (1962–1983, P ¼ 0.111, N ¼ 22, r2¼
0.12); again without a difference in slope between the
two time periods (ANCOVA, P ¼ 0.46). For white
perch, the observed mean lengths, like yellow perch,
overlapped between the two time periods (Fig. 6).
Comparisons with starvation models
Winter mortality was positively correlated with
predicted starvation for both yellow and white perch
during 1962–2002 (Fig. 7; for yellow perch, P , 0.001,
N¼37, r2¼0.28; for white perch, P¼0.035, N¼40, r2¼
0.11). For yellow perch, the predicted starvation was
mostly lower than the observed winter mortality, and
often substantially lower as indicated by the distribution
of data relative to the 1:1 line (Fig. 7). Because of the
index used for white perch, we directly compare the
absolute value of winter mortality index and predicted
starvation. Like the length–mortality relationships, the
correlations observed between winter mortality and
predicted starvation were not significant for both time
line for yellow perch is a linear regression (y¼17.1þ0.78x); no significant relationship was observed for white perch. A dotted 1:1
line and the 1993 observations are included for reference.
Mean lengths (6SE) of age-0 yellow perch and white perch in the trawl and walleye diet samples in October. The fitted
DEAN G. FITZGERALD ET AL.1494
Vol. 16, No. 4
periods for either species (Fig. 7). For yellow perch
during 1962–1983, the starvation predictions had a mean
of 0.07/winter (range ¼ 0.007–0.295, SE ¼ 0.01) that was
positively correlated (P ¼ 0.010, r2¼ 0.34) with winter
mortality. During 1984–2002, the starvation predictions
had a mean of 0.015 (range¼0.004–0.039, SE¼0.23) that
was no longer correlated (P . 0.500) with winter
mortality. For white perch during 1962–1983, the
starvation predictions had a mean of 0.40 (range ¼
0.16–0.73, SE¼0.03) that were not correlated (P . 0.500,
r2, 0.01) with winter mortality (Fig. 7). During 1984–
2002, the starvation predictions had a mean of 0.38
(range ¼ 0.11–0.74, SE ¼ 0.04) that were correlated (P ¼
0.003, r2¼ 0.44) with winter mortality.
Multiple regression analysis
The multiple regression models of winter mortality for
both age-0 yellow and white perch were highly
significant (for yellow perch, F6,30¼ 8.86, P , 0.001,
r2¼ 0.64, Table 2; for white perch, F6,32¼ 6.79, P ,
0.001, r2¼ 0.56, Table 3). Winter mortality for yellow
perch was negatively correlated with yellow perch length
in October (P , 0.001), number of gizzard shad in
walleye diets (P ¼ 0.009), and white perch density in
October (P ¼ 0.027); contrary to our hypothesis, winter
mortality was also negatively correlated with winter
duration (P ¼ 0.017, Table 2). In addition, density of
age-0 yellow perch and predator biomass was not
correlated with yellow perch winter mortality. For white
perch, winter mortality was negatively correlated with
white perch length in October (P ¼ 0.050) and yellow
perch density in October (P ¼ 0.041), and positively
correlated with predator biomass (P ¼ 0.005). Contrary
to our hypotheses, white perch mortality was positively
correlated with white perch density (P ¼ 0.002). Also,
winter duration and gizzard shad in the walleye diet
were not correlated with winter mortality (Table 3).
These analyses identified no significant interaction terms
for either species (Tables 2 and 3).
The existence of long-term observations on the
limnology and fishes of Oneida Lake provided an
opportunity to investigate the winter mortality dynamics
of age-0 yellow and white perch with the fall and rise of
nonnative gizzard shad. For both yellow and white
perch, winter mortality was negatively correlated with
mean length in autumn. These relationships corre-
sponded to our hypothesis of length-dependent mortal-
ity structuring the age-0 cohorts overwinter, and are
concordant with findings from past studies of these and
other temperate fishes (Post and Evans 1989, Johnson
and Evans 1991, Sogard 1997, Craig 2000). However,
length-dependent winter mortality can be caused by
either predation or starvation (Ricker 1969, Wootton
1998). To evaluate the importance of these two
mechanisms, we correlated winter mortality with pre-
dictions from models of starvation for both yellow and
white perch (Post and Evans 1989, Johnson and Evans
1991), assessed the diet of walleye for evidence of length-
dependent predation, and used multiple regression
analyses to assess the explanatory role of density of
prey species, fish body length in autumn, total predator
biomass, and winter duration on winter mortality. If
predation was important, we expected winter mortality
to be positively correlated with predator biomass and
negatively correlated with abundance of alternate prey.
sampled by trawl or in walleye diet, and walleye lengths in
October of 1993. The upper panel shows the length distribu-
tions for yellow perch from the trawl and from the walleye diet;
the middle panel shows the length distributions of white perch
from the trawl and from the walleye diet. Distributions showed
no differences based on Kolmogorov-Smirnov two-sample tests
(P . 0.400). All fish lengths were sorted to 4 mm length bins
prior to plotting the distributions. The lower panel shows the
lengths of walleye and age-0 yellow and white perch considered
in the diet analysis for 1993.
Lengths of age-0 yellow perch and white perch
August 20061495 ANALYSIS OF FISH MORTALITY OVER WINTER
If starvation was important, we expected mortality to be
positively correlated with winter duration.
For yellow perch, our interpretation of these analyses
is that starvation is not important for winter mortality in
Oneida Lake. Although the predictions from the Post
and Evans (1989) starvation model were correlated with
observed mortality, the predicted starvation mortality
was relatively low (maximum of 23%), even for the
cohorts with the smallest average length, whereas
observed mortality often exceeded 95%. Part of this
difference may be due to the Post and Evans (1989)
model, as it is not linear and winter mortality is low for
yellow perch larger than ;70 mm. Also, some difference
could be due to our winter index underestimating the
number of days when the temperature was less than the
68C used in the model. This approach underestimates
the number of days ,68C, and so the model prediction
represents a conservative estimate for yellow perch.
However, an increase of winter duration with one month
to account for this difference only increases the model
maximum mortality to 35%. Further, the starvation
model was derived for fish deprived of food, and so the
model is likely to overestimate risk of starvation in the
lake. Young yellow perch can consume invertebrates in
Oneida Lake between autumn and spring, and likely
feed at this time although no measurable growth has
been documented. Studies of the congener Eurasian
perch (Perca fluviatilis) at age-0 showed that fish of all
lengths fed through winter in different lakes and that
starvation was rare (Radke and Eckmann 1999,
Eckmann 2004). In addition, the multiple regression
showed that winter duration was not positively corre-
during 1962–2002. For yellow perch, the winter mortality was estimated from the change in density between autumn and spring.
For white perch, the winter mortality was estimated as an index derived from the change in density between autumn and the
subsequent catch of a cohort in the future. Refer to Methods for additional details. Fitted lines are linear regressions for yellow
perch (y ¼?0.12x þ 11.37) and white perch (y ¼?0.11x þ 10.19).
Analysis of winter mortality (per winter) in cohorts of age-0 yellow and white perch relative to mean length in October
(Post and Evans 1989) and white perch (Johnson and Evans 1991) during 1962–2002. The mortality estimates (per winter) for each
species (and cohort) are given as an instantaneous rate for each winter period. Refer to Methods for additional details. The fitted
lines are linear regressions for yellow perch (y¼15.01xþ1.50) and white perch (y¼3.40xþ0.21); a 1:1 line (dotted) is included for
yellow perch. Mortality is given as instantaneous rates per winter (see Methods).
Analysis of the relationship between winter mortality and predicted winter starvation in cohorts of age-0 yellow perch
DEAN G. FITZGERALD ET AL.1496
Vol. 16, No. 4
lated with winter mortality (it was significantly neg-
atively correlated), and there was no significant inter-
action between perch length and winter duration.
The available evidence for yellow perch suggests
predation is a more likely cause of winter mortality
then starvation in Oneida Lake. The inference of winter
predation is consistent with the multiple regression
analysis that showed lower mortality at large length and
at high density of alternate prey species, like age-0
gizzard shad and age-0 white perch, before winter. This
suggests winter predation on any abundant prey species,
and is also supported by the variable diet of walleye
observed during 1984–2002, independent of prey behav-
ior, as gizzard shad and emerald shiner are pelagic
schooling species compared with the benthic-oriented
yellow and white perch. Predator biomass was pos-
itively, but not significantly, correlated with mortality (P
¼ 0.24, Table 2). Previous studies have indicated that
walleye are the primary predator on age-0 yellow perch
in Oneida Lake up to autumn, and walleye continue to
feed on age-1 yellow perch through the spring and early
summer (Forney 1974, 1977, Nielsen 1980); walleye
probably also feed on young yellow perch through the
winter. However, we could not detect any length-
dependent predation on age-0 yellow perch by age-4
and older walleye in the autumn. In retrospect, this was
not unexpected, as little length selection of age-0 yellow
perch has been observed in Oneida Lake until perch
reach age-1 the following summer (Nielsen 1980). Also,
in Lake Erie, adult walleye in the same length range as
those in Oneida Lake were not length-selective on yellow
perch (Knight et al. 1984). Thus, although adult walleye
are likely a major source of winter mortality for age-0
yellow perch, they are probably not the cause of the
observed length-dependent mortality.
Other predators that are active during winter in
Oneida Lake may be more selective of small age-0
yellow perch. These predators include younger age
groups of walleye (Forney 1965), burbot (Lota lota;
Fratt et al. 1997; A. J. VanDeValk, personal observa-
tion), and adult yellow perch. Younger walleye are more
likely to be selective towards the smaller age-0 perch (as
observed in Lake Erie; Knight et al. 1984). Adult yellow
perch are also likely to be more length-selective (Post
and Rudstam 1992), and yellow perch are piscivorous in
the winter in Oneida Lake (L. Rudstam and T.
Brooking, unpublished data). In other lakes, adult yellow
perch and Eurasian perch frequently feed on small fishes
during winter, including young yellow perch (Kelso and
Ward 1977, Sanderson et al. 1999, Claessen et al. 2002,
Jacobsen et al. 2002).
Collectively, it is more likely that winter mortality of
yellow perch is regulated by predation, and that the
length dependence is the result of smaller fish taking
higher risks to feed during the winter (Walters 2000),
starving fish being more vulnerable to predators (Jonas
and Wahl 1998), or length-dependent predation by
predators other than adult walleye. Since growth of age-
0 yellow perch is density dependent in Oneida Lake,
length-dependent winter mortality is a compensatory
process and a likely explanation for the compensatory
mortality between age-0 in the autumn and age-2 yellow
perch described by Nielsen (1980). This mechanism
helps explain why the density of age-1 yellow perch has
not declined in the 1990s even though the density of age-
0 yellow perch has (Rudstam et al. 2004).
For white perch, the patterns of winter mortality (as
indicated by the mortality index) were more difficult to
interpret. Our observations of a negative correlation
between length in autumn and winter mortality and the
positive correlation between observed and predicted
starvation mortality from the Johnson and Evans (1991)
model indicate the possible presence of high starvation
rates. Johnson and Evans (1991) showed that during
simulated winters, the mortality of age-0 white perch
was higher than that observed for age-0 yellow perch,
and could be considerable for small white perch at 2.58C
(;50%). This high mortality was attributed to sensitivity
to low temperatures and higher activity of white perch
compared with yellow perch. Similarly, in the Bay of
Quinte, Canada, on Lake Ontario, the long winters of
1976 and 1977 led to catastrophic winter mortality of all
winter mortality as dependent variable, for 1962–2002
Multiple regression analysis with age-0 yellow perch
Notes: Independent variables considered included winter
duration (WINTER), age-0 yellow perch mean length in
October (YPOL), age-0 yellow perch density in October
(YPOD), age-0 white perch density in October (WPOD), total
predator biomass (PRBI), and the number of gizzard shad in
the diet per kg walleye (GSWA).
? Standardized partial regression coefficients, given to allow
for comparisons of the strength of the relationships among
independent variables measured with different units.
winter mortality index as dependent variable, 1962–2001
Multiple regression analysis with age-0 white perch
Note: Independent variables considered include those iden-
tified in Table 2, and age-0 white perch mean length in October
? Standardized partial regression coefficient, given to allow
for comparisons of the strength of the relationships among
independent variables measured with different units.
August 20061497 ANALYSIS OF FISH MORTALITY OVER WINTER
ages of white perch (Casselman and Scott 2003), and the
frequent long winters in Lake Superior limit the invasive
white perch to a very small population (Bronte et al.
2003). However, the multiple regression analysis for
white perch from Oneida Lake showed no effect of
The multiple regression analysis did also indicate an
importance for predation mortality over winter for age-0
white perch. Winter mortality was minimized with large
length in autumn, particularly during the recent years,
when age-0 yellow perch cohorts were present at high
density to presumably buffer white perch, and when
total predator biomass was low. Also, walleye show
selection towards smaller white perch when the average
length is large, and this trend is consistent with
predation by walleye as a cause for the observed
length-dependent winter mortality. However, the multi-
ple regression analysis identified a positive correlation
between white perch density and winter mortality, a
trend that is not consistent with a predation hypothesis
because white perch does not show density-dependent
growth in Oneida Lake (Table 1; Prout et al. 1990).
Thus, we were unable to separate the importance of
starvation and predation for overwinter mortality. The
use of an index based on the gill net survey instead of
direct estimates of age-1 density in the spring likely
contributed to our difficulties in determining causes for
winter mortality for this species.
Our analyses do indicate that age-0 white perch are
more tolerant of low temperatures than suggested from
past studies (e.g., Johnson and Evans 1991). First, all
but one white perch cohort led to fish in the subsequent
summer gill net surveys compared with zero winter
survival in age-0 gizzard shad during 1984–2002.
Second, the temperatures observed at a 10 m depth for
the 2000–2002 winters and recruitment index suggest
that age-0 white perch can tolerate long periods of low
temperatures in Oneida Lake. For example, age-0 white
perch survived well in the winter of 2002–2003 with 37
days with temperatures ,18C and also in the winter of
2001–2002 with no complete ice cover but with 67 days
with temperatures ,18C (Fig. 2). Third, age-1 white
perch occur in the diet of walleye in the spring (e.g.,
Forney 1974). Finally, white perch have maintained a
population continuously since they invaded during the
1950s. Low overwinter survival of white perch in the
laboratory could be diet related. Both the related striped
bass (Morone saxatilis) and white bass (Morone chrys-
ops) were able to tolerate low-temperature shock treat-
ments when fed a diet high in unsaturated fatty acids
(Kelly and Kohler 1999).
Age-0 gizzard shad have been abundant in Oneida
Lake in many years since 1984 (Roseman et al. 1996,
Hall and Rudstam 1999). In contrast to the negative
effects of gizzard shad on other fishes like sunfish
(Lepomis sp.) documented elsewhere (Adams et al. 1982,
View of ice mounds on the south shore of Oneida Lake, just after the ice breakup, during early April. Photo credit: Per
DEAN G. FITZGERALD ET AL.1498
Vol. 16, No. 4
Stein et al. 1995, Vanni et al. 2005; but see Jackson and
Noble 2000), the effect of gizzard shad on yellow perch
in Oneida Lake is largely positive. This inference reflects
minimal direct competition between age-0 gizzard shad
and yellow perch (Roseman et al. 1996), and gizzard
shad acting as a buffer for winter predation. The role of
a buffer species occurs even though age-0 gizzard shad in
Oneida Lake are not known to survive the winter, and,
when abundant, can be seen dead under the ice (J. L.
Forney, personal observation). Such length-independent
winter mortality is likely due to physiological stress of
cold temperatures on this southern species (Bodola
1965, Adams et al. 1982, Vanni et al. 2005). Even if
gizzard shad die relatively early in the winter period,
they would still buffer predation by walleye on age-0
yellow perch during late autumn. Thus, the rise of
gizzard shad in the lake has contributed to a decrease in
the winter mortality of yellow perch. This buffering,
together with increased growth rates in recent decades,
has compensated for increased summer mortality of age-
0 yellow perch in the 1990s and resulted in similar
density of age-1 yellow perch in the spring in years prior
to and after 1990 (Rudstam et al. 2004).
Fisheries management is forced to adapt to changing
ecosystems associated with factors like species invasions
and climate warming, and management decisions may
depend on how recruitment rates of key fishes respond
to these changes (Rose 2000, Magnuson 2002, Shuter et
al. 2002, Casselman and Scott 2003). In this study, the
application of models that predicted winter starvation of
age-0 cohorts helped us interpret observed correlations
between various independent variables and overwinter
mortality. Our analysis shows that winter mortality is
length-dependent for both yellow and white perch, and
that gizzard shad, a species that may become more
abundant with climate warming, act to buffer winter
mortality of yellow perch (although we did not detect
such a response for age-0 white perch). By extension,
this finding also indicates the larger lengths of age-0
yellow perch by autumn in recent decades, previously
attributed to a higher density of invertebrates following
the zebra mussel invasion (Mayer et al. 2000, 2002), only
explain a portion of the lower winter mortality during
these recent years. Shorter winters associated with
climate warming may also contribute to decreased
winter mortality of white perch, because starvation
may be important for this species.
The winter mortality dynamics of age-0 yellow and
white perch were shaped by density-dependent preda-
tor–prey interactions during low temperatures, and this
confirms that winter is not a quiescent period. Ecological
interactions during winter are known to shape fish
recruitment on both ecological and evolutionary time
scales (Campbell et al. 2005). Conover et al. (2005)
showed how length-dependent winter mortality can lead
to rapid changes in recruitment rates, and how
information on such processes is required for under-
standing the long-term dynamics of temperate fishes.
For Oneida Lake, our analyses indicate that predation is
the main mechanism shaping winter mortality of yellow
perch, while both predation and starvation are likely
important for white perch. Of course, starvation may
lead to an increased risk of predation. This study
revealed that nonnative fishes with fluctuating density
can modify the recruitment rates of a native fish due to
interactions in winter. It is also clear from this analysis
that early-stage mortality rates (e.g., egg, larvae,
juvenile; Koonce et al. 1977) do not set age-1 cohort
density for either perch species. That large mean length
in autumn alone does not consistently predict winter
mortality rates in age-0 yellow or white perch identifies
that the strength of the length-dependent mortality
relationships must vary on an interannual basis. The use
of identical sampling methods over a long-term period
provided the opportunity to identify and interpret these
species interactions. It is clear that winter duration
determines the northern limit of fish distributions (e.g.,
Shuter and Post 1990), but for the mid-latitude Oneida
Lake and the fish species described here, predator–prey
interactions seem to exert a greater influence on winter
mortality than starvation.
This study is one result of the long-term collaboration
between the Cornell Warmwater Fisheries Unit (CWFU) and
the New York Department of Environmental Conservation.
Recent interactions and analyses for this study also involved
individuals from the University of Windsor, Canada. Primary
funding to the CWFU was provided by New York Federal Aid
Project FA-48R. Additional funding was provided through a
Natural Sciences and Engineering Research Council (NSERC)
grant to P. Sale (#154284), NSERC Scholarship and Fellowship
to D. G. Fitzgerald, and a Great Lakes Fisheries Commission
grant to E. Mills and S. Millard. We also thank the many
undergraduate and graduate students who contributed to
different aspects of the data set. Edward Mills and Thomas
Brooking provided thoughtful comments and assistance in the
field; Edward Mills also provided access to his Oneida Lake
limnological data. The analyses, content, and presentation
format benefited from comments provided by anonymous
feedback and reviews. This is contribution number 230 from
the Cornell Biological Field Station.
Adams, S. M., R. B. McLean, and M. M. Huffman. 1982.
Structuring of a predator population through temperature-
mediated effects on prey availability. Canadian Journal of
Fisheries and Aquatic Sciences 39:1175–1184.
Aslop, R. G., and J. L. Forney. 1962. Growth and food of white
perch in Oneida Lake. New York Fish and Game Journal 9:
Beard, G. R., W. A. Scott, and J. K. Adamson. 1999. The value
of consistent methodology in long-term environmental
assessment. Environmental Monitoring and Assessment 54:
Bodola, A. 1965. Life history of the gizzard shad, Dorosoma
cepedianum (LeSueur), in western Lake Erie. U.S. Fish and
Wildlife Service Fisheries Bulletin 65:391–425.
Bronte, C. R., M. P. Ebener, D. R. Schreiner, D. S. DeVault,
M. M. Petzold, D. A. Jensen, C. Richards, and S. J. Lozano.
2003. Fish community change in Lake Superior, 1970–2000.
Canadian Journal of Fisheries and Aquatic Sciences 60:1552–
August 2006 1499ANALYSIS OF FISH MORTALITY OVER WINTER
Campbell, J. L., M. J. Mitchell, P. M. Groffman, L. M.
Christenson, and J. P. Hardy. 2005. Winter in northeastern
North America: a critical period for ecological processes.
Frontiers in Ecology and Evolution 3:314–322.
Carpenter, S. R. 1990. Large-scale perturbations: opportunities
for innovation. Ecology 71:2038–2043.
Casselman, J. M. 2002. Effects of temperature, global extremes,
and climate change on year-class production of warmwater,
coolwater, and coldwater fishes in the Great Lakes basin.
Pages 39–60 in N. A. McGinn, editor. Fisheries in a changing
climate. American Fisheries Society Symposium 32. Ameri-
can Fisheries Society, Bethesda, Marlyand, USA.
Casselman, J. M., and K. A. Scott. 2003. Fish community
dynamics of Lake Ontario: long-term trends in the fish
populations of eastern Lake Ontario and the Bay of Quinte.
Pages 349–384 in M. Munawar, editor. State of Lake
Ontario, past, present, future. Backhuys Publishers, Leiden,
Claessen, D., C. Van Oss, A. M. De Roos, and L. Persson.
2002. The impact of size-dependent predation on population
dynamics and individual life history. Ecology 83:1660–1675.
Connelly, N. A., and T. L. Brown. 1991. Net economic value of
the freshwater recreational fisheries of New York. Trans-
actions of the American Fisheries Society 120:770–775.
Conover, D. O., S. A. Arnott, M. R. Walsh, and S. B. Munch.
2005. Darwinian fishery science: lessons from Atlantic
silverside (Menidia menidia). Canadian Journal of Fisheries
and Aquatic Sciences 62:730–737.
Conover, D. O., and E. T. Schultz. 1995. Phenotypic similarity
and the evolutionary significance of countergradient varia-
tion. Trends in Ecology and Evolution 10:248–252.
Craig, J. 2000. Percid fishes systematics, ecology and exploita-
tion. Blackwell Science, Oxford, UK.
Cunjak, R. A. 1988. Physiological consequences of over-
wintering in streams: the cost of acclimitization? Canadian
Journal of Fisheries and Aquatic Sciences 45:443–452.
Eckmann, R. 2004. Overwinter changes in mass and lipid
content of Perca fluviatilis and Gymnocephalus cernuus.
Journal of Fish Biology 65:1498–1511.
Eddy, F. B. 1981. Effects of stress on osmotic and ionic
regulation in fish. Pages 77–102 in A. D. Pickering, editor.
Stress and fish. Academic Press, New York, New York,
Fitzgerald, D. G. 2000. Comparative hatching, growth and
overwinter survival of age-0 yellow perch (Perca flavescens
Mitchill) in temperate lakes of the Great Lake Basin.
Dissertation. University of Windsor, Windsor, Ontario,
Fitzgerald, D. G., D. F. Clapp, and B. J. Belonger. 2004.
Characterization of growth and winter survival of age-0
yellow perch in southeastern Lake Michigan. Journal of
Great Lakes Research 30:227–240.
Forney, J. L. 1965. Factors affecting first-year growth of
walleyes in Oneida Lake, New York. New York Fish and
Game Journal 13:146–167.
Forney, J. L. 1971. Development of dominant year classes in a
yellow perch population. Transactions of the American
Fisheries Society 101:739–749.
Forney, J. L. 1974. Interactions between yellow perch
abundance, walleye predation, and survival of alternate prey
in Oneida Lake, New York. Transactions of the American
Fisheries Society 103:15–24.
Forney, J. L. 1977. Reconstruction of yellow perch (Perca
flavescens) cohorts from examination of walleye (Stizostedion
vitreum vitreum) stomachs. Journal of the Fisheries Research
Board of Canada 34:925–932.
Forney, J. L. 1980. Evolution of a management strategy for the
walleye in Oneida Lake, New York. New York Fish and
Game Journal 27:105–141.
Fratt, T. W., D. W. Coble, F. Copes, and R. E. Bruesewitz.
1997. Diet of burbot in Green Bay and western Lake
Michigan with comparison to other waters. Journal of Great
Lakes Research 23:1–10.
Gotelli, N. J., and A. M. Ellison. 2004. A primer of ecological
statistics. Sinauer Associates, Sunderland, Massachusetts,
Graham, M. H. 2003. Confrounting multicollinearity in
ecological multiple regression. Ecology 84:2809–2815.
Hall, S. R., and L. G. Rudstam. 1999. Habitat use and
recruitment: a comparison of long-term recruitment patterns
among fish species in a shallow eutrophic lake, Oneida Lake,
NY, USA. Hydrobiologia 409:101–113.
Hazel, J. R. 1993. Thermal biology. Pages 427–467 in D. E.
Evans, editor. The physiology of fishes. CRC Press, Boca
Raton, Florida, USA.
Holt, R. D., and J. H. Lawton. 1994. The ecological
consequences of shared natural enemies. Annual Review of
Ecology and Systematics 25:495–520.
Idrisi, N., E. L. Mills, L. G. Rudstam, and D. J. Stewart. 2001.
Impact of zebra mussels, Dreissena polymorpha, on the
pelagic lower trophic levels of Oneida Lake, New York.
Canadian Journal of Fisheries and Aquatic Sciences 58:1430–
Jackson, J. R., and R. L. Noble. 2000. Relationships between
annual variations in reservoir conditions and age-0 large-
mouth bass year-class strength. Transactions of the American
Fisheries Society 129:699–715.
Jackson, R. B., S. R. Carpenter, C. N. Dahm, D. M.
McKnight, R. J. Naiman, S. L. Postel, and S. W. Running.
2001. Water in a changing world. Ecological Applications 11:
Johnson, T. B., and D. O. Evans. 1991. Behaviour, energetics,
and associated mortality of young-of-the-year white perch
(Morone americana) and yellow perch (Perca flavescens)
under simulated winter conditions. Canadian Journal of
Fisheries and Aquatic Sciences 48:672–680.
Jonas, J. L., and D. H. Wahl. 1998. Relative importance of
direct and indirect effects of starvation for young walleyes.
Transactions of the American Fisheries Society 127:192–205.
Kelly, A., and C. C. Kohler. 1999. Cold tolerance and fatty acid
composition in striped bass, white bass, and their hybrids.
North American Journal of Aquaculture 61:278–285.
Kelso, J. R., and F. J. Ward. 1977. Unexploited percid
populations of West Blue Lake, Manitoba, and their
interactions. Journal of the Fisheries Research Board of
Knight, R. L., F. J. Margraf, and R. F. Carline. 1984. Piscivory
by walleyes and yellow perch in western Lake Erie. Trans-
actions of the American Fisheries Society 113:677–693.
Knight, R. L., and B. Vondracek. 1993. Changes in prey fish
populations in western Lake Erie, 1969–88, as related to
walleye, Stizostedion vitreum, predation. Canadian Journal of
Fisheries and Aquatic Sciences 50:1289–1298.
Koonce, J. F., T. B. Bagenal, R. F. Carline, K. E. F. Hokanson,
and M. Nagi. 1977. Factors influencing year-class strength of
percids: a summary and model of temperature effects.
Journal of the Fisheries Research Board of Canada 34:
Lankford, T. E., Jr., and T. E. Targett. 2001. Low-temperature
tolerance of age-0 Atlantic croakers: recruitment implications
for U.S. mid-Atlantic Estuaries. Transactions of the Amer-
ican Fisheries Society 130:236–249.
Magnuson, J. J. 2002. Signals from ice cover trends and
variability. Pages 3–14 in N. A. McGinn, editor. Fisheries in
a changing climate. American Fisheries Society Symposium
32. American Fisheries Society, Bethesda, Marlyand, USA.
Magnuson, J. J., et al. 2000. Historical trends in lake and river
ice cover in the northern hemisphere. Science 289:1743–1746.
Mayer, C. M., R. A. Keats, L. G. Rudstam, and E. L. Mills.
2002. Zebra mussels as ecosystem engineers: scale dependent
effects on benthic invertebrates in a large eutrophic lake.
DEAN G. FITZGERALD ET AL.1500
Vol. 16, No. 4
Journal of the North American Benthological Society 21: Download full-text
Mayer, C. M., A. J. VanDeValk, J. L. Forney, L. G. Rudstam,
and E. L. Mills. 2000. Response of yellow perch (Perca
flavescens) in Oneida Lake, New York, to the establishment
of zebra mussels (Dreissena polymorpha). Canadian Journal
of Fisheries and Aquatic Sciences 57:742–754.
McCollum, A. B., D. B. Bunnell, and R. A. Stein. 2003. Cold,
northern winters: the importance of temperature to over-
winter mortality of age-0 white crappies. Transactions of the
American Fisheries Society 132:977–987.
Mills, E. L., and J. L. Forney. 1988. Trophic dynamics and
development of freshwater pelagic food webs. Pages 11– 29 in
S. R. Carpenter, editor. Complex interactions in lake
communities. Springer-Verlag, New York, New York, USA.
Mills, E. L., J. L. Forney, M. D. Clady, and W. R. Schaffner.
1978. Oneida Lake. Pages 367–451 in J. A. Bloomfield,
editor. Lakes of New York State. Volume II. Academic
Press, New York, New York, USA.
Moran, M. D. 2001. Arguments for rejecting the sequential
Bonferroni in ecological studies. Oikos 2003:403–405.
Munch, S. B., M. Mangel, and D. O. Conover. 2003.
Quantifying natural selection on body size from field data:
winter mortality in Menidia menidia. Ecology 84:2168–2177.
Nielsen, L. A. 1980. Effects of walleye (Stizostedion vitreum
vitreum) predation on juvenile mortality and recruitment of
yellow perch (Perca flavescens) in Oneida Lake, New York.
Canadian Journal of Fisheries and Aquatic Sciences 37:11–
Paloheimo, J. E., and L. M. Dickie. 1966. Food and growth of
fishes. III. Relations among food, body size, and growth
efficiency. Journal of the Fisheries Research Board of
Polis, G. A., and D. R. Strong. 1996. Food web complexity and
community dynamics. American Naturalist 147:813–846.
Post, J. R., and D. O. Evans. 1989. Size-dependent overwinter
mortality of young-of-the-year yellow perch (Perca flaves-
cens): laboratory, in situ enclosure, and field experiments.
Canadian Journal of Fisheries and Aquatic Sciences 46:1958–
Post, J. R., and L. G. Rudstam. 1992. Fisheries management
and the interactive dynamics of walleye and perch popula-
tions. Pages 381–406 in J. F. Kitchell, editor. Food web
management: a case study of Lake Mendota. Springer-
Verlag, New York, New York, USA.
Prout, M. W., E. L. Mills, and J. L. Forney. 1990. Diet, growth,
and potential competitive interactions between age-0 white
perch and yellow perch in Oneida Lake, New York.
Transactions of the American Fisheries Society 119:966–975.
Radke, R. J., and R. Eckmann. 1999. First-year overwinter
mortality in Eurasian perch (Perca fluviatilis L.). Ecology of
Freshwater Fishes 8:94–101.
Ricker, W. E. 1969. Effects of size-selective mortality and
sampling bias on estimates of growth, mortality, production,
and yield. Journal of the Fisheries Research Board of Canada
Rose, K. A. 2000. Why are quantitative relationships between
environmental quality and fish populations so evasive?
Ecological Applications 10:367–385.
Rose, K. A., E. S. Rutherford, D. S. McDermot, J. L. Forney,
and E. L. Mills. 1999. Individual-based model of yellow
perch and walleye populations in Oneida Lake. Ecological
Roseman, E. F., E. L. Mills, J. L. Forney, and L. G. Rudstam.
1996. Evaluation of competition between age-0 yellow perch
(Perca flavescens) and gizzard shad (Dorosoma cepedianum)
in Oneida Lake, New York. Canadian Journal of Fisheries
and Aquatic Sciences 53:865–874.
Rudstam, L. G., A. J. VanDeValk, C. M. Adams, J. T. H.
Coleman, J. L. Forney, and M. E. Richmond. 2004.
Cormorant predation and the population dynamics of
walleye and yellow perch in Oneida Lake. Ecological
Sanderson, B. C., T. R. Hrabik, J. J. Magnuson, and D. M.
Post. 1999. Cyclic dynamics of a yellow perch (Perca
flavescens) population in an oligotrophic lake: evidence for
the role of intraspecific interactions. Canadian Journal of
Fisheries and Aquatic Sciences 56:1534–1542.
SAS Institute. 2001. SAS statistical software. JMP version
4.0.4. SAS Institute, Cary, North Carolina, USA.
Schultz, E. T., and D. O. Conover. 1999. The allometry of
energy reserve depletion: test of a mechanism for size-
dependent winter mortality. Oecologia 119:474–483.
Schultz, E. T., D. O. Conover, and A. Ehtisham. 1998. The
dead of winter: size-dependent variation and genetic differ-
ences in seasonal mortality among Atlantic silveside (Athe-
rinidae: Menidia menidia) from different latitudes. Canadian
Journal of Fisheries and Aquatic Sciences 55:1149–1157.
Shuter, B. J., C. K. Minns, and N. Lester. 2002. Climate
change, freshwater fish, and fisheries: case studies from
Ontario and their use in assessing potential impacts. Pages
77–88 in N. A. McGinn, editor. Fisheries in a changing
climate. American Fisheries Society Symposium 32. Ameri-
can Fisheries Society, Bethesda, Marlyand, USA.
Shuter, B. J., and J. R. Post. 1990. Climate, population
variability, and the zoogeography of temperate fishes.
Transactions of the American Fisheries Society 119:314–336.
Sogard, S. 1997. Size-selective mortality in the juvenile stage of
teleost fishes: a review. Bulletin of Marine Science 60:1129–
Sokal, R. R., and F. J. Rohlf. 1995. Biometry. Third edition.
W.H. Freeman, New York, New York, USA.
Stein, R. A., D. R. Devries, and J. M. Dettmers. 1995. Food-
web regulation by a planktivore: exploring the generality of
the trophic cascade hypothesis. Canadian Journal of Fish-
eries and Aquatic Sciences 52:2518–2526.
VanDeValk, A. J., C. M. Adams, L. G. Rudstam, J. L. Forney,
T. E. Brooking, M. Gerken, B. Young, and J. Hooper. 2002.
Comparison of angler and cormorant harvest of walleye and
yellow perch in Oneida Lake, New York. Transactions of the
American Fisheries Society 131:27–39..
Vanni, M. J., K. K. Arend, M. T. Bremigan, D. B. Bunnell, J.
E. Garvey, M. J. Gonza ´ lez, W. H. Renwick, P. A. Soranno,
and R. A. Stein. 2005. Linking landscapes and food webs:
effects of omnivorous fish and watersheds on reservoir
ecosystems. BioScience 55:155–167.
Walters, C. 2000. Natural selection for predation avoidance
tactics: implications for marine population and community
dynamics. Marine Ecology-Progress Series 208:309–313.
Wootton, R. J. 1998. Ecology of teleost fishes. Second edition.
Kluwer Academic Publishers, Dordrecht, The Netherlands.
August 20061501ANALYSIS OF FISH MORTALITY OVER WINTER